sqlglot.dialects.tsql
1from __future__ import annotations 2 3import datetime 4import re 5import typing as t 6from functools import partial 7 8from sqlglot import exp, generator, parser, tokens, transforms 9from sqlglot.dialects.dialect import ( 10 Dialect, 11 NormalizationStrategy, 12 any_value_to_max_sql, 13 date_delta_sql, 14 datestrtodate_sql, 15 generatedasidentitycolumnconstraint_sql, 16 max_or_greatest, 17 min_or_least, 18 build_date_delta, 19 rename_func, 20 trim_sql, 21) 22from sqlglot.helper import seq_get 23from sqlglot.time import format_time 24from sqlglot.tokens import TokenType 25 26if t.TYPE_CHECKING: 27 from sqlglot._typing import E 28 29FULL_FORMAT_TIME_MAPPING = { 30 "weekday": "%A", 31 "dw": "%A", 32 "w": "%A", 33 "month": "%B", 34 "mm": "%B", 35 "m": "%B", 36} 37 38DATE_DELTA_INTERVAL = { 39 "year": "year", 40 "yyyy": "year", 41 "yy": "year", 42 "quarter": "quarter", 43 "qq": "quarter", 44 "q": "quarter", 45 "month": "month", 46 "mm": "month", 47 "m": "month", 48 "week": "week", 49 "ww": "week", 50 "wk": "week", 51 "day": "day", 52 "dd": "day", 53 "d": "day", 54} 55 56 57DATE_FMT_RE = re.compile("([dD]{1,2})|([mM]{1,2})|([yY]{1,4})|([hH]{1,2})|([sS]{1,2})") 58 59# N = Numeric, C=Currency 60TRANSPILE_SAFE_NUMBER_FMT = {"N", "C"} 61 62DEFAULT_START_DATE = datetime.date(1900, 1, 1) 63 64BIT_TYPES = {exp.EQ, exp.NEQ, exp.Is, exp.In, exp.Select, exp.Alias} 65 66# Unsupported options: 67# - OPTIMIZE FOR ( @variable_name { UNKNOWN | = <literal_constant> } [ , ...n ] ) 68# - TABLE HINT 69OPTIONS: parser.OPTIONS_TYPE = { 70 **dict.fromkeys( 71 ( 72 "DISABLE_OPTIMIZED_PLAN_FORCING", 73 "FAST", 74 "IGNORE_NONCLUSTERED_COLUMNSTORE_INDEX", 75 "LABEL", 76 "MAXDOP", 77 "MAXRECURSION", 78 "MAX_GRANT_PERCENT", 79 "MIN_GRANT_PERCENT", 80 "NO_PERFORMANCE_SPOOL", 81 "QUERYTRACEON", 82 "RECOMPILE", 83 ), 84 tuple(), 85 ), 86 "CONCAT": ("UNION",), 87 "DISABLE": ("EXTERNALPUSHDOWN", "SCALEOUTEXECUTION"), 88 "EXPAND": ("VIEWS",), 89 "FORCE": ("EXTERNALPUSHDOWN", "ORDER", "SCALEOUTEXECUTION"), 90 "HASH": ("GROUP", "JOIN", "UNION"), 91 "KEEP": ("PLAN",), 92 "KEEPFIXED": ("PLAN",), 93 "LOOP": ("JOIN",), 94 "MERGE": ("JOIN", "UNION"), 95 "OPTIMIZE": (("FOR", "UNKNOWN"),), 96 "ORDER": ("GROUP",), 97 "PARAMETERIZATION": ("FORCED", "SIMPLE"), 98 "ROBUST": ("PLAN",), 99 "USE": ("PLAN",), 100} 101 102OPTIONS_THAT_REQUIRE_EQUAL = ("MAX_GRANT_PERCENT", "MIN_GRANT_PERCENT", "LABEL") 103 104 105def _build_formatted_time( 106 exp_class: t.Type[E], full_format_mapping: t.Optional[bool] = None 107) -> t.Callable[[t.List], E]: 108 def _builder(args: t.List) -> E: 109 assert len(args) == 2 110 111 return exp_class( 112 this=exp.cast(args[1], exp.DataType.Type.DATETIME), 113 format=exp.Literal.string( 114 format_time( 115 args[0].name.lower(), 116 ( 117 {**TSQL.TIME_MAPPING, **FULL_FORMAT_TIME_MAPPING} 118 if full_format_mapping 119 else TSQL.TIME_MAPPING 120 ), 121 ) 122 ), 123 ) 124 125 return _builder 126 127 128def _build_format(args: t.List) -> exp.NumberToStr | exp.TimeToStr: 129 this = seq_get(args, 0) 130 fmt = seq_get(args, 1) 131 culture = seq_get(args, 2) 132 133 number_fmt = fmt and (fmt.name in TRANSPILE_SAFE_NUMBER_FMT or not DATE_FMT_RE.search(fmt.name)) 134 135 if number_fmt: 136 return exp.NumberToStr(this=this, format=fmt, culture=culture) 137 138 if fmt: 139 fmt = exp.Literal.string( 140 format_time(fmt.name, TSQL.FORMAT_TIME_MAPPING) 141 if len(fmt.name) == 1 142 else format_time(fmt.name, TSQL.TIME_MAPPING) 143 ) 144 145 return exp.TimeToStr(this=this, format=fmt, culture=culture) 146 147 148def _build_eomonth(args: t.List) -> exp.LastDay: 149 date = exp.TsOrDsToDate(this=seq_get(args, 0)) 150 month_lag = seq_get(args, 1) 151 152 if month_lag is None: 153 this: exp.Expression = date 154 else: 155 unit = DATE_DELTA_INTERVAL.get("month") 156 this = exp.DateAdd(this=date, expression=month_lag, unit=unit and exp.var(unit)) 157 158 return exp.LastDay(this=this) 159 160 161def _build_hashbytes(args: t.List) -> exp.Expression: 162 kind, data = args 163 kind = kind.name.upper() if kind.is_string else "" 164 165 if kind == "MD5": 166 args.pop(0) 167 return exp.MD5(this=data) 168 if kind in ("SHA", "SHA1"): 169 args.pop(0) 170 return exp.SHA(this=data) 171 if kind == "SHA2_256": 172 return exp.SHA2(this=data, length=exp.Literal.number(256)) 173 if kind == "SHA2_512": 174 return exp.SHA2(this=data, length=exp.Literal.number(512)) 175 176 return exp.func("HASHBYTES", *args) 177 178 179DATEPART_ONLY_FORMATS = {"DW", "HOUR", "QUARTER"} 180 181 182def _format_sql(self: TSQL.Generator, expression: exp.NumberToStr | exp.TimeToStr) -> str: 183 fmt = expression.args["format"] 184 185 if not isinstance(expression, exp.NumberToStr): 186 if fmt.is_string: 187 mapped_fmt = format_time(fmt.name, TSQL.INVERSE_TIME_MAPPING) 188 189 name = (mapped_fmt or "").upper() 190 if name in DATEPART_ONLY_FORMATS: 191 return self.func("DATEPART", name, expression.this) 192 193 fmt_sql = self.sql(exp.Literal.string(mapped_fmt)) 194 else: 195 fmt_sql = self.format_time(expression) or self.sql(fmt) 196 else: 197 fmt_sql = self.sql(fmt) 198 199 return self.func("FORMAT", expression.this, fmt_sql, expression.args.get("culture")) 200 201 202def _string_agg_sql(self: TSQL.Generator, expression: exp.GroupConcat) -> str: 203 this = expression.this 204 distinct = expression.find(exp.Distinct) 205 if distinct: 206 # exp.Distinct can appear below an exp.Order or an exp.GroupConcat expression 207 self.unsupported("T-SQL STRING_AGG doesn't support DISTINCT.") 208 this = distinct.pop().expressions[0] 209 210 order = "" 211 if isinstance(expression.this, exp.Order): 212 if expression.this.this: 213 this = expression.this.this.pop() 214 # Order has a leading space 215 order = f" WITHIN GROUP ({self.sql(expression.this)[1:]})" 216 217 separator = expression.args.get("separator") or exp.Literal.string(",") 218 return f"STRING_AGG({self.format_args(this, separator)}){order}" 219 220 221def _build_date_delta( 222 exp_class: t.Type[E], unit_mapping: t.Optional[t.Dict[str, str]] = None 223) -> t.Callable[[t.List], E]: 224 def _builder(args: t.List) -> E: 225 unit = seq_get(args, 0) 226 if unit and unit_mapping: 227 unit = exp.var(unit_mapping.get(unit.name.lower(), unit.name)) 228 229 start_date = seq_get(args, 1) 230 if start_date and start_date.is_number: 231 # Numeric types are valid DATETIME values 232 if start_date.is_int: 233 adds = DEFAULT_START_DATE + datetime.timedelta(days=int(start_date.this)) 234 start_date = exp.Literal.string(adds.strftime("%F")) 235 else: 236 # We currently don't handle float values, i.e. they're not converted to equivalent DATETIMEs. 237 # This is not a problem when generating T-SQL code, it is when transpiling to other dialects. 238 return exp_class(this=seq_get(args, 2), expression=start_date, unit=unit) 239 240 return exp_class( 241 this=exp.TimeStrToTime(this=seq_get(args, 2)), 242 expression=exp.TimeStrToTime(this=start_date), 243 unit=unit, 244 ) 245 246 return _builder 247 248 249def qualify_derived_table_outputs(expression: exp.Expression) -> exp.Expression: 250 """Ensures all (unnamed) output columns are aliased for CTEs and Subqueries.""" 251 alias = expression.args.get("alias") 252 253 if ( 254 isinstance(expression, (exp.CTE, exp.Subquery)) 255 and isinstance(alias, exp.TableAlias) 256 and not alias.columns 257 ): 258 from sqlglot.optimizer.qualify_columns import qualify_outputs 259 260 # We keep track of the unaliased column projection indexes instead of the expressions 261 # themselves, because the latter are going to be replaced by new nodes when the aliases 262 # are added and hence we won't be able to reach these newly added Alias parents 263 query = expression.this 264 unaliased_column_indexes = ( 265 i for i, c in enumerate(query.selects) if isinstance(c, exp.Column) and not c.alias 266 ) 267 268 qualify_outputs(query) 269 270 # Preserve the quoting information of columns for newly added Alias nodes 271 query_selects = query.selects 272 for select_index in unaliased_column_indexes: 273 alias = query_selects[select_index] 274 column = alias.this 275 if isinstance(column.this, exp.Identifier): 276 alias.args["alias"].set("quoted", column.this.quoted) 277 278 return expression 279 280 281# https://learn.microsoft.com/en-us/sql/t-sql/functions/datetimefromparts-transact-sql?view=sql-server-ver16#syntax 282def _build_datetimefromparts(args: t.List) -> exp.TimestampFromParts: 283 return exp.TimestampFromParts( 284 year=seq_get(args, 0), 285 month=seq_get(args, 1), 286 day=seq_get(args, 2), 287 hour=seq_get(args, 3), 288 min=seq_get(args, 4), 289 sec=seq_get(args, 5), 290 milli=seq_get(args, 6), 291 ) 292 293 294# https://learn.microsoft.com/en-us/sql/t-sql/functions/timefromparts-transact-sql?view=sql-server-ver16#syntax 295def _build_timefromparts(args: t.List) -> exp.TimeFromParts: 296 return exp.TimeFromParts( 297 hour=seq_get(args, 0), 298 min=seq_get(args, 1), 299 sec=seq_get(args, 2), 300 fractions=seq_get(args, 3), 301 precision=seq_get(args, 4), 302 ) 303 304 305def _build_with_arg_as_text( 306 klass: t.Type[exp.Expression], 307) -> t.Callable[[t.List[exp.Expression]], exp.Expression]: 308 def _parse(args: t.List[exp.Expression]) -> exp.Expression: 309 this = seq_get(args, 0) 310 311 if this and not this.is_string: 312 this = exp.cast(this, exp.DataType.Type.TEXT) 313 314 expression = seq_get(args, 1) 315 kwargs = {"this": this} 316 317 if expression: 318 kwargs["expression"] = expression 319 320 return klass(**kwargs) 321 322 return _parse 323 324 325def _build_json_query(args: t.List, dialect: Dialect) -> exp.JSONExtract: 326 if len(args) == 1: 327 # The default value for path is '$'. As a result, if you don't provide a 328 # value for path, JSON_QUERY returns the input expression. 