sqlglot.parser
1from __future__ import annotations 2 3import logging 4import typing as t 5from collections import defaultdict 6 7from sqlglot import exp 8from sqlglot.errors import ErrorLevel, ParseError, concat_messages, merge_errors 9from sqlglot.helper import apply_index_offset, ensure_list, seq_get 10from sqlglot.time import format_time 11from sqlglot.tokens import Token, Tokenizer, TokenType 12from sqlglot.trie import TrieResult, in_trie, new_trie 13 14if t.TYPE_CHECKING: 15 from sqlglot._typing import E, Lit 16 from sqlglot.dialects.dialect import Dialect, DialectType 17 18logger = logging.getLogger("sqlglot") 19 20 21def parse_var_map(args: t.List) -> exp.StarMap | exp.VarMap: 22 if len(args) == 1 and args[0].is_star: 23 return exp.StarMap(this=args[0]) 24 25 keys = [] 26 values = [] 27 for i in range(0, len(args), 2): 28 keys.append(args[i]) 29 values.append(args[i + 1]) 30 31 return exp.VarMap( 32 keys=exp.Array(expressions=keys), 33 values=exp.Array(expressions=values), 34 ) 35 36 37def parse_like(args: t.List) -> exp.Escape | exp.Like: 38 like = exp.Like(this=seq_get(args, 1), expression=seq_get(args, 0)) 39 return exp.Escape(this=like, expression=seq_get(args, 2)) if len(args) > 2 else like 40 41 42def binary_range_parser( 43 expr_type: t.Type[exp.Expression], 44) -> t.Callable[[Parser, t.Optional[exp.Expression]], t.Optional[exp.Expression]]: 45 return lambda self, this: self._parse_escape( 46 self.expression(expr_type, this=this, expression=self._parse_bitwise()) 47 ) 48 49 50def parse_logarithm(args: t.List, dialect: Dialect) -> exp.Func: 51 # Default argument order is base, expression 52 this = seq_get(args, 0) 53 expression = seq_get(args, 1) 54 55 if expression: 56 if not dialect.LOG_BASE_FIRST: 57 this, expression = expression, this 58 return exp.Log(this=this, expression=expression) 59 60 return (exp.Ln if dialect.parser_class.LOG_DEFAULTS_TO_LN else exp.Log)(this=this) 61 62 63def parse_extract_json_with_path(expr_type: t.Type[E]) -> t.Callable[[t.List, Dialect], E]: 64 def _parser(args: t.List, dialect: Dialect) -> E: 65 expression = expr_type( 66 this=seq_get(args, 0), expression=dialect.to_json_path(seq_get(args, 1)) 67 ) 68 if len(args) > 2 and expr_type is exp.JSONExtract: 69 expression.set("expressions", args[2:]) 70 71 return expression 72 73 return _parser 74 75 76class _Parser(type): 77 def __new__(cls, clsname, bases, attrs): 78 klass = super().__new__(cls, clsname, bases, attrs) 79 80 klass.SHOW_TRIE = new_trie(key.split(" ") for key in klass.SHOW_PARSERS) 81 klass.SET_TRIE = new_trie(key.split(" ") for key in klass.SET_PARSERS) 82 83 return klass 84 85 86class Parser(metaclass=_Parser): 87 """ 88 Parser consumes a list of tokens produced by the Tokenizer and produces a parsed syntax tree. 89 90 Args: 91 error_level: The desired error level. 92 Default: ErrorLevel.IMMEDIATE 93 error_message_context: Determines the amount of context to capture from a 94 query string when displaying the error message (in number of characters). 95 Default: 100 96 max_errors: Maximum number of error messages to include in a raised ParseError. 97 This is only relevant if error_level is ErrorLevel.RAISE. 98 Default: 3 99 """ 100 101 FUNCTIONS: t.Dict[str, t.Callable] = { 102 **{name: func.from_arg_list for name, func in exp.FUNCTION_BY_NAME.items()}, 103 "CONCAT": lambda args, dialect: exp.Concat( 104 expressions=args, 105 safe=not dialect.STRICT_STRING_CONCAT, 106 coalesce=dialect.CONCAT_COALESCE, 107 ), 108 "CONCAT_WS": lambda args, dialect: exp.ConcatWs( 109 expressions=args, 110 safe=not dialect.STRICT_STRING_CONCAT, 111 coalesce=dialect.CONCAT_COALESCE, 112 ), 113 "DATE_TO_DATE_STR": lambda args: exp.Cast( 114 this=seq_get(args, 0), 115 to=exp.DataType(this=exp.DataType.Type.TEXT), 116 ), 117 "GLOB": lambda args: exp.Glob(this=seq_get(args, 1), expression=seq_get(args, 0)), 118 "JSON_EXTRACT": parse_extract_json_with_path(exp.JSONExtract), 119 "JSON_EXTRACT_SCALAR": parse_extract_json_with_path(exp.JSONExtractScalar), 120 "JSON_EXTRACT_PATH_TEXT": parse_extract_json_with_path(exp.JSONExtractScalar), 121 "LIKE": parse_like, 122 "LOG": parse_logarithm, 123 "TIME_TO_TIME_STR": lambda args: exp.Cast( 124 this=seq_get(args, 0), 125 to=exp.DataType(this=exp.DataType.Type.TEXT), 126 ), 127 "TS_OR_DS_TO_DATE_STR": lambda args: exp.Substring( 128 this=exp.Cast( 129 this=seq_get(args, 0), 130 to=exp.DataType(this=exp.DataType.Type.TEXT), 131 ), 132 start=exp.Literal.number(1), 133 length=exp.Literal.number(10), 134 ), 135 "VAR_MAP": parse_var_map, 136 } 137 138 NO_PAREN_FUNCTIONS = { 139 TokenType.CURRENT_DATE: exp.CurrentDate, 140 TokenType.CURRENT_DATETIME: exp.CurrentDate, 141 TokenType.CURRENT_TIME: exp.CurrentTime, 142 TokenType.CURRENT_TIMESTAMP: exp.CurrentTimestamp, 143 TokenType.CURRENT_USER: exp.CurrentUser, 144 } 145 146 STRUCT_TYPE_TOKENS = { 147 TokenType.NESTED, 148 TokenType.STRUCT, 149 } 150 151 NESTED_TYPE_TOKENS = { 152 TokenType.ARRAY, 153 TokenType.LOWCARDINALITY, 154 TokenType.MAP, 155 TokenType.NULLABLE, 156 *STRUCT_TYPE_TOKENS, 157 } 158 159 ENUM_TYPE_TOKENS = { 160 TokenType.ENUM, 161 TokenType.ENUM8, 162 TokenType.ENUM16, 163 } 164 165 AGGREGATE_TYPE_TOKENS = { 166 TokenType.AGGREGATEFUNCTION, 167 TokenType.SIMPLEAGGREGATEFUNCTION, 168 } 169 170 TYPE_TOKENS = { 171 TokenType.BIT, 172 TokenType.BOOLEAN, 173 TokenType.TINYINT, 174 TokenType.UTINYINT, 175 TokenType.SMALLINT, 176 TokenType.USMALLINT, 177 TokenType.INT, 178 TokenType.UINT, 179 TokenType.BIGINT, 180 TokenType.UBIGINT, 181 TokenType.INT128, 182 TokenType.UINT128, 183 TokenType.INT256, 184 TokenType.UINT256, 185 TokenType.MEDIUMINT, 186 TokenType.UMEDIUMINT, 187 TokenType.FIXEDSTRING, 188 TokenType.FLOAT, 189 TokenType.DOUBLE, 190 TokenType.CHAR, 191 TokenType.NCHAR, 192 TokenType.VARCHAR, 193 TokenType.NVARCHAR, 194 TokenType.BPCHAR, 195 TokenType.TEXT, 196 TokenType.MEDIUMTEXT, 197 TokenType.LONGTEXT, 198 TokenType.MEDIUMBLOB, 199 TokenType.LONGBLOB, 200 TokenType.BINARY, 201 TokenType.VARBINARY, 202 TokenType.JSON, 203 TokenType.JSONB, 204 TokenType.INTERVAL, 205 TokenType.TINYBLOB, 206 TokenType.TINYTEXT, 207 TokenType.TIME, 208 TokenType.TIMETZ, 209 TokenType.TIMESTAMP, 210 TokenType.TIMESTAMP_S, 211 TokenType.TIMESTAMP_MS, 212 TokenType.TIMESTAMP_NS, 213 TokenType.TIMESTAMPTZ, 214 TokenType.TIMESTAMPLTZ, 215 TokenType.DATETIME, 216 TokenType.DATETIME64, 217 TokenType.DATE, 218 TokenType.DATE32, 219 TokenType.INT4RANGE, 220 TokenType.INT4MULTIRANGE, 221 TokenType.INT8RANGE, 222 TokenType.INT8MULTIRANGE, 223 TokenType.NUMRANGE, 224 TokenType.NUMMULTIRANGE, 225 TokenType.TSRANGE, 226 TokenType.TSMULTIRANGE, 227 TokenType.TSTZRANGE, 228 TokenType.TSTZMULTIRANGE, 229 TokenType.DATERANGE, 230 TokenType.DATEMULTIRANGE, 231 TokenType.DECIMAL, 232 TokenType.UDECIMAL, 233 TokenType.BIGDECIMAL, 234 TokenType.UUID, 235 TokenType.GEOGRAPHY, 236 TokenType.GEOMETRY, 237 TokenType.HLLSKETCH, 238 TokenType.HSTORE, 239 TokenType.PSEUDO_TYPE, 240 TokenType.SUPER, 241 TokenType.SERIAL, 242 TokenType.SMALLSERIAL, 243 TokenType.BIGSERIAL, 244 TokenType.XML, 245 TokenType.YEAR, 246 TokenType.UNIQUEIDENTIFIER, 247 TokenType.USERDEFINED, 248 TokenType.MONEY, 249 TokenType.SMALLMONEY, 250 TokenType.ROWVERSION, 251 TokenType.IMAGE, 252 TokenType.VARIANT, 253 TokenType.OBJECT, 254 TokenType.OBJECT_IDENTIFIER, 255 TokenType.INET, 256 TokenType.IPADDRESS, 257 TokenType.IPPREFIX, 258 TokenType.IPV4, 259 TokenType.IPV6, 260 TokenType.UNKNOWN, 261 TokenType.NULL, 262 *ENUM_TYPE_TOKENS, 263 *NESTED_TYPE_TOKENS, 264 *AGGREGATE_TYPE_TOKENS, 265 } 266 267 SIGNED_TO_UNSIGNED_TYPE_TOKEN = { 268 TokenType.BIGINT: TokenType.UBIGINT, 269 TokenType.INT: TokenType.UINT, 270 TokenType.MEDIUMINT: TokenType.UMEDIUMINT, 271 TokenType.SMALLINT: TokenType.USMALLINT, 272 TokenType.TINYINT: TokenType.UTINYINT, 273 TokenType.DECIMAL: TokenType.UDECIMAL, 274 } 275 276 SUBQUERY_PREDICATES = { 277 TokenType.ANY: exp.Any, 278 TokenType.ALL: exp.All, 279 TokenType.EXISTS: exp.Exists, 280 TokenType.SOME: exp.Any, 281 } 282 283 RESERVED_TOKENS = { 284 *Tokenizer.SINGLE_TOKENS.values(), 285 TokenType.SELECT, 286 } 287 288 DB_CREATABLES = { 289 TokenType.DATABASE, 290 TokenType.SCHEMA, 291 TokenType.TABLE, 292 TokenType.VIEW, 293 TokenType.MODEL, 294 TokenType.DICTIONARY, 295 } 296 297 CREATABLES = { 298 TokenType.COLUMN, 299 TokenType.CONSTRAINT, 300 TokenType.FUNCTION, 301 TokenType.INDEX, 302 TokenType.PROCEDURE, 303 TokenType.FOREIGN_KEY, 304 *DB_CREATABLES, 305 } 306 307 # Tokens that can represent identifiers 308 ID_VAR_TOKENS = { 309 TokenType.VAR, 310 TokenType.ANTI, 311 TokenType.APPLY, 312 TokenType.ASC, 313 TokenType.AUTO_INCREMENT, 314 TokenType.BEGIN, 315 TokenType.BPCHAR, 316 TokenType.CACHE, 317 TokenType.CASE, 318 TokenType.COLLATE, 319 TokenType.COMMAND, 320 TokenType.COMMENT, 321 TokenType.COMMIT, 322 TokenType.CONSTRAINT, 323 TokenType.DEFAULT, 324 TokenType.DELETE, 325 TokenType.DESC, 326 TokenType.DESCRIBE, 327 TokenType.DICTIONARY, 328 TokenType.DIV, 329 TokenType.END, 330 TokenType.EXECUTE, 331 TokenType.ESCAPE, 332 TokenType.FALSE, 333 TokenType.FIRST, 334 TokenType.FILTER, 335 TokenType.FINAL, 336 TokenType.FORMAT, 337 TokenType.FULL, 338 TokenType.IS, 339 TokenType.ISNULL, 340 TokenType.INTERVAL, 341 TokenType.KEEP, 342 TokenType.KILL, 343 TokenType.LEFT, 344 TokenType.LOAD, 345 TokenType.MERGE, 346 TokenType.NATURAL, 347 TokenType.NEXT, 348 TokenType.OFFSET, 349 TokenType.OPERATOR, 350 TokenType.ORDINALITY, 351 TokenType.OVERLAPS, 352 TokenType.OVERWRITE, 353 TokenType.PARTITION, 354 TokenType.PERCENT, 355 TokenType.PIVOT, 356 TokenType.PRAGMA, 357 TokenType.RANGE, 358 TokenType.RECURSIVE, 359 TokenType.REFERENCES, 360 TokenType.REFRESH, 361 TokenType.REPLACE, 362 TokenType.RIGHT, 363 TokenType.ROW, 364 TokenType.ROWS, 365 TokenType.SEMI, 366 TokenType.SET, 367 TokenType.SETTINGS, 368 TokenType.SHOW, 369 TokenType.TEMPORARY, 370 TokenType.TOP, 371 TokenType.TRUE, 372 TokenType.UNIQUE, 373 TokenType.UNPIVOT, 374 TokenType.UPDATE, 375 TokenType.USE, 376 TokenType.VOLATILE, 377 TokenType.WINDOW, 378 *CREATABLES, 379 *SUBQUERY_PREDICATES, 380 *TYPE_TOKENS, 381 *NO_PAREN_FUNCTIONS, 382 } 383 384 INTERVAL_VARS = ID_VAR_TOKENS - {TokenType.END} 385 386 TABLE_ALIAS_TOKENS = ID_VAR_TOKENS - { 387 TokenType.ANTI, 388 TokenType.APPLY, 389 TokenType.ASOF, 390 TokenType.FULL, 391 TokenType.LEFT, 392 TokenType.LOCK, 393 TokenType.NATURAL, 394 TokenType.OFFSET, 395 TokenType.RIGHT, 396 TokenType.SEMI, 397 TokenType.WINDOW, 398 } 399 400 COMMENT_TABLE_ALIAS_TOKENS = TABLE_ALIAS_TOKENS - {TokenType.IS} 401 402 UPDATE_ALIAS_TOKENS = TABLE_ALIAS_TOKENS - {TokenType.SET} 403 404 TRIM_TYPES = {"LEADING", "TRAILING", "BOTH"} 405 406 FUNC_TOKENS = { 407 TokenType.COLLATE, 408 TokenType.COMMAND, 409 TokenType.CURRENT_DATE, 410 TokenType.CURRENT_DATETIME, 411 TokenType.CURRENT_TIMESTAMP, 412 TokenType.CURRENT_TIME, 413 TokenType.CURRENT_USER, 414 TokenType.FILTER, 415 TokenType.FIRST, 416 TokenType.FORMAT, 417 TokenType.GLOB, 418 TokenType.IDENTIFIER, 419 TokenType.INDEX, 420 TokenType.ISNULL, 421 TokenType.ILIKE, 422 TokenType.INSERT, 423 TokenType.LIKE, 424 TokenType.MERGE, 425 TokenType.OFFSET, 426 TokenType.PRIMARY_KEY, 427 TokenType.RANGE, 428 TokenType.REPLACE, 429 TokenType.RLIKE, 430 TokenType.ROW, 431 TokenType.UNNEST, 432 TokenType.VAR, 433 TokenType.LEFT, 434 TokenType.RIGHT, 435 TokenType.DATE, 436 TokenType.DATETIME, 437 TokenType.TABLE, 438 TokenType.TIMESTAMP, 439 TokenType.TIMESTAMPTZ, 440 TokenType.WINDOW, 441 TokenType.XOR, 442 *TYPE_TOKENS, 443 *SUBQUERY_PREDICATES, 444 } 445 446 CONJUNCTION = { 447 TokenType.AND: exp.And, 448 TokenType.OR: exp.Or, 449 } 450 451 EQUALITY = { 452 TokenType.COLON_EQ: exp.PropertyEQ, 453 TokenType.EQ: exp.EQ, 454 TokenType.NEQ: exp.NEQ, 455 TokenType.NULLSAFE_EQ: exp.NullSafeEQ, 456 } 457 458 COMPARISON = { 459 TokenType.GT: exp.GT, 460 TokenType.GTE: exp.GTE, 461 TokenType.LT: exp.LT, 462 TokenType.LTE: exp.LTE, 463 } 464 465 BITWISE = { 466 TokenType.AMP: exp.BitwiseAnd, 467 TokenType.CARET: exp.BitwiseXor, 468 TokenType.PIPE: exp.BitwiseOr, 469 } 470 471 TERM = { 472 TokenType.DASH: exp.Sub, 473 TokenType.PLUS: exp.Add, 474 TokenType.MOD: exp.Mod, 475 TokenType.COLLATE: exp.Collate, 476 } 477 478 FACTOR = { 479 TokenType.DIV: exp.IntDiv, 480 TokenType.LR_ARROW: exp.Distance, 481 TokenType.SLASH: exp.Div, 482 TokenType.STAR: exp.Mul, 483 } 484 485 EXPONENT: t.Dict[TokenType, t.Type[exp.Expression]] = {} 486 487 TIMES = { 488 TokenType.TIME, 489 TokenType.TIMETZ, 490 } 491 492 TIMESTAMPS = { 493 TokenType.TIMESTAMP, 494 TokenType.TIMESTAMPTZ, 495 TokenType.TIMESTAMPLTZ, 496 *TIMES, 497 } 498 499 SET_OPERATIONS = { 500 TokenType.UNION, 501 TokenType.INTERSECT, 502 TokenType.EXCEPT, 503 } 504 505 JOIN_METHODS = { 506 TokenType.NATURAL, 507 TokenType.ASOF, 508 } 509 510 JOIN_SIDES = { 511 TokenType.LEFT, 512 TokenType.RIGHT, 513 TokenType.FULL, 514 } 515 516 JOIN_KINDS = { 517 TokenType.INNER, 518 TokenType.OUTER, 519 TokenType.CROSS, 520 TokenType.SEMI, 521 TokenType.ANTI, 522 } 523 524 JOIN_HINTS: t.Set[str] = set() 525 526 LAMBDAS = { 527 TokenType.ARROW: lambda self, expressions: self.expression( 528 exp.Lambda, 529 this=self._replace_lambda( 530 self._parse_conjunction(), 531 {node.name for node in expressions}, 532 ), 533 expressions=expressions, 534 ), 535 TokenType.FARROW: lambda self, expressions: self.expression( 536 exp.Kwarg, 537 this=exp.var(expressions[0].name), 538 expression=self._parse_conjunction(), 539 ), 540 } 541 542 COLUMN_OPERATORS = { 543 TokenType.DOT: None, 544 TokenType.DCOLON: lambda self, this, to: self.expression( 545 exp.Cast if self.STRICT_CAST else exp.TryCast, 546 this=this, 547 to=to, 548 ), 549 TokenType.ARROW: lambda self, this, path: self.expression( 550 exp.JSONExtract, 551 this=this, 552 expression=self.dialect.to_json_path(path), 553 ), 554 TokenType.DARROW: lambda self, this, path: self.expression( 555 exp.JSONExtractScalar, 556 this=this, 557 expression=self.dialect.to_json_path(path), 558 ), 559 TokenType.HASH_ARROW: lambda self, this, path: self.expression( 560 exp.JSONBExtract, 561 this=this, 562 expression=path, 563 ), 564 TokenType.DHASH_ARROW: lambda self, this, path: self.expression( 565 exp.JSONBExtractScalar, 566 this=this, 567 expression=path, 568 ), 569 TokenType.PLACEHOLDER: lambda self, this, key: self.expression( 570 exp.JSONBContains, 571 this=this, 572 expression=key, 573 ), 574 } 575 576 EXPRESSION_PARSERS = { 577 exp.Cluster: lambda self: self._parse_sort(exp.Cluster, TokenType.CLUSTER_BY), 578 exp.Column: lambda self: self._parse_column(), 579 exp.Condition: lambda self: self._parse_conjunction(), 580 exp.DataType: lambda self: self._parse_types(allow_identifiers=False), 581 exp.Expression: lambda self: self._parse_statement(), 582 exp.From: lambda self: self._parse_from(), 583 exp.Group: lambda self: self._parse_group(), 584 exp.Having: lambda self: self._parse_having(), 585 exp.Identifier: lambda self: self._parse_id_var(), 586 exp.Join: lambda self: self._parse_join(), 587 exp.Lambda: lambda self: self._parse_lambda(), 588 exp.Lateral: lambda self: self._parse_lateral(), 589 exp.Limit: lambda self: self._parse_limit(), 590 exp.Offset: lambda self: self._parse_offset(), 591 exp.Order: lambda self: self._parse_order(), 592 exp.Ordered: lambda self: self._parse_ordered(), 593 exp.Properties: lambda self: self._parse_properties(), 594 exp.Qualify: lambda self: self._parse_qualify(), 595 exp.Returning: lambda self: self._parse_returning(), 596 exp.Sort: lambda self: self._parse_sort(exp.Sort, TokenType.SORT_BY), 597 exp.Table: lambda self: self._parse_table_parts(), 598 exp.TableAlias: lambda self: self._parse_table_alias(), 599 exp.When: lambda self: seq_get(self._parse_when_matched(), 0), 600 exp.Where: lambda self: self._parse_where(), 601 exp.Window: lambda self: self._parse_named_window(), 602 exp.With: lambda self: self._parse_with(), 603 "JOIN_TYPE": lambda self: self._parse_join_parts(), 604 } 605 606 STATEMENT_PARSERS = { 607 TokenType.ALTER: lambda self: self._parse_alter(), 608 TokenType.BEGIN: lambda self: self._parse_transaction(), 609 TokenType.CACHE: lambda self: self._parse_cache(), 610 TokenType.COMMIT: lambda self: self._parse_commit_or_rollback(), 611 TokenType.COMMENT: lambda self: self._parse_comment(), 612 TokenType.CREATE: lambda self: self._parse_create(), 613 TokenType.DELETE: lambda self: self._parse_delete(), 614 TokenType.DESC: lambda self: self._parse_describe(), 615 TokenType.DESCRIBE: lambda self: self._parse_describe(), 616 TokenType.DROP: lambda self: self._parse_drop(), 617 TokenType.INSERT: lambda self: self._parse_insert(), 618 TokenType.KILL: lambda self: self._parse_kill(), 619 TokenType.LOAD: lambda self: self._parse_load(), 620 TokenType.MERGE: lambda self: self._parse_merge(), 621 TokenType.PIVOT: lambda self: self._parse_simplified_pivot(), 622 TokenType.PRAGMA: lambda self: self.expression(exp.Pragma, this=self._parse_expression()), 623 TokenType.REFRESH: lambda self: self._parse_refresh(), 624 TokenType.ROLLBACK: lambda self: self._parse_commit_or_rollback(), 625 TokenType.SET: lambda self: self._parse_set(), 626 TokenType.UNCACHE: lambda self: self._parse_uncache(), 627 TokenType.UPDATE: lambda self: self._parse_update(), 628 TokenType.USE: lambda self: self.expression( 629 exp.Use, 630 kind=self._match_texts(("ROLE", "WAREHOUSE", "DATABASE", "SCHEMA")) 631 and exp.var(self._prev.text), 632 this=self._parse_table(schema=False), 633 ), 634 } 635 636 UNARY_PARSERS = { 637 TokenType.PLUS: lambda self: self._parse_unary(), # Unary + is handled as a no-op 638 TokenType.NOT: lambda self: self.expression(exp.Not, this=self._parse_equality()), 639 TokenType.TILDA: lambda self: self.expression(exp.BitwiseNot, this=self._parse_unary()), 640 TokenType.DASH: lambda self: self.expression(exp.Neg, this=self._parse_unary()), 641 } 642 643 PRIMARY_PARSERS = { 644 TokenType.STRING: lambda self, token: self.expression( 645 exp.Literal, this=token.text, is_string=True 646 ), 647 TokenType.NUMBER: lambda self, token: self.expression( 648 exp.Literal, this=token.text, is_string=False 649 ), 650 TokenType.STAR: lambda self, _: self.expression( 651 exp.Star, **{"except": self._parse_except(), "replace": self._parse_replace()} 652 ), 653 TokenType.NULL: lambda self, _: self.expression(exp.Null), 654 TokenType.TRUE: lambda self, _: self.expression(exp.Boolean, this=True), 655 TokenType.FALSE: lambda self, _: self.expression(exp.Boolean, this=False), 656 TokenType.BIT_STRING: lambda self, token: self.expression(exp.BitString, this=token.text), 657 TokenType.HEX_STRING: lambda self, token: self.expression(exp.HexString, this=token.text), 658 TokenType.BYTE_STRING: lambda self, token: self.expression(exp.ByteString, this=token.text), 659 TokenType.INTRODUCER: lambda self, token: self._parse_introducer(token), 660 TokenType.NATIONAL_STRING: lambda self, token: self.expression( 661 exp.National, this=token.text 662 ), 663 TokenType.RAW_STRING: lambda self, token: self.expression(exp.RawString, this=token.text), 664 TokenType.HEREDOC_STRING: lambda self, token: self.expression( 665 exp.RawString, this=token.text 666 ), 667 TokenType.UNICODE_STRING: lambda self, token: self.expression( 668 exp.UnicodeString, 669 this=token.text, 670 escape=self._match_text_seq("UESCAPE") and self._parse_string(), 671 ), 672 TokenType.SESSION_PARAMETER: lambda self, _: self._parse_session_parameter(), 673 } 674 675 PLACEHOLDER_PARSERS = { 676 TokenType.PLACEHOLDER: lambda self: self.expression(exp.Placeholder), 677 TokenType.PARAMETER: lambda self: self._parse_parameter(), 678 TokenType.COLON: lambda self: ( 679 self.expression(exp.Placeholder, this=self._prev.text) 680 if self._match(TokenType.NUMBER) or self._match_set(self.ID_VAR_TOKENS) 681 else None 682 ), 683 } 684 685 RANGE_PARSERS = { 686 TokenType.BETWEEN: lambda self, this: self._parse_between(this), 687 TokenType.GLOB: binary_range_parser(exp.Glob), 688 TokenType.ILIKE: binary_range_parser(exp.ILike), 689 TokenType.IN: lambda self, this: self._parse_in(this), 690 TokenType.IRLIKE: binary_range_parser(exp.RegexpILike), 691 TokenType.IS: lambda self, this: self._parse_is(this), 692 TokenType.LIKE: binary_range_parser(exp.Like), 693 TokenType.OVERLAPS: binary_range_parser(exp.Overlaps), 694 TokenType.RLIKE: binary_range_parser(exp.RegexpLike), 695 TokenType.SIMILAR_TO: binary_range_parser(exp.SimilarTo), 696 TokenType.FOR: lambda self, this: self._parse_comprehension(this), 697 } 698 699 PROPERTY_PARSERS: t.Dict[str, t.Callable] = { 700 "ALGORITHM": lambda self: self._parse_property_assignment(exp.AlgorithmProperty), 701 "AUTO": lambda self: self._parse_auto_property(), 702 "AUTO_INCREMENT": lambda self: self._parse_property_assignment(exp.AutoIncrementProperty), 703 "BLOCKCOMPRESSION": lambda self: self._parse_blockcompression(), 704 "CHARSET": lambda self, **kwargs: self._parse_character_set(**kwargs), 705 "CHARACTER SET": lambda self, **kwargs: self._parse_character_set(**kwargs), 706 "CHECKSUM": lambda self: self._parse_checksum(), 707 "CLUSTER BY": lambda self: self._parse_cluster(), 708 "CLUSTERED": lambda self: self._parse_clustered_by(), 709 "COLLATE": lambda self, **kwargs: self._parse_property_assignment( 710 exp.CollateProperty, **kwargs 711 ), 712 "COMMENT": lambda self: self._parse_property_assignment(exp.SchemaCommentProperty), 713 "CONTAINS": lambda self: self._parse_contains_property(), 714 "COPY": lambda self: self._parse_copy_property(), 715 "DATABLOCKSIZE": lambda self, **kwargs: self._parse_datablocksize(**kwargs), 716 "DEFINER": lambda self: self._parse_definer(), 717 "DETERMINISTIC": lambda self: self.expression( 718 exp.StabilityProperty, this=exp.Literal.string("IMMUTABLE") 719 ), 720 "DISTKEY": lambda self: self._parse_distkey(), 721 "DISTSTYLE": lambda self: self._parse_property_assignment(exp.DistStyleProperty), 722 "ENGINE": lambda self: self._parse_property_assignment(exp.EngineProperty), 723 "EXECUTE": lambda self: self._parse_property_assignment(exp.ExecuteAsProperty), 724 "EXTERNAL": lambda self: self.expression(exp.ExternalProperty), 725 "FALLBACK": lambda self, **kwargs: self._parse_fallback(**kwargs), 726 "FORMAT": lambda self: self._parse_property_assignment(exp.FileFormatProperty), 727 "FREESPACE": lambda self: self._parse_freespace(), 728 "HEAP": lambda self: self.expression(exp.HeapProperty), 729 "IMMUTABLE": lambda self: self.expression( 730 exp.StabilityProperty, this=exp.Literal.string("IMMUTABLE") 731 ), 732 "INHERITS": lambda self: self.expression( 733 exp.InheritsProperty, expressions=self._parse_wrapped_csv(self._parse_table) 734 ), 735 "INPUT": lambda self: self.expression(exp.InputModelProperty, this=self._parse_schema()), 736 "JOURNAL": lambda self, **kwargs: self._parse_journal(**kwargs), 737 "LANGUAGE": lambda self: self._parse_property_assignment(exp.LanguageProperty), 738 "LAYOUT": lambda self: self._parse_dict_property(this="LAYOUT"), 739 "LIFETIME": lambda self: self._parse_dict_range(this="LIFETIME"), 740 "LIKE": lambda self: self._parse_create_like(), 741 "LOCATION": lambda self: self._parse_property_assignment(exp.LocationProperty), 742 "LOCK": lambda self: self._parse_locking(), 743 "LOCKING": lambda self: self._parse_locking(), 744 "LOG": lambda self, **kwargs: self._parse_log(**kwargs), 745 "MATERIALIZED": lambda self: self.expression(exp.MaterializedProperty), 746 "MERGEBLOCKRATIO": lambda self, **kwargs: self._parse_mergeblockratio(**kwargs), 747 "MODIFIES": lambda self: self._parse_modifies_property(), 748 "MULTISET": lambda self: self.expression(exp.SetProperty, multi=True), 749 "NO": lambda self: self._parse_no_property(), 750 "ON": lambda self: self._parse_on_property(), 751 "ORDER BY": lambda self: self._parse_order(skip_order_token=True), 752 "OUTPUT": lambda self: self.expression(exp.OutputModelProperty, this=self._parse_schema()), 753 "PARTITION": lambda self: self._parse_partitioned_of(), 754 "PARTITION BY": lambda self: self._parse_partitioned_by(), 755 "PARTITIONED BY": lambda self: self._parse_partitioned_by(), 756 "PARTITIONED_BY": lambda self: self._parse_partitioned_by(), 757 "PRIMARY KEY": lambda self: self._parse_primary_key(in_props=True), 758 "RANGE": lambda self: self._parse_dict_range(this="RANGE"), 759 "READS": lambda self: self._parse_reads_property(), 760 "REMOTE": lambda self: self._parse_remote_with_connection(), 761 "RETURNS": lambda self: self._parse_returns(), 762 "ROW": lambda self: self._parse_row(), 763 "ROW_FORMAT": lambda self: self._parse_property_assignment(exp.RowFormatProperty), 764 "SAMPLE": lambda self: self.expression( 765 exp.SampleProperty, this=self._match_text_seq("BY") and self._parse_bitwise() 766 ), 767 "SET": lambda self: self.expression(exp.SetProperty, multi=False), 768 "SETTINGS": lambda self: self.expression( 769 exp.SettingsProperty, expressions=self._parse_csv(self._parse_set_item) 770 ), 771 "SORTKEY": lambda self: self._parse_sortkey(), 772 "SOURCE": lambda self: self._parse_dict_property(this="SOURCE"), 773 "STABLE": lambda self: self.expression( 774 exp.StabilityProperty, this=exp.Literal.string("STABLE") 775 ), 776 "STORED": lambda self: self._parse_stored(), 777 "SYSTEM_VERSIONING": lambda self: self._parse_system_versioning_property(), 778 "TBLPROPERTIES": lambda self: self._parse_wrapped_csv(self._parse_property), 779 "TEMP": lambda self: self.expression(exp.TemporaryProperty), 780 "TEMPORARY": lambda self: self.expression(exp.TemporaryProperty), 781 "TO": lambda self: self._parse_to_table(), 782 "TRANSIENT": lambda self: self.expression(exp.TransientProperty), 783 "TRANSFORM": lambda self: self.expression( 784 exp.TransformModelProperty, expressions=self._parse_wrapped_csv(self._parse_expression) 785 ), 786 "TTL": lambda self: self._parse_ttl(), 787 "USING": lambda self: self._parse_property_assignment(exp.FileFormatProperty), 788 "VOLATILE": lambda self: self._parse_volatile_property(), 789 "WITH": lambda self: self._parse_with_property(), 790 } 791 792 CONSTRAINT_PARSERS = { 793 "AUTOINCREMENT": lambda self: self._parse_auto_increment(), 794 "AUTO_INCREMENT": lambda self: self._parse_auto_increment(), 795 "CASESPECIFIC": lambda self: self.expression(exp.CaseSpecificColumnConstraint, not_=False), 796 "CHARACTER SET": lambda self: self.expression( 797 exp.CharacterSetColumnConstraint, this=self._parse_var_or_string() 798 ), 799 "CHECK": lambda self: self.expression( 800 exp.CheckColumnConstraint, this=self._parse_wrapped(self._parse_conjunction) 801 ), 802 "COLLATE": lambda self: self.expression( 803 exp.CollateColumnConstraint, this=self._parse_var() 804 ), 805 "COMMENT": lambda self: self.expression( 806 exp.CommentColumnConstraint, this=self._parse_string() 807 ), 808 "COMPRESS": lambda self: self._parse_compress(), 809 "CLUSTERED": lambda self: self.expression( 810 exp.ClusteredColumnConstraint, this=self._parse_wrapped_csv(self._parse_ordered) 811 ), 812 "NONCLUSTERED": lambda self: self.expression( 813 exp.NonClusteredColumnConstraint, this=self._parse_wrapped_csv(self._parse_ordered) 814 ), 815 "DEFAULT": lambda self: self.expression( 816 exp.DefaultColumnConstraint, this=self._parse_bitwise() 817 ), 818 "ENCODE": lambda self: self.expression(exp.EncodeColumnConstraint, this=self._parse_var()), 819 "FOREIGN KEY": lambda self: self._parse_foreign_key(), 820 "FORMAT": lambda self: self.expression( 821 exp.DateFormatColumnConstraint, this=self._parse_var_or_string() 822 ), 823 "GENERATED": lambda self: self._parse_generated_as_identity(), 824 "IDENTITY": lambda self: self._parse_auto_increment(), 825 "INLINE": lambda self: self._parse_inline(), 826 "LIKE": lambda self: self._parse_create_like(), 827 "NOT": lambda self: self._parse_not_constraint(), 828 "NULL": lambda self: self.expression(exp.NotNullColumnConstraint, allow_null=True), 829 "ON": lambda self: ( 830 self._match(TokenType.UPDATE) 831 and self.expression(exp.OnUpdateColumnConstraint, this=self._parse_function()) 832 ) 833 or self.expression(exp.OnProperty, this=self._parse_id_var()), 834 "PATH": lambda self: self.expression(exp.PathColumnConstraint, this=self._parse_string()), 835 "PERIOD": lambda self: self._parse_period_for_system_time(), 836 "PRIMARY KEY": lambda self: self._parse_primary_key(), 837 "REFERENCES": lambda self: self._parse_references(match=False), 838 "TITLE": lambda self: self.expression( 839 exp.TitleColumnConstraint, this=self._parse_var_or_string() 840 ), 841 "TTL": lambda self: self.expression(exp.MergeTreeTTL, expressions=[self._parse_bitwise()]), 842 "UNIQUE": lambda self: self._parse_unique(), 843 "UPPERCASE": lambda self: self.expression(exp.UppercaseColumnConstraint), 844 "WITH": lambda self: self.expression( 845 exp.Properties, expressions=self._parse_wrapped_csv(self._parse_property) 846 ), 847 } 848 849 ALTER_PARSERS = { 850 "ADD": lambda self: self._parse_alter_table_add(), 851 "ALTER": lambda self: self._parse_alter_table_alter(), 852 "CLUSTER BY": lambda self: self._parse_cluster(wrapped=True), 853 "DELETE": lambda self: self.expression(exp.Delete, where=self._parse_where()), 854 "DROP": lambda self: self._parse_alter_table_drop(), 855 "RENAME": lambda self: self._parse_alter_table_rename(), 856 } 857 858 SCHEMA_UNNAMED_CONSTRAINTS = {"CHECK", "FOREIGN KEY", "LIKE", "PRIMARY KEY", "UNIQUE", "PERIOD"} 859 860 NO_PAREN_FUNCTION_PARSERS = { 861 "ANY": lambda self: self.expression(exp.Any, this=self._parse_bitwise()), 862 "CASE": lambda self: self._parse_case(), 863 "IF": lambda self: self._parse_if(), 864 "NEXT": lambda self: self._parse_next_value_for(), 865 } 866 867 INVALID_FUNC_NAME_TOKENS = { 868 TokenType.IDENTIFIER, 869 TokenType.STRING, 870 } 871 872 FUNCTIONS_WITH_ALIASED_ARGS = {"STRUCT"} 873 874 FUNCTION_PARSERS = { 875 "CAST": lambda self: self._parse_cast(self.STRICT_CAST), 876 "CONVERT": lambda self: self._parse_convert(self.STRICT_CAST), 877 "DECODE": lambda self: self._parse_decode(), 878 "EXTRACT": lambda self: self._parse_extract(), 879 "JSON_OBJECT": lambda self: self._parse_json_object(), 880 "JSON_OBJECTAGG": lambda self: self._parse_json_object(agg=True), 881 "JSON_TABLE": lambda self: self._parse_json_table(), 882 "MATCH": lambda self: self._parse_match_against(), 883 "OPENJSON": lambda self: self._parse_open_json(), 884 "POSITION": lambda self: self._parse_position(), 885 "PREDICT": lambda self: self._parse_predict(), 886 "SAFE_CAST": lambda self: self._parse_cast(False, safe=True), 887 "STRING_AGG": lambda self: self._parse_string_agg(), 888 "SUBSTRING": lambda self: self._parse_substring(), 889 "TRIM": lambda self: self._parse_trim(), 890 "TRY_CAST": lambda self: self._parse_cast(False, safe=True), 891 "TRY_CONVERT": lambda self: self._parse_convert(False, safe=True), 892 } 893 894 QUERY_MODIFIER_PARSERS = { 895 TokenType.MATCH_RECOGNIZE: lambda self: ("match", self._parse_match_recognize()), 896 TokenType.WHERE: lambda self: ("where", self._parse_where()), 897 TokenType.GROUP_BY: lambda self: ("group", self._parse_group()), 898 TokenType.HAVING: lambda self: ("having", self._parse_having()), 899 TokenType.QUALIFY: lambda self: ("qualify", self._parse_qualify()), 900 TokenType.WINDOW: lambda self: ("windows", self._parse_window_clause()), 901 TokenType.ORDER_BY: lambda self: ("order", self._parse_order()), 902 TokenType.LIMIT: lambda self: ("limit", self._parse_limit()), 903 TokenType.FETCH: lambda self: ("limit", self._parse_limit()), 904 TokenType.OFFSET: lambda self: ("offset", self._parse_offset()), 905 TokenType.FOR: lambda self: ("locks", self._parse_locks()), 906 TokenType.LOCK: lambda self: ("locks", self._parse_locks()), 907 TokenType.TABLE_SAMPLE: lambda self: ("sample", self._parse_table_sample(as_modifier=True)), 908 TokenType.USING: lambda self: ("sample", self._parse_table_sample(as_modifier=True)), 909 TokenType.CLUSTER_BY: lambda self: ( 910 "cluster", 911 self._parse_sort(exp.Cluster, TokenType.CLUSTER_BY), 912 ), 913 TokenType.DISTRIBUTE_BY: lambda self: ( 914 "distribute", 915 self._parse_sort(exp.Distribute, TokenType.DISTRIBUTE_BY), 916 ), 917 TokenType.SORT_BY: lambda self: ("sort", self._parse_sort(exp.Sort, TokenType.SORT_BY)), 918 TokenType.CONNECT_BY: lambda self: ("connect", self._parse_connect(skip_start_token=True)), 919 TokenType.START_WITH: lambda self: ("connect", self._parse_connect()), 920 } 921 922 SET_PARSERS = { 923 "GLOBAL": lambda self: self._parse_set_item_assignment("GLOBAL"), 924 "LOCAL": lambda self: self._parse_set_item_assignment("LOCAL"), 925 "SESSION": lambda self: self._parse_set_item_assignment("SESSION"), 926 "TRANSACTION": lambda self: self._parse_set_transaction(), 927 } 928 929 SHOW_PARSERS: t.Dict[str, t.Callable] = {} 930 931 TYPE_LITERAL_PARSERS = { 932 exp.DataType.Type.JSON: lambda self, this, _: self.expression(exp.ParseJSON, this=this), 933 } 934 935 MODIFIABLES = (exp.Subquery, exp.Subqueryable, exp.Table) 936 937 DDL_SELECT_TOKENS = {TokenType.SELECT, TokenType.WITH, TokenType.L_PAREN} 938 939 PRE_VOLATILE_TOKENS = {TokenType.CREATE, TokenType.REPLACE, TokenType.UNIQUE} 940 941 TRANSACTION_KIND = {"DEFERRED", "IMMEDIATE", "EXCLUSIVE"} 942 TRANSACTION_CHARACTERISTICS = { 943 "ISOLATION LEVEL REPEATABLE READ", 944 "ISOLATION LEVEL READ COMMITTED", 945 "ISOLATION LEVEL READ UNCOMMITTED", 946 "ISOLATION LEVEL SERIALIZABLE", 947 "READ WRITE", 948 "READ ONLY", 949 } 950 951 INSERT_ALTERNATIVES = {"ABORT", "FAIL", "IGNORE", "REPLACE", "ROLLBACK"} 952 953 CLONE_KEYWORDS = {"CLONE", "COPY"} 954 HISTORICAL_DATA_KIND = {"TIMESTAMP", "OFFSET", "STATEMENT", "STREAM"} 955 956 OPCLASS_FOLLOW_KEYWORDS = {"ASC", "DESC", "NULLS"} 957 OPTYPE_FOLLOW_TOKENS = {TokenType.COMMA, TokenType.R_PAREN} 958 959 TABLE_INDEX_HINT_TOKENS = {TokenType.FORCE, TokenType.IGNORE, TokenType.USE} 960 961 WINDOW_ALIAS_TOKENS = ID_VAR_TOKENS - {TokenType.ROWS} 962 WINDOW_BEFORE_PAREN_TOKENS = {TokenType.OVER} 963 WINDOW_SIDES = {"FOLLOWING", "PRECEDING"} 964 965 JSON_KEY_VALUE_SEPARATOR_TOKENS = {TokenType.COLON, TokenType.COMMA, TokenType.IS} 966 967 FETCH_TOKENS = ID_VAR_TOKENS - {TokenType.ROW, TokenType.ROWS, TokenType.PERCENT} 968 969 ADD_CONSTRAINT_TOKENS = {TokenType.CONSTRAINT, TokenType.PRIMARY_KEY, TokenType.FOREIGN_KEY} 970 971 DISTINCT_TOKENS = {TokenType.DISTINCT} 972 973 NULL_TOKENS = {TokenType.NULL} 974 975 UNNEST_OFFSET_ALIAS_TOKENS = ID_VAR_TOKENS - SET_OPERATIONS 976 977 STRICT_CAST = True 978 979 PREFIXED_PIVOT_COLUMNS = False 980 IDENTIFY_PIVOT_STRINGS = False 981 982 LOG_DEFAULTS_TO_LN = False 983 984 # Whether or not ADD is present for each column added by ALTER TABLE 985 ALTER_TABLE_ADD_REQUIRED_FOR_EACH_COLUMN = True 986 987 # Whether or not the table sample clause expects CSV syntax 988 TABLESAMPLE_CSV = False 989 990 # Whether or not the SET command needs a delimiter (e.g. "=") for assignments 991 SET_REQUIRES_ASSIGNMENT_DELIMITER = True 992 993 # Whether the TRIM function expects the characters to trim as its first argument 994 TRIM_PATTERN_FIRST = False 995 996 # Whether or not string aliases are supported `SELECT COUNT(*) 'count'` 997 STRING_ALIASES = False 998 999 # Whether query modifiers such as LIMIT are attached to the UNION node (vs its right operand) 1000 MODIFIERS_ATTACHED_TO_UNION = True 1001 UNION_MODIFIERS = {"order", "limit", "offset"} 1002 1003 # parses no parenthesis if statements as commands 1004 NO_PAREN_IF_COMMANDS = True 1005 1006 __slots__ = ( 1007 "error_level", 1008 "error_message_context", 1009 "max_errors", 1010 "dialect", 1011 "sql", 1012 "errors", 1013 "_tokens", 1014 "_index", 1015 "_curr", 1016 "_next", 1017 "_prev", 1018 "_prev_comments", 1019 ) 1020 1021 # Autofilled 1022 SHOW_TRIE: t.Dict = {} 1023 SET_TRIE: t.Dict = {} 1024 1025 def __init__( 1026 self, 1027 error_level: t.Optional[ErrorLevel] = None, 1028 error_message_context: int = 100, 1029 max_errors: int = 3, 1030 dialect: DialectType = None, 1031 ): 1032 from sqlglot.dialects import Dialect 1033 1034 self.error_level = error_level or ErrorLevel.IMMEDIATE 1035 self.error_message_context = error_message_context 1036 self.max_errors = max_errors 1037 self.dialect = Dialect.get_or_raise(dialect) 1038 self.reset() 1039 1040 def reset(self): 1041 self.sql = "" 1042 self.errors = [] 1043 self._tokens = [] 1044 self._index = 0 1045 self._curr = None 1046 self._next = None 1047 self._prev = None 1048 self._prev_comments = None 1049 1050 def parse( 1051 self, raw_tokens: t.List[Token], sql: t.Optional[str] = None 1052 ) -> t.List[t.Optional[exp.Expression]]: 1053 """ 1054 Parses a list of tokens and returns a list of syntax trees, one tree 1055 per parsed SQL statement. 1056 1057 Args: 1058 raw_tokens: The list of tokens. 1059 sql: The original SQL string, used to produce helpful debug messages. 1060 1061 Returns: 1062 The list of the produced syntax trees. 1063 """ 1064 return self._parse( 1065 parse_method=self.__class__._parse_statement, raw_tokens=raw_tokens, sql=sql 1066 ) 1067 1068 def parse_into( 1069 self, 1070 expression_types: exp.IntoType, 1071 raw_tokens: t.List[Token], 1072 sql: t.Optional[str] = None, 1073 ) -> t.List[t.Optional[exp.Expression]]: 1074 """ 1075 Parses a list of tokens into a given Expression type. If a collection of Expression 1076 types is given instead, this method will try to parse the token list into each one 1077 of them, stopping at the first for which the parsing succeeds. 1078 1079 Args: 1080 expression_types: The expression type(s) to try and parse the token list into. 1081 raw_tokens: The list of tokens. 1082 sql: The original SQL string, used to produce helpful debug messages. 1083 1084 Returns: 1085 The target Expression. 1086 """ 1087 errors = [] 1088 for expression_type in ensure_list(expression_types): 1089 parser = self.EXPRESSION_PARSERS.get(expression_type) 1090 if not parser: 1091 raise TypeError(f"No parser registered for {expression_type}") 1092 1093 try: 1094 return self._parse(parser, raw_tokens, sql) 1095 except ParseError as e: 1096 e.errors[0]["into_expression"] = expression_type 1097 errors.append(e) 1098 1099 raise ParseError( 1100 f"Failed to parse '{sql or raw_tokens}' into {expression_types}", 1101 errors=merge_errors(errors), 1102 ) from errors[-1] 1103 1104 def _parse( 1105 self, 1106 parse_method: t.Callable[[Parser], t.Optional[exp.Expression]], 1107 raw_tokens: t.List[Token], 1108 sql: t.Optional[str] = None, 1109 ) -> t.List[t.Optional[exp.Expression]]: 1110 self.reset() 1111 self.sql = sql or "" 1112 1113 total = len(raw_tokens) 1114 chunks: t.List[t.List[Token]] = [[]] 1115 1116 for i, token in enumerate(raw_tokens): 1117 if token.token_type == TokenType.SEMICOLON: 1118 if i < total - 1: 1119 chunks.append([]) 1120 else: 1121 chunks[-1].append(token) 1122 1123 expressions = [] 1124 1125 for tokens in chunks: 1126 self._index = -1 1127 self._tokens = tokens 1128 self._advance() 1129 1130 expressions.append(parse_method(self)) 1131 1132 if self._index < len(self._tokens): 1133 self.raise_error("Invalid expression / Unexpected token") 1134 1135 self.check_errors() 1136 1137 return expressions 1138 1139 def check_errors(self) -> None: 1140 """Logs or raises any found errors, depending on the chosen error level setting.""" 1141 if self.error_level == ErrorLevel.WARN: 1142 for error in self.errors: 1143 logger.error(str(error)) 1144 elif self.error_level == ErrorLevel.RAISE and self.errors: 1145 raise ParseError( 1146 concat_messages(self.errors, self.max_errors), 1147 errors=merge_errors(self.errors), 1148 ) 1149 1150 def raise_error(self, message: str, token: t.Optional[Token] = None) -> None: 1151 """ 1152 Appends an error in the list of recorded errors or raises it, depending on the chosen 1153 error level setting. 1154 """ 1155 token = token or self._curr or self._prev or Token.string("") 1156 start = token.start 1157 end = token.end + 1 1158 start_context = self.sql[max(start - self.error_message_context, 0) : start] 1159 highlight = self.sql[start:end] 1160 end_context = self.sql[end : end + self.error_message_context] 1161 1162 error = ParseError.new( 1163 f"{message}. Line {token.line}, Col: {token.col}.\n" 1164 f" {start_context}\033[4m{highlight}\033[0m{end_context}", 1165 description=message, 1166 line=token.line, 1167 col=token.col, 1168 start_context=start_context, 1169 highlight=highlight, 1170 end_context=end_context, 1171 ) 1172 1173 if self.error_level == ErrorLevel.IMMEDIATE: 1174 raise error 1175 1176 self.errors.append(error) 1177 1178 def expression( 1179 self, exp_class: t.Type[E], comments: t.Optional[t.List[str]] = None, **kwargs 1180 ) -> E: 1181 """ 1182 Creates a new, validated Expression. 1183 1184 Args: 1185 exp_class: The expression class to instantiate. 1186 comments: An optional list of comments to attach to the expression. 1187 kwargs: The arguments to set for the expression along with their respective values. 1188 1189 Returns: 1190 The target expression. 1191 """ 1192 instance = exp_class(**kwargs) 1193 instance.add_comments(comments) if comments else self._add_comments(instance) 1194 return self.validate_expression(instance) 1195 1196 def _add_comments(self, expression: t.Optional[exp.Expression]) -> None: 1197 if expression and self._prev_comments: 1198 expression.add_comments(self._prev_comments) 1199 self._prev_comments = None 1200 1201 def validate_expression(self, expression: E, args: t.Optional[t.List] = None) -> E: 1202 """ 1203 Validates an Expression, making sure that all its mandatory arguments are set. 1204 1205 Args: 1206 expression: The expression to validate. 1207 args: An optional list of items that was used to instantiate the expression, if it's a Func. 1208 1209 Returns: 1210 The validated expression. 1211 """ 1212 if self.error_level != ErrorLevel.IGNORE: 1213 for error_message in expression.error_messages(args): 1214 self.raise_error(error_message) 1215 1216 return expression 1217 1218 def _find_sql(self, start: Token, end: Token) -> str: 1219 return self.sql[start.start : end.end + 1] 1220 1221 def _is_connected(self) -> bool: 1222 return self._prev and self._curr and self._prev.end + 1 == self._curr.start 1223 1224 def _advance(self, times: int = 1) -> None: 1225 self._index += times 1226 self._curr = seq_get(self._tokens, self._index) 1227 self._next = seq_get(self._tokens, self._index + 1) 1228 1229 if self._index > 0: 1230 self._prev = self._tokens[self._index - 1] 1231 self._prev_comments = self._prev.comments 1232 else: 1233 self._prev = None 1234 self._prev_comments = None 1235 1236 def _retreat(self, index: int) -> None: 1237 if index != self._index: 1238 self._advance(index - self._index) 1239 1240 def _warn_unsupported(self) -> None: 1241 if len(self._tokens) <= 1: 1242 return 1243 1244 # We use _find_sql because self.sql may comprise multiple chunks, and we're only 1245 # interested in emitting a warning for the one being currently processed. 1246 sql = self._find_sql(self._tokens[0], self._tokens[-1])[: self.error_message_context] 1247 1248 logger.warning( 1249 f"'{sql}' contains unsupported syntax. Falling back to parsing as a 'Command'." 