sqlglot.tokens
1from __future__ import annotations 2 3import typing as t 4from enum import auto 5 6from sqlglot.errors import TokenError 7from sqlglot.helper import AutoName 8from sqlglot.trie import TrieResult, in_trie, new_trie 9 10 11class TokenType(AutoName): 12 L_PAREN = auto() 13 R_PAREN = auto() 14 L_BRACKET = auto() 15 R_BRACKET = auto() 16 L_BRACE = auto() 17 R_BRACE = auto() 18 COMMA = auto() 19 DOT = auto() 20 DASH = auto() 21 PLUS = auto() 22 COLON = auto() 23 DCOLON = auto() 24 DQMARK = auto() 25 SEMICOLON = auto() 26 STAR = auto() 27 BACKSLASH = auto() 28 SLASH = auto() 29 LT = auto() 30 LTE = auto() 31 GT = auto() 32 GTE = auto() 33 NOT = auto() 34 EQ = auto() 35 NEQ = auto() 36 NULLSAFE_EQ = auto() 37 COLON_EQ = auto() 38 AND = auto() 39 OR = auto() 40 AMP = auto() 41 DPIPE = auto() 42 PIPE = auto() 43 CARET = auto() 44 TILDA = auto() 45 ARROW = auto() 46 DARROW = auto() 47 FARROW = auto() 48 HASH = auto() 49 HASH_ARROW = auto() 50 DHASH_ARROW = auto() 51 LR_ARROW = auto() 52 DAT = auto() 53 LT_AT = auto() 54 AT_GT = auto() 55 DOLLAR = auto() 56 PARAMETER = auto() 57 SESSION_PARAMETER = auto() 58 DAMP = auto() 59 XOR = auto() 60 DSTAR = auto() 61 62 BLOCK_START = auto() 63 BLOCK_END = auto() 64 65 SPACE = auto() 66 BREAK = auto() 67 68 STRING = auto() 69 NUMBER = auto() 70 IDENTIFIER = auto() 71 DATABASE = auto() 72 COLUMN = auto() 73 COLUMN_DEF = auto() 74 SCHEMA = auto() 75 TABLE = auto() 76 VAR = auto() 77 BIT_STRING = auto() 78 HEX_STRING = auto() 79 BYTE_STRING = auto() 80 NATIONAL_STRING = auto() 81 RAW_STRING = auto() 82 HEREDOC_STRING = auto() 83 84 # types 85 BIT = auto() 86 BOOLEAN = auto() 87 TINYINT = auto() 88 UTINYINT = auto() 89 SMALLINT = auto() 90 USMALLINT = auto() 91 MEDIUMINT = auto() 92 UMEDIUMINT = auto() 93 INT = auto() 94 UINT = auto() 95 BIGINT = auto() 96 UBIGINT = auto() 97 INT128 = auto() 98 UINT128 = auto() 99 INT256 = auto() 100 UINT256 = auto() 101 FLOAT = auto() 102 DOUBLE = auto() 103 DECIMAL = auto() 104 UDECIMAL = auto() 105 BIGDECIMAL = auto() 106 CHAR = auto() 107 NCHAR = auto() 108 VARCHAR = auto() 109 NVARCHAR = auto() 110 TEXT = auto() 111 MEDIUMTEXT = auto() 112 LONGTEXT = auto() 113 MEDIUMBLOB = auto() 114 LONGBLOB = auto() 115 TINYBLOB = auto() 116 TINYTEXT = auto() 117 BINARY = auto() 118 VARBINARY = auto() 119 JSON = auto() 120 JSONB = auto() 121 TIME = auto() 122 TIMETZ = auto() 123 TIMESTAMP = auto() 124 TIMESTAMPTZ = auto() 125 TIMESTAMPLTZ = auto() 126 TIMESTAMP_S = auto() 127 TIMESTAMP_MS = auto() 128 TIMESTAMP_NS = auto() 129 DATETIME = auto() 130 DATETIME64 = auto() 131 DATE = auto() 132 INT4RANGE = auto() 133 INT4MULTIRANGE = auto() 134 INT8RANGE = auto() 135 INT8MULTIRANGE = auto() 136 NUMRANGE = auto() 137 NUMMULTIRANGE = auto() 138 TSRANGE = auto() 139 TSMULTIRANGE = auto() 140 TSTZRANGE = auto() 141 TSTZMULTIRANGE = auto() 142 DATERANGE = auto() 143 DATEMULTIRANGE = auto() 144 UUID = auto() 145 GEOGRAPHY = auto() 146 NULLABLE = auto() 147 GEOMETRY = auto() 148 HLLSKETCH = auto() 149 HSTORE = auto() 150 SUPER = auto() 151 SERIAL = auto() 152 SMALLSERIAL = auto() 153 BIGSERIAL = auto() 154 XML = auto() 155 YEAR = auto() 156 UNIQUEIDENTIFIER = auto() 157 USERDEFINED = auto() 158 MONEY = auto() 159 SMALLMONEY = auto() 160 ROWVERSION = auto() 161 IMAGE = auto() 162 VARIANT = auto() 163 OBJECT = auto() 164 INET = auto() 165 IPADDRESS = auto() 166 IPPREFIX = auto() 167 ENUM = auto() 168 ENUM8 = auto() 169 ENUM16 = auto() 170 FIXEDSTRING = auto() 171 LOWCARDINALITY = auto() 172 NESTED = auto() 173 UNKNOWN = auto() 174 175 # keywords 176 ALIAS = auto() 177 ALTER = auto() 178 ALWAYS = auto() 179 ALL = auto() 180 ANTI = auto() 181 ANY = auto() 182 APPLY = auto() 183 ARRAY = auto() 184 ASC = auto() 185 ASOF = auto() 186 AUTO_INCREMENT = auto() 187 BEGIN = auto() 188 BETWEEN = auto() 189 CACHE = auto() 190 CASE = auto() 191 CHARACTER_SET = auto() 192 CLUSTER_BY = auto() 193 COLLATE = auto() 194 COMMAND = auto() 195 COMMENT = auto() 196 COMMIT = auto() 197 CONNECT_BY = auto() 198 CONSTRAINT = auto() 199 CREATE = auto() 200 CROSS = auto() 201 CUBE = auto() 202 CURRENT_DATE = auto() 203 CURRENT_DATETIME = auto() 204 CURRENT_TIME = auto() 205 CURRENT_TIMESTAMP = auto() 206 CURRENT_USER = auto() 207 DEFAULT = auto() 208 DELETE = auto() 209 DESC = auto() 210 DESCRIBE = auto() 211 DICTIONARY = auto() 212 DISTINCT = auto() 213 DISTRIBUTE_BY = auto() 214 DIV = auto() 215 DROP = auto() 216 ELSE = auto() 217 END = auto() 218 ESCAPE = auto() 219 EXCEPT = auto() 220 EXECUTE = auto() 221 EXISTS = auto() 222 FALSE = auto() 223 FETCH = auto() 224 FILTER = auto() 225 FINAL = auto() 226 FIRST = auto() 227 FOR = auto() 228 FORCE = auto() 229 FOREIGN_KEY = auto() 230 FORMAT = auto() 231 FROM = auto() 232 FULL = auto() 233 FUNCTION = auto() 234 GLOB = auto() 235 GLOBAL = auto() 236 GROUP_BY = auto() 237 GROUPING_SETS = auto() 238 HAVING = auto() 239 HINT = auto() 240 IGNORE = auto() 241 ILIKE = auto() 242 ILIKE_ANY = auto() 243 IN = auto() 244 INDEX = auto() 245 INNER = auto() 246 INSERT = auto() 247 INTERSECT = auto() 248 INTERVAL = auto() 249 INTO = auto() 250 INTRODUCER = auto() 251 IRLIKE = auto() 252 IS = auto() 253 ISNULL = auto() 254 JOIN = auto() 255 JOIN_MARKER = auto() 256 KEEP = auto() 257 KILL = auto() 258 LANGUAGE = auto() 259 LATERAL = auto() 260 LEFT = auto() 261 LIKE = auto() 262 LIKE_ANY = auto() 263 LIMIT = auto() 264 LOAD = auto() 265 LOCK = auto() 266 MAP = auto() 267 MATCH_RECOGNIZE = auto() 268 MEMBER_OF = auto() 269 MERGE = auto() 270 MOD = auto() 271 MODEL = auto() 272 NATURAL = auto() 273 NEXT = auto() 274 NOTNULL = auto() 275 NULL = auto() 276 OBJECT_IDENTIFIER = auto() 277 OFFSET = auto() 278 ON = auto() 279 ORDER_BY = auto() 280 ORDERED = auto() 281 ORDINALITY = auto() 282 OUTER = auto() 283 OVER = auto() 284 OVERLAPS = auto() 285 OVERWRITE = auto() 286 PARTITION = auto() 287 PARTITION_BY = auto() 288 PERCENT = auto() 289 PIVOT = auto() 290 PLACEHOLDER = auto() 291 PRAGMA = auto() 292 PRIMARY_KEY = auto() 293 PROCEDURE = auto() 294 PROPERTIES = auto() 295 PSEUDO_TYPE = auto() 296 QUALIFY = auto() 297 QUOTE = auto() 298 RANGE = auto() 299 RECURSIVE = auto() 300 REFRESH = auto() 301 REPLACE = auto() 302 RETURNING = auto() 303 REFERENCES = auto() 304 RIGHT = auto() 305 RLIKE = auto() 306 ROLLBACK = auto() 307 ROLLUP = auto() 308 ROW = auto() 309 ROWS = auto() 310 SELECT = auto() 311 SEMI = auto() 312 SEPARATOR = auto() 313 SERDE_PROPERTIES = auto() 314 SET = auto() 315 SETTINGS = auto() 316 SHOW = auto() 317 SIMILAR_TO = auto() 318 SOME = auto() 319 SORT_BY = auto() 320 START_WITH = auto() 321 STRUCT = auto() 322 TABLE_SAMPLE = auto() 323 TEMPORARY = auto() 324 TOP = auto() 325 THEN = auto() 326 TRUE = auto() 327 UNCACHE = auto() 328 UNION = auto() 329 UNNEST = auto() 330 UNPIVOT = auto() 331 UPDATE = auto() 332 USE = auto() 333 USING = auto() 334 VALUES = auto() 335 VIEW = auto() 336 VOLATILE = auto() 337 WHEN = auto() 338 WHERE = auto() 339 WINDOW = auto() 340 WITH = auto() 341 UNIQUE = auto() 342 VERSION_SNAPSHOT = auto() 343 TIMESTAMP_SNAPSHOT = auto() 344 345 346class Token: 347 __slots__ = ("token_type", "text", "line", "col", "start", "end", "comments") 348 349 @classmethod 350 def number(cls, number: int) -> Token: 351 """Returns a NUMBER token with `number` as its text.""" 352 return cls(TokenType.NUMBER, str(number)) 353 354 @classmethod 355 def string(cls, string: str) -> Token: 356 """Returns a STRING token with `string` as its text.""" 357 return cls(TokenType.STRING, string) 358 359 @classmethod 360 def identifier(cls, identifier: str) -> Token: 361 """Returns an IDENTIFIER token with `identifier` as its text.""" 362 return cls(TokenType.IDENTIFIER, identifier) 363 364 @classmethod 365 def var(cls, var: str) -> Token: 366 """Returns an VAR token with `var` as its text.""" 367 return cls(TokenType.VAR, var) 368 369 def __init__( 370 self, 371 token_type: TokenType, 372 text: str, 373 line: int = 1, 374 col: int = 1, 375 start: int = 0, 376 end: int = 0, 377 comments: t.Optional[t.List[str]] = None, 378 ) -> None: 379 """Token initializer. 380 381 Args: 382 token_type: The TokenType Enum. 383 text: The text of the token. 384 line: The line that the token ends on. 385 col: The column that the token ends on. 386 start: The start index of the token. 387 end: The ending index of the token. 388 comments: The comments to attach to the token. 389 """ 390 self.token_type = token_type 391 self.text = text 392 self.line = line 393 self.col = col 394 self.start = start 395 self.end = end 396 self.comments = [] if comments is None else comments 397 398 def __repr__(self) -> str: 399 attributes = ", ".join(f"{k}: {getattr(self, k)}" for k in self.__slots__) 400 return f"<Token {attributes}>" 401 402 403class _Tokenizer(type): 404 def __new__(cls, clsname, bases, attrs): 405 klass = super().__new__(cls, clsname, bases, attrs) 406 407 def _convert_quotes(arr: t.List[str | t.Tuple[str, str]]) -> t.Dict[str, str]: 408 return dict( 409 (item, item) if isinstance(item, str) else (item[0], item[1]) for item in arr 410 ) 411 412 def _quotes_to_format( 413 token_type: TokenType, arr: t.List[str | t.Tuple[str, str]] 414 ) -> t.Dict[str, t.Tuple[str, TokenType]]: 415 return {k: (v, token_type) for k, v in _convert_quotes(arr).items()} 416 417 klass._QUOTES = _convert_quotes(klass.QUOTES) 418 klass._IDENTIFIERS = _convert_quotes(klass.IDENTIFIERS) 419 420 klass._FORMAT_STRINGS = { 421 **{ 422 p + s: (e, TokenType.NATIONAL_STRING) 423 for s, e in klass._QUOTES.items() 424 for p in ("n", "N") 425 }, 426 **_quotes_to_format(TokenType.BIT_STRING, klass.BIT_STRINGS), 427 **_quotes_to_format(TokenType.BYTE_STRING, klass.BYTE_STRINGS), 428 **_quotes_to_format(TokenType.HEX_STRING, klass.HEX_STRINGS), 429 **_quotes_to_format(TokenType.RAW_STRING, klass.RAW_STRINGS), 430 **_quotes_to_format(TokenType.HEREDOC_STRING, klass.HEREDOC_STRINGS), 431 } 432 433 klass._STRING_ESCAPES = set(klass.STRING_ESCAPES) 434 klass._IDENTIFIER_ESCAPES = set(klass.IDENTIFIER_ESCAPES) 435 klass._COMMENTS = { 436 **dict( 437 (comment, None) if isinstance(comment, str) else (comment[0], comment[1]) 438 for comment in klass.COMMENTS 439 ), 440 "{#": "#}", # Ensure Jinja comments are tokenized correctly in all dialects 441 } 442 443 klass._KEYWORD_TRIE = new_trie( 444 key.upper() 445 for key in ( 446 *klass.KEYWORDS, 447 *klass._COMMENTS, 448 *klass._QUOTES, 449 *klass._FORMAT_STRINGS, 450 ) 451 if " " in key or any(single in key for single in klass.SINGLE_TOKENS) 452 ) 453 454 return klass 455 456 457class Tokenizer(metaclass=_Tokenizer): 458 SINGLE_TOKENS = { 459 "(": TokenType.L_PAREN, 460 ")": TokenType.R_PAREN, 461 "[": TokenType.L_BRACKET, 462 "]": TokenType.R_BRACKET, 463 "{": TokenType.L_BRACE, 464 "}": TokenType.R_BRACE, 465 "&": TokenType.AMP, 466 "^": TokenType.CARET, 467 ":": TokenType.COLON, 468 ",": TokenType.COMMA, 469 ".": TokenType.DOT, 470 "-": TokenType.DASH, 471 "=": TokenType.EQ, 472 ">": TokenType.GT, 473 "<": TokenType.LT, 474 "%": TokenType.MOD, 475 "!": TokenType.NOT, 476 "|": TokenType.PIPE, 477 "+": TokenType.PLUS, 478 ";": TokenType.SEMICOLON, 479 "/": TokenType.SLASH, 480 "\\": TokenType.BACKSLASH, 481 "*": TokenType.STAR, 482 "~": TokenType.TILDA, 483 "?": TokenType.PLACEHOLDER, 484 "@": TokenType.PARAMETER, 485 # used for breaking a var like x'y' but nothing else 486 # the token type doesn't matter 487 "'": TokenType.QUOTE, 488 "`": TokenType.IDENTIFIER, 489 '"': TokenType.IDENTIFIER, 490 "#": TokenType.HASH, 491 } 492 493 BIT_STRINGS: t.List[str | t.Tuple[str, str]] = [] 494 BYTE_STRINGS: t.List[str | t.Tuple[str, str]] = [] 495 HEX_STRINGS: t.List[str | t.Tuple[str, str]] = [] 496 RAW_STRINGS: t.List[str | t.Tuple[str, str]] = [] 497 HEREDOC_STRINGS: t.List[str | t.Tuple[str, str]] = [] 498 IDENTIFIERS: t.List[str | t.Tuple[str, str]] = ['"'] 499 IDENTIFIER_ESCAPES = ['"'] 500 QUOTES: t.List[t.Tuple[str, str] | str] = ["'"] 501 STRING_ESCAPES = ["'"] 502 VAR_SINGLE_TOKENS: t.Set[str] = set() 503 ESCAPE_SEQUENCES: t.Dict[str, str] = {} 504 505 # Autofilled 506 IDENTIFIERS_CAN_START_WITH_DIGIT: bool = False 507 508 _COMMENTS: t.Dict[str, str] = {} 509 _FORMAT_STRINGS: t.Dict[str, t.Tuple[str, TokenType]] = {} 510 _IDENTIFIERS: t.Dict[str, str] = {} 511 _IDENTIFIER_ESCAPES: t.Set[str] = set() 512 _QUOTES: t.Dict[str, str] = {} 513 _STRING_ESCAPES: t.Set[str] = set() 514 _KEYWORD_TRIE: t.Dict = {} 515 516 KEYWORDS: t.Dict[str, TokenType] = { 517 **{f"{{%{postfix}": TokenType.BLOCK_START for postfix in ("", "+", "-")}, 518 **{f"{prefix}%}}": TokenType.BLOCK_END for prefix in ("", "+", "-")}, 519 **{f"{{{{{postfix}": TokenType.BLOCK_START for postfix in ("+", "-")}, 520 **{f"{prefix}}}}}": TokenType.BLOCK_END for prefix in ("+", "-")}, 521 "/*+": TokenType.HINT, 522 "==": TokenType.EQ, 523 "::": TokenType.DCOLON, 524 "||": TokenType.DPIPE, 525 ">=": TokenType.GTE, 526 "<=": TokenType.LTE, 527 "<>": TokenType.NEQ, 528 "!