329 args.append(exp.Literal.string("$")) 330 331 return parser.build_extract_json_with_path(exp.JSONExtract)(args, dialect) 332 333 334def _json_extract_sql( 335 self: TSQL.Generator, expression: exp.JSONExtract | exp.JSONExtractScalar 336) -> str: 337 json_query = self.func("JSON_QUERY", expression.this, expression.expression) 338 json_value = self.func("JSON_VALUE", expression.this, expression.expression) 339 return self.func("ISNULL", json_query, json_value) 340 341 342class TSQL(Dialect): 343 NORMALIZATION_STRATEGY = NormalizationStrategy.CASE_INSENSITIVE 344 TIME_FORMAT = "'yyyy-mm-dd hh:mm:ss'" 345 SUPPORTS_SEMI_ANTI_JOIN = False 346 LOG_BASE_FIRST = False 347 TYPED_DIVISION = True 348 CONCAT_COALESCE = True 349 350 TIME_MAPPING = { 351 "year": "%Y", 352 "dayofyear": "%j", 353 "day": "%d", 354 "dy": "%d", 355 "y": "%Y", 356 "week": "%W", 357 "ww": "%W", 358 "wk": "%W", 359 "hour": "%h", 360 "hh": "%I", 361 "minute": "%M", 362 "mi": "%M", 363 "n": "%M", 364 "second": "%S", 365 "ss": "%S", 366 "s": "%-S", 367 "millisecond": "%f", 368 "ms": "%f", 369 "weekday": "%W", 370 "dw": "%W", 371 "month": "%m", 372 "mm": "%M", 373 "m": "%-M", 374 "Y": "%Y", 375 "YYYY": "%Y", 376 "YY": "%y", 377 "MMMM": "%B", 378 "MMM": "%b", 379 "MM": "%m", 380 "M": "%-m", 381 "dddd": "%A", 382 "dd": "%d", 383 "d": "%-d", 384 "HH": "%H", 385 "H": "%-H", 386 "h": "%-I", 387 "S": "%f", 388 "yyyy": "%Y", 389 "yy": "%y", 390 } 391 392 CONVERT_FORMAT_MAPPING = { 393 "0": "%b %d %Y %-I:%M%p", 394 "1": "%m/%d/%y", 395 "2": "%y.%m.%d", 396 "3": "%d/%m/%y", 397 "4": "%d.%m.%y", 398 "5": "%d-%m-%y", 399 "6": "%d %b %y", 400 "7": "%b %d, %y", 401 "8": "%H:%M:%S", 402 "9": "%b %d %Y %-I:%M:%S:%f%p", 403 "10": "mm-dd-yy", 404 "11": "yy/mm/dd", 405 "12": "yymmdd", 406 "13": "%d %b %Y %H:%M:ss:%f", 407 "14": "%H:%M:%S:%f", 408 "20": "%Y-%m-%d %H:%M:%S", 409 "21": "%Y-%m-%d %H:%M:%S.%f", 410 "22": "%m/%d/%y %-I:%M:%S %p", 411 "23": "%Y-%m-%d", 412 "24": "%H:%M:%S", 413 "25": "%Y-%m-%d %H:%M:%S.%f", 414 "100": "%b %d %Y %-I:%M%p", 415 "101": "%m/%d/%Y", 416 "102": "%Y.%m.%d", 417 "103": "%d/%m/%Y", 418 "104": "%d.%m.%Y", 419 "105": "%d-%m-%Y", 420 "106": "%d %b %Y", 421 "107": "%b %d, %Y", 422 "108": "%H:%M:%S", 423 "109": "%b %d %Y %-I:%M:%S:%f%p", 424 "110": "%m-%d-%Y", 425 "111": "%Y/%m/%d", 426 "112": "%Y%m%d", 427 "113": "%d %b %Y %H:%M:%S:%f", 428 "114": "%H:%M:%S:%f", 429 "120": "%Y-%m-%d %H:%M:%S", 430 "121": "%Y-%m-%d %H:%M:%S.%f", 431 } 432 433 FORMAT_TIME_MAPPING = { 434 "y": "%B %Y", 435 "d": "%m/%d/%Y", 436 "H": "%-H", 437 "h": "%-I", 438 "s": "%Y-%m-%d %H:%M:%S", 439 "D": "%A,%B,%Y", 440 "f": "%A,%B,%Y %-I:%M %p", 441 "F": "%A,%B,%Y %-I:%M:%S %p", 442 "g": "%m/%d/%Y %-I:%M %p", 443 "G": "%m/%d/%Y %-I:%M:%S %p", 444 "M": "%B %-d", 445 "m": "%B %-d", 446 "O": "%Y-%m-%dT%H:%M:%S", 447 "u": "%Y-%M-%D %H:%M:%S%z", 448 "U": "%A, %B %D, %Y %H:%M:%S%z", 449 "T": "%-I:%M:%S %p", 450 "t": "%-I:%M", 451 "Y": "%a %Y", 452 } 453 454 class Tokenizer(tokens.Tokenizer): 455 IDENTIFIERS = [("[", "]"), '"'] 456 QUOTES = ["'", '"'] 457 HEX_STRINGS = [("0x", ""), ("0X", "")] 458 VAR_SINGLE_TOKENS = {"@", "$", "#"} 459 460 KEYWORDS = { 461 **tokens.Tokenizer.KEYWORDS, 462 "CLUSTERED INDEX": TokenType.INDEX, 463 "DATETIME2": TokenType.DATETIME, 464 "DATETIMEOFFSET": TokenType.TIMESTAMPTZ, 465 "DECLARE": TokenType.DECLARE, 466 "EXEC": TokenType.COMMAND, 467 "FOR SYSTEM_TIME": TokenType.TIMESTAMP_SNAPSHOT, 468 "IMAGE": TokenType.IMAGE, 469 "MONEY": TokenType.MONEY, 470 "NONCLUSTERED INDEX": TokenType.INDEX, 471 "NTEXT": TokenType.TEXT, 472 "OPTION": TokenType.OPTION, 473 "OUTPUT": TokenType.RETURNING, 474 "PRINT": TokenType.COMMAND, 475 "PROC": TokenType.PROCEDURE, 476 "REAL": TokenType.FLOAT, 477 "ROWVERSION": TokenType.ROWVERSION, 478 "SMALLDATETIME": TokenType.DATETIME, 479 "SMALLMONEY": TokenType.SMALLMONEY, 480 "SQL_VARIANT": TokenType.VARIANT, 481 "SYSTEM_USER": TokenType.CURRENT_USER, 482 "TOP": TokenType.TOP, 483 "TIMESTAMP": TokenType.ROWVERSION, 484 "TINYINT": TokenType.UTINYINT, 485 "UNIQUEIDENTIFIER": TokenType.UNIQUEIDENTIFIER, 486 "UPDATE STATISTICS": TokenType.COMMAND, 487 "XML": TokenType.XML, 488 } 489 KEYWORDS.pop("/*+") 490 491 COMMANDS = {*tokens.Tokenizer.COMMANDS, TokenType.END} 492 493 class Parser(parser.Parser): 494 SET_REQUIRES_ASSIGNMENT_DELIMITER = False 495 LOG_DEFAULTS_TO_LN = True 496 ALTER_TABLE_ADD_REQUIRED_FOR_EACH_COLUMN = False 497 STRING_ALIASES = True 498 NO_PAREN_IF_COMMANDS = False 499 500 QUERY_MODIFIER_PARSERS = { 501 **parser.Parser.QUERY_MODIFIER_PARSERS, 502 TokenType.OPTION: lambda self: ("options", self._parse_options()), 503 } 504 505 FUNCTIONS = { 506 **parser.Parser.FUNCTIONS, 507 "CHARINDEX": lambda args: exp.StrPosition( 508 this=seq_get(args, 1), 509 substr=seq_get(args, 0), 510 position=seq_get(args, 2), 511 ), 512 "DATEADD": build_date_delta(exp.DateAdd, unit_mapping=DATE_DELTA_INTERVAL), 513 "DATEDIFF": _build_date_delta(exp.DateDiff, unit_mapping=DATE_DELTA_INTERVAL), 514 "DATENAME": _build_formatted_time(exp.TimeToStr, full_format_mapping=True), 515 "DATEPART": _build_formatted_time(exp.TimeToStr), 516 "DATETIMEFROMPARTS": _build_datetimefromparts, 517 "EOMONTH": _build_eomonth, 518 "FORMAT": _build_format, 519 "GETDATE": exp.CurrentTimestamp.from_arg_list, 520 "HASHBYTES": _build_hashbytes, 521 "ISNULL": exp.Coalesce.from_arg_list, 522 "JSON_QUERY": _build_json_query, 523 "JSON_VALUE": parser.build_extract_json_with_path(exp.JSONExtractScalar), 524 "LEN": _build_with_arg_as_text(exp.Length), 525 "LEFT": _build_with_arg_as_text(exp.Left), 526 "RIGHT": _build_with_arg_as_text(exp.Right), 527 "REPLICATE": exp.Repeat.from_arg_list, 528 "SQUARE": lambda args: exp.Pow(this=seq_get(args, 0), expression=exp.Literal.number(2)), 529 "SYSDATETIME": exp.CurrentTimestamp.from_arg_list, 530 "SUSER_NAME": exp.CurrentUser.from_arg_list, 531 "SUSER_SNAME": exp.CurrentUser.from_arg_list, 532 "SYSTEM_USER": exp.CurrentUser.from_arg_list, 533 "TIMEFROMPARTS": _build_timefromparts, 534 } 535 536 JOIN_HINTS = {"LOOP", "HASH", "MERGE", "REMOTE"} 537 538 RETURNS_TABLE_TOKENS = parser.Parser.ID_VAR_TOKENS - { 539 TokenType.TABLE, 540 *parser.Parser.TYPE_TOKENS, 541 } 542 543 STATEMENT_PARSERS = { 544 **parser.Parser.STATEMENT_PARSERS, 545 TokenType.DECLARE: lambda self: self._parse_declare(), 546 } 547 548 RANGE_PARSERS = { 549 **parser.Parser.RANGE_PARSERS, 550 TokenType.DCOLON: lambda self, this: self.expression( 551 exp.ScopeResolution, 552 this=this, 553 expression=self._parse_function() or self._parse_var(any_token=True), 554 ), 555 } 556 557 # The DCOLON (::) operator serves as a scope resolution (exp.ScopeResolution) operator in T-SQL 558 COLUMN_OPERATORS = { 559 **parser.Parser.COLUMN_OPERATORS, 560 TokenType.DCOLON: lambda self, this, to: self.expression(exp.Cast, this=this, to=to) 561 if isinstance(to, exp.DataType) and to.this != exp.DataType.Type.USERDEFINED 562 else self.expression(exp.ScopeResolution, this=this, expression=to), 563 } 564 565 def _parse_dcolon(self) -> t.Optional[exp.Expression]: 566 # We want to use _parse_types() if the first token after :: is a known type, 567 # otherwise we could parse something like x::varchar(max) into a function 568 if self._match_set(self.TYPE_TOKENS, advance=False): 569 return self._parse_types() 570 571 return self._parse_function() or self._parse_types() 572 573 def _parse_options(self) -> t.Optional[t.List[exp.Expression]]: 574 if not self._match(TokenType.OPTION): 575 return None 576 577 def _parse_option() -> t.Optional[exp.Expression]: 578 option = self._parse_var_from_options(OPTIONS) 579 if not option: 580 return None 581 582 self._match(TokenType.EQ) 583 return self.expression( 584 exp.QueryOption, this=option, expression=self._parse_primary_or_var() 585 ) 586 587 return self._parse_wrapped_csv(_parse_option) 588 589 def _parse_projections(self) -> t.List[exp.Expression]: 590 """ 591 T-SQL supports the syntax alias = expression in the SELECT's projection list, 592 so we transform all parsed Selects to convert their EQ projections into Aliases. 593 594 See: https://learn.microsoft.com/en-us/sql/t-sql/queries/select-clause-transact-sql?view=sql-server-ver16#syntax 595 """ 596 return [ 597 ( 598 exp.alias_(projection.expression, projection.this.this, copy=False) 599 if isinstance(projection, exp.EQ) and isinstance(projection.this, exp.Column) 600 else projection 601 ) 602 for projection in super()._parse_projections() 603 ] 604 605 def _parse_commit_or_rollback(self) -> exp.Commit | exp.Rollback: 606 """Applies to SQL Server and Azure SQL Database 607 COMMIT [ { TRAN | TRANSACTION } 608 [ transaction_name | @tran_name_variable ] ] 609 [ WITH ( DELAYED_DURABILITY = { OFF | ON } ) ] 610 611 ROLLBACK { TRAN | TRANSACTION } 612 [ transaction_name | @tran_name_variable 613 | savepoint_name | @savepoint_variable ] 614 """ 615 rollback = self._prev.token_type == TokenType.ROLLBACK 616 617 self._match_texts(("TRAN", "TRANSACTION")) 618 this = self._parse_id_var() 619 620 if rollback: 621 return self.expression(exp.Rollback, this=this) 622 623 durability = None 624 if self._match_pair(TokenType.WITH, TokenType.L_PAREN): 625 self._match_text_seq("DELAYED_DURABILITY") 626 self._match(TokenType.EQ) 627 628 if self._match_text_seq("OFF"): 629 durability = False 630 else: 631 self._match(TokenType.ON) 632 durability = True 633 634 self._match_r_paren() 635 636 return self.expression(exp.Commit, this=this, durability=durability) 637 638 def _parse_transaction(self) -> exp.Transaction | exp.Command: 639 """Applies to SQL Server and Azure SQL Database 640 BEGIN { TRAN | TRANSACTION } 641 [ { transaction_name | @tran_name_variable } 642 [ WITH MARK [ 'description' ] ] 643 ] 644 """ 645 if self._match_texts(("TRAN", "TRANSACTION")): 646 transaction = self.expression(exp.Transaction, this=self._parse_id_var()) 647 if self._match_text_seq("WITH", "MARK"): 648 transaction.set("mark", self._parse_string()) 649 650 return transaction 651 652 return self._parse_as_command(self._prev) 653 654 def _parse_returns(self) -> exp.ReturnsProperty: 655 table = self._parse_id_var(any_token=False, tokens=self.RETURNS_TABLE_TOKENS) 656 returns = super()._parse_returns() 657 returns.set("table", table) 658 return returns 659 660 def _parse_convert( 661 self, strict: bool, safe: t.Optional[bool] = None 662 ) -> t.Optional[exp.Expression]: 663 this = self._parse_types() 664 self._match(TokenType.COMMA) 665 args = [this, *self._parse_csv(self._parse_assignment)] 666 convert = exp.Convert.from_arg_list(args) 667 convert.set("safe", safe) 668 convert.set("strict", strict) 669 return convert 670 671 def _parse_user_defined_function( 672 self, kind: t.Optional[TokenType] = None 673 ) -> t.Optional[exp.Expression]: 674 this = super()._parse_user_defined_function(kind=kind) 675 676 if ( 677 kind == TokenType.FUNCTION 678 or isinstance(this, exp.UserDefinedFunction) 679 or self._match(TokenType.ALIAS, advance=False) 680 ): 681 return this 682 683 expressions = self._parse_csv(self._parse_function_parameter) 684 return self.expression(exp.UserDefinedFunction, this=this, expressions=expressions) 685 686 def _parse_id_var( 687 self, 688 any_token: bool = True, 689 tokens: t.Optional[t.Collection[TokenType]] = None, 690 ) -> t.Optional[exp.Expression]: 691 is_temporary = self._match(TokenType.HASH) 692 is_global = is_temporary and self._match(TokenType.HASH) 693 694 this = super()._parse_id_var(any_token=any_token, tokens=tokens) 695 if this: 696 if is_global: 697 this.set("global", True) 698 elif is_temporary: 699 this.set("temporary", True) 700 701 return this 702 703 def _parse_create(self) -> exp.Create | exp.Command: 704 create = super()._parse_create() 705 706 if isinstance(create, exp.Create): 707 table = create.this.this if isinstance(create.this, exp.Schema) else create.this 708 if isinstance(table, exp.Table) and table.this.args.get("temporary"): 709 if not create.args.get("properties"): 710 create.set("properties", exp.Properties(expressions=[])) 711 712 create.args["properties"].append("expressions", exp.TemporaryProperty()) 713 714 return create 715 716 def _parse_if(self) -> t.Optional[exp.Expression]: 717 index = self._index 718 719 if self._match_text_seq("OBJECT_ID"): 720 self._parse_wrapped_csv(self._parse_string) 721 if self._match_text_seq("IS", "NOT", "NULL") and self._match(TokenType.DROP): 722 return self._parse_drop(exists=True) 723 self._retreat(index) 724 725 return super()._parse_if() 726 727 def _parse_unique(self) -> exp.UniqueColumnConstraint: 728 if self._match_texts(("CLUSTERED", "NONCLUSTERED")): 729 this = self.CONSTRAINT_PARSERS[self._prev.text.upper()](self) 730 else: 731 this = self._parse_schema(self._parse_id_var(any_token=False)) 732 733 return self.expression(exp.UniqueColumnConstraint, this=this) 734 735 def _parse_partition(self) -> t.Optional[exp.Partition]: 736 if not self._match_text_seq("WITH", "(", "PARTITIONS"): 737 return None 738 739 def parse_range(): 740 low = self._parse_bitwise() 741 high = self._parse_bitwise() if self._match_text_seq("TO") else None 742 743 return ( 744 self.expression(exp.PartitionRange, this=low, expression=high) if high else low 745 ) 746 747 partition = self.expression( 748 exp.Partition, expressions=self._parse_wrapped_csv(parse_range) 749 ) 750 751 self._match_r_paren() 752 753 return partition 754 755 def _parse_declare(self) -> exp.Declare | exp.Command: 756 index = self._index 757 expressions = self._try_parse(partial(self._parse_csv, self._parse_declareitem)) 758 759 if not expressions or self._curr: 760 self._retreat(index) 761 return self._parse_as_command(self._prev) 762 763 return self.expression(exp.Declare, expressions=expressions) 764 765 def _parse_declareitem(self) -> t.Optional[exp.DeclareItem]: 766 var = self._parse_id_var() 767 if not var: 768 return None 769 770 value = None 771 self._match(TokenType.ALIAS) 772 if self._match(TokenType.TABLE): 773 data_type = self._parse_schema() 774 else: 775 data_type = self._parse_types() 776 if self._match(TokenType.EQ): 777 value = self._parse_bitwise() 778 779 return self.expression(exp.DeclareItem, this=var, kind=data_type, default=value) 780 781 class Generator(generator.Generator): 782 LIMIT_IS_TOP = True 783 QUERY_HINTS = False 784 RETURNING_END = False 785 NVL2_SUPPORTED = False 786 ALTER_TABLE_INCLUDE_COLUMN_KEYWORD = False 787 LIMIT_FETCH = "FETCH" 788 COMPUTED_COLUMN_WITH_TYPE = False 789 CTE_RECURSIVE_KEYWORD_REQUIRED = False 790 ENSURE_BOOLS = True 791 NULL_ORDERING_SUPPORTED = None 792 SUPPORTS_SINGLE_ARG_CONCAT = False 793 TABLESAMPLE_SEED_KEYWORD = "REPEATABLE" 794 SUPPORTS_SELECT_INTO = True 795 JSON_PATH_BRACKETED_KEY_SUPPORTED = False 796 SUPPORTS_TO_NUMBER = False 797 SET_OP_MODIFIERS = False 798 COPY_PARAMS_EQ_REQUIRED = True 799 PARSE_JSON_NAME = None 800 801 EXPRESSIONS_WITHOUT_NESTED_CTES = { 802 exp.Create, 803 exp.Delete, 804 exp.Insert, 805 exp.Intersect, 806 exp.Except, 807 exp.Merge, 808 exp.Select, 809 exp.Subquery, 810 exp.Union, 811 exp.Update, 812 } 813 814 SUPPORTED_JSON_PATH_PARTS = { 815 exp.JSONPathKey, 816 exp.JSONPathRoot, 817 exp.JSONPathSubscript, 818 } 819 820 TYPE_MAPPING = { 821 **generator.Generator.TYPE_MAPPING, 822 exp.DataType.Type.BOOLEAN: "BIT", 823 exp.DataType.Type.DECIMAL: "NUMERIC", 824 exp.DataType.Type.DATETIME: "DATETIME2", 825 exp.DataType.Type.DOUBLE: "FLOAT", 826 exp.DataType.Type.INT: "INTEGER", 827 exp.DataType.Type.ROWVERSION: "ROWVERSION", 828 exp.DataType.Type.TEXT: "VARCHAR(MAX)", 829 exp.DataType.Type.TIMESTAMP: "DATETIME2", 830 exp.DataType.Type.TIMESTAMPTZ: "DATETIMEOFFSET", 831 exp.DataType.Type.UTINYINT: "TINYINT", 832 exp.DataType.Type.VARIANT: "SQL_VARIANT", 833 } 834 835 TYPE_MAPPING.pop(exp.DataType.Type.NCHAR) 836 TYPE_MAPPING.pop(exp.DataType.Type.NVARCHAR) 837 838 TRANSFORMS = { 839 **generator.Generator.TRANSFORMS, 840 exp.AnyValue: any_value_to_max_sql, 841 exp.ArrayToString: rename_func("STRING_AGG"), 842 exp.AutoIncrementColumnConstraint: lambda *_: "IDENTITY", 843 exp.DateAdd: date_delta_sql("DATEADD"), 844 exp.DateDiff: date_delta_sql("DATEDIFF"), 845 exp.CTE: transforms.preprocess([qualify_derived_table_outputs]), 846 exp.CurrentDate: rename_func("GETDATE"), 847 exp.CurrentTimestamp: rename_func("GETDATE"), 848 exp.DateStrToDate: datestrtodate_sql, 849 exp.Extract: rename_func("DATEPART"), 850 exp.GeneratedAsIdentityColumnConstraint: generatedasidentitycolumnconstraint_sql, 851 exp.GroupConcat: _string_agg_sql, 852 exp.If: rename_func("IIF"), 853 exp.JSONExtract: _json_extract_sql, 854 exp.JSONExtractScalar: _json_extract_sql, 855 exp.LastDay: lambda self, e: self.func("EOMONTH", e.this), 856 exp.Max: max_or_greatest, 857 exp.MD5: lambda self, e: self.func("HASHBYTES", exp.Literal.string("MD5"), e.this), 858 exp.Min: min_or_least, 859 exp.NumberToStr: _format_sql, 860 exp.Repeat: rename_func("REPLICATE"), 861 exp.Select: transforms.preprocess( 862 [ 863 transforms.eliminate_distinct_on, 864 transforms.eliminate_semi_and_anti_joins, 865 transforms.eliminate_qualify, 866 ] 867 ), 868 exp.Stddev: rename_func("STDEV"), 869 exp.StrPosition: lambda self, e: self.func( 870 "CHARINDEX", e.args.get("substr"), e.this, e.args.get("position") 871 ), 872 exp.Subquery: transforms.preprocess([qualify_derived_table_outputs]), 873 exp.SHA: lambda self, e: self.func("HASHBYTES", exp.Literal.string("SHA1"), e.this), 874 exp.SHA2: lambda self, e: self.func( 875 "HASHBYTES", exp.Literal.string(f"SHA2_{e.args.get('length', 256)}"), e.this 876 ), 877 exp.TemporaryProperty: lambda self, e: "", 878 exp.TimeStrToTime: lambda self, e: self.sql( 879 exp.cast(e.this, exp.DataType.Type.DATETIME) 880 ), 881 exp.TimeToStr: _format_sql, 882 exp.Trim: trim_sql, 883 exp.TsOrDsAdd: date_delta_sql("DATEADD", cast=True), 884 exp.TsOrDsDiff: date_delta_sql("DATEDIFF"), 885 } 886 887 TRANSFORMS.pop(exp.ReturnsProperty) 888 889 PROPERTIES_LOCATION = { 890 **generator.Generator.PROPERTIES_LOCATION, 891 exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED, 892 } 893 894 def scope_resolution(self, rhs: str, scope_name: str) -> str: 895 return f"{scope_name}::{rhs}" 896 897 def select_sql(self, expression: exp.Select) -> str: 898 if expression.args.get("offset"): 899 if not expression.args.get("order"): 900 # ORDER BY is required in order to use OFFSET in a query, so we use 901 # a noop order by, since we don't really care about the order. 902 # See: https://www.microsoftpressstore.com/articles/article.aspx?p=2314819 903 expression.order_by(exp.select(exp.null()).subquery(), copy=False) 904 905 limit = expression.args.get("limit") 906 if isinstance(limit, exp.Limit): 907 # TOP and OFFSET can't be combined, we need use FETCH instead of TOP 908 # we replace here because otherwise TOP would be generated in select_sql 909 limit.replace(exp.Fetch(direction="FIRST", count=limit.expression)) 910 911 return super().select_sql(expression) 912 913 def convert_sql(self, expression: exp.Convert) -> str: 914 name = "TRY_CONVERT" if expression.args.get("safe") else "CONVERT" 915 return self.func( 916 name, expression.this, expression.expression, expression.args.get("style") 917 ) 918 919 def queryoption_sql(self, expression: exp.QueryOption) -> str: 920 option = self.sql(expression, "this") 921 value = self.sql(expression, "expression") 922 if value: 923 optional_equal_sign = "= " if option in OPTIONS_THAT_REQUIRE_EQUAL else "" 924 return f"{option} {optional_equal_sign}{value}" 925 return option 926 927 def lateral_op(self, expression: exp.Lateral) -> str: 928 cross_apply = expression.args.get("cross_apply") 929 if cross_apply is True: 930 return "CROSS APPLY" 931 if cross_apply is False: 932 return "OUTER APPLY" 933 934 # TODO: perhaps we can check if the parent is a Join and transpile it appropriately 935 self.unsupported("LATERAL clause is not supported.") 936 return "LATERAL" 937 938 def timefromparts_sql(self, expression: exp.TimeFromParts) -> str: 939 nano = expression.args.get("nano") 940 if nano is not None: 941 nano.pop() 942 self.unsupported("Specifying nanoseconds is not supported in TIMEFROMPARTS.") 943 944 if expression.args.get("fractions") is None: 945 expression.set("fractions", exp.Literal.number(0)) 946 if expression.args.get("precision") is None: 947 expression.set("precision", exp.Literal.number(0)) 948 949 return rename_func("TIMEFROMPARTS")(self, expression) 950 951 def timestampfromparts_sql(self, expression: exp.TimestampFromParts) -> str: 952 zone = expression.args.get("zone") 953 if zone is not None: 954 zone.pop() 955 self.unsupported("Time zone is not supported in DATETIMEFROMPARTS.") 