1250 ) 1251 1252 def _parse_command(self) -> exp.Command: 1253 self._warn_unsupported() 1254 return self.expression( 1255 exp.Command, this=self._prev.text.upper(), expression=self._parse_string() 1256 ) 1257 1258 def _parse_comment(self, allow_exists: bool = True) -> exp.Expression: 1259 start = self._prev 1260 exists = self._parse_exists() if allow_exists else None 1261 1262 self._match(TokenType.ON) 1263 1264 kind = self._match_set(self.CREATABLES) and self._prev 1265 if not kind: 1266 return self._parse_as_command(start) 1267 1268 if kind.token_type in (TokenType.FUNCTION, TokenType.PROCEDURE): 1269 this = self._parse_user_defined_function(kind=kind.token_type) 1270 elif kind.token_type == TokenType.TABLE: 1271 this = self._parse_table(alias_tokens=self.COMMENT_TABLE_ALIAS_TOKENS) 1272 elif kind.token_type == TokenType.COLUMN: 1273 this = self._parse_column() 1274 else: 1275 this = self._parse_id_var() 1276 1277 self._match(TokenType.IS) 1278 1279 return self.expression( 1280 exp.Comment, this=this, kind=kind.text, expression=self._parse_string(), exists=exists 1281 ) 1282 1283 def _parse_to_table( 1284 self, 1285 ) -> exp.ToTableProperty: 1286 table = self._parse_table_parts(schema=True) 1287 return self.expression(exp.ToTableProperty, this=table) 1288 1289 # https://clickhouse.com/docs/en/engines/table-engines/mergetree-family/mergetree#mergetree-table-ttl 1290 def _parse_ttl(self) -> exp.Expression: 1291 def _parse_ttl_action() -> t.Optional[exp.Expression]: 1292 this = self._parse_bitwise() 1293 1294 if self._match_text_seq("DELETE"): 1295 return self.expression(exp.MergeTreeTTLAction, this=this, delete=True) 1296 if self._match_text_seq("RECOMPRESS"): 1297 return self.expression( 1298 exp.MergeTreeTTLAction, this=this, recompress=self._parse_bitwise() 1299 ) 1300 if self._match_text_seq("TO", "DISK"): 1301 return self.expression( 1302 exp.MergeTreeTTLAction, this=this, to_disk=self._parse_string() 1303 ) 1304 if self._match_text_seq("TO", "VOLUME"): 1305 return self.expression( 1306 exp.MergeTreeTTLAction, this=this, to_volume=self._parse_string() 1307 ) 1308 1309 return this 1310 1311 expressions = self._parse_csv(_parse_ttl_action) 1312 where = self._parse_where() 1313 group = self._parse_group() 1314 1315 aggregates = None 1316 if group and self._match(TokenType.SET): 1317 aggregates = self._parse_csv(self._parse_set_item) 1318 1319 return self.expression( 1320 exp.MergeTreeTTL, 1321 expressions=expressions, 1322 where=where, 1323 group=group, 1324 aggregates=aggregates, 1325 ) 1326 1327 def _parse_statement(self) -> t.Optional[exp.Expression]: 1328 if self._curr is None: 1329 return None 1330 1331 if self._match_set(self.STATEMENT_PARSERS): 1332 return self.STATEMENT_PARSERS[self._prev.token_type](self) 1333 1334 if self._match_set(Tokenizer.COMMANDS): 1335 return self._parse_command() 1336 1337 expression = self._parse_expression() 1338 expression = self._parse_set_operations(expression) if expression else self._parse_select() 1339 return self._parse_query_modifiers(expression) 1340 1341 def _parse_drop(self, exists: bool = False) -> exp.Drop | exp.Command: 1342 start = self._prev 1343 temporary = self._match(TokenType.TEMPORARY) 1344 materialized = self._match_text_seq("MATERIALIZED") 1345 1346 kind = self._match_set(self.CREATABLES) and self._prev.text 1347 if not kind: 1348 return self._parse_as_command(start) 1349 1350 return self.expression( 1351 exp.Drop, 1352 comments=start.comments, 1353 exists=exists or self._parse_exists(), 1354 this=self._parse_table( 1355 schema=True, is_db_reference=self._prev.token_type == TokenType.SCHEMA 1356 ), 1357 kind=kind, 1358 temporary=temporary, 1359 materialized=materialized, 1360 cascade=self._match_text_seq("CASCADE"), 1361 constraints=self._match_text_seq("CONSTRAINTS"), 1362 purge=self._match_text_seq("PURGE"), 1363 ) 1364 1365 def _parse_exists(self, not_: bool = False) -> t.Optional[bool]: 1366 return ( 1367 self._match_text_seq("IF") 1368 and (not not_ or self._match(TokenType.NOT)) 1369 and self._match(TokenType.EXISTS) 1370 ) 1371 1372 def _parse_create(self) -> exp.Create | exp.Command: 1373 # Note: this can't be None because we've matched a statement parser 1374 start = self._prev 1375 comments = self._prev_comments 1376 1377 replace = ( 1378 start.token_type == TokenType.REPLACE 1379 or self._match_pair(TokenType.OR, TokenType.REPLACE) 1380 or self._match_pair(TokenType.OR, TokenType.ALTER) 1381 ) 1382 unique = self._match(TokenType.UNIQUE) 1383 1384 if self._match_pair(TokenType.TABLE, TokenType.FUNCTION, advance=False): 1385 self._advance() 1386 1387 properties = None 1388 create_token = self._match_set(self.CREATABLES) and self._prev 1389 1390 if not create_token: 1391 # exp.Properties.Location.POST_CREATE 1392 properties = self._parse_properties() 1393 create_token = self._match_set(self.CREATABLES) and self._prev 1394 1395 if not properties or not create_token: 1396 return self._parse_as_command(start) 1397 1398 exists = self._parse_exists(not_=True) 1399 this = None 1400 expression: t.Optional[exp.Expression] = None 1401 indexes = None 1402 no_schema_binding = None 1403 begin = None 1404 end = None 1405 clone = None 1406 1407 def extend_props(temp_props: t.Optional[exp.Properties]) -> None: 1408 nonlocal properties 1409 if properties and temp_props: 1410 properties.expressions.extend(temp_props.expressions) 1411 elif temp_props: 1412 properties = temp_props 1413 1414 if create_token.token_type in (TokenType.FUNCTION, TokenType.PROCEDURE): 1415 this = self._parse_user_defined_function(kind=create_token.token_type) 1416 1417 # exp.Properties.Location.POST_SCHEMA ("schema" here is the UDF's type signature) 1418 extend_props(self._parse_properties()) 1419 1420 expression = self._match(TokenType.ALIAS) and self._parse_heredoc() 1421 1422 if not expression: 1423 if self._match(TokenType.COMMAND): 1424 expression = self._parse_as_command(self._prev) 1425 else: 1426 begin = self._match(TokenType.BEGIN) 1427 return_ = self._match_text_seq("RETURN") 1428 1429 if self._match(TokenType.STRING, advance=False): 1430 # Takes care of BigQuery's JavaScript UDF definitions that end in an OPTIONS property 1431 # # https://cloud.google.com/bigquery/docs/reference/standard-sql/data-definition-language#create_function_statement 1432 expression = self._parse_string() 1433 extend_props(self._parse_properties()) 1434 else: 1435 expression = self._parse_statement() 1436 1437 end = self._match_text_seq("END") 1438 1439 if return_: 1440 expression = self.expression(exp.Return, this=expression) 1441 elif create_token.token_type == TokenType.INDEX: 1442 this = self._parse_index(index=self._parse_id_var()) 1443 elif create_token.token_type in self.DB_CREATABLES: 1444 table_parts = self._parse_table_parts( 1445 schema=True, is_db_reference=create_token.token_type == TokenType.SCHEMA 1446 ) 1447 1448 # exp.Properties.Location.POST_NAME 1449 self._match(TokenType.COMMA) 1450 extend_props(self._parse_properties(before=True)) 1451 1452 this = self._parse_schema(this=table_parts) 1453 1454 # exp.Properties.Location.POST_SCHEMA and POST_WITH 1455 extend_props(self._parse_properties()) 1456 1457 self._match(TokenType.ALIAS) 1458 if not self._match_set(self.DDL_SELECT_TOKENS, advance=False): 1459 # exp.Properties.Location.POST_ALIAS 1460 extend_props(self._parse_properties()) 1461 1462 expression = self._parse_ddl_select() 1463 1464 if create_token.token_type == TokenType.TABLE: 1465 # exp.Properties.Location.POST_EXPRESSION 1466 extend_props(self._parse_properties()) 1467 1468 indexes = [] 1469 while True: 1470 index = self._parse_index() 1471 1472 # exp.Properties.Location.POST_INDEX 1473 extend_props(self._parse_properties()) 1474 1475 if not index: 1476 break 1477 else: 1478 self._match(TokenType.COMMA) 1479 indexes.append(index) 1480 elif create_token.token_type == TokenType.VIEW: 1481 if self._match_text_seq("WITH", "NO", "SCHEMA", "BINDING"): 1482 no_schema_binding = True 1483 1484 shallow = self._match_text_seq("SHALLOW") 1485 1486 if self._match_texts(self.CLONE_KEYWORDS): 1487 copy = self._prev.text.lower() == "copy" 1488 clone = self.expression( 1489 exp.Clone, this=self._parse_table(schema=True), shallow=shallow, copy=copy 1490 ) 1491 1492 if self._curr: 1493 return self._parse_as_command(start) 1494 1495 return self.expression( 1496 exp.Create, 1497 comments=comments, 1498 this=this, 1499 kind=create_token.text.upper(), 1500 replace=replace, 1501 unique=unique, 1502 expression=expression, 1503 exists=exists, 1504 properties=properties, 1505 indexes=indexes, 1506 no_schema_binding=no_schema_binding, 1507 begin=begin, 1508 end=end, 1509 clone=clone, 1510 ) 1511 1512 def _parse_property_before(self) -> t.Optional[exp.Expression]: 1513 # only used for teradata currently 1514 self._match(TokenType.COMMA) 1515 1516 kwargs = { 1517 "no": self._match_text_seq("NO"), 1518 "dual": self._match_text_seq("DUAL"), 1519 "before": self._match_text_seq("BEFORE"), 1520 "default": self._match_text_seq("DEFAULT"), 1521 "local": (self._match_text_seq("LOCAL") and "LOCAL") 1522 or (self._match_text_seq("NOT", "LOCAL") and "NOT LOCAL"), 1523 "after": self._match_text_seq("AFTER"), 1524 "minimum": self._match_texts(("MIN", "MINIMUM")), 1525 "maximum": self._match_texts(("MAX", "MAXIMUM")), 1526 } 1527 1528 if self._match_texts(self.PROPERTY_PARSERS): 1529 parser = self.PROPERTY_PARSERS[self._prev.text.upper()] 1530 try: 1531 return parser(self, **{k: v for k, v in kwargs.items() if v}) 1532 except TypeError: 1533 self.raise_error(f"Cannot parse property '{self._prev.text}'") 1534 1535 return None 1536 1537 def _parse_property(self) -> t.Optional[exp.Expression]: 1538 if self._match_texts(self.PROPERTY_PARSERS): 1539 return self.PROPERTY_PARSERS[self._prev.text.upper()](self) 1540 1541 if self._match(TokenType.DEFAULT) and self._match_texts(self.PROPERTY_PARSERS): 1542 return self.PROPERTY_PARSERS[self._prev.text.upper()](self, default=True) 1543 1544 if self._match_text_seq("COMPOUND", "SORTKEY"): 1545 return self._parse_sortkey(compound=True) 1546 1547 if self._match_text_seq("SQL", "SECURITY"): 1548 return self.expression(exp.SqlSecurityProperty, definer=self._match_text_seq("DEFINER")) 1549 1550 index = self._index 1551 key = self._parse_column() 1552 1553 if not self._match(TokenType.EQ): 1554 self._retreat(index) 1555 return None 1556 1557 return self.expression( 1558 exp.Property, 1559 this=key.to_dot() if isinstance(key, exp.Column) else key, 1560 value=self._parse_column() or self._parse_var(any_token=True), 1561 ) 1562 1563 def _parse_stored(self) -> exp.FileFormatProperty: 1564 self._match(TokenType.ALIAS) 1565 1566 input_format = self._parse_string() if self._match_text_seq("INPUTFORMAT") else None 1567 output_format = self._parse_string() if self._match_text_seq("OUTPUTFORMAT") else None 1568 1569 return self.expression( 1570 exp.FileFormatProperty, 1571 this=( 1572 self.expression( 1573 exp.InputOutputFormat, input_format=input_format, output_format=output_format 1574 ) 1575 if input_format or output_format 1576 else self._parse_var_or_string() or self._parse_number() or self._parse_id_var() 1577 ), 1578 ) 1579 1580 def _parse_property_assignment(self, exp_class: t.Type[E], **kwargs: t.Any) -> E: 1581 self._match(TokenType.EQ) 1582 self._match(TokenType.ALIAS) 1583 return self.expression(exp_class, this=self._parse_field(), **kwargs) 1584 1585 def _parse_properties(self, before: t.Optional[bool] = None) -> t.Optional[exp.Properties]: 1586 properties = [] 1587 while True: 1588 if before: 1589 prop = self._parse_property_before() 1590 else: 1591 prop = self._parse_property() 1592 1593 if not prop: 1594 break 1595 for p in ensure_list(prop): 1596 properties.append(p) 1597 1598 if properties: 1599 return self.expression(exp.Properties, expressions=properties) 1600 1601 return None 1602 1603 def _parse_fallback(self, no: bool = False) -> exp.FallbackProperty: 1604 return self.expression( 1605 exp.FallbackProperty, no=no, protection=self._match_text_seq("PROTECTION") 1606 ) 1607 1608 def _parse_volatile_property(self) -> exp.VolatileProperty | exp.StabilityProperty: 1609 if self._index >= 2: 1610 pre_volatile_token = self._tokens[self._index - 2] 1611 else: 1612 pre_volatile_token = None 1613 1614 if pre_volatile_token and pre_volatile_token.token_type in self.PRE_VOLATILE_TOKENS: 1615 return exp.VolatileProperty() 1616 1617 return self.expression(exp.StabilityProperty, this=exp.Literal.string("VOLATILE")) 1618 1619 def _parse_system_versioning_property(self) -> exp.WithSystemVersioningProperty: 1620 self._match_pair(TokenType.EQ, TokenType.ON) 1621 1622 prop = self.expression(exp.WithSystemVersioningProperty) 1623 if self._match(TokenType.L_PAREN): 1624 self._match_text_seq("HISTORY_TABLE", "=") 1625 prop.set("this", self._parse_table_parts()) 1626 1627 if self._match(TokenType.COMMA): 1628 self._match_text_seq("DATA_CONSISTENCY_CHECK", "=") 1629 prop.set("expression", self._advance_any() and self._prev.text.upper()) 1630 1631 self._match_r_paren() 1632 1633 return prop 1634 1635 def _parse_with_property( 1636 self, 1637 ) -> t.Optional[exp.Expression] | t.List[exp.Expression]: 1638 if self._match(TokenType.L_PAREN, advance=False): 1639 return self._parse_wrapped_csv(self._parse_property) 1640 1641 if self._match_text_seq("JOURNAL"): 1642 return self._parse_withjournaltable() 1643 1644 if self._match_text_seq("DATA"): 1645 return self._parse_withdata(no=False) 1646 elif self._match_text_seq("NO", "DATA"): 1647 return self._parse_withdata(no=True) 1648 1649 if not self._next: 1650 return None 1651 1652 return self._parse_withisolatedloading() 1653 1654 # https://dev.mysql.com/doc/refman/8.0/en/create-view.html 1655 def _parse_definer(self) -> t.Optional[exp.DefinerProperty]: 1656 self._match(TokenType.EQ) 1657 1658 user = self._parse_id_var() 1659 self._match(TokenType.PARAMETER) 1660 host = self._parse_id_var() or (self._match(TokenType.MOD) and self._prev.text) 1661 1662 if not user or not host: 1663 return None 1664 1665 return exp.DefinerProperty(this=f"{user}@{host}") 1666 1667 def _parse_withjournaltable(self) -> exp.WithJournalTableProperty: 1668 self._match(TokenType.TABLE) 1669 self._match(TokenType.EQ) 1670 return self.expression(exp.WithJournalTableProperty, this=self._parse_table_parts()) 1671 1672 def _parse_log(self, no: bool = False) -> exp.LogProperty: 1673 return self.expression(exp.LogProperty, no=no) 1674 1675 def _parse_journal(self, **kwargs) -> exp.JournalProperty: 1676 return self.expression(exp.JournalProperty, **kwargs) 1677 1678 def _parse_checksum(self) -> exp.ChecksumProperty: 1679 self._match(TokenType.EQ) 1680 1681 on = None 1682 if self._match(TokenType.ON): 1683 on = True 1684 elif self._match_text_seq("OFF"): 1685 on = False 1686 1687 return self.expression(exp.ChecksumProperty, on=on, default=self._match(TokenType.DEFAULT)) 1688 1689 def _parse_cluster(self, wrapped: bool = False) -> exp.Cluster: 1690 return self.expression( 1691 exp.Cluster, 1692 expressions=( 1693 self._parse_wrapped_csv(self._parse_ordered) 1694 if wrapped 1695 else self._parse_csv(self._parse_ordered) 1696 ), 1697 ) 1698 1699 def _parse_clustered_by(self) -> exp.ClusteredByProperty: 1700 self._match_text_seq("BY") 1701 1702 self._match_l_paren() 1703 expressions = self._parse_csv(self._parse_column) 1704 self._match_r_paren() 1705 1706 if self._match_text_seq("SORTED", "BY"): 1707 self._match_l_paren() 1708 sorted_by = self._parse_csv(self._parse_ordered) 1709 self._match_r_paren() 1710 else: 1711 sorted_by = None 1712 1713 self._match(TokenType.INTO) 1714 buckets = self._parse_number() 1715 self._match_text_seq("BUCKETS") 1716 1717 return self.expression( 1718 exp.ClusteredByProperty, 1719 expressions=expressions, 1720 sorted_by=sorted_by, 1721 buckets=buckets, 1722 ) 1723 1724 def _parse_copy_property(self) -> t.Optional[exp.CopyGrantsProperty]: 1725 if not self._match_text_seq("GRANTS"): 1726 self._retreat(self._index - 1) 1727 return None 1728 1729 return self.expression(exp.CopyGrantsProperty) 1730 1731 def _parse_freespace(self) -> exp.FreespaceProperty: 1732 self._match(TokenType.EQ) 1733 return self.expression( 1734 exp.FreespaceProperty, this=self._parse_number(), percent=self._match(TokenType.PERCENT) 1735 ) 1736 1737 def _parse_mergeblockratio( 1738 self, no: bool = False, default: bool = False 1739 ) -> exp.MergeBlockRatioProperty: 1740 if self._match(TokenType.EQ): 1741 return self.expression( 1742 exp.MergeBlockRatioProperty, 1743 this=self._parse_number(), 1744 percent=self._match(TokenType.PERCENT), 1745 ) 1746 1747 return self.expression(exp.MergeBlockRatioProperty, no=no, default=default) 1748 1749 def _parse_datablocksize( 1750 self, 1751 default: t.Optional[bool] = None, 1752 minimum: t.Optional[bool] = None, 1753 maximum: t.Optional[bool] = None, 1754 ) -> exp.DataBlocksizeProperty: 1755 self._match(TokenType.EQ) 1756 size = self._parse_number() 1757 1758 units = None 1759 if self._match_texts(("BYTES", "KBYTES", "KILOBYTES")): 1760 units = self._prev.text 1761 1762 return self.expression( 1763 exp.DataBlocksizeProperty, 1764 size=size, 1765 units=units, 1766 default=default, 1767 minimum=minimum, 1768 maximum=maximum, 1769 ) 1770 1771 def _parse_blockcompression(self) -> exp.BlockCompressionProperty: 1772 self._match(TokenType.EQ) 1773 always = self._match_text_seq("ALWAYS") 1774 manual = self._match_text_seq("MANUAL") 1775 never = self._match_text_seq("NEVER") 1776 default = self._match_text_seq("DEFAULT") 1777 1778 autotemp = None 1779 if self._match_text_seq("AUTOTEMP"): 1780 autotemp = self._parse_schema() 1781 1782 return self.expression( 1783 exp.BlockCompressionProperty, 1784 always=always, 1785 manual=manual, 1786 never=never, 1787 default=default, 1788 autotemp=autotemp, 1789 ) 1790 1791 def _parse_withisolatedloading(self) -> exp.IsolatedLoadingProperty: 1792 no = self._match_text_seq("NO") 1793 concurrent = self._match_text_seq("CONCURRENT") 1794 self._match_text_seq("ISOLATED", "LOADING") 1795 for_all = self._match_text_seq("FOR", "ALL") 1796 for_insert = self._match_text_seq("FOR", "INSERT") 1797 for_none = self._match_text_seq("FOR", "NONE") 1798 return self.expression( 1799 exp.IsolatedLoadingProperty, 1800 no=no, 1801 concurrent=concurrent, 1802 for_all=for_all, 1803 for_insert=for_insert, 1804 for_none=for_none, 1805 ) 1806 1807 def _parse_locking(self) -> exp.LockingProperty: 1808 if self._match(TokenType.TABLE): 1809 kind = "TABLE" 1810 elif self._match(TokenType.VIEW): 1811 kind = "VIEW" 1812 elif self._match(TokenType.ROW): 1813 kind = "ROW" 1814 elif self._match_text_seq("DATABASE"): 1815 kind = "DATABASE" 1816 else: 1817 kind = None 1818 1819 if kind in ("DATABASE", "TABLE", "VIEW"): 1820 this = self._parse_table_parts() 1821 else: 1822 this = None 1823 1824 if self._match(TokenType.FOR): 1825 for_or_in = "FOR" 1826 elif self._match(TokenType.IN): 1827 for_or_in = "IN" 1828 else: 1829 for_or_in = None 1830 1831 if self._match_text_seq("ACCESS"): 1832 lock_type = "ACCESS" 1833 elif self._match_texts(("EXCL", "EXCLUSIVE")): 1834 lock_type = "EXCLUSIVE" 1835 elif self._match_text_seq("SHARE"): 1836 lock_type = "SHARE" 1837 elif self._match_text_seq("READ"): 1838 lock_type = "READ" 1839 elif self._match_text_seq("WRITE"): 1840 lock_type = "WRITE" 1841 elif self._match_text_seq("CHECKSUM"): 1842 lock_type = "CHECKSUM" 1843 else: 1844 lock_type = None 1845 1846 override = self._match_text_seq("OVERRIDE") 1847 1848 return self.expression( 1849 exp.LockingProperty, 1850 this=this, 1851 kind=kind, 1852 for_or_in=for_or_in, 1853 lock_type=lock_type, 1854 override=override, 1855 ) 1856 1857 def _parse_partition_by(self) -> t.List[exp.Expression]: 1858 if self._match(TokenType.PARTITION_BY): 1859 return self._parse_csv(self._parse_conjunction) 1860 return [] 1861 1862 def _parse_partition_bound_spec(self) -> exp.PartitionBoundSpec: 1863 def _parse_partition_bound_expr() -> t.Optional[exp.Expression]: 1864 if self._match_text_seq("MINVALUE"): 1865 return exp.var("MINVALUE") 1866 if self._match_text_seq("MAXVALUE"): 1867 return exp.var("MAXVALUE") 1868 return self._parse_bitwise() 1869 1870 this: t.Optional[exp.Expression | t.List[exp.Expression]] = None 1871 expression = None 1872 from_expressions = None 1873 to_expressions = None 1874 1875 if self._match(TokenType.IN): 1876 this = self._parse_wrapped_csv(self._parse_bitwise) 1877 elif self._match(TokenType.FROM): 1878 from_expressions = self._parse_wrapped_csv(_parse_partition_bound_expr) 1879 self._match_text_seq("TO") 1880 to_expressions = self._parse_wrapped_csv(_parse_partition_bound_expr) 1881 elif self._match_text_seq("WITH", "(", "MODULUS"): 1882 this = self._parse_number() 1883 self._match_text_seq(",", "REMAINDER") 1884 expression = self._parse_number() 1885 self._match_r_paren() 1886 else: 1887 self.raise_error("Failed to parse partition bound spec.") 1888 1889 return self.expression( 1890 exp.PartitionBoundSpec, 1891 this=this, 1892 expression=expression, 1893 from_expressions=from_expressions, 1894 to_expressions=to_expressions, 1895 ) 1896 1897 # https://www.postgresql.org/docs/current/sql-createtable.html 1898 def _parse_partitioned_of(self) -> t.Optional[exp.PartitionedOfProperty]: 1899 if not self._match_text_seq("OF"): 1900 self._retreat(self._index - 1) 1901 return None 1902 1903 this = self._parse_table(schema=True) 1904 1905 if self._match(TokenType.DEFAULT): 1906 expression: exp.Var | exp.PartitionBoundSpec = exp.var("DEFAULT") 1907 elif self._match_text_seq("FOR", "VALUES"): 1908 expression = self._parse_partition_bound_spec() 1909 else: 1910 self.raise_error("Expecting either DEFAULT or FOR VALUES clause.") 1911 1912 return self.expression(exp.PartitionedOfProperty, this=this, expression=expression) 1913 1914 def _parse_partitioned_by(self) -> exp.PartitionedByProperty: 1915 self._match(TokenType.EQ) 1916 return self.expression( 1917 exp.PartitionedByProperty, 1918 this=self._parse_schema() or self._parse_bracket(self._parse_field()), 1919 ) 1920 1921 def _parse_withdata(self, no: bool = False) -> exp.WithDataProperty: 1922 if self._match_text_seq("AND", "STATISTICS"): 1923 statistics = True 1924 elif self._match_text_seq("AND", "NO", "STATISTICS"): 1925 statistics = False 1926 else: 1927 statistics = None 1928 1929 return self.expression(exp.WithDataProperty, no=no, statistics=statistics) 1930 1931 def _parse_contains_property(self) -> t.Optional[exp.SqlReadWriteProperty]: 1932 if self._match_text_seq("SQL"): 1933 return self.expression(exp.SqlReadWriteProperty, this="CONTAINS SQL") 1934 return None 1935 1936 def _parse_modifies_property(self) -> t.Optional[exp.SqlReadWriteProperty]: 1937 if self._match_text_seq("SQL", "DATA"): 1938 return self.expression(exp.SqlReadWriteProperty, this="MODIFIES SQL DATA") 1939 return None 1940 1941 def _parse_no_property(self) -> t.Optional[exp.Expression]: 1942 if self._match_text_seq("PRIMARY", "INDEX"): 1943 return exp.NoPrimaryIndexProperty() 1944 if self._match_text_seq("SQL"): 1945 return self.expression(exp.SqlReadWriteProperty, this="NO SQL") 1946 return None 1947 1948 def _parse_on_property(self) -> t.Optional[exp.Expression]: 1949 if self._match_text_seq("COMMIT", "PRESERVE", "ROWS"): 1950 return exp.OnCommitProperty() 1951 if self._match_text_seq("COMMIT", "DELETE", "ROWS"): 1952 return exp.OnCommitProperty(delete=True) 1953 return self.expression(exp.OnProperty, this=self._parse_schema(self._parse_id_var())) 1954 1955 def _parse_reads_property(self) -> t.Optional[exp.SqlReadWriteProperty]: 1956 if self._match_text_seq("SQL", "DATA"): 1957 return self.expression(exp.SqlReadWriteProperty, this="READS SQL DATA") 1958 return None 1959 1960 def _parse_distkey(self) -> exp.DistKeyProperty: 1961 return self.expression(exp.DistKeyProperty, this=self._parse_wrapped(self._parse_id_var)) 1962 1963 def _parse_create_like(self) -> t.Optional[exp.LikeProperty]: 1964 table = self._parse_table(schema=True) 1965 1966 options = [] 1967 while self._match_texts(("INCLUDING", "EXCLUDING")): 1968 this = self._prev.text.upper() 1969 1970 id_var = self._parse_id_var() 1971 if not id_var: 1972 return None 1973 1974 options.append( 1975 self.expression(exp.Property, this=this, value=exp.var(id_var.this.upper())) 1976 ) 1977 1978 return self.expression(exp.LikeProperty, this=table, expressions=options) 1979 1980 def _parse_sortkey(self, compound: bool = False) -> exp.SortKeyProperty: 1981 return self.expression( 1982 exp.SortKeyProperty, this=self._parse_wrapped_id_vars(), compound=compound 1983 ) 1984 1985 def _parse_character_set(self, default: bool = False) -> exp.CharacterSetProperty: 1986 self._match(TokenType.EQ) 1987 return self.expression( 1988 exp.CharacterSetProperty, this=self._parse_var_or_string(), default=default 1989 ) 1990 1991 def _parse_remote_with_connection(self) -> exp.RemoteWithConnectionModelProperty: 1992 self._match_text_seq("WITH", "CONNECTION") 1993 return self.expression( 1994 exp.RemoteWithConnectionModelProperty, this=self._parse_table_parts() 1995 ) 1996 1997 def _parse_returns(self) -> exp.ReturnsProperty: 1998 value: t.Optional[exp.Expression] 1999 is_table = self._match(TokenType.TABLE) 2000 2001 if is_table: 2002 if self._match(TokenType.LT): 2003 value = self.expression( 2004 exp.Schema, 2005 this="TABLE", 2006 expressions=self._parse_csv(self._parse_struct_types), 2007 ) 2008 if not self._match(TokenType.GT): 2009 self.raise_error("Expecting >") 2010 else: 2011 value = self._parse_schema(exp.var("TABLE")) 2012 else: 2013 value = self._parse_types() 2014 2015 return self.expression(exp.ReturnsProperty, this=value, is_table=is_table) 2016 2017 def _parse_describe(self) -> exp.Describe: 2018 kind = self._match_set(self.CREATABLES) and self._prev.text 2019 extended = self._match_text_seq("EXTENDED") 2020 this = self._parse_table(schema=True) 2021 properties = self._parse_properties() 2022 expressions = properties.expressions if properties else None 2023 return self.expression( 2024 exp.Describe, this=this, extended=extended, kind=kind, expressions=expressions 2025 ) 2026 2027 def _parse_insert(self) -> exp.Insert: 2028 comments = ensure_list(self._prev_comments) 2029 overwrite = self._match(TokenType.OVERWRITE) 2030 ignore = self._match(TokenType.IGNORE) 2031 local = self._match_text_seq("LOCAL") 2032 alternative = None 2033 2034 if self._match_text_seq("DIRECTORY"): 2035 this: t.Optional[exp.Expression] = self.expression( 2036 exp.Directory, 2037 this=self._parse_var_or_string(), 2038 local=local, 2039 row_format=self._parse_row_format(match_row=True), 2040 ) 2041 else: 2042 if self._match(TokenType.OR): 2043 alternative = self._match_texts(self.INSERT_ALTERNATIVES) and self._prev.text 2044 2045 self._match(TokenType.INTO) 2046 comments += ensure_list(self._prev_comments) 2047 self._match(TokenType.TABLE) 2048 this = self._parse_table(schema=True) 2049 2050 returning = self._parse_returning() 2051 2052 return self.expression( 2053 exp.Insert, 2054 comments=comments, 2055 this=this, 2056 by_name=self._match_text_seq("BY", "NAME"), 2057 exists=self._parse_exists(), 2058 partition=self._parse_partition(), 2059 where=self._match_pair(TokenType.REPLACE, TokenType.WHERE) 2060 and self._parse_conjunction(), 2061 expression=self._parse_ddl_select(), 2062 conflict=self._parse_on_conflict(), 2063 returning=returning or self._parse_returning(), 2064 overwrite=overwrite, 2065 alternative=alternative, 2066 ignore=ignore, 2067 ) 2068 2069 def _parse_kill(self) -> exp.Kill: 2070 kind = exp.var(self._prev.text) if self._match_texts(("CONNECTION", "QUERY")) else None 2071 2072 return self.expression( 2073 exp.Kill, 2074 this=self._parse_primary(), 2075 kind=kind, 2076 ) 2077 2078 def _parse_on_conflict(self) -> t.Optional[exp.OnConflict]: 2079 conflict = self._match_text_seq("ON", "CONFLICT") 2080 duplicate = self._match_text_seq("ON", "DUPLICATE", "KEY") 2081 2082 if not conflict and not duplicate: 2083 return None 2084 2085 nothing = None 2086 expressions = None 2087 key = None 2088 constraint = None 2089 2090 if conflict: 2091 if self._match_text_seq("ON", "CONSTRAINT"): 2092 constraint = self._parse_id_var() 2093 else: 2094 key = self._parse_csv(self._parse_value) 2095 2096 self._match_text_seq("DO") 2097 if self._match_text_seq("NOTHING"): 2098 nothing = True 2099 else: 2100 self._match(TokenType.UPDATE) 2101 self._match(TokenType.SET) 2102 expressions = self._parse_csv(self._parse_equality) 2103 2104 return self.expression( 2105 exp.OnConflict, 2106 duplicate=duplicate, 2107 expressions=expressions, 2108 nothing=nothing, 2109 key=key, 2110 constraint=constraint, 2111 ) 2112 2113 def _parse_returning(self) -> t.Optional[exp.Returning]: 2114 if not self._match(TokenType.RETURNING): 2115 return None 2116 return self.expression( 2117 exp.Returning, 2118 expressions=self._parse_csv(self._parse_expression), 2119 into=self._match(TokenType.INTO) and self._parse_table_part(), 2120 ) 2121 2122 def _parse_row(self) -> t.Optional[exp.RowFormatSerdeProperty | exp.RowFormatDelimitedProperty]: 2123 if not self._match(TokenType.FORMAT): 2124 return None 2125 return self._parse_row_format() 2126 2127 def _parse_row_format( 2128 self, match_row: bool = False 2129 ) -> t.Optional[exp.RowFormatSerdeProperty | exp.RowFormatDelimitedProperty]: 2130 if match_row and not self._match_pair(TokenType.ROW, TokenType.FORMAT): 2131 return None 2132 2133 if self._match_text_seq("SERDE"): 2134 this = self._parse_string() 2135 2136 serde_properties = None 2137 if self._match(TokenType.SERDE_PROPERTIES): 2138 serde_properties = self.expression( 2139 exp.SerdeProperties, expressions=self._parse_wrapped_csv(self._parse_property) 2140 ) 2141 2142 return self.expression( 2143 exp.RowFormatSerdeProperty, this=this, serde_properties=serde_properties 2144 ) 2145 2146 self._match_text_seq("DELIMITED") 2147 2148 kwargs = {} 2149 2150 if self._match_text_seq("FIELDS", "TERMINATED", "BY"): 2151 kwargs["fields"] = self._parse_string() 2152 if self._match_text_seq("ESCAPED", "BY"): 2153 kwargs["escaped"] = self._parse_string() 2154 if self._match_text_seq("COLLECTION", "ITEMS", "TERMINATED", "BY"): 2155 kwargs["collection_items"] = self._parse_string() 2156 if self._match_text_seq("MAP", "KEYS", "TERMINATED", "BY"): 2157 kwargs["map_keys"] = self._parse_string() 2158 if self._match_text_seq("LINES", "TERMINATED", "BY"): 2159 kwargs["lines"] = self._parse_string() 2160 if self._match_text_seq("NULL", "DEFINED", "AS"): 2161 kwargs["null"] = self._parse_string() 2162 2163 return self.expression(exp.RowFormatDelimitedProperty, **kwargs) # type: ignore 2164 2165 def _parse_load(self) -> exp.LoadData | exp.Command: 2166 if self._match_text_seq("DATA"): 2167 local = self._match_text_seq("LOCAL") 2168 self._match_text_seq("INPATH") 2169 inpath = self._parse_string() 2170 overwrite = self._match(TokenType.OVERWRITE) 2171 self._match_pair(TokenType.INTO, TokenType.TABLE) 2172 2173 return self.expression( 2174 exp.LoadData, 2175 this=self._parse_table(schema=True), 2176 local=local, 2177 overwrite=overwrite, 2178 inpath=inpath, 2179 partition=self._parse_partition(), 2180 input_format=self._match_text_seq("INPUTFORMAT") and self._parse_string(), 2181 serde=self._match_text_seq("SERDE") and self._parse_string(), 2182 ) 2183 return self._parse_as_command(self._prev) 2184 2185 def _parse_delete(self) -> exp.Delete: 2186 # This handles MySQL's "Multiple-Table Syntax" 2187 # https://dev.mysql.com/doc/refman/8.0/en/delete.html 2188 tables = None 2189 comments = self._prev_comments 2190 if not self._match(TokenType.FROM, advance=False): 2191 tables = self._parse_csv(self._parse_table) or None 2192 2193 returning = self._parse_returning() 2194 2195 return self.expression( 2196 exp.Delete, 2197 comments=comments, 2198 tables=tables, 2199 this=self._match(TokenType.FROM) and self._parse_table(joins=True), 2200 using=self._match(TokenType.USING) and self._parse_table(joins=True), 2201 where=self._parse_where(), 2202 returning=returning or self._parse_returning(), 2203 limit=self._parse_limit(), 2204 ) 2205 2206 def _parse_update(self) -> exp.Update: 2207 comments = self._prev_comments 2208 this = self._parse_table(joins=True, alias_tokens=self.UPDATE_ALIAS_TOKENS) 2209 expressions = self._match(TokenType.SET) and self._parse_csv(self._parse_equality) 2210 returning = self._parse_returning() 2211 return self.expression( 2212 exp.Update, 2213 comments=comments, 2214 **{ # type: ignore 2215 "this": this, 2216 "expressions": expressions, 2217 "from": self._parse_from(joins=True), 2218 "where": self._parse_where(), 2219 "returning": returning or self._parse_returning(), 2220 "order": self._parse_order(), 2221 "limit": self._parse_limit(), 2222 }, 2223 ) 2224 2225 def _parse_uncache(self) -> exp.Uncache: 2226 if not self._match(TokenType.TABLE): 2227 self.raise_error("Expecting TABLE after UNCACHE") 2228 2229 return self.expression( 2230 exp.Uncache, exists=self._parse_exists(), this=self._parse_table(schema=True) 2231 ) 2232 2233 def _parse_cache(self) -> exp.Cache: 2234 lazy = self._match_text_seq("LAZY") 2235 self._match(TokenType.TABLE) 2236 table = self._parse_table(schema=True) 2237 2238 options = [] 2239 if self._match_text_seq("OPTIONS"): 2240 self._match_l_paren() 2241 k = self._parse_string() 2242 self._match(TokenType.EQ) 2243 v = self._parse_string() 2244 options = [k, v] 2245 self._match_r_paren() 2246 2247 self._match(TokenType.ALIAS) 2248 return self.expression( 2249 exp.Cache, 2250 this=table, 2251 lazy=lazy, 2252 options=options, 2253 expression=self._parse_select(nested=True), 2254 ) 2255 2256 def _parse_partition(self) -> t.Optional[exp.Partition]: 2257 if not self._match(TokenType.PARTITION): 2258 return None 2259 2260 return self.expression( 2261 exp.Partition, expressions=self._parse_wrapped_csv(self._parse_conjunction) 2262 ) 2263 2264 def _parse_value(self) -> exp.Tuple: 2265 if self._match(TokenType.L_PAREN): 2266 expressions = self._parse_csv(self._parse_expression) 2267 self._match_r_paren() 2268 return self.expression(exp.Tuple, expressions=expressions) 2269 2270 # In presto we can have VALUES 1, 2 which results in 1 column & 2 rows. 2271 # https://prestodb.io/docs/current/sql/values.html 2272 return self.expression(exp.Tuple, expressions=[self._parse_expression()]) 2273 2274 def _parse_projections(self) -> t.List[exp.Expression]: 2275 return self._parse_expressions() 2276 2277 def _parse_select( 2278 self, 2279 nested: bool = False, 2280 table: bool = False, 2281 parse_subquery_alias: bool = True, 2282 parse_set_operation: bool = True, 2283 ) -> t.Optional[exp.Expression]: 2284 cte = self._parse_with() 2285 2286 if cte: 2287 this = self._parse_statement() 2288 2289 if not this: 2290 self.raise_error("Failed to parse any statement following CTE") 2291 return cte 2292 2293 if "with" in this.arg_types: 2294 this.set("with", cte) 2295 else: 2296 self.raise_error(f"{this.key} does not support CTE") 2297 this = cte 2298 2299 return this 2300 2301 # duckdb supports leading with FROM x 2302 from_ = self._parse_from() if self._match(TokenType.FROM, advance=False) else None 2303 2304 if self._match(TokenType.SELECT): 2305 comments = self._prev_comments 2306 2307 hint = self._parse_hint() 2308 all_ = self._match(TokenType.ALL) 2309 distinct = self._match_set(self.DISTINCT_TOKENS) 2310 2311 kind = ( 2312 self._match(TokenType.ALIAS) 2313 and self._match_texts(("STRUCT", "VALUE")) 2314 and self._prev.text.upper() 2315 ) 2316 2317 if distinct: 2318 distinct = self.expression( 2319 exp.Distinct, 2320 on=self._parse_value() if self._match(TokenType.ON) else None, 2321 ) 2322 2323 if all_ and distinct: 2324 self.raise_error("Cannot specify both ALL and DISTINCT after SELECT") 2325 2326 limit = self._parse_limit(top=True) 2327 projections = self._parse_projections() 2328 2329 this = self.expression( 2330 exp.Select, 2331 kind=kind, 2332 hint=hint, 2333 distinct=distinct, 2334 expressions=projections, 2335 limit=limit, 2336 ) 2337 this.comments = comments 2338 2339 into = self._parse_into() 2340 if into: 2341 this.set("into", into) 2342 2343 if not from_: 2344 from_ = self._parse_from() 2345 2346 if from_: 2347 this.set("from", from_) 2348 2349 this = self._parse_query_modifiers(this) 2350 elif (table or nested) and self._match(TokenType.L_PAREN): 2351 if self._match(TokenType.PIVOT): 2352 this = self._parse_simplified_pivot() 2353 elif self._match(TokenType.FROM): 2354 this = exp.select("*").from_( 2355 t.cast(exp.From, self._parse_from(skip_from_token=True)) 2356 ) 2357 else: 2358 this = ( 2359 self._parse_table() 2360 if table 2361 else self._parse_select(nested=True, parse_set_operation=False) 2362 ) 2363 this = self._parse_query_modifiers(self._parse_set_operations(this)) 2364 2365 self._match_r_paren() 2366 2367 # We return early here so that the UNION isn't attached to the subquery by the 2368 # following call to _parse_set_operations, but instead becomes the parent node 2369 return self._parse_subquery(this, parse_alias=parse_subquery_alias) 2370 elif self._match(TokenType.VALUES): 2371 this = self.expression( 2372 exp.Values, 2373 expressions=self._parse_csv(self._parse_value), 2374 alias=self._parse_table_alias(), 2375 ) 2376 elif from_: 2377 this = exp.select("*").from_(from_.this, copy=False) 2378 else: 2379 this = None 2380 2381 if parse_set_operation: 2382 return self._parse_set_operations(this) 2383 return this 2384 2385 def _parse_with(self, skip_with_token: bool = False) -> t.Optional[exp.With]: 2386 if not skip_with_token and not self._match(TokenType.WITH): 2387 return None 2388 2389 comments = self._prev_comments 2390 recursive = self._match(TokenType.RECURSIVE) 2391 2392 expressions = [] 2393 while True: 2394 expressions.append(self._parse_cte()) 2395 2396 if not self._match(TokenType.COMMA) and not self._match(TokenType.WITH): 2397 break 2398 else: 2399 self._match(TokenType.WITH) 2400 2401 return self.expression( 2402 exp.With, comments=comments, expressions=expressions, recursive=recursive 2403 ) 2404 2405 def _parse_cte(self) -> exp.CTE: 2406 alias = self._parse_table_alias(self.ID_VAR_TOKENS) 2407 if not alias or not alias.this: 2408 self.raise_error("Expected CTE to have alias") 2409 2410 self._match(TokenType.ALIAS) 2411 return self.expression( 2412 exp.CTE, this=self._parse_wrapped(self._parse_statement), alias=alias 2413 ) 2414 2415 def _parse_table_alias( 2416 self, alias_tokens: t.Optional[t.Collection[TokenType]] = None 2417 ) -> t.Optional[exp.TableAlias]: 2418 any_token = self._match(TokenType.ALIAS) 2419 alias = ( 2420 self._parse_id_var(any_token=any_token, tokens=alias_tokens or self.TABLE_ALIAS_TOKENS) 2421 or self._parse_string_as_identifier() 2422 ) 2423 2424 index = self._index 2425 if self._match(TokenType.L_PAREN): 2426 columns = self._parse_csv(self._parse_function_parameter) 2427 self._match_r_paren() if columns else self._retreat(index) 2428 else: 2429 columns = None 2430 2431 if not alias and not columns: 2432 return None 2433 2434 return self.expression(exp.TableAlias, this=alias, columns=columns) 2435 2436 def _parse_subquery( 2437 self, this: t.Optional[exp.Expression], parse_alias: bool = True 2438 ) -> t.Optional[exp.Subquery]: 2439 if not this: 2440 return None 2441 2442 return self.expression( 2443 exp.Subquery, 2444 this=this, 2445 pivots=self._parse_pivots(), 2446 alias=self._parse_table_alias() if parse_alias else None, 2447 ) 2448 2449 def _parse_query_modifiers( 2450 self, this: t.Optional[exp.Expression] 2451 ) -> t.Optional[exp.Expression]: 2452 if isinstance(this, self.MODIFIABLES): 2453 for join in iter(self._parse_join, None): 2454 this.append("joins", join) 2455 for lateral in iter(self._parse_lateral, None): 2456 this.append("laterals", lateral) 2457 2458 while True: 2459 if self._match_set(self.QUERY_MODIFIER_PARSERS, advance=False): 2460 parser = self.QUERY_MODIFIER_PARSERS[self._curr.token_type] 2461 key, expression = parser(self) 2462 2463 if expression: 2464 this.set(key, expression) 2465 if key == "limit": 2466 offset = expression.args.pop("offset", None) 2467 2468 if offset: 2469 offset = exp.Offset(expression=offset) 2470 this.set("offset", offset) 2471 2472 limit_by_expressions = expression.expressions 2473 expression.set("expressions", None) 2474 offset.set("expressions", limit_by_expressions) 2475 continue 2476 break 2477 return this 2478 2479 def _parse_hint(self) -> t.Optional[exp.Hint]: 2480 if self._match(TokenType.HINT): 2481 hints = [] 2482 for hint in iter(lambda: self._parse_csv(self._parse_function), []): 2483 hints.extend(hint) 2484 2485 if not self._match_pair(TokenType.STAR, TokenType.SLASH): 2486 self.raise_error("Expected */ after HINT") 2487 2488 return self.expression(exp.Hint, expressions=hints) 2489 2490 return None 2491 2492 def _parse_into(self) -> t.Optional[exp.Into]: 2493 if not self._match(TokenType.INTO): 2494 return None 2495 2496 temp = self._match(TokenType.TEMPORARY) 2497 unlogged = self._match_text_seq("UNLOGGED") 2498 self._match(TokenType.TABLE) 2499 2500 return self.expression( 2501 exp.Into, this=self._parse_table(schema=True), temporary=temp, unlogged=unlogged 2502 ) 2503 2504 def _parse_from( 2505 self, joins: bool = False, skip_from_token: bool = False 2506 ) -> t.Optional[exp.From]: 2507 if not skip_from_token and not self._match(TokenType.FROM): 2508 return None 2509 2510 return self.expression( 2511 exp.From, comments=self._prev_comments, this=self._parse_table(joins=joins) 2512 ) 2513 2514 def _parse_match_recognize(self) -> t.Optional[exp.MatchRecognize]: 2515 if not self._match(TokenType.MATCH_RECOGNIZE): 2516 return None 2517 2518 self._match_l_paren() 2519 2520 partition = self._parse_partition_by() 2521 order = self._parse_order() 2522 measures = self._parse_expressions() if self._match_text_seq("MEASURES") else None 2523 2524 if self._match_text_seq("ONE", "ROW", "PER", "MATCH"): 2525 rows = exp.var("ONE ROW PER MATCH") 2526 elif self._match_text_seq("ALL", "ROWS", "PER", "MATCH"): 2527 text = "ALL ROWS PER MATCH" 2528 if self._match_text_seq("SHOW", "EMPTY", "MATCHES"): 2529 text += " SHOW EMPTY MATCHES" 2530 elif self._match_text_seq("OMIT", "EMPTY", "MATCHES"): 2531 text += " OMIT EMPTY MATCHES" 2532 elif self._match_text_seq("WITH", "UNMATCHED", "ROWS"): 2533 text += " WITH UNMATCHED ROWS" 2534 rows = exp.var(text) 2535 else: 2536 rows = None 2537 2538 if self._match_text_seq("AFTER", "MATCH", "SKIP"): 2539 text = "AFTER MATCH SKIP" 2540 if self._match_text_seq("PAST", "LAST", "ROW"): 2541 text += " PAST LAST ROW" 2542 elif self._match_text_seq("TO", "NEXT", "ROW"): 2543 text += " TO NEXT ROW" 2544 elif self._match_text_seq("TO", "FIRST"): 2545 text += f" TO FIRST {self._advance_any().text}" # type: ignore 2546 elif self._match_text_seq("TO", "LAST"): 2547 text += f" TO LAST {self._advance_any().text}" # type: ignore 2548 after = exp.var(text) 2549 else: 2550 after = None 2551 2552 if self._match_text_seq("PATTERN"): 2553 self._match_l_paren() 2554 2555 if not self._curr: 2556 self.raise_error("Expecting )", self._curr) 2557 2558 paren = 1 2559 start = self._curr 2560 2561 while self._curr and paren > 0: 2562 if self._curr.token_type == TokenType.L_PAREN: 2563 paren += 1 2564 if self._curr.token_type == TokenType.R_PAREN: 2565 paren -= 1 2566 2567 end = self._prev 2568 self._advance() 2569 2570 if paren > 0: 2571 self.raise_error("Expecting )", self._curr) 2572 2573 pattern = exp.var(self._find_sql(start, end)) 2574 else: 2575 pattern = None 2576 2577 define = ( 2578 self._parse_csv(self._parse_name_as_expression) 2579 if self._match_text_seq("DEFINE") 2580 else None 2581 ) 2582 2583 self._match_r_paren() 2584 2585 return self.expression( 2586 exp.MatchRecognize, 2587 partition_by=partition, 2588 order=order, 2589 measures=measures, 2590 rows=rows, 2591 after=after, 2592 pattern=pattern, 2593 define=define, 2594 alias=self._parse_table_alias(), 2595 ) 2596 2597 def _parse_lateral(self) -> t.Optional[exp.Lateral]: 2598 cross_apply = self._match_pair(TokenType.CROSS, TokenType.APPLY) 2599 if not cross_apply and self._match_pair(TokenType.OUTER, TokenType.APPLY): 2600 cross_apply = False 2601 2602 if cross_apply is not None: 2603 this = self._parse_select(table=True) 2604 view = None 2605 outer = None 2606 elif self._match(TokenType.LATERAL): 2607 this = self._parse_select(table=True) 2608 view = self._match(TokenType.VIEW) 2609 outer = self._match(TokenType.OUTER) 2610 else: 2611 return None 2612 2613 if not this: 2614 this = ( 2615 self._parse_unnest() 2616 or self._parse_function() 2617 or self._parse_id_var(any_token=False) 2618 ) 2619 2620 while self._match(TokenType.DOT): 2621 this = exp.Dot( 2622 this=this, 2623 expression=self._parse_function() or self._parse_id_var(any_token=False), 2624 ) 2625 2626 if view: 2627 table = self._parse_id_var(any_token=False) 2628 columns = self._parse_csv(self._parse_id_var) if self._match(TokenType.ALIAS) else [] 2629 table_alias: t.Optional[exp.TableAlias] = self.expression( 2630 exp.TableAlias, this=table, columns=columns 2631 ) 2632 elif isinstance(this, (exp.Subquery, exp.Unnest)) and this.alias: 2633 # We move the alias from the lateral's child node to the lateral itself 2634 table_alias = this.args["alias"].pop() 2635 else: 2636 table_alias = self._parse_table_alias() 2637 2638 return self.expression( 2639 exp.Lateral, 2640 this=this, 2641 view=view, 2642 outer=outer, 2643 alias=table_alias, 2644 cross_apply=cross_apply, 2645 ) 2646 2647 def _parse_join_parts( 2648 self, 2649 ) -> t.Tuple[t.Optional[Token], t.Optional[Token], t.Optional[Token]]: 2650 return ( 2651 self._match_set(self.JOIN_METHODS) and self._prev, 2652 self._match_set(self.JOIN_SIDES) and self._prev, 2653 self._match_set(self.JOIN_KINDS) and self._prev, 2654 ) 2655 2656 def _parse_join( 2657 self, skip_join_token: bool = False, parse_bracket: bool = False 2658 ) -> t.Optional[exp.Join]: 2659 if self._match(TokenType.COMMA): 2660 return self.expression(exp.Join, this=self._parse_table()) 2661 2662 index = self._index 2663 method, side, kind = self._parse_join_parts() 2664 hint = self._prev.text if self._match_texts(self.JOIN_HINTS) else None 2665 join = self._match(TokenType.JOIN) 2666 2667 if not skip_join_token and not join: 2668 self._retreat(index) 2669 kind = None 2670 method = None 2671 side = None 2672 2673 outer_apply = self._match_pair(TokenType.OUTER, TokenType.APPLY, False) 2674 cross_apply = self._match_pair(TokenType.CROSS, TokenType.APPLY, False) 2675 2676 if not skip_join_token and not join and not outer_apply and not cross_apply: 2677 return None 2678 2679 kwargs: t.Dict[str, t.Any] = {"this": self._parse_table(parse_bracket=parse_bracket)} 2680 2681 if method: 2682 kwargs["method"] = method.text 2683 if side: 2684 kwargs["side"] = side.text 2685 if kind: 2686 kwargs["kind"] = kind.text 2687 if hint: 2688 kwargs["hint"] = hint 2689 2690 if self._match(TokenType.ON): 2691 kwargs["on"] = self._parse_conjunction() 2692 elif self._match(TokenType.USING): 2693 kwargs["using"] = self._parse_wrapped_id_vars() 2694 elif not (kind and kind.token_type == TokenType.CROSS): 2695 index = self._index 2696 join = self._parse_join() 2697 2698 if join and self._match(TokenType.ON): 2699 kwargs["on"] = self._parse_conjunction() 2700 elif join and self._match(TokenType.USING): 2701 kwargs["using"] = self._parse_wrapped_id_vars() 2702 else: 2703 join = None 2704 self._retreat(index) 2705 2706 kwargs["this"].set("joins", [join] if join else None) 2707 2708 comments = [c for token in (method, side, kind) if token for c in token.comments] 2709 return self.expression(exp.Join, comments=comments, **kwargs) 2710 2711 def _parse_opclass(self) -> t.Optional[exp.Expression]: 2712 this = self._parse_conjunction() 2713 if self._match_texts(self.OPCLASS_FOLLOW_KEYWORDS, advance=False): 2714 return this 2715 2716 if not self._match_set(self.OPTYPE_FOLLOW_TOKENS, advance=False): 2717 return self.expression(exp.Opclass, this=this, expression=self._parse_table_parts()) 2718 2719 return this 2720 2721 def _parse_index( 2722 self, 2723 index: t.Optional[exp.Expression] = None, 2724 ) -> t.Optional[exp.Index]: 2725 if index: 2726 unique = None 2727 primary = None 2728 amp = None 2729 2730 self._match(TokenType.ON) 2731 self._match(TokenType.TABLE) # hive 2732 table = self._parse_table_parts(schema=True) 2733 else: 2734 unique = self._match(TokenType.UNIQUE) 2735 primary = self._match_text_seq("PRIMARY") 2736 amp = self._match_text_seq("AMP") 2737 2738 if not self._match(TokenType.INDEX): 2739 return None 2740 2741 index = self._parse_id_var() 2742 table = None 2743 2744 using = self._parse_var(any_token=True) if self._match(TokenType.USING) else None 2745 2746 if self._match(TokenType.L_PAREN, advance=False): 2747 columns = self._parse_wrapped_csv(lambda: self._parse_ordered(self._parse_opclass)) 2748 else: 2749 columns = None 2750 2751 include = self._parse_wrapped_id_vars() if self._match_text_seq("INCLUDE") else None 2752 2753 return self.expression( 2754 exp.Index, 2755 this=index, 2756 table=table, 2757 using=using, 2758 columns=columns, 2759 unique=unique, 2760 primary=primary, 2761 amp=amp, 2762 include=include, 2763 partition_by=self._parse_partition_by(), 2764 where=self._parse_where(), 2765 ) 2766 2767 def _parse_table_hints(self) -> t.Optional[t.List[exp.Expression]]: 2768 hints: t.List[exp.Expression] = [] 2769 if self._match_pair(TokenType.WITH, TokenType.L_PAREN): 2770 # https://learn.microsoft.com/en-us/sql/t-sql/queries/hints-transact-sql-table?view=sql-server-ver16 2771 hints.append( 2772 self.expression( 2773 exp.WithTableHint, 2774 expressions=self._parse_csv( 2775 lambda: self._parse_function() or self._parse_var(any_token=True) 2776 ), 2777 ) 2778 ) 2779 self._match_r_paren() 2780 else: 2781 # https://dev.mysql.com/doc/refman/8.0/en/index-hints.html 2782 while self._match_set(self.TABLE_INDEX_HINT_TOKENS): 2783 hint = exp.IndexTableHint(this=self._prev.text.upper()) 2784 2785 self._match_texts(("INDEX", "KEY")) 2786 if self._match(TokenType.FOR): 2787 hint.set("target", self._advance_any() and self._prev.text.upper()) 2788 2789 hint.set("expressions", self._parse_wrapped_id_vars()) 2790 hints.append(hint) 2791 2792 return hints or None 2793 2794 def _parse_table_part(self, schema: bool = False) -> t.Optional[exp.Expression]: 2795 return ( 2796 (not schema and self._parse_function(optional_parens=False)) 2797 or self._parse_id_var(any_token=False) 2798 or self._parse_string_as_identifier() 2799 or self._parse_placeholder() 2800 ) 2801 2802 def _parse_table_parts(self, schema: bool = False, is_db_reference: bool = False) -> exp.Table: 2803 catalog = None 2804 db = None 2805 table: t.Optional[exp.Expression | str] = self._parse_table_part(schema=schema) 2806 2807 while self._match(TokenType.DOT): 2808 if catalog: 2809 # This allows nesting the table in arbitrarily many dot expressions if needed 2810 table = self.expression( 2811 exp.Dot, this=table, expression=self._parse_table_part(schema=schema) 2812 ) 2813 else: 2814 catalog = db 2815 db = table 2816 table = self._parse_table_part(schema=schema) or "" 2817 2818 if is_db_reference: 2819 catalog = db 2820 db = table 2821 table = None 2822 2823 if not table and not is_db_reference: 2824 self.raise_error(f"Expected table name but got {self._curr}") 2825 if not db and is_db_reference: 2826 self.raise_error(f"Expected database name but got {self._curr}") 2827 2828 return self.expression( 2829 exp.Table, this=table, db=db, catalog=catalog, pivots=self._parse_pivots() 2830 ) 2831 2832 def _parse_table( 2833 self, 2834 schema: bool = False, 2835 joins: bool = False, 2836 alias_tokens: t.Optional[t.Collection[TokenType]] = None, 2837 parse_bracket: bool = False, 2838 is_db_reference: bool = False, 2839 ) -> t.Optional[exp.Expression]: 2840 lateral = self._parse_lateral() 2841 if lateral: 2842 return lateral 2843 2844 unnest = self._parse_unnest() 2845 if unnest: 2846 return unnest 2847 2848 values = self._parse_derived_table_values() 2849 if values: 2850 return values 2851 2852 subquery = self._parse_select(table=True) 2853 if subquery: 2854 if not subquery.args.get("pivots"): 2855 subquery.set("pivots", self._parse_pivots()) 2856 return subquery 2857 2858 bracket = parse_bracket and self._parse_bracket(None) 2859 bracket = self.expression(exp.Table, this=bracket) if bracket else None 2860 this = t.cast( 2861 exp.Expression, 2862 bracket 2863 or self._parse_bracket( 2864 self._parse_table_parts(schema=schema, is_db_reference=is_db_reference) 2865 ), 2866 ) 2867 2868 if schema: 2869 return self._parse_schema(this=this) 2870 2871 version = self._parse_version() 2872 2873 if version: 2874 this.set("version", version) 2875 2876 if self.dialect.ALIAS_POST_TABLESAMPLE: 2877 table_sample = self._parse_table_sample() 2878 2879 alias = self._parse_table_alias(alias_tokens=alias_tokens or self.TABLE_ALIAS_TOKENS) 2880 if alias: 2881 this.set("alias", alias) 2882 2883 if isinstance(this, exp.Table) and self._match_text_seq("AT"): 2884 return self.expression( 2885 exp.AtIndex, this=this.to_column(copy=False), expression=self._parse_id_var() 2886 ) 2887 2888 this.set("hints", self._parse_table_hints()) 2889 2890 if not this.args.get("pivots"): 2891 this.set("pivots", self._parse_pivots()) 2892 2893 if not self.dialect.ALIAS_POST_TABLESAMPLE: 2894 table_sample = self._parse_table_sample() 2895 2896 if table_sample: 2897 table_sample.set("this", this) 2898 this = table_sample 2899 2900 if joins: 2901 for join in iter(self._parse_join, None): 2902 this.append("joins", join) 2903 2904 if self._match_pair(TokenType.WITH, TokenType.ORDINALITY): 2905 this.set("ordinality", True) 2906 this.set("alias", self._parse_table_alias()) 2907 2908 return this 2909 2910 def _parse_version(self) -> t.Optional[exp.Version]: 2911 if self._match(TokenType.TIMESTAMP_SNAPSHOT): 2912 this = "TIMESTAMP" 2913 elif self._match(TokenType.VERSION_SNAPSHOT): 2914 this = "VERSION" 2915 else: 2916 return None 2917 2918 if self._match_set((TokenType.FROM, TokenType.BETWEEN)): 2919 kind = self._prev.text.upper() 2920 start = self._parse_bitwise() 2921 self._match_texts(("TO", "AND")) 2922 end = self._parse_bitwise() 2923 expression: t.Optional[exp.Expression] = self.expression( 2924 exp.Tuple, expressions=[start, end] 2925 ) 2926 elif self._match_text_seq("CONTAINED", "IN"): 2927 kind = "CONTAINED IN" 2928 expression = self.expression( 2929 exp.Tuple, expressions=self._parse_wrapped_csv(self._parse_bitwise) 2930 ) 2931 elif self._match(TokenType.ALL): 2932 kind = "ALL" 2933 expression = None 2934 else: 2935 self._match_text_seq("AS", "OF") 2936 kind = "AS OF" 2937 expression = self._parse_type() 2938 2939 return self.expression(exp.Version, this=this, expression=expression, kind=kind) 2940 2941 def _parse_unnest(self, with_alias: bool = True) -> t.Optional[exp.Unnest]: 2942 if not self._match(TokenType.UNNEST): 2943 return None 2944 2945 expressions = self._parse_wrapped_csv(self._parse_equality) 2946 offset = self._match_pair(TokenType.WITH, TokenType.ORDINALITY) 2947 2948 alias = self._parse_table_alias() if with_alias else None 2949 2950 if alias: 2951 if self.dialect.UNNEST_COLUMN_ONLY: 2952 if alias.args.get("columns"): 2953 self.raise_error("Unexpected extra column alias in unnest.") 