=": TokenType.NEQ, 529 "<=>": TokenType.NULLSAFE_EQ, 530 "->": TokenType.ARROW, 531 "->>": TokenType.DARROW, 532 "=>": TokenType.FARROW, 533 "#>": TokenType.HASH_ARROW, 534 "#>>": TokenType.DHASH_ARROW, 535 "<->": TokenType.LR_ARROW, 536 "&&": TokenType.DAMP, 537 "??": TokenType.DQMARK, 538 "ALL": TokenType.ALL, 539 "ALWAYS": TokenType.ALWAYS, 540 "AND": TokenType.AND, 541 "ANTI": TokenType.ANTI, 542 "ANY": TokenType.ANY, 543 "ASC": TokenType.ASC, 544 "AS": TokenType.ALIAS, 545 "ASOF": TokenType.ASOF, 546 "AUTOINCREMENT": TokenType.AUTO_INCREMENT, 547 "AUTO_INCREMENT": TokenType.AUTO_INCREMENT, 548 "BEGIN": TokenType.BEGIN, 549 "BETWEEN": TokenType.BETWEEN, 550 "CACHE": TokenType.CACHE, 551 "UNCACHE": TokenType.UNCACHE, 552 "CASE": TokenType.CASE, 553 "CHARACTER SET": TokenType.CHARACTER_SET, 554 "CLUSTER BY": TokenType.CLUSTER_BY, 555 "COLLATE": TokenType.COLLATE, 556 "COLUMN": TokenType.COLUMN, 557 "COMMIT": TokenType.COMMIT, 558 "CONNECT BY": TokenType.CONNECT_BY, 559 "CONSTRAINT": TokenType.CONSTRAINT, 560 "CREATE": TokenType.CREATE, 561 "CROSS": TokenType.CROSS, 562 "CUBE": TokenType.CUBE, 563 "CURRENT_DATE": TokenType.CURRENT_DATE, 564 "CURRENT_TIME": TokenType.CURRENT_TIME, 565 "CURRENT_TIMESTAMP": TokenType.CURRENT_TIMESTAMP, 566 "CURRENT_USER": TokenType.CURRENT_USER, 567 "DATABASE": TokenType.DATABASE, 568 "DEFAULT": TokenType.DEFAULT, 569 "DELETE": TokenType.DELETE, 570 "DESC": TokenType.DESC, 571 "DESCRIBE": TokenType.DESCRIBE, 572 "DISTINCT": TokenType.DISTINCT, 573 "DISTRIBUTE BY": TokenType.DISTRIBUTE_BY, 574 "DIV": TokenType.DIV, 575 "DROP": TokenType.DROP, 576 "ELSE": TokenType.ELSE, 577 "END": TokenType.END, 578 "ESCAPE": TokenType.ESCAPE, 579 "EXCEPT": TokenType.EXCEPT, 580 "EXECUTE": TokenType.EXECUTE, 581 "EXISTS": TokenType.EXISTS, 582 "FALSE": TokenType.FALSE, 583 "FETCH": TokenType.FETCH, 584 "FILTER": TokenType.FILTER, 585 "FIRST": TokenType.FIRST, 586 "FULL": TokenType.FULL, 587 "FUNCTION": TokenType.FUNCTION, 588 "FOR": TokenType.FOR, 589 "FOREIGN KEY": TokenType.FOREIGN_KEY, 590 "FORMAT": TokenType.FORMAT, 591 "FROM": TokenType.FROM, 592 "GEOGRAPHY": TokenType.GEOGRAPHY, 593 "GEOMETRY": TokenType.GEOMETRY, 594 "GLOB": TokenType.GLOB, 595 "GROUP BY": TokenType.GROUP_BY, 596 "GROUPING SETS": TokenType.GROUPING_SETS, 597 "HAVING": TokenType.HAVING, 598 "ILIKE": TokenType.ILIKE, 599 "IN": TokenType.IN, 600 "INDEX": TokenType.INDEX, 601 "INET": TokenType.INET, 602 "INNER": TokenType.INNER, 603 "INSERT": TokenType.INSERT, 604 "INTERVAL": TokenType.INTERVAL, 605 "INTERSECT": TokenType.INTERSECT, 606 "INTO": TokenType.INTO, 607 "IS": TokenType.IS, 608 "ISNULL": TokenType.ISNULL, 609 "JOIN": TokenType.JOIN, 610 "KEEP": TokenType.KEEP, 611 "KILL": TokenType.KILL, 612 "LATERAL": TokenType.LATERAL, 613 "LEFT": TokenType.LEFT, 614 "LIKE": TokenType.LIKE, 615 "LIMIT": TokenType.LIMIT, 616 "LOAD": TokenType.LOAD, 617 "LOCK": TokenType.LOCK, 618 "MERGE": TokenType.MERGE, 619 "NATURAL": TokenType.NATURAL, 620 "NEXT": TokenType.NEXT, 621 "NOT": TokenType.NOT, 622 "NOTNULL": TokenType.NOTNULL, 623 "NULL": TokenType.NULL, 624 "OBJECT": TokenType.OBJECT, 625 "OFFSET": TokenType.OFFSET, 626 "ON": TokenType.ON, 627 "OR": TokenType.OR, 628 "XOR": TokenType.XOR, 629 "ORDER BY": TokenType.ORDER_BY, 630 "ORDINALITY": TokenType.ORDINALITY, 631 "OUTER": TokenType.OUTER, 632 "OVER": TokenType.OVER, 633 "OVERLAPS": TokenType.OVERLAPS, 634 "OVERWRITE": TokenType.OVERWRITE, 635 "PARTITION": TokenType.PARTITION, 636 "PARTITION BY": TokenType.PARTITION_BY, 637 "PARTITIONED BY": TokenType.PARTITION_BY, 638 "PARTITIONED_BY": TokenType.PARTITION_BY, 639 "PERCENT": TokenType.PERCENT, 640 "PIVOT": TokenType.PIVOT, 641 "PRAGMA": TokenType.PRAGMA, 642 "PRIMARY KEY": TokenType.PRIMARY_KEY, 643 "PROCEDURE": TokenType.PROCEDURE, 644 "QUALIFY": TokenType.QUALIFY, 645 "RANGE": TokenType.RANGE, 646 "RECURSIVE": TokenType.RECURSIVE, 647 "REGEXP": TokenType.RLIKE, 648 "REPLACE": TokenType.REPLACE, 649 "RETURNING": TokenType.RETURNING, 650 "REFERENCES": TokenType.REFERENCES, 651 "RIGHT": TokenType.RIGHT, 652 "RLIKE": TokenType.RLIKE, 653 "ROLLBACK": TokenType.ROLLBACK, 654 "ROLLUP": TokenType.ROLLUP, 655 "ROW": TokenType.ROW, 656 "ROWS": TokenType.ROWS, 657 "SCHEMA": TokenType.SCHEMA, 658 "SELECT": TokenType.SELECT, 659 "SEMI": TokenType.SEMI, 660 "SET": TokenType.SET, 661 "SETTINGS": TokenType.SETTINGS, 662 "SHOW": TokenType.SHOW, 663 "SIMILAR TO": TokenType.SIMILAR_TO, 664 "SOME": TokenType.SOME, 665 "SORT BY": TokenType.SORT_BY, 666 "START WITH": TokenType.START_WITH, 667 "TABLE": TokenType.TABLE, 668 "TABLESAMPLE": TokenType.TABLE_SAMPLE, 669 "TEMP": TokenType.TEMPORARY, 670 "TEMPORARY": TokenType.TEMPORARY, 671 "THEN": TokenType.THEN, 672 "TRUE": TokenType.TRUE, 673 "UNION": TokenType.UNION, 674 "UNKNOWN": TokenType.UNKNOWN, 675 "UNNEST": TokenType.UNNEST, 676 "UNPIVOT": TokenType.UNPIVOT, 677 "UPDATE": TokenType.UPDATE, 678 "USE": TokenType.USE, 679 "USING": TokenType.USING, 680 "UUID": TokenType.UUID, 681 "VALUES": TokenType.VALUES, 682 "VIEW": TokenType.VIEW, 683 "VOLATILE": TokenType.VOLATILE, 684 "WHEN": TokenType.WHEN, 685 "WHERE": TokenType.WHERE, 686 "WINDOW": TokenType.WINDOW, 687 "WITH": TokenType.WITH, 688 "APPLY": TokenType.APPLY, 689 "ARRAY": TokenType.ARRAY, 690 "BIT": TokenType.BIT, 691 "BOOL": TokenType.BOOLEAN, 692 "BOOLEAN": TokenType.BOOLEAN, 693 "BYTE": TokenType.TINYINT, 694 "MEDIUMINT": TokenType.MEDIUMINT, 695 "INT1": TokenType.TINYINT, 696 "TINYINT": TokenType.TINYINT, 697 "INT16": TokenType.SMALLINT, 698 "SHORT": TokenType.SMALLINT, 699 "SMALLINT": TokenType.SMALLINT, 700 "INT128": TokenType.INT128, 701 "HUGEINT": TokenType.INT128, 702 "INT2": TokenType.SMALLINT, 703 "INTEGER": TokenType.INT, 704 "INT": TokenType.INT, 705 "INT4": TokenType.INT, 706 "INT32": TokenType.INT, 707 "INT64": TokenType.BIGINT, 708 "LONG": TokenType.BIGINT, 709 "BIGINT": TokenType.BIGINT, 710 "INT8": TokenType.TINYINT, 711 "DEC": TokenType.DECIMAL, 712 "DECIMAL": TokenType.DECIMAL, 713 "BIGDECIMAL": TokenType.BIGDECIMAL, 714 "BIGNUMERIC": TokenType.BIGDECIMAL, 715 "MAP": TokenType.MAP, 716 "NULLABLE": TokenType.NULLABLE, 717 "NUMBER": TokenType.DECIMAL, 718 "NUMERIC": TokenType.DECIMAL, 719 "FIXED": TokenType.DECIMAL, 720 "REAL": TokenType.FLOAT, 721 "FLOAT": TokenType.FLOAT, 722 "FLOAT4": TokenType.FLOAT, 723 "FLOAT8": TokenType.DOUBLE, 724 "DOUBLE": TokenType.DOUBLE, 725 "DOUBLE PRECISION": TokenType.DOUBLE, 726 "JSON": TokenType.JSON, 727 "CHAR": TokenType.CHAR, 728 "CHARACTER": TokenType.CHAR, 729 "NCHAR": TokenType.NCHAR, 730 "VARCHAR": TokenType.VARCHAR, 731 "VARCHAR2": TokenType.VARCHAR, 732 "NVARCHAR": TokenType.NVARCHAR, 733 "NVARCHAR2": TokenType.NVARCHAR, 734 "STR": TokenType.TEXT, 735 "STRING": TokenType.TEXT, 736 "TEXT": TokenType.TEXT, 737 "LONGTEXT": TokenType.LONGTEXT, 738 "MEDIUMTEXT": TokenType.MEDIUMTEXT, 739 "TINYTEXT": TokenType.TINYTEXT, 740 "CLOB": TokenType.TEXT, 741 "LONGVARCHAR": TokenType.TEXT, 742 "BINARY": TokenType.BINARY, 743 "BLOB": TokenType.VARBINARY, 744 "LONGBLOB": TokenType.LONGBLOB, 745 "MEDIUMBLOB": TokenType.MEDIUMBLOB, 746 "TINYBLOB": TokenType.TINYBLOB, 747 "BYTEA": TokenType.VARBINARY, 748 "VARBINARY": TokenType.VARBINARY, 749 "TIME": TokenType.TIME, 750 "TIMETZ": TokenType.TIMETZ, 751 "TIMESTAMP": TokenType.TIMESTAMP, 752 "TIMESTAMPTZ": TokenType.TIMESTAMPTZ, 753 "TIMESTAMPLTZ": TokenType.TIMESTAMPLTZ, 754 "DATE": TokenType.DATE, 755 "DATETIME": TokenType.DATETIME, 756 "INT4RANGE": TokenType.INT4RANGE, 757 "INT4MULTIRANGE": TokenType.INT4MULTIRANGE, 758 "INT8RANGE": TokenType.INT8RANGE, 759 "INT8MULTIRANGE": TokenType.INT8MULTIRANGE, 760 "NUMRANGE": TokenType.NUMRANGE, 761 "NUMMULTIRANGE": TokenType.NUMMULTIRANGE, 762 "TSRANGE": TokenType.TSRANGE, 763 "TSMULTIRANGE": TokenType.TSMULTIRANGE, 764 "TSTZRANGE": TokenType.TSTZRANGE, 765 "TSTZMULTIRANGE": TokenType.TSTZMULTIRANGE, 766 "DATERANGE": TokenType.DATERANGE, 767 "DATEMULTIRANGE": TokenType.DATEMULTIRANGE, 768 "UNIQUE": TokenType.UNIQUE, 769 "STRUCT": TokenType.STRUCT, 770 "VARIANT": TokenType.VARIANT, 771 "ALTER": TokenType.ALTER, 772 "ANALYZE": TokenType.COMMAND, 773 "CALL": TokenType.COMMAND, 774 "COMMENT": TokenType.COMMENT, 775 "COPY": TokenType.COMMAND, 776 "EXPLAIN": TokenType.COMMAND, 777 "GRANT": TokenType.COMMAND, 778 "OPTIMIZE": TokenType.COMMAND, 779 "PREPARE": TokenType.COMMAND, 780 "TRUNCATE": TokenType.COMMAND, 781 "VACUUM": TokenType.COMMAND, 782 "USER-DEFINED": TokenType.USERDEFINED, 783 "FOR VERSION": TokenType.VERSION_SNAPSHOT, 784 "FOR TIMESTAMP": TokenType.TIMESTAMP_SNAPSHOT, 785 } 786 787 WHITE_SPACE: t.Dict[t.Optional[str], TokenType] = { 788 " ": TokenType.SPACE, 789 "\t": TokenType.SPACE, 790 "\n": TokenType.BREAK, 791 "\r": TokenType.BREAK, 792 "\r\n": TokenType.BREAK, 793 } 794 795 COMMANDS = { 796 TokenType.COMMAND, 797 TokenType.EXECUTE, 798 TokenType.FETCH, 799 TokenType.SHOW, 800 } 801 802 COMMAND_PREFIX_TOKENS = {TokenType.SEMICOLON, TokenType.BEGIN} 803 804 # handle numeric literals like in hive (3L = BIGINT) 805 NUMERIC_LITERALS: t.Dict[str, str] = {} 806 ENCODE: t.Optional[str] = None 807 808 COMMENTS = ["--", ("/*", "*/")] 809 810 __slots__ = ( 811 "sql", 812 "size", 813 "tokens", 814 "_start", 815 "_current", 816 "_line", 817 "_col", 818 "_comments", 819 "_char", 820 "_end", 821 "_peek", 822 "_prev_token_line", 823 ) 824 825 def __init__(self) -> None: 826 self.reset() 827 828 def reset(self) -> None: 829 self.sql = "" 830 self.size = 0 831 self.tokens: t.List[Token] = [] 832 self._start = 0 833 self._current = 0 834 self._line = 1 835 self._col = 0 836 self._comments: t.List[str] = [] 837 838 self._char = "" 839 self._end = False 840 self._peek = "" 841 self._prev_token_line = -1 842 843 def tokenize(self, sql: str) -> t.List[Token]: 844 """Returns a list of tokens corresponding to the SQL string `sql`.""" 845 self.reset() 846 self.sql = sql 847 self.size = len(sql) 848 849 try: 850 self._scan() 851 except Exception as e: 852 start = max(self._current - 50, 0) 853 end = min(self._current + 50, self.size - 1) 854 context = self.sql[start:end] 855 raise TokenError(f"Error tokenizing '{context}'") from e 856 857 return self.tokens 858 859 def _scan(self, until: t.Optional[t.Callable] = None) -> None: 860 while self.size and not self._end: 861 current = self._current 862 863 # skip spaces inline rather than iteratively call advance() 864 # for performance reasons 865 while current < self.size: 866 char = self.sql[current] 867 868 if char.isspace() and (char == " " or char == "\t"): 869 current += 1 870 else: 871 break 872 873 n = current - self._current 874 self._start = current 875 self._advance(n if n > 1 else 1) 876 877 if self._char is None: 878 break 879 880 if not self._char.isspace(): 881 if self._char.isdigit(): 882 self._scan_number() 883 elif self._char in self._IDENTIFIERS: 884 self._scan_identifier(self._IDENTIFIERS[self._char]) 885 else: 886 self._scan_keywords() 887 888 if until and until(): 889 break 890 891 if self.tokens and self._comments: 892 self.tokens[-1].comments.extend(self._comments) 893 894 def _chars(self, size: int) -> str: 895 if size == 1: 896 return self._char 897 898 start = self._current - 1 899 end = start + size 900 901 return self.sql[start:end] if end <= self.size else "" 902 903 def _advance(self, i: int = 1, alnum: bool = False) -> None: 904 if self.WHITE_SPACE.get(self._char) is TokenType.BREAK: 905 self._col = 1 906 self._line += 1 907 else: 908 self._col += i 909 910 self._current += i 911 self._end = self._current >= self.size 912 self._char = self.sql[self._current - 1] 913 self._peek = "" if self._end else self.sql[self._current] 914 915 if alnum and self._char.isalnum(): 916 # Here we use local variables instead of attributes for better performance 917 _col = self._col 918 _current = self._current 919 _end = self._end 920 _peek = self._peek 921 922 while _peek.isalnum(): 923 _col += 1 924 _current += 1 925 _end = _current >= self.size 926 _peek = "" if _end else self.sql[_current] 927 928 self._col = _col 929 self._current = _current 930 self._end = _end 931 self._peek = _peek 932 self._char = self.sql[_current - 1] 933 934 @property 935 def _text(self) -> str: 936 return self.sql[self._start : self._current] 937 938 def peek(self, i: int = 0) -> str: 939 i = self._current + i 940 if i < self.size: 941 return self.sql[i] 942 return "" 943 944 def _add(self, token_type: TokenType, text: t.Optional[str] = None) -> None: 945 self._prev_token_line = self._line 946 947 if self._comments and token_type == TokenType.SEMICOLON and self.tokens: 948 self.tokens[-1].comments.extend(self._comments) 949 self._comments = [] 950 951 self.tokens.append( 952 Token( 953 token_type, 954 text=self._text if text is None else text, 955 line=self._line, 956 col=self._col, 957 start=self._start, 958 end=self._current - 1, 959 comments=self._comments, 960 ) 961 ) 962 self._comments = [] 963 964 # If we have either a semicolon or a begin token before the command's token, we'll parse 965 # whatever follows the command's token as a string 966 if ( 967 token_type in self.COMMANDS 968 and self._peek != ";" 969 and (len(self.tokens) == 1 or self.tokens[-2].token_type in self.COMMAND_PREFIX_TOKENS) 970 ): 971 start = self._current 972 tokens = len(self.tokens) 973 self._scan(lambda: self._peek == ";") 974 self.tokens = self.tokens[:tokens] 975 text = self.sql[start : self._current].strip() 976 if text: 977 self._add(TokenType.STRING, text) 978 979 def _scan_keywords(self) -> None: 980 size = 0 981 word = None 982 chars = self._text 983 char = chars 984 prev_space = False 985 skip = False 986 trie = self._KEYWORD_TRIE 987 single_token = char in self.SINGLE_TOKENS 988 989 while chars: 990 if skip: 991 result = TrieResult.PREFIX 992 else: 993 result, trie = in_trie(trie, char.upper()) 994 995 if result == TrieResult.FAILED: 996 break 997 if result == TrieResult.EXISTS: 998 word = chars 999 1000 end = self._current + size 1001 size += 1 1002 1003 if end < self.size: 1004 char = self.sql[end] 1005 single_token = single_token or char in self.SINGLE_TOKENS 1006 is_space = char.isspace() 1007 1008 if not is_space or not prev_space: 1009 if is_space: 1010 char = " " 1011 chars += char 1012 prev_space = is_space 1013 skip = False 1014 else: 1015 skip = True 1016 else: 1017 char = "" 1018 chars = " " 1019 1020 if word: 1021 if self._scan_string(word): 1022 return 1023 if self._scan_comment(word): 1024 return 1025 if prev_space or single_token or not char: 1026 self._advance(size - 1) 1027 word = word.upper() 1028 self._add(self.KEYWORDS[word], text=word) 1029 return 1030 1031 if self._char in self.SINGLE_TOKENS: 1032 self._add(self.SINGLE_TOKENS[self._char], text=self._char) 1033 return 1034 1035 self._scan_var() 1036 1037 def _scan_comment(self, comment_start: str) -> bool: 1038 if comment_start not in self._COMMENTS: 1039 return False 1040 1041 comment_start_line = self._line 1042 comment_start_size = len(comment_start) 1043 comment_end = self._COMMENTS[comment_start] 1044 1045 if comment_end: 1046 # Skip the comment's start delimiter 1047 self._advance(comment_start_size) 1048 1049 comment_end_size = len(comment_end) 1050 while not self._end and self._chars(comment_end_size) != comment_end: 1051 self._advance(alnum=True) 1052 1053 self._comments.append(self._text[comment_start_size : -comment_end_size + 1]) 1054 self._advance(comment_end_size - 1) 1055 else: 1056 while not self._end and not self.WHITE_SPACE.get(self._peek) is TokenType.BREAK: 1057 self._advance(alnum=True) 1058 self._comments.append(self._text[comment_start_size:]) 1059 1060 # Leading comment is attached to the succeeding token, whilst trailing comment to the preceding. 