956 957 nano = expression.args.get("nano") 958 if nano is not None: 959 nano.pop() 960 self.unsupported("Specifying nanoseconds is not supported in DATETIMEFROMPARTS.") 961 962 if expression.args.get("milli") is None: 963 expression.set("milli", exp.Literal.number(0)) 964 965 return rename_func("DATETIMEFROMPARTS")(self, expression) 966 967 def setitem_sql(self, expression: exp.SetItem) -> str: 968 this = expression.this 969 if isinstance(this, exp.EQ) and not isinstance(this.left, exp.Parameter): 970 # T-SQL does not use '=' in SET command, except when the LHS is a variable. 971 return f"{self.sql(this.left)} {self.sql(this.right)}" 972 973 return super().setitem_sql(expression) 974 975 def boolean_sql(self, expression: exp.Boolean) -> str: 976 if type(expression.parent) in BIT_TYPES: 977 return "1" if expression.this else "0" 978 979 return "(1 = 1)" if expression.this else "(1 = 0)" 980 981 def is_sql(self, expression: exp.Is) -> str: 982 if isinstance(expression.expression, exp.Boolean): 983 return self.binary(expression, "=") 984 return self.binary(expression, "IS") 985 986 def createable_sql(self, expression: exp.Create, locations: t.DefaultDict) -> str: 987 sql = self.sql(expression, "this") 988 properties = expression.args.get("properties") 989 990 if sql[:1] != "#" and any( 991 isinstance(prop, exp.TemporaryProperty) 992 for prop in (properties.expressions if properties else []) 993 ): 994 sql = f"[#{sql[1:]}" if sql.startswith("[") else f"#{sql}" 995 996 return sql 997 998 def create_sql(self, expression: exp.Create) -> str: 999 kind = expression.kind 1000 exists = expression.args.pop("exists", None) 1001 1002 like_property = expression.find(exp.LikeProperty) 1003 if like_property: 1004 ctas_expression = like_property.this 1005 else: 1006 ctas_expression = expression.expression 1007 1008 if kind == "VIEW": 1009 expression.this.set("catalog", None) 1010 with_ = expression.args.get("with") 1011 if ctas_expression and with_: 1012 # We've already preprocessed the Create expression to bubble up any nested CTEs, 1013 # but CREATE VIEW actually requires the WITH clause to come after it so we need 1014 # to amend the AST by moving the CTEs to the CREATE VIEW statement's query. 1015 ctas_expression.set("with", with_.pop()) 1016 1017 sql = super().create_sql(expression) 1018 1019 table = expression.find(exp.Table) 1020 1021 # Convert CTAS statement to SELECT .. INTO .. 1022 if kind == "TABLE" and ctas_expression: 1023 if isinstance(ctas_expression, exp.UNWRAPPED_QUERIES): 1024 ctas_expression = ctas_expression.subquery() 1025 1026 select_into = exp.select("*").from_(exp.alias_(ctas_expression, "temp", table=True)) 1027 select_into.set("into", exp.Into(this=table)) 1028 1029 if like_property: 1030 select_into.limit(0, copy=False) 1031 1032 sql = self.sql(select_into) 1033 1034 if exists: 1035 identifier = self.sql(exp.Literal.string(exp.table_name(table) if table else "")) 1036 sql = self.sql(exp.Literal.string(sql)) 1037 if kind == "SCHEMA": 1038 sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.schemata WHERE schema_name = {identifier}) EXEC({sql})""" 1039 elif kind == "TABLE": 1040 assert table 1041 where = exp.and_( 1042 exp.column("table_name").eq(table.name), 1043 exp.column("table_schema").eq(table.db) if table.db else None, 1044 exp.column("table_catalog").eq(table.catalog) if table.catalog else None, 1045 ) 1046 sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.tables WHERE {where}) EXEC({sql})""" 1047 elif kind == "INDEX": 1048 index = self.sql(exp.Literal.string(expression.this.text("this"))) 1049 sql = f"""IF NOT EXISTS (SELECT * FROM sys.indexes WHERE object_id = object_id({identifier}) AND name = {index}) EXEC({sql})""" 1050 elif expression.args.get("replace"): 1051 sql = sql.replace("CREATE OR REPLACE ", "CREATE OR ALTER ", 1) 1052 1053 return self.prepend_ctes(expression, sql) 1054 1055 def offset_sql(self, expression: exp.Offset) -> str: 1056 return f"{super().offset_sql(expression)} ROWS" 1057 1058 def version_sql(self, expression: exp.Version) -> str: 1059 name = "SYSTEM_TIME" if expression.name == "TIMESTAMP" else expression.name 1060 this = f"FOR {name}" 1061 expr = expression.expression 1062 kind = expression.text("kind") 1063 if kind in ("FROM", "BETWEEN"): 1064 args = expr.expressions 1065 sep = "TO" if kind == "FROM" else "AND" 1066 expr_sql = f"{self.sql(seq_get(args, 0))} {sep} {self.sql(seq_get(args, 1))}" 1067 else: 1068 expr_sql = self.sql(expr) 1069 1070 expr_sql = f" {expr_sql}" if expr_sql else "" 1071 return f"{this} {kind}{expr_sql}" 1072 1073 def returnsproperty_sql(self, expression: exp.ReturnsProperty) -> str: 1074 table = expression.args.get("table") 1075 table = f"{table} " if table else "" 1076 return f"RETURNS {table}{self.sql(expression, 'this')}" 1077 1078 def returning_sql(self, expression: exp.Returning) -> str: 1079 into = self.sql(expression, "into") 1080 into = self.seg(f"INTO {into}") if into else "" 1081 return f"{self.seg('OUTPUT')} {self.expressions(expression, flat=True)}{into}" 1082 1083 def transaction_sql(self, expression: exp.Transaction) -> str: 1084 this = self.sql(expression, "this") 1085 this = f" {this}" if this else "" 1086 mark = self.sql(expression, "mark") 1087 mark = f" WITH MARK {mark}" if mark else "" 1088 return f"BEGIN TRANSACTION{this}{mark}" 1089 1090 def commit_sql(self, expression: exp.Commit) -> str: 1091 this = self.sql(expression, "this") 1092 this = f" {this}" if this else "" 1093 durability = expression.args.get("durability") 1094 durability = ( 1095 f" WITH (DELAYED_DURABILITY = {'ON' if durability else 'OFF'})" 1096 if durability is not None 1097 else "" 1098 ) 1099 return f"COMMIT TRANSACTION{this}{durability}" 1100 1101 def rollback_sql(self, expression: exp.Rollback) -> str: 1102 this = self.sql(expression, "this") 1103 this = f" {this}" if this else "" 1104 return f"ROLLBACK TRANSACTION{this}" 1105 1106 def identifier_sql(self, expression: exp.Identifier) -> str: 1107 identifier = super().identifier_sql(expression) 1108 1109 if expression.args.get("global"): 1110 identifier = f"##{identifier}" 1111 elif expression.args.get("temporary"): 1112 identifier = f"#{identifier}" 1113 1114 return identifier 1115 1116 def constraint_sql(self, expression: exp.Constraint) -> str: 1117 this = self.sql(expression, "this") 1118 expressions = self.expressions(expression, flat=True, sep=" ") 1119 return f"CONSTRAINT {this} {expressions}" 1120 1121 def length_sql(self, expression: exp.Length) -> str: 1122 return self._uncast_text(expression, "LEN") 1123 1124 def right_sql(self, expression: exp.Right) -> str: 1125 return self._uncast_text(expression, "RIGHT") 1126 1127 def left_sql(self, expression: exp.Left) -> str: 1128 return self._uncast_text(expression, "LEFT") 1129 1130 def _uncast_text(self, expression: exp.Expression, name: str) -> str: 1131 this = expression.this 1132 if isinstance(this, exp.Cast) and this.is_type(exp.DataType.Type.TEXT): 1133 this_sql = self.sql(this, "this") 1134 else: 1135 this_sql = self.sql(this) 1136 expression_sql = self.sql(expression, "expression") 1137 return self.func(name, this_sql, expression_sql if expression_sql else None) 1138 1139 def partition_sql(self, expression: exp.Partition) -> str: 1140 return f"WITH (PARTITIONS({self.expressions(expression, flat=True)}))" 1141 1142 def altertable_sql(self, expression: exp.AlterTable) -> str: 1143 action = seq_get(expression.args.get("actions") or [], 0) 1144 if isinstance(action, exp.RenameTable): 1145 return f"EXEC sp_rename '{self.sql(expression.this)}', '{action.this.name}'" 1146 return super().altertable_sql(expression) 1147 1148 def drop_sql(self, expression: exp.Drop) -> str: 1149 if expression.args["kind"] == "VIEW": 1150 expression.this.set("catalog", None) 1151 return super().drop_sql(expression) 1152 1153 def declare_sql(self, expression: exp.Declare) -> str: 1154 return f"DECLARE {self.expressions(expression, flat=True)}" 1155 1156 def declareitem_sql(self, expression: exp.DeclareItem) -> str: 1157 variable = self.sql(expression, "this") 1158 default = self.sql(expression, "default") 1159 default = f" = {default}" if default else "" 1160 1161 kind = self.sql(expression, "kind") 1162 if isinstance(expression.args.get("kind"), exp.Schema): 1163 kind = f"TABLE {kind}" 1164 1165 return f"{variable} AS {kind}{default}" 1166 1167 def options_modifier(self, expression: exp.Expression) -> str: 1168 options = self.expressions(expression, key="options") 1169 return f" OPTION{self.wrap(options)}" if options else ""
250def qualify_derived_table_outputs(expression: exp.Expression) -> exp.Expression: 251 """Ensures all (unnamed) output columns are aliased for CTEs and Subqueries.""" 252 alias = expression.args.get("alias") 253 254 if ( 255 isinstance(expression, (exp.CTE, exp.Subquery)) 256 and isinstance(alias, exp.TableAlias) 257 and not alias.columns 258 ): 259 from sqlglot.optimizer.qualify_columns import qualify_outputs 260 261 # We keep track of the unaliased column projection indexes instead of the expressions 262 # themselves, because the latter are going to be replaced by new nodes when the aliases 263 # are added and hence we won't be able to reach these newly added Alias parents 264 query = expression.this 265 unaliased_column_indexes = ( 266 i for i, c in enumerate(query.selects) if isinstance(c, exp.Column) and not c.alias 267 ) 268 269 qualify_outputs(query) 270 271 # Preserve the quoting information of columns for newly added Alias nodes 272 query_selects = query.selects 273 for select_index in unaliased_column_indexes: 274 alias = query_selects[select_index] 275 column = alias.this 276 if isinstance(column.this, exp.Identifier): 277 alias.args["alias"].set("quoted", column.this.quoted) 278 279 return expression
Ensures all (unnamed) output columns are aliased for CTEs and Subqueries.