2954 2955 alias.set("columns", [alias.this]) 2956 alias.set("this", None) 2957 2958 columns = alias.args.get("columns") or [] 2959 if offset and len(expressions) < len(columns): 2960 offset = columns.pop() 2961 2962 if not offset and self._match_pair(TokenType.WITH, TokenType.OFFSET): 2963 self._match(TokenType.ALIAS) 2964 offset = self._parse_id_var( 2965 any_token=False, tokens=self.UNNEST_OFFSET_ALIAS_TOKENS 2966 ) or exp.to_identifier("offset") 2967 2968 return self.expression(exp.Unnest, expressions=expressions, alias=alias, offset=offset) 2969 2970 def _parse_derived_table_values(self) -> t.Optional[exp.Values]: 2971 is_derived = self._match_pair(TokenType.L_PAREN, TokenType.VALUES) 2972 if not is_derived and not self._match(TokenType.VALUES): 2973 return None 2974 2975 expressions = self._parse_csv(self._parse_value) 2976 alias = self._parse_table_alias() 2977 2978 if is_derived: 2979 self._match_r_paren() 2980 2981 return self.expression( 2982 exp.Values, expressions=expressions, alias=alias or self._parse_table_alias() 2983 ) 2984 2985 def _parse_table_sample(self, as_modifier: bool = False) -> t.Optional[exp.TableSample]: 2986 if not self._match(TokenType.TABLE_SAMPLE) and not ( 2987 as_modifier and self._match_text_seq("USING", "SAMPLE") 2988 ): 2989 return None 2990 2991 bucket_numerator = None 2992 bucket_denominator = None 2993 bucket_field = None 2994 percent = None 2995 size = None 2996 seed = None 2997 2998 method = self._parse_var(tokens=(TokenType.ROW,), upper=True) 2999 matched_l_paren = self._match(TokenType.L_PAREN) 3000 3001 if self.TABLESAMPLE_CSV: 3002 num = None 3003 expressions = self._parse_csv(self._parse_primary) 3004 else: 3005 expressions = None 3006 num = ( 3007 self._parse_factor() 3008 if self._match(TokenType.NUMBER, advance=False) 3009 else self._parse_primary() or self._parse_placeholder() 3010 ) 3011 3012 if self._match_text_seq("BUCKET"): 3013 bucket_numerator = self._parse_number() 3014 self._match_text_seq("OUT", "OF") 3015 bucket_denominator = bucket_denominator = self._parse_number() 3016 self._match(TokenType.ON) 3017 bucket_field = self._parse_field() 3018 elif self._match_set((TokenType.PERCENT, TokenType.MOD)): 3019 percent = num 3020 elif self._match(TokenType.ROWS) or not self.dialect.TABLESAMPLE_SIZE_IS_PERCENT: 3021 size = num 3022 else: 3023 percent = num 3024 3025 if matched_l_paren: 3026 self._match_r_paren() 3027 3028 if self._match(TokenType.L_PAREN): 3029 method = self._parse_var(upper=True) 3030 seed = self._match(TokenType.COMMA) and self._parse_number() 3031 self._match_r_paren() 3032 elif self._match_texts(("SEED", "REPEATABLE")): 3033 seed = self._parse_wrapped(self._parse_number) 3034 3035 return self.expression( 3036 exp.TableSample, 3037 expressions=expressions, 3038 method=method, 3039 bucket_numerator=bucket_numerator, 3040 bucket_denominator=bucket_denominator, 3041 bucket_field=bucket_field, 3042 percent=percent, 3043 size=size, 3044 seed=seed, 3045 ) 3046 3047 def _parse_pivots(self) -> t.Optional[t.List[exp.Pivot]]: 3048 return list(iter(self._parse_pivot, None)) or None 3049 3050 def _parse_joins(self) -> t.Optional[t.List[exp.Join]]: 3051 return list(iter(self._parse_join, None)) or None 3052 3053 # https://duckdb.org/docs/sql/statements/pivot 3054 def _parse_simplified_pivot(self) -> exp.Pivot: 3055 def _parse_on() -> t.Optional[exp.Expression]: 3056 this = self._parse_bitwise() 3057 return self._parse_in(this) if self._match(TokenType.IN) else this 3058 3059 this = self._parse_table() 3060 expressions = self._match(TokenType.ON) and self._parse_csv(_parse_on) 3061 using = self._match(TokenType.USING) and self._parse_csv( 3062 lambda: self._parse_alias(self._parse_function()) 3063 ) 3064 group = self._parse_group() 3065 return self.expression( 3066 exp.Pivot, this=this, expressions=expressions, using=using, group=group 3067 ) 3068 3069 def _parse_pivot_in(self) -> exp.In: 3070 def _parse_aliased_expression() -> t.Optional[exp.Expression]: 3071 this = self._parse_conjunction() 3072 3073 self._match(TokenType.ALIAS) 3074 alias = self._parse_field() 3075 if alias: 3076 return self.expression(exp.PivotAlias, this=this, alias=alias) 3077 3078 return this 3079 3080 value = self._parse_column() 3081 3082 if not self._match_pair(TokenType.IN, TokenType.L_PAREN): 3083 self.raise_error("Expecting IN (") 3084 3085 aliased_expressions = self._parse_csv(_parse_aliased_expression) 3086 3087 self._match_r_paren() 3088 return self.expression(exp.In, this=value, expressions=aliased_expressions) 3089 3090 def _parse_pivot(self) -> t.Optional[exp.Pivot]: 3091 index = self._index 3092 include_nulls = None 3093 3094 if self._match(TokenType.PIVOT): 3095 unpivot = False 3096 elif self._match(TokenType.UNPIVOT): 3097 unpivot = True 3098 3099 # https://docs.databricks.com/en/sql/language-manual/sql-ref-syntax-qry-select-unpivot.html#syntax 3100 if self._match_text_seq("INCLUDE", "NULLS"): 3101 include_nulls = True 3102 elif self._match_text_seq("EXCLUDE", "NULLS"): 3103 include_nulls = False 3104 else: 3105 return None 3106 3107 expressions = [] 3108 3109 if not self._match(TokenType.L_PAREN): 3110 self._retreat(index) 3111 return None 3112 3113 if unpivot: 3114 expressions = self._parse_csv(self._parse_column) 3115 else: 3116 expressions = self._parse_csv(lambda: self._parse_alias(self._parse_function())) 3117 3118 if not expressions: 3119 self.raise_error("Failed to parse PIVOT's aggregation list") 3120 3121 if not self._match(TokenType.FOR): 3122 self.raise_error("Expecting FOR") 3123 3124 field = self._parse_pivot_in() 3125 3126 self._match_r_paren() 3127 3128 pivot = self.expression( 3129 exp.Pivot, 3130 expressions=expressions, 3131 field=field, 3132 unpivot=unpivot, 3133 include_nulls=include_nulls, 3134 ) 3135 3136 if not self._match_set((TokenType.PIVOT, TokenType.UNPIVOT), advance=False): 3137 pivot.set("alias", self._parse_table_alias()) 3138 3139 if not unpivot: 3140 names = self._pivot_column_names(t.cast(t.List[exp.Expression], expressions)) 3141 3142 columns: t.List[exp.Expression] = [] 3143 for fld in pivot.args["field"].expressions: 3144 field_name = fld.sql() if self.IDENTIFY_PIVOT_STRINGS else fld.alias_or_name 3145 for name in names: 3146 if self.PREFIXED_PIVOT_COLUMNS: 3147 name = f"{name}_{field_name}" if name else field_name 3148 else: 3149 name = f"{field_name}_{name}" if name else field_name 3150 3151 columns.append(exp.to_identifier(name)) 3152 3153 pivot.set("columns", columns) 3154 3155 return pivot 3156 3157 def _pivot_column_names(self, aggregations: t.List[exp.Expression]) -> t.List[str]: 3158 return [agg.alias for agg in aggregations] 3159 3160 def _parse_where(self, skip_where_token: bool = False) -> t.Optional[exp.Where]: 3161 if not skip_where_token and not self._match(TokenType.WHERE): 3162 return None 3163 3164 return self.expression( 3165 exp.Where, comments=self._prev_comments, this=self._parse_conjunction() 3166 ) 3167 3168 def _parse_group(self, skip_group_by_token: bool = False) -> t.Optional[exp.Group]: 3169 if not skip_group_by_token and not self._match(TokenType.GROUP_BY): 3170 return None 3171 3172 elements = defaultdict(list) 3173 3174 if self._match(TokenType.ALL): 3175 return self.expression(exp.Group, all=True) 3176 3177 while True: 3178 expressions = self._parse_csv(self._parse_conjunction) 3179 if expressions: 3180 elements["expressions"].extend(expressions) 3181 3182 grouping_sets = self._parse_grouping_sets() 3183 if grouping_sets: 3184 elements["grouping_sets"].extend(grouping_sets) 3185 3186 rollup = None 3187 cube = None 3188 totals = None 3189 3190 index = self._index 3191 with_ = self._match(TokenType.WITH) 3192 if self._match(TokenType.ROLLUP): 3193 rollup = with_ or self._parse_wrapped_csv(self._parse_column) 3194 elements["rollup"].extend(ensure_list(rollup)) 3195 3196 if self._match(TokenType.CUBE): 3197 cube = with_ or self._parse_wrapped_csv(self._parse_column) 3198 elements["cube"].extend(ensure_list(cube)) 3199 3200 if self._match_text_seq("TOTALS"): 3201 totals = True 3202 elements["totals"] = True # type: ignore 3203 3204 if not (grouping_sets or rollup or cube or totals): 3205 if with_: 3206 self._retreat(index) 3207 break 3208 3209 return self.expression(exp.Group, **elements) # type: ignore 3210 3211 def _parse_grouping_sets(self) -> t.Optional[t.List[exp.Expression]]: 3212 if not self._match(TokenType.GROUPING_SETS): 3213 return None 3214 3215 return self._parse_wrapped_csv(self._parse_grouping_set) 3216 3217 def _parse_grouping_set(self) -> t.Optional[exp.Expression]: 3218 if self._match(TokenType.L_PAREN): 3219 grouping_set = self._parse_csv(self._parse_column) 3220 self._match_r_paren() 3221 return self.expression(exp.Tuple, expressions=grouping_set) 3222 3223 return self._parse_column() 3224 3225 def _parse_having(self, skip_having_token: bool = False) -> t.Optional[exp.Having]: 3226 if not skip_having_token and not self._match(TokenType.HAVING): 3227 return None 3228 return self.expression(exp.Having, this=self._parse_conjunction()) 3229 3230 def _parse_qualify(self) -> t.Optional[exp.Qualify]: 3231 if not self._match(TokenType.QUALIFY): 3232 return None 3233 return self.expression(exp.Qualify, this=self._parse_conjunction()) 3234 3235 def _parse_connect(self, skip_start_token: bool = False) -> t.Optional[exp.Connect]: 3236 if skip_start_token: 3237 start = None 3238 elif self._match(TokenType.START_WITH): 3239 start = self._parse_conjunction() 3240 else: 3241 return None 3242 3243 self._match(TokenType.CONNECT_BY) 3244 self.NO_PAREN_FUNCTION_PARSERS["PRIOR"] = lambda self: self.expression( 3245 exp.Prior, this=self._parse_bitwise() 3246 ) 3247 connect = self._parse_conjunction() 3248 self.NO_PAREN_FUNCTION_PARSERS.pop("PRIOR") 3249 3250 if not start and self._match(TokenType.START_WITH): 3251 start = self._parse_conjunction() 3252 3253 return self.expression(exp.Connect, start=start, connect=connect) 3254 3255 def _parse_name_as_expression(self) -> exp.Alias: 3256 return self.expression( 3257 exp.Alias, 3258 alias=self._parse_id_var(any_token=True), 3259 this=self._match(TokenType.ALIAS) and self._parse_conjunction(), 3260 ) 3261 3262 def _parse_interpolate(self) -> t.Optional[t.List[exp.Expression]]: 3263 if self._match_text_seq("INTERPOLATE"): 3264 return self._parse_wrapped_csv(self._parse_name_as_expression) 3265 return None 3266 3267 def _parse_order( 3268 self, this: t.Optional[exp.Expression] = None, skip_order_token: bool = False 3269 ) -> t.Optional[exp.Expression]: 3270 siblings = None 3271 if not skip_order_token and not self._match(TokenType.ORDER_BY): 3272 if not self._match(TokenType.ORDER_SIBLINGS_BY): 3273 return this 3274 3275 siblings = True 3276 3277 return self.expression( 3278 exp.Order, 3279 this=this, 3280 expressions=self._parse_csv(self._parse_ordered), 3281 interpolate=self._parse_interpolate(), 3282 siblings=siblings, 3283 ) 3284 3285 def _parse_sort(self, exp_class: t.Type[E], token: TokenType) -> t.Optional[E]: 3286 if not self._match(token): 3287 return None 3288 return self.expression(exp_class, expressions=self._parse_csv(self._parse_ordered)) 3289 3290 def _parse_ordered(self, parse_method: t.Optional[t.Callable] = None) -> exp.Ordered: 3291 this = parse_method() if parse_method else self._parse_conjunction() 3292 3293 asc = self._match(TokenType.ASC) 3294 desc = self._match(TokenType.DESC) or (asc and False) 3295 3296 is_nulls_first = self._match_text_seq("NULLS", "FIRST") 3297 is_nulls_last = self._match_text_seq("NULLS", "LAST") 3298 3299 nulls_first = is_nulls_first or False 3300 explicitly_null_ordered = is_nulls_first or is_nulls_last 3301 3302 if ( 3303 not explicitly_null_ordered 3304 and ( 3305 (not desc and self.dialect.NULL_ORDERING == "nulls_are_small") 3306 or (desc and self.dialect.NULL_ORDERING != "nulls_are_small") 3307 ) 3308 and self.dialect.NULL_ORDERING != "nulls_are_last" 3309 ): 3310 nulls_first = True 3311 3312 if self._match_text_seq("WITH", "FILL"): 3313 with_fill = self.expression( 3314 exp.WithFill, 3315 **{ # type: ignore 3316 "from": self._match(TokenType.FROM) and self._parse_bitwise(), 3317 "to": self._match_text_seq("TO") and self._parse_bitwise(), 3318 "step": self._match_text_seq("STEP") and self._parse_bitwise(), 3319 }, 3320 ) 3321 else: 3322 with_fill = None 3323 3324 return self.expression( 3325 exp.Ordered, this=this, desc=desc, nulls_first=nulls_first, with_fill=with_fill 3326 ) 3327 3328 def _parse_limit( 3329 self, this: t.Optional[exp.Expression] = None, top: bool = False 3330 ) -> t.Optional[exp.Expression]: 3331 if self._match(TokenType.TOP if top else TokenType.LIMIT): 3332 comments = self._prev_comments 3333 if top: 3334 limit_paren = self._match(TokenType.L_PAREN) 3335 expression = self._parse_term() if limit_paren else self._parse_number() 3336 3337 if limit_paren: 3338 self._match_r_paren() 3339 else: 3340 expression = self._parse_term() 3341 3342 if self._match(TokenType.COMMA): 3343 offset = expression 3344 expression = self._parse_term() 3345 else: 3346 offset = None 3347 3348 limit_exp = self.expression( 3349 exp.Limit, 3350 this=this, 3351 expression=expression, 3352 offset=offset, 3353 comments=comments, 3354 expressions=self._parse_limit_by(), 3355 ) 3356 3357 return limit_exp 3358 3359 if self._match(TokenType.FETCH): 3360 direction = self._match_set((TokenType.FIRST, TokenType.NEXT)) 3361 direction = self._prev.text.upper() if direction else "FIRST" 3362 3363 count = self._parse_field(tokens=self.FETCH_TOKENS) 3364 percent = self._match(TokenType.PERCENT) 3365 3366 self._match_set((TokenType.ROW, TokenType.ROWS)) 3367 3368 only = self._match_text_seq("ONLY") 3369 with_ties = self._match_text_seq("WITH", "TIES") 3370 3371 if only and with_ties: 3372 self.raise_error("Cannot specify both ONLY and WITH TIES in FETCH clause") 3373 3374 return self.expression( 3375 exp.Fetch, 3376 direction=direction, 3377 count=count, 3378 percent=percent, 3379 with_ties=with_ties, 3380 ) 3381 3382 return this 3383 3384 def _parse_offset(self, this: t.Optional[exp.Expression] = None) -> t.Optional[exp.Expression]: 3385 if not self._match(TokenType.OFFSET): 3386 return this 3387 3388 count = self._parse_term() 3389 self._match_set((TokenType.ROW, TokenType.ROWS)) 3390 3391 return self.expression( 3392 exp.Offset, this=this, expression=count, expressions=self._parse_limit_by() 3393 ) 3394 3395 def _parse_limit_by(self) -> t.Optional[t.List[exp.Expression]]: 3396 return self._match_text_seq("BY") and self._parse_csv(self._parse_bitwise) 3397 3398 def _parse_locks(self) -> t.List[exp.Lock]: 3399 locks = [] 3400 while True: 3401 if self._match_text_seq("FOR", "UPDATE"): 3402 update = True 3403 elif self._match_text_seq("FOR", "SHARE") or self._match_text_seq( 3404 "LOCK", "IN", "SHARE", "MODE" 3405 ): 3406 update = False 3407 else: 3408 break 3409 3410 expressions = None 3411 if self._match_text_seq("OF"): 3412 expressions = self._parse_csv(lambda: self._parse_table(schema=True)) 3413 3414 wait: t.Optional[bool | exp.Expression] = None 3415 if self._match_text_seq("NOWAIT"): 3416 wait = True 3417 elif self._match_text_seq("WAIT"): 3418 wait = self._parse_primary() 3419 elif self._match_text_seq("SKIP", "LOCKED"): 3420 wait = False 3421 3422 locks.append( 3423 self.expression(exp.Lock, update=update, expressions=expressions, wait=wait) 3424 ) 3425 3426 return locks 3427 3428 def _parse_set_operations(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 3429 while this and self._match_set(self.SET_OPERATIONS): 3430 token_type = self._prev.token_type 3431 3432 if token_type == TokenType.UNION: 3433 operation = exp.Union 3434 elif token_type == TokenType.EXCEPT: 3435 operation = exp.Except 3436 else: 3437 operation = exp.Intersect 3438 3439 comments = self._prev.comments 3440 distinct = self._match(TokenType.DISTINCT) or not self._match(TokenType.ALL) 3441 by_name = self._match_text_seq("BY", "NAME") 3442 expression = self._parse_select(nested=True, parse_set_operation=False) 3443 3444 this = self.expression( 3445 operation, 3446 comments=comments, 3447 this=this, 3448 distinct=distinct, 3449 by_name=by_name, 3450 expression=expression, 3451 ) 3452 3453 if isinstance(this, exp.Union) and self.MODIFIERS_ATTACHED_TO_UNION: 3454 expression = this.expression 3455 3456 if expression: 3457 for arg in self.UNION_MODIFIERS: 3458 expr = expression.args.get(arg) 3459 if expr: 3460 this.set(arg, expr.pop()) 3461 3462 return this 3463 3464 def _parse_expression(self) -> t.Optional[exp.Expression]: 3465 return self._parse_alias(self._parse_conjunction()) 3466 3467 def _parse_conjunction(self) -> t.Optional[exp.Expression]: 3468 return self._parse_tokens(self._parse_equality, self.CONJUNCTION) 3469 3470 def _parse_equality(self) -> t.Optional[exp.Expression]: 3471 return self._parse_tokens(self._parse_comparison, self.EQUALITY) 3472 3473 def _parse_comparison(self) -> t.Optional[exp.Expression]: 3474 return self._parse_tokens(self._parse_range, self.COMPARISON) 3475 3476 def _parse_range(self, this: t.Optional[exp.Expression] = None) -> t.Optional[exp.Expression]: 3477 this = this or self._parse_bitwise() 3478 negate = self._match(TokenType.NOT) 3479 3480 if self._match_set(self.RANGE_PARSERS): 3481 expression = self.RANGE_PARSERS[self._prev.token_type](self, this) 3482 if not expression: 3483 return this 3484 3485 this = expression 3486 elif self._match(TokenType.ISNULL): 3487 this = self.expression(exp.Is, this=this, expression=exp.Null()) 3488 3489 # Postgres supports ISNULL and NOTNULL for conditions. 3490 # https://blog.andreiavram.ro/postgresql-null-composite-type/ 3491 if self._match(TokenType.NOTNULL): 3492 this = self.expression(exp.Is, this=this, expression=exp.Null()) 3493 this = self.expression(exp.Not, this=this) 3494 3495 if negate: 3496 this = self.expression(exp.Not, this=this) 3497 3498 if self._match(TokenType.IS): 3499 this = self._parse_is(this) 3500 3501 return this 3502 3503 def _parse_is(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 3504 index = self._index - 1 3505 negate = self._match(TokenType.NOT) 3506 3507 if self._match_text_seq("DISTINCT", "FROM"): 3508 klass = exp.NullSafeEQ if negate else exp.NullSafeNEQ 3509 return self.expression(klass, this=this, expression=self._parse_conjunction()) 3510 3511 expression = self._parse_null() or self._parse_boolean() 3512 if not expression: 3513 self._retreat(index) 3514 return None 3515 3516 this = self.expression(exp.Is, this=this, expression=expression) 3517 return self.expression(exp.Not, this=this) if negate else this 3518 3519 def _parse_in(self, this: t.Optional[exp.Expression], alias: bool = False) -> exp.In: 3520 unnest = self._parse_unnest(with_alias=False) 3521 if unnest: 3522 this = self.expression(exp.In, this=this, unnest=unnest) 3523 elif self._match_set((TokenType.L_PAREN, TokenType.L_BRACKET)): 3524 matched_l_paren = self._prev.token_type == TokenType.L_PAREN 3525 expressions = self._parse_csv(lambda: self._parse_select_or_expression(alias=alias)) 3526 3527 if len(expressions) == 1 and isinstance(expressions[0], exp.Subqueryable): 3528 this = self.expression(exp.In, this=this, query=expressions[0]) 3529 else: 3530 this = self.expression(exp.In, this=this, expressions=expressions) 3531 3532 if matched_l_paren: 3533 self._match_r_paren(this) 3534 elif not self._match(TokenType.R_BRACKET, expression=this): 3535 self.raise_error("Expecting ]") 3536 else: 3537 this = self.expression(exp.In, this=this, field=self._parse_field()) 3538 3539 return this 3540 3541 def _parse_between(self, this: t.Optional[exp.Expression]) -> exp.Between: 3542 low = self._parse_bitwise() 3543 self._match(TokenType.AND) 3544 high = self._parse_bitwise() 3545 return self.expression(exp.Between, this=this, low=low, high=high) 3546 3547 def _parse_escape(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 3548 if not self._match(TokenType.ESCAPE): 3549 return this 3550 return self.expression(exp.Escape, this=this, expression=self._parse_string()) 3551 3552 def _parse_interval(self, match_interval: bool = True) -> t.Optional[exp.Interval]: 3553 index = self._index 3554 3555 if not self._match(TokenType.INTERVAL) and match_interval: 3556 return None 3557 3558 if self._match(TokenType.STRING, advance=False): 3559 this = self._parse_primary() 3560 else: 3561 this = self._parse_term() 3562 3563 if not this or ( 3564 isinstance(this, exp.Column) 3565 and not this.table 3566 and not this.this.quoted 3567 and this.name.upper() == "IS" 3568 ): 3569 self._retreat(index) 3570 return None 3571 3572 unit = self._parse_function() or ( 3573 not self._match(TokenType.ALIAS, advance=False) 3574 and self._parse_var(any_token=True, upper=True) 3575 ) 3576 3577 # Most dialects support, e.g., the form INTERVAL '5' day, thus we try to parse 3578 # each INTERVAL expression into this canonical form so it's easy to transpile 3579 if this and this.is_number: 3580 this = exp.Literal.string(this.name) 3581 elif this and this.is_string: 3582 parts = this.name.split() 3583 3584 if len(parts) == 2: 3585 if unit: 3586 # This is not actually a unit, it's something else (e.g. a "window side") 3587 unit = None 3588 self._retreat(self._index - 1) 3589 3590 this = exp.Literal.string(parts[0]) 3591 unit = self.expression(exp.Var, this=parts[1].upper()) 3592 3593 return self.expression(exp.Interval, this=this, unit=unit) 3594 3595 def _parse_bitwise(self) -> t.Optional[exp.Expression]: 3596 this = self._parse_term() 3597 3598 while True: 3599 if self._match_set(self.BITWISE): 3600 this = self.expression( 3601 self.BITWISE[self._prev.token_type], 3602 this=this, 3603 expression=self._parse_term(), 3604 ) 3605 elif self.dialect.DPIPE_IS_STRING_CONCAT and self._match(TokenType.DPIPE): 3606 this = self.expression( 3607 exp.DPipe, 3608 this=this, 3609 expression=self._parse_term(), 3610 safe=not self.dialect.STRICT_STRING_CONCAT, 3611 ) 3612 elif self._match(TokenType.DQMARK): 3613 this = self.expression(exp.Coalesce, this=this, expressions=self._parse_term()) 3614 elif self._match_pair(TokenType.LT, TokenType.LT): 3615 this = self.expression( 3616 exp.BitwiseLeftShift, this=this, expression=self._parse_term() 3617 ) 3618 elif self._match_pair(TokenType.GT, TokenType.GT): 3619 this = self.expression( 3620 exp.BitwiseRightShift, this=this, expression=self._parse_term() 3621 ) 3622 else: 3623 break 3624 3625 return this 3626 3627 def _parse_term(self) -> t.Optional[exp.Expression]: 3628 return self._parse_tokens(self._parse_factor, self.TERM) 3629 3630 def _parse_factor(self) -> t.Optional[exp.Expression]: 3631 parse_method = self._parse_exponent if self.EXPONENT else self._parse_unary 3632 this = parse_method() 3633 3634 while self._match_set(self.FACTOR): 3635 this = self.expression( 3636 self.FACTOR[self._prev.token_type], 3637 this=this, 3638 comments=self._prev_comments, 3639 expression=parse_method(), 3640 ) 3641 if isinstance(this, exp.Div): 3642 this.args["typed"] = self.dialect.TYPED_DIVISION 3643 this.args["safe"] = self.dialect.SAFE_DIVISION 3644 3645 return this 3646 3647 def _parse_exponent(self) -> t.Optional[exp.Expression]: 3648 return self._parse_tokens(self._parse_unary, self.EXPONENT) 3649 3650 def _parse_unary(self) -> t.Optional[exp.Expression]: 3651 if self._match_set(self.UNARY_PARSERS): 3652 return self.UNARY_PARSERS[self._prev.token_type](self) 3653 return self._parse_at_time_zone(self._parse_type()) 3654 3655 def _parse_type(self, parse_interval: bool = True) -> t.Optional[exp.Expression]: 3656 interval = parse_interval and self._parse_interval() 3657 if interval: 3658 # Convert INTERVAL 'val_1' unit_1 ... 'val_n' unit_n into a sum of intervals 3659 while self._match_set((TokenType.STRING, TokenType.NUMBER), advance=False): 3660 interval = self.expression( # type: ignore 3661 exp.Add, this=interval, expression=self._parse_interval(match_interval=False) 3662 ) 3663 3664 return interval 3665 3666 index = self._index 3667 data_type = self._parse_types(check_func=True, allow_identifiers=False) 3668 this = self._parse_column() 3669 3670 if data_type: 3671 if isinstance(this, exp.Literal): 3672 parser = self.TYPE_LITERAL_PARSERS.get(data_type.this) 3673 if parser: 3674 return parser(self, this, data_type) 3675 return self.expression(exp.Cast, this=this, to=data_type) 3676 if not data_type.expressions: 3677 self._retreat(index) 3678 return self._parse_column() 3679 return self._parse_column_ops(data_type) 3680 3681 return this and self._parse_column_ops(this) 3682 3683 def _parse_type_size(self) -> t.Optional[exp.DataTypeParam]: 3684 this = self._parse_type() 3685 if not this: 3686 return None 3687 3688 return self.expression( 3689 exp.DataTypeParam, this=this, expression=self._parse_var(any_token=True) 3690 ) 3691 3692 def _parse_types( 3693 self, check_func: bool = False, schema: bool = False, allow_identifiers: bool = True 3694 ) -> t.Optional[exp.Expression]: 3695 index = self._index 3696 3697 prefix = self._match_text_seq("SYSUDTLIB", ".") 3698 3699 if not self._match_set(self.TYPE_TOKENS): 3700 identifier = allow_identifiers and self._parse_id_var( 3701 any_token=False, tokens=(TokenType.VAR,) 3702 ) 3703 if identifier: 3704 tokens = self.dialect.tokenize(identifier.name) 3705 3706 if len(tokens) != 1: 3707 self.raise_error("Unexpected identifier", self._prev) 3708 3709 if tokens[0].token_type in self.TYPE_TOKENS: 3710 self._prev = tokens[0] 3711 elif self.dialect.SUPPORTS_USER_DEFINED_TYPES: 3712 type_name = identifier.name 3713 3714 while self._match(TokenType.DOT): 3715 type_name = f"{type_name}.{self._advance_any() and self._prev.text}" 3716 3717 return exp.DataType.build(type_name, udt=True) 3718 else: 3719 self._retreat(self._index - 1) 3720 return None 3721 else: 3722 return None 3723 3724 type_token = self._prev.token_type 3725 3726 if type_token == TokenType.PSEUDO_TYPE: 3727 return self.expression(exp.PseudoType, this=self._prev.text.upper()) 3728 3729 if type_token == TokenType.OBJECT_IDENTIFIER: 3730 return self.expression(exp.ObjectIdentifier, this=self._prev.text.upper()) 3731 3732 nested = type_token in self.NESTED_TYPE_TOKENS 3733 is_struct = type_token in self.STRUCT_TYPE_TOKENS 3734 is_aggregate = type_token in self.AGGREGATE_TYPE_TOKENS 3735 expressions = None 3736 maybe_func = False 3737 3738 if self._match(TokenType.L_PAREN): 3739 if is_struct: 3740 expressions = self._parse_csv(self._parse_struct_types) 3741 elif nested: 3742 expressions = self._parse_csv( 3743 lambda: self._parse_types( 3744 check_func=check_func, schema=schema, allow_identifiers=allow_identifiers 3745 ) 3746 ) 3747 elif type_token in self.ENUM_TYPE_TOKENS: 3748 expressions = self._parse_csv(self._parse_equality) 3749 elif is_aggregate: 3750 func_or_ident = self._parse_function(anonymous=True) or self._parse_id_var( 3751 any_token=False, tokens=(TokenType.VAR,) 3752 ) 3753 if not func_or_ident or not self._match(TokenType.COMMA): 3754 return None 3755 expressions = self._parse_csv( 3756 lambda: self._parse_types( 3757 check_func=check_func, schema=schema, allow_identifiers=allow_identifiers 3758 ) 3759 ) 3760 expressions.insert(0, func_or_ident) 3761 else: 3762 expressions = self._parse_csv(self._parse_type_size) 3763 3764 if not expressions or not self._match(TokenType.R_PAREN): 3765 self._retreat(index) 3766 return None 3767 3768 maybe_func = True 3769 3770 this: t.Optional[exp.Expression] = None 3771 values: t.Optional[t.List[exp.Expression]] = None 3772 3773 if nested and self._match(TokenType.LT): 3774 if is_struct: 3775 expressions = self._parse_csv(lambda: self._parse_struct_types(type_required=True)) 3776 else: 3777 expressions = self._parse_csv( 3778 lambda: self._parse_types( 3779 check_func=check_func, schema=schema, allow_identifiers=allow_identifiers 3780 ) 3781 ) 3782 3783 if not self._match(TokenType.GT): 3784 self.raise_error("Expecting >") 3785 3786 if self._match_set((TokenType.L_BRACKET, TokenType.L_PAREN)): 3787 values = self._parse_csv(self._parse_conjunction) 3788 self._match_set((TokenType.R_BRACKET, TokenType.R_PAREN)) 3789 3790 if type_token in self.TIMESTAMPS: 3791 if self._match_text_seq("WITH", "TIME", "ZONE"): 3792 maybe_func = False 3793 tz_type = ( 3794 exp.DataType.Type.TIMETZ 3795 if type_token in self.TIMES 3796 else exp.DataType.Type.TIMESTAMPTZ 3797 ) 3798 this = exp.DataType(this=tz_type, expressions=expressions) 3799 elif self._match_text_seq("WITH", "LOCAL", "TIME", "ZONE"): 3800 maybe_func = False 3801 this = exp.DataType(this=exp.DataType.Type.TIMESTAMPLTZ, expressions=expressions) 3802 elif self._match_text_seq("WITHOUT", "TIME", "ZONE"): 3803 maybe_func = False 3804 elif type_token == TokenType.INTERVAL: 3805 unit = self._parse_var() 3806 3807 if self._match_text_seq("TO"): 3808 span = [exp.IntervalSpan(this=unit, expression=self._parse_var())] 3809 else: 3810 span = None 3811 3812 if span or not unit: 3813 this = self.expression( 3814 exp.DataType, this=exp.DataType.Type.INTERVAL, expressions=span 3815 ) 3816 else: 3817 this = self.expression(exp.DataType, this=self.expression(exp.Interval, unit=unit)) 3818 3819 if maybe_func and check_func: 3820 index2 = self._index 3821 peek = self._parse_string() 3822 3823 if not peek: 3824 self._retreat(index) 3825 return None 3826 3827 self._retreat(index2) 3828 3829 if not this: 3830 if self._match_text_seq("UNSIGNED"): 3831 unsigned_type_token = self.SIGNED_TO_UNSIGNED_TYPE_TOKEN.get(type_token) 3832 if not unsigned_type_token: 3833 self.raise_error(f"Cannot convert {type_token.value} to unsigned.") 3834 3835 type_token = unsigned_type_token or type_token 3836 3837 this = exp.DataType( 3838 this=exp.DataType.Type[type_token.value], 3839 expressions=expressions, 3840 nested=nested, 3841 values=values, 3842 prefix=prefix, 3843 ) 3844 3845 while self._match_pair(TokenType.L_BRACKET, TokenType.R_BRACKET): 3846 this = exp.DataType(this=exp.DataType.Type.ARRAY, expressions=[this], nested=True) 3847 3848 return this 3849 3850 def _parse_struct_types(self, type_required: bool = False) -> t.Optional[exp.Expression]: 3851 index = self._index 3852 this = self._parse_type(parse_interval=False) or self._parse_id_var() 3853 self._match(TokenType.COLON) 3854 column_def = self._parse_column_def(this) 3855 3856 if type_required and ( 3857 (isinstance(this, exp.Column) and this.this is column_def) or this is column_def 3858 ): 3859 self._retreat(index) 3860 return self._parse_types() 3861 3862 return column_def 3863 3864 def _parse_at_time_zone(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 3865 if not self._match_text_seq("AT", "TIME", "ZONE"): 3866 return this 3867 return self.expression(exp.AtTimeZone, this=this, zone=self._parse_unary()) 3868 3869 def _parse_column(self) -> t.Optional[exp.Expression]: 3870 this = self._parse_column_reference() 3871 return self._parse_column_ops(this) if this else self._parse_bracket(this) 3872 3873 def _parse_column_reference(self) -> t.Optional[exp.Expression]: 3874 this = self._parse_field() 3875 if isinstance(this, exp.Identifier): 3876 this = self.expression(exp.Column, this=this) 3877 return this 3878 3879 def _parse_column_ops(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 3880 this = self._parse_bracket(this) 3881 3882 while self._match_set(self.COLUMN_OPERATORS): 3883 op_token = self._prev.token_type 3884 op = self.COLUMN_OPERATORS.get(op_token) 3885 3886 if op_token == TokenType.DCOLON: 3887 field = self._parse_types() 3888 if not field: 3889 self.raise_error("Expected type") 3890 elif op and self._curr: 3891 field = self._parse_column_reference() 3892 else: 3893 field = self._parse_field(anonymous_func=True, any_token=True) 3894 3895 if isinstance(field, exp.Func): 3896 # bigquery allows function calls like x.y.count(...) 3897 # SAFE.SUBSTR(...) 3898 # https://cloud.google.com/bigquery/docs/reference/standard-sql/functions-reference#function_call_rules 3899 this = self._replace_columns_with_dots(this) 3900 3901 if op: 3902 this = op(self, this, field) 3903 elif isinstance(this, exp.Column) and not this.args.get("catalog"): 3904 this = self.expression( 3905 exp.Column, 3906 this=field, 3907 table=this.this, 3908 db=this.args.get("table"), 3909 catalog=this.args.get("db"), 3910 ) 3911 else: 3912 this = self.expression(exp.Dot, this=this, expression=field) 3913 this = self._parse_bracket(this) 3914 return this 3915 3916 def _parse_primary(self) -> t.Optional[exp.Expression]: 3917 if self._match_set(self.PRIMARY_PARSERS): 3918 token_type = self._prev.token_type 3919 primary = self.PRIMARY_PARSERS[token_type](self, self._prev) 3920 3921 if token_type == TokenType.STRING: 3922 expressions = [primary] 3923 while self._match(TokenType.STRING): 3924 expressions.append(exp.Literal.string(self._prev.text)) 3925 3926 if len(expressions) > 1: 3927 return self.expression(exp.Concat, expressions=expressions) 3928 3929 return primary 3930 3931 if self._match_pair(TokenType.DOT, TokenType.NUMBER): 3932 return exp.Literal.number(f"0.{self._prev.text}") 3933 3934 if self._match(TokenType.L_PAREN): 3935 comments = self._prev_comments 3936 query = self._parse_select() 3937 3938 if query: 3939 expressions = [query] 3940 else: 3941 expressions = self._parse_expressions() 3942 3943 this = self._parse_query_modifiers(seq_get(expressions, 0)) 3944 3945 if isinstance(this, exp.Subqueryable): 3946 this = self._parse_set_operations( 3947 self._parse_subquery(this=this, parse_alias=False) 3948 ) 3949 elif len(expressions) > 1: 3950 this = self.expression(exp.Tuple, expressions=expressions) 3951 else: 3952 this = self.expression(exp.Paren, this=self._parse_set_operations(this)) 3953 3954 if this: 3955 this.add_comments(comments) 3956 3957 self._match_r_paren(expression=this) 3958 return this 3959 3960 return None 3961 3962 def _parse_field( 3963 self, 3964 any_token: bool = False, 3965 tokens: t.Optional[t.Collection[TokenType]] = None, 3966 anonymous_func: bool = False, 3967 ) -> t.Optional[exp.Expression]: 3968 return ( 3969 self._parse_primary() 3970 or self._parse_function(anonymous=anonymous_func) 3971 or self._parse_id_var(any_token=any_token, tokens=tokens) 3972 ) 3973 3974 def _parse_function( 3975 self, 3976 functions: t.Optional[t.Dict[str, t.Callable]] = None, 3977 anonymous: bool = False, 3978 optional_parens: bool = True, 3979 ) -> t.Optional[exp.Expression]: 3980 # This allows us to also parse {fn <function>} syntax (Snowflake, MySQL support this) 3981 # See: https://community.snowflake.com/s/article/SQL-Escape-Sequences 3982 fn_syntax = False 3983 if ( 3984 self._match(TokenType.L_BRACE, advance=False) 3985 and self._next 3986 and self._next.text.upper() == "FN" 3987 ): 3988 self._advance(2) 3989 fn_syntax = True 3990 3991 func = self._parse_function_call( 3992 functions=functions, anonymous=anonymous, optional_parens=optional_parens 3993 ) 3994 3995 if fn_syntax: 3996 self._match(TokenType.R_BRACE) 3997 3998 return func 3999 4000 def _parse_function_call( 4001 self, 4002 functions: t.Optional[t.Dict[str, t.Callable]] = None, 4003 anonymous: bool = False, 4004 optional_parens: bool = True, 4005 ) -> t.Optional[exp.Expression]: 4006 if not self._curr: 4007 return None 4008 4009 comments = self._curr.comments 4010 token_type = self._curr.token_type 4011 this = self._curr.text 4012 upper = this.upper() 4013 4014 parser = self.NO_PAREN_FUNCTION_PARSERS.get(upper) 4015 if optional_parens and parser and token_type not in self.INVALID_FUNC_NAME_TOKENS: 4016 self._advance() 4017 return parser(self) 4018 4019 if not self._next or self._next.token_type != TokenType.L_PAREN: 4020 if optional_parens and token_type in self.NO_PAREN_FUNCTIONS: 4021 self._advance() 4022 return self.expression(self.NO_PAREN_FUNCTIONS[token_type]) 4023 4024 return None 4025 4026 if token_type not in self.FUNC_TOKENS: 4027 return None 4028 4029 self._advance(2) 4030 4031 parser = self.FUNCTION_PARSERS.get(upper) 4032 if parser and not anonymous: 4033 this = parser(self) 4034 else: 4035 subquery_predicate = self.SUBQUERY_PREDICATES.get(token_type) 4036 4037 if subquery_predicate and self._curr.token_type in (TokenType.SELECT, TokenType.WITH): 4038 this = self.expression(subquery_predicate, this=self._parse_select()) 4039 self._match_r_paren() 4040 return this 4041 4042 if functions is None: 4043 functions = self.FUNCTIONS 4044 4045 function = functions.get(upper) 4046 4047 alias = upper in self.FUNCTIONS_WITH_ALIASED_ARGS 4048 args = self._parse_csv(lambda: self._parse_lambda(alias=alias)) 4049 4050 if function and not anonymous: 4051 if "dialect" in function.__code__.co_varnames: 4052 func = function(args, dialect=self.dialect) 4053 else: 4054 func = function(args) 4055 4056 func = self.validate_expression(func, args) 4057 if not self.dialect.NORMALIZE_FUNCTIONS: 4058 func.meta["name"] = this 4059 4060 this = func 4061 else: 4062 this = self.expression(exp.Anonymous, this=this, expressions=args) 4063 4064 if isinstance(this, exp.Expression): 4065 this.add_comments(comments) 4066 4067 self._match_r_paren(this) 4068 return self._parse_window(this) 4069 4070 def _parse_function_parameter(self) -> t.Optional[exp.Expression]: 4071 return self._parse_column_def(self._parse_id_var()) 4072 4073 def _parse_user_defined_function( 4074 self, kind: t.Optional[TokenType] = None 4075 ) -> t.Optional[exp.Expression]: 4076 this = self._parse_id_var() 4077 4078 while self._match(TokenType.DOT): 4079 this = self.expression(exp.Dot, this=this, expression=self._parse_id_var()) 4080 4081 if not self._match(TokenType.L_PAREN): 4082 return this 4083 4084 expressions = self._parse_csv(self._parse_function_parameter) 4085 self._match_r_paren() 4086 return self.expression( 4087 exp.UserDefinedFunction, this=this, expressions=expressions, wrapped=True 4088 ) 4089 4090 def _parse_introducer(self, token: Token) -> exp.Introducer | exp.Identifier: 4091 literal = self._parse_primary() 4092 if literal: 4093 return self.expression(exp.Introducer, this=token.text, expression=literal) 4094 4095 return self.expression(exp.Identifier, this=token.text) 4096 4097 def _parse_session_parameter(self) -> exp.SessionParameter: 4098 kind = None 4099 this = self._parse_id_var() or self._parse_primary() 4100 4101 if this and self._match(TokenType.DOT): 4102 kind = this.name 4103 this = self._parse_var() or self._parse_primary() 4104 4105 return self.expression(exp.SessionParameter, this=this, kind=kind) 4106 4107 def _parse_lambda(self, alias: bool = False) -> t.Optional[exp.Expression]: 4108 index = self._index 4109 4110 if self._match(TokenType.L_PAREN): 4111 expressions = t.cast( 4112 t.List[t.Optional[exp.Expression]], self._parse_csv(self._parse_id_var) 4113 ) 4114 4115 if not self._match(TokenType.R_PAREN): 4116 self._retreat(index) 4117 else: 4118 expressions = [self._parse_id_var()] 4119 4120 if self._match_set(self.LAMBDAS): 4121 return self.LAMBDAS[self._prev.token_type](self, expressions) 4122 4123 self._retreat(index) 4124 4125 this: t.Optional[exp.Expression] 4126 4127 if self._match(TokenType.DISTINCT): 4128 this = self.expression( 4129 exp.Distinct, expressions=self._parse_csv(self._parse_conjunction) 4130 ) 4131 else: 4132 this = self._parse_select_or_expression(alias=alias) 4133 4134 return self._parse_limit( 4135 self._parse_order(self._parse_having_max(self._parse_respect_or_ignore_nulls(this))) 4136 ) 4137 4138 def _parse_schema(self, this: t.Optional[exp.Expression] = None) -> t.Optional[exp.Expression]: 4139 index = self._index 4140 4141 if not self.errors: 4142 try: 4143 if self._parse_select(nested=True): 4144 return this 4145 except ParseError: 4146 pass 4147 finally: 4148 self.errors.clear() 4149 self._retreat(index) 4150 4151 if not self._match(TokenType.L_PAREN): 4152 return this 4153 4154 args = self._parse_csv(lambda: self._parse_constraint() or self._parse_field_def()) 4155 4156 self._match_r_paren() 4157 return self.expression(exp.Schema, this=this, expressions=args) 4158 4159 def _parse_field_def(self) -> t.Optional[exp.Expression]: 4160 return self._parse_column_def(self._parse_field(any_token=True)) 4161 4162 def _parse_column_def(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 4163 # column defs are not really columns, they're identifiers 4164 if isinstance(this, exp.Column): 4165 this = this.this 4166 4167 kind = self._parse_types(schema=True) 4168 4169 if self._match_text_seq("FOR", "ORDINALITY"): 4170 return self.expression(exp.ColumnDef, this=this, ordinality=True) 4171 4172 constraints: t.List[exp.Expression] = [] 4173 4174 if not kind and self._match(TokenType.ALIAS): 4175 constraints.append( 4176 self.expression( 4177 exp.ComputedColumnConstraint, 4178 this=self._parse_conjunction(), 4179 persisted=self._match_text_seq("PERSISTED"), 4180 not_null=self._match_pair(TokenType.NOT, TokenType.NULL), 4181 ) 4182 ) 4183 elif kind and self._match_pair(TokenType.ALIAS, TokenType.L_PAREN, advance=False): 4184 self._match(TokenType.ALIAS) 4185 constraints.append( 4186 self.expression(exp.TransformColumnConstraint, this=self._parse_field()) 4187 ) 4188 4189 while True: 4190 constraint = self._parse_column_constraint() 4191 if not constraint: 4192 break 4193 constraints.append(constraint) 4194 4195 if not kind and not constraints: 4196 return this 4197 4198 return self.expression(exp.ColumnDef, this=this, kind=kind, constraints=constraints) 4199 4200 def _parse_auto_increment( 4201 self, 4202 ) -> exp.GeneratedAsIdentityColumnConstraint | exp.AutoIncrementColumnConstraint: 4203 start = None 4204 increment = None 4205 4206 if self._match(TokenType.L_PAREN, advance=False): 4207 args = self._parse_wrapped_csv(self._parse_bitwise) 4208 start = seq_get(args, 0) 4209 increment = seq_get(args, 1) 4210 elif self._match_text_seq("START"): 4211 start = self._parse_bitwise() 4212 self._match_text_seq("INCREMENT") 4213 increment = self._parse_bitwise() 4214 4215 if start and increment: 4216 return exp.GeneratedAsIdentityColumnConstraint(start=start, increment=increment) 4217 4218 return exp.AutoIncrementColumnConstraint() 4219 4220 def _parse_auto_property(self) -> t.Optional[exp.AutoRefreshProperty]: 4221 if not self._match_text_seq("REFRESH"): 4222 self._retreat(self._index - 1) 4223 return None 4224 return self.expression(exp.AutoRefreshProperty, this=self._parse_var(upper=True)) 4225 4226 def _parse_compress(self) -> exp.CompressColumnConstraint: 4227 if self._match(TokenType.L_PAREN, advance=False): 4228 return self.expression( 4229 exp.CompressColumnConstraint, this=self._parse_wrapped_csv(self._parse_bitwise) 4230 ) 4231 4232 return self.expression(exp.CompressColumnConstraint, this=self._parse_bitwise()) 4233 4234 def _parse_generated_as_identity( 4235 self, 4236 ) -> ( 4237 exp.GeneratedAsIdentityColumnConstraint 4238 | exp.ComputedColumnConstraint 4239 | exp.GeneratedAsRowColumnConstraint 4240 ): 4241 if self._match_text_seq("BY", "DEFAULT"): 4242 on_null = self._match_pair(TokenType.ON, TokenType.NULL) 4243 this = self.expression( 4244 exp.GeneratedAsIdentityColumnConstraint, this=False, on_null=on_null 4245 ) 4246 else: 4247 self._match_text_seq("ALWAYS") 4248 this = self.expression(exp.GeneratedAsIdentityColumnConstraint, this=True) 4249 4250 self._match(TokenType.ALIAS) 4251 4252 if self._match_text_seq("ROW"): 4253 start = self._match_text_seq("START") 4254 if not start: 4255 self._match(TokenType.END) 4256 hidden = self._match_text_seq("HIDDEN") 4257 return self.expression(exp.GeneratedAsRowColumnConstraint, start=start, hidden=hidden) 4258 4259 identity = self._match_text_seq("IDENTITY") 4260 4261 if self._match(TokenType.L_PAREN): 4262 if self._match(TokenType.START_WITH): 4263 this.set("start", self._parse_bitwise()) 4264 if self._match_text_seq("INCREMENT", "BY"): 4265 this.set("increment", self._parse_bitwise()) 4266 if self._match_text_seq("MINVALUE"): 4267 this.set("minvalue", self._parse_bitwise()) 4268 if self._match_text_seq("MAXVALUE"): 4269 this.set("maxvalue", self._parse_bitwise()) 4270 4271 if self._match_text_seq("CYCLE"): 4272 this.set("cycle", True) 4273 elif self._match_text_seq("NO", "CYCLE"): 4274 this.set("cycle", False) 4275 4276 if not identity: 4277 this.set("expression", self._parse_bitwise()) 4278 elif not this.args.get("start") and self._match(TokenType.NUMBER, advance=False): 4279 args = self._parse_csv(self._parse_bitwise) 4280 this.set("start", seq_get(args, 0)) 4281 this.set("increment", seq_get(args, 1)) 4282 4283 self._match_r_paren() 4284 4285 return this 4286 4287 def _parse_inline(self) -> exp.InlineLengthColumnConstraint: 4288 self._match_text_seq("LENGTH") 4289 return self.expression(exp.InlineLengthColumnConstraint, this=self._parse_bitwise()) 4290 4291 def _parse_not_constraint( 4292 self, 4293 ) -> t.Optional[exp.Expression]: 4294 if self._match_text_seq("NULL"): 4295 return self.expression(exp.NotNullColumnConstraint) 4296 if self._match_text_seq("CASESPECIFIC"): 4297 return self.expression(exp.CaseSpecificColumnConstraint, not_=True) 4298 if self._match_text_seq("FOR", "REPLICATION"): 4299 return self.expression(exp.NotForReplicationColumnConstraint) 4300 return None 4301 4302 def _parse_column_constraint(self) -> t.Optional[exp.Expression]: 4303 if self._match(TokenType.CONSTRAINT): 4304 this = self._parse_id_var() 4305 else: 4306 this = None 4307 4308 if self._match_texts(self.CONSTRAINT_PARSERS): 4309 return self.expression( 4310 exp.ColumnConstraint, 4311 this=this, 4312 kind=self.CONSTRAINT_PARSERS[self._prev.text.upper()](self), 4313 ) 4314 4315 return this 4316 4317 def _parse_constraint(self) -> t.Optional[exp.Expression]: 4318 if not self._match(TokenType.CONSTRAINT): 4319 return self._parse_unnamed_constraint(constraints=self.SCHEMA_UNNAMED_CONSTRAINTS) 4320 4321 this = self._parse_id_var() 4322 expressions = [] 4323 4324 while True: 4325 constraint = self._parse_unnamed_constraint() or self._parse_function() 4326 if not constraint: 4327 break 4328 expressions.append(constraint) 4329 4330 return self.expression(exp.Constraint, this=this, expressions=expressions) 4331 4332 def _parse_unnamed_constraint( 4333 self, constraints: t.Optional[t.Collection[str]] = None 4334 ) -> t.Optional[exp.Expression]: 4335 if self._match(TokenType.IDENTIFIER, advance=False) or not self._match_texts( 4336 constraints or self.CONSTRAINT_PARSERS 4337 ): 4338 return None 4339 4340 constraint = self._prev.text.upper() 4341 if constraint not in self.CONSTRAINT_PARSERS: 4342 self.raise_error(f"No parser found for schema constraint {constraint}.") 