1061 # Multiple consecutive comments are preserved by appending them to the current comments list. 1062 if comment_start_line == self._prev_token_line: 1063 self.tokens[-1].comments.extend(self._comments) 1064 self._comments = [] 1065 self._prev_token_line = self._line 1066 1067 return True 1068 1069 def _scan_number(self) -> None: 1070 if self._char == "0": 1071 peek = self._peek.upper() 1072 if peek == "B": 1073 return self._scan_bits() if self.BIT_STRINGS else self._add(TokenType.NUMBER) 1074 elif peek == "X": 1075 return self._scan_hex() if self.HEX_STRINGS else self._add(TokenType.NUMBER) 1076 1077 decimal = False 1078 scientific = 0 1079 1080 while True: 1081 if self._peek.isdigit(): 1082 self._advance() 1083 elif self._peek == "." and not decimal: 1084 after = self.peek(1) 1085 if after.isdigit() or not after.isalpha(): 1086 decimal = True 1087 self._advance() 1088 else: 1089 return self._add(TokenType.VAR) 1090 elif self._peek in ("-", "+") and scientific == 1: 1091 scientific += 1 1092 self._advance() 1093 elif self._peek.upper() == "E" and not scientific: 1094 scientific += 1 1095 self._advance() 1096 elif self._peek.isidentifier(): 1097 number_text = self._text 1098 literal = "" 1099 1100 while self._peek.strip() and self._peek not in self.SINGLE_TOKENS: 1101 literal += self._peek 1102 self._advance() 1103 1104 token_type = self.KEYWORDS.get(self.NUMERIC_LITERALS.get(literal.upper(), "")) 1105 1106 if token_type: 1107 self._add(TokenType.NUMBER, number_text) 1108 self._add(TokenType.DCOLON, "::") 1109 return self._add(token_type, literal) 1110 elif self.IDENTIFIERS_CAN_START_WITH_DIGIT: 1111 return self._add(TokenType.VAR) 1112 1113 self._advance(-len(literal)) 1114 return self._add(TokenType.NUMBER, number_text) 1115 else: 1116 return self._add(TokenType.NUMBER) 1117 1118 def _scan_bits(self) -> None: 1119 self._advance() 1120 value = self._extract_value() 1121 try: 1122 # If `value` can't be converted to a binary, fallback to tokenizing it as an identifier 1123 int(value, 2) 1124 self._add(TokenType.BIT_STRING, value[2:]) # Drop the 0b 1125 except ValueError: 1126 self._add(TokenType.IDENTIFIER) 1127 1128 def _scan_hex(self) -> None: 1129 self._advance() 1130 value = self._extract_value() 1131 try: 1132 # If `value` can't be converted to a hex, fallback to tokenizing it as an identifier 1133 int(value, 16) 1134 self._add(TokenType.HEX_STRING, value[2:]) # Drop the 0x 1135 except ValueError: 1136 self._add(TokenType.IDENTIFIER) 1137 1138 def _extract_value(self) -> str: 1139 while True: 1140 char = self._peek.strip() 1141 if char and char not in self.SINGLE_TOKENS: 1142 self._advance(alnum=True) 1143 else: 1144 break 1145 1146 return self._text 1147 1148 def _scan_string(self, start: str) -> bool: 1149 base = None 1150 token_type = TokenType.STRING 1151 1152 if start in self._QUOTES: 1153 end = self._QUOTES[start] 1154 elif start in self._FORMAT_STRINGS: 1155 end, token_type = self._FORMAT_STRINGS[start] 1156 1157 if token_type == TokenType.HEX_STRING: 1158 base = 16 1159 elif token_type == TokenType.BIT_STRING: 1160 base = 2 1161 elif token_type == TokenType.HEREDOC_STRING: 1162 self._advance() 1163 tag = "" if self._char == end else self._extract_string(end) 1164 end = f"{start}{tag}{end}" 1165 else: 1166 return False 1167 1168 self._advance(len(start)) 1169 text = self._extract_string(end) 1170 1171 if base: 1172 try: 1173 int(text, base) 1174 except: 1175 raise TokenError( 1176 f"Numeric string contains invalid characters from {self._line}:{self._start}" 1177 ) 1178 else: 1179 text = text.encode(self.ENCODE).decode(self.ENCODE) if self.ENCODE else text 1180 1181 self._add(token_type, text) 1182 return True 1183 1184 def _scan_identifier(self, identifier_end: str) -> None: 1185 self._advance() 1186 text = self._extract_string(identifier_end, self._IDENTIFIER_ESCAPES) 1187 self._add(TokenType.IDENTIFIER, text) 1188 1189 def _scan_var(self) -> None: 1190 while True: 1191 char = self._peek.strip() 1192 if char and (char in self.VAR_SINGLE_TOKENS or char not in self.SINGLE_TOKENS): 1193 self._advance(alnum=True) 1194 else: 1195 break 1196 1197 self._add( 1198 TokenType.VAR 1199 if self.tokens and self.tokens[-1].token_type == TokenType.PARAMETER 1200 else self.KEYWORDS.get(self._text.upper(), TokenType.VAR) 1201 ) 1202 1203 def _extract_string(self, delimiter: str, escapes=None) -> str: 1204 text = "" 1205 delim_size = len(delimiter) 1206 escapes = self._STRING_ESCAPES if escapes is None else escapes 1207 1208 while True: 1209 if ( 1210 self._char in escapes 1211 and (self._peek == delimiter or self._peek in escapes) 1212 and (self._char not in self._QUOTES or self._char == self._peek) 1213 ): 1214 if self._peek == delimiter: 1215 text += self._peek 1216 else: 1217 text += self._char + self._peek 1218 1219 if self._current + 1 < self.size: 1220 self._advance(2) 1221 else: 1222 raise TokenError(f"Missing {delimiter} from {self._line}:{self._current}") 1223 else: 1224 if self._chars(delim_size) == delimiter: 1225 if delim_size > 1: 1226 self._advance(delim_size - 1) 1227 break 1228 1229 if self._end: 1230 raise TokenError(f"Missing {delimiter} from {self._line}:{self._start}") 1231 1232 if self.ESCAPE_SEQUENCES and self._peek and self._char in self.STRING_ESCAPES: 1233 escaped_sequence = self.ESCAPE_SEQUENCES.get(self._char + self._peek) 1234 if escaped_sequence: 1235 self._advance(2) 1236 text += escaped_sequence 1237 continue 1238 1239 current = self._current - 1 1240 self._advance(alnum=True) 1241 text += self.sql[current : self._current - 1] 1242 1243 return text
12class TokenType(AutoName): 13 L_PAREN = auto() 14 R_PAREN = auto() 15 L_BRACKET = auto() 16 R_BRACKET = auto() 17 L_BRACE = auto() 18 R_BRACE = auto() 19 COMMA = auto() 20 DOT = auto() 21 DASH = auto() 22 PLUS = auto() 23 COLON = auto() 24 DCOLON = auto() 25 DQMARK = auto() 26 SEMICOLON = auto() 27 STAR = auto() 28 BACKSLASH = auto() 29 SLASH = auto() 30 LT = auto() 31 LTE = auto() 32 GT = auto() 33 GTE = auto() 34 NOT = auto() 35 EQ = auto() 36 NEQ = auto() 37 NULLSAFE_EQ = auto() 38 COLON_EQ = auto() 39 AND = auto() 40 OR = auto() 41 AMP = auto() 42 DPIPE = auto() 43 PIPE = auto() 44 CARET = auto() 45 TILDA = auto() 46 ARROW = auto() 47 DARROW = auto() 48 FARROW = auto() 49 HASH = auto() 50 HASH_ARROW = auto() 51 DHASH_ARROW = auto() 52 LR_ARROW = auto() 53 DAT = auto() 54 LT_AT = auto() 55 AT_GT = auto() 56 DOLLAR = auto() 57 PARAMETER = auto() 58 SESSION_PARAMETER = auto() 59 DAMP = auto() 60 XOR = auto() 61 DSTAR = auto() 62 63 BLOCK_START = auto() 64 BLOCK_END = auto() 65 66 SPACE = auto() 67 BREAK = auto() 68 69 STRING = auto() 70 NUMBER = auto() 71 IDENTIFIER = auto() 72 DATABASE = auto() 73 COLUMN = auto() 74 COLUMN_DEF = auto() 75 SCHEMA = auto() 76 TABLE = auto() 77 VAR = auto() 78 BIT_STRING = auto() 79 HEX_STRING = auto() 80 BYTE_STRING = auto() 81 NATIONAL_STRING = auto() 82 RAW_STRING = auto() 83 HEREDOC_STRING = auto() 84 85 # types 86 BIT = auto() 87 BOOLEAN = auto() 88 TINYINT = auto() 89 UTINYINT = auto() 90 SMALLINT = auto() 91 USMALLINT = auto() 92 MEDIUMINT = auto() 93 UMEDIUMINT = auto() 94 INT = auto() 95 UINT = auto() 96 BIGINT = auto() 97 UBIGINT = auto() 98 INT128 = auto() 99 UINT128 = auto() 100 INT256 = auto() 101 UINT256 = auto() 102 FLOAT = auto() 103 DOUBLE = auto() 104 DECIMAL = auto() 105 UDECIMAL = auto() 106 BIGDECIMAL = auto() 107 CHAR = auto() 108 NCHAR = auto() 109 VARCHAR = auto() 110 NVARCHAR = auto() 111 TEXT = auto() 112 MEDIUMTEXT = auto() 113 LONGTEXT = auto() 114 MEDIUMBLOB = auto() 115 LONGBLOB = auto() 116 TINYBLOB = auto() 117 TINYTEXT = auto() 118 BINARY = auto() 119 VARBINARY = auto() 120 JSON = auto() 121 JSONB = auto() 122 TIME = auto() 123 TIMETZ = auto() 124 TIMESTAMP = auto() 125 TIMESTAMPTZ = auto() 126 TIMESTAMPLTZ = auto() 127 TIMESTAMP_S = auto() 128 TIMESTAMP_MS = auto() 129 TIMESTAMP_NS = auto() 130 DATETIME = auto() 131 DATETIME64 = auto() 132 DATE = auto() 133 INT4RANGE = auto() 134 INT4MULTIRANGE = auto() 135 INT8RANGE = auto() 136 INT8MULTIRANGE = auto() 137 NUMRANGE = auto() 138 NUMMULTIRANGE = auto() 139 TSRANGE = auto() 140 TSMULTIRANGE = auto() 141 TSTZRANGE = auto() 142 TSTZMULTIRANGE = auto() 143 DATERANGE = auto() 144 DATEMULTIRANGE = auto() 145 UUID = auto() 146 GEOGRAPHY = auto() 147 NULLABLE = auto() 148 GEOMETRY = auto() 149 HLLSKETCH = auto() 150 HSTORE = auto() 151 SUPER = auto() 152 SERIAL = auto() 153 SMALLSERIAL = auto() 154 BIGSERIAL = auto() 155 XML = auto() 156 YEAR = auto() 157 UNIQUEIDENTIFIER = auto() 158 USERDEFINED = auto() 159 MONEY = auto() 160 SMALLMONEY = auto() 161 ROWVERSION = auto() 162 IMAGE = auto() 163 VARIANT = auto() 164 OBJECT = auto() 165 INET = auto() 166 IPADDRESS = auto() 167 IPPREFIX = auto() 168 ENUM = auto() 169 ENUM8 = auto() 170 ENUM16 = auto() 171 FIXEDSTRING = auto() 172 LOWCARDINALITY = auto() 173 NESTED = auto() 174 UNKNOWN = auto() 175 176 # keywords 177 ALIAS = auto() 178 ALTER = auto() 179 ALWAYS = auto() 180 ALL = auto() 181 ANTI = auto() 182 ANY = auto() 183 APPLY = auto() 184 ARRAY = auto() 185 ASC = auto() 186 ASOF = auto() 187 AUTO_INCREMENT = auto() 188 BEGIN = auto() 189 BETWEEN = auto() 190 CACHE = auto() 191 CASE = auto() 192 CHARACTER_SET = auto() 193 CLUSTER_BY = auto() 194 COLLATE = auto() 195 COMMAND = auto() 196 COMMENT = auto() 197 COMMIT = auto() 198 CONNECT_BY = auto() 199 CONSTRAINT = auto() 200 CREATE = auto() 201 CROSS = auto() 202 CUBE = auto() 203 CURRENT_DATE = auto() 204 CURRENT_DATETIME = auto() 205 CURRENT_TIME = auto() 206 CURRENT_TIMESTAMP = auto() 207 CURRENT_USER = auto() 208 DEFAULT = auto() 209 DELETE = auto() 210 DESC = auto() 211 DESCRIBE = auto() 212 DICTIONARY = auto() 213 DISTINCT = auto() 214 DISTRIBUTE_BY = auto() 215 DIV = auto() 216 DROP = auto() 217 ELSE = auto() 218 END = auto() 219 ESCAPE = auto() 220 EXCEPT = auto() 221 EXECUTE = auto() 222 EXISTS = auto() 223 FALSE = auto() 224 FETCH = auto() 225 FILTER = auto() 226 FINAL = auto() 227 FIRST = auto() 228 FOR = auto() 229 FORCE = auto() 230 FOREIGN_KEY = auto() 231 FORMAT = auto() 232 FROM = auto() 233 FULL = auto() 234 FUNCTION = auto() 235 GLOB = auto() 236 GLOBAL = auto() 237 GROUP_BY = auto() 238 GROUPING_SETS = auto() 239 HAVING = auto() 240 HINT = auto() 241 IGNORE = auto() 242 ILIKE = auto() 243 ILIKE_ANY = auto() 244 IN = auto() 245 INDEX = auto() 246 INNER = auto() 247 INSERT = auto() 248 INTERSECT = auto() 249 INTERVAL = auto() 250 INTO = auto() 251 INTRODUCER = auto() 252 IRLIKE = auto() 253 IS = auto() 254 ISNULL = auto() 255 JOIN = auto() 256 JOIN_MARKER = auto() 257 KEEP = auto() 258 KILL = auto() 259 LANGUAGE = auto() 260 LATERAL = auto() 261 LEFT = auto() 262 LIKE = auto() 263 LIKE_ANY = auto() 264 LIMIT = auto() 265 LOAD = auto() 266 LOCK = auto() 267 MAP = auto() 268 MATCH_RECOGNIZE = auto() 269 MEMBER_OF = auto() 270 MERGE = auto() 271 MOD = auto() 272 MODEL = auto() 273 NATURAL = auto() 274 NEXT = auto() 275 NOTNULL = auto() 276 NULL = auto() 277 OBJECT_IDENTIFIER = auto() 278 OFFSET = auto() 279 ON = auto() 280 ORDER_BY = auto() 281 ORDERED = auto() 282 ORDINALITY = auto() 283 OUTER = auto() 284 OVER = auto() 285 OVERLAPS = auto() 286 OVERWRITE = auto() 287 PARTITION = auto() 288 PARTITION_BY = auto() 289 PERCENT = auto() 290 PIVOT = auto() 291 PLACEHOLDER = auto() 292 PRAGMA = auto() 293 PRIMARY_KEY = auto() 294 PROCEDURE = auto() 295 PROPERTIES = auto() 296 PSEUDO_TYPE = auto() 297 QUALIFY = auto() 298 QUOTE = auto() 299 RANGE = auto() 300 RECURSIVE = auto() 301 REFRESH = auto() 302 REPLACE = auto() 303 RETURNING = auto() 304 REFERENCES = auto() 305 RIGHT = auto() 306 RLIKE = auto() 307 ROLLBACK = auto() 308 ROLLUP = auto() 309 ROW = auto() 310 ROWS = auto() 311 SELECT = auto() 312 SEMI = auto() 313 SEPARATOR = auto() 314 SERDE_PROPERTIES = auto() 315 SET = auto() 316 SETTINGS = auto() 317 SHOW = auto() 318 SIMILAR_TO = auto() 319 SOME = auto() 320 SORT_BY = auto() 321 START_WITH = auto() 322 STRUCT = auto() 323 TABLE_SAMPLE = auto() 324 TEMPORARY = auto() 325 TOP = auto() 326 THEN = auto() 327 TRUE = auto() 328 UNCACHE = auto() 329 UNION = auto() 330 UNNEST = auto() 331 UNPIVOT = auto() 332 UPDATE = auto() 333 USE = auto() 334 USING = auto() 335 VALUES = auto() 336 VIEW = auto() 337 VOLATILE = auto() 338 WHEN = auto() 339 WHERE = auto() 340 WINDOW = auto() 341 WITH = auto() 342 UNIQUE = auto() 343 VERSION_SNAPSHOT = auto() 344 TIMESTAMP_SNAPSHOT = auto()
An enumeration.