343class TSQL(Dialect): 344 NORMALIZATION_STRATEGY = NormalizationStrategy.CASE_INSENSITIVE 345 TIME_FORMAT = "'yyyy-mm-dd hh:mm:ss'" 346 SUPPORTS_SEMI_ANTI_JOIN = False 347 LOG_BASE_FIRST = False 348 TYPED_DIVISION = True 349 CONCAT_COALESCE = True 350 351 TIME_MAPPING = { 352 "year": "%Y", 353 "dayofyear": "%j", 354 "day": "%d", 355 "dy": "%d", 356 "y": "%Y", 357 "week": "%W", 358 "ww": "%W", 359 "wk": "%W", 360 "hour": "%h", 361 "hh": "%I", 362 "minute": "%M", 363 "mi": "%M", 364 "n": "%M", 365 "second": "%S", 366 "ss": "%S", 367 "s": "%-S", 368 "millisecond": "%f", 369 "ms": "%f", 370 "weekday": "%W", 371 "dw": "%W", 372 "month": "%m", 373 "mm": "%M", 374 "m": "%-M", 375 "Y": "%Y", 376 "YYYY": "%Y", 377 "YY": "%y", 378 "MMMM": "%B", 379 "MMM": "%b", 380 "MM": "%m", 381 "M": "%-m", 382 "dddd": "%A", 383 "dd": "%d", 384 "d": "%-d", 385 "HH": "%H", 386 "H": "%-H", 387 "h": "%-I", 388 "S": "%f", 389 "yyyy": "%Y", 390 "yy": "%y", 391 } 392 393 CONVERT_FORMAT_MAPPING = { 394 "0": "%b %d %Y %-I:%M%p", 395 "1": "%m/%d/%y", 396 "2": "%y.%m.%d", 397 "3": "%d/%m/%y", 398 "4": "%d.%m.%y", 399 "5": "%d-%m-%y", 400 "6": "%d %b %y", 401 "7": "%b %d, %y", 402 "8": "%H:%M:%S", 403 "9": "%b %d %Y %-I:%M:%S:%f%p", 404 "10": "mm-dd-yy", 405 "11": "yy/mm/dd", 406 "12": "yymmdd", 407 "13": "%d %b %Y %H:%M:ss:%f", 408 "14": "%H:%M:%S:%f", 409 "20": "%Y-%m-%d %H:%M:%S", 410 "21": "%Y-%m-%d %H:%M:%S.%f", 411 "22": "%m/%d/%y %-I:%M:%S %p", 412 "23": "%Y-%m-%d", 413 "24": "%H:%M:%S", 414 "25": "%Y-%m-%d %H:%M:%S.%f", 415 "100": "%b %d %Y %-I:%M%p", 416 "101": "%m/%d/%Y", 417 "102": "%Y.%m.%d", 418 "103": "%d/%m/%Y", 419 "104": "%d.%m.%Y", 420 "105": "%d-%m-%Y", 421 "106": "%d %b %Y", 422 "107": "%b %d, %Y", 423 "108": "%H:%M:%S", 424 "109": "%b %d %Y %-I:%M:%S:%f%p", 425 "110": "%m-%d-%Y", 426 "111": "%Y/%m/%d", 427 "112": "%Y%m%d", 428 "113": "%d %b %Y %H:%M:%S:%f", 429 "114": "%H:%M:%S:%f", 430 "120": "%Y-%m-%d %H:%M:%S", 431 "121": "%Y-%m-%d %H:%M:%S.%f", 432 } 433 434 FORMAT_TIME_MAPPING = { 435 "y": "%B %Y", 436 "d": "%m/%d/%Y", 437 "H": "%-H", 438 "h": "%-I", 439 "s": "%Y-%m-%d %H:%M:%S", 440 "D": "%A,%B,%Y", 441 "f": "%A,%B,%Y %-I:%M %p", 442 "F": "%A,%B,%Y %-I:%M:%S %p", 443 "g": "%m/%d/%Y %-I:%M %p", 444 "G": "%m/%d/%Y %-I:%M:%S %p", 445 "M": "%B %-d", 446 "m": "%B %-d", 447 "O": "%Y-%m-%dT%H:%M:%S", 448 "u": "%Y-%M-%D %H:%M:%S%z", 449 "U": "%A, %B %D, %Y %H:%M:%S%z", 450 "T": "%-I:%M:%S %p", 451 "t": "%-I:%M", 452 "Y": "%a %Y", 453 } 454 455 class Tokenizer(tokens.Tokenizer): 456 IDENTIFIERS = [("[", "]"), '"'] 457 QUOTES = ["'", '"'] 458 HEX_STRINGS = [("0x", ""), ("0X", "")] 459 VAR_SINGLE_TOKENS = {"@", "$", "#"} 460 461 KEYWORDS = { 462 **tokens.Tokenizer.KEYWORDS, 463 "CLUSTERED INDEX": TokenType.INDEX, 464 "DATETIME2": TokenType.DATETIME, 465 "DATETIMEOFFSET": TokenType.TIMESTAMPTZ, 466 "DECLARE": TokenType.DECLARE, 467 "EXEC": TokenType.COMMAND, 468 "FOR SYSTEM_TIME": TokenType.TIMESTAMP_SNAPSHOT, 469 "IMAGE": TokenType.IMAGE, 470 "MONEY": TokenType.MONEY, 471 "NONCLUSTERED INDEX": TokenType.INDEX, 472 "NTEXT": TokenType.TEXT, 473 "OPTION": TokenType.OPTION, 474 "OUTPUT": TokenType.RETURNING, 475 "PRINT": TokenType.COMMAND, 476 "PROC": TokenType.PROCEDURE, 477 "REAL": TokenType.FLOAT, 478 "ROWVERSION": TokenType.ROWVERSION, 479 "SMALLDATETIME": TokenType.DATETIME, 480 "SMALLMONEY": TokenType.SMALLMONEY, 481 "SQL_VARIANT": TokenType.VARIANT, 482 "SYSTEM_USER": TokenType.CURRENT_USER, 483 "TOP": TokenType.TOP, 484 "TIMESTAMP": TokenType.ROWVERSION, 485 "TINYINT": TokenType.UTINYINT, 486 "UNIQUEIDENTIFIER": TokenType.UNIQUEIDENTIFIER, 487 "UPDATE STATISTICS": TokenType.COMMAND, 488 "XML": TokenType.XML, 489 } 490 KEYWORDS.pop("/*+") 491 492 COMMANDS = {*tokens.Tokenizer.COMMANDS, TokenType.END} 493 494 class Parser(parser.Parser): 495 SET_REQUIRES_ASSIGNMENT_DELIMITER = False 496 LOG_DEFAULTS_TO_LN = True 497 ALTER_TABLE_ADD_REQUIRED_FOR_EACH_COLUMN = False 498 STRING_ALIASES = True 499 NO_PAREN_IF_COMMANDS = False 500 501 QUERY_MODIFIER_PARSERS = { 502 **parser.Parser.QUERY_MODIFIER_PARSERS, 503 TokenType.OPTION: lambda self: ("options", self._parse_options()), 504 } 505 506 FUNCTIONS = { 507 **parser.Parser.FUNCTIONS, 508 "CHARINDEX": lambda args: exp.StrPosition( 509 this=seq_get(args, 1), 510 substr=seq_get(args, 0), 511 position=seq_get(args, 2), 512 ), 513 "DATEADD": build_date_delta(exp.DateAdd, unit_mapping=DATE_DELTA_INTERVAL), 514 "DATEDIFF": _build_date_delta(exp.DateDiff, unit_mapping=DATE_DELTA_INTERVAL), 515 "DATENAME": _build_formatted_time(exp.TimeToStr, full_format_mapping=True), 516 "DATEPART": _build_formatted_time(exp.TimeToStr), 517 "DATETIMEFROMPARTS": _build_datetimefromparts, 518 "EOMONTH": _build_eomonth, 519 "FORMAT": _build_format, 520 "GETDATE": exp.CurrentTimestamp.from_arg_list, 521 "HASHBYTES": _build_hashbytes, 522 "ISNULL": exp.Coalesce.from_arg_list, 523 "JSON_QUERY": _build_json_query, 524 "JSON_VALUE": parser.build_extract_json_with_path(exp.JSONExtractScalar), 525 "LEN": _build_with_arg_as_text(exp.Length), 526 "LEFT": _build_with_arg_as_text(exp.Left), 527 "RIGHT": _build_with_arg_as_text(exp.Right), 528 "REPLICATE": exp.Repeat.from_arg_list, 529 "SQUARE": lambda args: exp.Pow(this=seq_get(args, 0), expression=exp.Literal.number(2)), 530 "SYSDATETIME": exp.CurrentTimestamp.from_arg_list, 531 "SUSER_NAME": exp.CurrentUser.from_arg_list, 532 "SUSER_SNAME": exp.CurrentUser.from_arg_list, 533 "SYSTEM_USER": exp.CurrentUser.from_arg_list, 534 "TIMEFROMPARTS": _build_timefromparts, 535 } 536 537 JOIN_HINTS = {"LOOP", "HASH", "MERGE", "REMOTE"} 538 539 RETURNS_TABLE_TOKENS = parser.Parser.ID_VAR_TOKENS - { 540 TokenType.TABLE, 541 *parser.Parser.TYPE_TOKENS, 542 } 543 544 STATEMENT_PARSERS = { 545 **parser.Parser.STATEMENT_PARSERS, 546 TokenType.DECLARE: lambda self: self._parse_declare(), 547 } 548 549 RANGE_PARSERS = { 550 **parser.Parser.RANGE_PARSERS, 551 TokenType.DCOLON: lambda self, this: self.expression( 552 exp.ScopeResolution, 553 this=this, 554 expression=self._parse_function() or self._parse_var(any_token=True), 555 ), 556 } 557 558 # The DCOLON (::) operator serves as a scope resolution (exp.ScopeResolution) operator in T-SQL 559 COLUMN_OPERATORS = { 560 **parser.Parser.COLUMN_OPERATORS, 561 TokenType.DCOLON: lambda self, this, to: self.expression(exp.Cast, this=this, to=to) 562 if isinstance(to, exp.DataType) and to.this != exp.DataType.Type.USERDEFINED 563 else self.expression(exp.ScopeResolution, this=this, expression=to), 564 } 565 566 def _parse_dcolon(self) -> t.Optional[exp.Expression]: 567 # We want to use _parse_types() if the first token after :: is a known type, 568 # otherwise we could parse something like x::varchar(max) into a function 569 if self._match_set(self.TYPE_TOKENS, advance=False): 570 return self._parse_types() 571 572 return self._parse_function() or self._parse_types() 573 574 def _parse_options(self) -> t.Optional[t.List[exp.Expression]]: 575 if not self._match(TokenType.OPTION): 576 return None 577 578 def _parse_option() -> t.Optional[exp.Expression]: 579 option = self._parse_var_from_options(OPTIONS) 580 if not option: 581 return None 582 583 self._match(TokenType.EQ) 584 return self.expression( 585 exp.QueryOption, this=option, expression=self._parse_primary_or_var() 586 ) 587 588 return self._parse_wrapped_csv(_parse_option) 589 590 def _parse_projections(self) -> t.List[exp.Expression]: 591 """ 592 T-SQL supports the syntax alias = expression in the SELECT's projection list, 593 so we transform all parsed Selects to convert their EQ projections into Aliases. 594 595 See: https://learn.microsoft.com/en-us/sql/t-sql/queries/select-clause-transact-sql?view=sql-server-ver16#syntax 596 """ 597 return [ 598 ( 599 exp.alias_(projection.expression, projection.this.this, copy=False) 600 if isinstance(projection, exp.EQ) and isinstance(projection.this, exp.Column) 601 else projection 602 ) 603 for projection in super()._parse_projections() 604 ] 605 606 def _parse_commit_or_rollback(self) -> exp.Commit | exp.Rollback: 607 """Applies to SQL Server and Azure SQL Database 608 COMMIT [ { TRAN | TRANSACTION } 609 [ transaction_name | @tran_name_variable ] ] 610 [ WITH ( DELAYED_DURABILITY = { OFF | ON } ) ] 611 612 ROLLBACK { TRAN | TRANSACTION } 613 [ transaction_name | @tran_name_variable 614 | savepoint_name | @savepoint_variable ] 615 """ 616 rollback = self._prev.token_type == TokenType.ROLLBACK 617 618 self._match_texts(("TRAN", "TRANSACTION")) 619 this = self._parse_id_var() 620 621 if rollback: 622 return self.expression(exp.Rollback, this=this) 623 624 durability = None 625 if self._match_pair(TokenType.WITH, TokenType.L_PAREN): 626 self._match_text_seq("DELAYED_DURABILITY") 627 self._match(TokenType.EQ) 628 629 if self._match_text_seq("OFF"): 630 durability = False 631 else: 632 self._match(TokenType.ON) 633 durability = True 634 635 self._match_r_paren() 636 637 return self.expression(exp.Commit, this=this, durability=durability) 638 639 def _parse_transaction(self) -> exp.Transaction | exp.Command: 640 """Applies to SQL Server and Azure SQL Database 641 BEGIN { TRAN | TRANSACTION } 642 [ { transaction_name | @tran_name_variable } 643 [ WITH MARK [ 'description' ] ] 644 ] 645 """ 646 if self._match_texts(("TRAN", "TRANSACTION")): 647 transaction = self.expression(exp.Transaction, this=self._parse_id_var()) 648 if self._match_text_seq("WITH", "MARK"): 649 transaction.set("mark", self._parse_string()) 650 651 return transaction 652 653 return self._parse_as_command(self._prev) 654 655 def _parse_returns(self) -> exp.ReturnsProperty: 656 table = self._parse_id_var(any_token=False, tokens=self.RETURNS_TABLE_TOKENS) 657 returns = super()._parse_returns() 658 returns.set("table", table) 659 return returns 660 661 def _parse_convert( 662 self, strict: bool, safe: t.Optional[bool] = None 663 ) -> t.Optional[exp.Expression]: 664 this = self._parse_types() 665 self._match(TokenType.COMMA) 666 args = [this, *self._parse_csv(self._parse_assignment)] 667 convert = exp.Convert.from_arg_list(args) 668 convert.set("safe", safe) 669 convert.set("strict", strict) 670 return convert 671 672 def _parse_user_defined_function( 673 self, kind: t.Optional[TokenType] = None 674 ) -> t.Optional[exp.Expression]: 675 this = super()._parse_user_defined_function(kind=kind) 676 677 if ( 678 kind == TokenType.FUNCTION 679 or isinstance(this, exp.UserDefinedFunction) 680 or self._match(TokenType.ALIAS, advance=False) 681 ): 682 return this 683 684 expressions = self._parse_csv(self._parse_function_parameter) 685 return self.expression(exp.UserDefinedFunction, this=this, expressions=expressions) 686 687 def _parse_id_var( 688 self, 689 any_token: bool = True, 690 tokens: t.Optional[t.Collection[TokenType]] = None, 691 ) -> t.Optional[exp.Expression]: 692 is_temporary = self._match(TokenType.HASH) 693 is_global = is_temporary and self._match(TokenType.HASH) 694 695 this = super()._parse_id_var(any_token=any_token, tokens=tokens) 696 if this: 697 if is_global: 698 this.set("global", True) 699 elif is_temporary: 700 this.set("temporary", True) 701 702 return this 703 704 def _parse_create(self) -> exp.Create | exp.Command: 705 create = super()._parse_create() 706 707 if isinstance(create, exp.Create): 708 table = create.this.this if isinstance(create.this, exp.Schema) else create.this 709 if isinstance(table, exp.Table) and table.this.args.get("temporary"): 710 if not create.args.get("properties"): 711 create.set("properties", exp.Properties(expressions=[])) 712 713 create.args["properties"].append("expressions", exp.TemporaryProperty()) 714 715 return create 716 717 def _parse_if(self) -> t.Optional[exp.Expression]: 718 index = self._index 719 720 if self._match_text_seq("OBJECT_ID"): 721 self._parse_wrapped_csv(self._parse_string) 722 if self._match_text_seq("IS", "NOT", "NULL") and self._match(TokenType.DROP): 723 return self._parse_drop(exists=True) 724 self._retreat(index) 725 726 return super()._parse_if() 727 728 def _parse_unique(self) -> exp.UniqueColumnConstraint: 729 if self._match_texts(("CLUSTERED", "NONCLUSTERED")): 730 this = self.CONSTRAINT_PARSERS[self._prev.text.upper()](self) 731 else: 732 this = self._parse_schema(self._parse_id_var(any_token=False)) 733 734 return self.expression(exp.UniqueColumnConstraint, this=this) 735 736 def _parse_partition(self) -> t.Optional[exp.Partition]: 737 if not self._match_text_seq("WITH", "(", "PARTITIONS"): 738 return None 739 740 def parse_range(): 741 low = self._parse_bitwise() 742 high = self._parse_bitwise() if self._match_text_seq("TO") else None 743 744 return ( 745 self.expression(exp.PartitionRange, this=low, expression=high) if high else low 746 ) 747 748 partition = self.expression( 749 exp.Partition, expressions=self._parse_wrapped_csv(parse_range) 750 ) 751 752 self._match_r_paren() 753 754 return partition 755 756 def _parse_declare(self) -> exp.Declare | exp.Command: 757 index = self._index 758 expressions = self._try_parse(partial(self._parse_csv, self._parse_declareitem)) 759 760 if not expressions or self._curr: 761 self._retreat(index) 762 return self._parse_as_command(self._prev) 763 764 return self.expression(exp.Declare, expressions=expressions) 765 766 def _parse_declareitem(self) -> t.Optional[exp.DeclareItem]: 767 var = self._parse_id_var() 768 if not var: 769 return None 770 771 value = None 772 self._match(TokenType.ALIAS) 773 if self._match(TokenType.TABLE): 774 data_type = self._parse_schema() 775 else: 776 data_type = self._parse_types() 777 if self._match(TokenType.EQ): 778 value = self._parse_bitwise() 779 780 return self.expression(exp.DeclareItem, this=var, kind=data_type, default=value) 781 782 class Generator(generator.Generator): 783 LIMIT_IS_TOP = True 784 QUERY_HINTS = False 785 RETURNING_END = False 786 NVL2_SUPPORTED = False 787 ALTER_TABLE_INCLUDE_COLUMN_KEYWORD = False 788 LIMIT_FETCH = "FETCH" 789 COMPUTED_COLUMN_WITH_TYPE = False 790 CTE_RECURSIVE_KEYWORD_REQUIRED = False 791 ENSURE_BOOLS = True 792 NULL_ORDERING_SUPPORTED = None 793 SUPPORTS_SINGLE_ARG_CONCAT = False 794 TABLESAMPLE_SEED_KEYWORD = "REPEATABLE" 795 SUPPORTS_SELECT_INTO = True 796 JSON_PATH_BRACKETED_KEY_SUPPORTED = False 797 SUPPORTS_TO_NUMBER = False 798 SET_OP_MODIFIERS = False 799 COPY_PARAMS_EQ_REQUIRED = True 800 PARSE_JSON_NAME = None 801 802 EXPRESSIONS_WITHOUT_NESTED_CTES = { 803 exp.Create, 804 exp.Delete, 805 exp.Insert, 806 exp.Intersect, 807 exp.Except, 808 exp.Merge, 809 exp.Select, 810 exp.Subquery, 811 exp.Union, 812 exp.Update, 813 } 814 815 SUPPORTED_JSON_PATH_PARTS = { 816 exp.JSONPathKey, 817 exp.JSONPathRoot, 818 exp.JSONPathSubscript, 819 } 820 821 TYPE_MAPPING = { 822 **generator.Generator.TYPE_MAPPING, 823 exp.DataType.Type.BOOLEAN: "BIT", 824 exp.DataType.Type.DECIMAL: "NUMERIC", 825 exp.DataType.Type.DATETIME: "DATETIME2", 826 exp.DataType.Type.DOUBLE: "FLOAT", 827 exp.DataType.Type.INT: "INTEGER", 828 exp.DataType.Type.ROWVERSION: "ROWVERSION", 829 exp.DataType.Type.TEXT: "VARCHAR(MAX)", 830 exp.DataType.Type.TIMESTAMP: "DATETIME2", 831 exp.DataType.Type.TIMESTAMPTZ: "DATETIMEOFFSET", 832 exp.DataType.Type.UTINYINT: "TINYINT", 833 exp.DataType.Type.VARIANT: "SQL_VARIANT", 834 } 835 836 TYPE_MAPPING.pop(exp.DataType.Type.NCHAR) 837 TYPE_MAPPING.pop(exp.DataType.Type.NVARCHAR) 838 839 TRANSFORMS = { 840 **generator.Generator.TRANSFORMS, 841 exp.AnyValue: any_value_to_max_sql, 842 exp.ArrayToString: rename_func("STRING_AGG"), 843 exp.AutoIncrementColumnConstraint: lambda *_: "IDENTITY", 844 exp.DateAdd: date_delta_sql("DATEADD"), 845 exp.DateDiff: date_delta_sql("DATEDIFF"), 846 exp.CTE: transforms.preprocess([qualify_derived_table_outputs]), 847 exp.CurrentDate: rename_func("GETDATE"), 848 exp.CurrentTimestamp: rename_func("GETDATE"), 849 exp.DateStrToDate: datestrtodate_sql, 850 exp.Extract: rename_func("DATEPART"), 851 exp.GeneratedAsIdentityColumnConstraint: generatedasidentitycolumnconstraint_sql, 852 exp.GroupConcat: _string_agg_sql, 853 exp.If: rename_func("IIF"), 854 exp.JSONExtract: _json_extract_sql, 855 exp.JSONExtractScalar: _json_extract_sql, 856 exp.LastDay: lambda self, e: self.func("EOMONTH", e.this), 857 exp.Max: max_or_greatest, 858 exp.MD5: lambda self, e: self.func("HASHBYTES", exp.Literal.string("MD5"), e.this), 859 exp.Min: min_or_least, 860 exp.NumberToStr: _format_sql, 861 exp.Repeat: rename_func("REPLICATE"), 862 exp.Select: transforms.preprocess( 863 [ 864 transforms.eliminate_distinct_on, 865 transforms.eliminate_semi_and_anti_joins, 866 transforms.eliminate_qualify, 867 ] 868 ), 869 exp.Stddev: rename_func("STDEV"), 870 exp.StrPosition: lambda self, e: self.func( 871 "CHARINDEX", e.args.get("substr"), e.this, e.args.get("position") 872 ), 873 exp.Subquery: transforms.preprocess([qualify_derived_table_outputs]), 874 exp.SHA: lambda self, e: self.func("HASHBYTES", exp.Literal.string("SHA1"), e.this), 875 exp.SHA2: lambda self, e: self.func( 876 "HASHBYTES", exp.Literal.string(f"SHA2_{e.args.get('length', 256)}"), e.this 877 ), 878 exp.TemporaryProperty: lambda self, e: "", 879 exp.TimeStrToTime: lambda self, e: self.sql( 880 exp.cast(e.this, exp.DataType.Type.DATETIME) 881 ), 882 exp.TimeToStr: _format_sql, 883 exp.Trim: trim_sql, 884 exp.TsOrDsAdd: date_delta_sql("DATEADD", cast=True), 885 exp.TsOrDsDiff: date_delta_sql("DATEDIFF"), 886 } 887 888 TRANSFORMS.pop(exp.ReturnsProperty) 889 890 PROPERTIES_LOCATION = { 891 **generator.Generator.PROPERTIES_LOCATION, 892 exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED, 893 } 894 895 def scope_resolution(self, rhs: str, scope_name: str) -> str: 896 return f"{scope_name}::{rhs}" 897 898 def select_sql(self, expression: exp.Select) -> str: 899 if expression.args.get("offset"): 900 if not expression.args.get("order"): 901 # ORDER BY is required in order to use OFFSET in a query, so we use 902 # a noop order by, since we don't really care about the order. 903 # See: https://www.microsoftpressstore.com/articles/article.aspx?p=2314819 904 expression.order_by(exp.select(exp.null()).subquery(), copy=False) 905 906 limit = expression.args.get("limit") 907 if isinstance(limit, exp.Limit): 908 # TOP and OFFSET can't be combined, we need use FETCH instead of TOP 909 # we replace here because otherwise TOP would be generated in select_sql 910 limit.replace(exp.Fetch(direction="FIRST", count=limit.expression)) 911 912 return super().select_sql(expression) 913 914 def convert_sql(self, expression: exp.Convert) -> str: 915 name = "TRY_CONVERT" if expression.args.get("safe") else "CONVERT" 916 return self.func( 917 name, expression.this, expression.expression, expression.args.get("style") 918 ) 919 920 def queryoption_sql(self, expression: exp.QueryOption) -> str: 921 option = self.sql(expression, "this") 922 value = self.sql(expression, "expression") 923 if value: 924 optional_equal_sign = "= " if option in OPTIONS_THAT_REQUIRE_EQUAL else "" 925 return f"{option} {optional_equal_sign}{value}" 926 return option 927 928 def lateral_op(self, expression: exp.Lateral) -> str: 929 cross_apply = expression.args.get("cross_apply") 930 if cross_apply is True: 931 return "CROSS APPLY" 932 if cross_apply is False: 933 return "OUTER APPLY" 934 935 # TODO: perhaps we can check if the parent is a Join and transpile it appropriately 936 self.unsupported("LATERAL clause is not supported.") 937 return "LATERAL" 938 939 def timefromparts_sql(self, expression: exp.TimeFromParts) -> str: 940 nano = expression.args.get("nano") 941 if nano is not None: 942 nano.pop() 943 self.unsupported("Specifying nanoseconds is not supported in TIMEFROMPARTS.") 944 945 if expression.args.get("fractions") is None: 946 expression.set("fractions", exp.Literal.number(0)) 947 if expression.args.get("precision") is None: 948 expression.set("precision", exp.Literal.number(0)) 949 950 return rename_func("TIMEFROMPARTS")(self, expression) 951 952 def timestampfromparts_sql(self, expression: exp.TimestampFromParts) -> str: 953 zone = expression.args.get("zone") 954 if zone is not None: 955 zone.pop() 956 self.unsupported("Time zone is not supported in DATETIMEFROMPARTS.") 957 958 nano = expression.args.get("nano") 959 if nano is not None: 960 nano.pop() 961 self.unsupported("Specifying nanoseconds is not supported in DATETIMEFROMPARTS.") 962 963 if expression.args.get("milli") is None: 964 expression.set("milli", exp.Literal.number(0)) 965 966 return rename_func("DATETIMEFROMPARTS")(self, expression) 967 968 def setitem_sql(self, expression: exp.SetItem) -> str: 969 this = expression.this 970 if isinstance(this, exp.EQ) and not isinstance(this.left, exp.Parameter): 971 # T-SQL does not use '=' in SET command, except when the LHS is a variable. 972 return f"{self.sql(this.left)} {self.sql(this.right)}" 973 974 return super().setitem_sql(expression) 975 976 def boolean_sql(self, expression: exp.Boolean) -> str: 977 if type(expression.parent) in BIT_TYPES: 978 return "1" if expression.this else "0" 979 980 return "(1 = 1)" if expression.this else "(1 = 0)" 981 982 def is_sql(self, expression: exp.Is) -> str: 983 if isinstance(expression.expression, exp.Boolean): 984 return self.binary(expression, "=") 985 return self.binary(expression, "IS") 986 987 def createable_sql(self, expression: exp.Create, locations: t.DefaultDict) -> str: 988 sql = self.sql(expression, "this") 989 properties = expression.args.get("properties") 990 991 if sql[:1] != "#" and any( 992 isinstance(prop, exp.TemporaryProperty) 993 for prop in (properties.expressions if properties else []) 994 ): 995 sql = f"[#{sql[1:]}" if sql.startswith("[") else f"#{sql}" 996 997 return sql 998 999 def create_sql(self, expression: exp.Create) -> str: 1000 kind = expression.kind 1001 exists = expression.args.pop("exists", None) 1002 1003 like_property = expression.find(exp.LikeProperty) 1004 if like_property: 1005 ctas_expression = like_property.this 1006 else: 1007 ctas_expression = expression.expression 1008 1009 if kind == "VIEW": 1010 expression.this.set("catalog", None) 1011 with_ = expression.args.get("with") 1012 if ctas_expression and with_: 1013 # We've already preprocessed the Create expression to bubble up any nested CTEs, 1014 # but CREATE VIEW actually requires the WITH clause to come after it so we need 1015 # to amend the AST by moving the CTEs to the CREATE VIEW statement's query. 