4343 4344 return self.CONSTRAINT_PARSERS[constraint](self) 4345 4346 def _parse_unique(self) -> exp.UniqueColumnConstraint: 4347 self._match_text_seq("KEY") 4348 return self.expression( 4349 exp.UniqueColumnConstraint, 4350 this=self._parse_schema(self._parse_id_var(any_token=False)), 4351 index_type=self._match(TokenType.USING) and self._advance_any() and self._prev.text, 4352 ) 4353 4354 def _parse_key_constraint_options(self) -> t.List[str]: 4355 options = [] 4356 while True: 4357 if not self._curr: 4358 break 4359 4360 if self._match(TokenType.ON): 4361 action = None 4362 on = self._advance_any() and self._prev.text 4363 4364 if self._match_text_seq("NO", "ACTION"): 4365 action = "NO ACTION" 4366 elif self._match_text_seq("CASCADE"): 4367 action = "CASCADE" 4368 elif self._match_text_seq("RESTRICT"): 4369 action = "RESTRICT" 4370 elif self._match_pair(TokenType.SET, TokenType.NULL): 4371 action = "SET NULL" 4372 elif self._match_pair(TokenType.SET, TokenType.DEFAULT): 4373 action = "SET DEFAULT" 4374 else: 4375 self.raise_error("Invalid key constraint") 4376 4377 options.append(f"ON {on} {action}") 4378 elif self._match_text_seq("NOT", "ENFORCED"): 4379 options.append("NOT ENFORCED") 4380 elif self._match_text_seq("DEFERRABLE"): 4381 options.append("DEFERRABLE") 4382 elif self._match_text_seq("INITIALLY", "DEFERRED"): 4383 options.append("INITIALLY DEFERRED") 4384 elif self._match_text_seq("NORELY"): 4385 options.append("NORELY") 4386 elif self._match_text_seq("MATCH", "FULL"): 4387 options.append("MATCH FULL") 4388 else: 4389 break 4390 4391 return options 4392 4393 def _parse_references(self, match: bool = True) -> t.Optional[exp.Reference]: 4394 if match and not self._match(TokenType.REFERENCES): 4395 return None 4396 4397 expressions = None 4398 this = self._parse_table(schema=True) 4399 options = self._parse_key_constraint_options() 4400 return self.expression(exp.Reference, this=this, expressions=expressions, options=options) 4401 4402 def _parse_foreign_key(self) -> exp.ForeignKey: 4403 expressions = self._parse_wrapped_id_vars() 4404 reference = self._parse_references() 4405 options = {} 4406 4407 while self._match(TokenType.ON): 4408 if not self._match_set((TokenType.DELETE, TokenType.UPDATE)): 4409 self.raise_error("Expected DELETE or UPDATE") 4410 4411 kind = self._prev.text.lower() 4412 4413 if self._match_text_seq("NO", "ACTION"): 4414 action = "NO ACTION" 4415 elif self._match(TokenType.SET): 4416 self._match_set((TokenType.NULL, TokenType.DEFAULT)) 4417 action = "SET " + self._prev.text.upper() 4418 else: 4419 self._advance() 4420 action = self._prev.text.upper() 4421 4422 options[kind] = action 4423 4424 return self.expression( 4425 exp.ForeignKey, 4426 expressions=expressions, 4427 reference=reference, 4428 **options, # type: ignore 4429 ) 4430 4431 def _parse_primary_key_part(self) -> t.Optional[exp.Expression]: 4432 return self._parse_field() 4433 4434 def _parse_period_for_system_time(self) -> t.Optional[exp.PeriodForSystemTimeConstraint]: 4435 if not self._match(TokenType.TIMESTAMP_SNAPSHOT): 4436 self._retreat(self._index - 1) 4437 return None 4438 4439 id_vars = self._parse_wrapped_id_vars() 4440 return self.expression( 4441 exp.PeriodForSystemTimeConstraint, 4442 this=seq_get(id_vars, 0), 4443 expression=seq_get(id_vars, 1), 4444 ) 4445 4446 def _parse_primary_key( 4447 self, wrapped_optional: bool = False, in_props: bool = False 4448 ) -> exp.PrimaryKeyColumnConstraint | exp.PrimaryKey: 4449 desc = ( 4450 self._match_set((TokenType.ASC, TokenType.DESC)) 4451 and self._prev.token_type == TokenType.DESC 4452 ) 4453 4454 if not in_props and not self._match(TokenType.L_PAREN, advance=False): 4455 return self.expression(exp.PrimaryKeyColumnConstraint, desc=desc) 4456 4457 expressions = self._parse_wrapped_csv( 4458 self._parse_primary_key_part, optional=wrapped_optional 4459 ) 4460 options = self._parse_key_constraint_options() 4461 return self.expression(exp.PrimaryKey, expressions=expressions, options=options) 4462 4463 def _parse_bracket_key_value(self, is_map: bool = False) -> t.Optional[exp.Expression]: 4464 return self._parse_slice(self._parse_alias(self._parse_conjunction(), explicit=True)) 4465 4466 def _parse_bracket(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 4467 if not self._match_set((TokenType.L_BRACKET, TokenType.L_BRACE)): 4468 return this 4469 4470 bracket_kind = self._prev.token_type 4471 expressions = self._parse_csv( 4472 lambda: self._parse_bracket_key_value(is_map=bracket_kind == TokenType.L_BRACE) 4473 ) 4474 4475 if not self._match(TokenType.R_BRACKET) and bracket_kind == TokenType.L_BRACKET: 4476 self.raise_error("Expected ]") 4477 elif not self._match(TokenType.R_BRACE) and bracket_kind == TokenType.L_BRACE: 4478 self.raise_error("Expected }") 4479 4480 # https://duckdb.org/docs/sql/data_types/struct.html#creating-structs 4481 if bracket_kind == TokenType.L_BRACE: 4482 this = self.expression(exp.Struct, expressions=expressions) 4483 elif not this or this.name.upper() == "ARRAY": 4484 this = self.expression(exp.Array, expressions=expressions) 4485 else: 4486 expressions = apply_index_offset(this, expressions, -self.dialect.INDEX_OFFSET) 4487 this = self.expression(exp.Bracket, this=this, expressions=expressions) 4488 4489 self._add_comments(this) 4490 return self._parse_bracket(this) 4491 4492 def _parse_slice(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 4493 if self._match(TokenType.COLON): 4494 return self.expression(exp.Slice, this=this, expression=self._parse_conjunction()) 4495 return this 4496 4497 def _parse_case(self) -> t.Optional[exp.Expression]: 4498 ifs = [] 4499 default = None 4500 4501 comments = self._prev_comments 4502 expression = self._parse_conjunction() 4503 4504 while self._match(TokenType.WHEN): 4505 this = self._parse_conjunction() 4506 self._match(TokenType.THEN) 4507 then = self._parse_conjunction() 4508 ifs.append(self.expression(exp.If, this=this, true=then)) 4509 4510 if self._match(TokenType.ELSE): 4511 default = self._parse_conjunction() 4512 4513 if not self._match(TokenType.END): 4514 if isinstance(default, exp.Interval) and default.this.sql().upper() == "END": 4515 default = exp.column("interval") 4516 else: 4517 self.raise_error("Expected END after CASE", self._prev) 4518 4519 return self._parse_window( 4520 self.expression(exp.Case, comments=comments, this=expression, ifs=ifs, default=default) 4521 ) 4522 4523 def _parse_if(self) -> t.Optional[exp.Expression]: 4524 if self._match(TokenType.L_PAREN): 4525 args = self._parse_csv(self._parse_conjunction) 4526 this = self.validate_expression(exp.If.from_arg_list(args), args) 4527 self._match_r_paren() 4528 else: 4529 index = self._index - 1 4530 4531 if self.NO_PAREN_IF_COMMANDS and index == 0: 4532 return self._parse_as_command(self._prev) 4533 4534 condition = self._parse_conjunction() 4535 4536 if not condition: 4537 self._retreat(index) 4538 return None 4539 4540 self._match(TokenType.THEN) 4541 true = self._parse_conjunction() 4542 false = self._parse_conjunction() if self._match(TokenType.ELSE) else None 4543 self._match(TokenType.END) 4544 this = self.expression(exp.If, this=condition, true=true, false=false) 4545 4546 return self._parse_window(this) 4547 4548 def _parse_next_value_for(self) -> t.Optional[exp.Expression]: 4549 if not self._match_text_seq("VALUE", "FOR"): 4550 self._retreat(self._index - 1) 4551 return None 4552 4553 return self.expression( 4554 exp.NextValueFor, 4555 this=self._parse_column(), 4556 order=self._match(TokenType.OVER) and self._parse_wrapped(self._parse_order), 4557 ) 4558 4559 def _parse_extract(self) -> exp.Extract: 4560 this = self._parse_function() or self._parse_var() or self._parse_type() 4561 4562 if self._match(TokenType.FROM): 4563 return self.expression(exp.Extract, this=this, expression=self._parse_bitwise()) 4564 4565 if not self._match(TokenType.COMMA): 4566 self.raise_error("Expected FROM or comma after EXTRACT", self._prev) 4567 4568 return self.expression(exp.Extract, this=this, expression=self._parse_bitwise()) 4569 4570 def _parse_cast(self, strict: bool, safe: t.Optional[bool] = None) -> exp.Expression: 4571 this = self._parse_conjunction() 4572 4573 if not self._match(TokenType.ALIAS): 4574 if self._match(TokenType.COMMA): 4575 return self.expression(exp.CastToStrType, this=this, to=self._parse_string()) 4576 4577 self.raise_error("Expected AS after CAST") 4578 4579 fmt = None 4580 to = self._parse_types() 4581 4582 if self._match(TokenType.FORMAT): 4583 fmt_string = self._parse_string() 4584 fmt = self._parse_at_time_zone(fmt_string) 4585 4586 if not to: 4587 to = exp.DataType.build(exp.DataType.Type.UNKNOWN) 4588 if to.this in exp.DataType.TEMPORAL_TYPES: 4589 this = self.expression( 4590 exp.StrToDate if to.this == exp.DataType.Type.DATE else exp.StrToTime, 4591 this=this, 4592 format=exp.Literal.string( 4593 format_time( 4594 fmt_string.this if fmt_string else "", 4595 self.dialect.FORMAT_MAPPING or self.dialect.TIME_MAPPING, 4596 self.dialect.FORMAT_TRIE or self.dialect.TIME_TRIE, 4597 ) 4598 ), 4599 ) 4600 4601 if isinstance(fmt, exp.AtTimeZone) and isinstance(this, exp.StrToTime): 4602 this.set("zone", fmt.args["zone"]) 4603 return this 4604 elif not to: 4605 self.raise_error("Expected TYPE after CAST") 4606 elif isinstance(to, exp.Identifier): 4607 to = exp.DataType.build(to.name, udt=True) 4608 elif to.this == exp.DataType.Type.CHAR: 4609 if self._match(TokenType.CHARACTER_SET): 4610 to = self.expression(exp.CharacterSet, this=self._parse_var_or_string()) 4611 4612 return self.expression( 4613 exp.Cast if strict else exp.TryCast, this=this, to=to, format=fmt, safe=safe 4614 ) 4615 4616 def _parse_string_agg(self) -> exp.Expression: 4617 if self._match(TokenType.DISTINCT): 4618 args: t.List[t.Optional[exp.Expression]] = [ 4619 self.expression(exp.Distinct, expressions=[self._parse_conjunction()]) 4620 ] 4621 if self._match(TokenType.COMMA): 4622 args.extend(self._parse_csv(self._parse_conjunction)) 4623 else: 4624 args = self._parse_csv(self._parse_conjunction) # type: ignore 4625 4626 index = self._index 4627 if not self._match(TokenType.R_PAREN) and args: 4628 # postgres: STRING_AGG([DISTINCT] expression, separator [ORDER BY expression1 {ASC | DESC} [, ...]]) 4629 # bigquery: STRING_AGG([DISTINCT] expression [, separator] [ORDER BY key [{ASC | DESC}] [, ... ]] [LIMIT n]) 4630 args[-1] = self._parse_limit(this=self._parse_order(this=args[-1])) 4631 return self.expression(exp.GroupConcat, this=args[0], separator=seq_get(args, 1)) 4632 4633 # Checks if we can parse an order clause: WITHIN GROUP (ORDER BY <order_by_expression_list> [ASC | DESC]). 4634 # This is done "manually", instead of letting _parse_window parse it into an exp.WithinGroup node, so that 4635 # the STRING_AGG call is parsed like in MySQL / SQLite and can thus be transpiled more easily to them. 4636 if not self._match_text_seq("WITHIN", "GROUP"): 4637 self._retreat(index) 4638 return self.validate_expression(exp.GroupConcat.from_arg_list(args), args) 4639 4640 self._match_l_paren() # The corresponding match_r_paren will be called in parse_function (caller) 4641 order = self._parse_order(this=seq_get(args, 0)) 4642 return self.expression(exp.GroupConcat, this=order, separator=seq_get(args, 1)) 4643 4644 def _parse_convert( 4645 self, strict: bool, safe: t.Optional[bool] = None 4646 ) -> t.Optional[exp.Expression]: 4647 this = self._parse_bitwise() 4648 4649 if self._match(TokenType.USING): 4650 to: t.Optional[exp.Expression] = self.expression( 4651 exp.CharacterSet, this=self._parse_var() 4652 ) 4653 elif self._match(TokenType.COMMA): 4654 to = self._parse_types() 4655 else: 4656 to = None 4657 4658 return self.expression(exp.Cast if strict else exp.TryCast, this=this, to=to, safe=safe) 4659 4660 def _parse_decode(self) -> t.Optional[exp.Decode | exp.Case]: 4661 """ 4662 There are generally two variants of the DECODE function: 4663 4664 - DECODE(bin, charset) 4665 - DECODE(expression, search, result [, search, result] ... [, default]) 4666 4667 The second variant will always be parsed into a CASE expression. Note that NULL 4668 needs special treatment, since we need to explicitly check for it with `IS NULL`, 4669 instead of relying on pattern matching. 4670 """ 4671 args = self._parse_csv(self._parse_conjunction) 4672 4673 if len(args) < 3: 4674 return self.expression(exp.Decode, this=seq_get(args, 0), charset=seq_get(args, 1)) 4675 4676 expression, *expressions = args 4677 if not expression: 4678 return None 4679 4680 ifs = [] 4681 for search, result in zip(expressions[::2], expressions[1::2]): 4682 if not search or not result: 4683 return None 4684 4685 if isinstance(search, exp.Literal): 4686 ifs.append( 4687 exp.If(this=exp.EQ(this=expression.copy(), expression=search), true=result) 4688 ) 4689 elif isinstance(search, exp.Null): 4690 ifs.append( 4691 exp.If(this=exp.Is(this=expression.copy(), expression=exp.Null()), true=result) 4692 ) 4693 else: 4694 cond = exp.or_( 4695 exp.EQ(this=expression.copy(), expression=search), 4696 exp.and_( 4697 exp.Is(this=expression.copy(), expression=exp.Null()), 4698 exp.Is(this=search.copy(), expression=exp.Null()), 4699 copy=False, 4700 ), 4701 copy=False, 4702 ) 4703 ifs.append(exp.If(this=cond, true=result)) 4704 4705 return exp.Case(ifs=ifs, default=expressions[-1] if len(expressions) % 2 == 1 else None) 4706 4707 def _parse_json_key_value(self) -> t.Optional[exp.JSONKeyValue]: 4708 self._match_text_seq("KEY") 4709 key = self._parse_column() 4710 self._match_set(self.JSON_KEY_VALUE_SEPARATOR_TOKENS) 4711 self._match_text_seq("VALUE") 4712 value = self._parse_bitwise() 4713 4714 if not key and not value: 4715 return None 4716 return self.expression(exp.JSONKeyValue, this=key, expression=value) 4717 4718 def _parse_format_json(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 4719 if not this or not self._match_text_seq("FORMAT", "JSON"): 4720 return this 4721 4722 return self.expression(exp.FormatJson, this=this) 4723 4724 def _parse_on_handling(self, on: str, *values: str) -> t.Optional[str]: 4725 # Parses the "X ON Y" syntax, i.e. NULL ON NULL (Oracle, T-SQL) 4726 for value in values: 4727 if self._match_text_seq(value, "ON", on): 4728 return f"{value} ON {on}" 4729 4730 return None 4731 4732 @t.overload 4733 def _parse_json_object(self, agg: Lit[False]) -> exp.JSONObject: 4734 ... 4735 4736 @t.overload 4737 def _parse_json_object(self, agg: Lit[True]) -> exp.JSONObjectAgg: 4738 ... 4739 4740 def _parse_json_object(self, agg=False): 4741 star = self._parse_star() 4742 expressions = ( 4743 [star] 4744 if star 4745 else self._parse_csv(lambda: self._parse_format_json(self._parse_json_key_value())) 4746 ) 4747 null_handling = self._parse_on_handling("NULL", "NULL", "ABSENT") 4748 4749 unique_keys = None 4750 if self._match_text_seq("WITH", "UNIQUE"): 4751 unique_keys = True 4752 elif self._match_text_seq("WITHOUT", "UNIQUE"): 4753 unique_keys = False 4754 4755 self._match_text_seq("KEYS") 4756 4757 return_type = self._match_text_seq("RETURNING") and self._parse_format_json( 4758 self._parse_type() 4759 ) 4760 encoding = self._match_text_seq("ENCODING") and self._parse_var() 4761 4762 return self.expression( 4763 exp.JSONObjectAgg if agg else exp.JSONObject, 4764 expressions=expressions, 4765 null_handling=null_handling, 4766 unique_keys=unique_keys, 4767 return_type=return_type, 4768 encoding=encoding, 4769 ) 4770 4771 # Note: this is currently incomplete; it only implements the "JSON_value_column" part 4772 def _parse_json_column_def(self) -> exp.JSONColumnDef: 4773 if not self._match_text_seq("NESTED"): 4774 this = self._parse_id_var() 4775 kind = self._parse_types(allow_identifiers=False) 4776 nested = None 4777 else: 4778 this = None 4779 kind = None 4780 nested = True 4781 4782 path = self._match_text_seq("PATH") and self._parse_string() 4783 nested_schema = nested and self._parse_json_schema() 4784 4785 return self.expression( 4786 exp.JSONColumnDef, 4787 this=this, 4788 kind=kind, 4789 path=path, 4790 nested_schema=nested_schema, 4791 ) 4792 4793 def _parse_json_schema(self) -> exp.JSONSchema: 4794 self._match_text_seq("COLUMNS") 4795 return self.expression( 4796 exp.JSONSchema, 4797 expressions=self._parse_wrapped_csv(self._parse_json_column_def, optional=True), 4798 ) 4799 4800 def _parse_json_table(self) -> exp.JSONTable: 4801 this = self._parse_format_json(self._parse_bitwise()) 4802 path = self._match(TokenType.COMMA) and self._parse_string() 4803 error_handling = self._parse_on_handling("ERROR", "ERROR", "NULL") 4804 empty_handling = self._parse_on_handling("EMPTY", "ERROR", "NULL") 4805 schema = self._parse_json_schema() 4806 4807 return exp.JSONTable( 4808 this=this, 4809 schema=schema, 4810 path=path, 4811 error_handling=error_handling, 4812 empty_handling=empty_handling, 4813 ) 4814 4815 def _parse_match_against(self) -> exp.MatchAgainst: 4816 expressions = self._parse_csv(self._parse_column) 4817 4818 self._match_text_seq(")", "AGAINST", "(") 4819 4820 this = self._parse_string() 4821 4822 if self._match_text_seq("IN", "NATURAL", "LANGUAGE", "MODE"): 4823 modifier = "IN NATURAL LANGUAGE MODE" 4824 if self._match_text_seq("WITH", "QUERY", "EXPANSION"): 4825 modifier = f"{modifier} WITH QUERY EXPANSION" 4826 elif self._match_text_seq("IN", "BOOLEAN", "MODE"): 4827 modifier = "IN BOOLEAN MODE" 4828 elif self._match_text_seq("WITH", "QUERY", "EXPANSION"): 4829 modifier = "WITH QUERY EXPANSION" 4830 else: 4831 modifier = None 4832 4833 return self.expression( 4834 exp.MatchAgainst, this=this, expressions=expressions, modifier=modifier 4835 ) 4836 4837 # https://learn.microsoft.com/en-us/sql/t-sql/functions/openjson-transact-sql?view=sql-server-ver16 4838 def _parse_open_json(self) -> exp.OpenJSON: 4839 this = self._parse_bitwise() 4840 path = self._match(TokenType.COMMA) and self._parse_string() 4841 4842 def _parse_open_json_column_def() -> exp.OpenJSONColumnDef: 4843 this = self._parse_field(any_token=True) 4844 kind = self._parse_types() 4845 path = self._parse_string() 4846 as_json = self._match_pair(TokenType.ALIAS, TokenType.JSON) 4847 4848 return self.expression( 4849 exp.OpenJSONColumnDef, this=this, kind=kind, path=path, as_json=as_json 4850 ) 4851 4852 expressions = None 4853 if self._match_pair(TokenType.R_PAREN, TokenType.WITH): 4854 self._match_l_paren() 4855 expressions = self._parse_csv(_parse_open_json_column_def) 4856 4857 return self.expression(exp.OpenJSON, this=this, path=path, expressions=expressions) 4858 4859 def _parse_position(self, haystack_first: bool = False) -> exp.StrPosition: 4860 args = self._parse_csv(self._parse_bitwise) 4861 4862 if self._match(TokenType.IN): 4863 return self.expression( 4864 exp.StrPosition, this=self._parse_bitwise(), substr=seq_get(args, 0) 4865 ) 4866 4867 if haystack_first: 4868 haystack = seq_get(args, 0) 4869 needle = seq_get(args, 1) 4870 else: 4871 needle = seq_get(args, 0) 4872 haystack = seq_get(args, 1) 4873 4874 return self.expression( 4875 exp.StrPosition, this=haystack, substr=needle, position=seq_get(args, 2) 4876 ) 4877 4878 def _parse_predict(self) -> exp.Predict: 4879 self._match_text_seq("MODEL") 4880 this = self._parse_table() 4881 4882 self._match(TokenType.COMMA) 4883 self._match_text_seq("TABLE") 4884 4885 return self.expression( 4886 exp.Predict, 4887 this=this, 4888 expression=self._parse_table(), 4889 params_struct=self._match(TokenType.COMMA) and self._parse_bitwise(), 4890 ) 4891 4892 def _parse_join_hint(self, func_name: str) -> exp.JoinHint: 4893 args = self._parse_csv(self._parse_table) 4894 return exp.JoinHint(this=func_name.upper(), expressions=args) 4895 4896 def _parse_substring(self) -> exp.Substring: 4897 # Postgres supports the form: substring(string [from int] [for int]) 4898 # https://www.postgresql.org/docs/9.1/functions-string.html @ Table 9-6 4899 4900 args = t.cast(t.List[t.Optional[exp.Expression]], self._parse_csv(self._parse_bitwise)) 4901 4902 if self._match(TokenType.FROM): 4903 args.append(self._parse_bitwise()) 4904 if self._match(TokenType.FOR): 4905 args.append(self._parse_bitwise()) 4906 4907 return self.validate_expression(exp.Substring.from_arg_list(args), args) 4908 4909 def _parse_trim(self) -> exp.Trim: 4910 # https://www.w3resource.com/sql/character-functions/trim.php 4911 # https://docs.oracle.com/javadb/10.8.3.0/ref/rreftrimfunc.html 4912 4913 position = None 4914 collation = None 4915 expression = None 4916 4917 if self._match_texts(self.TRIM_TYPES): 4918 position = self._prev.text.upper() 4919 4920 this = self._parse_bitwise() 4921 if self._match_set((TokenType.FROM, TokenType.COMMA)): 4922 invert_order = self._prev.token_type == TokenType.FROM or self.TRIM_PATTERN_FIRST 4923 expression = self._parse_bitwise() 4924 4925 if invert_order: 4926 this, expression = expression, this 4927 4928 if self._match(TokenType.COLLATE): 4929 collation = self._parse_bitwise() 4930 4931 return self.expression( 4932 exp.Trim, this=this, position=position, expression=expression, collation=collation 4933 ) 4934 4935 def _parse_window_clause(self) -> t.Optional[t.List[exp.Expression]]: 4936 return self._match(TokenType.WINDOW) and self._parse_csv(self._parse_named_window) 4937 4938 def _parse_named_window(self) -> t.Optional[exp.Expression]: 4939 return self._parse_window(self._parse_id_var(), alias=True) 4940 4941 def _parse_respect_or_ignore_nulls( 4942 self, this: t.Optional[exp.Expression] 4943 ) -> t.Optional[exp.Expression]: 4944 if self._match_text_seq("IGNORE", "NULLS"): 4945 return self.expression(exp.IgnoreNulls, this=this) 4946 if self._match_text_seq("RESPECT", "NULLS"): 4947 return self.expression(exp.RespectNulls, this=this) 4948 return this 4949 4950 def _parse_having_max(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 4951 if self._match(TokenType.HAVING): 4952 self._match_texts(("MAX", "MIN")) 4953 max = self._prev.text.upper() != "MIN" 4954 return self.expression( 4955 exp.HavingMax, this=this, expression=self._parse_column(), max=max 4956 ) 4957 4958 return this 4959 4960 def _parse_window( 4961 self, this: t.Optional[exp.Expression], alias: bool = False 4962 ) -> t.Optional[exp.Expression]: 4963 if self._match_pair(TokenType.FILTER, TokenType.L_PAREN): 4964 self._match(TokenType.WHERE) 4965 this = self.expression( 4966 exp.Filter, this=this, expression=self._parse_where(skip_where_token=True) 4967 ) 4968 self._match_r_paren() 4969 4970 # T-SQL allows the OVER (...) syntax after WITHIN GROUP. 4971 # https://learn.microsoft.com/en-us/sql/t-sql/functions/percentile-disc-transact-sql?view=sql-server-ver16 4972 if self._match_text_seq("WITHIN", "GROUP"): 4973 order = self._parse_wrapped(self._parse_order) 4974 this = self.expression(exp.WithinGroup, this=this, expression=order) 4975 4976 # SQL spec defines an optional [ { IGNORE | RESPECT } NULLS ] OVER 4977 # Some dialects choose to implement and some do not. 4978 # https://dev.mysql.com/doc/refman/8.0/en/window-function-descriptions.html 4979 4980 # There is some code above in _parse_lambda that handles 4981 # SELECT FIRST_VALUE(TABLE.COLUMN IGNORE|RESPECT NULLS) OVER ... 4982 4983 # The below changes handle 4984 # SELECT FIRST_VALUE(TABLE.COLUMN) IGNORE|RESPECT NULLS OVER ... 4985 4986 # Oracle allows both formats 4987 # (https://docs.oracle.com/en/database/oracle/oracle-database/19/sqlrf/img_text/first_value.html) 4988 # and Snowflake chose to do the same for familiarity 4989 # https://docs.snowflake.com/en/sql-reference/functions/first_value.html#usage-notes 4990 if isinstance(this, exp.AggFunc): 4991 ignore_respect = this.find(exp.IgnoreNulls, exp.RespectNulls) 4992 4993 if ignore_respect and ignore_respect is not this: 4994 ignore_respect.replace(ignore_respect.this) 4995 this = self.expression(ignore_respect.__class__, this=this) 4996 4997 this = self._parse_respect_or_ignore_nulls(this) 4998 4999 # bigquery select from window x AS (partition by ...) 5000 if alias: 5001 over = None 5002 self._match(TokenType.ALIAS) 5003 elif not self._match_set(self.WINDOW_BEFORE_PAREN_TOKENS): 5004 return this 5005 else: 5006 over = self._prev.text.upper() 5007 5008 if not self._match(TokenType.L_PAREN): 5009 return self.expression( 5010 exp.Window, this=this, alias=self._parse_id_var(False), over=over 5011 ) 5012 5013 window_alias = self._parse_id_var(any_token=False, tokens=self.WINDOW_ALIAS_TOKENS) 5014 5015 first = self._match(TokenType.FIRST) 5016 if self._match_text_seq("LAST"): 5017 first = False 5018 5019 partition, order = self._parse_partition_and_order() 5020 kind = self._match_set((TokenType.ROWS, TokenType.RANGE)) and self._prev.text 5021 5022 if kind: 5023 self._match(TokenType.BETWEEN) 5024 start = self._parse_window_spec() 5025 self._match(TokenType.AND) 5026 end = self._parse_window_spec() 5027 5028 spec = self.expression( 5029 exp.WindowSpec, 5030 kind=kind, 5031 start=start["value"], 5032 start_side=start["side"], 5033 end=end["value"], 5034 end_side=end["side"], 5035 ) 5036 else: 5037 spec = None 5038 5039 self._match_r_paren() 5040 5041 window = self.expression( 5042 exp.Window, 5043 this=this, 5044 partition_by=partition, 5045 order=order, 5046 spec=spec, 5047 alias=window_alias, 5048 over=over, 5049 first=first, 5050 ) 5051 5052 # This covers Oracle's FIRST/LAST syntax: aggregate KEEP (...) OVER (...) 5053 if self._match_set(self.WINDOW_BEFORE_PAREN_TOKENS, advance=False): 5054 return self._parse_window(window, alias=alias) 5055 5056 return window 5057 5058 def _parse_partition_and_order( 5059 self, 5060 ) -> t.Tuple[t.List[exp.Expression], t.Optional[exp.Expression]]: 5061 return self._parse_partition_by(), self._parse_order() 5062 5063 def _parse_window_spec(self) -> t.Dict[str, t.Optional[str | exp.Expression]]: 5064 self._match(TokenType.BETWEEN) 5065 5066 return { 5067 "value": ( 5068 (self._match_text_seq("UNBOUNDED") and "UNBOUNDED") 5069 or (self._match_text_seq("CURRENT", "ROW") and "CURRENT ROW") 5070 or self._parse_bitwise() 5071 ), 5072 "side": self._match_texts(self.WINDOW_SIDES) and self._prev.text, 5073 } 5074 5075 def _parse_alias( 5076 self, this: t.Optional[exp.Expression], explicit: bool = False 5077 ) -> t.Optional[exp.Expression]: 5078 any_token = self._match(TokenType.ALIAS) 5079 comments = self._prev_comments 5080 5081 if explicit and not any_token: 5082 return this 5083 5084 if self._match(TokenType.L_PAREN): 5085 aliases = self.expression( 5086 exp.Aliases, 5087 comments=comments, 5088 this=this, 5089 expressions=self._parse_csv(lambda: self._parse_id_var(any_token)), 5090 ) 5091 self._match_r_paren(aliases) 5092 return aliases 5093 5094 alias = self._parse_id_var(any_token) or ( 5095 self.STRING_ALIASES and self._parse_string_as_identifier() 5096 ) 5097 5098 if alias: 5099 this = self.expression(exp.Alias, comments=comments, this=this, alias=alias) 5100 column = this.this 5101 5102 # Moves the comment next to the alias in `expr /* comment */ AS alias` 5103 if not this.comments and column and column.comments: 5104 this.comments = column.comments 5105 column.comments = None 5106 5107 return this 5108 5109 def _parse_id_var( 5110 self, 5111 any_token: bool = True, 5112 tokens: t.Optional[t.Collection[TokenType]] = None, 5113 ) -> t.Optional[exp.Expression]: 5114 identifier = self._parse_identifier() 5115 5116 if identifier: 5117 return identifier 5118 5119 if (any_token and self._advance_any()) or self._match_set(tokens or self.ID_VAR_TOKENS): 5120 quoted = self._prev.token_type == TokenType.STRING 5121 return exp.Identifier(this=self._prev.text, quoted=quoted) 5122 5123 return None 5124 5125 def _parse_string(self) -> t.Optional[exp.Expression]: 5126 if self._match_set((TokenType.STRING, TokenType.RAW_STRING)): 5127 return self.PRIMARY_PARSERS[self._prev.token_type](self, self._prev) 5128 return self._parse_placeholder() 5129 5130 def _parse_string_as_identifier(self) -> t.Optional[exp.Identifier]: 5131 return exp.to_identifier(self._match(TokenType.STRING) and self._prev.text, quoted=True) 5132 5133 def _parse_number(self) -> t.Optional[exp.Expression]: 5134 if self._match(TokenType.NUMBER): 5135 return self.PRIMARY_PARSERS[TokenType.NUMBER](self, self._prev) 5136 return self._parse_placeholder() 5137 5138 def _parse_identifier(self) -> t.Optional[exp.Expression]: 5139 if self._match(TokenType.IDENTIFIER): 5140 return self.expression(exp.Identifier, this=self._prev.text, quoted=True) 5141 return self._parse_placeholder() 5142 5143 def _parse_var( 5144 self, 5145 any_token: bool = False, 5146 tokens: t.Optional[t.Collection[TokenType]] = None, 5147 upper: bool = False, 5148 ) -> t.Optional[exp.Expression]: 5149 if ( 5150 (any_token and self._advance_any()) 5151 or self._match(TokenType.VAR) 5152 or (self._match_set(tokens) if tokens else False) 5153 ): 5154 return self.expression( 5155 exp.Var, this=self._prev.text.upper() if upper else self._prev.text 5156 ) 5157 return self._parse_placeholder() 5158 5159 def _advance_any(self, ignore_reserved: bool = False) -> t.Optional[Token]: 5160 if self._curr and (ignore_reserved or self._curr.token_type not in self.RESERVED_TOKENS): 5161 self._advance() 5162 return self._prev 5163 return None 5164 5165 def _parse_var_or_string(self) -> t.Optional[exp.Expression]: 5166 return self._parse_var() or self._parse_string() 5167 5168 def _parse_null(self) -> t.Optional[exp.Expression]: 5169 if self._match_set(self.NULL_TOKENS): 5170 return self.PRIMARY_PARSERS[TokenType.NULL](self, self._prev) 5171 return self._parse_placeholder() 5172 5173 def _parse_boolean(self) -> t.Optional[exp.Expression]: 5174 if self._match(TokenType.TRUE): 5175 return self.PRIMARY_PARSERS[TokenType.TRUE](self, self._prev) 5176 if self._match(TokenType.FALSE): 5177 return self.PRIMARY_PARSERS[TokenType.FALSE](self, self._prev) 5178 return self._parse_placeholder() 5179 5180 def _parse_star(self) -> t.Optional[exp.Expression]: 5181 if self._match(TokenType.STAR): 5182 return self.PRIMARY_PARSERS[TokenType.STAR](self, self._prev) 5183 return self._parse_placeholder() 5184 5185 def _parse_parameter(self) -> exp.Parameter: 5186 def _parse_parameter_part() -> t.Optional[exp.Expression]: 5187 return ( 5188 self._parse_identifier() or self._parse_primary() or self._parse_var(any_token=True) 5189 ) 5190 5191 self._match(TokenType.L_BRACE) 5192 this = _parse_parameter_part() 5193 expression = self._match(TokenType.COLON) and _parse_parameter_part() 5194 self._match(TokenType.R_BRACE) 5195 5196 return self.expression(exp.Parameter, this=this, expression=expression) 5197 5198 def _parse_placeholder(self) -> t.Optional[exp.Expression]: 5199 if self._match_set(self.PLACEHOLDER_PARSERS): 5200 placeholder = self.PLACEHOLDER_PARSERS[self._prev.token_type](self) 5201 if placeholder: 5202 return placeholder 5203 self._advance(-1) 5204 return None 5205 5206 def _parse_except(self) -> t.Optional[t.List[exp.Expression]]: 5207 if not self._match(TokenType.EXCEPT): 5208 return None 5209 if self._match(TokenType.L_PAREN, advance=False): 5210 return self._parse_wrapped_csv(self._parse_column) 5211 5212 except_column = self._parse_column() 5213 return [except_column] if except_column else None 5214 5215 def _parse_replace(self) -> t.Optional[t.List[exp.Expression]]: 5216 if not self._match(TokenType.REPLACE): 5217 return None 5218 if self._match(TokenType.L_PAREN, advance=False): 5219 return self._parse_wrapped_csv(self._parse_expression) 5220 5221 replace_expression = self._parse_expression() 5222 return [replace_expression] if replace_expression else None 5223 5224 def _parse_csv( 5225 self, parse_method: t.Callable, sep: TokenType = TokenType.COMMA 5226 ) -> t.List[exp.Expression]: 5227 parse_result = parse_method() 5228 items = [parse_result] if parse_result is not None else [] 5229 5230 while self._match(sep): 5231 self._add_comments(parse_result) 5232 parse_result = parse_method() 5233 if parse_result is not None: 5234 items.append(parse_result) 5235 5236 return items 5237 5238 def _parse_tokens( 5239 self, parse_method: t.Callable, expressions: t.Dict 5240 ) -> t.Optional[exp.Expression]: 5241 this = parse_method() 5242 5243 while self._match_set(expressions): 5244 this = self.expression( 5245 expressions[self._prev.token_type], 5246 this=this, 5247 comments=self._prev_comments, 5248 expression=parse_method(), 5249 ) 5250 5251 return this 5252 5253 def _parse_wrapped_id_vars(self, optional: bool = False) -> t.List[exp.Expression]: 5254 return self._parse_wrapped_csv(self._parse_id_var, optional=optional) 5255 5256 def _parse_wrapped_csv( 5257 self, parse_method: t.Callable, sep: TokenType = TokenType.COMMA, optional: bool = False 5258 ) -> t.List[exp.Expression]: 5259 return self._parse_wrapped( 5260 lambda: self._parse_csv(parse_method, sep=sep), optional=optional 5261 ) 5262 5263 def _parse_wrapped(self, parse_method: t.Callable, optional: bool = False) -> t.Any: 5264 wrapped = self._match(TokenType.L_PAREN) 5265 if not wrapped and not optional: 5266 self.raise_error("Expecting (") 5267 parse_result = parse_method() 5268 if wrapped: 5269 self._match_r_paren() 5270 return parse_result 5271 5272 def _parse_expressions(self) -> t.List[exp.Expression]: 5273 return self._parse_csv(self._parse_expression) 5274 5275 def _parse_select_or_expression(self, alias: bool = False) -> t.Optional[exp.Expression]: 5276 return self._parse_select() or self._parse_set_operations( 5277 self._parse_expression() if alias else self._parse_conjunction() 5278 ) 5279 5280 def _parse_ddl_select(self) -> t.Optional[exp.Expression]: 5281 return self._parse_query_modifiers( 5282 self._parse_set_operations(self._parse_select(nested=True, parse_subquery_alias=False)) 5283 ) 5284 5285 def _parse_transaction(self) -> exp.Transaction | exp.Command: 5286 this = None 5287 if self._match_texts(self.TRANSACTION_KIND): 5288 this = self._prev.text 5289 5290 self._match_texts(("TRANSACTION", "WORK")) 5291 5292 modes = [] 5293 while True: 5294 mode = [] 5295 while self._match(TokenType.VAR): 5296 mode.append(self._prev.text) 5297 5298 if mode: 5299 modes.append(" ".join(mode)) 5300 if not self._match(TokenType.COMMA): 5301 break 5302 5303 return self.expression(exp.Transaction, this=this, modes=modes) 5304 5305 def _parse_commit_or_rollback(self) -> exp.Commit | exp.Rollback: 5306 chain = None 5307 savepoint = None 5308 is_rollback = self._prev.token_type == TokenType.ROLLBACK 5309 5310 self._match_texts(("TRANSACTION", "WORK")) 5311 5312 if self._match_text_seq("TO"): 5313 self._match_text_seq("SAVEPOINT") 5314 savepoint = self._parse_id_var() 5315 5316 if self._match(TokenType.AND): 5317 chain = not self._match_text_seq("NO") 5318 self._match_text_seq("CHAIN") 5319 5320 if is_rollback: 5321 return self.expression(exp.Rollback, savepoint=savepoint) 5322 5323 return self.expression(exp.Commit, chain=chain) 5324 5325 def _parse_refresh(self) -> exp.Refresh: 5326 self._match(TokenType.TABLE) 5327 return self.expression(exp.Refresh, this=self._parse_string() or self._parse_table()) 5328 5329 def _parse_add_column(self) -> t.Optional[exp.Expression]: 5330 if not self._match_text_seq("ADD"): 5331 return None 5332 5333 self._match(TokenType.COLUMN) 5334 exists_column = self._parse_exists(not_=True) 5335 expression = self._parse_field_def() 5336 5337 if expression: 5338 expression.set("exists", exists_column) 5339 5340 # https://docs.databricks.com/delta/update-schema.html#explicitly-update-schema-to-add-columns 5341 if self._match_texts(("FIRST", "AFTER")): 5342 position = self._prev.text 5343 column_position = self.expression( 5344 exp.ColumnPosition, this=self._parse_column(), position=position 5345 ) 5346 expression.set("position", column_position) 5347 5348 return expression 5349 5350 def _parse_drop_column(self) -> t.Optional[exp.Drop | exp.Command]: 5351 drop = self._match(TokenType.DROP) and self._parse_drop() 5352 if drop and not isinstance(drop, exp.Command): 5353 drop.set("kind", drop.args.get("kind", "COLUMN")) 5354 return drop 5355 5356 # https://docs.aws.amazon.com/athena/latest/ug/alter-table-drop-partition.html 5357 def _parse_drop_partition(self, exists: t.Optional[bool] = None) -> exp.DropPartition: 5358 return self.expression( 5359 exp.DropPartition, expressions=self._parse_csv(self._parse_partition), exists=exists 5360 ) 5361 5362 def _parse_add_constraint(self) -> exp.AddConstraint: 5363 this = None 5364 kind = self._prev.token_type 5365 5366 if kind == TokenType.CONSTRAINT: 5367 this = self._parse_id_var() 5368 5369 if self._match_text_seq("CHECK"): 5370 expression = self._parse_wrapped(self._parse_conjunction) 5371 enforced = self._match_text_seq("ENFORCED") or False 5372 5373 return self.expression( 5374 exp.AddConstraint, this=this, expression=expression, enforced=enforced 5375 ) 5376 5377 if kind == TokenType.FOREIGN_KEY or self._match(TokenType.FOREIGN_KEY): 5378 expression = self._parse_foreign_key() 5379 elif kind == TokenType.PRIMARY_KEY or self._match(TokenType.PRIMARY_KEY): 5380 expression = self._parse_primary_key() 5381 else: 5382 expression = None 5383 5384 return self.expression(exp.AddConstraint, this=this, expression=expression) 5385 5386 def _parse_alter_table_add(self) -> t.List[exp.Expression]: 5387 index = self._index - 1 5388 5389 if self._match_set(self.ADD_CONSTRAINT_TOKENS): 5390 return self._parse_csv(self._parse_add_constraint) 5391 5392 self._retreat(index) 5393 if not self.ALTER_TABLE_ADD_REQUIRED_FOR_EACH_COLUMN and self._match_text_seq("ADD"): 5394 return self._parse_wrapped_csv(self._parse_field_def, optional=True) 5395 return self._parse_wrapped_csv(self._parse_add_column, optional=True) 5396 5397 def _parse_alter_table_alter(self) -> exp.AlterColumn: 5398 self._match(TokenType.COLUMN) 5399 column = self._parse_field(any_token=True) 5400 5401 if self._match_pair(TokenType.DROP, TokenType.DEFAULT): 5402 return self.expression(exp.AlterColumn, this=column, drop=True) 5403 if self._match_pair(TokenType.SET, TokenType.DEFAULT): 5404 return self.expression(exp.AlterColumn, this=column, default=self._parse_conjunction()) 5405 if self._match(TokenType.COMMENT): 5406 return self.expression(exp.AlterColumn, this=column, comment=self._parse_string()) 5407 5408 self._match_text_seq("SET", "DATA") 5409 return self.expression( 5410 exp.AlterColumn, 5411 this=column, 5412 dtype=self._match_text_seq("TYPE") and self._parse_types(), 5413 collate=self._match(TokenType.COLLATE) and self._parse_term(), 5414 using=self._match(TokenType.USING) and self._parse_conjunction(), 5415 ) 5416 5417 def _parse_alter_table_drop(self) -> t.List[exp.Expression]: 5418 index = self._index - 1 5419 5420 partition_exists = self._parse_exists() 5421 if self._match(TokenType.PARTITION, advance=False): 5422 return self._parse_csv(lambda: self._parse_drop_partition(exists=partition_exists)) 5423 5424 self._retreat(index) 5425 return self._parse_csv(self._parse_drop_column) 5426 5427 def _parse_alter_table_rename(self) -> t.Optional[exp.RenameTable | exp.RenameColumn]: 5428 if self._match(TokenType.COLUMN): 5429 exists = self._parse_exists() 5430 old_column = self._parse_column() 5431 to = self._match_text_seq("TO") 5432 new_column = self._parse_column() 5433 5434 if old_column is None or to is None or new_column is None: 5435 return None 5436 5437 return self.expression(exp.RenameColumn, this=old_column, to=new_column, exists=exists) 5438 5439 self._match_text_seq("TO") 5440 return self.expression(exp.RenameTable, this=self._parse_table(schema=True)) 5441 5442 def _parse_alter(self) -> exp.AlterTable | exp.Command: 5443 start = self._prev 5444 5445 if not self._match(TokenType.TABLE): 5446 return self._parse_as_command(start) 5447 5448 exists = self._parse_exists() 5449 only = self._match_text_seq("ONLY") 5450 this = self._parse_table(schema=True) 5451 5452 if self._next: 5453 self._advance() 5454 5455 parser = self.ALTER_PARSERS.get(self._prev.text.upper()) if self._prev else None 5456 if parser: 5457 actions = ensure_list(parser(self)) 5458 5459 if not self._curr and actions: 5460 return self.expression( 5461 exp.AlterTable, 5462 this=this, 5463 exists=exists, 5464 actions=actions, 5465 only=only, 5466 ) 5467 5468 return self._parse_as_command(start) 5469 5470 def _parse_merge(self) -> exp.Merge: 5471 self._match(TokenType.INTO) 5472 target = self._parse_table() 5473 5474 if target and self._match(TokenType.ALIAS, advance=False): 5475 target.set("alias", self._parse_table_alias()) 5476 5477 self._match(TokenType.USING) 5478 using = self._parse_table() 5479 5480 self._match(TokenType.ON) 5481 on = self._parse_conjunction() 5482 5483 return self.expression( 5484 exp.Merge, 5485 this=target, 5486 using=using, 5487 on=on, 5488 expressions=self._parse_when_matched(), 5489 ) 5490 5491 def _parse_when_matched(self) -> t.List[exp.When]: 5492 whens = [] 5493 5494 while self._match(TokenType.WHEN): 5495 matched = not self._match(TokenType.NOT) 5496 self._match_text_seq("MATCHED") 5497 source = ( 5498 False 5499 if self._match_text_seq("BY", "TARGET") 5500 else self._match_text_seq("BY", "SOURCE") 5501 ) 5502 condition = self._parse_conjunction() if self._match(TokenType.AND) else None 5503 5504 self._match(TokenType.THEN) 5505 5506 if self._match(TokenType.INSERT): 5507 _this = self._parse_star() 5508 if _this: 5509 then: t.Optional[exp.Expression] = self.expression(exp.Insert, this=_this) 5510 else: 5511 then = self.expression( 5512 exp.Insert, 5513 this=self._parse_value(), 5514 expression=self._match(TokenType.VALUES) and self._parse_value(), 5515 ) 5516 elif self._match(TokenType.UPDATE): 5517 expressions = self._parse_star() 5518 if expressions: 5519 then = self.expression(exp.Update, expressions=expressions) 5520 else: 5521 then = self.expression( 5522 exp.Update, 5523 expressions=self._match(TokenType.SET) 5524 and self._parse_csv(self._parse_equality), 5525 ) 5526 elif self._match(TokenType.DELETE): 5527 then = self.expression(exp.Var, this=self._prev.text) 5528 else: 5529 then = None 5530 5531 whens.append( 5532 self.expression( 5533 exp.When, 5534 matched=matched, 5535 source=source, 5536 condition=condition, 5537 then=then, 5538 ) 5539 ) 5540 return whens 5541 5542 def _parse_show(self) -> t.Optional[exp.Expression]: 5543 parser = self._find_parser(self.SHOW_PARSERS, self.SHOW_TRIE) 5544 if parser: 5545 return parser(self) 5546 return self._parse_as_command(self._prev) 5547 5548 def _parse_set_item_assignment( 5549 self, kind: t.Optional[str] = None 5550 ) -> t.Optional[exp.Expression]: 5551 index = self._index 5552 5553 if kind in ("GLOBAL", "SESSION") and self._match_text_seq("TRANSACTION"): 5554 return self._parse_set_transaction(global_=kind == "GLOBAL") 5555 5556 left = self._parse_primary() or self._parse_id_var() 5557 assignment_delimiter = self._match_texts(("=", "TO")) 5558 5559 if not left or (self.SET_REQUIRES_ASSIGNMENT_DELIMITER and not assignment_delimiter): 5560 self._retreat(index) 5561 return None 5562 5563 right = self._parse_statement() or self._parse_id_var() 5564 this = self.expression(exp.EQ, this=left, expression=right) 5565 5566 return self.expression(exp.SetItem, this=this, kind=kind) 5567 5568 def _parse_set_transaction(self, global_: bool = False) -> exp.Expression: 5569 self._match_text_seq("TRANSACTION") 5570 characteristics = self._parse_csv( 5571 lambda: self._parse_var_from_options(self.TRANSACTION_CHARACTERISTICS) 5572 ) 5573 return self.expression( 5574 exp.SetItem, 5575 expressions=characteristics, 5576 kind="TRANSACTION", 5577 **{"global": global_}, # type: ignore 5578 ) 5579 5580 def _parse_set_item(self) -> t.Optional[exp.Expression]: 5581 parser = self._find_parser(self.SET_PARSERS, self.SET_TRIE) 5582 return parser(self) if parser else self._parse_set_item_assignment(kind=None) 5583 5584 def _parse_set(self, unset: bool = False, tag: bool = False) -> exp.Set | exp.Command: 5585 index = self._index 5586 set_ = self.expression( 5587 exp.Set, expressions=self._parse_csv(self._parse_set_item), unset=unset, tag=tag 5588 ) 5589 5590 if self._curr: 5591 self._retreat(index) 5592 return self._parse_as_command(self._prev) 5593 5594 return set_ 5595 5596 def _parse_var_from_options(self, options: t.Collection[str]) -> t.Optional[exp.Var]: 5597 for option in options: 5598 if self._match_text_seq(*option.split(" ")): 5599 return exp.var(option) 5600 return None 5601 5602 def _parse_as_command(self, start: Token) -> exp.Command: 5603 while self._curr: 5604 self._advance() 5605 text = self._find_sql(start, self._prev) 5606 size = len(start.text) 5607 self._warn_unsupported() 5608 return exp.Command(this=text[:size], expression=text[size:]) 5609 5610 def _parse_dict_property(self, this: str) -> exp.DictProperty: 5611 settings = [] 5612 5613 self._match_l_paren() 5614 kind = self._parse_id_var() 5615 5616 if self._match(TokenType.L_PAREN): 5617 while True: 5618 key = self._parse_id_var() 5619 value = self._parse_primary() 5620 5621 if not key and value is None: 5622 break 5623 settings.append(self.expression(exp.DictSubProperty, this=key, value=value)) 5624 self._match(TokenType.R_PAREN) 5625 5626 self._match_r_paren() 5627 5628 return self.expression( 5629 exp.DictProperty, 5630 this=this, 5631 kind=kind.this if kind else None, 5632 settings=settings, 5633 ) 5634 5635 def _parse_dict_range(self, this: str) -> exp.DictRange: 5636 self._match_l_paren() 5637 has_min = self._match_text_seq("MIN") 5638 if has_min: 5639 min = self._parse_var() or self._parse_primary() 5640 self._match_text_seq("MAX") 5641 max = self._parse_var() or self._parse_primary() 5642 else: 5643 max = self._parse_var() or self._parse_primary() 5644 min = exp.Literal.number(0) 5645 self._match_r_paren() 5646 return self.expression(exp.DictRange, this=this, min=min, max=max) 5647 5648 def _parse_comprehension( 5649 self, this: t.Optional[exp.Expression] 5650 ) -> t.Optional[exp.Comprehension]: 5651 index = self._index 5652 expression = self._parse_column() 5653 if not self._match(TokenType.IN): 5654 self._retreat(index - 1) 5655 return None 5656 iterator = self._parse_column() 5657 condition = self._parse_conjunction() if self._match_text_seq("IF") else None 5658 return self.expression( 5659 exp.Comprehension, 5660 this=this, 5661 expression=expression, 5662 iterator=iterator, 5663 condition=condition, 5664 ) 5665 5666 def _parse_heredoc(self) -> t.Optional[exp.Heredoc]: 5667 if self._match(TokenType.HEREDOC_STRING): 5668 return self.expression(exp.Heredoc, this=self._prev.text) 5669 5670 if not self._match_text_seq("$"): 5671 return None 5672 5673 tags = ["$"] 5674 tag_text = None 5675 5676 if self._is_connected(): 5677 self._advance() 5678 tags.append(self._prev.text.upper()) 5679 else: 5680 self.raise_error("No closing $ found") 5681 5682 if tags[-1] != "$": 5683 if self._is_connected() and self._match_text_seq("$"): 5684 tag_text = tags[-1] 5685 tags.append("$") 5686 else: 5687 self.raise_error("No closing $ found") 5688 5689 heredoc_start = self._curr 5690 5691 while self._curr: 5692 if self._match_text_seq(*tags, advance=False): 5693 this = self._find_sql(heredoc_start, self._prev) 5694 self._advance(len(tags)) 5695 return self.expression(exp.Heredoc, this=this, tag=tag_text) 5696 5697 self._advance() 5698 5699 self.raise_error(f"No closing {''.join(tags)} found") 5700 return None 5701 5702 def _find_parser( 5703 self, parsers: t.Dict[str, t.Callable], trie: t.Dict 5704 ) -> t.Optional[t.Callable]: 5705 if not self._curr: 5706 return None 5707 5708 index = self._index 5709 this = [] 5710 while True: 5711 # The current token might be multiple words 5712 curr = self._curr.text.upper() 5713 key = curr.split(" ") 5714 this.append(curr) 5715 5716 self._advance() 5717 result, trie = in_trie(trie, key) 5718 if result == TrieResult.FAILED: 5719 break 5720 5721 if result == TrieResult.EXISTS: 5722 subparser = parsers[" ".join(this)] 5723 return subparser 5724 5725 self._retreat(index) 5726 return None 5727 5728 def _match(self, token_type, advance=True, expression=None): 5729 if not self._curr: 5730 return None 5731 5732 if self._curr.token_type == token_type: 5733 if advance: 5734 self._advance() 5735 self._add_comments(expression) 5736 return True 5737 5738 return None 5739 5740 def _match_set(self, types, advance=True): 5741 if not self._curr: 5742 return None 5743 5744 if self._curr.token_type in types: 5745 if advance: 5746 self._advance() 5747 return True 5748 5749 return None 5750 5751 def _match_pair(self, token_type_a, token_type_b, advance=True): 5752 if not self._curr or not self._next: 5753 return None 5754 5755 if self._curr.token_type == token_type_a and self._next.token_type == token_type_b: 5756 if advance: 5757 self._advance(2) 5758 return True 5759 5760 return None 5761 5762 def _match_l_paren(self, expression: t.Optional[exp.Expression] = None) -> None: 5763 if not self._match(TokenType.L_PAREN, expression=expression): 5764 self.raise_error("Expecting (") 5765 5766 def _match_r_paren(self, expression: t.Optional[exp.Expression] = None) -> None: 5767 if not self._match(TokenType.R_PAREN, expression=expression): 5768 self.raise_error("Expecting )") 5769 5770 def _match_texts(self, texts, advance=True): 5771 if self._curr and self._curr.text.upper() in texts: 5772 if advance: 5773 self._advance() 5774 return True 5775 return None 5776 5777 def _match_text_seq(self, *texts, advance=True): 5778 index = self._index 5779 for text in texts: 5780 if self._curr and self._curr.text.upper() == text: 5781 self._advance() 5782 else: 5783 self._retreat(index) 5784 return None 5785 5786 if not advance: 5787 self._retreat(index) 5788 5789 return True 5790 5791 @t.overload 5792 def _replace_columns_with_dots(self, this: exp.Expression) -> exp.Expression: 5793 ... 5794 5795 @t.overload 5796 def _replace_columns_with_dots( 5797 self, this: t.Optional[exp.Expression] 5798 ) -> t.Optional[exp.Expression]: 5799 ... 5800 5801 def _replace_columns_with_dots(self, this): 5802 if isinstance(this, exp.Dot): 5803 exp.replace_children(this, self._replace_columns_with_dots) 5804 elif isinstance(this, exp.Column): 5805 exp.replace_children(this, self._replace_columns_with_dots) 5806 table = this.args.get("table") 5807 this = ( 5808 self.expression(exp.Dot, this=table, expression=this.this) if table else this.this 5809 ) 5810 5811 return this 5812 5813 def _replace_lambda( 5814 self, node: t.Optional[exp.Expression], lambda_variables: t.Set[str] 5815 ) -> t.Optional[exp.Expression]: 5816 if not node: 5817 return node 5818 5819 for column in node.find_all(exp.Column): 5820 if column.parts[0].name in lambda_variables: 5821 dot_or_id = column.to_dot() if column.table else column.this 5822 parent = column.parent 5823 5824 while isinstance(parent, exp.Dot): 5825 if not isinstance(parent.parent, exp.Dot): 5826 parent.replace(dot_or_id) 5827 break 5828 parent = parent.parent 5829 else: 5830 if column is node: 5831 node = dot_or_id 5832 else: 5833 column.replace(dot_or_id) 5834 return node
22def parse_var_map(args: t.List) -> exp.StarMap | exp.VarMap: 23 if len(args) == 1 and args[0].is_star: 24 return exp.StarMap(this=args[0]) 25 26 keys = [] 27 values = [] 28 for i in range(0, len(args), 2): 29 keys.append(args[i]) 30 values.append(args[i + 1]) 31 32 return exp.VarMap( 33 keys=exp.Array(expressions=keys), 34 values=exp.Array(expressions=values), 35 )
51def parse_logarithm(args: t.List, dialect: Dialect) -> exp.Func: 52 # Default argument order is base, expression 53 this = seq_get(args, 0) 54 expression = seq_get(args, 1) 55 56 if expression: 57 if not dialect.LOG_BASE_FIRST: 58 this, expression = expression, this 59 return exp.Log(this=this, expression=expression) 60 61 return (exp.Ln if dialect.parser_class.LOG_DEFAULTS_TO_LN else exp.Log)(this=this)
64def parse_extract_json_with_path(expr_type: t.Type[E]) -> t.Callable[[t.List, Dialect], E]: 65 def _parser(args: t.List, dialect: Dialect) -> E: 66 expression = expr_type( 67 this=seq_get(args, 0), expression=dialect.to_json_path(seq_get(args, 1)) 68 ) 69 if len(args) > 2 and expr_type is exp.JSONExtract: 70 expression.set("expressions", args[2:]) 71 72 return expression 73 74 return _parser