L_PAREN =
<TokenType.L_PAREN: 'L_PAREN'>
R_PAREN =
<TokenType.R_PAREN: 'R_PAREN'>
L_BRACKET =
<TokenType.L_BRACKET: 'L_BRACKET'>
R_BRACKET =
<TokenType.R_BRACKET: 'R_BRACKET'>
L_BRACE =
<TokenType.L_BRACE: 'L_BRACE'>
R_BRACE =
<TokenType.R_BRACE: 'R_BRACE'>
COMMA =
<TokenType.COMMA: 'COMMA'>
DOT =
<TokenType.DOT: 'DOT'>
DASH =
<TokenType.DASH: 'DASH'>
PLUS =
<TokenType.PLUS: 'PLUS'>
COLON =
<TokenType.COLON: 'COLON'>
DCOLON =
<TokenType.DCOLON: 'DCOLON'>
DQMARK =
<TokenType.DQMARK: 'DQMARK'>
SEMICOLON =
<TokenType.SEMICOLON: 'SEMICOLON'>
STAR =
<TokenType.STAR: 'STAR'>
BACKSLASH =
<TokenType.BACKSLASH: 'BACKSLASH'>
SLASH =
<TokenType.SLASH: 'SLASH'>
LT =
<TokenType.LT: 'LT'>
LTE =
<TokenType.LTE: 'LTE'>
GT =
<TokenType.GT: 'GT'>
GTE =
<TokenType.GTE: 'GTE'>
NOT =
<TokenType.NOT: 'NOT'>
EQ =
<TokenType.EQ: 'EQ'>
NEQ =
<TokenType.NEQ: 'NEQ'>
NULLSAFE_EQ =
<TokenType.NULLSAFE_EQ: 'NULLSAFE_EQ'>
COLON_EQ =
<TokenType.COLON_EQ: 'COLON_EQ'>
AND =
<TokenType.AND: 'AND'>
OR =
<TokenType.OR: 'OR'>
AMP =
<TokenType.AMP: 'AMP'>
DPIPE =
<TokenType.DPIPE: 'DPIPE'>
PIPE =
<TokenType.PIPE: 'PIPE'>
CARET =
<TokenType.CARET: 'CARET'>
TILDA =
<TokenType.TILDA: 'TILDA'>
ARROW =
<TokenType.ARROW: 'ARROW'>
DARROW =
<TokenType.DARROW: 'DARROW'>
FARROW =
<TokenType.FARROW: 'FARROW'>
HASH =
<TokenType.HASH: 'HASH'>
HASH_ARROW =
<TokenType.HASH_ARROW: 'HASH_ARROW'>
DHASH_ARROW =
<TokenType.DHASH_ARROW: 'DHASH_ARROW'>
LR_ARROW =
<TokenType.LR_ARROW: 'LR_ARROW'>
DAT =
<TokenType.DAT: 'DAT'>
LT_AT =
<TokenType.LT_AT: 'LT_AT'>
AT_GT =
<TokenType.AT_GT: 'AT_GT'>
DOLLAR =
<TokenType.DOLLAR: 'DOLLAR'>
PARAMETER =
<TokenType.PARAMETER: 'PARAMETER'>
SESSION_PARAMETER =
<TokenType.SESSION_PARAMETER: 'SESSION_PARAMETER'>
DAMP =
<TokenType.DAMP: 'DAMP'>
XOR =
<TokenType.XOR: 'XOR'>
DSTAR =
<TokenType.DSTAR: 'DSTAR'>
BLOCK_START =
<TokenType.BLOCK_START: 'BLOCK_START'>
BLOCK_END =
<TokenType.BLOCK_END: 'BLOCK_END'>
SPACE =
<TokenType.SPACE: 'SPACE'>
BREAK =
<TokenType.BREAK: 'BREAK'>
STRING =
<TokenType.STRING: 'STRING'>
NUMBER =
<TokenType.NUMBER: 'NUMBER'>
IDENTIFIER =
<TokenType.IDENTIFIER: 'IDENTIFIER'>
DATABASE =
<TokenType.DATABASE: 'DATABASE'>
COLUMN =
<TokenType.COLUMN: 'COLUMN'>
COLUMN_DEF =
<TokenType.COLUMN_DEF: 'COLUMN_DEF'>
SCHEMA =
<TokenType.SCHEMA: 'SCHEMA'>
TABLE =
<TokenType.TABLE: 'TABLE'>
VAR =
<TokenType.VAR: 'VAR'>
BIT_STRING =
<TokenType.BIT_STRING: 'BIT_STRING'>
HEX_STRING =
<TokenType.HEX_STRING: 'HEX_STRING'>
BYTE_STRING =
<TokenType.BYTE_STRING: 'BYTE_STRING'>
NATIONAL_STRING =
<TokenType.NATIONAL_STRING: 'NATIONAL_STRING'>
RAW_STRING =
<TokenType.RAW_STRING: 'RAW_STRING'>
HEREDOC_STRING =
<TokenType.HEREDOC_STRING: 'HEREDOC_STRING'>
BIT =
<TokenType.BIT: 'BIT'>
BOOLEAN =
<TokenType.BOOLEAN: 'BOOLEAN'>
TINYINT =
<TokenType.TINYINT: 'TINYINT'>
UTINYINT =
<TokenType.UTINYINT: 'UTINYINT'>
SMALLINT =
<TokenType.SMALLINT: 'SMALLINT'>
USMALLINT =
<TokenType.USMALLINT: 'USMALLINT'>
MEDIUMINT =
<TokenType.MEDIUMINT: 'MEDIUMINT'>
UMEDIUMINT =
<TokenType.UMEDIUMINT: 'UMEDIUMINT'>
INT =
<TokenType.INT: 'INT'>
UINT =
<TokenType.UINT: 'UINT'>
BIGINT =
<TokenType.BIGINT: 'BIGINT'>
UBIGINT =
<TokenType.UBIGINT: 'UBIGINT'>
INT128 =
<TokenType.INT128: 'INT128'>
UINT128 =
<TokenType.UINT128: 'UINT128'>
INT256 =
<TokenType.INT256: 'INT256'>
UINT256 =
<TokenType.UINT256: 'UINT256'>
FLOAT =
<TokenType.FLOAT: 'FLOAT'>
DOUBLE =
<TokenType.DOUBLE: 'DOUBLE'>
DECIMAL =
<TokenType.DECIMAL: 'DECIMAL'>
UDECIMAL =
<TokenType.UDECIMAL: 'UDECIMAL'>
BIGDECIMAL =
<TokenType.BIGDECIMAL: 'BIGDECIMAL'>
CHAR =
<TokenType.CHAR: 'CHAR'>
NCHAR =
<TokenType.NCHAR: 'NCHAR'>
VARCHAR =
<TokenType.VARCHAR: 'VARCHAR'>
NVARCHAR =
<TokenType.NVARCHAR: 'NVARCHAR'>
TEXT =
<TokenType.TEXT: 'TEXT'>
MEDIUMTEXT =
<TokenType.MEDIUMTEXT: 'MEDIUMTEXT'>
LONGTEXT =
<TokenType.LONGTEXT: 'LONGTEXT'>
MEDIUMBLOB =
<TokenType.MEDIUMBLOB: 'MEDIUMBLOB'>
LONGBLOB =
<TokenType.LONGBLOB: 'LONGBLOB'>
TINYBLOB =
<TokenType.TINYBLOB: 'TINYBLOB'>
TINYTEXT =
<TokenType.TINYTEXT: 'TINYTEXT'>
BINARY =
<TokenType.BINARY: 'BINARY'>
VARBINARY =
<TokenType.VARBINARY: 'VARBINARY'>
JSON =
<TokenType.JSON: 'JSON'>
JSONB =
<TokenType.JSONB: 'JSONB'>
TIME =
<TokenType.TIME: 'TIME'>
TIMETZ =
<TokenType.TIMETZ: 'TIMETZ'>
TIMESTAMP =
<TokenType.TIMESTAMP: 'TIMESTAMP'>
TIMESTAMPTZ =
<TokenType.TIMESTAMPTZ: 'TIMESTAMPTZ'>
TIMESTAMPLTZ =
<TokenType.TIMESTAMPLTZ: 'TIMESTAMPLTZ'>
TIMESTAMP_S =
<TokenType.TIMESTAMP_S: 'TIMESTAMP_S'>
TIMESTAMP_MS =
<TokenType.TIMESTAMP_MS: 'TIMESTAMP_MS'>
TIMESTAMP_NS =
<TokenType.TIMESTAMP_NS: 'TIMESTAMP_NS'>
DATETIME =
<TokenType.DATETIME: 'DATETIME'>
DATETIME64 =
<TokenType.DATETIME64: 'DATETIME64'>
DATE =
<TokenType.DATE: 'DATE'>
INT4RANGE =
<TokenType.INT4RANGE: 'INT4RANGE'>
INT4MULTIRANGE =
<TokenType.INT4MULTIRANGE: 'INT4MULTIRANGE'>
INT8RANGE =
<TokenType.INT8RANGE: 'INT8RANGE'>
INT8MULTIRANGE =
<TokenType.INT8MULTIRANGE: 'INT8MULTIRANGE'>
NUMRANGE =
<TokenType.NUMRANGE: 'NUMRANGE'>
NUMMULTIRANGE =
<TokenType.NUMMULTIRANGE: 'NUMMULTIRANGE'>
TSRANGE =
<TokenType.TSRANGE: 'TSRANGE'>
TSMULTIRANGE =
<TokenType.TSMULTIRANGE: 'TSMULTIRANGE'>
TSTZRANGE =
<TokenType.TSTZRANGE: 'TSTZRANGE'>
TSTZMULTIRANGE =
<TokenType.TSTZMULTIRANGE: 'TSTZMULTIRANGE'>
DATERANGE =
<TokenType.DATERANGE: 'DATERANGE'>
DATEMULTIRANGE =
<TokenType.DATEMULTIRANGE: 'DATEMULTIRANGE'>
UUID =
<TokenType.UUID: 'UUID'>
GEOGRAPHY =
<TokenType.GEOGRAPHY: 'GEOGRAPHY'>
NULLABLE =
<TokenType.NULLABLE: 'NULLABLE'>
GEOMETRY =
<TokenType.GEOMETRY: 'GEOMETRY'>
HLLSKETCH =
<TokenType.HLLSKETCH: 'HLLSKETCH'>
HSTORE =
<TokenType.HSTORE: 'HSTORE'>
SUPER =
<TokenType.SUPER: 'SUPER'>
SERIAL =
<TokenType.SERIAL: 'SERIAL'>
SMALLSERIAL =
<TokenType.SMALLSERIAL: 'SMALLSERIAL'>
BIGSERIAL =
<TokenType.BIGSERIAL: 'BIGSERIAL'>
XML =
<TokenType.XML: 'XML'>
YEAR =
<TokenType.YEAR: 'YEAR'>
UNIQUEIDENTIFIER =
<TokenType.UNIQUEIDENTIFIER: 'UNIQUEIDENTIFIER'>
USERDEFINED =
<TokenType.USERDEFINED: 'USERDEFINED'>
MONEY =
<TokenType.MONEY: 'MONEY'>
SMALLMONEY =
<TokenType.SMALLMONEY: 'SMALLMONEY'>
ROWVERSION =
<TokenType.ROWVERSION: 'ROWVERSION'>
IMAGE =
<TokenType.IMAGE: 'IMAGE'>
VARIANT =
<TokenType.VARIANT: 'VARIANT'>
OBJECT =
<TokenType.OBJECT: 'OBJECT'>
INET =
<TokenType.INET: 'INET'>
IPADDRESS =
<TokenType.IPADDRESS: 'IPADDRESS'>
IPPREFIX =
<TokenType.IPPREFIX: 'IPPREFIX'>
ENUM =
<TokenType.ENUM: 'ENUM'>
ENUM8 =
<TokenType.ENUM8: 'ENUM8'>
ENUM16 =
<TokenType.ENUM16: 'ENUM16'>
FIXEDSTRING =
<TokenType.FIXEDSTRING: 'FIXEDSTRING'>
LOWCARDINALITY =
<TokenType.LOWCARDINALITY: 'LOWCARDINALITY'>
NESTED =
<TokenType.NESTED: 'NESTED'>
UNKNOWN =
<TokenType.UNKNOWN: 'UNKNOWN'>
ALIAS =
<TokenType.ALIAS: 'ALIAS'>
ALTER =
<TokenType.ALTER: 'ALTER'>
ALWAYS =
<TokenType.ALWAYS: 'ALWAYS'>
ALL =
<TokenType.ALL: 'ALL'>
ANTI =
<TokenType.ANTI: 'ANTI'>
ANY =
<TokenType.ANY: 'ANY'>
APPLY =
<TokenType.APPLY: 'APPLY'>
ARRAY =
<TokenType.ARRAY: 'ARRAY'>
ASC =
<TokenType.ASC: 'ASC'>
ASOF =
<TokenType.ASOF: 'ASOF'>
AUTO_INCREMENT =
<TokenType.AUTO_INCREMENT: 'AUTO_INCREMENT'>
BEGIN =
<TokenType.BEGIN: 'BEGIN'>
BETWEEN =
<TokenType.BETWEEN: 'BETWEEN'>
CACHE =
<TokenType.CACHE: 'CACHE'>
CASE =
<TokenType.CASE: 'CASE'>
CHARACTER_SET =
<TokenType.CHARACTER_SET: 'CHARACTER_SET'>
CLUSTER_BY =
<TokenType.CLUSTER_BY: 'CLUSTER_BY'>
COLLATE =
<TokenType.COLLATE: 'COLLATE'>
COMMAND =
<TokenType.COMMAND: 'COMMAND'>
COMMENT =
<TokenType.COMMENT: 'COMMENT'>
COMMIT =
<TokenType.COMMIT: 'COMMIT'>
CONNECT_BY =
<TokenType.CONNECT_BY: 'CONNECT_BY'>
CONSTRAINT =
<TokenType.CONSTRAINT: 'CONSTRAINT'>
CREATE =
<TokenType.CREATE: 'CREATE'>
CROSS =
<TokenType.CROSS: 'CROSS'>
CUBE =
<TokenType.CUBE: 'CUBE'>
CURRENT_DATE =
<TokenType.CURRENT_DATE: 'CURRENT_DATE'>
CURRENT_DATETIME =
<TokenType.CURRENT_DATETIME: 'CURRENT_DATETIME'>
CURRENT_TIME =
<TokenType.CURRENT_TIME: 'CURRENT_TIME'>
CURRENT_TIMESTAMP =
<TokenType.CURRENT_TIMESTAMP: 'CURRENT_TIMESTAMP'>
CURRENT_USER =
<TokenType.CURRENT_USER: 'CURRENT_USER'>
DEFAULT =
<TokenType.DEFAULT: 'DEFAULT'>
DELETE =
<TokenType.DELETE: 'DELETE'>
DESC =
<TokenType.DESC: 'DESC'>
DESCRIBE =
<TokenType.DESCRIBE: 'DESCRIBE'>
DICTIONARY =
<TokenType.DICTIONARY: 'DICTIONARY'>
DISTINCT =
<TokenType.DISTINCT: 'DISTINCT'>
DISTRIBUTE_BY =
<TokenType.DISTRIBUTE_BY: 'DISTRIBUTE_BY'>
DIV =
<TokenType.DIV: 'DIV'>
DROP =
<TokenType.DROP: 'DROP'>
ELSE =
<TokenType.ELSE: 'ELSE'>
END =
<TokenType.END: 'END'>
ESCAPE =
<TokenType.ESCAPE: 'ESCAPE'>
EXCEPT =
<TokenType.EXCEPT: 'EXCEPT'>
EXECUTE =
<TokenType.EXECUTE: 'EXECUTE'>
EXISTS =
<TokenType.EXISTS: 'EXISTS'>