1016 ctas_expression.set("with", with_.pop()) 1017 1018 sql = super().create_sql(expression) 1019 1020 table = expression.find(exp.Table) 1021 1022 # Convert CTAS statement to SELECT .. INTO .. 1023 if kind == "TABLE" and ctas_expression: 1024 if isinstance(ctas_expression, exp.UNWRAPPED_QUERIES): 1025 ctas_expression = ctas_expression.subquery() 1026 1027 select_into = exp.select("*").from_(exp.alias_(ctas_expression, "temp", table=True)) 1028 select_into.set("into", exp.Into(this=table)) 1029 1030 if like_property: 1031 select_into.limit(0, copy=False) 1032 1033 sql = self.sql(select_into) 1034 1035 if exists: 1036 identifier = self.sql(exp.Literal.string(exp.table_name(table) if table else "")) 1037 sql = self.sql(exp.Literal.string(sql)) 1038 if kind == "SCHEMA": 1039 sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.schemata WHERE schema_name = {identifier}) EXEC({sql})""" 1040 elif kind == "TABLE": 1041 assert table 1042 where = exp.and_( 1043 exp.column("table_name").eq(table.name), 1044 exp.column("table_schema").eq(table.db) if table.db else None, 1045 exp.column("table_catalog").eq(table.catalog) if table.catalog else None, 1046 ) 1047 sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.tables WHERE {where}) EXEC({sql})""" 1048 elif kind == "INDEX": 1049 index = self.sql(exp.Literal.string(expression.this.text("this"))) 1050 sql = f"""IF NOT EXISTS (SELECT * FROM sys.indexes WHERE object_id = object_id({identifier}) AND name = {index}) EXEC({sql})""" 1051 elif expression.args.get("replace"): 1052 sql = sql.replace("CREATE OR REPLACE ", "CREATE OR ALTER ", 1) 1053 1054 return self.prepend_ctes(expression, sql) 1055 1056 def offset_sql(self, expression: exp.Offset) -> str: 1057 return f"{super().offset_sql(expression)} ROWS" 1058 1059 def version_sql(self, expression: exp.Version) -> str: 1060 name = "SYSTEM_TIME" if expression.name == "TIMESTAMP" else expression.name 1061 this = f"FOR {name}" 1062 expr = expression.expression 1063 kind = expression.text("kind") 1064 if kind in ("FROM", "BETWEEN"): 1065 args = expr.expressions 1066 sep = "TO" if kind == "FROM" else "AND" 1067 expr_sql = f"{self.sql(seq_get(args, 0))} {sep} {self.sql(seq_get(args, 1))}" 1068 else: 1069 expr_sql = self.sql(expr) 1070 1071 expr_sql = f" {expr_sql}" if expr_sql else "" 1072 return f"{this} {kind}{expr_sql}" 1073 1074 def returnsproperty_sql(self, expression: exp.ReturnsProperty) -> str: 1075 table = expression.args.get("table") 1076 table = f"{table} " if table else "" 1077 return f"RETURNS {table}{self.sql(expression, 'this')}" 1078 1079 def returning_sql(self, expression: exp.Returning) -> str: 1080 into = self.sql(expression, "into") 1081 into = self.seg(f"INTO {into}") if into else "" 1082 return f"{self.seg('OUTPUT')} {self.expressions(expression, flat=True)}{into}" 1083 1084 def transaction_sql(self, expression: exp.Transaction) -> str: 1085 this = self.sql(expression, "this") 1086 this = f" {this}" if this else "" 1087 mark = self.sql(expression, "mark") 1088 mark = f" WITH MARK {mark}" if mark else "" 1089 return f"BEGIN TRANSACTION{this}{mark}" 1090 1091 def commit_sql(self, expression: exp.Commit) -> str: 1092 this = self.sql(expression, "this") 1093 this = f" {this}" if this else "" 1094 durability = expression.args.get("durability") 1095 durability = ( 1096 f" WITH (DELAYED_DURABILITY = {'ON' if durability else 'OFF'})" 1097 if durability is not None 1098 else "" 1099 ) 1100 return f"COMMIT TRANSACTION{this}{durability}" 1101 1102 def rollback_sql(self, expression: exp.Rollback) -> str: 1103 this = self.sql(expression, "this") 1104 this = f" {this}" if this else "" 1105 return f"ROLLBACK TRANSACTION{this}" 1106 1107 def identifier_sql(self, expression: exp.Identifier) -> str: 1108 identifier = super().identifier_sql(expression) 1109 1110 if expression.args.get("global"): 1111 identifier = f"##{identifier}" 1112 elif expression.args.get("temporary"): 1113 identifier = f"#{identifier}" 1114 1115 return identifier 1116 1117 def constraint_sql(self, expression: exp.Constraint) -> str: 1118 this = self.sql(expression, "this") 1119 expressions = self.expressions(expression, flat=True, sep=" ") 1120 return f"CONSTRAINT {this} {expressions}" 1121 1122 def length_sql(self, expression: exp.Length) -> str: 1123 return self._uncast_text(expression, "LEN") 1124 1125 def right_sql(self, expression: exp.Right) -> str: 1126 return self._uncast_text(expression, "RIGHT") 1127 1128 def left_sql(self, expression: exp.Left) -> str: 1129 return self._uncast_text(expression, "LEFT") 1130 1131 def _uncast_text(self, expression: exp.Expression, name: str) -> str: 1132 this = expression.this 1133 if isinstance(this, exp.Cast) and this.is_type(exp.DataType.Type.TEXT): 1134 this_sql = self.sql(this, "this") 1135 else: 1136 this_sql = self.sql(this) 1137 expression_sql = self.sql(expression, "expression") 1138 return self.func(name, this_sql, expression_sql if expression_sql else None) 1139 1140 def partition_sql(self, expression: exp.Partition) -> str: 1141 return f"WITH (PARTITIONS({self.expressions(expression, flat=True)}))" 1142 1143 def altertable_sql(self, expression: exp.AlterTable) -> str: 1144 action = seq_get(expression.args.get("actions") or [], 0) 1145 if isinstance(action, exp.RenameTable): 1146 return f"EXEC sp_rename '{self.sql(expression.this)}', '{action.this.name}'" 1147 return super().altertable_sql(expression) 1148 1149 def drop_sql(self, expression: exp.Drop) -> str: 1150 if expression.args["kind"] == "VIEW": 1151 expression.this.set("catalog", None) 1152 return super().drop_sql(expression) 1153 1154 def declare_sql(self, expression: exp.Declare) -> str: 1155 return f"DECLARE {self.expressions(expression, flat=True)}" 1156 1157 def declareitem_sql(self, expression: exp.DeclareItem) -> str: 1158 variable = self.sql(expression, "this") 1159 default = self.sql(expression, "default") 1160 default = f" = {default}" if default else "" 1161 1162 kind = self.sql(expression, "kind") 1163 if isinstance(expression.args.get("kind"), exp.Schema): 1164 kind = f"TABLE {kind}" 1165 1166 return f"{variable} AS {kind}{default}" 1167 1168 def options_modifier(self, expression: exp.Expression) -> str: 1169 options = self.expressions(expression, key="options") 1170 return f" OPTION{self.wrap(options)}" if options else ""
Specifies the strategy according to which identifiers should be normalized.
Whether the base comes first in the LOG
function.
Possible values: True
, False
, None
(two arguments are not supported by LOG
)
Whether the behavior of a / b
depends on the types of a
and b
.
False means a / b
is always float division.
True means a / b
is integer division if both a
and b
are integers.
A NULL
arg in CONCAT
yields NULL
by default, but in some dialects it yields an empty string.
Associates this dialect's time formats with their equivalent Python strftime
formats.
Inherited Members
- sqlglot.dialects.dialect.Dialect
- Dialect
- INDEX_OFFSET
- WEEK_OFFSET
- UNNEST_COLUMN_ONLY
- ALIAS_POST_TABLESAMPLE
- TABLESAMPLE_SIZE_IS_PERCENT
- IDENTIFIERS_CAN_START_WITH_DIGIT
- DPIPE_IS_STRING_CONCAT
- STRICT_STRING_CONCAT
- SUPPORTS_USER_DEFINED_TYPES
- COPY_PARAMS_ARE_CSV
- NORMALIZE_FUNCTIONS
- NULL_ORDERING
- SAFE_DIVISION
- HEX_LOWERCASE
- DATE_FORMAT
- DATEINT_FORMAT
- FORMAT_MAPPING
- UNESCAPED_SEQUENCES
- PSEUDOCOLUMNS
- PREFER_CTE_ALIAS_COLUMN
- FORCE_EARLY_ALIAS_REF_EXPANSION
- EXPAND_ALIAS_REFS_EARLY_ONLY_IN_GROUP_BY
- SUPPORTS_ORDER_BY_ALL
- DATE_PART_MAPPING
- TYPE_TO_EXPRESSIONS
- ANNOTATORS
- get_or_raise
- format_time
- settings
- normalize_identifier
- case_sensitive
- can_identify
- quote_identifier
- to_json_path
- parse
- parse_into
- generate
- transpile
- tokenize
- tokenizer
- jsonpath_tokenizer
- parser
- generator
455 class Tokenizer(tokens.Tokenizer): 456 IDENTIFIERS = [("[", "]"), '"'] 457 QUOTES = ["'", '"'] 458 HEX_STRINGS = [("0x", ""), ("0X", "")] 459 VAR_SINGLE_TOKENS = {"@", "$", "#"} 460 461 KEYWORDS = { 462 **tokens.Tokenizer.KEYWORDS, 463 "CLUSTERED INDEX": TokenType.INDEX, 464 "DATETIME2": TokenType.DATETIME, 465 "DATETIMEOFFSET": TokenType.TIMESTAMPTZ, 466 "DECLARE": TokenType.DECLARE, 467 "EXEC": TokenType.COMMAND, 468 "FOR SYSTEM_TIME": TokenType.TIMESTAMP_SNAPSHOT, 469 "IMAGE": TokenType.IMAGE, 470 "MONEY": TokenType.MONEY, 471 "NONCLUSTERED INDEX": TokenType.INDEX, 472 "NTEXT": TokenType.TEXT, 473 "OPTION": TokenType.OPTION, 474 "OUTPUT": TokenType.RETURNING, 475 "PRINT": TokenType.COMMAND, 476 "PROC": TokenType.PROCEDURE, 477 "REAL": TokenType.FLOAT, 478 "ROWVERSION": TokenType.ROWVERSION, 479 "SMALLDATETIME": TokenType.DATETIME, 480 "SMALLMONEY": TokenType.SMALLMONEY, 481 "SQL_VARIANT": TokenType.VARIANT, 482 "SYSTEM_USER": TokenType.CURRENT_USER, 483 "TOP": TokenType.TOP, 484 "TIMESTAMP": TokenType.ROWVERSION, 485 "TINYINT": TokenType.UTINYINT, 486 "UNIQUEIDENTIFIER": TokenType.UNIQUEIDENTIFIER, 487 "UPDATE STATISTICS": TokenType.COMMAND, 488 "XML": TokenType.XML, 489 } 490 KEYWORDS.pop("/*+") 491 492 COMMANDS = {*tokens.Tokenizer.COMMANDS, TokenType.END}
Inherited Members
- sqlglot.tokens.Tokenizer
- Tokenizer
- SINGLE_TOKENS
- BIT_STRINGS
- BYTE_STRINGS
- RAW_STRINGS
- HEREDOC_STRINGS
- UNICODE_STRINGS
- IDENTIFIER_ESCAPES
- STRING_ESCAPES
- HEREDOC_TAG_IS_IDENTIFIER
- HEREDOC_STRING_ALTERNATIVE
- STRING_ESCAPES_ALLOWED_IN_RAW_STRINGS
- WHITE_SPACE
- COMMAND_PREFIX_TOKENS
- NUMERIC_LITERALS
- COMMENTS
- dialect
- reset
- tokenize
- tokenize_rs
- size
- sql
- tokens