87class Parser(metaclass=_Parser): 88 """ 89 Parser consumes a list of tokens produced by the Tokenizer and produces a parsed syntax tree. 90 91 Args: 92 error_level: The desired error level. 93 Default: ErrorLevel.IMMEDIATE 94 error_message_context: Determines the amount of context to capture from a 95 query string when displaying the error message (in number of characters). 96 Default: 100 97 max_errors: Maximum number of error messages to include in a raised ParseError. 98 This is only relevant if error_level is ErrorLevel.RAISE. 99 Default: 3 100 """ 101 102 FUNCTIONS: t.Dict[str, t.Callable] = { 103 **{name: func.from_arg_list for name, func in exp.FUNCTION_BY_NAME.items()}, 104 "CONCAT": lambda args, dialect: exp.Concat( 105 expressions=args, 106 safe=not dialect.STRICT_STRING_CONCAT, 107 coalesce=dialect.CONCAT_COALESCE, 108 ), 109 "CONCAT_WS": lambda args, dialect: exp.ConcatWs( 110 expressions=args, 111 safe=not dialect.STRICT_STRING_CONCAT, 112 coalesce=dialect.CONCAT_COALESCE, 113 ), 114 "DATE_TO_DATE_STR": lambda args: exp.Cast( 115 this=seq_get(args, 0), 116 to=exp.DataType(this=exp.DataType.Type.TEXT), 117 ), 118 "GLOB": lambda args: exp.Glob(this=seq_get(args, 1), expression=seq_get(args, 0)), 119 "JSON_EXTRACT": parse_extract_json_with_path(exp.JSONExtract), 120 "JSON_EXTRACT_SCALAR": parse_extract_json_with_path(exp.JSONExtractScalar), 121 "JSON_EXTRACT_PATH_TEXT": parse_extract_json_with_path(exp.JSONExtractScalar), 122 "LIKE": parse_like, 123 "LOG": parse_logarithm, 124 "TIME_TO_TIME_STR": lambda args: exp.Cast( 125 this=seq_get(args, 0), 126 to=exp.DataType(this=exp.DataType.Type.TEXT), 127 ), 128 "TS_OR_DS_TO_DATE_STR": lambda args: exp.Substring( 129 this=exp.Cast( 130 this=seq_get(args, 0), 131 to=exp.DataType(this=exp.DataType.Type.TEXT), 132 ), 133 start=exp.Literal.number(1), 134 length=exp.Literal.number(10), 135 ), 136 "VAR_MAP": parse_var_map, 137 } 138 139 NO_PAREN_FUNCTIONS = { 140 TokenType.CURRENT_DATE: exp.CurrentDate, 141 TokenType.CURRENT_DATETIME: exp.CurrentDate, 142 TokenType.CURRENT_TIME: exp.CurrentTime, 143 TokenType.CURRENT_TIMESTAMP: exp.CurrentTimestamp, 144 TokenType.CURRENT_USER: exp.CurrentUser, 145 } 146 147 STRUCT_TYPE_TOKENS = { 148 TokenType.NESTED, 149 TokenType.STRUCT, 150 } 151 152 NESTED_TYPE_TOKENS = { 153 TokenType.ARRAY, 154 TokenType.LOWCARDINALITY, 155 TokenType.MAP, 156 TokenType.NULLABLE, 157 *STRUCT_TYPE_TOKENS, 158 } 159 160 ENUM_TYPE_TOKENS = { 161 TokenType.ENUM, 162 TokenType.ENUM8, 163 TokenType.ENUM16, 164 } 165 166 AGGREGATE_TYPE_TOKENS = { 167 TokenType.AGGREGATEFUNCTION, 168 TokenType.SIMPLEAGGREGATEFUNCTION, 169 } 170 171 TYPE_TOKENS = { 172 TokenType.BIT, 173 TokenType.BOOLEAN, 174 TokenType.TINYINT, 175 TokenType.UTINYINT, 176 TokenType.SMALLINT, 177 TokenType.USMALLINT, 178 TokenType.INT, 179 TokenType.UINT, 180 TokenType.BIGINT, 181 TokenType.UBIGINT, 182 TokenType.INT128, 183 TokenType.UINT128, 184 TokenType.INT256, 185 TokenType.UINT256, 186 TokenType.MEDIUMINT, 187 TokenType.UMEDIUMINT, 188 TokenType.FIXEDSTRING, 189 TokenType.FLOAT, 190 TokenType.DOUBLE, 191 TokenType.CHAR, 192 TokenType.NCHAR, 193 TokenType.VARCHAR, 194 TokenType.NVARCHAR, 195 TokenType.BPCHAR, 196 TokenType.TEXT, 197 TokenType.MEDIUMTEXT, 198 TokenType.LONGTEXT, 199 TokenType.MEDIUMBLOB, 200 TokenType.LONGBLOB, 201 TokenType.BINARY, 202 TokenType.VARBINARY, 203 TokenType.JSON, 204 TokenType.JSONB, 205 TokenType.INTERVAL, 206 TokenType.TINYBLOB, 207 TokenType.TINYTEXT, 208 TokenType.TIME, 209 TokenType.TIMETZ, 210 TokenType.TIMESTAMP, 211 TokenType.TIMESTAMP_S, 212 TokenType.TIMESTAMP_MS, 213 TokenType.TIMESTAMP_NS, 214 TokenType.TIMESTAMPTZ, 215 TokenType.TIMESTAMPLTZ, 216 TokenType.DATETIME, 217 TokenType.DATETIME64, 218 TokenType.DATE, 219 TokenType.DATE32, 220 TokenType.INT4RANGE, 221 TokenType.INT4MULTIRANGE, 222 TokenType.INT8RANGE, 223 TokenType.INT8MULTIRANGE, 224 TokenType.NUMRANGE, 225 TokenType.NUMMULTIRANGE, 226 TokenType.TSRANGE, 227 TokenType.TSMULTIRANGE, 228 TokenType.TSTZRANGE, 229 TokenType.TSTZMULTIRANGE, 230 TokenType.DATERANGE, 231 TokenType.DATEMULTIRANGE, 232 TokenType.DECIMAL, 233 TokenType.UDECIMAL, 234 TokenType.BIGDECIMAL, 235 TokenType.UUID, 236 TokenType.GEOGRAPHY, 237 TokenType.GEOMETRY, 238 TokenType.HLLSKETCH, 239 TokenType.HSTORE, 240 TokenType.PSEUDO_TYPE, 241 TokenType.SUPER, 242 TokenType.SERIAL, 243 TokenType.SMALLSERIAL, 244 TokenType.BIGSERIAL, 245 TokenType.XML, 246 TokenType.YEAR, 247 TokenType.UNIQUEIDENTIFIER, 248 TokenType.USERDEFINED, 249 TokenType.MONEY, 250 TokenType.SMALLMONEY, 251 TokenType.ROWVERSION, 252 TokenType.IMAGE, 253 TokenType.VARIANT, 254 TokenType.OBJECT, 255 TokenType.OBJECT_IDENTIFIER, 256 TokenType.INET, 257 TokenType.IPADDRESS, 258 TokenType.IPPREFIX, 259 TokenType.IPV4, 260 TokenType.IPV6, 261 TokenType.UNKNOWN, 262 TokenType.NULL, 263 *ENUM_TYPE_TOKENS, 264 *NESTED_TYPE_TOKENS, 265 *AGGREGATE_TYPE_TOKENS, 266 } 267 268 SIGNED_TO_UNSIGNED_TYPE_TOKEN = { 269 TokenType.BIGINT: TokenType.UBIGINT, 270 TokenType.INT: TokenType.UINT, 271 TokenType.MEDIUMINT: TokenType.UMEDIUMINT, 272 TokenType.SMALLINT: TokenType.USMALLINT, 273 TokenType.TINYINT: TokenType.UTINYINT, 274 TokenType.DECIMAL: TokenType.UDECIMAL, 275 } 276 277 SUBQUERY_PREDICATES = { 278 TokenType.ANY: exp.Any, 279 TokenType.ALL: exp.All, 280 TokenType.EXISTS: exp.Exists, 281 TokenType.SOME: exp.Any, 282 } 283 284 RESERVED_TOKENS = { 285 *Tokenizer.SINGLE_TOKENS.values(), 286 TokenType.SELECT, 287 } 288 289 DB_CREATABLES = { 290 TokenType.DATABASE, 291 TokenType.SCHEMA, 292 TokenType.TABLE, 293 TokenType.VIEW, 294 TokenType.MODEL, 295 TokenType.DICTIONARY, 296 } 297 298 CREATABLES = { 299 TokenType.COLUMN, 300 TokenType.CONSTRAINT, 301 TokenType.FUNCTION, 302 TokenType.INDEX, 303 TokenType.PROCEDURE, 304 TokenType.FOREIGN_KEY, 305 *DB_CREATABLES, 306 } 307 308 # Tokens that can represent identifiers 309 ID_VAR_TOKENS = { 310 TokenType.VAR, 311 TokenType.ANTI, 312 TokenType.APPLY, 313 TokenType.ASC, 314 TokenType.AUTO_INCREMENT, 315 TokenType.BEGIN, 316 TokenType.BPCHAR, 317 TokenType.CACHE, 318 TokenType.CASE, 319 TokenType.COLLATE, 320 TokenType.COMMAND, 321 TokenType.COMMENT, 322 TokenType.COMMIT, 323 TokenType.CONSTRAINT, 324 TokenType.DEFAULT, 325 TokenType.DELETE, 326 TokenType.DESC, 327 TokenType.DESCRIBE, 328 TokenType.DICTIONARY, 329 TokenType.DIV, 330 TokenType.END, 331 TokenType.EXECUTE, 332 TokenType.ESCAPE, 333 TokenType.FALSE, 334 TokenType.FIRST, 335 TokenType.FILTER, 336 TokenType.FINAL, 337 TokenType.FORMAT, 338 TokenType.FULL, 339 TokenType.IS, 340 TokenType.ISNULL, 341 TokenType.INTERVAL, 342 TokenType.KEEP, 343 TokenType.KILL, 344 TokenType.LEFT, 345 TokenType.LOAD, 346 TokenType.MERGE, 347 TokenType.NATURAL, 348 TokenType.NEXT, 349 TokenType.OFFSET, 350 TokenType.OPERATOR, 351 TokenType.ORDINALITY, 352 TokenType.OVERLAPS, 353 TokenType.OVERWRITE, 354 TokenType.PARTITION, 355 TokenType.PERCENT, 356 TokenType.PIVOT, 357 TokenType.PRAGMA, 358 TokenType.RANGE, 359 TokenType.RECURSIVE, 360 TokenType.REFERENCES, 361 TokenType.REFRESH, 362 TokenType.REPLACE, 363 TokenType.RIGHT, 364 TokenType.ROW, 365 TokenType.ROWS, 366 TokenType.SEMI, 367 TokenType.SET, 368 TokenType.SETTINGS, 369 TokenType.SHOW, 370 TokenType.TEMPORARY, 371 TokenType.TOP, 372 TokenType.TRUE, 373 TokenType.UNIQUE, 374 TokenType.UNPIVOT, 375 TokenType.UPDATE, 376 TokenType.USE, 377 TokenType.VOLATILE, 378 TokenType.WINDOW, 379 *CREATABLES, 380 *SUBQUERY_PREDICATES, 381 *TYPE_TOKENS, 382 *NO_PAREN_FUNCTIONS, 383 } 384 385 INTERVAL_VARS = ID_VAR_TOKENS - {TokenType.END} 386 387 TABLE_ALIAS_TOKENS = ID_VAR_TOKENS - { 388 TokenType.ANTI, 389 TokenType.APPLY, 390 TokenType.ASOF, 391 TokenType.FULL, 392 TokenType.LEFT, 393 TokenType.LOCK, 394 TokenType.NATURAL, 395 TokenType.OFFSET, 396 TokenType.RIGHT, 397 TokenType.SEMI, 398 TokenType.WINDOW, 399 } 400 401 COMMENT_TABLE_ALIAS_TOKENS = TABLE_ALIAS_TOKENS - {TokenType.IS} 402 403 UPDATE_ALIAS_TOKENS = TABLE_ALIAS_TOKENS - {TokenType.SET} 404 405 TRIM_TYPES = {"LEADING", "TRAILING", "BOTH"} 406 407 FUNC_TOKENS = { 408 TokenType.COLLATE, 409 TokenType.COMMAND, 410 TokenType.CURRENT_DATE, 411 TokenType.CURRENT_DATETIME, 412 TokenType.CURRENT_TIMESTAMP, 413 TokenType.CURRENT_TIME, 414 TokenType.CURRENT_USER, 415 TokenType.FILTER, 416 TokenType.FIRST, 417 TokenType.FORMAT, 418 TokenType.GLOB, 419 TokenType.IDENTIFIER, 420 TokenType.INDEX, 421 TokenType.ISNULL, 422 TokenType.ILIKE, 423 TokenType.INSERT, 424 TokenType.LIKE, 425 TokenType.MERGE, 426 TokenType.OFFSET, 427 TokenType.PRIMARY_KEY, 428 TokenType.RANGE, 429 TokenType.REPLACE, 430 TokenType.RLIKE, 431 TokenType.ROW, 432 TokenType.UNNEST, 433 TokenType.VAR, 434 TokenType.LEFT, 435 TokenType.RIGHT, 436 TokenType.DATE, 437 TokenType.DATETIME, 438 TokenType.TABLE, 439 TokenType.TIMESTAMP, 440 TokenType.TIMESTAMPTZ, 441 TokenType.WINDOW, 442 TokenType.XOR, 443 *TYPE_TOKENS, 444 *SUBQUERY_PREDICATES, 445 } 446 447 CONJUNCTION = { 448 TokenType.AND: exp.And, 449 TokenType.OR: exp.Or, 450 } 451 452 EQUALITY = { 453 TokenType.COLON_EQ: exp.PropertyEQ, 454 TokenType.EQ: exp.EQ, 455 TokenType.NEQ: exp.NEQ, 456 TokenType.NULLSAFE_EQ: exp.NullSafeEQ, 457 } 458 459 COMPARISON = { 460 TokenType.GT: exp.GT, 461 TokenType.GTE: exp.GTE, 462 TokenType.LT: exp.LT, 463 TokenType.LTE: exp.LTE, 464 } 465 466 BITWISE = { 467 TokenType.AMP: exp.BitwiseAnd, 468 TokenType.CARET: exp.BitwiseXor, 469 TokenType.PIPE: exp.BitwiseOr, 470 } 471 472 TERM = { 473 TokenType.DASH: exp.Sub, 474 TokenType.PLUS: exp.Add, 475 TokenType.MOD: exp.Mod, 476 TokenType.COLLATE: exp.Collate, 477 } 478 479 FACTOR = { 480 TokenType.DIV: exp.IntDiv, 481 TokenType.LR_ARROW: exp.Distance, 482 TokenType.SLASH: exp.Div, 483 TokenType.STAR: exp.Mul, 484 } 485 486 EXPONENT: t.Dict[TokenType, t.Type[exp.Expression]] = {} 487 488 TIMES = { 489 TokenType.TIME, 490 TokenType.TIMETZ, 491 } 492 493 TIMESTAMPS = { 494 TokenType.TIMESTAMP, 495 TokenType.TIMESTAMPTZ, 496 TokenType.TIMESTAMPLTZ, 497 *TIMES, 498 } 499 500 SET_OPERATIONS = { 501 TokenType.UNION, 502 TokenType.INTERSECT, 503 TokenType.EXCEPT, 504 } 505 506 JOIN_METHODS = { 507 TokenType.NATURAL, 508 TokenType.ASOF, 509 } 510 511 JOIN_SIDES = { 512 TokenType.LEFT, 513 TokenType.RIGHT, 514 TokenType.FULL, 515 } 516 517 JOIN_KINDS = { 518 TokenType.INNER, 519 TokenType.OUTER, 520 TokenType.CROSS, 521 TokenType.SEMI, 522 TokenType.ANTI, 523 } 524 525 JOIN_HINTS: t.Set[str] = set() 526 527 LAMBDAS = { 528 TokenType.ARROW: lambda self, expressions: self.expression( 529 exp.Lambda, 530 this=self._replace_lambda( 531 self._parse_conjunction(), 532 {node.name for node in expressions}, 533 ), 534 expressions=expressions, 535 ), 536 TokenType.FARROW: lambda self, expressions: self.expression( 537 exp.Kwarg, 538 this=exp.var(expressions[0].name), 539 expression=self._parse_conjunction(), 540 ), 541 } 542 543 COLUMN_OPERATORS = { 544 TokenType.DOT: None, 545 TokenType.DCOLON: lambda self, this, to: self.expression( 546 exp.Cast if self.STRICT_CAST else exp.TryCast, 547 this=this, 548 to=to, 549 ), 550 TokenType.ARROW: lambda self, this, path: self.expression( 551 exp.JSONExtract, 552 this=this, 553 expression=self.dialect.to_json_path(path), 554 ), 555 TokenType.DARROW: lambda self, this, path: self.expression( 556 exp.JSONExtractScalar, 557 this=this, 558 expression=self.dialect.to_json_path(path), 559 ), 560 TokenType.HASH_ARROW: lambda self, this, path: self.expression( 561 exp.JSONBExtract, 562 this=this, 563 expression=path, 564 ), 565 TokenType.DHASH_ARROW: lambda self, this, path: self.expression( 566 exp.JSONBExtractScalar, 567 this=this, 568 expression=path, 569 ), 570 TokenType.PLACEHOLDER: lambda self, this, key: self.expression( 571 exp.JSONBContains, 572 this=this, 573 expression=key, 574 ), 575 } 576 577 EXPRESSION_PARSERS = { 578 exp.Cluster: lambda self: self._parse_sort(exp.Cluster, TokenType.CLUSTER_BY), 579 exp.Column: lambda self: self._parse_column(), 580 exp.Condition: lambda self: self._parse_conjunction(), 581 exp.DataType: lambda self: self._parse_types(allow_identifiers=False), 582 exp.Expression: lambda self: self._parse_statement(), 583 exp.From: lambda self: self._parse_from(), 584 exp.Group: lambda self: self._parse_group(), 585 exp.Having: lambda self: self._parse_having(), 586 exp.Identifier: lambda self: self._parse_id_var(), 587 exp.Join: lambda self: self._parse_join(), 588 exp.Lambda: lambda self: self._parse_lambda(), 589 exp.Lateral: lambda self: self._parse_lateral(), 590 exp.Limit: lambda self: self._parse_limit(), 591 exp.Offset: lambda self: self._parse_offset(), 592 exp.Order: lambda self: self._parse_order(), 593 exp.Ordered: lambda self: self._parse_ordered(), 594 exp.Properties: lambda self: self._parse_properties(), 595 exp.Qualify: lambda self: self._parse_qualify(), 596 exp.Returning: lambda self: self._parse_returning(), 597 exp.Sort: lambda self: self._parse_sort(exp.Sort, TokenType.SORT_BY), 598 exp.Table: lambda self: self._parse_table_parts(), 599 exp.TableAlias: lambda self: self._parse_table_alias(), 600 exp.When: lambda self: seq_get(self._parse_when_matched(), 0), 601 exp.Where: lambda self: self._parse_where(), 602 exp.Window: lambda self: self._parse_named_window(), 603 exp.With: lambda self: self._parse_with(), 604 "JOIN_TYPE": lambda self: self._parse_join_parts(), 605 } 606 607 STATEMENT_PARSERS = { 608 TokenType.ALTER: lambda self: self._parse_alter(), 609 TokenType.BEGIN: lambda self: self._parse_transaction(), 610 TokenType.CACHE: lambda self: self._parse_cache(), 611 TokenType.COMMIT: lambda self: self._parse_commit_or_rollback(), 612 TokenType.COMMENT: lambda self: self._parse_comment(), 613 TokenType.CREATE: lambda self: self._parse_create(), 614 TokenType.DELETE: lambda self: self._parse_delete(), 615 TokenType.DESC: lambda self: self._parse_describe(), 616 TokenType.DESCRIBE: lambda self: self._parse_describe(), 617 TokenType.DROP: lambda self: self._parse_drop(), 618 TokenType.INSERT: lambda self: self._parse_insert(), 619 TokenType.KILL: lambda self: self._parse_kill(), 620 TokenType.LOAD: lambda self: self._parse_load(), 621 TokenType.MERGE: lambda self: self._parse_merge(), 622 TokenType.PIVOT: lambda self: self._parse_simplified_pivot(), 623 TokenType.PRAGMA: lambda self: self.expression(exp.Pragma, this=self._parse_expression()), 624 TokenType.REFRESH: lambda self: self._parse_refresh(), 625 TokenType.ROLLBACK: lambda self: self._parse_commit_or_rollback(), 626 TokenType.SET: lambda self: self._parse_set(), 627 TokenType.UNCACHE: lambda self: self._parse_uncache(), 628 TokenType.UPDATE: lambda self: self._parse_update(), 629 TokenType.USE: lambda self: self.expression( 630 exp.Use, 631 kind=self._match_texts(("ROLE", "WAREHOUSE", "DATABASE", "SCHEMA")) 632 and exp.var(self._prev.text), 633 this=self._parse_table(schema=False), 634 ), 635 } 636 637 UNARY_PARSERS = { 638 TokenType.PLUS: lambda self: self._parse_unary(), # Unary + is handled as a no-op 639 TokenType.NOT: lambda self: self.expression(exp.Not, this=self._parse_equality()), 640 TokenType.TILDA: lambda self: self.expression(exp.BitwiseNot, this=self._parse_unary()), 641 TokenType.DASH: lambda self: self.expression(exp.Neg, this=self._parse_unary()), 642 } 643 644 PRIMARY_PARSERS = { 645 TokenType.STRING: lambda self, token: self.expression( 646 exp.Literal, this=token.text, is_string=True 647 ), 648 TokenType.NUMBER: lambda self, token: self.expression( 649 exp.Literal, this=token.text, is_string=False 650 ), 651 TokenType.STAR: lambda self, _: self.expression( 652 exp.Star, **{"except": self._parse_except(), "replace": self._parse_replace()} 653 ), 654 TokenType.NULL: lambda self, _: self.expression(exp.Null), 655 TokenType.TRUE: lambda self, _: self.expression(exp.Boolean, this=True), 656 TokenType.FALSE: lambda self, _: self.expression(exp.Boolean, this=False), 657 TokenType.BIT_STRING: lambda self, token: self.expression(exp.BitString, this=token.text), 658 TokenType.HEX_STRING: lambda self, token: self.expression(exp.HexString, this=token.text), 659 TokenType.BYTE_STRING: lambda self, token: self.expression(exp.ByteString, this=token.text), 660 TokenType.INTRODUCER: lambda self, token: self._parse_introducer(token), 661 TokenType.NATIONAL_STRING: lambda self, token: self.expression( 662 exp.National, this=token.text 663 ), 664 TokenType.RAW_STRING: lambda self, token: self.expression(exp.RawString, this=token.text), 665 TokenType.HEREDOC_STRING: lambda self, token: self.expression( 666 exp.RawString, this=token.text 667 ), 668 TokenType.UNICODE_STRING: lambda self, token: self.expression( 669 exp.UnicodeString, 670 this=token.text, 671 escape=self._match_text_seq("UESCAPE") and self._parse_string(), 672 ), 673 TokenType.SESSION_PARAMETER: lambda self, _: self._parse_session_parameter(), 674 } 675 676 PLACEHOLDER_PARSERS = { 677 TokenType.PLACEHOLDER: lambda self: self.expression(exp.Placeholder), 678 TokenType.PARAMETER: lambda self: self._parse_parameter(), 679 TokenType.COLON: lambda self: ( 680 self.expression(exp.Placeholder, this=self._prev.text) 681 if self._match(TokenType.NUMBER) or self._match_set(self.ID_VAR_TOKENS) 682 else None 683 ), 684 } 685 686 RANGE_PARSERS = { 687 TokenType.BETWEEN: lambda self, this: self._parse_between(this), 688 TokenType.GLOB: binary_range_parser(exp.Glob), 689 TokenType.ILIKE: binary_range_parser(exp.ILike), 690 TokenType.IN: lambda self, this: self._parse_in(this), 691 TokenType.IRLIKE: binary_range_parser(exp.RegexpILike), 692 TokenType.IS: lambda self, this: self._parse_is(this), 693 TokenType.LIKE: binary_range_parser(exp.Like), 694 TokenType.OVERLAPS: binary_range_parser(exp.Overlaps), 695 TokenType.RLIKE: binary_range_parser(exp.RegexpLike), 696 TokenType.SIMILAR_TO: binary_range_parser(exp.SimilarTo), 697 TokenType.FOR: lambda self, this: self._parse_comprehension(this), 698 } 699 700 PROPERTY_PARSERS: t.Dict[str, t.Callable] = { 701 "ALGORITHM": lambda self: self._parse_property_assignment(exp.AlgorithmProperty), 702 "AUTO": lambda self: self._parse_auto_property(), 703 "AUTO_INCREMENT": lambda self: self._parse_property_assignment(exp.AutoIncrementProperty), 704 "BLOCKCOMPRESSION": lambda self: self._parse_blockcompression(), 705 "CHARSET": lambda self, **kwargs: self._parse_character_set(**kwargs), 706 "CHARACTER SET": lambda self, **kwargs: self._parse_character_set(**kwargs), 707 "CHECKSUM": lambda self: self._parse_checksum(), 708 "CLUSTER BY": lambda self: self._parse_cluster(), 709 "CLUSTERED": lambda self: self._parse_clustered_by(), 710 "COLLATE": lambda self, **kwargs: self._parse_property_assignment( 711 exp.CollateProperty, **kwargs 712 ), 713 "COMMENT": lambda self: self._parse_property_assignment(exp.SchemaCommentProperty), 714 "CONTAINS": lambda self: self._parse_contains_property(), 715 "COPY": lambda self: self._parse_copy_property(), 716 "DATABLOCKSIZE": lambda self, **kwargs: self._parse_datablocksize(**kwargs), 717 "DEFINER": lambda self: self._parse_definer(), 718 "DETERMINISTIC": lambda self: self.expression( 719 exp.StabilityProperty, this=exp.Literal.string("IMMUTABLE") 720 ), 721 "DISTKEY": lambda self: self._parse_distkey(), 722 "DISTSTYLE": lambda self: self._parse_property_assignment(exp.DistStyleProperty), 723 "ENGINE": lambda self: self._parse_property_assignment(exp.EngineProperty), 724 "EXECUTE": lambda self: self._parse_property_assignment(exp.ExecuteAsProperty), 725 "EXTERNAL": lambda self: self.expression(exp.ExternalProperty), 726 "FALLBACK": lambda self, **kwargs: self._parse_fallback(**kwargs), 727 "FORMAT": lambda self: self._parse_property_assignment(exp.FileFormatProperty), 728 "FREESPACE": lambda self: self._parse_freespace(), 729 "HEAP": lambda self: self.expression(exp.HeapProperty), 730 "IMMUTABLE": lambda self: self.expression( 731 exp.StabilityProperty, this=exp.Literal.string("IMMUTABLE") 732 ), 733 "INHERITS": lambda self: self.expression( 734 exp.InheritsProperty, expressions=self._parse_wrapped_csv(self._parse_table) 735 ), 736 "INPUT": lambda self: self.expression(exp.InputModelProperty, this=self._parse_schema()), 737 "JOURNAL": lambda self, **kwargs: self._parse_journal(**kwargs), 738 "LANGUAGE": lambda self: self._parse_property_assignment(exp.LanguageProperty), 739 "LAYOUT": lambda self: self._parse_dict_property(this="LAYOUT"), 740 "LIFETIME": lambda self: self._parse_dict_range(this="LIFETIME"), 741 "LIKE": lambda self: self._parse_create_like(), 742 "LOCATION": lambda self: self._parse_property_assignment(exp.LocationProperty), 743 "LOCK": lambda self: self._parse_locking(), 744 "LOCKING": lambda self: self._parse_locking(), 745 "LOG": lambda self, **kwargs: self._parse_log(**kwargs), 746 "MATERIALIZED": lambda self: self.expression(exp.MaterializedProperty), 747 "MERGEBLOCKRATIO": lambda self, **kwargs: self._parse_mergeblockratio(**kwargs), 748 "MODIFIES": lambda self: self._parse_modifies_property(), 749 "MULTISET": lambda self: self.expression(exp.SetProperty, multi=True), 750 "NO": lambda self: self._parse_no_property(), 751 "ON": lambda self: self._parse_on_property(), 752 "ORDER BY": lambda self: self._parse_order(skip_order_token=True), 753 "OUTPUT": lambda self: self.expression(exp.OutputModelProperty, this=self._parse_schema()), 754 "PARTITION": lambda self: self._parse_partitioned_of(), 755 "PARTITION BY": lambda self: self._parse_partitioned_by(), 756 "PARTITIONED BY": lambda self: self._parse_partitioned_by(), 757 "PARTITIONED_BY": lambda self: self._parse_partitioned_by(), 758 "PRIMARY KEY": lambda self: self._parse_primary_key(in_props=True), 759 "RANGE": lambda self: self._parse_dict_range(this="RANGE"), 760 "READS": lambda self: self._parse_reads_property(), 761 "REMOTE": lambda self: self._parse_remote_with_connection(), 762 "RETURNS": lambda self: self._parse_returns(), 763 "ROW": lambda self: self._parse_row(), 764 "ROW_FORMAT": lambda self: self._parse_property_assignment(exp.RowFormatProperty), 765 "SAMPLE": lambda self: self.expression( 766 exp.SampleProperty, this=self._match_text_seq("BY") and self._parse_bitwise() 767 ), 768 "SET": lambda self: self.expression(exp.SetProperty, multi=False), 769 "SETTINGS": lambda self: self.expression( 770 exp.SettingsProperty, expressions=self._parse_csv(self._parse_set_item) 771 ), 772 "SORTKEY": lambda self: self._parse_sortkey(), 773 "SOURCE": lambda self: self._parse_dict_property(this="SOURCE"), 774 "STABLE": lambda self: self.expression( 775 exp.StabilityProperty, this=exp.Literal.string("STABLE") 776 ), 777 "STORED": lambda self: self._parse_stored(), 778 "SYSTEM_VERSIONING": lambda self: self._parse_system_versioning_property(), 779 "TBLPROPERTIES": lambda self: self._parse_wrapped_csv(self._parse_property), 780 "TEMP": lambda self: self.expression(exp.TemporaryProperty), 781 "TEMPORARY": lambda self: self.expression(exp.TemporaryProperty), 782 "TO": lambda self: self._parse_to_table(), 783 "TRANSIENT": lambda self: self.expression(exp.TransientProperty), 784 "TRANSFORM": lambda self: self.expression( 785 exp.TransformModelProperty, expressions=self._parse_wrapped_csv(self._parse_expression) 786 ), 787 "TTL": lambda self: self._parse_ttl(), 788 "USING": lambda self: self._parse_property_assignment(exp.FileFormatProperty), 789 "VOLATILE": lambda self: self._parse_volatile_property(), 790 "WITH": lambda self: self._parse_with_property(), 791 } 792 793 CONSTRAINT_PARSERS = { 794 "AUTOINCREMENT": lambda self: self._parse_auto_increment(), 795 "AUTO_INCREMENT": lambda self: self._parse_auto_increment(), 796 "CASESPECIFIC": lambda self: self.expression(exp.CaseSpecificColumnConstraint, not_=False), 797 "CHARACTER SET": lambda self: self.expression( 798 exp.CharacterSetColumnConstraint, this=self._parse_var_or_string() 799 ), 800 "CHECK": lambda self: self.expression( 801 exp.CheckColumnConstraint, this=self._parse_wrapped(self._parse_conjunction) 802 ), 803 "COLLATE": lambda self: self.expression( 804 exp.CollateColumnConstraint, this=self._parse_var() 805 ), 806 "COMMENT": lambda self: self.expression( 807 exp.CommentColumnConstraint, this=self._parse_string() 808 ), 809 "COMPRESS": lambda self: self._parse_compress(), 810 "CLUSTERED": lambda self: self.expression( 811 exp.ClusteredColumnConstraint, this=self._parse_wrapped_csv(self._parse_ordered) 812 ), 813 "NONCLUSTERED": lambda self: self.expression( 814 exp.NonClusteredColumnConstraint, this=self._parse_wrapped_csv(self._parse_ordered) 815 ), 816 "DEFAULT": lambda self: self.expression( 817 exp.DefaultColumnConstraint, this=self._parse_bitwise() 818 ), 819 "ENCODE": lambda self: self.expression(exp.EncodeColumnConstraint, this=self._parse_var()), 820 "FOREIGN KEY": lambda self: self._parse_foreign_key(), 821 "FORMAT": lambda self: self.expression( 822 exp.DateFormatColumnConstraint, this=self._parse_var_or_string() 823 ), 824 "GENERATED": lambda self: self._parse_generated_as_identity(), 825 "IDENTITY": lambda self: self._parse_auto_increment(), 826 "INLINE": lambda self: self._parse_inline(), 827 "LIKE": lambda self: self._parse_create_like(), 828 "NOT": lambda self: self._parse_not_constraint(), 829 "NULL": lambda self: self.expression(exp.NotNullColumnConstraint, allow_null=True), 830 "ON": lambda self: ( 831 self._match(TokenType.UPDATE) 832 and self.expression(exp.OnUpdateColumnConstraint, this=self._parse_function()) 833 ) 834 or self.expression(exp.OnProperty, this=self._parse_id_var()), 835 "PATH": lambda self: self.expression(exp.PathColumnConstraint, this=self._parse_string()), 836 "PERIOD": lambda self: self._parse_period_for_system_time(), 837 "PRIMARY KEY": lambda self: self._parse_primary_key(), 838 "REFERENCES": lambda self: self._parse_references(match=False), 839 "TITLE": lambda self: self.expression( 840 exp.TitleColumnConstraint, this=self._parse_var_or_string() 841 ), 842 "TTL": lambda self: self.expression(exp.MergeTreeTTL, expressions=[self._parse_bitwise()]), 843 "UNIQUE": lambda self: self._parse_unique(), 844 "UPPERCASE": lambda self: self.expression(exp.UppercaseColumnConstraint), 845 "WITH": lambda self: self.expression( 846 exp.Properties, expressions=self._parse_wrapped_csv(self._parse_property) 847 ), 848 } 849 850 ALTER_PARSERS = { 851 "ADD": lambda self: self._parse_alter_table_add(), 852 "ALTER": lambda self: self._parse_alter_table_alter(), 853 "CLUSTER BY": lambda self: self._parse_cluster(wrapped=True), 854 "DELETE": lambda self: self.expression(exp.Delete, where=self._parse_where()), 855 "DROP": lambda self: self._parse_alter_table_drop(), 856 "RENAME": lambda self: self._parse_alter_table_rename(), 857 } 858 859 SCHEMA_UNNAMED_CONSTRAINTS = {"CHECK", "FOREIGN KEY", "LIKE", "PRIMARY KEY", "UNIQUE", "PERIOD"} 860 861 NO_PAREN_FUNCTION_PARSERS = { 862 "ANY": lambda self: self.expression(exp.Any, this=self._parse_bitwise()), 863 "CASE": lambda self: self._parse_case(), 864 "IF": lambda self: self._parse_if(), 865 "NEXT": lambda self: self._parse_next_value_for(), 866 } 867 868 INVALID_FUNC_NAME_TOKENS = { 869 TokenType.IDENTIFIER, 870 TokenType.STRING, 871 } 872 873 FUNCTIONS_WITH_ALIASED_ARGS = {"STRUCT"} 874 875 FUNCTION_PARSERS = { 876 "CAST": lambda self: self._parse_cast(self.STRICT_CAST), 877 "CONVERT": lambda self: self._parse_convert(self.STRICT_CAST), 878 "DECODE": lambda self: self._parse_decode(), 879 "EXTRACT": lambda self: self._parse_extract(), 880 "JSON_OBJECT": lambda self: self._parse_json_object(), 881 "JSON_OBJECTAGG": lambda self: self._parse_json_object(agg=True), 882 "JSON_TABLE": lambda self: self._parse_json_table(), 883 "MATCH": lambda self: self._parse_match_against(), 884 "OPENJSON": lambda self: self._parse_open_json(), 885 "POSITION": lambda self: self._parse_position(), 886 "PREDICT": lambda self: self._parse_predict(), 887 "SAFE_CAST": lambda self: self._parse_cast(False, safe=True), 888 "STRING_AGG": lambda self: self._parse_string_agg(), 889 "SUBSTRING": lambda self: self._parse_substring(), 890 "TRIM": lambda self: self._parse_trim(), 891 "TRY_CAST": lambda self: self._parse_cast(False, safe=True), 892 "TRY_CONVERT": lambda self: self._parse_convert(False, safe=True), 893 } 894 895 QUERY_MODIFIER_PARSERS = { 896 TokenType.MATCH_RECOGNIZE: lambda self: ("match", self._parse_match_recognize()), 897 TokenType.WHERE: lambda self: ("where", self._parse_where()), 898 TokenType.GROUP_BY: lambda self: ("group", self._parse_group()), 899 TokenType.HAVING: lambda self: ("having", self._parse_having()), 900 TokenType.QUALIFY: lambda self: ("qualify", self._parse_qualify()), 901 TokenType.WINDOW: lambda self: ("windows", self._parse_window_clause()), 902 TokenType.ORDER_BY: lambda self: ("order", self._parse_order()), 903 TokenType.LIMIT: lambda self: ("limit", self._parse_limit()), 904 TokenType.FETCH: lambda self: ("limit", self._parse_limit()), 905 TokenType.OFFSET: lambda self: ("offset", self._parse_offset()), 906 TokenType.FOR: lambda self: ("locks", self._parse_locks()), 907 TokenType.LOCK: lambda self: ("locks", self._parse_locks()), 908 TokenType.TABLE_SAMPLE: lambda self: ("sample", self._parse_table_sample(as_modifier=True)), 909 TokenType.USING: lambda self: ("sample", self._parse_table_sample(as_modifier=True)), 910 TokenType.CLUSTER_BY: lambda self: ( 911 "cluster", 912 self._parse_sort(exp.Cluster, TokenType.CLUSTER_BY), 913 ), 914 TokenType.DISTRIBUTE_BY: lambda self: ( 915 "distribute", 916 self._parse_sort(exp.Distribute, TokenType.DISTRIBUTE_BY), 917 ), 918 TokenType.SORT_BY: lambda self: ("sort", self._parse_sort(exp.Sort, TokenType.SORT_BY)), 919 TokenType.CONNECT_BY: lambda self: ("connect", self._parse_connect(skip_start_token=True)), 920 TokenType.START_WITH: lambda self: ("connect", self._parse_connect()), 921 } 922 923 SET_PARSERS = { 924 "GLOBAL": lambda self: self._parse_set_item_assignment("GLOBAL"), 925 "LOCAL": lambda self: self._parse_set_item_assignment("LOCAL"), 926 "SESSION": lambda self: self._parse_set_item_assignment("SESSION"), 927 "TRANSACTION": lambda self: self._parse_set_transaction(), 928 } 929 930 SHOW_PARSERS: t.Dict[str, t.Callable] = {} 931 932 TYPE_LITERAL_PARSERS = { 933 exp.DataType.Type.JSON: lambda self, this, _: self.expression(exp.ParseJSON, this=this), 934 } 935 936 MODIFIABLES = (exp.Subquery, exp.Subqueryable, exp.Table) 937 938 DDL_SELECT_TOKENS = {TokenType.SELECT, TokenType.WITH, TokenType.L_PAREN} 939 940 PRE_VOLATILE_TOKENS = {TokenType.CREATE, TokenType.REPLACE, TokenType.UNIQUE} 941 942 TRANSACTION_KIND = {"DEFERRED", "IMMEDIATE", "EXCLUSIVE"} 943 TRANSACTION_CHARACTERISTICS = { 944 "ISOLATION LEVEL REPEATABLE READ", 945 "ISOLATION LEVEL READ COMMITTED", 946 "ISOLATION LEVEL READ UNCOMMITTED", 947 "ISOLATION LEVEL SERIALIZABLE", 948 "READ WRITE", 949 "READ ONLY", 950 } 951 952 INSERT_ALTERNATIVES = {"ABORT", "FAIL", "IGNORE", "REPLACE", "ROLLBACK"} 953 954 CLONE_KEYWORDS = {"CLONE", "COPY"} 955 HISTORICAL_DATA_KIND = {"TIMESTAMP", "OFFSET", "STATEMENT", "STREAM"} 956 957 OPCLASS_FOLLOW_KEYWORDS = {"ASC", "DESC", "NULLS"} 958 OPTYPE_FOLLOW_TOKENS = {TokenType.COMMA, TokenType.R_PAREN} 959 960 TABLE_INDEX_HINT_TOKENS = {TokenType.FORCE, TokenType.IGNORE, TokenType.USE} 961 962 WINDOW_ALIAS_TOKENS = ID_VAR_TOKENS - {TokenType.ROWS} 963 WINDOW_BEFORE_PAREN_TOKENS = {TokenType.OVER} 964 WINDOW_SIDES = {"FOLLOWING", "PRECEDING"} 965 966 JSON_KEY_VALUE_SEPARATOR_TOKENS = {TokenType.COLON, TokenType.COMMA, TokenType.IS} 967 968 FETCH_TOKENS = ID_VAR_TOKENS - {TokenType.ROW, TokenType.ROWS, TokenType.PERCENT} 969 970 ADD_CONSTRAINT_TOKENS = {TokenType.CONSTRAINT, TokenType.PRIMARY_KEY, TokenType.FOREIGN_KEY} 971 972 DISTINCT_TOKENS = {TokenType.DISTINCT} 973 974 NULL_TOKENS = {TokenType.NULL} 975 976 UNNEST_OFFSET_ALIAS_TOKENS = ID_VAR_TOKENS - SET_OPERATIONS 977 978 STRICT_CAST = True 979 980 PREFIXED_PIVOT_COLUMNS = False 981 IDENTIFY_PIVOT_STRINGS = False 982 983 LOG_DEFAULTS_TO_LN = False 984 985 # Whether or not ADD is present for each column added by ALTER TABLE 986 ALTER_TABLE_ADD_REQUIRED_FOR_EACH_COLUMN = True 987 988 # Whether or not the table sample clause expects CSV syntax 989 TABLESAMPLE_CSV = False 990 991 # Whether or not the SET command needs a delimiter (e.g. "=") for assignments 992 SET_REQUIRES_ASSIGNMENT_DELIMITER = True 993 994 # Whether the TRIM function expects the characters to trim as its first argument 995 TRIM_PATTERN_FIRST = False 996 997 # Whether or not string aliases are supported `SELECT COUNT(*) 'count'` 998 STRING_ALIASES = False 999 1000 # Whether query modifiers such as LIMIT are attached to the UNION node (vs its right operand) 1001 MODIFIERS_ATTACHED_TO_UNION = True 1002 UNION_MODIFIERS = {"order", "limit", "offset"} 1003 1004 # parses no parenthesis if statements as commands 1005 NO_PAREN_IF_COMMANDS = True 1006 1007 __slots__ = ( 1008 "error_level", 1009 "error_message_context", 1010 "max_errors", 1011 "dialect", 1012 "sql", 1013 "errors", 1014 "_tokens", 1015 "_index", 1016 "_curr", 1017 "_next", 1018 "_prev", 1019 "_prev_comments", 1020 ) 1021 1022 # Autofilled 1023 SHOW_TRIE: t.Dict = {} 1024 SET_TRIE: t.Dict = {} 1025 1026 def __init__( 1027 self, 1028 error_level: t.Optional[ErrorLevel] = None, 1029 error_message_context: int = 100, 1030 max_errors: int = 3, 1031 dialect: DialectType = None, 1032 ): 1033 from sqlglot.dialects import Dialect 1034 1035 self.error_level = error_level or ErrorLevel.IMMEDIATE 1036 self.error_message_context = error_message_context 1037 self.max_errors = max_errors 1038 self.dialect = Dialect.get_or_raise(dialect) 1039 self.reset() 1040 1041 def reset(self): 1042 self.sql = "" 1043 self.errors = [] 1044 self._tokens = [] 1045 self._index = 0 1046 self._curr = None 1047 self._next = None 1048 self._prev = None 1049 self._prev_comments = None 1050 1051 def parse( 1052 self, raw_tokens: t.List[Token], sql: t.Optional[str] = None 1053 ) -> t.List[t.Optional[exp.Expression]]: 1054 """ 1055 Parses a list of tokens and returns a list of syntax trees, one tree 1056 per parsed SQL statement. 1057 1058 Args: 1059 raw_tokens: The list of tokens. 1060 sql: The original SQL string, used to produce helpful debug messages. 1061 1062 Returns: 1063 The list of the produced syntax trees. 1064 """ 1065 return self._parse( 1066 parse_method=self.__class__._parse_statement, raw_tokens=raw_tokens, sql=sql 1067 ) 1068 1069 def parse_into( 1070 self, 1071 expression_types: exp.IntoType, 1072 raw_tokens: t.List[Token], 1073 sql: t.Optional[str] = None, 1074 ) -> t.List[t.Optional[exp.Expression]]: 1075 """ 1076 Parses a list of tokens into a given Expression type. If a collection of Expression 1077 types is given instead, this method will try to parse the token list into each one 1078 of them, stopping at the first for which the parsing succeeds. 1079 1080 Args: 1081 expression_types: The expression type(s) to try and parse the token list into. 1082 raw_tokens: The list of tokens. 1083 sql: The original SQL string, used to produce helpful debug messages. 1084 1085 Returns: 1086 The target Expression. 1087 """ 1088 errors = [] 1089 for expression_type in ensure_list(expression_types): 1090 parser = self.EXPRESSION_PARSERS.get(expression_type) 1091 if not parser: 1092 raise TypeError(f"No parser registered for {expression_type}") 1093 1094 try: 1095 return self._parse(parser, raw_tokens, sql) 1096 except ParseError as e: 1097 e.errors[0]["into_expression"] = expression_type 1098 errors.append(e) 1099 1100 raise ParseError( 1101 f"Failed to parse '{sql or raw_tokens}' into {expression_types}", 1102 errors=merge_errors(errors), 1103 ) from errors[-1] 1104 1105 def _parse( 1106 self, 1107 parse_method: t.Callable[[Parser], t.Optional[exp.Expression]], 1108 raw_tokens: t.List[Token], 1109 sql: t.Optional[str] = None, 1110 ) -> t.List[t.Optional[exp.Expression]]: 1111 self.reset() 1112 self.sql = sql or "" 1113 1114 total = len(raw_tokens) 1115 chunks: t.List[t.List[Token]] = [[]] 1116 1117 for i, token in enumerate(raw_tokens): 1118 if token.token_type == TokenType.SEMICOLON: 1119 if i < total - 1: 1120 chunks.append([]) 1121 else: 1122 chunks[-1].append(token) 1123 1124 expressions = [] 1125 1126 for tokens in chunks: 1127 self._index = -1 1128 self._tokens = tokens 1129 self._advance() 1130 1131 expressions.append(parse_method(self)) 1132 1133 if self._index < len(self._tokens): 1134 self.raise_error("Invalid expression / Unexpected token") 1135 1136 self.check_errors() 1137 1138 return expressions 1139 1140 def check_errors(self) -> None: 1141 """Logs or raises any found errors, depending on the chosen error level setting.""" 1142 if self.error_level == ErrorLevel.WARN: 1143 for error in self.errors: 1144 logger.error(str(error)) 1145 elif self.error_level == ErrorLevel.RAISE and self.errors: 1146 raise ParseError( 1147 concat_messages(self.errors, self.max_errors), 1148 errors=merge_errors(self.errors), 1149 ) 1150 1151 def raise_error(self, message: str, token: t.Optional[Token] = None) -> None: 1152 """ 1153 Appends an error in the list of recorded errors or raises it, depending on the chosen 1154 error level setting. 1155 """ 1156 token = token or self._curr or self._prev or Token.string("") 1157 start = token.start 1158 end = token.end + 1 1159 start_context = self.sql[max(start - self.error_message_context, 0) : start] 1160 highlight = self.sql[start:end] 1161 end_context = self.sql[end : end + self.error_message_context] 1162 1163 error = ParseError.new( 1164 f"{message}. Line {token.line}, Col: {token.col}.\n" 1165 f" {start_context}\033[4m{highlight}\033[0m{end_context}", 1166 description=message, 1167 line=token.line, 1168 col=token.col, 1169 start_context=start_context, 1170 highlight=highlight, 1171 end_context=end_context, 1172 ) 1173 1174 if self.error_level == ErrorLevel.IMMEDIATE: 1175 raise error 1176 1177 self.errors.append(error) 1178 1179 def expression( 1180 self, exp_class: t.Type[E], comments: t.Optional[t.List[str]] = None, **kwargs 1181 ) -> E: 1182 """ 1183 Creates a new, validated Expression. 1184 1185 Args: 1186 exp_class: The expression class to instantiate. 1187 comments: An optional list of comments to attach to the expression. 1188 kwargs: The arguments to set for the expression along with their respective values. 1189 1190 Returns: 1191 The target expression. 1192 """ 1193 instance = exp_class(**kwargs) 1194 instance.add_comments(comments) if comments else self._add_comments(instance) 1195 return self.validate_expression(instance) 1196 1197 def _add_comments(self, expression: t.Optional[exp.Expression]) -> None: 1198 if expression and self._prev_comments: 1199 expression.add_comments(self._prev_comments) 1200 self._prev_comments = None 1201 1202 def validate_expression(self, expression: E, args: t.Optional[t.List] = None) -> E: 1203 """ 1204 Validates an Expression, making sure that all its mandatory arguments are set. 1205 1206 Args: 1207 expression: The expression to validate. 1208 args: An optional list of items that was used to instantiate the expression, if it's a Func. 1209 1210 Returns: 1211 The validated expression. 1212 """ 1213 if self.error_level != ErrorLevel.IGNORE: 1214 for error_message in expression.error_messages(args): 1215 self.raise_error(error_message) 1216 1217 return expression 1218 1219 def _find_sql(self, start: Token, end: Token) -> str: 1220 return self.sql[start.start : end.end + 1] 1221 1222 def _is_connected(self) -> bool: 1223 return self._prev and self._curr and self._prev.end + 1 == self._curr.start 1224 1225 def _advance(self, times: int = 1) -> None: 1226 self._index += times 1227 self._curr = seq_get(self._tokens, self._index) 1228 self._next = seq_get(self._tokens, self._index + 1) 1229 1230 if self._index > 0: 1231 self._prev = self._tokens[self._index - 1] 1232 self._prev_comments = self._prev.comments 1233 else: 1234 self._prev = None 1235 self._prev_comments = None 1236 1237 def _retreat(self, index: int) -> None: 1238 if index != self._index: 1239 self._advance(index - self._index) 1240 1241 def _warn_unsupported(self) -> None: 1242 if len(self._tokens) <= 1: 1243 return 1244 1245 # We use _find_sql because self.sql may comprise multiple chunks, and we're only 1246 # interested in emitting a warning for the one being currently processed. 1247 sql = self._find_sql(self._tokens[0], self._tokens[-1])[: self.error_message_context] 1248 1249 logger.warning( 1250 f"'{sql}' contains unsupported syntax. Falling back to parsing as a 'Command'." 1251 ) 1252 1253 def _parse_command(self) -> exp.Command: 1254 self._warn_unsupported() 1255 return self.expression( 1256 exp.Command, this=self._prev.text.upper(), expression=self._parse_string() 1257 ) 1258 1259 def _parse_comment(self, allow_exists: bool = True) -> exp.Expression: 1260 start = self._prev 1261 exists = self._parse_exists() if allow_exists else None 1262 1263 self._match(TokenType.ON) 1264 1265 kind = self._match_set(self.CREATABLES) and self._prev 1266 if not kind: 1267 return self._parse_as_command(start) 1268 1269 if kind.token_type in (TokenType.FUNCTION, TokenType.PROCEDURE): 1270 this = self._parse_user_defined_function(kind=kind.token_type) 1271 elif kind.token_type == TokenType.TABLE: 1272 this = self._parse_table(alias_tokens=self.COMMENT_TABLE_ALIAS_TOKENS) 1273 elif kind.token_type == TokenType.COLUMN: 1274 this = self._parse_column() 1275 else: 1276 this = self._parse_id_var() 1277 1278 self._match(TokenType.IS) 1279 1280 return self.expression( 1281 exp.Comment, this=this, kind=kind.text, expression=self._parse_string(), exists=exists 1282 ) 1283 1284 def _parse_to_table( 1285 self, 1286 ) -> exp.ToTableProperty: 1287 table = self._parse_table_parts(schema=True) 1288 return self.expression(exp.ToTableProperty, this=table) 1289 1290 # https://clickhouse.com/docs/en/engines/table-engines/mergetree-family/mergetree#mergetree-table-ttl 1291 def _parse_ttl(self) -> exp.Expression: 1292 def _parse_ttl_action() -> t.Optional[exp.Expression]: 1293 this = self._parse_bitwise() 1294 1295 if self._match_text_seq("DELETE"): 1296 return self.expression(exp.MergeTreeTTLAction, this=this, delete=True) 1297 if self._match_text_seq("RECOMPRESS"): 1298 return self.expression( 1299 exp.MergeTreeTTLAction, this=this, recompress=self._parse_bitwise() 1300 ) 1301 if self._match_text_seq("TO", "DISK"): 1302 return self.expression( 1303 exp.MergeTreeTTLAction, this=this, to_disk=self._parse_string() 1304 ) 1305 if self._match_text_seq("TO", "VOLUME"): 1306 return self.expression( 1307 exp.MergeTreeTTLAction, this=this, to_volume=self._parse_string() 1308 ) 1309 1310 return this 1311 1312 expressions = self._parse_csv(_parse_ttl_action) 1313 where = self._parse_where() 1314 group = self._parse_group() 1315 1316 aggregates = None 1317 if group and self._match(TokenType.SET): 1318 aggregates = self._parse_csv(self._parse_set_item) 1319 1320 return self.expression( 1321 exp.MergeTreeTTL, 1322 expressions=expressions, 1323 where=where, 1324 group=group, 1325 aggregates=aggregates, 1326 ) 1327 1328 def _parse_statement(self) -> t.Optional[exp.Expression]: 1329 if self._curr is None: 1330 return None 1331 1332 if self._match_set(self.STATEMENT_PARSERS): 1333 return self.STATEMENT_PARSERS[self._prev.token_type](self) 1334 1335 if self._match_set(Tokenizer.COMMANDS): 1336 return self._parse_command() 1337 1338 expression = self._parse_expression() 1339 expression = self._parse_set_operations(expression) if expression else self._parse_select() 1340 return self._parse_query_modifiers(expression) 1341 1342 def _parse_drop(self, exists: bool = False) -> exp.Drop | exp.Command: 1343 start = self._prev 1344 temporary = self._match(TokenType.TEMPORARY) 1345 materialized = self._match_text_seq("MATERIALIZED") 1346 1347 kind = self._match_set(self.CREATABLES) and self._prev.text 1348 if not kind: 1349 return self._parse_as_command(start) 1350 1351 return self.expression( 1352 exp.Drop, 1353 comments=start.comments, 1354 exists=exists or self._parse_exists(), 1355 this=self._parse_table( 1356 schema=True, is_db_reference=self._prev.token_type == TokenType.SCHEMA 1357 ), 1358 kind=kind, 1359 temporary=temporary, 1360 materialized=materialized, 1361 cascade=self._match_text_seq("CASCADE"), 1362 constraints=self._match_text_seq("CONSTRAINTS"), 1363 purge=self._match_text_seq("PURGE"), 1364 ) 1365 1366 def _parse_exists(self, not_: bool = False) -> t.Optional[bool]: 1367 return ( 1368 self._match_text_seq("IF") 1369 and (not not_ or self._match(TokenType.NOT)) 1370 and self._match(TokenType.EXISTS) 1371 ) 1372 1373 def _parse_create(self) -> exp.Create | exp.Command: 1374 # Note: this can't be None because we've matched a statement parser 1375 start = self._prev 1376 comments = self._prev_comments 1377 1378 replace = ( 1379 start.token_type == TokenType.REPLACE 1380 or self._match_pair(TokenType.OR, TokenType.REPLACE) 1381 or self._match_pair(TokenType.OR, TokenType.ALTER) 1382 ) 1383 unique = self._match(TokenType.UNIQUE) 1384 1385 if self._match_pair(TokenType.TABLE, TokenType.FUNCTION, advance=False): 1386 self._advance() 1387 1388 properties = None 1389 create_token = self._match_set(self.CREATABLES) and self._prev 1390 1391 if not create_token: 1392 # exp.Properties.Location.POST_CREATE 1393 properties = self._parse_properties() 1394 create_token = self._match_set(self.CREATABLES) and self._prev 1395 1396 if not properties or not create_token: 1397 return self._parse_as_command(start) 1398 1399 exists = self._parse_exists(not_=True) 1400 this = None 1401 expression: t.Optional[exp.Expression] = None 1402 indexes = None 1403 no_schema_binding = None 1404 begin = None 1405 end = None 1406 clone = None 1407 1408 def extend_props(temp_props: t.Optional[exp.Properties]) -> None: 1409 nonlocal properties 1410 if properties and temp_props: 1411 properties.expressions.extend(temp_props.expressions) 1412 elif temp_props: 1413 properties = temp_props 1414 1415 if create_token.token_type in (TokenType.FUNCTION, TokenType.PROCEDURE): 1416 this = self._parse_user_defined_function(kind=create_token.token_type) 1417 1418 # exp.Properties.Location.POST_SCHEMA ("schema" here is the UDF's type signature) 1419 extend_props(self._parse_properties()) 1420 1421 expression = self._match(TokenType.ALIAS) and self._parse_heredoc() 1422 1423 if not expression: 1424 if self._match(TokenType.COMMAND): 1425 expression = self._parse_as_command(self._prev) 1426 else: 1427 begin = self._match(TokenType.BEGIN) 1428 return_ = self._match_text_seq("RETURN") 1429 1430 if self._match(TokenType.STRING, advance=False): 1431 # Takes care of BigQuery's JavaScript UDF definitions that end in an OPTIONS property 1432 # # https://cloud.google.com/bigquery/docs/reference/standard-sql/data-definition-language#create_function_statement 1433 expression = self._parse_string() 1434 extend_props(self._parse_properties()) 1435 else: 1436 expression = self._parse_statement() 1437 1438 end = self._match_text_seq("END") 1439 1440 if return_: 1441 expression = self.expression(exp.Return, this=expression) 1442 elif create_token.token_type == TokenType.INDEX: 1443 this = self._parse_index(index=self._parse_id_var()) 1444 elif create_token.token_type in self.DB_CREATABLES: 1445 table_parts = self._parse_table_parts( 1446 schema=True, is_db_reference=create_token.token_type == TokenType.SCHEMA 1447 ) 1448 1449 # exp.Properties.Location.POST_NAME 1450 self._match(TokenType.COMMA) 1451 extend_props(self._parse_properties(before=True)) 1452 1453 this = self._parse_schema(this=table_parts) 1454 1455 # exp.Properties.Location.POST_SCHEMA and POST_WITH 1456 extend_props(self._parse_properties()) 1457 1458 self._match(TokenType.ALIAS) 1459 if not self._match_set(self.DDL_SELECT_TOKENS, advance=False): 1460 # exp.Properties.Location.POST_ALIAS 1461 extend_props(self._parse_properties()) 1462 1463 expression = self._parse_ddl_select() 1464 1465 if create_token.token_type == TokenType.TABLE: 1466 # exp.Properties.Location.POST_EXPRESSION 1467 extend_props(self._parse_properties()) 1468 1469 indexes = [] 1470 while True: 1471 index = self._parse_index() 1472 1473 # exp.Properties.Location.POST_INDEX 1474 extend_props(self._parse_properties()) 1475 1476 if not index: 1477 break 1478 else: 1479 self._match(TokenType.COMMA) 1480 indexes.append(index) 1481 elif create_token.token_type == TokenType.VIEW: 1482 if self._match_text_seq("WITH", "NO", "SCHEMA", "BINDING"): 1483 no_schema_binding = True 1484 1485 shallow = self._match_text_seq("SHALLOW") 1486 1487 if self._match_texts(self.CLONE_KEYWORDS): 1488 copy = self._prev.text.lower() == "copy" 1489 clone = self.expression( 1490 exp.Clone, this=self._parse_table(schema=True), shallow=shallow, copy=copy 1491 ) 1492 1493 if self._curr: 1494 return self._parse_as_command(start) 1495 1496 return self.expression( 1497 exp.Create, 1498 comments=comments, 1499 this=this, 1500 kind=create_token.text.upper(), 1501 replace=replace, 1502 unique=unique, 1503 expression=expression, 1504 exists=exists, 1505 properties=properties, 1506 indexes=indexes, 1507 no_schema_binding=no_schema_binding, 1508 begin=begin, 1509 end=end, 1510 clone=clone, 1511 ) 1512 1513 def _parse_property_before(self) -> t.Optional[exp.Expression]: 1514 # only used for teradata currently 1515 self._match(TokenType.COMMA) 1516 1517 kwargs = { 1518 "no": self._match_text_seq("NO"), 1519 "dual": self._match_text_seq("DUAL"), 1520 "before": self._match_text_seq("BEFORE"), 1521 "default": self._match_text_seq("DEFAULT"), 1522 "local": (self._match_text_seq("LOCAL") and "LOCAL") 1523 or (self._match_text_seq("NOT", "LOCAL") and "NOT LOCAL"), 1524 "after": self._match_text_seq("AFTER"), 1525 "minimum": self._match_texts(("MIN", "MINIMUM")), 1526 "maximum": self._match_texts(("MAX", "MAXIMUM")), 1527 } 1528 1529 if self._match_texts(self.PROPERTY_PARSERS): 1530 parser = self.PROPERTY_PARSERS[self._prev.text.upper()] 1531 try: 1532 return parser(self, **{k: v for k, v in kwargs.items() if v}) 1533 except TypeError: 1534 self.raise_error(f"Cannot parse property '{self._prev.text}'") 1535 1536 return None 1537 1538 def _parse_property(self) -> t.Optional[exp.Expression]: 1539 if self._match_texts(self.PROPERTY_PARSERS): 1540 return self.PROPERTY_PARSERS[self._prev.text.upper()](self) 1541 1542 if self._match(TokenType.DEFAULT) and self._match_texts(self.PROPERTY_PARSERS): 1543 return self.PROPERTY_PARSERS[self._prev.text.upper()](self, default=True) 1544 1545 if self._match_text_seq("COMPOUND", "SORTKEY"): 1546 return self._parse_sortkey(compound=True) 1547 1548 if self._match_text_seq("SQL", "SECURITY"): 1549 return self.expression(exp.SqlSecurityProperty, definer=self._match_text_seq("DEFINER")) 1550 1551 index = self._index 1552 key = self._parse_column() 1553 1554 if not self._match(TokenType.EQ): 1555 self._retreat(index) 1556 return None 1557 1558 return self.expression( 1559 exp.Property, 1560 this=key.to_dot() if isinstance(key, exp.Column) else key, 1561 value=self._parse_column() or self._parse_var(any_token=True), 1562 ) 1563 1564 def _parse_stored(self) -> exp.FileFormatProperty: 1565 self._match(TokenType.ALIAS) 1566 