FALSE =
<TokenType.FALSE: 'FALSE'>
FETCH =
<TokenType.FETCH: 'FETCH'>
FILTER =
<TokenType.FILTER: 'FILTER'>
FINAL =
<TokenType.FINAL: 'FINAL'>
FIRST =
<TokenType.FIRST: 'FIRST'>
FOR =
<TokenType.FOR: 'FOR'>
FORCE =
<TokenType.FORCE: 'FORCE'>
FOREIGN_KEY =
<TokenType.FOREIGN_KEY: 'FOREIGN_KEY'>
FORMAT =
<TokenType.FORMAT: 'FORMAT'>
FROM =
<TokenType.FROM: 'FROM'>
FULL =
<TokenType.FULL: 'FULL'>
FUNCTION =
<TokenType.FUNCTION: 'FUNCTION'>
GLOB =
<TokenType.GLOB: 'GLOB'>
GLOBAL =
<TokenType.GLOBAL: 'GLOBAL'>
GROUP_BY =
<TokenType.GROUP_BY: 'GROUP_BY'>
GROUPING_SETS =
<TokenType.GROUPING_SETS: 'GROUPING_SETS'>
HAVING =
<TokenType.HAVING: 'HAVING'>
HINT =
<TokenType.HINT: 'HINT'>
IGNORE =
<TokenType.IGNORE: 'IGNORE'>
ILIKE =
<TokenType.ILIKE: 'ILIKE'>
ILIKE_ANY =
<TokenType.ILIKE_ANY: 'ILIKE_ANY'>
IN =
<TokenType.IN: 'IN'>
INDEX =
<TokenType.INDEX: 'INDEX'>
INNER =
<TokenType.INNER: 'INNER'>
INSERT =
<TokenType.INSERT: 'INSERT'>
INTERSECT =
<TokenType.INTERSECT: 'INTERSECT'>
INTERVAL =
<TokenType.INTERVAL: 'INTERVAL'>
INTO =
<TokenType.INTO: 'INTO'>
INTRODUCER =
<TokenType.INTRODUCER: 'INTRODUCER'>
IRLIKE =
<TokenType.IRLIKE: 'IRLIKE'>
IS =
<TokenType.IS: 'IS'>
ISNULL =
<TokenType.ISNULL: 'ISNULL'>
JOIN =
<TokenType.JOIN: 'JOIN'>
JOIN_MARKER =
<TokenType.JOIN_MARKER: 'JOIN_MARKER'>
KEEP =
<TokenType.KEEP: 'KEEP'>
KILL =
<TokenType.KILL: 'KILL'>
LANGUAGE =
<TokenType.LANGUAGE: 'LANGUAGE'>
LATERAL =
<TokenType.LATERAL: 'LATERAL'>
LEFT =
<TokenType.LEFT: 'LEFT'>
LIKE =
<TokenType.LIKE: 'LIKE'>
LIKE_ANY =
<TokenType.LIKE_ANY: 'LIKE_ANY'>
LIMIT =
<TokenType.LIMIT: 'LIMIT'>
LOAD =
<TokenType.LOAD: 'LOAD'>
LOCK =
<TokenType.LOCK: 'LOCK'>
MAP =
<TokenType.MAP: 'MAP'>
MATCH_RECOGNIZE =
<TokenType.MATCH_RECOGNIZE: 'MATCH_RECOGNIZE'>
MEMBER_OF =
<TokenType.MEMBER_OF: 'MEMBER_OF'>
MERGE =
<TokenType.MERGE: 'MERGE'>
MOD =
<TokenType.MOD: 'MOD'>
MODEL =
<TokenType.MODEL: 'MODEL'>
NATURAL =
<TokenType.NATURAL: 'NATURAL'>
NEXT =
<TokenType.NEXT: 'NEXT'>
NOTNULL =
<TokenType.NOTNULL: 'NOTNULL'>
NULL =
<TokenType.NULL: 'NULL'>
OBJECT_IDENTIFIER =
<TokenType.OBJECT_IDENTIFIER: 'OBJECT_IDENTIFIER'>
OFFSET =
<TokenType.OFFSET: 'OFFSET'>
ON =
<TokenType.ON: 'ON'>
ORDER_BY =
<TokenType.ORDER_BY: 'ORDER_BY'>
ORDERED =
<TokenType.ORDERED: 'ORDERED'>
ORDINALITY =
<TokenType.ORDINALITY: 'ORDINALITY'>
OUTER =
<TokenType.OUTER: 'OUTER'>
OVER =
<TokenType.OVER: 'OVER'>
OVERLAPS =
<TokenType.OVERLAPS: 'OVERLAPS'>
OVERWRITE =
<TokenType.OVERWRITE: 'OVERWRITE'>
PARTITION =
<TokenType.PARTITION: 'PARTITION'>
PARTITION_BY =
<TokenType.PARTITION_BY: 'PARTITION_BY'>
PERCENT =
<TokenType.PERCENT: 'PERCENT'>
PIVOT =
<TokenType.PIVOT: 'PIVOT'>
PLACEHOLDER =
<TokenType.PLACEHOLDER: 'PLACEHOLDER'>
PRAGMA =
<TokenType.PRAGMA: 'PRAGMA'>
PRIMARY_KEY =
<TokenType.PRIMARY_KEY: 'PRIMARY_KEY'>
PROCEDURE =
<TokenType.PROCEDURE: 'PROCEDURE'>
PROPERTIES =
<TokenType.PROPERTIES: 'PROPERTIES'>
PSEUDO_TYPE =
<TokenType.PSEUDO_TYPE: 'PSEUDO_TYPE'>
QUALIFY =
<TokenType.QUALIFY: 'QUALIFY'>
QUOTE =
<TokenType.QUOTE: 'QUOTE'>
RANGE =
<TokenType.RANGE: 'RANGE'>
RECURSIVE =
<TokenType.RECURSIVE: 'RECURSIVE'>
REFRESH =
<TokenType.REFRESH: 'REFRESH'>
REPLACE =
<TokenType.REPLACE: 'REPLACE'>
RETURNING =
<TokenType.RETURNING: 'RETURNING'>
REFERENCES =
<TokenType.REFERENCES: 'REFERENCES'>
RIGHT =
<TokenType.RIGHT: 'RIGHT'>
RLIKE =
<TokenType.RLIKE: 'RLIKE'>
ROLLBACK =
<TokenType.ROLLBACK: 'ROLLBACK'>
ROLLUP =
<TokenType.ROLLUP: 'ROLLUP'>
ROW =
<TokenType.ROW: 'ROW'>
ROWS =
<TokenType.ROWS: 'ROWS'>
SELECT =
<TokenType.SELECT: 'SELECT'>
SEMI =
<TokenType.SEMI: 'SEMI'>
SEPARATOR =
<TokenType.SEPARATOR: 'SEPARATOR'>
SERDE_PROPERTIES =
<TokenType.SERDE_PROPERTIES: 'SERDE_PROPERTIES'>
SET =
<TokenType.SET: 'SET'>
SETTINGS =
<TokenType.SETTINGS: 'SETTINGS'>
SHOW =
<TokenType.SHOW: 'SHOW'>
SIMILAR_TO =
<TokenType.SIMILAR_TO: 'SIMILAR_TO'>
SOME =
<TokenType.SOME: 'SOME'>
SORT_BY =
<TokenType.SORT_BY: 'SORT_BY'>
START_WITH =
<TokenType.START_WITH: 'START_WITH'>
STRUCT =
<TokenType.STRUCT: 'STRUCT'>
TABLE_SAMPLE =
<TokenType.TABLE_SAMPLE: 'TABLE_SAMPLE'>
TEMPORARY =
<TokenType.TEMPORARY: 'TEMPORARY'>
TOP =
<TokenType.TOP: 'TOP'>
THEN =
<TokenType.THEN: 'THEN'>
TRUE =
<TokenType.TRUE: 'TRUE'>
UNCACHE =
<TokenType.UNCACHE: 'UNCACHE'>
UNION =
<TokenType.UNION: 'UNION'>
UNNEST =
<TokenType.UNNEST: 'UNNEST'>
UNPIVOT =
<TokenType.UNPIVOT: 'UNPIVOT'>
UPDATE =
<TokenType.UPDATE: 'UPDATE'>
USE =
<TokenType.USE: 'USE'>
USING =
<TokenType.USING: 'USING'>
VALUES =
<TokenType.VALUES: 'VALUES'>
VIEW =
<TokenType.VIEW: 'VIEW'>
VOLATILE =
<TokenType.VOLATILE: 'VOLATILE'>
WHEN =
<TokenType.WHEN: 'WHEN'>
WHERE =
<TokenType.WHERE: 'WHERE'>
WINDOW =
<TokenType.WINDOW: 'WINDOW'>
WITH =
<TokenType.WITH: 'WITH'>
UNIQUE =
<TokenType.UNIQUE: 'UNIQUE'>
VERSION_SNAPSHOT =
<TokenType.VERSION_SNAPSHOT: 'VERSION_SNAPSHOT'>
TIMESTAMP_SNAPSHOT =
<TokenType.TIMESTAMP_SNAPSHOT: 'TIMESTAMP_SNAPSHOT'>
Inherited Members
- enum.Enum
- name
- value
class
Token:
347class Token: 348 __slots__ = ("token_type", "text", "line", "col", "start", "end", "comments") 349 350 @classmethod 351 def number(cls, number: int) -> Token: 352 """Returns a NUMBER token with `number` as its text.""" 353 return cls(TokenType.NUMBER, str(number)) 354 355 @classmethod 356 def string(cls, string: str) -> Token: 357 """Returns a STRING token with `string` as its text.""" 358 return cls(TokenType.STRING, string) 359 360 @classmethod 361 def identifier(cls, identifier: str) -> Token: 362 """Returns an IDENTIFIER token with `identifier` as its text.""" 363 return cls(TokenType.IDENTIFIER, identifier) 364 365 @classmethod 366 def var(cls, var: str) -> Token: 367 """Returns an VAR token with `var` as its text.""" 368 return cls(TokenType.VAR, var) 369 370 def __init__( 371 self, 372 token_type: TokenType, 373 text: str, 374 line: int = 1, 375 col: int = 1, 376 start: int = 0, 377 end: int = 0, 378 comments: t.Optional[t.List[str]] = None, 379 ) -> None: 380 """Token initializer. 381 382 Args: 383 token_type: The TokenType Enum. 384 text: The text of the token. 385 line: The line that the token ends on. 386 col: The column that the token ends on. 387 start: The start index of the token. 388 end: The ending index of the token. 389 comments: The comments to attach to the token. 390 """ 391 self.token_type = token_type 392 self.text = text 393 self.line = line 394 self.col = col 395 self.start = start 396 self.end = end 397 self.comments = [] if comments is None else comments 398 399 def __repr__(self) -> str: 400 attributes = ", ".join(f"{k}: {getattr(self, k)}" for k in self.__slots__) 401 return f"<Token {attributes}>"
Token( token_type: TokenType, text: str, line: int = 1, col: int = 1, start: int = 0, end: int = 0, comments: Optional[List[str]] = None)
370 def __init__( 371 self, 372 token_type: TokenType, 373 text: str, 374 line: int = 1, 375 col: int = 1, 376 start: int = 0, 377 end: int = 0, 378 comments: t.Optional[t.List[str]] = None, 379 ) -> None: 380 """Token initializer. 381 382 Args: 383 token_type: The TokenType Enum. 384 text: The text of the token. 385 line: The line that the token ends on. 386 col: The column that the token ends on. 387 start: The start index of the token. 388 end: The ending index of the token. 389 comments: The comments to attach to the token. 390 """ 391 self.token_type = token_type 392 self.text = text 393 self.line = line 394 self.col = col 395 self.start = start 396 self.end = end 397 self.comments = [] if comments is None else comments
Token initializer.
Arguments:
- token_type: The TokenType Enum.
- text: The text of the token.
- line: The line that the token ends on.
- col: The column that the token ends on.
- start: The start index of the token.
- end: The ending index of the token.
- comments: The comments to attach to the token.
350 @classmethod 351 def number(cls, number: int) -> Token: 352 """Returns a NUMBER token with `number` as its text.""" 353 return cls(TokenType.NUMBER, str(number))
Returns a NUMBER token with number
as its text.
355 @classmethod 356 def string(cls, string: str) -> Token: 357 """Returns a STRING token with `string` as its text.""" 358 return cls(TokenType.STRING, string)
Returns a STRING token with string
as its text.
360 @classmethod 361 def identifier(cls, identifier: str) -> Token: 362 """Returns an IDENTIFIER token with `identifier` as its text.""" 363 return cls(TokenType.IDENTIFIER, identifier)
Returns an IDENTIFIER token with identifier
as its text.