1567 input_format = self._parse_string() if self._match_text_seq("INPUTFORMAT") else None 1568 output_format = self._parse_string() if self._match_text_seq("OUTPUTFORMAT") else None 1569 1570 return self.expression( 1571 exp.FileFormatProperty, 1572 this=( 1573 self.expression( 1574 exp.InputOutputFormat, input_format=input_format, output_format=output_format 1575 ) 1576 if input_format or output_format 1577 else self._parse_var_or_string() or self._parse_number() or self._parse_id_var() 1578 ), 1579 ) 1580 1581 def _parse_property_assignment(self, exp_class: t.Type[E], **kwargs: t.Any) -> E: 1582 self._match(TokenType.EQ) 1583 self._match(TokenType.ALIAS) 1584 return self.expression(exp_class, this=self._parse_field(), **kwargs) 1585 1586 def _parse_properties(self, before: t.Optional[bool] = None) -> t.Optional[exp.Properties]: 1587 properties = [] 1588 while True: 1589 if before: 1590 prop = self._parse_property_before() 1591 else: 1592 prop = self._parse_property() 1593 1594 if not prop: 1595 break 1596 for p in ensure_list(prop): 1597 properties.append(p) 1598 1599 if properties: 1600 return self.expression(exp.Properties, expressions=properties) 1601 1602 return None 1603 1604 def _parse_fallback(self, no: bool = False) -> exp.FallbackProperty: 1605 return self.expression( 1606 exp.FallbackProperty, no=no, protection=self._match_text_seq("PROTECTION") 1607 ) 1608 1609 def _parse_volatile_property(self) -> exp.VolatileProperty | exp.StabilityProperty: 1610 if self._index >= 2: 1611 pre_volatile_token = self._tokens[self._index - 2] 1612 else: 1613 pre_volatile_token = None 1614 1615 if pre_volatile_token and pre_volatile_token.token_type in self.PRE_VOLATILE_TOKENS: 1616 return exp.VolatileProperty() 1617 1618 return self.expression(exp.StabilityProperty, this=exp.Literal.string("VOLATILE")) 1619 1620 def _parse_system_versioning_property(self) -> exp.WithSystemVersioningProperty: 1621 self._match_pair(TokenType.EQ, TokenType.ON) 1622 1623 prop = self.expression(exp.WithSystemVersioningProperty) 1624 if self._match(TokenType.L_PAREN): 1625 self._match_text_seq("HISTORY_TABLE", "=") 1626 prop.set("this", self._parse_table_parts()) 1627 1628 if self._match(TokenType.COMMA): 1629 self._match_text_seq("DATA_CONSISTENCY_CHECK", "=") 1630 prop.set("expression", self._advance_any() and self._prev.text.upper()) 1631 1632 self._match_r_paren() 1633 1634 return prop 1635 1636 def _parse_with_property( 1637 self, 1638 ) -> t.Optional[exp.Expression] | t.List[exp.Expression]: 1639 if self._match(TokenType.L_PAREN, advance=False): 1640 return self._parse_wrapped_csv(self._parse_property) 1641 1642 if self._match_text_seq("JOURNAL"): 1643 return self._parse_withjournaltable() 1644 1645 if self._match_text_seq("DATA"): 1646 return self._parse_withdata(no=False) 1647 elif self._match_text_seq("NO", "DATA"): 1648 return self._parse_withdata(no=True) 1649 1650 if not self._next: 1651 return None 1652 1653 return self._parse_withisolatedloading() 1654 1655 # https://dev.mysql.com/doc/refman/8.0/en/create-view.html 1656 def _parse_definer(self) -> t.Optional[exp.DefinerProperty]: 1657 self._match(TokenType.EQ) 1658 1659 user = self._parse_id_var() 1660 self._match(TokenType.PARAMETER) 1661 host = self._parse_id_var() or (self._match(TokenType.MOD) and self._prev.text) 1662 1663 if not user or not host: 1664 return None 1665 1666 return exp.DefinerProperty(this=f"{user}@{host}") 1667 1668 def _parse_withjournaltable(self) -> exp.WithJournalTableProperty: 1669 self._match(TokenType.TABLE) 1670 self._match(TokenType.EQ) 1671 return self.expression(exp.WithJournalTableProperty, this=self._parse_table_parts()) 1672 1673 def _parse_log(self, no: bool = False) -> exp.LogProperty: 1674 return self.expression(exp.LogProperty, no=no) 1675 1676 def _parse_journal(self, **kwargs) -> exp.JournalProperty: 1677 return self.expression(exp.JournalProperty, **kwargs) 1678 1679 def _parse_checksum(self) -> exp.ChecksumProperty: 1680 self._match(TokenType.EQ) 1681 1682 on = None 1683 if self._match(TokenType.ON): 1684 on = True 1685 elif self._match_text_seq("OFF"): 1686 on = False 1687 1688 return self.expression(exp.ChecksumProperty, on=on, default=self._match(TokenType.DEFAULT)) 1689 1690 def _parse_cluster(self, wrapped: bool = False) -> exp.Cluster: 1691 return self.expression( 1692 exp.Cluster, 1693 expressions=( 1694 self._parse_wrapped_csv(self._parse_ordered) 1695 if wrapped 1696 else self._parse_csv(self._parse_ordered) 1697 ), 1698 ) 1699 1700 def _parse_clustered_by(self) -> exp.ClusteredByProperty: 1701 self._match_text_seq("BY") 1702 1703 self._match_l_paren() 1704 expressions = self._parse_csv(self._parse_column) 1705 self._match_r_paren() 1706 1707 if self._match_text_seq("SORTED", "BY"): 1708 self._match_l_paren() 1709 sorted_by = self._parse_csv(self._parse_ordered) 1710 self._match_r_paren() 1711 else: 1712 sorted_by = None 1713 1714 self._match(TokenType.INTO) 1715 buckets = self._parse_number() 1716 self._match_text_seq("BUCKETS") 1717 1718 return self.expression( 1719 exp.ClusteredByProperty, 1720 expressions=expressions, 1721 sorted_by=sorted_by, 1722 buckets=buckets, 1723 ) 1724 1725 def _parse_copy_property(self) -> t.Optional[exp.CopyGrantsProperty]: 1726 if not self._match_text_seq("GRANTS"): 1727 self._retreat(self._index - 1) 1728 return None 1729 1730 return self.expression(exp.CopyGrantsProperty) 1731 1732 def _parse_freespace(self) -> exp.FreespaceProperty: 1733 self._match(TokenType.EQ) 1734 return self.expression( 1735 exp.FreespaceProperty, this=self._parse_number(), percent=self._match(TokenType.PERCENT) 1736 ) 1737 1738 def _parse_mergeblockratio( 1739 self, no: bool = False, default: bool = False 1740 ) -> exp.MergeBlockRatioProperty: 1741 if self._match(TokenType.EQ): 1742 return self.expression( 1743 exp.MergeBlockRatioProperty, 1744 this=self._parse_number(), 1745 percent=self._match(TokenType.PERCENT), 1746 ) 1747 1748 return self.expression(exp.MergeBlockRatioProperty, no=no, default=default) 1749 1750 def _parse_datablocksize( 1751 self, 1752 default: t.Optional[bool] = None, 1753 minimum: t.Optional[bool] = None, 1754 maximum: t.Optional[bool] = None, 1755 ) -> exp.DataBlocksizeProperty: 1756 self._match(TokenType.EQ) 1757 size = self._parse_number() 1758 1759 units = None 1760 if self._match_texts(("BYTES", "KBYTES", "KILOBYTES")): 1761 units = self._prev.text 1762 1763 return self.expression( 1764 exp.DataBlocksizeProperty, 1765 size=size, 1766 units=units, 1767 default=default, 1768 minimum=minimum, 1769 maximum=maximum, 1770 ) 1771 1772 def _parse_blockcompression(self) -> exp.BlockCompressionProperty: 1773 self._match(TokenType.EQ) 1774 always = self._match_text_seq("ALWAYS") 1775 manual = self._match_text_seq("MANUAL") 1776 never = self._match_text_seq("NEVER") 1777 default = self._match_text_seq("DEFAULT") 1778 1779 autotemp = None 1780 if self._match_text_seq("AUTOTEMP"): 1781 autotemp = self._parse_schema() 1782 1783 return self.expression( 1784 exp.BlockCompressionProperty, 1785 always=always, 1786 manual=manual, 1787 never=never, 1788 default=default, 1789 autotemp=autotemp, 1790 ) 1791 1792 def _parse_withisolatedloading(self) -> exp.IsolatedLoadingProperty: 1793 no = self._match_text_seq("NO") 1794 concurrent = self._match_text_seq("CONCURRENT") 1795 self._match_text_seq("ISOLATED", "LOADING") 1796 for_all = self._match_text_seq("FOR", "ALL") 1797 for_insert = self._match_text_seq("FOR", "INSERT") 1798 for_none = self._match_text_seq("FOR", "NONE") 1799 return self.expression( 1800 exp.IsolatedLoadingProperty, 1801 no=no, 1802 concurrent=concurrent, 1803 for_all=for_all, 1804 for_insert=for_insert, 1805 for_none=for_none, 1806 ) 1807 1808 def _parse_locking(self) -> exp.LockingProperty: 1809 if self._match(TokenType.TABLE): 1810 kind = "TABLE" 1811 elif self._match(TokenType.VIEW): 1812 kind = "VIEW" 1813 elif self._match(TokenType.ROW): 1814 kind = "ROW" 1815 elif self._match_text_seq("DATABASE"): 1816 kind = "DATABASE" 1817 else: 1818 kind = None 1819 1820 if kind in ("DATABASE", "TABLE", "VIEW"): 1821 this = self._parse_table_parts() 1822 else: 1823 this = None 1824 1825 if self._match(TokenType.FOR): 1826 for_or_in = "FOR" 1827 elif self._match(TokenType.IN): 1828 for_or_in = "IN" 1829 else: 1830 for_or_in = None 1831 1832 if self._match_text_seq("ACCESS"): 1833 lock_type = "ACCESS" 1834 elif self._match_texts(("EXCL", "EXCLUSIVE")): 1835 lock_type = "EXCLUSIVE" 1836 elif self._match_text_seq("SHARE"): 1837 lock_type = "SHARE" 1838 elif self._match_text_seq("READ"): 1839 lock_type = "READ" 1840 elif self._match_text_seq("WRITE"): 1841 lock_type = "WRITE" 1842 elif self._match_text_seq("CHECKSUM"): 1843 lock_type = "CHECKSUM" 1844 else: 1845 lock_type = None 1846 1847 override = self._match_text_seq("OVERRIDE") 1848 1849 return self.expression( 1850 exp.LockingProperty, 1851 this=this, 1852 kind=kind, 1853 for_or_in=for_or_in, 1854 lock_type=lock_type, 1855 override=override, 1856 ) 1857 1858 def _parse_partition_by(self) -> t.List[exp.Expression]: 1859 if self._match(TokenType.PARTITION_BY): 1860 return self._parse_csv(self._parse_conjunction) 1861 return [] 1862 1863 def _parse_partition_bound_spec(self) -> exp.PartitionBoundSpec: 1864 def _parse_partition_bound_expr() -> t.Optional[exp.Expression]: 1865 if self._match_text_seq("MINVALUE"): 1866 return exp.var("MINVALUE") 1867 if self._match_text_seq("MAXVALUE"): 1868 return exp.var("MAXVALUE") 1869 return self._parse_bitwise() 1870 1871 this: t.Optional[exp.Expression | t.List[exp.Expression]] = None 1872 expression = None 1873 from_expressions = None 1874 to_expressions = None 1875 1876 if self._match(TokenType.IN): 1877 this = self._parse_wrapped_csv(self._parse_bitwise) 1878 elif self._match(TokenType.FROM): 1879 from_expressions = self._parse_wrapped_csv(_parse_partition_bound_expr) 1880 self._match_text_seq("TO") 1881 to_expressions = self._parse_wrapped_csv(_parse_partition_bound_expr) 1882 elif self._match_text_seq("WITH", "(", "MODULUS"): 1883 this = self._parse_number() 1884 self._match_text_seq(",", "REMAINDER") 1885 expression = self._parse_number() 1886 self._match_r_paren() 1887 else: 1888 self.raise_error("Failed to parse partition bound spec.") 1889 1890 return self.expression( 1891 exp.PartitionBoundSpec, 1892 this=this, 1893 expression=expression, 1894 from_expressions=from_expressions, 1895 to_expressions=to_expressions, 1896 ) 1897 1898 # https://www.postgresql.org/docs/current/sql-createtable.html 1899 def _parse_partitioned_of(self) -> t.Optional[exp.PartitionedOfProperty]: 1900 if not self._match_text_seq("OF"): 1901 self._retreat(self._index - 1) 1902 return None 1903 1904 this = self._parse_table(schema=True) 1905 1906 if self._match(TokenType.DEFAULT): 1907 expression: exp.Var | exp.PartitionBoundSpec = exp.var("DEFAULT") 1908 elif self._match_text_seq("FOR", "VALUES"): 1909 expression = self._parse_partition_bound_spec() 1910 else: 1911 self.raise_error("Expecting either DEFAULT or FOR VALUES clause.") 1912 1913 return self.expression(exp.PartitionedOfProperty, this=this, expression=expression) 1914 1915 def _parse_partitioned_by(self) -> exp.PartitionedByProperty: 1916 self._match(TokenType.EQ) 1917 return self.expression( 1918 exp.PartitionedByProperty, 1919 this=self._parse_schema() or self._parse_bracket(self._parse_field()), 1920 ) 1921 1922 def _parse_withdata(self, no: bool = False) -> exp.WithDataProperty: 1923 if self._match_text_seq("AND", "STATISTICS"): 1924 statistics = True 1925 elif self._match_text_seq("AND", "NO", "STATISTICS"): 1926 statistics = False 1927 else: 1928 statistics = None 1929 1930 return self.expression(exp.WithDataProperty, no=no, statistics=statistics) 1931 1932 def _parse_contains_property(self) -> t.Optional[exp.SqlReadWriteProperty]: 1933 if self._match_text_seq("SQL"): 1934 return self.expression(exp.SqlReadWriteProperty, this="CONTAINS SQL") 1935 return None 1936 1937 def _parse_modifies_property(self) -> t.Optional[exp.SqlReadWriteProperty]: 1938 if self._match_text_seq("SQL", "DATA"): 1939 return self.expression(exp.SqlReadWriteProperty, this="MODIFIES SQL DATA") 1940 return None 1941 1942 def _parse_no_property(self) -> t.Optional[exp.Expression]: 1943 if self._match_text_seq("PRIMARY", "INDEX"): 1944 return exp.NoPrimaryIndexProperty() 1945 if self._match_text_seq("SQL"): 1946 return self.expression(exp.SqlReadWriteProperty, this="NO SQL") 1947 return None 1948 1949 def _parse_on_property(self) -> t.Optional[exp.Expression]: 1950 if self._match_text_seq("COMMIT", "PRESERVE", "ROWS"): 1951 return exp.OnCommitProperty() 1952 if self._match_text_seq("COMMIT", "DELETE", "ROWS"): 1953 return exp.OnCommitProperty(delete=True) 1954 return self.expression(exp.OnProperty, this=self._parse_schema(self._parse_id_var())) 1955 1956 def _parse_reads_property(self) -> t.Optional[exp.SqlReadWriteProperty]: 1957 if self._match_text_seq("SQL", "DATA"): 1958 return self.expression(exp.SqlReadWriteProperty, this="READS SQL DATA") 1959 return None 1960 1961 def _parse_distkey(self) -> exp.DistKeyProperty: 1962 return self.expression(exp.DistKeyProperty, this=self._parse_wrapped(self._parse_id_var)) 1963 1964 def _parse_create_like(self) -> t.Optional[exp.LikeProperty]: 1965 table = self._parse_table(schema=True) 1966 1967 options = [] 1968 while self._match_texts(("INCLUDING", "EXCLUDING")): 1969 this = self._prev.text.upper() 1970 1971 id_var = self._parse_id_var() 1972 if not id_var: 1973 return None 1974 1975 options.append( 1976 self.expression(exp.Property, this=this, value=exp.var(id_var.this.upper())) 1977 ) 1978 1979 return self.expression(exp.LikeProperty, this=table, expressions=options) 1980 1981 def _parse_sortkey(self, compound: bool = False) -> exp.SortKeyProperty: 1982 return self.expression( 1983 exp.SortKeyProperty, this=self._parse_wrapped_id_vars(), compound=compound 1984 ) 1985 1986 def _parse_character_set(self, default: bool = False) -> exp.CharacterSetProperty: 1987 self._match(TokenType.EQ) 1988 return self.expression( 1989 exp.CharacterSetProperty, this=self._parse_var_or_string(), default=default 1990 ) 1991 1992 def _parse_remote_with_connection(self) -> exp.RemoteWithConnectionModelProperty: 1993 self._match_text_seq("WITH", "CONNECTION") 1994 return self.expression( 1995 exp.RemoteWithConnectionModelProperty, this=self._parse_table_parts() 1996 ) 1997 1998 def _parse_returns(self) -> exp.ReturnsProperty: 1999 value: t.Optional[exp.Expression] 2000 is_table = self._match(TokenType.TABLE) 2001 2002 if is_table: 2003 if self._match(TokenType.LT): 2004 value = self.expression( 2005 exp.Schema, 2006 this="TABLE", 2007 expressions=self._parse_csv(self._parse_struct_types), 2008 ) 2009 if not self._match(TokenType.GT): 2010 self.raise_error("Expecting >") 2011 else: 2012 value = self._parse_schema(exp.var("TABLE")) 2013 else: 2014 value = self._parse_types() 2015 2016 return self.expression(exp.ReturnsProperty, this=value, is_table=is_table) 2017 2018 def _parse_describe(self) -> exp.Describe: 2019 kind = self._match_set(self.CREATABLES) and self._prev.text 2020 extended = self._match_text_seq("EXTENDED") 2021 this = self._parse_table(schema=True) 2022 properties = self._parse_properties() 2023 expressions = properties.expressions if properties else None 2024 return self.expression( 2025 exp.Describe, this=this, extended=extended, kind=kind, expressions=expressions 2026 ) 2027 2028 def _parse_insert(self) -> exp.Insert: 2029 comments = ensure_list(self._prev_comments) 2030 overwrite = self._match(TokenType.OVERWRITE) 2031 ignore = self._match(TokenType.IGNORE) 2032 local = self._match_text_seq("LOCAL") 2033 alternative = None 2034 2035 if self._match_text_seq("DIRECTORY"): 2036 this: t.Optional[exp.Expression] = self.expression( 2037 exp.Directory, 2038 this=self._parse_var_or_string(), 2039 local=local, 2040 row_format=self._parse_row_format(match_row=True), 2041 ) 2042 else: 2043 if self._match(TokenType.OR): 2044 alternative = self._match_texts(self.INSERT_ALTERNATIVES) and self._prev.text 2045 2046 self._match(TokenType.INTO) 2047 comments += ensure_list(self._prev_comments) 2048 self._match(TokenType.TABLE) 2049 this = self._parse_table(schema=True) 2050 2051 returning = self._parse_returning() 2052 2053 return self.expression( 2054 exp.Insert, 2055 comments=comments, 2056 this=this, 2057 by_name=self._match_text_seq("BY", "NAME"), 2058 exists=self._parse_exists(), 2059 partition=self._parse_partition(), 2060 where=self._match_pair(TokenType.REPLACE, TokenType.WHERE) 2061 and self._parse_conjunction(), 2062 expression=self._parse_ddl_select(), 2063 conflict=self._parse_on_conflict(), 2064 returning=returning or self._parse_returning(), 2065 overwrite=overwrite, 2066 alternative=alternative, 2067 ignore=ignore, 2068 ) 2069 2070 def _parse_kill(self) -> exp.Kill: 2071 kind = exp.var(self._prev.text) if self._match_texts(("CONNECTION", "QUERY")) else None 2072 2073 return self.expression( 2074 exp.Kill, 2075 this=self._parse_primary(), 2076 kind=kind, 2077 ) 2078 2079 def _parse_on_conflict(self) -> t.Optional[exp.OnConflict]: 2080 conflict = self._match_text_seq("ON", "CONFLICT") 2081 duplicate = self._match_text_seq("ON", "DUPLICATE", "KEY") 2082 2083 if not conflict and not duplicate: 2084 return None 2085 2086 nothing = None 2087 expressions = None 2088 key = None 2089 constraint = None 2090 2091 if conflict: 2092 if self._match_text_seq("ON", "CONSTRAINT"): 2093 constraint = self._parse_id_var() 2094 else: 2095 key = self._parse_csv(self._parse_value) 2096 2097 self._match_text_seq("DO") 2098 if self._match_text_seq("NOTHING"): 2099 nothing = True 2100 else: 2101 self._match(TokenType.UPDATE) 2102 self._match(TokenType.SET) 2103 expressions = self._parse_csv(self._parse_equality) 2104 2105 return self.expression( 2106 exp.OnConflict, 2107 duplicate=duplicate, 2108 expressions=expressions, 2109 nothing=nothing, 2110 key=key, 2111 constraint=constraint, 2112 ) 2113 2114 def _parse_returning(self) -> t.Optional[exp.Returning]: 2115 if not self._match(TokenType.RETURNING): 2116 return None 2117 return self.expression( 2118 exp.Returning, 2119 expressions=self._parse_csv(self._parse_expression), 2120 into=self._match(TokenType.INTO) and self._parse_table_part(), 2121 ) 2122 2123 def _parse_row(self) -> t.Optional[exp.RowFormatSerdeProperty | exp.RowFormatDelimitedProperty]: 2124 if not self._match(TokenType.FORMAT): 2125 return None 2126 return self._parse_row_format() 2127 2128 def _parse_row_format( 2129 self, match_row: bool = False 2130 ) -> t.Optional[exp.RowFormatSerdeProperty | exp.RowFormatDelimitedProperty]: 2131 if match_row and not self._match_pair(TokenType.ROW, TokenType.FORMAT): 2132 return None 2133 2134 if self._match_text_seq("SERDE"): 2135 this = self._parse_string() 2136 2137 serde_properties = None 2138 if self._match(TokenType.SERDE_PROPERTIES): 2139 serde_properties = self.expression( 2140 exp.SerdeProperties, expressions=self._parse_wrapped_csv(self._parse_property) 2141 ) 2142 2143 return self.expression( 2144 exp.RowFormatSerdeProperty, this=this, serde_properties=serde_properties 2145 ) 2146 2147 self._match_text_seq("DELIMITED") 2148 2149 kwargs = {} 2150 2151 if self._match_text_seq("FIELDS", "TERMINATED", "BY"): 2152 kwargs["fields"] = self._parse_string() 2153 if self._match_text_seq("ESCAPED", "BY"): 2154 kwargs["escaped"] = self._parse_string() 2155 if self._match_text_seq("COLLECTION", "ITEMS", "TERMINATED", "BY"): 2156 kwargs["collection_items"] = self._parse_string() 2157 if self._match_text_seq("MAP", "KEYS", "TERMINATED", "BY"): 2158 kwargs["map_keys"] = self._parse_string() 2159 if self._match_text_seq("LINES", "TERMINATED", "BY"): 2160 kwargs["lines"] = self._parse_string() 2161 if self._match_text_seq("NULL", "DEFINED", "AS"): 2162 kwargs["null"] = self._parse_string() 2163 2164 return self.expression(exp.RowFormatDelimitedProperty, **kwargs) # type: ignore 2165 2166 def _parse_load(self) -> exp.LoadData | exp.Command: 2167 if self._match_text_seq("DATA"): 2168 local = self._match_text_seq("LOCAL") 2169 self._match_text_seq("INPATH") 2170 inpath = self._parse_string() 2171 overwrite = self._match(TokenType.OVERWRITE) 2172 self._match_pair(TokenType.INTO, TokenType.TABLE) 2173 2174 return self.expression( 2175 exp.LoadData, 2176 this=self._parse_table(schema=True), 2177 local=local, 2178 overwrite=overwrite, 2179 inpath=inpath, 2180 partition=self._parse_partition(), 2181 input_format=self._match_text_seq("INPUTFORMAT") and self._parse_string(), 2182 serde=self._match_text_seq("SERDE") and self._parse_string(), 2183 ) 2184 return self._parse_as_command(self._prev) 2185 2186 def _parse_delete(self) -> exp.Delete: 2187 # This handles MySQL's "Multiple-Table Syntax" 2188 # https://dev.mysql.com/doc/refman/8.0/en/delete.html 2189 tables = None 2190 comments = self._prev_comments 2191 if not self._match(TokenType.FROM, advance=False): 2192 tables = self._parse_csv(self._parse_table) or None 2193 2194 returning = self._parse_returning() 2195 2196 return self.expression( 2197 exp.Delete, 2198 comments=comments, 2199 tables=tables, 2200 this=self._match(TokenType.FROM) and self._parse_table(joins=True), 2201 using=self._match(TokenType.USING) and self._parse_table(joins=True), 2202 where=self._parse_where(), 2203 returning=returning or self._parse_returning(), 2204 limit=self._parse_limit(), 2205 ) 2206 2207 def _parse_update(self) -> exp.Update: 2208 comments = self._prev_comments 2209 this = self._parse_table(joins=True, alias_tokens=self.UPDATE_ALIAS_TOKENS) 2210 expressions = self._match(TokenType.SET) and self._parse_csv(self._parse_equality) 2211 returning = self._parse_returning() 2212 return self.expression( 2213 exp.Update, 2214 comments=comments, 2215 **{ # type: ignore 2216 "this": this, 2217 "expressions": expressions, 2218 "from": self._parse_from(joins=True), 2219 "where": self._parse_where(), 2220 "returning": returning or self._parse_returning(), 2221 "order": self._parse_order(), 2222 "limit": self._parse_limit(), 2223 }, 2224 ) 2225 2226 def _parse_uncache(self) -> exp.Uncache: 2227 if not self._match(TokenType.TABLE): 2228 self.raise_error("Expecting TABLE after UNCACHE") 2229 2230 return self.expression( 2231 exp.Uncache, exists=self._parse_exists(), this=self._parse_table(schema=True) 2232 ) 2233 2234 def _parse_cache(self) -> exp.Cache: 2235 lazy = self._match_text_seq("LAZY") 2236 self._match(TokenType.TABLE) 2237 table = self._parse_table(schema=True) 2238 2239 options = [] 2240 if self._match_text_seq("OPTIONS"): 2241 self._match_l_paren() 2242 k = self._parse_string() 2243 self._match(TokenType.EQ) 2244 v = self._parse_string() 2245 options = [k, v] 2246 self._match_r_paren() 2247 2248 self._match(TokenType.ALIAS) 2249 return self.expression( 2250 exp.Cache, 2251 this=table, 2252 lazy=lazy, 2253 options=options, 2254 expression=self._parse_select(nested=True), 2255 ) 2256 2257 def _parse_partition(self) -> t.Optional[exp.Partition]: 2258 if not self._match(TokenType.PARTITION): 2259 return None 2260 2261 return self.expression( 2262 exp.Partition, expressions=self._parse_wrapped_csv(self._parse_conjunction) 2263 ) 2264 2265 def _parse_value(self) -> exp.Tuple: 2266 if self._match(TokenType.L_PAREN): 2267 expressions = self._parse_csv(self._parse_expression) 2268 self._match_r_paren() 2269 return self.expression(exp.Tuple, expressions=expressions) 2270 2271 # In presto we can have VALUES 1, 2 which results in 1 column & 2 rows. 2272 # https://prestodb.io/docs/current/sql/values.html 2273 return self.expression(exp.Tuple, expressions=[self._parse_expression()]) 2274 2275 def _parse_projections(self) -> t.List[exp.Expression]: 2276 return self._parse_expressions() 2277 2278 def _parse_select( 2279 self, 2280 nested: bool = False, 2281 table: bool = False, 2282 parse_subquery_alias: bool = True, 2283 parse_set_operation: bool = True, 2284 ) -> t.Optional[exp.Expression]: 2285 cte = self._parse_with() 2286 2287 if cte: 2288 this = self._parse_statement() 2289 2290 if not this: 2291 self.raise_error("Failed to parse any statement following CTE") 2292 return cte 2293 2294 if "with" in this.arg_types: 2295 this.set("with", cte) 2296 else: 2297 self.raise_error(f"{this.key} does not support CTE") 2298 this = cte 2299 2300 return this 2301 2302 # duckdb supports leading with FROM x 2303 from_ = self._parse_from() if self._match(TokenType.FROM, advance=False) else None 2304 2305 if self._match(TokenType.SELECT): 2306 comments = self._prev_comments 2307 2308 hint = self._parse_hint() 2309 all_ = self._match(TokenType.ALL) 2310 distinct = self._match_set(self.DISTINCT_TOKENS) 2311 2312 kind = ( 2313 self._match(TokenType.ALIAS) 2314 and self._match_texts(("STRUCT", "VALUE")) 2315 and self._prev.text.upper() 2316 ) 2317 2318 if distinct: 2319 distinct = self.expression( 2320 exp.Distinct, 2321 on=self._parse_value() if self._match(TokenType.ON) else None, 2322 ) 2323 2324 if all_ and distinct: 2325 self.raise_error("Cannot specify both ALL and DISTINCT after SELECT") 2326 2327 limit = self._parse_limit(top=True) 2328 projections = self._parse_projections() 2329 2330 this = self.expression( 2331 exp.Select, 2332 kind=kind, 2333 hint=hint, 2334 distinct=distinct, 2335 expressions=projections, 2336 limit=limit, 2337 ) 2338 this.comments = comments 2339 2340 into = self._parse_into() 2341 if into: 2342 this.set("into", into) 2343 2344 if not from_: 2345 from_ = self._parse_from() 2346 2347 if from_: 2348 this.set("from", from_) 2349 2350 this = self._parse_query_modifiers(this) 2351 elif (table or nested) and self._match(TokenType.L_PAREN): 2352 if self._match(TokenType.PIVOT): 2353 this = self._parse_simplified_pivot() 2354 elif self._match(TokenType.FROM): 2355 this = exp.select("*").from_( 2356 t.cast(exp.From, self._parse_from(skip_from_token=True)) 2357 ) 2358 else: 2359 this = ( 2360 self._parse_table() 2361 if table 2362 else self._parse_select(nested=True, parse_set_operation=False) 2363 ) 2364 this = self._parse_query_modifiers(self._parse_set_operations(this)) 2365 2366 self._match_r_paren() 2367 2368 # We return early here so that the UNION isn't attached to the subquery by the 2369 # following call to _parse_set_operations, but instead becomes the parent node 2370 return self._parse_subquery(this, parse_alias=parse_subquery_alias) 2371 elif self._match(TokenType.VALUES): 2372 this = self.expression( 2373 exp.Values, 2374 expressions=self._parse_csv(self._parse_value), 2375 alias=self._parse_table_alias(), 2376 ) 2377 elif from_: 2378 this = exp.select("*").from_(from_.this, copy=False) 2379 else: 2380 this = None 2381 2382 if parse_set_operation: 2383 return self._parse_set_operations(this) 2384 return this 2385 2386 def _parse_with(self, skip_with_token: bool = False) -> t.Optional[exp.With]: 2387 if not skip_with_token and not self._match(TokenType.WITH): 2388 return None 2389 2390 comments = self._prev_comments 2391 recursive = self._match(TokenType.RECURSIVE) 2392 2393 expressions = [] 2394 while True: 2395 expressions.append(self._parse_cte()) 2396 2397 if not self._match(TokenType.COMMA) and not self._match(TokenType.WITH): 2398 break 2399 else: 2400 self._match(TokenType.WITH) 2401 2402 return self.expression( 2403 exp.With, comments=comments, expressions=expressions, recursive=recursive 2404 ) 2405 2406 def _parse_cte(self) -> exp.CTE: 2407 alias = self._parse_table_alias(self.ID_VAR_TOKENS) 2408 if not alias or not alias.this: 2409 self.raise_error("Expected CTE to have alias") 2410 2411 self._match(TokenType.ALIAS) 2412 return self.expression( 2413 exp.CTE, this=self._parse_wrapped(self._parse_statement), alias=alias 2414 ) 2415 2416 def _parse_table_alias( 2417 self, alias_tokens: t.Optional[t.Collection[TokenType]] = None 2418 ) -> t.Optional[exp.TableAlias]: 2419 any_token = self._match(TokenType.ALIAS) 2420 alias = ( 2421 self._parse_id_var(any_token=any_token, tokens=alias_tokens or self.TABLE_ALIAS_TOKENS) 2422 or self._parse_string_as_identifier() 2423 ) 2424 2425 index = self._index 2426 if self._match(TokenType.L_PAREN): 2427 columns = self._parse_csv(self._parse_function_parameter) 2428 self._match_r_paren() if columns else self._retreat(index) 2429 else: 2430 columns = None 2431 2432 if not alias and not columns: 2433 return None 2434 2435 return self.expression(exp.TableAlias, this=alias, columns=columns) 2436 2437 def _parse_subquery( 2438 self, this: t.Optional[exp.Expression], parse_alias: bool = True 2439 ) -> t.Optional[exp.Subquery]: 2440 if not this: 2441 return None 2442 2443 return self.expression( 2444 exp.Subquery, 2445 this=this, 2446 pivots=self._parse_pivots(), 2447 alias=self._parse_table_alias() if parse_alias else None, 2448 ) 2449 2450 def _parse_query_modifiers( 2451 self, this: t.Optional[exp.Expression] 2452 ) -> t.Optional[exp.Expression]: 2453 if isinstance(this, self.MODIFIABLES): 2454 for join in iter(self._parse_join, None): 2455 this.append("joins", join) 2456 for lateral in iter(self._parse_lateral, None): 2457 this.append("laterals", lateral) 2458 2459 while True: 2460 if self._match_set(self.QUERY_MODIFIER_PARSERS, advance=False): 2461 parser = self.QUERY_MODIFIER_PARSERS[self._curr.token_type] 2462 key, expression = parser(self) 2463 2464 if expression: 2465 this.set(key, expression) 2466 if key == "limit": 2467 offset = expression.args.pop("offset", None) 2468 2469 if offset: 2470 offset = exp.Offset(expression=offset) 2471 this.set("offset", offset) 2472 2473 limit_by_expressions = expression.expressions 2474 expression.set("expressions", None) 2475 offset.set("expressions", limit_by_expressions) 2476 continue 2477 break 2478 return this 2479 2480 def _parse_hint(self) -> t.Optional[exp.Hint]: 2481 if self._match(TokenType.HINT): 2482 hints = [] 2483 for hint in iter(lambda: self._parse_csv(self._parse_function), []): 2484 hints.extend(hint) 2485 2486 if not self._match_pair(TokenType.STAR, TokenType.SLASH): 2487 self.raise_error("Expected */ after HINT") 2488 2489 return self.expression(exp.Hint, expressions=hints) 2490 2491 return None 2492 2493 def _parse_into(self) -> t.Optional[exp.Into]: 2494 if not self._match(TokenType.INTO): 2495 return None 2496 2497 temp = self._match(TokenType.TEMPORARY) 2498 unlogged = self._match_text_seq("UNLOGGED") 2499 self._match(TokenType.TABLE) 2500 2501 return self.expression( 2502 exp.Into, this=self._parse_table(schema=True), temporary=temp, unlogged=unlogged 2503 ) 2504 2505 def _parse_from( 2506 self, joins: bool = False, skip_from_token: bool = False 2507 ) -> t.Optional[exp.From]: 2508 if not skip_from_token and not self._match(TokenType.FROM): 2509 return None 2510 2511 return self.expression( 2512 exp.From, comments=self._prev_comments, this=self._parse_table(joins=joins) 2513 ) 2514 2515 def _parse_match_recognize(self) -> t.Optional[exp.MatchRecognize]: 2516 if not self._match(TokenType.MATCH_RECOGNIZE): 2517 return None 2518 2519 self._match_l_paren() 2520 2521 partition = self._parse_partition_by() 2522 order = self._parse_order() 2523 measures = self._parse_expressions() if self._match_text_seq("MEASURES") else None 2524 2525 if self._match_text_seq("ONE", "ROW", "PER", "MATCH"): 2526 rows = exp.var("ONE ROW PER MATCH") 2527 elif self._match_text_seq("ALL", "ROWS", "PER", "MATCH"): 2528 text = "ALL ROWS PER MATCH" 2529 if self._match_text_seq("SHOW", "EMPTY", "MATCHES"): 2530 text += " SHOW EMPTY MATCHES" 2531 elif self._match_text_seq("OMIT", "EMPTY", "MATCHES"): 2532 text += " OMIT EMPTY MATCHES" 2533 elif self._match_text_seq("WITH", "UNMATCHED", "ROWS"): 2534 text += " WITH UNMATCHED ROWS" 2535 rows = exp.var(text) 2536 else: 2537 rows = None 2538 2539 if self._match_text_seq("AFTER", "MATCH", "SKIP"): 2540 text = "AFTER MATCH SKIP" 2541 if self._match_text_seq("PAST", "LAST", "ROW"): 2542 text += " PAST LAST ROW" 2543 elif self._match_text_seq("TO", "NEXT", "ROW"): 2544 text += " TO NEXT ROW" 2545 elif self._match_text_seq("TO", "FIRST"): 2546 text += f" TO FIRST {self._advance_any().text}" # type: ignore 2547 elif self._match_text_seq("TO", "LAST"): 2548 text += f" TO LAST {self._advance_any().text}" # type: ignore 2549 after = exp.var(text) 2550 else: 2551 after = None 2552 2553 if self._match_text_seq("PATTERN"): 2554 self._match_l_paren() 2555 2556 if not self._curr: 2557 self.raise_error("Expecting )", self._curr) 2558 2559 paren = 1 2560 start = self._curr 2561 2562 while self._curr and paren > 0: 2563 if self._curr.token_type == TokenType.L_PAREN: 2564 paren += 1 2565 if self._curr.token_type == TokenType.R_PAREN: 2566 paren -= 1 2567 2568 end = self._prev 2569 self._advance() 2570 2571 if paren > 0: 2572 self.raise_error("Expecting )", self._curr) 2573 2574 pattern = exp.var(self._find_sql(start, end)) 2575 else: 2576 pattern = None 2577 2578 define = ( 2579 self._parse_csv(self._parse_name_as_expression) 2580 if self._match_text_seq("DEFINE") 2581 else None 2582 ) 2583 2584 self._match_r_paren() 2585 2586 return self.expression( 2587 exp.MatchRecognize, 2588 partition_by=partition, 2589 order=order, 2590 measures=measures, 2591 rows=rows, 2592 after=after, 2593 pattern=pattern, 2594 define=define, 2595 alias=self._parse_table_alias(), 2596 ) 2597 2598 def _parse_lateral(self) -> t.Optional[exp.Lateral]: 2599 cross_apply = self._match_pair(TokenType.CROSS, TokenType.APPLY) 2600 if not cross_apply and self._match_pair(TokenType.OUTER, TokenType.APPLY): 2601 cross_apply = False 2602 2603 if cross_apply is not None: 2604 this = self._parse_select(table=True) 2605 view = None 2606 outer = None 2607 elif self._match(TokenType.LATERAL): 2608 this = self._parse_select(table=True) 2609 view = self._match(TokenType.VIEW) 2610 outer = self._match(TokenType.OUTER) 2611 else: 2612 return None 2613 2614 if not this: 2615 this = ( 2616 self._parse_unnest() 2617 or self._parse_function() 2618 or self._parse_id_var(any_token=False) 2619 ) 2620 2621 while self._match(TokenType.DOT): 2622 this = exp.Dot( 2623 this=this, 2624 expression=self._parse_function() or self._parse_id_var(any_token=False), 2625 ) 2626 2627 if view: 2628 table = self._parse_id_var(any_token=False) 2629 columns = self._parse_csv(self._parse_id_var) if self._match(TokenType.ALIAS) else [] 2630 table_alias: t.Optional[exp.TableAlias] = self.expression( 2631 exp.TableAlias, this=table, columns=columns 2632 ) 2633 elif isinstance(this, (exp.Subquery, exp.Unnest)) and this.alias: 2634 # We move the alias from the lateral's child node to the lateral itself 2635 table_alias = this.args["alias"].pop() 2636 else: 2637 table_alias = self._parse_table_alias() 2638 2639 return self.expression( 2640 exp.Lateral, 2641 this=this, 2642 view=view, 2643 outer=outer, 2644 alias=table_alias, 2645 cross_apply=cross_apply, 2646 ) 2647 2648 def _parse_join_parts( 2649 self, 2650 ) -> t.Tuple[t.Optional[Token], t.Optional[Token], t.Optional[Token]]: 2651 return ( 2652 self._match_set(self.JOIN_METHODS) and self._prev, 2653 self._match_set(self.JOIN_SIDES) and self._prev, 2654 self._match_set(self.JOIN_KINDS) and self._prev, 2655 ) 2656 2657 def _parse_join( 2658 self, skip_join_token: bool = False, parse_bracket: bool = False 2659 ) -> t.Optional[exp.Join]: 2660 if self._match(TokenType.COMMA): 2661 return self.expression(exp.Join, this=self._parse_table()) 2662 2663 index = self._index 2664 method, side, kind = self._parse_join_parts() 2665 hint = self._prev.text if self._match_texts(self.JOIN_HINTS) else None 2666 join = self._match(TokenType.JOIN) 2667 2668 if not skip_join_token and not join: 2669 self._retreat(index) 2670 kind = None 2671 method = None 2672 side = None 2673 2674 outer_apply = self._match_pair(TokenType.OUTER, TokenType.APPLY, False) 2675 cross_apply = self._match_pair(TokenType.CROSS, TokenType.APPLY, False) 2676 2677 if not skip_join_token and not join and not outer_apply and not cross_apply: 2678 return None 2679 2680 kwargs: t.Dict[str, t.Any] = {"this": self._parse_table(parse_bracket=parse_bracket)} 2681 2682 if method: 2683 kwargs["method"] = method.text 2684 if side: 2685 kwargs["side"] = side.text 2686 if kind: 2687 kwargs["kind"] = kind.text 2688 if hint: 2689 kwargs["hint"] = hint 2690 2691 if self._match(TokenType.ON): 2692 kwargs["on"] = self._parse_conjunction() 2693 elif self._match(TokenType.USING): 2694 kwargs["using"] = self._parse_wrapped_id_vars() 2695 elif not (kind and kind.token_type == TokenType.CROSS): 2696 index = self._index 2697 join = self._parse_join() 2698 2699 if join and self._match(TokenType.ON): 2700 kwargs["on"] = self._parse_conjunction() 2701 elif join and self._match(TokenType.USING): 2702 kwargs["using"] = self._parse_wrapped_id_vars() 2703 else: 2704 join = None 2705 self._retreat(index) 2706 2707 kwargs["this"].set("joins", [join] if join else None) 2708 2709 comments = [c for token in (method, side, kind) if token for c in token.comments] 2710 return self.expression(exp.Join, comments=comments, **kwargs) 2711 2712 def _parse_opclass(self) -> t.Optional[exp.Expression]: 2713 this = self._parse_conjunction() 2714 if self._match_texts(self.OPCLASS_FOLLOW_KEYWORDS, advance=False): 2715 return this 2716 2717 if not self._match_set(self.OPTYPE_FOLLOW_TOKENS, advance=False): 2718 return self.expression(exp.Opclass, this=this, expression=self._parse_table_parts()) 2719 2720 return this 2721 2722 def _parse_index( 2723 self, 2724 index: t.Optional[exp.Expression] = None, 2725 ) -> t.Optional[exp.Index]: 2726 if index: 2727 unique = None 2728 primary = None 2729 amp = None 2730 2731 self._match(TokenType.ON) 2732 self._match(TokenType.TABLE) # hive 2733 table = self._parse_table_parts(schema=True) 2734 else: 2735 unique = self._match(TokenType.UNIQUE) 2736 primary = self._match_text_seq("PRIMARY") 2737 amp = self._match_text_seq("AMP") 2738 2739 if not self._match(TokenType.INDEX): 2740 return None 2741 2742 index = self._parse_id_var() 2743 table = None 2744 2745 using = self._parse_var(any_token=True) if self._match(TokenType.USING) else None 2746 2747 if self._match(TokenType.L_PAREN, advance=False): 2748 columns = self._parse_wrapped_csv(lambda: self._parse_ordered(self._parse_opclass)) 2749 else: 2750 columns = None 2751 2752 include = self._parse_wrapped_id_vars() if self._match_text_seq("INCLUDE") else None 2753 2754 return self.expression( 2755 exp.Index, 2756 this=index, 2757 table=table, 2758 using=using, 2759 columns=columns, 2760 unique=unique, 2761 primary=primary, 2762 amp=amp, 2763 include=include, 2764 partition_by=self._parse_partition_by(), 2765 where=self._parse_where(), 2766 ) 2767 2768 def _parse_table_hints(self) -> t.Optional[t.List[exp.Expression]]: 2769 hints: t.List[exp.Expression] = [] 2770 if self._match_pair(TokenType.WITH, TokenType.L_PAREN): 2771 # https://learn.microsoft.com/en-us/sql/t-sql/queries/hints-transact-sql-table?view=sql-server-ver16 2772 hints.append( 2773 self.expression( 2774 exp.WithTableHint, 2775 expressions=self._parse_csv( 2776 lambda: self._parse_function() or self._parse_var(any_token=True) 2777 ), 2778 ) 2779 ) 2780 self._match_r_paren() 2781 else: 2782 # https://dev.mysql.com/doc/refman/8.0/en/index-hints.html 2783 while self._match_set(self.TABLE_INDEX_HINT_TOKENS): 2784 hint = exp.IndexTableHint(this=self._prev.text.upper()) 2785 2786 self._match_texts(("INDEX", "KEY")) 2787 if self._match(TokenType.FOR): 2788 hint.set("target", self._advance_any() and self._prev.text.upper()) 2789 2790 hint.set("expressions", self._parse_wrapped_id_vars()) 2791 hints.append(hint) 2792 2793 return hints or None 2794 2795 def _parse_table_part(self, schema: bool = False) -> t.Optional[exp.Expression]: 2796 return ( 2797 (not schema and self._parse_function(optional_parens=False)) 2798 or self._parse_id_var(any_token=False) 2799 or self._parse_string_as_identifier() 2800 or self._parse_placeholder() 2801 ) 2802 2803 def _parse_table_parts(self, schema: bool = False, is_db_reference: bool = False) -> exp.Table: 2804 catalog = None 2805 db = None 2806 table: t.Optional[exp.Expression | str] = self._parse_table_part(schema=schema) 2807 2808 while self._match(TokenType.DOT): 2809 if catalog: 2810 # This allows nesting the table in arbitrarily many dot expressions if needed 2811 table = self.expression( 2812 exp.Dot, this=table, expression=self._parse_table_part(schema=schema) 2813 ) 2814 else: 2815 catalog = db 2816 db = table 2817 table = self._parse_table_part(schema=schema) or "" 2818 2819 if is_db_reference: 2820 catalog = db 2821 db = table 2822 table = None 2823 2824 if not table and not is_db_reference: 2825 self.raise_error(f"Expected table name but got {self._curr}") 2826 if not db and is_db_reference: 2827 self.raise_error(f"Expected database name but got {self._curr}") 2828 2829 return self.expression( 2830 exp.Table, this=table, db=db, catalog=catalog, pivots=self._parse_pivots() 2831 ) 2832 2833 def _parse_table( 2834 self, 2835 schema: bool = False, 2836 joins: bool = False, 2837 alias_tokens: t.Optional[t.Collection[TokenType]] = None, 2838 parse_bracket: bool = False, 2839 is_db_reference: bool = False, 2840 ) -> t.Optional[exp.Expression]: 2841 lateral = self._parse_lateral() 2842 if lateral: 2843 return lateral 2844 2845 unnest = self._parse_unnest() 2846 if unnest: 2847 return unnest 2848 2849 values = self._parse_derived_table_values() 2850 if values: 2851 return values 2852 2853 subquery = self._parse_select(table=True) 2854 if subquery: 2855 if not subquery.args.get("pivots"): 2856 subquery.set("pivots", self._parse_pivots()) 2857 return subquery 2858 2859 bracket = parse_bracket and self._parse_bracket(None) 2860 bracket = self.expression(exp.Table, this=bracket) if bracket else None 2861 this = t.cast( 2862 exp.Expression, 2863 bracket 2864 or self._parse_bracket( 2865 self._parse_table_parts(schema=schema, is_db_reference=is_db_reference) 2866 ), 2867 ) 2868 2869 if schema: 2870 return self._parse_schema(this=this) 2871 2872 version = self._parse_version() 2873 2874 if version: 2875 this.set("version", version) 2876 2877 if self.dialect.ALIAS_POST_TABLESAMPLE: 2878 table_sample = self._parse_table_sample() 2879 2880 alias = self._parse_table_alias(alias_tokens=alias_tokens or self.TABLE_ALIAS_TOKENS) 2881 if alias: 2882 this.set("alias", alias) 2883 2884 if isinstance(this, exp.Table) and self._match_text_seq("AT"): 2885 return self.expression( 2886 exp.AtIndex, this=this.to_column(copy=False), expression=self._parse_id_var() 2887 ) 2888 2889 this.set("hints", self._parse_table_hints()) 2890 2891 if not this.args.get("pivots"): 2892 this.set("pivots", self._parse_pivots()) 2893 2894 if not self.dialect.ALIAS_POST_TABLESAMPLE: 2895 table_sample = self._parse_table_sample() 2896 2897 if table_sample: 2898 table_sample.set("this", this) 2899 this = table_sample 2900 2901 if joins: 2902 for join in iter(self._parse_join, None): 2903 this.append("joins", join) 2904 2905 if self._match_pair(TokenType.WITH, TokenType.ORDINALITY): 2906 this.set("ordinality", True) 2907 this.set("alias", self._parse_table_alias()) 2908 2909 return this 2910 2911 def _parse_version(self) -> t.Optional[exp.Version]: 2912 if self._match(TokenType.TIMESTAMP_SNAPSHOT): 2913 this = "TIMESTAMP" 2914 elif self._match(TokenType.VERSION_SNAPSHOT): 2915 this = "VERSION" 2916 else: 2917 return None 2918 2919 if self._match_set((TokenType.FROM, TokenType.BETWEEN)): 2920 kind = self._prev.text.upper() 2921 start = self._parse_bitwise() 2922 self._match_texts(("TO", "AND")) 2923 end = self._parse_bitwise() 2924 expression: t.Optional[exp.Expression] = self.expression( 2925 exp.Tuple, expressions=[start, end] 2926 ) 2927 elif self._match_text_seq("CONTAINED", "IN"): 2928 kind = "CONTAINED IN" 2929 expression = self.expression( 2930 exp.Tuple, expressions=self._parse_wrapped_csv(self._parse_bitwise) 2931 ) 2932 elif self._match(TokenType.ALL): 2933 kind = "ALL" 2934 expression = None 2935 else: 2936 self._match_text_seq("AS", "OF") 2937 kind = "AS OF" 2938 expression = self._parse_type() 2939 2940 return self.expression(exp.Version, this=this, expression=expression, kind=kind) 2941 2942 def _parse_unnest(self, with_alias: bool = True) -> t.Optional[exp.Unnest]: 2943 if not self._match(TokenType.UNNEST): 2944 return None 2945 2946 expressions = self._parse_wrapped_csv(self._parse_equality) 2947 offset = self._match_pair(TokenType.WITH, TokenType.ORDINALITY) 2948 2949 alias = self._parse_table_alias() if with_alias else None 2950 2951 if alias: 2952 if self.dialect.UNNEST_COLUMN_ONLY: 2953 if alias.args.get("columns"): 2954 self.raise_error("Unexpected extra column alias in unnest.") 2955 2956 alias.set("columns", [alias.this]) 2957 alias.set("this", None) 2958 2959 columns = alias.args.get("columns") or [] 2960 if offset and len(expressions) < len(columns): 2961 offset = columns.pop() 2962 2963 if not offset and self._match_pair(TokenType.WITH, TokenType.OFFSET): 2964 self._match(TokenType.ALIAS) 2965 offset = self._parse_id_var( 2966 any_token=False, tokens=self.UNNEST_OFFSET_ALIAS_TOKENS 2967 ) or exp.to_identifier("offset") 2968 2969 return self.expression(exp.Unnest, expressions=expressions, alias=alias, offset=offset) 2970 2971 def _parse_derived_table_values(self) -> t.Optional[exp.Values]: 2972 is_derived = self._match_pair(TokenType.L_PAREN, TokenType.VALUES) 2973 if not is_derived and not self._match(TokenType.VALUES): 2974 return None 2975 2976 expressions = self._parse_csv(self._parse_value) 2977 alias = self._parse_table_alias() 2978 2979 if is_derived: 2980 self._match_r_paren() 2981 2982 return self.expression( 2983 exp.Values, expressions=expressions, alias=alias or self._parse_table_alias() 2984 ) 2985 2986 def _parse_table_sample(self, as_modifier: bool = False) -> t.Optional[exp.TableSample]: 2987 if not self._match(TokenType.TABLE_SAMPLE) and not ( 2988 as_modifier and self._match_text_seq("USING", "SAMPLE") 2989 ): 2990 return None 2991 2992 bucket_numerator = None 2993 bucket_denominator = None 2994 bucket_field = None 2995 percent = None 2996 size = None 2997 seed = None 2998 2999 method = self._parse_var(tokens=(TokenType.ROW,), upper=True) 3000 matched_l_paren = self._match(TokenType.L_PAREN) 3001 3002 if self.TABLESAMPLE_CSV: 3003 num = None 3004 expressions = self._parse_csv(self._parse_primary) 3005 else: 3006 expressions = None 3007 num = ( 3008 self._parse_factor() 3009 if self._match(TokenType.NUMBER, advance=False) 3010 else self._parse_primary() or self._parse_placeholder() 3011 ) 3012 3013 if self._match_text_seq("BUCKET"): 3014 bucket_numerator = self._parse_number() 3015 self._match_text_seq("OUT", "OF") 3016 bucket_denominator = bucket_denominator = self._parse_number() 3017 self._match(TokenType.ON) 3018 bucket_field = self._parse_field() 3019 elif self._match_set((TokenType.PERCENT, TokenType.MOD)): 3020 percent = num 3021 elif self._match(TokenType.ROWS) or not self.dialect.TABLESAMPLE_SIZE_IS_PERCENT: 3022 size = num 3023 else: 3024 percent = num 3025 3026 if matched_l_paren: 3027 self._match_r_paren() 3028 3029 if self._match(TokenType.L_PAREN): 3030 method = self._parse_var(upper=True) 3031 seed = self._match(TokenType.COMMA) and self._parse_number() 3032 self._match_r_paren() 3033 elif self._match_texts(("SEED", "REPEATABLE")): 3034 seed = self._parse_wrapped(self._parse_number) 3035 3036 return self.expression( 3037 exp.TableSample, 3038 expressions=expressions, 3039 method=method, 3040 bucket_numerator=bucket_numerator, 3041 bucket_denominator=bucket_denominator, 3042 bucket_field=bucket_field, 3043 percent=percent, 3044 size=size, 3045 seed=seed, 3046 ) 3047 3048 def _parse_pivots(self) -> t.Optional[t.List[exp.Pivot]]: 3049 return list(iter(self._parse_pivot, None)) or None 3050 3051 def _parse_joins(self) -> t.Optional[t.List[exp.Join]]: 3052 return list(iter(self._parse_join, None)) or None 3053 3054 # https://duckdb.org/docs/sql/statements/pivot 3055 def _parse_simplified_pivot(self) -> exp.Pivot: 3056 def _parse_on() -> t.Optional[exp.Expression]: 3057 this = self._parse_bitwise() 3058 return self._parse_in(this) if self._match(TokenType.IN) else this 3059 3060 this = self._parse_table() 3061 expressions = self._match(TokenType.ON) and self._parse_csv(_parse_on) 3062 using = self._match(TokenType.USING) and self._parse_csv( 3063 lambda: self._parse_alias(self._parse_function()) 3064 ) 3065 group = self._parse_group() 3066 return self.expression( 3067 exp.Pivot, this=this, expressions=expressions, using=using, group=group 3068 ) 3069 3070 def _parse_pivot_in(self) -> exp.In: 3071 def _parse_aliased_expression() -> t.Optional[exp.Expression]: 3072 this = self._parse_conjunction() 3073 3074 self._match(TokenType.ALIAS) 3075 alias = self._parse_field() 3076 if alias: 3077 return self.expression(exp.PivotAlias, this=this, alias=alias) 3078 3079 return this 3080 3081 value = self._parse_column() 3082 3083 if not self._match_pair(TokenType.IN, TokenType.L_PAREN): 3084 self.raise_error("Expecting IN (") 3085 3086 aliased_expressions = self._parse_csv(_parse_aliased_expression) 3087 3088 self._match_r_paren() 3089 return self.expression(exp.In, this=value, expressions=aliased_expressions) 3090 3091 def _parse_pivot(self) -> t.Optional[exp.Pivot]: 3092 index = self._index 3093 include_nulls = None 3094 3095 if self._match(TokenType.PIVOT): 3096 unpivot = False 3097 elif self._match(TokenType.UNPIVOT): 3098 unpivot = True 3099 3100 # https://docs.databricks.com/en/sql/language-manual/sql-ref-syntax-qry-select-unpivot.html#syntax 3101 if self._match_text_seq("INCLUDE", "NULLS"): 3102 include_nulls = True 3103 elif self._match_text_seq("EXCLUDE", "NULLS"): 3104 include_nulls = False 3105 else: 3106 return None 3107 3108 expressions = [] 3109 3110 if not self._match(TokenType.L_PAREN): 3111 self._retreat(index) 3112 return None 3113 3114 if unpivot: 3115 expressions = self._parse_csv(self._parse_column) 3116 else: 3117 expressions = self._parse_csv(lambda: self._parse_alias(self._parse_function())) 3118 3119 if not expressions: 3120 self.raise_error("Failed to parse PIVOT's aggregation list") 3121 3122 if not self._match(TokenType.FOR): 3123 self.raise_error("Expecting FOR") 3124 3125 field = self._parse_pivot_in() 3126 3127 self._match_r_paren() 3128 3129 pivot = self.expression( 3130 exp.Pivot, 3131 expressions=expressions, 3132 field=field, 3133 unpivot=unpivot, 3134 include_nulls=include_nulls, 3135 ) 3136 3137 if not self._match_set((TokenType.PIVOT, TokenType.UNPIVOT), advance=False): 3138 pivot.set("alias", self._parse_table_alias()) 3139 3140 if not unpivot: 3141 names = self._pivot_column_names(t.cast(t.List[exp.Expression], expressions)) 3142 3143 columns: t.List[exp.Expression] = [] 3144 for fld in pivot.args["field"].expressions: 3145 field_name = fld.sql() if self.IDENTIFY_PIVOT_STRINGS else fld.alias_or_name 3146 for name in names: 3147 if self.PREFIXED_PIVOT_COLUMNS: 3148 name = f"{name}_{field_name}" if name else field_name 3149 else: 3150 name = f"{field_name}_{name}" if name else field_name 3151 3152 columns.append(exp.to_identifier(name)) 3153 3154 pivot.set("columns", columns) 3155 3156 return pivot 3157 3158 def _pivot_column_names(self, aggregations: t.List[exp.Expression]) -> t.List[str]: 3159 return [agg.alias for agg in aggregations] 3160 3161 def _parse_where(self, skip_where_token: bool = False) -> t.Optional[exp.Where]: 3162 if not skip_where_token and not self._match(TokenType.WHERE): 3163 return None 3164 3165 return self.expression( 3166 exp.Where, comments=self._prev_comments, this=self._parse_conjunction() 3167 ) 3168 3169 def _parse_group(self, skip_group_by_token: bool = False) -> t.Optional[exp.Group]: 3170 if not skip_group_by_token and not self._match(TokenType.GROUP_BY): 3171 return None 3172 3173 elements = defaultdict(list) 3174 3175 if self._match(TokenType.ALL): 3176 return self.expression(exp.Group, all=True) 3177 3178 while True: 3179 expressions = self._parse_csv(self._parse_conjunction) 3180 if expressions: 3181 elements["expressions"].extend(expressions) 3182 3183 grouping_sets = self._parse_grouping_sets() 3184 if grouping_sets: 3185 elements["grouping_sets"].extend(grouping_sets) 3186 3187 rollup = None 3188 cube = None 3189 totals = None 3190 3191 index = self._index 3192 with_ = self._match(TokenType.WITH) 3193 if self._match(TokenType.ROLLUP): 3194 rollup = with_ or self._parse_wrapped_csv(self._parse_column) 3195 elements["rollup"].extend(ensure_list(rollup)) 3196 3197 if self._match(TokenType.CUBE): 3198 cube = with_ or self._parse_wrapped_csv(self._parse_column) 3199 elements["cube"].extend(ensure_list(cube)) 3200 3201 if self._match_text_seq("TOTALS"): 3202 totals = True 3203 elements["totals"] = True # type: ignore 3204 3205 if not (grouping_sets or rollup or cube or totals): 3206 if with_: 3207 self._retreat(index) 3208 break 3209 3210 return self.expression(exp.Group, **elements) # type: ignore 3211 3212 def _parse_grouping_sets(self) -> t.Optional[t.List[exp.Expression]]: 3213 if not self._match(TokenType.GROUPING_SETS): 3214 return None 3215 3216 return self._parse_wrapped_csv(self._parse_grouping_set) 3217 3218 def _parse_grouping_set(self) -> t.Optional[exp.Expression]: 3219 if self._match(TokenType.L_PAREN): 3220 grouping_set = self._parse_csv(self._parse_column) 3221 self._match_r_paren() 3222 return self.expression(exp.Tuple, expressions=grouping_set) 3223 3224 return self._parse_column() 3225 3226 def _parse_having(self, skip_having_token: bool = False) -> t.Optional[exp.Having]: 3227 if not skip_having_token and not self._match(TokenType.HAVING): 3228 return None 3229 return self.expression(exp.Having, this=self._parse_conjunction()) 3230 3231 def _parse_qualify(self) -> t.Optional[exp.Qualify]: 3232 if not self._match(TokenType.QUALIFY): 3233 return None 3234 return self.expression(exp.Qualify, this=self._parse_conjunction()) 3235 3236 def _parse_connect(self, skip_start_token: bool = False) -> t.Optional[exp.Connect]: 3237 if skip_start_token: 3238 start = None 3239 elif self._match(TokenType.START_WITH): 3240 start = self._parse_conjunction() 3241 else: 3242 return None 3243 3244 self._match(TokenType.CONNECT_BY) 3245 self.NO_PAREN_FUNCTION_PARSERS["PRIOR"] = lambda self: self.expression( 3246 exp.Prior, this=self._parse_bitwise() 3247 ) 3248 connect = self._parse_conjunction() 3249 self.NO_PAREN_FUNCTION_PARSERS.pop("PRIOR") 3250 3251 if not start and self._match(TokenType.START_WITH): 3252 start = self._parse_conjunction() 3253 3254 return self.expression(exp.Connect, start=start, connect=connect) 3255 3256 def _parse_name_as_expression(self) -> exp.Alias: 3257 return self.expression( 3258 exp.Alias, 3259 alias=self._parse_id_var(any_token=True), 3260 this=self._match(TokenType.ALIAS) and self._parse_conjunction(), 3261 ) 3262 3263 def _parse_interpolate(self) -> t.Optional[t.List[exp.Expression]]: 3264 if self._match_text_seq("INTERPOLATE"): 3265 return self._parse_wrapped_csv(self._parse_name_as_expression) 3266 return None 3267 3268 def _parse_order( 3269 self, this: t.Optional[exp.Expression] = None, skip_order_token: bool = False 3270 ) -> t.Optional[exp.Expression]: 3271 siblings = None 3272 if not skip_order_token and not self._match(TokenType.ORDER_BY): 3273 if not self._match(TokenType.ORDER_SIBLINGS_BY): 3274 return this 3275 3276 siblings = True 3277 3278 return self.expression( 3279 exp.Order, 3280 this=this, 3281 expressions=self._parse_csv(self._parse_ordered), 3282 interpolate=self._parse_interpolate(), 3283 siblings=siblings, 3284 ) 3285 3286 def _parse_sort(self, exp_class: t.Type[E], token: TokenType) -> t.Optional[E]: 3287 if not self._match(token): 3288 return None 3289 return self.expression(exp_class, expressions=self._parse_csv(self._parse_ordered)) 3290 3291 def _parse_ordered(self, parse_method: t.Optional[t.Callable] = None) -> exp.Ordered: 3292 this = parse_method() if parse_method else self._parse_conjunction() 3293 3294 asc = self._match(TokenType.ASC) 3295 desc = self._match(TokenType.DESC) or (asc and False) 3296 3297 is_nulls_first = self._match_text_seq("NULLS", "FIRST") 3298 is_nulls_last = self._match_text_seq("NULLS", "LAST") 3299 3300 nulls_first = is_nulls_first or False 3301 explicitly_null_ordered = is_nulls_first or is_nulls_last 3302 3303 if ( 3304 not explicitly_null_ordered 3305 and ( 3306 (not desc and self.dialect.NULL_ORDERING == "nulls_are_small") 3307 or (desc and self.dialect.NULL_ORDERING != "nulls_are_small") 3308 ) 3309 and self.dialect.NULL_ORDERING != "nulls_are_last" 3310 ): 3311 nulls_first = True 3312 3313 if self._match_text_seq("WITH", "FILL"): 3314 with_fill = self.expression( 3315 exp.WithFill, 3316 **{ # type: ignore 3317 "from": self._match(TokenType.FROM) and self._parse_bitwise(), 3318 "to": self._match_text_seq("TO") and self._parse_bitwise(), 3319 "step": self._match_text_seq("STEP") and self._parse_bitwise(), 3320 }, 3321 ) 3322 else: 3323 with_fill = None 3324 3325 return self.expression( 3326 exp.Ordered, this=this, desc=desc, nulls_first=nulls_first, with_fill=with_fill 3327 ) 3328 3329 def _parse_limit( 3330 self, this: t.Optional[exp.Expression] = None, top: bool = False 3331 ) -> t.Optional[exp.Expression]: 3332 if self._match(TokenType.TOP if top else TokenType.LIMIT): 3333 comments = self._prev_comments 3334 if top: 3335 limit_paren = self._match(TokenType.L_PAREN) 3336 expression = self._parse_term() if limit_paren else self._parse_number() 3337 3338 if limit_paren: 3339 self._match_r_paren() 3340 else: 3341 expression = self._parse_term() 3342 3343 if self._match(TokenType.COMMA): 3344 offset = expression 3345 expression = self._parse_term() 3346 else: 3347 offset = None 3348 3349 limit_exp = self.expression( 3350 exp.Limit, 3351 this=this, 3352 expression=expression, 3353 offset=offset, 3354 comments=comments, 3355 expressions=self._parse_limit_by(), 3356 ) 3357 3358 return limit_exp 3359 3360 if self._match(TokenType.FETCH): 3361 direction = self._match_set((TokenType.FIRST, TokenType.NEXT)) 3362 direction = self._prev.text.upper() if direction else "FIRST" 3363 3364 count = self._parse_field(tokens=self.FETCH_TOKENS) 3365 percent = self._match(TokenType.PERCENT) 3366 3367 self._match_set((TokenType.ROW, TokenType.ROWS)) 3368 3369 only = self._match_text_seq("ONLY") 3370 with_ties = self._match_text_seq("WITH", "TIES") 3371 3372 if only and with_ties: 3373 self.raise_error("Cannot specify both ONLY and WITH TIES in FETCH clause") 3374 3375 return self.expression( 3376 exp.Fetch, 3377 direction=direction, 3378 count=count, 3379 percent=percent, 3380 with_ties=with_ties, 3381 ) 3382 3383 return this 3384 3385 def _parse_offset(self, this: t.Optional[exp.Expression] = None) -> t.Optional[exp.Expression]: 3386 if not self._match(TokenType.OFFSET): 3387 return this 3388 3389 count = self._parse_term() 3390 self._match_set((TokenType.ROW, TokenType.ROWS)) 3391 3392 return self.expression( 3393 exp.Offset, this=this, expression=count, expressions=self._parse_limit_by() 3394 ) 3395 3396 def _parse_limit_by(self) -> t.Optional[t.List[exp.Expression]]: 3397 return self._match_text_seq("BY") and self._parse_csv(self._parse_bitwise) 3398 3399 def _parse_locks(self) -> t.List[exp.Lock]: 3400 locks = [] 3401 while True: 3402 if self._match_text_seq("FOR", "UPDATE"): 3403 update = True 3404 elif self._match_text_seq("FOR", "SHARE") or self._match_text_seq( 3405 "LOCK", "IN", "SHARE", "MODE" 3406 ): 3407 update = False 3408 else: 3409 break 3410 3411 expressions = None 3412 if self._match_text_seq("OF"): 3413 expressions = self._parse_csv(lambda: self._parse_table(schema=True)) 3414 3415 wait: t.Optional[bool | exp.Expression] = None 3416 if self._match_text_seq("NOWAIT"): 3417 wait = True 3418 elif self._match_text_seq("WAIT"): 3419 wait = self._parse_primary() 3420 elif self._match_text_seq("SKIP", "LOCKED"): 3421 wait = False 3422 3423 locks.append( 3424 self.expression(exp.Lock, update=update, expressions=expressions, wait=wait) 3425 ) 3426 3427 return locks 3428 3429 def _parse_set_operations(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 3430 while this and self._match_set(self.SET_OPERATIONS): 3431 token_type = self._prev.token_type 3432 3433 if token_type == TokenType.UNION: 3434 operation = exp.Union 3435 elif token_type == TokenType.EXCEPT: 3436 operation = exp.Except 3437 else: 3438 operation = exp.Intersect 3439 3440 comments = self._prev.comments 3441 distinct = self._match(TokenType.DISTINCT) or not self._match(TokenType.ALL) 3442 by_name = self._match_text_seq("BY", "NAME") 3443 expression = self._parse_select(nested=True, parse_set_operation=False) 3444 3445 this = self.expression( 3446 operation, 3447 comments=comments, 3448 this=this, 3449 distinct=distinct, 3450 by_name=by_name, 3451 expression=expression, 3452 ) 3453 3454 if isinstance(this, exp.Union) and self.MODIFIERS_ATTACHED_TO_UNION: 3455 expression = this.expression 3456 3457 if expression: 3458 for arg in self.UNION_MODIFIERS: 3459 expr = expression.args.get(arg) 3460 if expr: 3461 this.set(arg, expr.pop()) 3462 3463 return this 3464 3465 def _parse_expression(self) -> t.Optional[exp.Expression]: 3466 return self._parse_alias(self._parse_conjunction()) 3467 3468 def _parse_conjunction(self) -> t.Optional[exp.Expression]: 3469 return self._parse_tokens(self._parse_equality, self.CONJUNCTION) 3470 3471 def _parse_equality(self) -> t.Optional[exp.Expression]: 3472 return self._parse_tokens(self._parse_comparison, self.EQUALITY) 3473 3474 def _parse_comparison(self) -> t.Optional[exp.Expression]: 3475 return self._parse_tokens(self._parse_range, self.COMPARISON) 3476 3477 def _parse_range(self, this: t.Optional[exp.Expression] = None) -> t.Optional[exp.Expression]: 3478 this = this or self._parse_bitwise() 3479 negate = self._match(TokenType.NOT) 3480 3481 if self._match_set(self.RANGE_PARSERS): 3482 expression = self.RANGE_PARSERS[self._prev.token_type](self, this) 3483 if not expression: 3484 return this 3485 3486 this = expression 3487 elif self._match(TokenType.ISNULL): 3488 this = self.expression(exp.Is, this=this, expression=exp.Null()) 3489 3490 # Postgres supports ISNULL and NOTNULL for conditions. 3491 # https://blog.andreiavram.ro/postgresql-null-composite-type/ 3492 if self._match(TokenType.NOTNULL): 3493 this = self.expression(exp.Is, this=this, expression=exp.Null()) 3494 this = self.expression(exp.Not, this=this) 3495 3496 if negate: 3497 this = self.expression(exp.Not, this=this) 3498 3499 if self._match(TokenType.IS): 3500 this = self._parse_is(this) 3501 3502 return this 3503 3504 def _parse_is(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 3505 index = self._index - 1 3506 negate = self._match(TokenType.NOT) 3507 3508 if self._match_text_seq("DISTINCT", "FROM"): 3509 klass = exp.NullSafeEQ if negate else exp.NullSafeNEQ 3510 return self.expression(klass, this=this, expression=self._parse_conjunction()) 3511 3512 expression = self._parse_null() or self._parse_boolean() 3513 if not expression: 3514 self._retreat(index) 3515 return None 3516 3517 this = self.expression(exp.Is, this=this, expression=expression) 3518 return self.expression(exp.Not, this=this) if negate else this 3519 3520 def _parse_in(self, this: t.Optional[exp.Expression], alias: bool = False) -> exp.In: 3521 unnest = self._parse_unnest(with_alias=False) 3522 if unnest: 3523 this = self.expression(exp.In, this=this, unnest=unnest) 3524 elif self._match_set((TokenType.L_PAREN, TokenType.L_BRACKET)): 3525 matched_l_paren = self._prev.token_type == TokenType.L_PAREN 3526 expressions = self._parse_csv(lambda: self._parse_select_or_expression(alias=alias)) 3527 3528 if len(expressions) == 1 and isinstance(expressions[0], exp.Subqueryable): 3529 this = self.expression(exp.In, this=this, query=expressions[0]) 3530 else: 3531 this = self.expression(exp.In, this=this, expressions=expressions) 3532 3533 if matched_l_paren: 3534 self._match_r_paren(this) 3535 elif not self._match(TokenType.R_BRACKET, expression=this): 3536 self.raise_error("Expecting ]") 3537 else: 3538 this = self.expression(exp.In, this=this, field=self._parse_field()) 3539 3540 return this 3541 3542 def _parse_between(self, this: t.Optional[exp.Expression]) -> exp.Between: 3543 low = self._parse_bitwise() 3544 self._match(TokenType.AND) 3545 high = self._parse_bitwise() 3546 return self.expression(exp.Between, this=this, low=low, high=high) 3547 3548 def _parse_escape(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 3549 if not self._match(TokenType.ESCAPE): 3550 return this 3551 return self.expression(exp.Escape, this=this, expression=self._parse_string()) 3552 3553 def _parse_interval(self, match_interval: bool = True) -> t.Optional[exp.Interval]: 3554 index = self._index 3555 3556 if not self._match(TokenType.INTERVAL) and match_interval: 3557 return None 3558 3559 if self._match(TokenType.STRING, advance=False): 3560 this = self._parse_primary() 3561 else: 3562 this = self._parse_term() 3563 3564 if not this or ( 3565 isinstance(this, exp.Column) 3566 and not this.table 3567 and not this.this.quoted 3568 and this.name.upper() == "IS" 3569 ): 3570 self._retreat(index) 3571 return None 3572 3573 unit = self._parse_function() or ( 3574 not self._match(TokenType.ALIAS, advance=False) 3575 and self._parse_var(any_token=True, upper=True) 3576 ) 3577 3578 # Most dialects support, e.g., the form INTERVAL '5' day, thus we try to parse 3579 # each INTERVAL expression into this canonical form so it's easy to transpile 3580 if this and this.is_number: 3581 this = exp.Literal.string(this.name) 3582 elif this and this.is_string: 3583 parts = this.name.split() 3584 3585 if len(parts) == 2: 3586 if unit: 3587 # This is not actually a unit, it's something else (e.g. a "window side") 3588 unit = None 3589 self._retreat(self._index - 1) 3590 3591 this = exp.Literal.string(parts[0]) 3592 unit = self.expression(exp.Var, this=parts[1].upper()) 3593 3594 return self.expression(exp.Interval, this=this, unit=unit) 3595 3596 def _parse_bitwise(self) -> t.Optional[exp.Expression]: 3597 this = self._parse_term() 3598 3599 while True: 3600 if self._match_set(self.BITWISE): 3601 this = self.expression( 3602 self.BITWISE[self._prev.token_type], 3603 this=this, 3604 expression=self._parse_term(), 3605 ) 3606 elif self.dialect.DPIPE_IS_STRING_CONCAT and self._match(TokenType.DPIPE): 3607 this = self.expression( 3608 exp.DPipe, 3609 this=this, 3610 expression=self._parse_term(), 3611 safe=not self.dialect.STRICT_STRING_CONCAT, 3612 ) 3613 elif self._match(TokenType.DQMARK): 3614 this = self.expression(exp.Coalesce, this=this, expressions=self._parse_term()) 3615 elif self._match_pair(TokenType.LT, TokenType.LT): 3616 this = self.expression( 3617 exp.BitwiseLeftShift, this=this, expression=self._parse_term() 3618 ) 3619 elif self._match_pair(TokenType.GT, TokenType.GT): 3620 this = self.expression( 3621 exp.BitwiseRightShift, this=this, expression=self._parse_term() 3622 ) 3623 else: 3624 break 3625 3626 return this 3627 3628 def _parse_term(self) -> t.Optional[exp.Expression]: 3629 return self._parse_tokens(self._parse_factor, self.TERM) 3630 3631 def _parse_factor(self) -> t.Optional[exp.Expression]: 3632 parse_method = self._parse_exponent if self.EXPONENT else self._parse_unary 3633 this = parse_method() 3634 3635 while self._match_set(self.FACTOR): 3636 this = self.expression( 3637 self.FACTOR[self._prev.token_type], 3638 this=this, 3639 comments=self._prev_comments, 3640 expression=parse_method(), 3641 ) 3642 if isinstance(this, exp.Div): 3643 this.args["typed"] = self.dialect.TYPED_DIVISION 3644 this.args["safe"] = self.dialect.SAFE_DIVISION 3645 3646 return this 3647 3648 def _parse_exponent(self) -> t.Optional[exp.Expression]: 3649 return self._parse_tokens(self._parse_unary, self.EXPONENT) 3650 3651 def _parse_unary(self) -> t.Optional[exp.Expression]: 3652 if self._match_set(self.UNARY_PARSERS): 3653 return self.UNARY_PARSERS[self._prev.token_type](self) 3654 return self._parse_at_time_zone(self._parse_type()) 3655 3656 def _parse_type(self, parse_interval: bool = True) -> t.Optional[exp.Expression]: 3657 interval = parse_interval and self._parse_interval() 3658 if interval: 3659 # Convert INTERVAL 'val_1' unit_1 ... 'val_n' unit_n into a sum of intervals 3660 while self._match_set((TokenType.STRING, TokenType.NUMBER), advance=False): 3661 interval = self.expression( # type: ignore 3662 exp.Add, this=interval, expression=self._parse_interval(match_interval=False) 3663 ) 3664 3665 return interval 3666 3667 index = self._index 3668 data_type = self._parse_types(check_func=True, allow_identifiers=False) 3669 this = self._parse_column() 3670 3671 if data_type: 3672 if isinstance(this, exp.Literal): 3673 parser = self.TYPE_LITERAL_PARSERS.get(data_type.this) 3674 if parser: 3675 return parser(self, this, data_type) 3676 return self.expression(exp.Cast, this=this, to=data_type) 3677 if not data_type.expressions: 3678 self._retreat(index) 3679 return self._parse_column() 3680 return self._parse_column_ops(data_type) 3681 3682 return this and self._parse_column_ops(this) 3683 3684 def _parse_type_size(self) -> t.Optional[exp.DataTypeParam]: 3685 this = self._parse_type() 3686 if not this: 3687 return None 3688 3689 return self.expression( 3690 exp.DataTypeParam, this=this, expression=self._parse_var(any_token=True) 3691 ) 3692 3693 def _parse_types( 3694 self, check_func: bool = False, schema: bool = False, allow_identifiers: bool = True 3695 ) -> t.Optional[exp.Expression]: 3696 index = self._index 3697 3698 prefix = self._match_text_seq("SYSUDTLIB", ".") 3699 3700 if not self._match_set(self.TYPE_TOKENS): 3701 identifier = allow_identifiers and self._parse_id_var( 3702 any_token=False, tokens=(TokenType.VAR,) 3703 ) 3704 if identifier: 3705 tokens = self.dialect.tokenize(identifier.name) 3706 3707 if len(tokens) != 1: 3708 self.raise_error("Unexpected identifier", self._prev) 3709 3710 if tokens[0].token_type in self.TYPE_TOKENS: 3711 self._prev = tokens[0] 3712 elif self.dialect.SUPPORTS_USER_DEFINED_TYPES: 3713 type_name = identifier.name 3714 3715 while self._match(TokenType.DOT): 3716 type_name = f"{type_name}.{self._advance_any() and self._prev.text}" 3717 3718 return exp.DataType.build(type_name, udt=True) 3719 else: 3720 self._retreat(self._index - 1) 3721 return None 3722 else: 3723 return None 3724 3725 type_token = self._prev.token_type 3726 3727 if type_token == TokenType.PSEUDO_TYPE: 3728 return self.expression(exp.PseudoType, this=self._prev.text.upper()) 3729 3730 if type_token == TokenType.OBJECT_IDENTIFIER: 3731 return self.expression(exp.ObjectIdentifier, this=self._prev.text.upper()) 3732 3733 nested = type_token in self.NESTED_TYPE_TOKENS 3734 is_struct = type_token in self.STRUCT_TYPE_TOKENS 3735 is_aggregate = type_token in self.AGGREGATE_TYPE_TOKENS 3736 expressions = None 3737 maybe_func = False 3738 3739 if self._match(TokenType.L_PAREN): 3740 if is_struct: 3741 expressions = self._parse_csv(self._parse_struct_types) 3742 elif nested: 3743 expressions = self._parse_csv( 3744 lambda: self._parse_types( 3745 check_func=check_func, schema=schema, allow_identifiers=allow_identifiers 3746 ) 3747 ) 3748 elif type_token in self.ENUM_TYPE_TOKENS: 3749 expressions = self._parse_csv(self._parse_equality) 3750 elif is_aggregate: 3751 func_or_ident = self._parse_function(anonymous=True) or self._parse_id_var( 3752 any_token=False, tokens=(TokenType.VAR,) 3753 ) 3754 if not func_or_ident or not self._match(TokenType.COMMA): 3755 return None 3756 expressions = self._parse_csv( 3757 lambda: self._parse_types( 3758 check_func=check_func, schema=schema, allow_identifiers=allow_identifiers 3759 ) 3760 ) 3761 expressions.insert(0, func_or_ident) 3762 else: 3763 expressions = self._parse_csv(self._parse_type_size) 3764 3765 if not expressions or not self._match(TokenType.R_PAREN): 3766 self._retreat(index) 3767 return None 3768 3769 maybe_func = True 3770 3771 this: t.Optional[exp.Expression] = None 3772 values: t.Optional[t.List[exp.Expression]] = None 3773 3774 if nested and self._match(TokenType.LT): 3775 if is_struct: 3776 expressions = self._parse_csv(lambda: self._parse_struct_types(type_required=True)) 3777 else: 3778 expressions = self._parse_csv( 3779 lambda: self._parse_types( 3780 check_func=check_func, schema=schema, allow_identifiers=allow_identifiers 3781 ) 3782 ) 3783 3784 if not self._match(TokenType.GT): 3785 self.raise_error("Expecting >") 3786 3787 if self._match_set((TokenType.L_BRACKET, TokenType.L_PAREN)): 3788 values = self._parse_csv(self._parse_conjunction) 3789 self._match_set((TokenType.R_BRACKET, TokenType.R_PAREN)) 3790 3791 if type_token in self.TIMESTAMPS: 3792 if self._match_text_seq("WITH", "TIME", "ZONE"): 3793 maybe_func = False 3794 tz_type = ( 3795 exp.DataType.Type.TIMETZ 3796 if type_token in self.TIMES 3797 else exp.DataType.Type.TIMESTAMPTZ 3798 ) 3799 this = exp.DataType(this=tz_type, expressions=expressions) 3800 elif self._match_text_seq("WITH", "LOCAL", "TIME", "ZONE"): 3801 maybe_func = False 3802 this = exp.DataType(this=exp.DataType.Type.TIMESTAMPLTZ, expressions=expressions) 3803 elif self._match_text_seq("WITHOUT", "TIME", "ZONE"): 3804 maybe_func = False 3805 elif type_token == TokenType.INTERVAL: 3806 unit = self._parse_var() 3807 3808 if self._match_text_seq("TO"): 3809 span = [exp.IntervalSpan(this=unit, expression=self._parse_var())] 3810 else: 3811 span = None 3812 3813 if span or not unit: 3814 this = self.expression( 3815 exp.DataType, this=exp.DataType.Type.INTERVAL, expressions=span 3816 ) 3817 else: 3818 this = self.expression(exp.DataType, this=self.expression(exp.Interval, unit=unit)) 3819 3820 if maybe_func and check_func: 3821 index2 = self._index 3822 peek = self._parse_string() 3823 3824 if not peek: 3825 self._retreat(index) 3826 return None 3827 3828 self._retreat(index2) 3829 3830 if not this: 3831 if self._match_text_seq("UNSIGNED"): 3832 unsigned_type_token = self.SIGNED_TO_UNSIGNED_TYPE_TOKEN.get(type_token) 3833 if not unsigned_type_token: 3834 self.raise_error(f"Cannot convert {type_token.value} to unsigned.") 3835 3836 type_token = unsigned_type_token or type_token 3837 3838 this = exp.DataType( 3839 this=exp.DataType.Type[type_token.value], 3840 expressions=expressions, 3841 nested=nested, 3842 values=values, 3843 prefix=prefix, 3844 ) 3845 3846 while self._match_pair(TokenType.L_BRACKET, TokenType.R_BRACKET): 3847 this = exp.DataType(this=exp.DataType.Type.ARRAY, expressions=[this], nested=True) 3848 3849 return this 3850 3851 def _parse_struct_types(self, type_required: bool = False) -> t.Optional[exp.Expression]: 3852 index = self._index 3853 this = self._parse_type(parse_interval=False) or self._parse_id_var() 3854 self._match(TokenType.COLON) 3855 column_def = self._parse_column_def(this) 3856 3857 if type_required and ( 3858 (isinstance(this, exp.Column) and this.this is column_def) or this is column_def 3859 ): 3860 self._retreat(index) 3861 return self._parse_types() 3862 3863 return column_def 3864 3865 def _parse_at_time_zone(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 3866 if not self._match_text_seq("AT", "TIME", "ZONE"): 3867 return this 3868 return self.expression(exp.AtTimeZone, this=this, zone=self._parse_unary()) 3869 3870 def _parse_column(self) -> t.Optional[exp.Expression]: 3871 this = self._parse_column_reference() 3872 return self._parse_column_ops(this) if this else self._parse_bracket(this) 3873 3874 def _parse_column_reference(self) -> t.Optional[exp.Expression]: 3875 this = self._parse_field() 3876 if isinstance(this, exp.Identifier): 3877 this = self.expression(exp.Column, this=this) 3878 return this 3879 3880 def _parse_column_ops(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 3881 this = self._parse_bracket(this) 3882 3883 while self._match_set(self.COLUMN_OPERATORS): 3884 op_token = self._prev.token_type 3885 op = self.COLUMN_OPERATORS.get(op_token) 3886 3887 if op_token == TokenType.DCOLON: 3888 field = self._parse_types() 3889 if not field: 3890 self.raise_error("Expected type") 3891 elif op and self._curr: 3892 field = self._parse_column_reference() 3893 else: 3894 field = self._parse_field(anonymous_func=True, any_token=True) 3895 3896 if isinstance(field, exp.Func): 3897 # bigquery allows function calls like x.y.count(...) 3898 # SAFE.SUBSTR(...) 3899 # https://cloud.google.com/bigquery/docs/reference/standard-sql/functions-reference#function_call_rules 3900 this = self._replace_columns_with_dots(this) 3901 3902 if op: 3903 this = op(self, this, field) 3904 elif isinstance(this, exp.Column) and not this.args.get("catalog"): 3905 this = self.expression( 3906 exp.Column, 3907 this=field, 3908 table=this.this, 3909 db=this.args.get("table"), 3910 catalog=this.args.get("db"), 3911 ) 3912 else: 3913 this = self.expression(exp.Dot, this=this, expression=field) 3914 this = self._parse_bracket(this) 3915 return this 3916 3917 def _parse_primary(self) -> t.Optional[exp.Expression]: 3918 if self._match_set(self.PRIMARY_PARSERS): 3919 token_type = self._prev.token_type 3920 primary = self.PRIMARY_PARSERS[token_type](self, self._prev) 3921 3922 if token_type == TokenType.STRING: 3923 expressions = [primary] 3924 while self._match(TokenType.STRING): 3925 expressions.append(exp.Literal.string(self._prev.text)) 3926 3927 if len(expressions) > 1: 3928 return self.expression(exp.Concat, expressions=expressions) 3929 3930 return primary 3931 3932 if self._match_pair(TokenType.DOT, TokenType.NUMBER): 3933 return exp.Literal.number(f"0.{self._prev.text}") 3934 3935 if self._match(TokenType.L_PAREN): 3936 comments = self._prev_comments 3937 query = self._parse_select() 3938 3939 if query: 3940 expressions = [query] 3941 else: 3942 expressions = self._parse_expressions() 3943 3944 this = self._parse_query_modifiers(seq_get(expressions, 0)) 3945 3946 if isinstance(this, exp.Subqueryable): 3947 this = self._parse_set_operations( 3948 self._parse_subquery(this=this, parse_alias=False) 3949 ) 3950 elif len(expressions) > 1: 3951 this = self.expression(exp.Tuple, expressions=expressions) 3952 else: 3953 this = self.expression(exp.Paren, this=self._parse_set_operations(this)) 3954 3955 if this: 3956 this.add_comments(comments) 3957 3958 self._match_r_paren(expression=this) 3959 return this 3960 3961 return None 3962 3963 def _parse_field( 3964 self, 3965 any_token: bool = False, 3966 tokens: t.Optional[t.Collection[TokenType]] = None, 3967 anonymous_func: bool = False, 3968 ) -> t.Optional[exp.Expression]: 3969 return ( 3970 self._parse_primary() 3971 or self._parse_function(anonymous=anonymous_func) 3972 or self._parse_id_var(any_token=any_token, tokens=tokens) 3973 ) 3974 3975 def _parse_function( 3976 self, 3977 functions: t.Optional[t.Dict[str, t.Callable]] = None, 3978 anonymous: bool = False, 3979 optional_parens: bool = True, 3980 ) -> t.Optional[exp.Expression]: 3981 # This allows us to also parse {fn <function>} syntax (Snowflake, MySQL support this) 3982 # See: https://community.snowflake.com/s/article/SQL-Escape-Sequences 3983 fn_syntax = False 3984 if ( 3985 self._match(TokenType.L_BRACE, advance=False) 3986 and self._next 3987 and self._next.text.upper() == "FN" 3988 ): 3989 self._advance(2) 3990 fn_syntax = True 3991 3992 func = self._parse_function_call( 3993 functions=functions, anonymous=anonymous, optional_parens=optional_parens 3994 ) 3995 3996 if fn_syntax: 3997 self._match(TokenType.R_BRACE) 3998 3999 return func 4000 4001 def _parse_function_call( 4002 self, 4003 functions: t.Optional[t.Dict[str, t.Callable]] = None, 4004 anonymous: bool = False, 4005 optional_parens: bool = True, 4006 ) -> t.Optional[exp.Expression]: 4007 if not self._curr: 4008 return None 4009 4010 comments = self._curr.comments 4011 token_type = self._curr.token_type 4012 this = self._curr.text 4013 upper = this.upper() 4014 4015 parser = self.NO_PAREN_FUNCTION_PARSERS.get(upper) 4016 if optional_parens and parser and token_type not in self.INVALID_FUNC_NAME_TOKENS: 4017 self._advance() 4018 return parser(self) 4019 4020 if not self._next or self._next.token_type != TokenType.L_PAREN: 4021 if optional_parens and token_type in self.NO_PAREN_FUNCTIONS: 4022 self._advance() 4023 return self.expression(self.NO_PAREN_FUNCTIONS[token_type]) 4024 4025 return None 4026 4027 if token_type not in self.FUNC_TOKENS: 4028 return None 4029 4030 self._advance(2) 4031 4032 parser = self.FUNCTION_PARSERS.get(upper) 4033 if parser and not anonymous: 4034 this = parser(self) 4035 else: 4036 subquery_predicate = self.SUBQUERY_PREDICATES.get(token_type) 4037 4038 if subquery_predicate and self._curr.token_type in (TokenType.SELECT, TokenType.WITH): 4039 this = self.expression(subquery_predicate, this=self._parse_select()) 4040 self._match_r_paren() 4041 return this 4042 4043 if functions is None: 4044 functions = self.FUNCTIONS 4045 4046 function = functions.get(upper) 4047 4048 alias = upper in self.FUNCTIONS_WITH_ALIASED_ARGS 4049 args = self._parse_csv(lambda: self._parse_lambda(alias=alias)) 4050 4051 if function and not anonymous: 4052 if "dialect" in function.__code__.co_varnames: 4053 func = function(args, dialect=self.dialect) 4054 else: 4055 func = function(args) 4056 4057 func = self.validate_expression(func, args) 4058 if not self.dialect.NORMALIZE_FUNCTIONS: 4059 func.meta["name"] = this 4060 4061 this = func 4062 else: 4063 this = self.expression(exp.Anonymous, this=this, expressions=args) 4064 4065 if isinstance(this, exp.Expression): 4066 this.add_comments(comments) 4067 4068 self._match_r_paren(this) 4069 return self._parse_window(this) 4070 4071 def _parse_function_parameter(self) -> t.Optional[exp.Expression]: 4072 return self._parse_column_def(self._parse_id_var()) 4073 4074 def _parse_user_defined_function( 4075 self, kind: t.Optional[TokenType] = None 4076 ) -> t.Optional[exp.Expression]: 4077 this = self._parse_id_var() 4078 4079 while self._match(TokenType.DOT): 4080 this = self.expression(exp.Dot, this=this, expression=self._parse_id_var()) 4081 4082 if not self._match(TokenType.L_PAREN): 4083 return this 4084 4085 expressions = self._parse_csv(self._parse_function_parameter) 4086 self._match_r_paren() 4087 return self.expression( 4088 exp.UserDefinedFunction, this=this, expressions=expressions, wrapped=True 4089 ) 4090 4091 def _parse_introducer(self, token: Token) -> exp.Introducer | exp.Identifier: 4092 literal = self._parse_primary() 4093 if literal: 4094 return self.expression(exp.Introducer, this=token.text, expression=literal) 4095 4096 return self.expression(exp.Identifier, this=token.text) 4097 4098 def _parse_session_parameter(self) -> exp.SessionParameter: 4099 kind = None 4100 this = self._parse_id_var() or self._parse_primary() 4101 4102 if this and self._match(TokenType.DOT): 4103 kind = this.name 4104 this = self._parse_var() or self._parse_primary() 4105 4106 return self.expression(exp.SessionParameter, this=this, kind=kind) 4107 4108 def _parse_lambda(self, alias: bool = False) -> t.Optional[exp.Expression]: 4109 index = self._index 4110 4111 if self._match(TokenType.L_PAREN): 4112 expressions = t.cast( 4113 t.List[t.Optional[exp.Expression]], self._parse_csv(self._parse_id_var) 4114 ) 4115 4116 if not self._match(TokenType.R_PAREN): 4117 self._retreat(index) 4118 else: 4119 expressions = [self._parse_id_var()] 4120 4121 if self._match_set(self.LAMBDAS): 4122 return self.LAMBDAS[self._prev.token_type](self, expressions) 4123 4124 self._retreat(index) 4125 4126 this: t.Optional[exp.Expression] 4127 4128 if self._match(TokenType.DISTINCT): 4129 this = self.expression( 4130 exp.Distinct, expressions=self._parse_csv(self._parse_conjunction) 4131 ) 4132 else: 4133 this = self._parse_select_or_expression(alias=alias) 4134 4135 return self._parse_limit( 4136 self._parse_order(self._parse_having_max(self._parse_respect_or_ignore_nulls(this))) 4137 ) 4138 4139 def _parse_schema(self, this: t.Optional[exp.Expression] = None) -> t.Optional[exp.Expression]: 4140 index = self._index 4141 4142 if not self.errors: 4143 try: 4144 if self._parse_select(nested=True): 4145 return this 4146 except ParseError: 4147 pass 4148 finally: 4149 self.errors.clear() 4150 self._retreat(index) 4151 4152 if not self._match(TokenType.L_PAREN): 4153 return this 4154 4155 args = self._parse_csv(lambda: self._parse_constraint() or self._parse_field_def()) 4156 4157 self._match_r_paren() 4158 return self.expression(exp.Schema, this=this, expressions=args) 4159 4160 def _parse_field_def(self) -> t.Optional[exp.Expression]: 4161 return self._parse_column_def(self._parse_field(any_token=True)) 4162 4163 def _parse_column_def(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 4164 # column defs are not really columns, they're identifiers 4165 if isinstance(this, exp.Column): 4166 this = this.this 4167 4168 kind = self._parse_types(schema=True) 4169 4170 if self._match_text_seq("FOR", "ORDINALITY"): 4171 return self.expression(exp.ColumnDef, this=this, ordinality=True) 4172 4173 constraints: t.List[exp.Expression] = [] 4174 4175 if not kind and self._match(TokenType.ALIAS): 4176 constraints.append( 4177 self.expression( 4178 exp.ComputedColumnConstraint, 4179 this=self._parse_conjunction(), 4180 persisted=self._match_text_seq("PERSISTED"), 4181 not_null=self._match_pair(TokenType.NOT, TokenType.NULL), 4182 ) 4183 ) 4184 elif kind and self._match_pair(TokenType.ALIAS, TokenType.L_PAREN, advance=False): 4185 self._match(TokenType.ALIAS) 4186 constraints.append( 4187 self.expression(exp.TransformColumnConstraint, this=self._parse_field()) 4188 ) 4189 4190 while True: 4191 constraint = self._parse_column_constraint() 4192 if not constraint: 4193 break 4194 constraints.append(constraint) 4195 4196 if not kind and not constraints: 4197 return this 4198 4199 return self.expression(exp.ColumnDef, this=this, kind=kind, constraints=constraints) 4200 4201 def _parse_auto_increment( 4202 self, 4203 ) -> exp.GeneratedAsIdentityColumnConstraint | exp.AutoIncrementColumnConstraint: 4204 start = None 4205 increment = None 4206 4207 if self._match(TokenType.L_PAREN, advance=False): 4208 args = self._parse_wrapped_csv(self._parse_bitwise) 4209 start = seq_get(args, 0) 4210 increment = seq_get(args, 1) 4211 elif self._match_text_seq("START"): 4212 start = self._parse_bitwise() 4213 self._match_text_seq("INCREMENT") 4214 increment = self._parse_bitwise() 4215 4216 if start and increment: 4217 return exp.GeneratedAsIdentityColumnConstraint(start=start, increment=increment) 4218 4219 return exp.AutoIncrementColumnConstraint() 4220 4221 def _parse_auto_property(self) -> t.Optional[exp.AutoRefreshProperty]: 4222 if not self._match_text_seq("REFRESH"): 4223 self._retreat(self._index - 1) 4224 return None 4225 return self.expression(exp.AutoRefreshProperty, this=self._parse_var(upper=True)) 4226 4227 def _parse_compress(self) -> exp.CompressColumnConstraint: 4228 if self._match(TokenType.L_PAREN, advance=False): 4229 return self.expression( 4230 exp.CompressColumnConstraint, this=self._parse_wrapped_csv(self._parse_bitwise) 4231 ) 4232 4233 return self.expression(exp.CompressColumnConstraint, this=self._parse_bitwise()) 4234 4235 def _parse_generated_as_identity( 4236 self, 4237 ) -> ( 4238 exp.GeneratedAsIdentityColumnConstraint 4239 | exp.ComputedColumnConstraint 4240 | exp.GeneratedAsRowColumnConstraint 4241 ): 4242 if self._match_text_seq("BY", "DEFAULT"): 4243 on_null = self._match_pair(TokenType.ON, TokenType.NULL) 4244 this = self.expression( 4245 exp.GeneratedAsIdentityColumnConstraint, this=False, on_null=on_null 4246 ) 4247 else: 4248 self._match_text_seq("ALWAYS") 4249 this = self.expression(exp.GeneratedAsIdentityColumnConstraint, this=True) 4250 4251 self._match(TokenType.ALIAS) 4252 4253 if self._match_text_seq("ROW"): 4254 start = self._match_text_seq("START") 4255 if not start: 4256 self._match(TokenType.END) 4257 hidden = self._match_text_seq("HIDDEN") 4258 return self.expression(exp.GeneratedAsRowColumnConstraint, start=start, hidden=hidden) 4259 4260 identity = self._match_text_seq("IDENTITY") 4261 4262 if self._match(TokenType.L_PAREN): 4263 if self._match(TokenType.START_WITH): 4264 this.set("start", self._parse_bitwise()) 4265 if self._match_text_seq("INCREMENT", "BY"): 4266 this.set("increment", self._parse_bitwise()) 4267 if self._match_text_seq("MINVALUE"): 4268 this.set("minvalue", self._parse_bitwise()) 4269 if self._match_text_seq("MAXVALUE"): 4270 this.set("maxvalue", self._parse_bitwise()) 4271 4272 if self._match_text_seq("CYCLE"): 4273 this.set("cycle", True) 4274 elif self._match_text_seq("NO", "CYCLE"): 4275 this.set("cycle", False) 4276 4277 if not identity: 4278 this.set("expression", self._parse_bitwise()) 4279 elif not this.args.get("start") and self._match(TokenType.NUMBER, advance=False): 4280 args = self._parse_csv(self._parse_bitwise) 4281 this.set("start", seq_get(args, 0)) 4282 this.set("increment", seq_get(args, 1)) 4283 4284 self._match_r_paren() 4285 4286 return this 4287 4288 def _parse_inline(self) -> exp.InlineLengthColumnConstraint: 4289 self._match_text_seq("LENGTH") 4290 return self.expression(exp.InlineLengthColumnConstraint, this=self._parse_bitwise()) 4291 4292 def _parse_not_constraint( 4293 self, 4294 ) -> t.Optional[exp.Expression]: 4295 if self._match_text_seq("NULL"): 4296 return self.expression(exp.NotNullColumnConstraint) 4297 if self._match_text_seq("CASESPECIFIC"): 4298 return self.expression(exp.CaseSpecificColumnConstraint, not_=True) 4299 if self._match_text_seq("FOR", "REPLICATION"): 4300 return self.expression(exp.NotForReplicationColumnConstraint) 4301 return None 4302 4303 def _parse_column_constraint(self) -> t.Optional[exp.Expression]: 4304 if self._match(TokenType.CONSTRAINT): 4305 this = self._parse_id_var() 4306 else: 4307 this = None 4308 4309 if self._match_texts(self.CONSTRAINT_PARSERS): 4310 return self.expression( 4311 exp.ColumnConstraint, 4312 this=this, 4313 kind=self.CONSTRAINT_PARSERS[self._prev.text.upper()](self), 4314 ) 4315 4316 return this 4317 4318 def _parse_constraint(self) -> t.Optional[exp.Expression]: 4319 if not self._match(TokenType.CONSTRAINT): 4320 return self._parse_unnamed_constraint(constraints=self.SCHEMA_UNNAMED_CONSTRAINTS) 4321 4322 this = self._parse_id_var() 4323 expressions = [] 4324 4325 while True: 4326 constraint = self._parse_unnamed_constraint() or self._parse_function() 4327 if not constraint: 4328 break 4329 expressions.append(constraint) 4330 4331 return self.expression(exp.Constraint, this=this, expressions=expressions) 4332 4333 def _parse_unnamed_constraint( 4334 self, constraints: t.Optional[t.Collection[str]] = None 4335 ) -> t.Optional[exp.Expression]: 4336 if self._match(TokenType.IDENTIFIER, advance=False) or not self._match_texts( 4337 constraints or self.CONSTRAINT_PARSERS 4338 ): 4339 return None 4340 4341 constraint = self._prev.text.upper() 4342 if constraint not in self.CONSTRAINT_PARSERS: 4343 self.raise_error(f"No parser found for schema constraint {constraint}.") 4344 4345 return self.CONSTRAINT_PARSERS[constraint](self) 4346 4347 def _parse_unique(self) -> exp.UniqueColumnConstraint: 4348 self._match_text_seq("KEY") 4349 return self.expression( 4350 exp.UniqueColumnConstraint, 4351 this=self._parse_schema(self._parse_id_var(any_token=False)), 4352 index_type=self._match(TokenType.USING) and self._advance_any() and self._prev.text, 4353 ) 4354 4355 def _parse_key_constraint_options(self) -> t.List[str]: 4356 options = [] 4357 while True: 4358 if not self._curr: 4359 break 4360 4361 if self._match(TokenType.ON): 4362 action = None 4363 on = self._advance_any() and self._prev.text 4364 4365 if self._match_text_seq("NO", "ACTION"): 4366 action = "NO ACTION" 4367 elif self._match_text_seq("CASCADE"): 4368 action = "CASCADE" 4369 elif self._match_text_seq("RESTRICT"): 4370 action = "RESTRICT" 4371 elif self._match_pair(TokenType.SET, TokenType.NULL): 4372 action = "SET NULL" 4373 elif self._match_pair(TokenType.SET, TokenType.DEFAULT): 4374 action = "SET DEFAULT" 4375 else: 4376 self.raise_error("Invalid key constraint") 4377 4378 options.append(f"ON {on} {action}") 4379 elif self._match_text_seq("NOT", "ENFORCED"): 4380 options.append("NOT ENFORCED") 4381 elif self._match_text_seq("DEFERRABLE"): 4382 options.append("DEFERRABLE") 4383 elif self._match_text_seq("INITIALLY", "DEFERRED"): 4384 options.append("INITIALLY DEFERRED") 4385 elif self._match_text_seq("NORELY"): 4386 options.append("NORELY") 4387 elif self._match_text_seq("MATCH", "FULL"): 4388 options.append("MATCH FULL") 4389 else: 4390 break 4391 4392 return options 4393 4394 def _parse_references(self, match: bool = True) -> t.Optional[exp.Reference]: 4395 if match and not self._match(TokenType.REFERENCES): 4396 return None 4397 4398 expressions = None 4399 this = self._parse_table(schema=True) 4400 options = self._parse_key_constraint_options() 4401 return self.expression(exp.Reference, this=this, expressions=expressions, options=options) 4402 4403 def _parse_foreign_key(self) -> exp.ForeignKey: 4404 expressions = self._parse_wrapped_id_vars() 4405 reference = self._parse_references() 4406 options = {} 4407 4408 while self._match(TokenType.ON): 4409 if not self._match_set((TokenType.DELETE, TokenType.UPDATE)): 4410 self.raise_error("Expected DELETE or UPDATE") 4411 4412 kind = self._prev.text.lower() 4413 4414 if self._match_text_seq("NO", "ACTION"): 4415 action = "NO ACTION" 4416 elif self._match(TokenType.SET): 4417 self._match_set((TokenType.NULL, TokenType.DEFAULT)) 4418 action = "SET " + self._prev.text.upper() 4419 else: 4420 self._advance() 4421 action = self._prev.text.upper() 4422 4423 options[kind] = action 4424 4425 return self.expression( 4426 exp.ForeignKey, 4427 expressions=expressions, 4428 reference=reference, 4429 **options, # type: ignore 4430 ) 4431 4432 def _parse_primary_key_part(self) -> t.Optional[exp.Expression]: 4433 return self._parse_field() 4434 4435 def _parse_period_for_system_time(self) -> t.Optional[exp.PeriodForSystemTimeConstraint]: 4436 if not self._match(TokenType.TIMESTAMP_SNAPSHOT): 4437 self._retreat(self._index - 1) 4438 return None 4439 4440 id_vars = self._parse_wrapped_id_vars() 4441 return self.expression( 4442 exp.PeriodForSystemTimeConstraint, 4443 this=seq_get(id_vars, 0), 4444 expression=seq_get(id_vars, 1), 4445 ) 4446 4447 def _parse_primary_key( 4448 self, wrapped_optional: bool = False, in_props: bool = False 4449 ) -> exp.PrimaryKeyColumnConstraint | exp.PrimaryKey: 4450 desc = ( 4451 self._match_set((TokenType.ASC, TokenType.DESC)) 4452 and self._prev.token_type == TokenType.DESC 4453 ) 4454 4455 if not in_props and not self._match(TokenType.L_PAREN, advance=False): 4456 return self.expression(exp.PrimaryKeyColumnConstraint, desc=desc) 4457 4458 expressions = self._parse_wrapped_csv( 4459 self._parse_primary_key_part, optional=wrapped_optional 4460 ) 4461 options = self._parse_key_constraint_options() 4462 return self.expression(exp.PrimaryKey, expressions=expressions, options=options) 4463 4464 def _parse_bracket_key_value(self, is_map: bool = False) -> t.Optional[exp.Expression]: 4465 return self._parse_slice(self._parse_alias(self._parse_conjunction(), explicit=True)) 4466 4467 def _parse_bracket(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 4468 if not self._match_set((TokenType.L_BRACKET, TokenType.L_BRACE)): 4469 return this 4470 4471 bracket_kind = self._prev.token_type 4472 expressions = self._parse_csv( 4473 lambda: self._parse_bracket_key_value(is_map=bracket_kind == TokenType.L_BRACE) 4474 ) 4475 4476 if not self._match(TokenType.R_BRACKET) and bracket_kind == TokenType.L_BRACKET: 4477 self.raise_error("Expected ]") 4478 elif not self._match(TokenType.R_BRACE) and bracket_kind == TokenType.L_BRACE: 4479 self.raise_error("Expected }") 4480 4481 # https://duckdb.org/docs/sql/data_types/struct.html#creating-structs 4482 if bracket_kind == TokenType.L_BRACE: 4483 this = self.expression(exp.Struct, expressions=expressions) 4484 elif not this or this.name.upper() == "ARRAY": 4485 this = self.expression(exp.Array, expressions=expressions) 4486 else: 4487 expressions = apply_index_offset(this, expressions, -self.dialect.INDEX_OFFSET) 4488 this = self.expression(exp.Bracket, this=this, expressions=expressions) 4489 4490 self._add_comments(this) 4491 return self._parse_bracket(this) 4492 4493 def _parse_slice(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 4494 if self._match(TokenType.COLON): 4495 return self.expression(exp.Slice, this=this, expression=self._parse_conjunction()) 4496 return this 4497 4498 def _parse_case(self) -> t.Optional[exp.Expression]: 4499 ifs = [] 4500 default = None 4501 4502 comments = self._prev_comments 4503 expression = self._parse_conjunction() 4504 4505 while self._match(TokenType.WHEN): 4506 this = self._parse_conjunction() 4507 self._match(TokenType.THEN) 4508 then = self._parse_conjunction() 4509 ifs.append(self.expression(exp.If, this=this, true=then)) 4510 4511 if self._match(TokenType.ELSE): 4512 default = self._parse_conjunction() 4513 4514 if not self._match(TokenType.END): 4515 if isinstance(default, exp.Interval) and default.this.sql().upper() == "END": 4516 default = exp.column("interval") 4517 else: 4518 self.raise_error("Expected END after CASE", self._prev) 4519 4520 return self._parse_window( 4521 self.expression(exp.Case, comments=comments, this=expression, ifs=ifs, default=default) 4522 ) 4523 4524 def _parse_if(self) -> t.Optional[exp.Expression]: 4525 if self._match(TokenType.L_PAREN): 4526 args = self._parse_csv(self._parse_conjunction) 4527 this = self.validate_expression(exp.If.from_arg_list(args), args) 4528 self._match_r_paren() 4529 else: 4530 index = self._index - 1 4531 4532 if self.NO_PAREN_IF_COMMANDS and index == 0: 4533 return self._parse_as_command(self._prev) 4534 4535 condition = self._parse_conjunction() 4536 4537 if not condition: 4538 self._retreat(index) 4539 return None 4540 4541 self._match(TokenType.THEN) 4542 true = self._parse_conjunction() 4543 false = self._parse_conjunction() if self._match(TokenType.ELSE) else None 4544 self._match(TokenType.END) 4545 this = self.expression(exp.If, this=condition, true=true, false=false) 4546 4547 return self._parse_window(this) 4548 4549 def _parse_next_value_for(self) -> t.Optional[exp.Expression]: 4550 if not self._match_text_seq("VALUE", "FOR"): 4551 self._retreat(self._index - 1) 4552 return None 4553 4554 return self.expression( 4555 exp.NextValueFor, 4556 this=self._parse_column(), 4557 order=self._match(TokenType.OVER) and self._parse_wrapped(self._parse_order), 4558 ) 4559 4560 def _parse_extract(self) -> exp.Extract: 4561 this = self._parse_function() or self._parse_var() or self._parse_type() 4562 4563 if self._match(TokenType.FROM): 4564 return self.expression(exp.Extract, this=this, expression=self._parse_bitwise()) 4565 4566 if not self._match(TokenType.COMMA): 4567 self.raise_error("Expected FROM or comma after EXTRACT", self._prev) 4568 4569 return self.expression(exp.Extract, this=this, expression=self._parse_bitwise()) 4570 4571 def _parse_cast(self, strict: bool, safe: t.Optional[bool] = None) -> exp.Expression: 4572 this = self._parse_conjunction() 4573 4574 if not self._match(TokenType.ALIAS): 4575 if self._match(TokenType.COMMA): 4576 return self.expression(exp.CastToStrType, this=this, to=self._parse_string()) 4577 4578 self.raise_error("Expected AS after CAST") 4579 4580 fmt = None 4581 to = self._parse_types() 4582 4583 if self._match(TokenType.FORMAT): 4584 fmt_string = self._parse_string() 4585 fmt = self._parse_at_time_zone(fmt_string) 4586 4587 if not to: 4588 to = exp.DataType.build(exp.DataType.Type.UNKNOWN) 4589 if to.this in exp.DataType.TEMPORAL_TYPES: 4590 this = self.expression( 4591 exp.StrToDate if to.this == exp.DataType.Type.DATE else exp.StrToTime, 4592 this=this, 4593 format=exp.Literal.string( 4594 format_time( 4595 fmt_string.this if fmt_string else "", 4596 self.dialect.FORMAT_MAPPING or self.dialect.TIME_MAPPING, 4597 self.dialect.FORMAT_TRIE or self.dialect.TIME_TRIE, 4598 ) 4599 ), 4600 ) 4601 4602 if isinstance(fmt, exp.AtTimeZone) and isinstance(this, exp.StrToTime): 4603 this.set("zone", fmt.args["zone"]) 4604 return this 4605 elif not to: 4606 self.raise_error("Expected TYPE after CAST") 4607 elif isinstance(to, exp.Identifier): 4608 to = exp.DataType.build(to.name, udt=True) 4609 elif to.this == exp.DataType.Type.CHAR: 4610 if self._match(TokenType.CHARACTER_SET): 4611 to = self.expression(exp.CharacterSet, this=self._parse_var_or_string()) 4612 4613 return self.expression( 4614 exp.Cast if strict else exp.TryCast, this=this, to=to, format=fmt, safe=safe 4615 ) 4616 4617 def _parse_string_agg(self) -> exp.Expression: 4618 if self._match(TokenType.DISTINCT): 4619 args: t.List[t.Optional[exp.Expression]] = [ 4620 self.expression(exp.Distinct, expressions=[self._parse_conjunction()]) 4621 ] 4622 if self._match(TokenType.COMMA): 4623 args.extend(self._parse_csv(self._parse_conjunction)) 4624 else: 4625 args = self._parse_csv(self._parse_conjunction) # type: ignore 4626 4627 index = self._index 4628 if not self._match(TokenType.R_PAREN) and args: 4629 # postgres: STRING_AGG([DISTINCT] expression, separator [ORDER BY expression1 {ASC | DESC} [, ...]]) 