class
Tokenizer:
458class Tokenizer(metaclass=_Tokenizer): 459 SINGLE_TOKENS = { 460 "(": TokenType.L_PAREN, 461 ")": TokenType.R_PAREN, 462 "[": TokenType.L_BRACKET, 463 "]": TokenType.R_BRACKET, 464 "{": TokenType.L_BRACE, 465 "}": TokenType.R_BRACE, 466 "&": TokenType.AMP, 467 "^": TokenType.CARET, 468 ":": TokenType.COLON, 469 ",": TokenType.COMMA, 470 ".": TokenType.DOT, 471 "-": TokenType.DASH, 472 "=": TokenType.EQ, 473 ">": TokenType.GT, 474 "<": TokenType.LT, 475 "%": TokenType.MOD, 476 "!": TokenType.NOT, 477 "|": TokenType.PIPE, 478 "+": TokenType.PLUS, 479 ";": TokenType.SEMICOLON, 480 "/": TokenType.SLASH, 481 "\\": TokenType.BACKSLASH, 482 "*": TokenType.STAR, 483 "~": TokenType.TILDA, 484 "?": TokenType.PLACEHOLDER, 485 "@": TokenType.PARAMETER, 486 # used for breaking a var like x'y' but nothing else 487 # the token type doesn't matter 488 "'": TokenType.QUOTE, 489 "`": TokenType.IDENTIFIER, 490 '"': TokenType.IDENTIFIER, 491 "#": TokenType.HASH, 492 } 493 494 BIT_STRINGS: t.List[str | t.Tuple[str, str]] = [] 495 BYTE_STRINGS: t.List[str | t.Tuple[str, str]] = [] 496 HEX_STRINGS: t.List[str | t.Tuple[str, str]] = [] 497 RAW_STRINGS: t.List[str | t.Tuple[str, str]] = [] 498 HEREDOC_STRINGS: t.List[str | t.Tuple[str, str]] = [] 499 IDENTIFIERS: t.List[str | t.Tuple[str, str]] = ['"'] 500 IDENTIFIER_ESCAPES = ['"'] 501 QUOTES: t.List[t.Tuple[str, str] | str] = ["'"] 502 STRING_ESCAPES = ["'"] 503 VAR_SINGLE_TOKENS: t.Set[str] = set() 504 ESCAPE_SEQUENCES: t.Dict[str, str] = {} 505 506 # Autofilled 507 IDENTIFIERS_CAN_START_WITH_DIGIT: bool = False 508 509 _COMMENTS: t.Dict[str, str] = {} 510 _FORMAT_STRINGS: t.Dict[str, t.Tuple[str, TokenType]] = {} 511 _IDENTIFIERS: t.Dict[str, str] = {} 512 _IDENTIFIER_ESCAPES: t.Set[str] = set() 513 _QUOTES: t.Dict[str, str] = {} 514 _STRING_ESCAPES: t.Set[str] = set() 515 _KEYWORD_TRIE: t.Dict = {} 516 517 KEYWORDS: t.Dict[str, TokenType] = { 518 **{f"{{%{postfix}": TokenType.BLOCK_START for postfix in ("", "+", "-")}, 519 **{f"{prefix}%}}": TokenType.BLOCK_END for prefix in ("", "+", "-")}, 520 **{f"{{{{{postfix}": TokenType.BLOCK_START for postfix in ("+", "-")}, 521 **{f"{prefix}}}}}": TokenType.BLOCK_END for prefix in ("+", "-")}, 522 "/*+": TokenType.HINT, 523 "==": TokenType.EQ, 524 "::": TokenType.DCOLON, 525 "||": TokenType.DPIPE, 526 ">=": TokenType.GTE, 527 "<=": TokenType.LTE, 528 "<>": TokenType.NEQ, 529 "!=": TokenType.NEQ, 530 "<=>": TokenType.NULLSAFE_EQ, 531 "->": TokenType.ARROW, 532 "->>": TokenType.DARROW, 533 "=>": TokenType.FARROW, 534 "#>": TokenType.HASH_ARROW, 535 "#>>": TokenType.DHASH_ARROW, 536 "<->": TokenType.LR_ARROW, 537 "&&": TokenType.DAMP, 538 "??": TokenType.DQMARK, 539 "ALL": TokenType.ALL, 540 "ALWAYS": TokenType.ALWAYS, 541 "AND": TokenType.AND, 542 "ANTI": TokenType.ANTI, 543 "ANY": TokenType.ANY, 544 "ASC": TokenType.ASC, 545 "AS": TokenType.ALIAS, 546 "ASOF": TokenType.ASOF, 547 "AUTOINCREMENT": TokenType.AUTO_INCREMENT, 548 "AUTO_INCREMENT": TokenType.AUTO_INCREMENT, 549 "BEGIN": TokenType.BEGIN, 550 "BETWEEN": TokenType.BETWEEN, 551 "CACHE": TokenType.CACHE, 552 "UNCACHE": TokenType.UNCACHE, 553 "CASE": TokenType.CASE, 554 "CHARACTER SET": TokenType.CHARACTER_SET, 555 "CLUSTER BY": TokenType.CLUSTER_BY, 556 "COLLATE": TokenType.COLLATE, 557 "COLUMN": TokenType.COLUMN, 558 "COMMIT": TokenType.COMMIT, 559 "CONNECT BY": TokenType.CONNECT_BY, 560 "CONSTRAINT": TokenType.CONSTRAINT, 561 "CREATE": TokenType.CREATE, 562 "CROSS": TokenType.CROSS, 563 "CUBE": TokenType.CUBE, 564 "CURRENT_DATE": TokenType.CURRENT_DATE, 565 "CURRENT_TIME": TokenType.CURRENT_TIME, 566 "CURRENT_TIMESTAMP": TokenType.CURRENT_TIMESTAMP, 567 "CURRENT_USER": TokenType.CURRENT_USER, 568 "DATABASE": TokenType.DATABASE, 569 "DEFAULT": TokenType.DEFAULT, 570 "DELETE": TokenType.DELETE, 571 "DESC": TokenType.DESC, 572 "DESCRIBE": TokenType.DESCRIBE, 573 "DISTINCT": TokenType.DISTINCT, 574 "DISTRIBUTE BY": TokenType.DISTRIBUTE_BY, 575 "DIV": TokenType.DIV, 576 "DROP": TokenType.DROP, 577 "ELSE": TokenType.ELSE, 578 "END": TokenType.END, 579 "ESCAPE": TokenType.ESCAPE, 580 "EXCEPT": TokenType.EXCEPT, 581 "EXECUTE": TokenType.EXECUTE, 582 "EXISTS": TokenType.EXISTS, 583 "FALSE": TokenType.FALSE, 584 "FETCH": TokenType.FETCH, 585 "FILTER": TokenType.FILTER, 586 "FIRST": TokenType.FIRST, 587 "FULL": TokenType.FULL, 588 "FUNCTION": TokenType.FUNCTION, 589 "FOR": TokenType.FOR, 590 "FOREIGN KEY": TokenType.FOREIGN_KEY, 591 "FORMAT": TokenType.FORMAT, 592 "FROM": TokenType.FROM, 593 "GEOGRAPHY": TokenType.GEOGRAPHY, 594 "GEOMETRY": TokenType.GEOMETRY, 595 "GLOB": TokenType.GLOB, 596 "GROUP BY": TokenType.GROUP_BY, 597 "GROUPING SETS": TokenType.GROUPING_SETS, 598 "HAVING": TokenType.HAVING, 599 "ILIKE": TokenType.ILIKE, 600 "IN": TokenType.IN, 601 "INDEX": TokenType.INDEX, 602 "INET": TokenType.INET, 603 "INNER": TokenType.INNER, 604 "INSERT": TokenType.INSERT, 605 "INTERVAL": TokenType.INTERVAL, 606 "INTERSECT": TokenType.INTERSECT, 607 "INTO": TokenType.INTO, 608 "IS": TokenType.IS, 609 "ISNULL": TokenType.ISNULL, 610 "JOIN": TokenType.JOIN, 611 "KEEP": TokenType.KEEP, 612 "KILL": TokenType.KILL, 613 "LATERAL": TokenType.LATERAL, 614 "LEFT": TokenType.LEFT, 615 "LIKE": TokenType.LIKE, 616 "LIMIT": TokenType.LIMIT, 617 "LOAD": TokenType.LOAD, 618 "LOCK": TokenType.LOCK, 619 "MERGE": TokenType.MERGE, 620 "NATURAL": TokenType.NATURAL, 621 "NEXT": TokenType.NEXT, 622 "NOT": TokenType.NOT, 623 "NOTNULL": TokenType.NOTNULL, 624 "NULL": TokenType.NULL, 625 "OBJECT": TokenType.OBJECT, 626 "OFFSET": TokenType.OFFSET, 627 "ON": TokenType.ON, 628 "OR": TokenType.OR, 629 "XOR": TokenType.XOR, 630 "ORDER BY": TokenType.ORDER_BY, 631 "ORDINALITY": TokenType.ORDINALITY, 632 "OUTER": TokenType.OUTER, 633 "OVER": TokenType.OVER, 634 "OVERLAPS": TokenType.OVERLAPS, 635 "OVERWRITE": TokenType.OVERWRITE, 636 "PARTITION": TokenType.PARTITION, 637 "PARTITION BY": TokenType.PARTITION_BY, 638 "PARTITIONED BY": TokenType.PARTITION_BY, 639 "PARTITIONED_BY": TokenType.PARTITION_BY, 640 "PERCENT": TokenType.PERCENT, 641 "PIVOT": TokenType.PIVOT, 642 "PRAGMA": TokenType.PRAGMA, 643 "PRIMARY KEY": TokenType.PRIMARY_KEY, 644 "PROCEDURE": TokenType.PROCEDURE, 645 "QUALIFY": TokenType.QUALIFY, 646 "RANGE": TokenType.RANGE, 647 "RECURSIVE": TokenType.RECURSIVE, 648 "REGEXP": TokenType.RLIKE, 649 "REPLACE": TokenType.REPLACE, 650 "RETURNING": TokenType.RETURNING, 651 "REFERENCES": TokenType.REFERENCES, 652 "RIGHT": TokenType.RIGHT, 653 "RLIKE": TokenType.RLIKE, 654 "ROLLBACK": TokenType.ROLLBACK, 655 "ROLLUP": TokenType.ROLLUP, 656 "ROW": TokenType.ROW, 657 "ROWS": TokenType.ROWS, 658 "SCHEMA": TokenType.SCHEMA, 659 "SELECT": TokenType.SELECT, 660 "SEMI": TokenType.SEMI, 661 "SET": TokenType.SET, 662 "SETTINGS": TokenType.SETTINGS, 663 "SHOW": TokenType.SHOW, 664 "SIMILAR TO": TokenType.SIMILAR_TO, 665 "SOME": TokenType.SOME, 666 "SORT BY": TokenType.SORT_BY, 667 "START WITH": TokenType.START_WITH, 668 "TABLE": TokenType.TABLE, 669 "TABLESAMPLE": TokenType.TABLE_SAMPLE, 670 "TEMP": TokenType.TEMPORARY, 671 "TEMPORARY": TokenType.TEMPORARY, 672 "THEN": TokenType.THEN, 673 "TRUE": TokenType.TRUE, 674 "UNION": TokenType.UNION, 675 "UNKNOWN": TokenType.UNKNOWN, 676 "UNNEST": TokenType.UNNEST, 677 "UNPIVOT": TokenType.UNPIVOT, 678 "UPDATE": TokenType.UPDATE, 679 "USE": TokenType.USE, 680 "USING": TokenType.USING, 681 "UUID": TokenType.UUID, 682 "VALUES": TokenType.VALUES, 683 "VIEW": TokenType.VIEW, 684 "VOLATILE": TokenType.VOLATILE, 685 "WHEN": TokenType.WHEN, 686 "WHERE": TokenType.WHERE, 687 "WINDOW": TokenType.WINDOW, 688 "WITH": TokenType.WITH, 689 "APPLY": TokenType.APPLY, 690 "ARRAY": TokenType.ARRAY, 691 "BIT": TokenType.BIT, 692 "BOOL": TokenType.BOOLEAN, 693 "BOOLEAN": TokenType.BOOLEAN, 694 "BYTE": TokenType.TINYINT, 695 "MEDIUMINT": TokenType.MEDIUMINT, 696 "INT1": TokenType.TINYINT, 697 "TINYINT": TokenType.TINYINT, 698 "INT16": TokenType.SMALLINT, 699 "SHORT": TokenType.SMALLINT, 700 "SMALLINT": TokenType.SMALLINT, 701 "INT128": TokenType.INT128, 702 "HUGEINT": TokenType.INT128, 703 "INT2": TokenType.SMALLINT, 704 "INTEGER": TokenType.INT, 705 "INT": TokenType.INT, 706 "INT4": TokenType.INT, 707 "INT32": TokenType.INT, 708 "INT64": TokenType.BIGINT, 709 "LONG": TokenType.BIGINT, 710 "BIGINT": TokenType.BIGINT, 711 "INT8": TokenType.TINYINT, 712 "DEC": TokenType.DECIMAL, 713 "DECIMAL": TokenType.DECIMAL, 714 "BIGDECIMAL": TokenType.BIGDECIMAL, 715 "BIGNUMERIC": TokenType.BIGDECIMAL, 716 "MAP": TokenType.MAP, 717 "NULLABLE": TokenType.NULLABLE, 718 "NUMBER": TokenType.DECIMAL, 719 "NUMERIC": TokenType.DECIMAL, 720 "FIXED": TokenType.DECIMAL, 721 "REAL": TokenType.FLOAT, 722 "FLOAT": TokenType.FLOAT, 723 "FLOAT4": TokenType.FLOAT, 724 "FLOAT8": TokenType.DOUBLE, 725 "DOUBLE": TokenType.DOUBLE, 726 "DOUBLE PRECISION": TokenType.DOUBLE, 727 "JSON": TokenType.JSON, 728 "CHAR": TokenType.CHAR, 729 "CHARACTER": TokenType.CHAR, 730 "NCHAR": TokenType.NCHAR, 731 "VARCHAR": TokenType.VARCHAR, 732 "VARCHAR2": TokenType.VARCHAR, 733 "NVARCHAR": TokenType.NVARCHAR, 734 "NVARCHAR2": TokenType.NVARCHAR, 735 "STR": TokenType.TEXT, 736 "STRING": TokenType.TEXT, 737 "TEXT": TokenType.TEXT, 738 "LONGTEXT": TokenType.LONGTEXT, 739 "MEDIUMTEXT": TokenType.MEDIUMTEXT, 740 "TINYTEXT": TokenType.TINYTEXT, 741 "CLOB": TokenType.TEXT, 742 "LONGVARCHAR": TokenType.TEXT, 743 "BINARY": TokenType.BINARY, 744 "BLOB": TokenType.VARBINARY, 745 "LONGBLOB": TokenType.LONGBLOB, 746 "MEDIUMBLOB": TokenType.MEDIUMBLOB, 747 "TINYBLOB": TokenType.TINYBLOB, 748 "BYTEA": TokenType.VARBINARY, 749 "VARBINARY": TokenType.VARBINARY, 750 "TIME": TokenType.TIME, 751 "TIMETZ": TokenType.TIMETZ, 752 "TIMESTAMP": TokenType.TIMESTAMP, 753 "TIMESTAMPTZ": TokenType.TIMESTAMPTZ, 754 "TIMESTAMPLTZ": TokenType.TIMESTAMPLTZ, 755 "DATE": TokenType.DATE, 756 "DATETIME": TokenType.DATETIME, 757 "INT4RANGE": TokenType.INT4RANGE, 758 "INT4MULTIRANGE": TokenType.INT4MULTIRANGE, 759 "INT8RANGE": TokenType.INT8RANGE, 760 "INT8MULTIRANGE": TokenType.INT8MULTIRANGE, 761 "NUMRANGE": TokenType.NUMRANGE, 762 "NUMMULTIRANGE": TokenType.NUMMULTIRANGE, 763 "TSRANGE": TokenType.TSRANGE, 764 "TSMULTIRANGE": TokenType.TSMULTIRANGE, 765 "TSTZRANGE": TokenType.TSTZRANGE, 766 "TSTZMULTIRANGE": TokenType.TSTZMULTIRANGE, 767 "DATERANGE": TokenType.DATERANGE, 768 "DATEMULTIRANGE": TokenType.DATEMULTIRANGE, 769 "UNIQUE": TokenType.UNIQUE, 770 "STRUCT": TokenType.STRUCT, 771 "VARIANT": TokenType.VARIANT, 772 "ALTER": TokenType.ALTER, 773 "ANALYZE": TokenType.COMMAND, 774 "CALL": TokenType.COMMAND, 775 "COMMENT": TokenType.COMMENT, 776 "COPY": TokenType.COMMAND, 777 "EXPLAIN": TokenType.COMMAND, 778 "GRANT": TokenType.COMMAND, 779 "OPTIMIZE": TokenType.COMMAND, 780 "PREPARE": TokenType.COMMAND, 781 "TRUNCATE": TokenType.COMMAND, 782 "VACUUM": TokenType.COMMAND, 783 "USER-DEFINED": TokenType.USERDEFINED, 784 "FOR VERSION": TokenType.VERSION_SNAPSHOT, 785 "FOR TIMESTAMP": TokenType.TIMESTAMP_SNAPSHOT, 786 } 787 788 WHITE_SPACE: t.Dict[t.Optional[str], TokenType] = { 789 " ": TokenType.SPACE, 790 "\t": TokenType.SPACE, 791 "\n": TokenType.BREAK, 792 "\r": TokenType.BREAK, 793 "\r\n": TokenType.BREAK, 794 } 795 796 COMMANDS = { 797 TokenType.COMMAND, 798 TokenType.EXECUTE, 799 TokenType.FETCH, 800 TokenType.SHOW, 801 } 802 803 COMMAND_PREFIX_TOKENS = {TokenType.SEMICOLON, TokenType.BEGIN} 804 805 # handle numeric literals like in hive (3L = BIGINT) 806 NUMERIC_LITERALS: t.Dict[str, str] = {} 807 ENCODE: t.Optional[str] = None 808 809 COMMENTS = ["--", ("/*", "*/")] 810 811 __slots__ = ( 812 "sql", 813 "size", 814 "tokens", 815 "_start", 816 "_current", 817 "_line", 818 "_col", 819 "_comments", 820 "_char", 821 "_end", 822 "_peek", 823 "_prev_token_line", 824 ) 825 826 def __init__(self) -> None: 827 self.reset() 828 829 def reset(self) -> None: 830 self.sql = "" 831 self.size = 0 832 self.tokens: t.List[Token] = [] 833 self._start = 0 834 self._current = 0 835 self._line = 1 836 self._col = 0 837 self._comments: t.List[str] = [] 838 839 self._char = "" 840 self._end = False 841 self._peek = "" 842 self._prev_token_line = -1 843 844 def tokenize(self, sql: str) -> t.List[Token]: 845 """Returns a list of tokens corresponding to the SQL string `sql`.""" 