4630 # bigquery: STRING_AGG([DISTINCT] expression [, separator] [ORDER BY key [{ASC | DESC}] [, ... ]] [LIMIT n]) 4631 args[-1] = self._parse_limit(this=self._parse_order(this=args[-1])) 4632 return self.expression(exp.GroupConcat, this=args[0], separator=seq_get(args, 1)) 4633 4634 # Checks if we can parse an order clause: WITHIN GROUP (ORDER BY <order_by_expression_list> [ASC | DESC]). 4635 # This is done "manually", instead of letting _parse_window parse it into an exp.WithinGroup node, so that 4636 # the STRING_AGG call is parsed like in MySQL / SQLite and can thus be transpiled more easily to them. 4637 if not self._match_text_seq("WITHIN", "GROUP"): 4638 self._retreat(index) 4639 return self.validate_expression(exp.GroupConcat.from_arg_list(args), args) 4640 4641 self._match_l_paren() # The corresponding match_r_paren will be called in parse_function (caller) 4642 order = self._parse_order(this=seq_get(args, 0)) 4643 return self.expression(exp.GroupConcat, this=order, separator=seq_get(args, 1)) 4644 4645 def _parse_convert( 4646 self, strict: bool, safe: t.Optional[bool] = None 4647 ) -> t.Optional[exp.Expression]: 4648 this = self._parse_bitwise() 4649 4650 if self._match(TokenType.USING): 4651 to: t.Optional[exp.Expression] = self.expression( 4652 exp.CharacterSet, this=self._parse_var() 4653 ) 4654 elif self._match(TokenType.COMMA): 4655 to = self._parse_types() 4656 else: 4657 to = None 4658 4659 return self.expression(exp.Cast if strict else exp.TryCast, this=this, to=to, safe=safe) 4660 4661 def _parse_decode(self) -> t.Optional[exp.Decode | exp.Case]: 4662 """ 4663 There are generally two variants of the DECODE function: 4664 4665 - DECODE(bin, charset) 4666 - DECODE(expression, search, result [, search, result] ... [, default]) 4667 4668 The second variant will always be parsed into a CASE expression. Note that NULL 4669 needs special treatment, since we need to explicitly check for it with `IS NULL`, 4670 instead of relying on pattern matching. 4671 """ 4672 args = self._parse_csv(self._parse_conjunction) 4673 4674 if len(args) < 3: 4675 return self.expression(exp.Decode, this=seq_get(args, 0), charset=seq_get(args, 1)) 4676 4677 expression, *expressions = args 4678 if not expression: 4679 return None 4680 4681 ifs = [] 4682 for search, result in zip(expressions[::2], expressions[1::2]): 4683 if not search or not result: 4684 return None 4685 4686 if isinstance(search, exp.Literal): 4687 ifs.append( 4688 exp.If(this=exp.EQ(this=expression.copy(), expression=search), true=result) 4689 ) 4690 elif isinstance(search, exp.Null): 4691 ifs.append( 4692 exp.If(this=exp.Is(this=expression.copy(), expression=exp.Null()), true=result) 4693 ) 4694 else: 4695 cond = exp.or_( 4696 exp.EQ(this=expression.copy(), expression=search), 4697 exp.and_( 4698 exp.Is(this=expression.copy(), expression=exp.Null()), 4699 exp.Is(this=search.copy(), expression=exp.Null()), 4700 copy=False, 4701 ), 4702 copy=False, 4703 ) 4704 ifs.append(exp.If(this=cond, true=result)) 4705 4706 return exp.Case(ifs=ifs, default=expressions[-1] if len(expressions) % 2 == 1 else None) 4707 4708 def _parse_json_key_value(self) -> t.Optional[exp.JSONKeyValue]: 4709 self._match_text_seq("KEY") 4710 key = self._parse_column() 4711 self._match_set(self.JSON_KEY_VALUE_SEPARATOR_TOKENS) 4712 self._match_text_seq("VALUE") 4713 value = self._parse_bitwise() 4714 4715 if not key and not value: 4716 return None 4717 return self.expression(exp.JSONKeyValue, this=key, expression=value) 4718 4719 def _parse_format_json(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 4720 if not this or not self._match_text_seq("FORMAT", "JSON"): 4721 return this 4722 4723 return self.expression(exp.FormatJson, this=this) 4724 4725 def _parse_on_handling(self, on: str, *values: str) -> t.Optional[str]: 4726 # Parses the "X ON Y" syntax, i.e. NULL ON NULL (Oracle, T-SQL) 4727 for value in values: 4728 if self._match_text_seq(value, "ON", on): 4729 return f"{value} ON {on}" 4730 4731 return None 4732 4733 @t.overload 4734 def _parse_json_object(self, agg: Lit[False]) -> exp.JSONObject: 4735 ... 4736 4737 @t.overload 4738 def _parse_json_object(self, agg: Lit[True]) -> exp.JSONObjectAgg: 4739 ... 4740 4741 def _parse_json_object(self, agg=False): 4742 star = self._parse_star() 4743 expressions = ( 4744 [star] 4745 if star 4746 else self._parse_csv(lambda: self._parse_format_json(self._parse_json_key_value())) 4747 ) 4748 null_handling = self._parse_on_handling("NULL", "NULL", "ABSENT") 4749 4750 unique_keys = None 4751 if self._match_text_seq("WITH", "UNIQUE"): 4752 unique_keys = True 4753 elif self._match_text_seq("WITHOUT", "UNIQUE"): 4754 unique_keys = False 4755 4756 self._match_text_seq("KEYS") 4757 4758 return_type = self._match_text_seq("RETURNING") and self._parse_format_json( 4759 self._parse_type() 4760 ) 4761 encoding = self._match_text_seq("ENCODING") and self._parse_var() 4762 4763 return self.expression( 4764 exp.JSONObjectAgg if agg else exp.JSONObject, 4765 expressions=expressions, 4766 null_handling=null_handling, 4767 unique_keys=unique_keys, 4768 return_type=return_type, 4769 encoding=encoding, 4770 ) 4771 4772 # Note: this is currently incomplete; it only implements the "JSON_value_column" part 4773 def _parse_json_column_def(self) -> exp.JSONColumnDef: 4774 if not self._match_text_seq("NESTED"): 4775 this = self._parse_id_var() 4776 kind = self._parse_types(allow_identifiers=False) 4777 nested = None 4778 else: 4779 this = None 4780 kind = None 4781 nested = True 4782 4783 path = self._match_text_seq("PATH") and self._parse_string() 4784 nested_schema = nested and self._parse_json_schema() 4785 4786 return self.expression( 4787 exp.JSONColumnDef, 4788 this=this, 4789 kind=kind, 4790 path=path, 4791 nested_schema=nested_schema, 4792 ) 4793 4794 def _parse_json_schema(self) -> exp.JSONSchema: 4795 self._match_text_seq("COLUMNS") 4796 return self.expression( 4797 exp.JSONSchema, 4798 expressions=self._parse_wrapped_csv(self._parse_json_column_def, optional=True), 4799 ) 4800 4801 def _parse_json_table(self) -> exp.JSONTable: 4802 this = self._parse_format_json(self._parse_bitwise()) 4803 path = self._match(TokenType.COMMA) and self._parse_string() 4804 error_handling = self._parse_on_handling("ERROR", "ERROR", "NULL") 4805 empty_handling = self._parse_on_handling("EMPTY", "ERROR", "NULL") 4806 schema = self._parse_json_schema() 4807 4808 return exp.JSONTable( 4809 this=this, 4810 schema=schema, 4811 path=path, 4812 error_handling=error_handling, 4813 empty_handling=empty_handling, 4814 ) 4815 4816 def _parse_match_against(self) -> exp.MatchAgainst: 4817 expressions = self._parse_csv(self._parse_column) 4818 4819 self._match_text_seq(")", "AGAINST", "(") 4820 4821 this = self._parse_string() 4822 4823 if self._match_text_seq("IN", "NATURAL", "LANGUAGE", "MODE"): 4824 modifier = "IN NATURAL LANGUAGE MODE" 4825 if self._match_text_seq("WITH", "QUERY", "EXPANSION"): 4826 modifier = f"{modifier} WITH QUERY EXPANSION" 4827 elif self._match_text_seq("IN", "BOOLEAN", "MODE"): 4828 modifier = "IN BOOLEAN MODE" 4829 elif self._match_text_seq("WITH", "QUERY", "EXPANSION"): 4830 modifier = "WITH QUERY EXPANSION" 4831 else: 4832 modifier = None 4833 4834 return self.expression( 4835 exp.MatchAgainst, this=this, expressions=expressions, modifier=modifier 4836 ) 4837 4838 # https://learn.microsoft.com/en-us/sql/t-sql/functions/openjson-transact-sql?view=sql-server-ver16 4839 def _parse_open_json(self) -> exp.OpenJSON: 4840 this = self._parse_bitwise() 4841 path = self._match(TokenType.COMMA) and self._parse_string() 4842 4843 def _parse_open_json_column_def() -> exp.OpenJSONColumnDef: 4844 this = self._parse_field(any_token=True) 4845 kind = self._parse_types() 4846 path = self._parse_string() 4847 as_json = self._match_pair(TokenType.ALIAS, TokenType.JSON) 4848 4849 return self.expression( 4850 exp.OpenJSONColumnDef, this=this, kind=kind, path=path, as_json=as_json 4851 ) 4852 4853 expressions = None 4854 if self._match_pair(TokenType.R_PAREN, TokenType.WITH): 4855 self._match_l_paren() 4856 expressions = self._parse_csv(_parse_open_json_column_def) 4857 4858 return self.expression(exp.OpenJSON, this=this, path=path, expressions=expressions) 4859 4860 def _parse_position(self, haystack_first: bool = False) -> exp.StrPosition: 4861 args = self._parse_csv(self._parse_bitwise) 4862 4863 if self._match(TokenType.IN): 4864 return self.expression( 4865 exp.StrPosition, this=self._parse_bitwise(), substr=seq_get(args, 0) 4866 ) 4867 4868 if haystack_first: 4869 haystack = seq_get(args, 0) 4870 needle = seq_get(args, 1) 4871 else: 4872 needle = seq_get(args, 0) 4873 haystack = seq_get(args, 1) 4874 4875 return self.expression( 4876 exp.StrPosition, this=haystack, substr=needle, position=seq_get(args, 2) 4877 ) 4878 4879 def _parse_predict(self) -> exp.Predict: 4880 self._match_text_seq("MODEL") 4881 this = self._parse_table() 4882 4883 self._match(TokenType.COMMA) 4884 self._match_text_seq("TABLE") 4885 4886 return self.expression( 4887 exp.Predict, 4888 this=this, 4889 expression=self._parse_table(), 4890 params_struct=self._match(TokenType.COMMA) and self._parse_bitwise(), 4891 ) 4892 4893 def _parse_join_hint(self, func_name: str) -> exp.JoinHint: 4894 args = self._parse_csv(self._parse_table) 4895 return exp.JoinHint(this=func_name.upper(), expressions=args) 4896 4897 def _parse_substring(self) -> exp.Substring: 4898 # Postgres supports the form: substring(string [from int] [for int]) 4899 # https://www.postgresql.org/docs/9.1/functions-string.html @ Table 9-6 4900 4901 args = t.cast(t.List[t.Optional[exp.Expression]], self._parse_csv(self._parse_bitwise)) 4902 4903 if self._match(TokenType.FROM): 4904 args.append(self._parse_bitwise()) 4905 if self._match(TokenType.FOR): 4906 args.append(self._parse_bitwise()) 4907 4908 return self.validate_expression(exp.Substring.from_arg_list(args), args) 4909 4910 def _parse_trim(self) -> exp.Trim: 4911 # https://www.w3resource.com/sql/character-functions/trim.php 4912 # https://docs.oracle.com/javadb/10.8.3.0/ref/rreftrimfunc.html 4913 4914 position = None 4915 collation = None 4916 expression = None 4917 4918 if self._match_texts(self.TRIM_TYPES): 4919 position = self._prev.text.upper() 4920 4921 this = self._parse_bitwise() 4922 if self._match_set((TokenType.FROM, TokenType.COMMA)): 4923 invert_order = self._prev.token_type == TokenType.FROM or self.TRIM_PATTERN_FIRST 4924 expression = self._parse_bitwise() 4925 4926 if invert_order: 4927 this, expression = expression, this 4928 4929 if self._match(TokenType.COLLATE): 4930 collation = self._parse_bitwise() 4931 4932 return self.expression( 4933 exp.Trim, this=this, position=position, expression=expression, collation=collation 4934 ) 4935 4936 def _parse_window_clause(self) -> t.Optional[t.List[exp.Expression]]: 4937 return self._match(TokenType.WINDOW) and self._parse_csv(self._parse_named_window) 4938 4939 def _parse_named_window(self) -> t.Optional[exp.Expression]: 4940 return self._parse_window(self._parse_id_var(), alias=True) 4941 4942 def _parse_respect_or_ignore_nulls( 4943 self, this: t.Optional[exp.Expression] 4944 ) -> t.Optional[exp.Expression]: 4945 if self._match_text_seq("IGNORE", "NULLS"): 4946 return self.expression(exp.IgnoreNulls, this=this) 4947 if self._match_text_seq("RESPECT", "NULLS"): 4948 return self.expression(exp.RespectNulls, this=this) 4949 return this 4950 4951 def _parse_having_max(self, this: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]: 4952 if self._match(TokenType.HAVING): 4953 self._match_texts(("MAX", "MIN")) 4954 max = self._prev.text.upper() != "MIN" 4955 return self.expression( 4956 exp.HavingMax, this=this, expression=self._parse_column(), max=max 4957 ) 4958 4959 return this 4960 4961 def _parse_window( 4962 self, this: t.Optional[exp.Expression], alias: bool = False 4963 ) -> t.Optional[exp.Expression]: 4964 if self._match_pair(TokenType.FILTER, TokenType.L_PAREN): 4965 self._match(TokenType.WHERE) 4966 this = self.expression( 4967 exp.Filter, this=this, expression=self._parse_where(skip_where_token=True) 4968 ) 4969 self._match_r_paren() 4970 4971 # T-SQL allows the OVER (...) syntax after WITHIN GROUP. 4972 # https://learn.microsoft.com/en-us/sql/t-sql/functions/percentile-disc-transact-sql?view=sql-server-ver16 4973 if self._match_text_seq("WITHIN", "GROUP"): 4974 order = self._parse_wrapped(self._parse_order) 4975 this = self.expression(exp.WithinGroup, this=this, expression=order) 4976 4977 # SQL spec defines an optional [ { IGNORE | RESPECT } NULLS ] OVER 4978 # Some dialects choose to implement and some do not. 4979 # https://dev.mysql.com/doc/refman/8.0/en/window-function-descriptions.html 4980 4981 # There is some code above in _parse_lambda that handles 4982 # SELECT FIRST_VALUE(TABLE.COLUMN IGNORE|RESPECT NULLS) OVER ... 4983 4984 # The below changes handle 4985 # SELECT FIRST_VALUE(TABLE.COLUMN) IGNORE|RESPECT NULLS OVER ... 4986 4987 # Oracle allows both formats 4988 # (https://docs.oracle.com/en/database/oracle/oracle-database/19/sqlrf/img_text/first_value.html) 4989 # and Snowflake chose to do the same for familiarity 4990 # https://docs.snowflake.com/en/sql-reference/functions/first_value.html#usage-notes 4991 if isinstance(this, exp.AggFunc): 4992 ignore_respect = this.find(exp.IgnoreNulls, exp.RespectNulls) 4993 4994 if ignore_respect and ignore_respect is not this: 4995 ignore_respect.replace(ignore_respect.this) 4996 this = self.expression(ignore_respect.__class__, this=this) 4997 4998 this = self._parse_respect_or_ignore_nulls(this) 4999 5000 # bigquery select from window x AS (partition by ...) 5001 if alias: 5002 over = None 5003 self._match(TokenType.ALIAS) 5004 elif not self._match_set(self.WINDOW_BEFORE_PAREN_TOKENS): 5005 return this 5006 else: 5007 over = self._prev.text.upper() 5008 5009 if not self._match(TokenType.L_PAREN): 5010 return self.expression( 5011 exp.Window, this=this, alias=self._parse_id_var(False), over=over 5012 ) 5013 5014 window_alias = self._parse_id_var(any_token=False, tokens=self.WINDOW_ALIAS_TOKENS) 5015 5016 first = self._match(TokenType.FIRST) 5017 if self._match_text_seq("LAST"): 5018 first = False 5019 5020 partition, order = self._parse_partition_and_order() 5021 kind = self._match_set((TokenType.ROWS, TokenType.RANGE)) and self._prev.text 5022 5023 if kind: 5024 self._match(TokenType.BETWEEN) 5025 start = self._parse_window_spec() 5026 self._match(TokenType.AND) 5027 end = self._parse_window_spec() 5028 5029 spec = self.expression( 5030 exp.WindowSpec, 5031 kind=kind, 5032 start=start["value"], 5033 start_side=start["side"], 5034 end=end["value"], 5035 end_side=end["side"], 5036 ) 5037 else: 5038 spec = None 5039 5040 self._match_r_paren() 5041 5042 window = self.expression( 5043 exp.Window, 5044 this=this, 5045 partition_by=partition, 5046 order=order, 5047 spec=spec, 5048 alias=window_alias, 5049 over=over, 5050 first=first, 5051 ) 5052 5053 # This covers Oracle's FIRST/LAST syntax: aggregate KEEP (...) OVER (...) 5054 if self._match_set(self.WINDOW_BEFORE_PAREN_TOKENS, advance=False): 5055 return self._parse_window(window, alias=alias) 5056 5057 return window 5058 5059 def _parse_partition_and_order( 5060 self, 5061 ) -> t.Tuple[t.List[exp.Expression], t.Optional[exp.Expression]]: 5062 return self._parse_partition_by(), self._parse_order() 5063 5064 def _parse_window_spec(self) -> t.Dict[str, t.Optional[str | exp.Expression]]: 5065 self._match(TokenType.BETWEEN) 5066 5067 return { 5068 "value": ( 5069 (self._match_text_seq("UNBOUNDED") and "UNBOUNDED") 5070 or (self._match_text_seq("CURRENT", "ROW") and "CURRENT ROW") 5071 or self._parse_bitwise() 5072 ), 5073 "side": self._match_texts(self.WINDOW_SIDES) and self._prev.text, 5074 } 5075 5076 def _parse_alias( 5077 self, this: t.Optional[exp.Expression], explicit: bool = False 5078 ) -> t.Optional[exp.Expression]: 5079 any_token = self._match(TokenType.ALIAS) 5080 comments = self._prev_comments 5081 5082 if explicit and not any_token: 5083 return this 5084 5085 if self._match(TokenType.L_PAREN): 5086 aliases = self.expression( 5087 exp.Aliases, 5088 comments=comments, 5089 this=this, 5090 expressions=self._parse_csv(lambda: self._parse_id_var(any_token)), 5091 ) 5092 self._match_r_paren(aliases) 5093 return aliases 5094 5095 alias = self._parse_id_var(any_token) or ( 5096 self.STRING_ALIASES and self._parse_string_as_identifier() 5097 ) 5098 5099 if alias: 5100 this = self.expression(exp.Alias, comments=comments, this=this, alias=alias) 5101 column = this.this 5102 5103 # Moves the comment next to the alias in `expr /* comment */ AS alias` 5104 if not this.comments and column and column.comments: 5105 this.comments = column.comments 5106 column.comments = None 5107 5108 return this 5109 5110 def _parse_id_var( 5111 self, 5112 any_token: bool = True, 5113 tokens: t.Optional[t.Collection[TokenType]] = None, 5114 ) -> t.Optional[exp.Expression]: 5115 identifier = self._parse_identifier() 5116 5117 if identifier: 5118 return identifier 5119 5120 if (any_token and self._advance_any()) or self._match_set(tokens or self.ID_VAR_TOKENS): 5121 quoted = self._prev.token_type == TokenType.STRING 5122 return exp.Identifier(this=self._prev.text, quoted=quoted) 5123 5124 return None 5125 5126 def _parse_string(self) -> t.Optional[exp.Expression]: 5127 if self._match_set((TokenType.STRING, TokenType.RAW_STRING)): 5128 return self.PRIMARY_PARSERS[self._prev.token_type](self, self._prev) 5129 return self._parse_placeholder() 5130 5131 def _parse_string_as_identifier(self) -> t.Optional[exp.Identifier]: 5132 return exp.to_identifier(self._match(TokenType.STRING) and self._prev.text, quoted=True) 5133 5134 def _parse_number(self) -> t.Optional[exp.Expression]: 5135 if self._match(TokenType.NUMBER): 5136 return self.PRIMARY_PARSERS[TokenType.NUMBER](self, self._prev) 5137 return self._parse_placeholder() 5138 5139 def _parse_identifier(self) -> t.Optional[exp.Expression]: 5140 if self._match(TokenType.IDENTIFIER): 5141 return self.expression(exp.Identifier, this=self._prev.text, quoted=True) 5142 return self._parse_placeholder() 5143 5144 def _parse_var( 5145 self, 5146 any_token: bool = False, 5147 tokens: t.Optional[t.Collection[TokenType]] = None, 5148 upper: bool = False, 5149 ) -> t.Optional[exp.Expression]: 5150 if ( 5151 (any_token and self._advance_any()) 5152 or self._match(TokenType.VAR) 5153 or (self._match_set(tokens) if tokens else False) 5154 ): 5155 return self.expression( 5156 exp.Var, this=self._prev.text.upper() if upper else self._prev.text 5157 ) 5158 return self._parse_placeholder() 5159 5160 def _advance_any(self, ignore_reserved: bool = False) -> t.Optional[Token]: 5161 if self._curr and (ignore_reserved or self._curr.token_type not in self.RESERVED_TOKENS): 5162 self._advance() 5163 return self._prev 5164 return None 5165 5166 def _parse_var_or_string(self) -> t.Optional[exp.Expression]: 5167 return self._parse_var() or self._parse_string() 5168 5169 def _parse_null(self) -> t.Optional[exp.Expression]: 5170 if self._match_set(self.NULL_TOKENS): 5171 return self.PRIMARY_PARSERS[TokenType.NULL](self, self._prev) 5172 return self._parse_placeholder() 5173 5174 def _parse_boolean(self) -> t.Optional[exp.Expression]: 5175 if self._match(TokenType.TRUE): 5176 return self.PRIMARY_PARSERS[TokenType.TRUE](self, self._prev) 5177 if self._match(TokenType.FALSE): 5178 return self.PRIMARY_PARSERS[TokenType.FALSE](self, self._prev) 5179 return self._parse_placeholder() 5180 5181 def _parse_star(self) -> t.Optional[exp.Expression]: 5182 if self._match(TokenType.STAR): 5183 return self.PRIMARY_PARSERS[TokenType.STAR](self, self._prev) 5184 return self._parse_placeholder() 5185 5186 def _parse_parameter(self) -> exp.Parameter: 5187 def _parse_parameter_part() -> t.Optional[exp.Expression]: 5188 return ( 5189 self._parse_identifier() or self._parse_primary() or self._parse_var(any_token=True) 5190 ) 5191 5192 self._match(TokenType.L_BRACE) 5193 this = _parse_parameter_part() 5194 expression = self._match(TokenType.COLON) and _parse_parameter_part() 5195 self._match(TokenType.R_BRACE) 5196 5197 return self.expression(exp.Parameter, this=this, expression=expression) 5198 5199 def _parse_placeholder(self) -> t.Optional[exp.Expression]: 5200 if self._match_set(self.PLACEHOLDER_PARSERS): 5201 placeholder = self.PLACEHOLDER_PARSERS[self._prev.token_type](self) 5202 if placeholder: 5203 return placeholder 5204 self._advance(-1) 5205 return None 5206 5207 def _parse_except(self) -> t.Optional[t.List[exp.Expression]]: 5208 if not self._match(TokenType.EXCEPT): 5209 return None 5210 if self._match(TokenType.L_PAREN, advance=False): 5211 return self._parse_wrapped_csv(self._parse_column) 5212 5213 except_column = self._parse_column() 5214 return [except_column] if except_column else None 5215 5216 def _parse_replace(self) -> t.Optional[t.List[exp.Expression]]: 5217 if not self._match(TokenType.REPLACE): 5218 return None 5219 if self._match(TokenType.L_PAREN, advance=False): 5220 return self._parse_wrapped_csv(self._parse_expression) 5221 5222 replace_expression = self._parse_expression() 5223 return [replace_expression] if replace_expression else None 5224 5225 def _parse_csv( 5226 self, parse_method: t.Callable, sep: TokenType = TokenType.COMMA 5227 ) -> t.List[exp.Expression]: 5228 parse_result = parse_method() 5229 items = [parse_result] if parse_result is not None else [] 5230 5231 while self._match(sep): 5232 self._add_comments(parse_result) 5233 parse_result = parse_method() 5234 if parse_result is not None: 5235 items.append(parse_result) 5236 5237 return items 5238 5239 def _parse_tokens( 5240 self, parse_method: t.Callable, expressions: t.Dict 5241 ) -> t.Optional[exp.Expression]: 5242 this = parse_method() 5243 5244 while self._match_set(expressions): 5245 this = self.expression( 5246 expressions[self._prev.token_type], 5247 this=this, 5248 comments=self._prev_comments, 5249 expression=parse_method(), 5250 ) 5251 5252 return this 5253 5254 def _parse_wrapped_id_vars(self, optional: bool = False) -> t.List[exp.Expression]: 5255 return self._parse_wrapped_csv(self._parse_id_var, optional=optional) 5256 5257 def _parse_wrapped_csv( 5258 self, parse_method: t.Callable, sep: TokenType = TokenType.COMMA, optional: bool = False 5259 ) -> t.List[exp.Expression]: 5260 return self._parse_wrapped( 5261 lambda: self._parse_csv(parse_method, sep=sep), optional=optional 5262 ) 5263 5264 def _parse_wrapped(self, parse_method: t.Callable, optional: bool = False) -> t.Any: 5265 wrapped = self._match(TokenType.L_PAREN) 5266 if not wrapped and not optional: 5267 self.raise_error("Expecting (") 5268 parse_result = parse_method() 5269 if wrapped: 5270 self._match_r_paren() 5271 return parse_result 5272 5273 def _parse_expressions(self) -> t.List[exp.Expression]: 5274 return self._parse_csv(self._parse_expression) 5275 5276 def _parse_select_or_expression(self, alias: bool = False) -> t.Optional[exp.Expression]: 5277 return self._parse_select() or self._parse_set_operations( 5278 self._parse_expression() if alias else self._parse_conjunction() 5279 ) 5280 5281 def _parse_ddl_select(self) -> t.Optional[exp.Expression]: 5282 return self._parse_query_modifiers( 5283 self._parse_set_operations(self._parse_select(nested=True, parse_subquery_alias=False)) 5284 ) 5285 5286 def _parse_transaction(self) -> exp.Transaction | exp.Command: 5287 this = None 5288 if self._match_texts(self.TRANSACTION_KIND): 5289 this = self._prev.text 5290 5291 self._match_texts(("TRANSACTION", "WORK")) 5292 5293 modes = [] 5294 while True: 5295 mode = [] 5296 while self._match(TokenType.VAR): 5297 mode.append(self._prev.text) 5298 5299 if mode: 5300 modes.append(" ".join(mode)) 5301 if not self._match(TokenType.COMMA): 5302 break 5303 5304 return self.expression(exp.Transaction, this=this, modes=modes) 5305 5306 def _parse_commit_or_rollback(self) -> exp.Commit | exp.Rollback: 5307 chain = None 5308 savepoint = None 5309 is_rollback = self._prev.token_type == TokenType.ROLLBACK 5310 5311 self._match_texts(("TRANSACTION", "WORK")) 5312 5313 if self._match_text_seq("TO"): 5314 self._match_text_seq("SAVEPOINT") 5315 savepoint = self._parse_id_var() 5316 5317 if self._match(TokenType.AND): 5318 chain = not self._match_text_seq("NO") 5319 self._match_text_seq("CHAIN") 5320 5321 if is_rollback: 5322 return self.expression(exp.Rollback, savepoint=savepoint) 5323 5324 return self.expression(exp.Commit, chain=chain) 5325 5326 def _parse_refresh(self) -> exp.Refresh: 5327 self._match(TokenType.TABLE) 5328 return self.expression(exp.Refresh, this=self._parse_string() or self._parse_table()) 5329 5330 def _parse_add_column(self) -> t.Optional[exp.Expression]: 5331 if not self._match_text_seq("ADD"): 5332 return None 5333 5334 self._match(TokenType.COLUMN) 5335 exists_column = self._parse_exists(not_=True) 5336 expression = self._parse_field_def() 5337 5338 if expression: 5339 expression.set("exists", exists_column) 5340 5341 # https://docs.databricks.com/delta/update-schema.html#explicitly-update-schema-to-add-columns 5342 if self._match_texts(("FIRST", "AFTER")): 5343 position = self._prev.text 5344 column_position = self.expression( 5345 exp.ColumnPosition, this=self._parse_column(), position=position 5346 ) 5347 expression.set("position", column_position) 5348 5349 return expression 5350 5351 def _parse_drop_column(self) -> t.Optional[exp.Drop | exp.Command]: 5352 drop = self._match(TokenType.DROP) and self._parse_drop() 5353 if drop and not isinstance(drop, exp.Command): 5354 drop.set("kind", drop.args.get("kind", "COLUMN")) 5355 return drop 5356 5357 # https://docs.aws.amazon.com/athena/latest/ug/alter-table-drop-partition.html 5358 def _parse_drop_partition(self, exists: t.Optional[bool] = None) -> exp.DropPartition: 5359 return self.expression( 5360 exp.DropPartition, expressions=self._parse_csv(self._parse_partition), exists=exists 5361 ) 5362 5363 def _parse_add_constraint(self) -> exp.AddConstraint: 5364 this = None 5365 kind = self._prev.token_type 5366 5367 if kind == TokenType.CONSTRAINT: 5368 this = self._parse_id_var() 5369 5370 if self._match_text_seq("CHECK"): 5371 expression = self._parse_wrapped(self._parse_conjunction) 5372 enforced = self._match_text_seq("ENFORCED") or False 5373 5374 return self.expression( 5375 exp.AddConstraint, this=this, expression=expression, enforced=enforced 5376 ) 5377 5378 if kind == TokenType.FOREIGN_KEY or self._match(TokenType.FOREIGN_KEY): 5379 expression = self._parse_foreign_key() 5380 elif kind == TokenType.PRIMARY_KEY or self._match(TokenType.PRIMARY_KEY): 5381 expression = self._parse_primary_key() 5382 else: 5383 expression = None 5384 5385 return self.expression(exp.AddConstraint, this=this, expression=expression) 5386 5387 def _parse_alter_table_add(self) -> t.List[exp.Expression]: 5388 index = self._index - 1 5389 5390 if self._match_set(self.ADD_CONSTRAINT_TOKENS): 5391 return self._parse_csv(self._parse_add_constraint) 5392 5393 self._retreat(index) 5394 if not self.ALTER_TABLE_ADD_REQUIRED_FOR_EACH_COLUMN and self._match_text_seq("ADD"): 5395 return self._parse_wrapped_csv(self._parse_field_def, optional=True) 5396 return self._parse_wrapped_csv(self._parse_add_column, optional=True) 5397 5398 def _parse_alter_table_alter(self) -> exp.AlterColumn: 5399 self._match(TokenType.COLUMN) 5400 column = self._parse_field(any_token=True) 5401 5402 if self._match_pair(TokenType.DROP, TokenType.DEFAULT): 5403 return self.expression(exp.AlterColumn, this=column, drop=True) 5404 if self._match_pair(TokenType.SET, TokenType.DEFAULT): 5405 return self.expression(exp.AlterColumn, this=column, default=self._parse_conjunction()) 5406 if self._match(TokenType.COMMENT): 5407 return self.expression(exp.AlterColumn, this=column, comment=self._parse_string()) 5408 5409 self._match_text_seq("SET", "DATA") 5410 return self.expression( 5411 exp.AlterColumn, 5412 this=column, 5413 dtype=self._match_text_seq("TYPE") and self._parse_types(), 5414 collate=self._match(TokenType.COLLATE) and self._parse_term(), 5415 using=self._match(TokenType.USING) and self._parse_conjunction(), 5416 ) 5417 5418 def _parse_alter_table_drop(self) -> t.List[exp.Expression]: 5419 index = self._index - 1 5420 5421 partition_exists = self._parse_exists() 5422 if self._match(TokenType.PARTITION, advance=False): 5423 return self._parse_csv(lambda: self._parse_drop_partition(exists=partition_exists)) 5424 5425 self._retreat(index) 5426 return self._parse_csv(self._parse_drop_column) 5427 5428 def _parse_alter_table_rename(self) -> t.Optional[exp.RenameTable | exp.RenameColumn]: 5429 if self._match(TokenType.COLUMN): 5430 exists = self._parse_exists() 5431 old_column = self._parse_column() 5432 to = self._match_text_seq("TO") 5433 new_column = self._parse_column() 5434 5435 if old_column is None or to is None or new_column is None: 5436 return None 5437 5438 return self.expression(exp.RenameColumn, this=old_column, to=new_column, exists=exists) 5439 5440 self._match_text_seq("TO") 5441 return self.expression(exp.RenameTable, this=self._parse_table(schema=True)) 5442 5443 def _parse_alter(self) -> exp.AlterTable | exp.Command: 5444 start = self._prev 5445 5446 if not self._match(TokenType.TABLE): 5447 return self._parse_as_command(start) 5448 5449 exists = self._parse_exists() 5450 only = self._match_text_seq("ONLY") 5451 this = self._parse_table(schema=True) 5452 5453 if self._next: 5454 self._advance() 5455 5456 parser = self.ALTER_PARSERS.get(self._prev.text.upper()) if self._prev else None 5457 if parser: 5458 actions = ensure_list(parser(self)) 5459 5460 if not self._curr and actions: 5461 return self.expression( 5462 exp.AlterTable, 5463 this=this, 5464 exists=exists, 5465 actions=actions, 5466 only=only, 5467 ) 5468 5469 return self._parse_as_command(start) 5470 5471 def _parse_merge(self) -> exp.Merge: 5472 self._match(TokenType.INTO) 5473 target = self._parse_table() 5474 5475 if target and self._match(TokenType.ALIAS, advance=False): 5476 target.set("alias", self._parse_table_alias()) 5477 5478 self._match(TokenType.USING) 5479 using = self._parse_table() 5480 5481 self._match(TokenType.ON) 5482 on = self._parse_conjunction() 5483 5484 return self.expression( 5485 exp.Merge, 5486 this=target, 5487 using=using, 5488 on=on, 5489 expressions=self._parse_when_matched(), 5490 ) 5491 5492 def _parse_when_matched(self) -> t.List[exp.When]: 5493 whens = [] 5494 5495 while self._match(TokenType.WHEN): 5496 matched = not self._match(TokenType.NOT) 5497 self._match_text_seq("MATCHED") 5498 source = ( 5499 False 5500 if self._match_text_seq("BY", "TARGET") 5501 else self._match_text_seq("BY", "SOURCE") 5502 ) 5503 condition = self._parse_conjunction() if self._match(TokenType.AND) else None 5504 5505 self._match(TokenType.THEN) 5506 5507 if self._match(TokenType.INSERT): 5508 _this = self._parse_star() 5509 if _this: 5510 then: t.Optional[exp.Expression] = self.expression(exp.Insert, this=_this) 5511 else: 5512 then = self.expression( 5513 exp.Insert, 5514 this=self._parse_value(), 5515 expression=self._match(TokenType.VALUES) and self._parse_value(), 5516 ) 5517 elif self._match(TokenType.UPDATE): 5518 expressions = self._parse_star() 5519 if expressions: 5520 then = self.expression(exp.Update, expressions=expressions) 5521 else: 5522 then = self.expression( 5523 exp.Update, 5524 expressions=self._match(TokenType.SET) 5525 and self._parse_csv(self._parse_equality), 5526 ) 5527 elif self._match(TokenType.DELETE): 5528 then = self.expression(exp.Var, this=self._prev.text) 5529 else: 5530 then = None 5531 5532 whens.append( 5533 self.expression( 5534 exp.When, 5535 matched=matched, 5536 source=source, 5537 condition=condition, 5538 then=then, 5539 ) 5540 ) 5541 return whens 5542 5543 def _parse_show(self) -> t.Optional[exp.Expression]: 5544 parser = self._find_parser(self.SHOW_PARSERS, self.SHOW_TRIE) 5545 if parser: 5546 return parser(self) 5547 return self._parse_as_command(self._prev) 5548 5549 def _parse_set_item_assignment( 5550 self, kind: t.Optional[str] = None 5551 ) -> t.Optional[exp.Expression]: 5552 index = self._index 5553 5554 if kind in ("GLOBAL", "SESSION") and self._match_text_seq("TRANSACTION"): 5555 return self._parse_set_transaction(global_=kind == "GLOBAL") 5556 5557 left = self._parse_primary() or self._parse_id_var() 5558 assignment_delimiter = self._match_texts(("=", "TO")) 5559 5560 if not left or (self.SET_REQUIRES_ASSIGNMENT_DELIMITER and not assignment_delimiter): 5561 self._retreat(index) 5562 return None 5563 5564 right = self._parse_statement() or self._parse_id_var() 5565 this = self.expression(exp.EQ, this=left, expression=right) 5566 5567 return self.expression(exp.SetItem, this=this, kind=kind) 5568 5569 def _parse_set_transaction(self, global_: bool = False) -> exp.Expression: 5570 self._match_text_seq("TRANSACTION") 5571 characteristics = self._parse_csv( 5572 lambda: self._parse_var_from_options(self.TRANSACTION_CHARACTERISTICS) 5573 ) 5574 return self.expression( 5575 exp.SetItem, 5576 expressions=characteristics, 5577 kind="TRANSACTION", 5578 **{"global": global_}, # type: ignore 5579 ) 5580 5581 def _parse_set_item(self) -> t.Optional[exp.Expression]: 5582 parser = self._find_parser(self.SET_PARSERS, self.SET_TRIE) 5583 return parser(self) if parser else self._parse_set_item_assignment(kind=None) 5584 5585 def _parse_set(self, unset: bool = False, tag: bool = False) -> exp.Set | exp.Command: 5586 index = self._index 5587 set_ = self.expression( 5588 exp.Set, expressions=self._parse_csv(self._parse_set_item), unset=unset, tag=tag 5589 ) 5590 5591 if self._curr: 5592 self._retreat(index) 5593 return self._parse_as_command(self._prev) 5594 5595 return set_ 5596 5597 def _parse_var_from_options(self, options: t.Collection[str]) -> t.Optional[exp.Var]: 5598 for option in options: 5599 if self._match_text_seq(*option.split(" ")): 5600 return exp.var(option) 5601 return None 5602 5603 def _parse_as_command(self, start: Token) -> exp.Command: 5604 while self._curr: 5605 self._advance() 5606 text = self._find_sql(start, self._prev) 5607 size = len(start.text) 5608 self._warn_unsupported() 5609 return exp.Command(this=text[:size], expression=text[size:]) 5610 5611 def _parse_dict_property(self, this: str) -> exp.DictProperty: 5612 settings = [] 5613 5614 self._match_l_paren() 5615 kind = self._parse_id_var() 5616 5617 if self._match(TokenType.L_PAREN): 5618 while True: 5619 key = self._parse_id_var() 5620 value = self._parse_primary() 5621 5622 if not key and value is None: 5623 break 5624 settings.append(self.expression(exp.DictSubProperty, this=key, value=value)) 5625 self._match(TokenType.R_PAREN) 5626 5627 self._match_r_paren() 5628 5629 return self.expression( 5630 exp.DictProperty, 5631 this=this, 5632 kind=kind.this if kind else None, 5633 settings=settings, 5634 ) 5635 5636 def _parse_dict_range(self, this: str) -> exp.DictRange: 5637 self._match_l_paren() 5638 has_min = self._match_text_seq("MIN") 5639 if has_min: 5640 min = self._parse_var() or self._parse_primary() 5641 self._match_text_seq("MAX") 5642 max = self._parse_var() or self._parse_primary() 5643 else: 5644 max = self._parse_var() or self._parse_primary() 5645 min = exp.Literal.number(0) 5646 self._match_r_paren() 5647 return self.expression(exp.DictRange, this=this, min=min, max=max) 5648 5649 def _parse_comprehension( 5650 self, this: t.Optional[exp.Expression] 5651 ) -> t.Optional[exp.Comprehension]: 5652 index = self._index 5653 expression = self._parse_column() 5654 if not self._match(TokenType.IN): 5655 self._retreat(index - 1) 5656 return None 5657 iterator = self._parse_column() 5658 condition = self._parse_conjunction() if self._match_text_seq("IF") else None 5659 return self.expression( 5660 exp.Comprehension, 5661 this=this, 5662 expression=expression, 5663 iterator=iterator, 5664 condition=condition, 5665 ) 5666 5667 def _parse_heredoc(self) -> t.Optional[exp.Heredoc]: 5668 if self._match(TokenType.HEREDOC_STRING): 5669 return self.expression(exp.Heredoc, this=self._prev.text) 5670 5671 if not self._match_text_seq("$"): 5672 return None 5673 5674 tags = ["$"] 5675 tag_text = None 5676 5677 if self._is_connected(): 5678 self._advance() 5679 tags.append(self._prev.text.upper()) 5680 else: 5681 self.raise_error("No closing $ found") 5682 5683 if tags[-1] != "$": 5684 if self._is_connected() and self._match_text_seq("$"): 5685 tag_text = tags[-1] 5686 tags.append("$") 5687 else: 5688 self.raise_error("No closing $ found") 5689 5690 heredoc_start = self._curr 5691 5692 while self._curr: 5693 if self._match_text_seq(*tags, advance=False): 5694 this = self._find_sql(heredoc_start, self._prev) 5695 self._advance(len(tags)) 5696 return self.expression(exp.Heredoc, this=this, tag=tag_text) 5697 5698 self._advance() 5699 5700 self.raise_error(f"No closing {''.join(tags)} found") 5701 return None 5702 5703 def _find_parser( 5704 self, parsers: t.Dict[str, t.Callable], trie: t.Dict 5705 ) -> t.Optional[t.Callable]: 5706 if not self._curr: 5707 return None 5708 5709 index = self._index 5710 this = [] 5711 while True: 5712 # The current token might be multiple words 5713 curr = self._curr.text.upper() 5714 key = curr.split(" ") 5715 this.append(curr) 5716 5717 self._advance() 5718 result, trie = in_trie(trie, key) 5719 if result == TrieResult.FAILED: 5720 break 5721 5722 if result == TrieResult.EXISTS: 5723 subparser = parsers[" ".join(this)] 5724 return subparser 5725 5726 self._retreat(index) 5727 return None 5728 5729 def _match(self, token_type, advance=True, expression=None): 5730 if not self._curr: 5731 return None 5732 5733 if self._curr.token_type == token_type: 5734 if advance: 5735 self._advance() 5736 self._add_comments(expression) 5737 return True 5738 5739 return None 5740 5741 def _match_set(self, types, advance=True): 5742 if not self._curr: 5743 return None 5744 5745 if self._curr.token_type in types: 5746 if advance: 5747 self._advance() 5748 return True 5749 5750 return None 5751 5752 def _match_pair(self, token_type_a, token_type_b, advance=True): 5753 if not self._curr or not self._next: 5754 return None 5755 5756 if self._curr.token_type == token_type_a and self._next.token_type == token_type_b: 5757 if advance: 5758 self._advance(2) 5759 return True 5760 5761 return None 5762 5763 def _match_l_paren(self, expression: t.Optional[exp.Expression] = None) -> None: 5764 if not self._match(TokenType.L_PAREN, expression=expression): 5765 self.raise_error("Expecting (") 5766 5767 def _match_r_paren(self, expression: t.Optional[exp.Expression] = None) -> None: 5768 if not self._match(TokenType.R_PAREN, expression=expression): 5769 self.raise_error("Expecting )") 5770 5771 def _match_texts(self, texts, advance=True): 5772 if self._curr and self._curr.text.upper() in texts: 5773 if advance: 5774 self._advance() 5775 return True 5776 return None 5777 5778 def _match_text_seq(self, *texts, advance=True): 5779 index = self._index 5780 for text in texts: 5781 if self._curr and self._curr.text.upper() == text: 5782 self._advance() 5783 else: 5784 self._retreat(index) 5785 return None 5786 5787 if not advance: 5788 self._retreat(index) 5789 5790 return True 5791 5792 @t.overload 5793 def _replace_columns_with_dots(self, this: exp.Expression) -> exp.Expression: 5794 ... 5795 5796 @t.overload 5797 def _replace_columns_with_dots( 5798 self, this: t.Optional[exp.Expression] 5799 ) -> t.Optional[exp.Expression]: 5800 ... 5801 5802 def _replace_columns_with_dots(self, this): 5803 if isinstance(this, exp.Dot): 5804 exp.replace_children(this, self._replace_columns_with_dots) 5805 elif isinstance(this, exp.Column): 5806 exp.replace_children(this, self._replace_columns_with_dots) 5807 table = this.args.get("table") 5808 this = ( 5809 self.expression(exp.Dot, this=table, expression=this.this) if table else this.this 5810 ) 5811 5812 return this 5813 5814 def _replace_lambda( 5815 self, node: t.Optional[exp.Expression], lambda_variables: t.Set[str] 5816 ) -> t.Optional[exp.Expression]: 5817 if not node: 5818 return node 5819 5820 for column in node.find_all(exp.Column): 5821 if column.parts[0].name in lambda_variables: 5822 dot_or_id = column.to_dot() if column.table else column.this 5823 parent = column.parent 5824 5825 while isinstance(parent, exp.Dot): 5826 if not isinstance(parent.parent, exp.Dot): 5827 parent.replace(dot_or_id) 5828 break 5829 parent = parent.parent 5830 else: 5831 if column is node: 5832 node = dot_or_id 5833 else: 5834 column.replace(dot_or_id) 5835 return node
Parser consumes a list of tokens produced by the Tokenizer and produces a parsed syntax tree.
Arguments:
- error_level: The desired error level. Default: ErrorLevel.IMMEDIATE
- error_message_context: Determines the amount of context to capture from a query string when displaying the error message (in number of characters). Default: 100
- max_errors: Maximum number of error messages to include in a raised ParseError. This is only relevant if error_level is ErrorLevel.RAISE. Default: 3
1026 def __init__( 1027 self, 1028 error_level: t.Optional[ErrorLevel] = None, 1029 error_message_context: int = 100, 1030 max_errors: int = 3, 1031 dialect: DialectType = None, 1032 ): 1033 from sqlglot.dialects import Dialect 1034 1035 self.error_level = error_level or ErrorLevel.IMMEDIATE 1036 self.error_message_context = error_message_context 1037 self.max_errors = max_errors 1038 self.dialect = Dialect.get_or_raise(dialect) 1039 self.reset()
1051 def parse( 1052 self, raw_tokens: t.List[Token], sql: t.Optional[str] = None 1053 ) -> t.List[t.Optional[exp.Expression]]: 1054 """ 1055 Parses a list of tokens and returns a list of syntax trees, one tree 1056 per parsed SQL statement. 1057 1058 Args: 1059 raw_tokens: The list of tokens. 1060 sql: The original SQL string, used to produce helpful debug messages. 1061 1062 Returns: 1063 The list of the produced syntax trees. 1064 """ 1065 return self._parse( 1066 parse_method=self.__class__._parse_statement, raw_tokens=raw_tokens, sql=sql 1067 )
Parses a list of tokens and returns a list of syntax trees, one tree per parsed SQL statement.
Arguments:
- raw_tokens: The list of tokens.
- sql: The original SQL string, used to produce helpful debug messages.
Returns:
The list of the produced syntax trees.
1069 def parse_into( 1070 self, 1071 expression_types: exp.IntoType, 1072 raw_tokens: t.List[Token], 1073 sql: t.Optional[str] = None, 1074 ) -> t.List[t.Optional[exp.Expression]]: 1075 """ 1076 Parses a list of tokens into a given Expression type. If a collection of Expression 1077 types is given instead, this method will try to parse the token list into each one 1078 of them, stopping at the first for which the parsing succeeds. 1079 1080 Args: 1081 expression_types: The expression type(s) to try and parse the token list into. 1082 raw_tokens: The list of tokens. 1083 sql: The original SQL string, used to produce helpful debug messages. 1084 1085 Returns: 1086 The target Expression. 1087 """ 1088 errors = [] 1089 for expression_type in ensure_list(expression_types): 1090 parser = self.EXPRESSION_PARSERS.get(expression_type) 1091 if not parser: 1092 raise TypeError(f"No parser registered for {expression_type}") 1093 1094 try: 1095 return self._parse(parser, raw_tokens, sql) 1096 except ParseError as e: 1097 e.errors[0]["into_expression"] = expression_type 1098 errors.append(e) 1099 1100 raise ParseError( 1101 f"Failed to parse '{sql or raw_tokens}' into {expression_types}", 1102 errors=merge_errors(errors), 1103 ) from errors[-1]
Parses a list of tokens into a given Expression type. If a collection of Expression types is given instead, this method will try to parse the token list into each one of them, stopping at the first for which the parsing succeeds.
Arguments:
- expression_types: The expression type(s) to try and parse the token list into.
- raw_tokens: The list of tokens.
- sql: The original SQL string, used to produce helpful debug messages.
Returns:
The target Expression.
1140 def check_errors(self) -> None: 1141 """Logs or raises any found errors, depending on the chosen error level setting.""" 1142 if self.error_level == ErrorLevel.WARN: 1143 for error in self.errors: 1144 logger.error(str(error)) 1145 elif self.error_level == ErrorLevel.RAISE and self.errors: 1146 raise ParseError( 1147 concat_messages(self.errors, self.max_errors), 1148 errors=merge_errors(self.errors), 1149 )
Logs or raises any found errors, depending on the chosen error level setting.
1151 def raise_error(self, message: str, token: t.Optional[Token] = None) -> None: 1152 """ 1153 Appends an error in the list of recorded errors or raises it, depending on the chosen 1154 error level setting. 1155 """ 1156 token = token or self._curr or self._prev or Token.string("") 1157 start = token.start 1158 end = token.end + 1 1159 start_context = self.sql[max(start - self.error_message_context, 0) : start] 1160 highlight = self.sql[start:end] 1161 end_context = self.sql[end : end + self.error_message_context] 1162 1163 error = ParseError.new( 1164 f"{message}. Line {token.line}, Col: {token.col}.\n" 1165 f" {start_context}\033[4m{highlight}\033[0m{end_context}", 1166 description=message, 1167 line=token.line, 1168 col=token.col, 1169 start_context=start_context, 1170 highlight=highlight, 1171 end_context=end_context, 1172 ) 1173 1174 if self.error_level == ErrorLevel.IMMEDIATE: 1175 raise error 1176 1177 self.errors.append(error)
Appends an error in the list of recorded errors or raises it, depending on the chosen error level setting.
1179 def expression( 1180 self, exp_class: t.Type[E], comments: t.Optional[t.List[str]] = None, **kwargs 1181 ) -> E: 1182 """ 1183 Creates a new, validated Expression. 1184 1185 Args: 1186 exp_class: The expression class to instantiate. 1187 comments: An optional list of comments to attach to the expression. 1188 kwargs: The arguments to set for the expression along with their respective values. 1189 1190 Returns: 1191 The target expression. 1192 """ 1193 instance = exp_class(**kwargs) 1194 instance.add_comments(comments) if comments else self._add_comments(instance) 1195 return self.validate_expression(instance)
Creates a new, validated Expression.
Arguments:
- exp_class: The expression class to instantiate.
- comments: An optional list of comments to attach to the expression.
- kwargs: The arguments to set for the expression along with their respective values.
Returns:
The target expression.
1202 def validate_expression(self, expression: E, args: t.Optional[t.List] = None) -> E: 1203 """ 1204 Validates an Expression, making sure that all its mandatory arguments are set. 1205 1206 Args: 1207 expression: The expression to validate. 1208 args: An optional list of items that was used to instantiate the expression, if it's a Func. 1209 1210 Returns: 1211 The validated expression. 1212 """ 1213 if self.error_level != ErrorLevel.IGNORE: 1214 for error_message in expression.error_messages(args): 1215 self.raise_error(error_message) 1216 1217 return expression
Validates an Expression, making sure that all its mandatory arguments are set.
Arguments:
- expression: The expression to validate.
- args: An optional list of items that was used to instantiate the expression, if it's a Func.
Returns:
The validated expression.