846 self.reset() 847 self.sql = sql 848 self.size = len(sql) 849 850 try: 851 self._scan() 852 except Exception as e: 853 start = max(self._current - 50, 0) 854 end = min(self._current + 50, self.size - 1) 855 context = self.sql[start:end] 856 raise TokenError(f"Error tokenizing '{context}'") from e 857 858 return self.tokens 859 860 def _scan(self, until: t.Optional[t.Callable] = None) -> None: 861 while self.size and not self._end: 862 current = self._current 863 864 # skip spaces inline rather than iteratively call advance() 865 # for performance reasons 866 while current < self.size: 867 char = self.sql[current] 868 869 if char.isspace() and (char == " " or char == "\t"): 870 current += 1 871 else: 872 break 873 874 n = current - self._current 875 self._start = current 876 self._advance(n if n > 1 else 1) 877 878 if self._char is None: 879 break 880 881 if not self._char.isspace(): 882 if self._char.isdigit(): 883 self._scan_number() 884 elif self._char in self._IDENTIFIERS: 885 self._scan_identifier(self._IDENTIFIERS[self._char]) 886 else: 887 self._scan_keywords() 888 889 if until and until(): 890 break 891 892 if self.tokens and self._comments: 893 self.tokens[-1].comments.extend(self._comments) 894 895 def _chars(self, size: int) -> str: 896 if size == 1: 897 return self._char 898 899 start = self._current - 1 900 end = start + size 901 902 return self.sql[start:end] if end <= self.size else "" 903 904 def _advance(self, i: int = 1, alnum: bool = False) -> None: 905 if self.WHITE_SPACE.get(self._char) is TokenType.BREAK: 906 self._col = 1 907 self._line += 1 908 else: 909 self._col += i 910 911 self._current += i 912 self._end = self._current >= self.size 913 self._char = self.sql[self._current - 1] 914 self._peek = "" if self._end else self.sql[self._current] 915 916 if alnum and self._char.isalnum(): 917 # Here we use local variables instead of attributes for better performance 918 _col = self._col 919 _current = self._current 920 _end = self._end 921 _peek = self._peek 922 923 while _peek.isalnum(): 924 _col += 1 925 _current += 1 926 _end = _current >= self.size 927 _peek = "" if _end else self.sql[_current] 928 929 self._col = _col 930 self._current = _current 931 self._end = _end 932 self._peek = _peek 933 self._char = self.sql[_current - 1] 934 935 @property 936 def _text(self) -> str: 937 return self.sql[self._start : self._current] 938 939 def peek(self, i: int = 0) -> str: 940 i = self._current + i 941 if i < self.size: 942 return self.sql[i] 943 return "" 944 945 def _add(self, token_type: TokenType, text: t.Optional[str] = None) -> None: 946 self._prev_token_line = self._line 947 948 if self._comments and token_type == TokenType.SEMICOLON and self.tokens: 949 self.tokens[-1].comments.extend(self._comments) 950 self._comments = [] 951 952 self.tokens.append( 953 Token( 954 token_type, 955 text=self._text if text is None else text, 956 line=self._line, 957 col=self._col, 958 start=self._start, 959 end=self._current - 1, 960 comments=self._comments, 961 ) 962 ) 963 self._comments = [] 964 965 # If we have either a semicolon or a begin token before the command's token, we'll parse 966 # whatever follows the command's token as a string 967 if ( 968 token_type in self.COMMANDS 969 and self._peek != ";" 970 and (len(self.tokens) == 1 or self.tokens[-2].token_type in self.COMMAND_PREFIX_TOKENS) 971 ): 972 start = self._current 973 tokens = len(self.tokens) 974 self._scan(lambda: self._peek == ";") 975 self.tokens = self.tokens[:tokens] 976 text = self.sql[start : self._current].strip() 977 if text: 978 self._add(TokenType.STRING, text) 979 980 def _scan_keywords(self) -> None: 981 size = 0 982 word = None 983 chars = self._text 984 char = chars 985 prev_space = False 986 skip = False 987 trie = self._KEYWORD_TRIE 988 single_token = char in self.SINGLE_TOKENS 989 990 while chars: 991 if skip: 992 result = TrieResult.PREFIX 993 else: 994 result, trie = in_trie(trie, char.upper()) 995 996 if result == TrieResult.FAILED: 997 break 998 if result == TrieResult.EXISTS: 999 word = chars 1000 1001 end = self._current + size 1002 size += 1 1003 1004 if end < self.size: 1005 char = self.sql[end] 1006 single_token = single_token or char in self.SINGLE_TOKENS 1007 is_space = char.isspace() 1008 1009 if not is_space or not prev_space: 1010 if is_space: 1011 char = " " 1012 chars += char 1013 prev_space = is_space 1014 skip = False 1015 else: 1016 skip = True 1017 else: 1018 char = "" 1019 chars = " " 1020 1021 if word: 1022 if self._scan_string(word): 1023 return 1024 if self._scan_comment(word): 1025 return 1026 if prev_space or single_token or not char: 1027 self._advance(size - 1) 1028 word = word.upper() 1029 self._add(self.KEYWORDS[word], text=word) 1030 return 1031 1032 if self._char in self.SINGLE_TOKENS: 1033 self._add(self.SINGLE_TOKENS[self._char], text=self._char) 1034 return 1035 1036 self._scan_var() 1037 1038 def _scan_comment(self, comment_start: str) -> bool: 1039 if comment_start not in self._COMMENTS: 1040 return False 1041 1042 comment_start_line = self._line 1043 comment_start_size = len(comment_start) 1044 comment_end = self._COMMENTS[comment_start] 1045 1046 if comment_end: 1047 # Skip the comment's start delimiter 1048 self._advance(comment_start_size) 1049 1050 comment_end_size = len(comment_end) 1051 while not self._end and self._chars(comment_end_size) != comment_end: 1052 self._advance(alnum=True) 1053 1054 self._comments.append(self._text[comment_start_size : -comment_end_size + 1]) 1055 self._advance(comment_end_size - 1) 1056 else: 1057 while not self._end and not self.WHITE_SPACE.get(self._peek) is TokenType.BREAK: 1058 self._advance(alnum=True) 1059 self._comments.append(self._text[comment_start_size:]) 1060 1061 # Leading comment is attached to the succeeding token, whilst trailing comment to the preceding. 1062 # Multiple consecutive comments are preserved by appending them to the current comments list. 1063 if comment_start_line == self._prev_token_line: 1064 self.tokens[-1].comments.extend(self._comments) 1065 self._comments = [] 1066 self._prev_token_line = self._line 1067 1068 return True 1069 1070 def _scan_number(self) -> None: 1071 if self._char == "0": 1072 peek = self._peek.upper() 1073 if peek == "B": 1074 return self._scan_bits() if self.BIT_STRINGS else self._add(TokenType.NUMBER) 1075 elif peek == "X": 1076 return self._scan_hex() if self.HEX_STRINGS else self._add(TokenType.NUMBER) 1077 1078 decimal = False 1079 scientific = 0 1080 1081 while True: 1082 if self._peek.isdigit(): 1083 self._advance() 1084 elif self._peek == "." and not decimal: 1085 after = self.peek(1) 1086 if after.isdigit() or not after.isalpha(): 1087 decimal = True 1088 self._advance() 1089 else: 1090 return self._add(TokenType.VAR) 1091 elif self._peek in ("-", "+") and scientific == 1: 1092 scientific += 1 1093 self._advance() 1094 elif self._peek.upper() == "E" and not scientific: 1095 scientific += 1 1096 self._advance() 1097 elif self._peek.isidentifier(): 1098 number_text = self._text 1099 literal = "" 1100 1101 while self._peek.strip() and self._peek not in self.SINGLE_TOKENS: 1102 literal += self._peek 1103 self._advance() 1104 1105 token_type = self.KEYWORDS.get(self.NUMERIC_LITERALS.get(literal.upper(), "")) 1106 1107 if token_type: 1108 self._add(TokenType.NUMBER, number_text) 1109 self._add(TokenType.DCOLON, "::") 1110 return self._add(token_type, literal) 1111 elif self.IDENTIFIERS_CAN_START_WITH_DIGIT: 1112 return self._add(TokenType.VAR) 1113 1114 self._advance(-len(literal)) 1115 return self._add(TokenType.NUMBER, number_text) 1116 else: 1117 return self._add(TokenType.NUMBER) 1118 1119 def _scan_bits(self) -> None: 1120 self._advance() 1121 value = self._extract_value() 1122 try: 1123 # If `value` can't be converted to a binary, fallback to tokenizing it as an identifier 1124 int(value, 2) 1125 self._add(TokenType.BIT_STRING, value[2:]) # Drop the 0b 1126 except ValueError: 1127 self._add(TokenType.IDENTIFIER) 1128 1129 def _scan_hex(self) -> None: 1130 self._advance() 1131 value = self._extract_value() 1132 try: 1133 # If `value` can't be converted to a hex, fallback to tokenizing it as an identifier 1134 int(value, 16) 1135 self._add(TokenType.HEX_STRING, value[2:]) # Drop the 0x 1136 except ValueError: 1137 self._add(TokenType.IDENTIFIER) 1138 1139 def _extract_value(self) -> str: 1140 while True: 1141 char = self._peek.strip() 1142 if char and char not in self.SINGLE_TOKENS: 1143 self._advance(alnum=True) 1144 else: 1145 break 1146 1147 return self._text 1148 1149 def _scan_string(self, start: str) -> bool: 1150 base = None 1151 token_type = TokenType.STRING 1152 1153 if start in self._QUOTES: 1154 end = self._QUOTES[start] 1155 elif start in self._FORMAT_STRINGS: 1156 end, token_type = self._FORMAT_STRINGS[start] 1157 1158 if token_type == TokenType.HEX_STRING: 1159 base = 16 1160 elif token_type == TokenType.BIT_STRING: 1161 base = 2 1162 elif token_type == TokenType.HEREDOC_STRING: 1163 self._advance() 1164 tag = "" if self._char == end else self._extract_string(end) 1165 end = f"{start}{tag}{end}" 1166 else: 1167 return False 1168 1169 self._advance(len(start)) 1170 text = self._extract_string(end) 1171 1172 if base: 1173 try: 1174 int(text, base) 1175 except: 1176 raise TokenError( 1177 f"Numeric string contains invalid characters from {self._line}:{self._start}" 1178 ) 1179 else: 1180 text = text.encode(self.ENCODE).decode(self.ENCODE) if self.ENCODE else text 1181 1182 self._add(token_type, text) 1183 return True 1184 1185 def _scan_identifier(self, identifier_end: str) -> None: 1186 self._advance() 1187 text = self._extract_string(identifier_end, self._IDENTIFIER_ESCAPES) 1188 self._add(TokenType.IDENTIFIER, text) 1189 1190 def _scan_var(self) -> None: 1191 while True: 1192 char = self._peek.strip() 1193 if char and (char in self.VAR_SINGLE_TOKENS or char not in self.SINGLE_TOKENS): 1194 self._advance(alnum=True) 1195 else: 1196 break 1197 1198 self._add( 1199 TokenType.VAR 1200 if self.tokens and self.tokens[-1].token_type == TokenType.PARAMETER 1201 else self.KEYWORDS.get(self._text.upper(), TokenType.VAR) 1202 ) 1203 1204 def _extract_string(self, delimiter: str, escapes=None) -> str: 1205 text = "" 1206 delim_size = len(delimiter) 1207 escapes = self._STRING_ESCAPES if escapes is None else escapes 1208 1209 while True: 1210 if ( 1211 self._char in escapes 1212 and (self._peek == delimiter or self._peek in escapes) 1213 and (self._char not in self._QUOTES or self._char == self._peek) 1214 ): 1215 if self._peek == delimiter: 1216 text += self._peek 1217 else: 1218 text += self._char + self._peek 1219 1220 if self._current + 1 < self.size: 1221 self._advance(2) 1222 else: 1223 raise TokenError(f"Missing {delimiter} from {self._line}:{self._current}") 1224 else: 1225 if self._chars(delim_size) == delimiter: 1226 if delim_size > 1: 1227 self._advance(delim_size - 1) 1228 break 1229 1230 if self._end: 1231 raise TokenError(f"Missing {delimiter} from {self._line}:{self._start}") 1232 1233 if self.ESCAPE_SEQUENCES and self._peek and self._char in self.STRING_ESCAPES: 1234 escaped_sequence = self.ESCAPE_SEQUENCES.get(self._char + self._peek) 1235 if escaped_sequence: 1236 self._advance(2) 1237 text += escaped_sequence 1238 continue 1239 1240 current = self._current - 1 1241 self._advance(alnum=True) 1242 text += self.sql[current : self._current - 1] 1243 1244 return text
SINGLE_TOKENS =
{'(': <TokenType.L_PAREN: 'L_PAREN'>, ')': <TokenType.R_PAREN: 'R_PAREN'>, '[': <TokenType.L_BRACKET: 'L_BRACKET'>, ']': <TokenType.R_BRACKET: 'R_BRACKET'>, '{': <TokenType.L_BRACE: 'L_BRACE'>, '}': <TokenType.R_BRACE: 'R_BRACE'>, '&': <TokenType.AMP: 'AMP'>, '^': <TokenType.CARET: 'CARET'>, ':': <TokenType.COLON: 'COLON'>, ',': <TokenType.COMMA: 'COMMA'>, '.': <TokenType.DOT: 'DOT'>, '-': <TokenType.DASH: 'DASH'>, '=': <TokenType.EQ: 'EQ'>, '>': <TokenType.GT: 'GT'>, '<': <TokenType.LT: 'LT'>, '%': <TokenType.MOD: 'MOD'>, '!': <TokenType.NOT: 'NOT'>, '|': <TokenType.PIPE: 'PIPE'>, '+': <TokenType.PLUS: 'PLUS'>, ';': <TokenType.SEMICOLON: 'SEMICOLON'>, '/': <TokenType.SLASH: 'SLASH'>, '\\': <TokenType.BACKSLASH: 'BACKSLASH'>, '*': <TokenType.STAR: 'STAR'>, '~': <TokenType.TILDA: 'TILDA'>, '?': <TokenType.PLACEHOLDER: 'PLACEHOLDER'>, '@': <TokenType.PARAMETER: 'PARAMETER'>, "'": <TokenType.QUOTE: 'QUOTE'>, '`': <TokenType.IDENTIFIER: 'IDENTIFIER'>, '"': <TokenType.IDENTIFIER: 'IDENTIFIER'>, '#': <TokenType.HASH: 'HASH'>}
KEYWORDS: Dict[str, TokenType] =
{'{%': <TokenType.BLOCK_START: 'BLOCK_START'>, '{%+': <TokenType.BLOCK_START: 'BLOCK_START'>, '{%-': <TokenType.BLOCK_START: 'BLOCK_START'>, '%}': <TokenType.BLOCK_END: 'BLOCK_END'>, '+%}': <TokenType.BLOCK_END: 'BLOCK_END'>, '-%}': <TokenType.BLOCK_END: 'BLOCK_END'>, '{{+': <TokenType.BLOCK_START: 'BLOCK_START'>, '{{-': <TokenType.BLOCK_START: 'BLOCK_START'>, '+}}': <TokenType.BLOCK_END: 'BLOCK_END'>, '-}}': <TokenType.BLOCK_END: 'BLOCK_END'>, '/*+': <TokenType.HINT: 'HINT'>, '==': <TokenType.EQ: 'EQ'>, '::': <TokenType.DCOLON: 'DCOLON'>, '||': <TokenType.DPIPE: 'DPIPE'>, '>=': <TokenType.GTE: 'GTE'>, '<=': <TokenType.LTE: 'LTE'>, '<>': <TokenType.NEQ: 'NEQ'>, '!=': <TokenType.NEQ: 'NEQ'>, '<=>': <TokenType.NULLSAFE_EQ: 'NULLSAFE_EQ'>, '->': <TokenType.ARROW: 'ARROW'>, '->>': <TokenType.DARROW: 'DARROW'>, '=>': <TokenType.FARROW: 'FARROW'>, '#>': <TokenType.HASH_ARROW: 'HASH_ARROW'>, '#>>': <TokenType.DHASH_ARROW: 'DHASH_ARROW'>, '<->': <TokenType.LR_ARROW: 'LR_ARROW'>, '&&': <TokenType.DAMP: 'DAMP'>, '??': <TokenType.DQMARK: 'DQMARK'>, 'ALL': <TokenType.ALL: 'ALL'>, 'ALWAYS': <TokenType.ALWAYS: 'ALWAYS'>, 'AND': <TokenType.AND: 'AND'>, 'ANTI': <TokenType.ANTI: 'ANTI'>, 'ANY': <TokenType.ANY: 'ANY'>, 'ASC': <TokenType.ASC: 'ASC'>, 'AS': <TokenType.ALIAS: 'ALIAS'>, 'ASOF': <TokenType.ASOF: 'ASOF'>, 'AUTOINCREMENT': <TokenType.AUTO_INCREMENT: 'AUTO_INCREMENT'>, 'AUTO_INCREMENT': <TokenType.AUTO_INCREMENT: 'AUTO_INCREMENT'>, 'BEGIN': <TokenType.BEGIN: 'BEGIN'>, 'BETWEEN': <TokenType.BETWEEN: 'BETWEEN'>, 'CACHE': <TokenType.CACHE: 'CACHE'>, 'UNCACHE': <TokenType.UNCACHE: 'UNCACHE'>, 'CASE': <TokenType.CASE: 'CASE'>, 'CHARACTER SET': <TokenType.CHARACTER_SET: 'CHARACTER_SET'>, 'CLUSTER BY': <TokenType.CLUSTER_BY: 'CLUSTER_BY'>, 'COLLATE': <TokenType.COLLATE: 'COLLATE'>, 'COLUMN': <TokenType.COLUMN: 'COLUMN'>, 'COMMIT': <TokenType.COMMIT: 'COMMIT'>, 'CONNECT BY': <TokenType.CONNECT_BY: 'CONNECT_BY'>, 'CONSTRAINT': <TokenType.CONSTRAINT: 'CONSTRAINT'>, 'CREATE': <TokenType.CREATE: 'CREATE'>, 'CROSS': <TokenType.CROSS: 'CROSS'>, 'CUBE': <TokenType.CUBE: 'CUBE'>, 'CURRENT_DATE': <TokenType.CURRENT_DATE: 'CURRENT_DATE'>, 'CURRENT_TIME': <TokenType.CURRENT_TIME: 'CURRENT_TIME'>, 'CURRENT_TIMESTAMP': <TokenType.CURRENT_TIMESTAMP: 'CURRENT_TIMESTAMP'>, 'CURRENT_USER': <TokenType.CURRENT_USER: 'CURRENT_USER'>, 'DATABASE': <TokenType.DATABASE: 'DATABASE'>, 'DEFAULT': <TokenType.DEFAULT: 'DEFAULT'>, 'DELETE': <TokenType.DELETE: 'DELETE'>, 'DESC': <TokenType.DESC: 'DESC'>, 'DESCRIBE': <TokenType.DESCRIBE: 'DESCRIBE'>, 'DISTINCT': <TokenType.DISTINCT: 'DISTINCT'>, 'DISTRIBUTE BY': <TokenType.DISTRIBUTE_BY: 'DISTRIBUTE_BY'>, 'DIV': <TokenType.DIV: 'DIV'>, 'DROP': <TokenType.DROP: 'DROP'>, 'ELSE': <TokenType.ELSE: 'ELSE'>, 'END': <TokenType.END: 'END'>, 'ESCAPE': <TokenType.ESCAPE: 'ESCAPE'>, 'EXCEPT': <TokenType.EXCEPT: 'EXCEPT'>, 'EXECUTE': <TokenType.EXECUTE: 'EXECUTE'>, 'EXISTS': <TokenType.EXISTS: 'EXISTS'>, 'FALSE': <TokenType.FALSE: 'FALSE'>, 'FETCH': <TokenType.FETCH: 'FETCH'>, 'FILTER': <TokenType.FILTER: 'FILTER'>, 'FIRST': <TokenType.FIRST: 'FIRST'>, 'FULL': <TokenType.FULL: 'FULL'>, 'FUNCTION': <TokenType.FUNCTION: 'FUNCTION'>, 'FOR': <TokenType.FOR: 'FOR'>, 'FOREIGN KEY': <TokenType.FOREIGN_KEY: 'FOREIGN_KEY'>, 'FORMAT': <TokenType.FORMAT: 'FORMAT'>, 'FROM': <TokenType.FROM: 'FROM'>, 'GEOGRAPHY': <TokenType.GEOGRAPHY: 'GEOGRAPHY'>, 'GEOMETRY': <TokenType.GEOMETRY: 'GEOMETRY'>, 'GLOB': <TokenType.GLOB: 'GLOB'>, 'GROUP BY': <TokenType.GROUP_BY: 'GROUP_BY'>, 'GROUPING SETS': <TokenType.GROUPING_SETS: 'GROUPING_SETS'>, 'HAVING': <TokenType.HAVING: 'HAVING'>, 'ILIKE': <TokenType.ILIKE: 'ILIKE'>, 'IN': <TokenType.IN: 'IN'>, 'INDEX': <TokenType.INDEX: 'INDEX'>, 'INET': <TokenType.INET: 'INET'>, 'INNER': <TokenType.INNER: 'INNER'>, 'INSERT': <TokenType.INSERT: 'INSERT'>, 'INTERVAL': <TokenType.INTERVAL: 'INTERVAL'>, 'INTERSECT': <TokenType.INTERSECT: 'INTERSECT'>, 'INTO': <TokenType.INTO: 'INTO'>, 'IS': <TokenType.IS: 'IS'>, 'ISNULL': <TokenType.ISNULL: 'ISNULL'>, 'JOIN': <TokenType.JOIN: 'JOIN'>, 'KEEP': <TokenType.KEEP: 'KEEP'>, 'KILL': <TokenType.KILL: 'KILL'>, 'LATERAL': <TokenType.LATERAL: 'LATERAL'>, 'LEFT': <TokenType.LEFT: 'LEFT'>, 'LIKE': <TokenType.LIKE: 'LIKE'>, 'LIMIT': <TokenType.LIMIT: 'LIMIT'>, 'LOAD': <TokenType.LOAD: 'LOAD'>, 'LOCK': <TokenType.LOCK: 'LOCK'>, 'MERGE': <TokenType.MERGE: 'MERGE'>, 'NATURAL': <TokenType.NATURAL: 'NATURAL'>, 'NEXT': <TokenType.NEXT: 'NEXT'>, 'NOT': <TokenType.NOT: 'NOT'>, 'NOTNULL': <TokenType.NOTNULL: 'NOTNULL'>, 'NULL': <TokenType.NULL: 'NULL'>, 'OBJECT': <TokenType.OBJECT: 'OBJECT'>, 'OFFSET': <TokenType.OFFSET: 'OFFSET'>, 'ON': <TokenType.ON: 'ON'>, 'OR': <TokenType.OR: 'OR'>, 'XOR': <TokenType.XOR: 'XOR'>, 'ORDER BY': <TokenType.ORDER_BY: 'ORDER_BY'>, 'ORDINALITY': <TokenType.ORDINALITY: 'ORDINALITY'>, 'OUTER': <TokenType.OUTER: 'OUTER'>, 'OVER': <TokenType.OVER: 'OVER'>, 'OVERLAPS': <TokenType.OVERLAPS: 'OVERLAPS'>, 'OVERWRITE': <TokenType.OVERWRITE: 'OVERWRITE'>, 'PARTITION': <TokenType.PARTITION: 'PARTITION'>, 'PARTITION BY': <TokenType.PARTITION_BY: 'PARTITION_BY'>, 'PARTITIONED BY': <TokenType.PARTITION_BY: 'PARTITION_BY'>, 'PARTITIONED_BY': <TokenType.PARTITION_BY: 'PARTITION_BY'>, 'PERCENT': <TokenType.PERCENT: 'PERCENT'>, 'PIVOT': <TokenType.PIVOT: 'PIVOT'>, 'PRAGMA': <TokenType.PRAGMA: 'PRAGMA'>, 'PRIMARY KEY': <TokenType.PRIMARY_KEY: 'PRIMARY_KEY'>, 'PROCEDURE': <TokenType.PROCEDURE: 'PROCEDURE'>, 'QUALIFY': <TokenType.QUALIFY: 'QUALIFY'>, 'RANGE': <TokenType.RANGE: 'RANGE'>, 'RECURSIVE': <TokenType.RECURSIVE: 'RECURSIVE'>, 'REGEXP': <TokenType.RLIKE: 'RLIKE'>, 'REPLACE': <TokenType.REPLACE: 'REPLACE'>, 'RETURNING': <TokenType.RETURNING: 'RETURNING'>, 'REFERENCES': <TokenType.REFERENCES: 'REFERENCES'>, 'RIGHT': <TokenType.RIGHT: 'RIGHT'>, 'RLIKE': <TokenType.RLIKE: 'RLIKE'>, 'ROLLBACK': <TokenType.ROLLBACK: 'ROLLBACK'>, 'ROLLUP': <TokenType.ROLLUP: 'ROLLUP'>, 'ROW': <TokenType.ROW: 'ROW'>, 'ROWS': <TokenType.ROWS: 'ROWS'>, 'SCHEMA': <TokenType.SCHEMA: 'SCHEMA'>, 'SELECT': <TokenType.SELECT: 'SELECT'>, 'SEMI': <TokenType.SEMI: 'SEMI'>, 'SET': <TokenType.SET: 'SET'>, 'SETTINGS': <TokenType.SETTINGS: 'SETTINGS'>, 'SHOW': <TokenType.SHOW: 'SHOW'>, 'SIMILAR TO': <TokenType.SIMILAR_TO: 'SIMILAR_TO'>, 'SOME': <TokenType.SOME: 'SOME'>, 'SORT BY': <TokenType.SORT_BY: 'SORT_BY'>, 'START WITH': <TokenType.START_WITH: 'START_WITH'>, 'TABLE': <TokenType.TABLE: 'TABLE'>, 'TABLESAMPLE': <TokenType.TABLE_SAMPLE: 'TABLE_SAMPLE'>, 'TEMP': <TokenType.TEMPORARY: 'TEMPORARY'>, 'TEMPORARY': <TokenType.TEMPORARY: 'TEMPORARY'>, 'THEN': <TokenType.THEN: 'THEN'>, 'TRUE': <TokenType.TRUE: 'TRUE'>, 'UNION': <TokenType.UNION: 'UNION'>, 'UNKNOWN': <TokenType.UNKNOWN: 'UNKNOWN'>, 'UNNEST': <TokenType.UNNEST: 'UNNEST'>, 'UNPIVOT': <TokenType.UNPIVOT: 'UNPIVOT'>, 'UPDATE': <TokenType.UPDATE: 'UPDATE'>, 'USE': <TokenType.USE: 'USE'>, 'USING': <TokenType.USING: 'USING'>, 'UUID': <TokenType.UUID: 'UUID'>, 'VALUES': <TokenType.VALUES: 'VALUES'>, 'VIEW': <TokenType.VIEW: 'VIEW'>, 'VOLATILE': <TokenType.VOLATILE: 'VOLATILE'>, 'WHEN': <TokenType.WHEN: 'WHEN'>, 'WHERE': <TokenType.WHERE: 'WHERE'>, 'WINDOW': <TokenType.WINDOW: 'WINDOW'>, 'WITH': <TokenType.WITH: 'WITH'>, 'APPLY': <TokenType.APPLY: 'APPLY'>, 'ARRAY': <TokenType.ARRAY: 'ARRAY'>, 'BIT': <TokenType.BIT: 'BIT'>, 'BOOL': <TokenType.BOOLEAN: 'BOOLEAN'>, 'BOOLEAN': <TokenType.BOOLEAN: 'BOOLEAN'>, 'BYTE': <TokenType.TINYINT: 'TINYINT'>, 'MEDIUMINT': <TokenType.MEDIUMINT: 'MEDIUMINT'>, 'INT1': <TokenType.TINYINT: 'TINYINT'>, 'TINYINT': <TokenType.TINYINT: 'TINYINT'>, 'INT16': <TokenType.SMALLINT: 'SMALLINT'>, 'SHORT': <TokenType.SMALLINT: 'SMALLINT'>, 'SMALLINT': <TokenType.SMALLINT: 'SMALLINT'>, 'INT128': <TokenType.INT128: 'INT128'>, 'HUGEINT': <TokenType.INT128: 'INT128'>, 'INT2': <TokenType.SMALLINT: 'SMALLINT'>, 'INTEGER': <TokenType.INT: 'INT'>, 'INT': <TokenType.INT: 'INT'>, 'INT4': <TokenType.INT: 'INT'>, 'INT32': <TokenType.INT: 'INT'>, 'INT64': <TokenType.BIGINT: 'BIGINT'>, 'LONG': <TokenType.BIGINT: 'BIGINT'>, 'BIGINT': <TokenType.BIGINT: 'BIGINT'>, 'INT8': <TokenType.TINYINT: 'TINYINT'>, 'DEC': <TokenType.DECIMAL: 'DECIMAL'>, 'DECIMAL': <TokenType.DECIMAL: 'DECIMAL'>, 'BIGDECIMAL': <TokenType.BIGDECIMAL: 'BIGDECIMAL'>, 'BIGNUMERIC': <TokenType.BIGDECIMAL: 'BIGDECIMAL'>, 'MAP': <TokenType.MAP: 'MAP'>, 'NULLABLE': <TokenType.NULLABLE: 'NULLABLE'>, 'NUMBER': <TokenType.DECIMAL: 'DECIMAL'>, 'NUMERIC': <TokenType.DECIMAL: 'DECIMAL'>, 'FIXED': <TokenType.DECIMAL: 'DECIMAL'>, 'REAL': <TokenType.FLOAT: 'FLOAT'>, 'FLOAT': <TokenType.FLOAT: 'FLOAT'>, 'FLOAT4': <TokenType.FLOAT: 'FLOAT'>, 'FLOAT8': <TokenType.DOUBLE: 'DOUBLE'>, 'DOUBLE': <TokenType.DOUBLE: 'DOUBLE'>, 'DOUBLE PRECISION': <TokenType.DOUBLE: 'DOUBLE'>, 'JSON': <TokenType.JSON: 'JSON'>, 'CHAR': <TokenType.CHAR: 'CHAR'>, 'CHARACTER': <TokenType.CHAR: 'CHAR'>, 'NCHAR': <TokenType.NCHAR: 'NCHAR'>, 'VARCHAR': <TokenType.VARCHAR: 'VARCHAR'>, 'VARCHAR2': <TokenType.VARCHAR: 'VARCHAR'>, 'NVARCHAR': <TokenType.NVARCHAR: 'NVARCHAR'>, 'NVARCHAR2': <TokenType.NVARCHAR: 'NVARCHAR'>, 'STR': <TokenType.TEXT: 'TEXT'>, 'STRING': <TokenType.TEXT: 'TEXT'>, 'TEXT': <TokenType.TEXT: 'TEXT'>, 'LONGTEXT': <TokenType.LONGTEXT: 'LONGTEXT'>, 'MEDIUMTEXT': <TokenType.MEDIUMTEXT: 'MEDIUMTEXT'>, 'TINYTEXT': <TokenType.TINYTEXT: 'TINYTEXT'>, 'CLOB': <TokenType.TEXT: 'TEXT'>, 'LONGVARCHAR': <TokenType.TEXT: 'TEXT'>, 'BINARY': <TokenType.BINARY: 'BINARY'>, 'BLOB': <TokenType.VARBINARY: 'VARBINARY'>, 'LONGBLOB': <TokenType.LONGBLOB: 'LONGBLOB'>, 'MEDIUMBLOB': <TokenType.MEDIUMBLOB: 'MEDIUMBLOB'>, 'TINYBLOB': <TokenType.TINYBLOB: 'TINYBLOB'>, 'BYTEA': <TokenType.VARBINARY: 'VARBINARY'>, 'VARBINARY': <TokenType.VARBINARY: 'VARBINARY'>, 'TIME': <TokenType.TIME: 'TIME'>, 'TIMETZ': <TokenType.TIMETZ: 'TIMETZ'>, 'TIMESTAMP': <TokenType.TIMESTAMP: 'TIMESTAMP'>, 'TIMESTAMPTZ': <TokenType.TIMESTAMPTZ: 'TIMESTAMPTZ'>, 'TIMESTAMPLTZ': <TokenType.TIMESTAMPLTZ: 'TIMESTAMPLTZ'>, 'DATE': <TokenType.DATE: 'DATE'>, 'DATETIME': <TokenType.DATETIME: 'DATETIME'>, 'INT4RANGE': <TokenType.INT4RANGE: 'INT4RANGE'>, 'INT4MULTIRANGE': <TokenType.INT4MULTIRANGE: 'INT4MULTIRANGE'>, 'INT8RANGE': <TokenType.INT8RANGE: 'INT8RANGE'>, 'INT8MULTIRANGE': <TokenType.INT8MULTIRANGE: 'INT8MULTIRANGE'>, 'NUMRANGE': <TokenType.NUMRANGE: 'NUMRANGE'>, 'NUMMULTIRANGE': <TokenType.NUMMULTIRANGE: 'NUMMULTIRANGE'>, 'TSRANGE': <TokenType.TSRANGE: 'TSRANGE'>, 'TSMULTIRANGE': <TokenType.TSMULTIRANGE: 'TSMULTIRANGE'>, 'TSTZRANGE': <TokenType.TSTZRANGE: 'TSTZRANGE'>, 'TSTZMULTIRANGE': <TokenType.TSTZMULTIRANGE: 'TSTZMULTIRANGE'>, 'DATERANGE': <TokenType.DATERANGE: 'DATERANGE'>, 'DATEMULTIRANGE': <TokenType.DATEMULTIRANGE: 'DATEMULTIRANGE'>, 'UNIQUE': <TokenType.UNIQUE: 'UNIQUE'>, 'STRUCT': <TokenType.STRUCT: 'STRUCT'>, 'VARIANT': <TokenType.VARIANT: 'VARIANT'>, 'ALTER': <TokenType.ALTER: 'ALTER'>, 'ANALYZE': <TokenType.COMMAND: 'COMMAND'>, 'CALL': <TokenType.COMMAND: 'COMMAND'>, 'COMMENT': <TokenType.COMMENT: 'COMMENT'>, 'COPY': <TokenType.COMMAND: 'COMMAND'>, 'EXPLAIN': <TokenType.COMMAND: 'COMMAND'>, 'GRANT': <TokenType.COMMAND: 'COMMAND'>, 'OPTIMIZE': <TokenType.COMMAND: 'COMMAND'>, 'PREPARE': <TokenType.COMMAND: 'COMMAND'>, 'TRUNCATE': <TokenType.COMMAND: 'COMMAND'>, 'VACUUM': <TokenType.COMMAND: 'COMMAND'>, 'USER-DEFINED': <TokenType.USERDEFINED: 'USERDEFINED'>, 'FOR VERSION': <TokenType.VERSION_SNAPSHOT: 'VERSION_SNAPSHOT'>, 'FOR TIMESTAMP': <TokenType.TIMESTAMP_SNAPSHOT: 'TIMESTAMP_SNAPSHOT'>}
WHITE_SPACE: Dict[Optional[str], TokenType] =
{' ': <TokenType.SPACE: 'SPACE'>, '\t': <TokenType.SPACE: 'SPACE'>, '\n': <TokenType.BREAK: 'BREAK'>, '\r': <TokenType.BREAK: 'BREAK'>, '\r\n': <TokenType.BREAK: 'BREAK'>}
COMMANDS =
{<TokenType.COMMAND: 'COMMAND'>, <TokenType.SHOW: 'SHOW'>, <TokenType.EXECUTE: 'EXECUTE'>, <TokenType.FETCH: 'FETCH'>}
COMMAND_PREFIX_TOKENS =
{<TokenType.SEMICOLON: 'SEMICOLON'>, <TokenType.BEGIN: 'BEGIN'>}
def
reset(self) -> None:
829 def reset(self) -> None: 830 self.sql = "" 831 self.size = 0 832 self.tokens: t.List[Token] = [] 833 self._start = 0 834 self._current = 0 835 self._line = 1 836 self._col = 0 837 self._comments: t.List[str] = [] 838 839 self._char = "" 840 self._end = False 841 self._peek = "" 842 self._prev_token_line = -1
844 def tokenize(self, sql: str) -> t.List[Token]: 845 """Returns a list of tokens corresponding to the SQL string `sql`.""" 846 self.reset() 847 self.sql = sql 848 self.size = len(sql) 849 850 try: 851 self._scan() 852 except Exception as e: 853 start = max(self._current - 50, 0) 854 end = min(self._current + 50, self.size - 1) 855 context = self.sql[start:end] 856 raise TokenError(f"Error tokenizing '{context}'") from e 857 858 return self.tokens
Returns a list of tokens corresponding to the SQL string sql
.