494 class Parser(parser.Parser): 495 SET_REQUIRES_ASSIGNMENT_DELIMITER = False 496 LOG_DEFAULTS_TO_LN = True 497 ALTER_TABLE_ADD_REQUIRED_FOR_EACH_COLUMN = False 498 STRING_ALIASES = True 499 NO_PAREN_IF_COMMANDS = False 500 501 QUERY_MODIFIER_PARSERS = { 502 **parser.Parser.QUERY_MODIFIER_PARSERS, 503 TokenType.OPTION: lambda self: ("options", self._parse_options()), 504 } 505 506 FUNCTIONS = { 507 **parser.Parser.FUNCTIONS, 508 "CHARINDEX": lambda args: exp.StrPosition( 509 this=seq_get(args, 1), 510 substr=seq_get(args, 0), 511 position=seq_get(args, 2), 512 ), 513 "DATEADD": build_date_delta(exp.DateAdd, unit_mapping=DATE_DELTA_INTERVAL), 514 "DATEDIFF": _build_date_delta(exp.DateDiff, unit_mapping=DATE_DELTA_INTERVAL), 515 "DATENAME": _build_formatted_time(exp.TimeToStr, full_format_mapping=True), 516 "DATEPART": _build_formatted_time(exp.TimeToStr), 517 "DATETIMEFROMPARTS": _build_datetimefromparts, 518 "EOMONTH": _build_eomonth, 519 "FORMAT": _build_format, 520 "GETDATE": exp.CurrentTimestamp.from_arg_list, 521 "HASHBYTES": _build_hashbytes, 522 "ISNULL": exp.Coalesce.from_arg_list, 523 "JSON_QUERY": _build_json_query, 524 "JSON_VALUE": parser.build_extract_json_with_path(exp.JSONExtractScalar), 525 "LEN": _build_with_arg_as_text(exp.Length), 526 "LEFT": _build_with_arg_as_text(exp.Left), 527 "RIGHT": _build_with_arg_as_text(exp.Right), 528 "REPLICATE": exp.Repeat.from_arg_list, 529 "SQUARE": lambda args: exp.Pow(this=seq_get(args, 0), expression=exp.Literal.number(2)), 530 "SYSDATETIME": exp.CurrentTimestamp.from_arg_list, 531 "SUSER_NAME": exp.CurrentUser.from_arg_list, 532 "SUSER_SNAME": exp.CurrentUser.from_arg_list, 533 "SYSTEM_USER": exp.CurrentUser.from_arg_list, 534 "TIMEFROMPARTS": _build_timefromparts, 535 } 536 537 JOIN_HINTS = {"LOOP", "HASH", "MERGE", "REMOTE"} 538 539 RETURNS_TABLE_TOKENS = parser.Parser.ID_VAR_TOKENS - { 540 TokenType.TABLE, 541 *parser.Parser.TYPE_TOKENS, 542 } 543 544 STATEMENT_PARSERS = { 545 **parser.Parser.STATEMENT_PARSERS, 546 TokenType.DECLARE: lambda self: self._parse_declare(), 547 } 548 549 RANGE_PARSERS = { 550 **parser.Parser.RANGE_PARSERS, 551 TokenType.DCOLON: lambda self, this: self.expression( 552 exp.ScopeResolution, 553 this=this, 554 expression=self._parse_function() or self._parse_var(any_token=True), 555 ), 556 } 557 558 # The DCOLON (::) operator serves as a scope resolution (exp.ScopeResolution) operator in T-SQL 559 COLUMN_OPERATORS = { 560 **parser.Parser.COLUMN_OPERATORS, 561 TokenType.DCOLON: lambda self, this, to: self.expression(exp.Cast, this=this, to=to) 562 if isinstance(to, exp.DataType) and to.this != exp.DataType.Type.USERDEFINED 563 else self.expression(exp.ScopeResolution, this=this, expression=to), 564 } 565 566 def _parse_dcolon(self) -> t.Optional[exp.Expression]: 567 # We want to use _parse_types() if the first token after :: is a known type, 568 # otherwise we could parse something like x::varchar(max) into a function 569 if self._match_set(self.TYPE_TOKENS, advance=False): 570 return self._parse_types() 571 572 return self._parse_function() or self._parse_types() 573 574 def _parse_options(self) -> t.Optional[t.List[exp.Expression]]: 575 if not self._match(TokenType.OPTION): 576 return None 577 578 def _parse_option() -> t.Optional[exp.Expression]: 579 option = self._parse_var_from_options(OPTIONS) 580 if not option: 581 return None 582 583 self._match(TokenType.EQ) 584 return self.expression( 585 exp.QueryOption, this=option, expression=self._parse_primary_or_var() 586 ) 587 588 return self._parse_wrapped_csv(_parse_option) 589 590 def _parse_projections(self) -> t.List[exp.Expression]: 591 """ 592 T-SQL supports the syntax alias = expression in the SELECT's projection list, 593 so we transform all parsed Selects to convert their EQ projections into Aliases. 594 595 See: https://learn.microsoft.com/en-us/sql/t-sql/queries/select-clause-transact-sql?view=sql-server-ver16#syntax 596 """ 597 return [ 598 ( 599 exp.alias_(projection.expression, projection.this.this, copy=False) 600 if isinstance(projection, exp.EQ) and isinstance(projection.this, exp.Column) 601 else projection 602 ) 603 for projection in super()._parse_projections() 604 ] 605 606 def _parse_commit_or_rollback(self) -> exp.Commit | exp.Rollback: 607 """Applies to SQL Server and Azure SQL Database 608 COMMIT [ { TRAN | TRANSACTION } 609 [ transaction_name | @tran_name_variable ] ] 610 [ WITH ( DELAYED_DURABILITY = { OFF | ON } ) ] 611 612 ROLLBACK { TRAN | TRANSACTION } 613 [ transaction_name | @tran_name_variable 614 | savepoint_name | @savepoint_variable ] 615 """ 616 rollback = self._prev.token_type == TokenType.ROLLBACK 617 618 self._match_texts(("TRAN", "TRANSACTION")) 619 this = self._parse_id_var() 620 621 if rollback: 622 return self.expression(exp.Rollback, this=this) 623 624 durability = None 625 if self._match_pair(TokenType.WITH, TokenType.L_PAREN): 626 self._match_text_seq("DELAYED_DURABILITY") 627 self._match(TokenType.EQ) 628 629 if self._match_text_seq("OFF"): 630 durability = False 631 else: 632 self._match(TokenType.ON) 633 durability = True 634 635 self._match_r_paren() 636 637 return self.expression(exp.Commit, this=this, durability=durability) 638 639 def _parse_transaction(self) -> exp.Transaction | exp.Command: 640 """Applies to SQL Server and Azure SQL Database 641 BEGIN { TRAN | TRANSACTION } 642 [ { transaction_name | @tran_name_variable } 643 [ WITH MARK [ 'description' ] ] 644 ] 645 """ 646 if self._match_texts(("TRAN", "TRANSACTION")): 647 transaction = self.expression(exp.Transaction, this=self._parse_id_var()) 648 if self._match_text_seq("WITH", "MARK"): 649 transaction.set("mark", self._parse_string()) 650 651 return transaction 652 653 return self._parse_as_command(self._prev) 654 655 def _parse_returns(self) -> exp.ReturnsProperty: 656 table = self._parse_id_var(any_token=False, tokens=self.RETURNS_TABLE_TOKENS) 657 returns = super()._parse_returns() 658 returns.set("table", table) 659 return returns 660 661 def _parse_convert( 662 self, strict: bool, safe: t.Optional[bool] = None 663 ) -> t.Optional[exp.Expression]: 664 this = self._parse_types() 665 self._match(TokenType.COMMA) 666 args = [this, *self._parse_csv(self._parse_assignment)] 667 convert = exp.Convert.from_arg_list(args) 668 convert.set("safe", safe) 669 convert.set("strict", strict) 670 return convert 671 672 def _parse_user_defined_function( 673 self, kind: t.Optional[TokenType] = None 674 ) -> t.Optional[exp.Expression]: 675 this = super()._parse_user_defined_function(kind=kind) 676 677 if ( 678 kind == TokenType.FUNCTION 679 or isinstance(this, exp.UserDefinedFunction) 680 or self._match(TokenType.ALIAS, advance=False) 681 ): 682 return this 683 684 expressions = self._parse_csv(self._parse_function_parameter) 685 return self.expression(exp.UserDefinedFunction, this=this, expressions=expressions) 686 687 def _parse_id_var( 688 self, 689 any_token: bool = True, 690 tokens: t.Optional[t.Collection[TokenType]] = None, 691 ) -> t.Optional[exp.Expression]: 692 is_temporary = self._match(TokenType.HASH) 693 is_global = is_temporary and self._match(TokenType.HASH) 694 695 this = super()._parse_id_var(any_token=any_token, tokens=tokens) 696 if this: 697 if is_global: 698 this.set("global", True) 699 elif is_temporary: 700 this.set("temporary", True) 701 702 return this 703 704 def _parse_create(self) -> exp.Create | exp.Command: 705 create = super()._parse_create() 706 707 if isinstance(create, exp.Create): 708 table = create.this.this if isinstance(create.this, exp.Schema) else create.this 709 if isinstance(table, exp.Table) and table.this.args.get("temporary"): 710 if not create.args.get("properties"): 711 create.set("properties", exp.Properties(expressions=[])) 712 713 create.args["properties"].append("expressions", exp.TemporaryProperty()) 714 715 return create 716 717 def _parse_if(self) -> t.Optional[exp.Expression]: 718 index = self._index 719 720 if self._match_text_seq("OBJECT_ID"): 721 self._parse_wrapped_csv(self._parse_string) 722 if self._match_text_seq("IS", "NOT", "NULL") and self._match(TokenType.DROP): 723 return self._parse_drop(exists=True) 724 self._retreat(index) 725 726 return super()._parse_if() 727 728 def _parse_unique(self) -> exp.UniqueColumnConstraint: 729 if self._match_texts(("CLUSTERED", "NONCLUSTERED")): 730 this = self.CONSTRAINT_PARSERS[self._prev.text.upper()](self) 731 else: 732 this = self._parse_schema(self._parse_id_var(any_token=False)) 733 734 return self.expression(exp.UniqueColumnConstraint, this=this) 735 736 def _parse_partition(self) -> t.Optional[exp.Partition]: 737 if not self._match_text_seq("WITH", "(", "PARTITIONS"): 738 return None 739 740 def parse_range(): 741 low = self._parse_bitwise() 742 high = self._parse_bitwise() if self._match_text_seq("TO") else None 743 744 return ( 745 self.expression(exp.PartitionRange, this=low, expression=high) if high else low 746 ) 747 748 partition = self.expression( 749 exp.Partition, expressions=self._parse_wrapped_csv(parse_range) 750 ) 751 752 self._match_r_paren() 753 754 return partition 755 756 def _parse_declare(self) -> exp.Declare | exp.Command: 757 index = self._index 758 expressions = self._try_parse(partial(self._parse_csv, self._parse_declareitem)) 759 760 if not expressions or self._curr: 761 self._retreat(index) 762 return self._parse_as_command(self._prev) 763 764 return self.expression(exp.Declare, expressions=expressions) 765 766 def _parse_declareitem(self) -> t.Optional[exp.DeclareItem]: 767 var = self._parse_id_var() 768 if not var: 769 return None 770 771 value = None 772 self._match(TokenType.ALIAS) 773 if self._match(TokenType.TABLE): 774 data_type = self._parse_schema() 775 else: 776 data_type = self._parse_types() 777 if self._match(TokenType.EQ): 778 value = self._parse_bitwise() 779 780 return self.expression(exp.DeclareItem, this=var, kind=data_type, default=value)
Parser consumes a list of tokens produced by the Tokenizer and produces a parsed syntax tree.
Arguments:
- error_level: The desired error level. Default: ErrorLevel.IMMEDIATE
- error_message_context: The amount of context to capture from a query string when displaying the error message (in number of characters). Default: 100
- max_errors: Maximum number of error messages to include in a raised ParseError. This is only relevant if error_level is ErrorLevel.RAISE. Default: 3
Inherited Members
- sqlglot.parser.Parser
- Parser
- NO_PAREN_FUNCTIONS
- STRUCT_TYPE_TOKENS
- NESTED_TYPE_TOKENS
- ENUM_TYPE_TOKENS
- AGGREGATE_TYPE_TOKENS
- TYPE_TOKENS
- SIGNED_TO_UNSIGNED_TYPE_TOKEN
- SUBQUERY_PREDICATES
- RESERVED_TOKENS
- DB_CREATABLES
- CREATABLES
- INTERVAL_VARS
- ALIAS_TOKENS
- ARRAY_CONSTRUCTORS
- COMMENT_TABLE_ALIAS_TOKENS
- UPDATE_ALIAS_TOKENS
- TRIM_TYPES
- FUNC_TOKENS
- CONJUNCTION
- ASSIGNMENT
- DISJUNCTION
- EQUALITY
- COMPARISON
- BITWISE
- TERM
- FACTOR
- EXPONENT
- TIMES
- TIMESTAMPS
- SET_OPERATIONS
- JOIN_METHODS
- JOIN_SIDES
- JOIN_KINDS
- LAMBDAS
- EXPRESSION_PARSERS
- UNARY_PARSERS
- STRING_PARSERS
- NUMERIC_PARSERS
- PRIMARY_PARSERS
- PLACEHOLDER_PARSERS
- PROPERTY_PARSERS
- CONSTRAINT_PARSERS
- ALTER_PARSERS
- ALTER_ALTER_PARSERS
- SCHEMA_UNNAMED_CONSTRAINTS
- NO_PAREN_FUNCTION_PARSERS
- INVALID_FUNC_NAME_TOKENS
- FUNCTIONS_WITH_ALIASED_ARGS
- KEY_VALUE_DEFINITIONS
- FUNCTION_PARSERS
- SET_PARSERS
- SHOW_PARSERS
- TYPE_LITERAL_PARSERS
- TYPE_CONVERTERS
- DDL_SELECT_TOKENS
- PRE_VOLATILE_TOKENS
- TRANSACTION_KIND
- TRANSACTION_CHARACTERISTICS
- CONFLICT_ACTIONS
- CREATE_SEQUENCE
- ISOLATED_LOADING_OPTIONS
- USABLES
- CAST_ACTIONS
- SCHEMA_BINDING_OPTIONS
- KEY_CONSTRAINT_OPTIONS
- INSERT_ALTERNATIVES
- CLONE_KEYWORDS
- HISTORICAL_DATA_PREFIX
- HISTORICAL_DATA_KIND
- OPCLASS_FOLLOW_KEYWORDS
- OPTYPE_FOLLOW_TOKENS
- TABLE_INDEX_HINT_TOKENS
- VIEW_ATTRIBUTES
- WINDOW_ALIAS_TOKENS
- WINDOW_BEFORE_PAREN_TOKENS
- WINDOW_SIDES
- JSON_KEY_VALUE_SEPARATOR_TOKENS
- FETCH_TOKENS
- ADD_CONSTRAINT_TOKENS
- DISTINCT_TOKENS
- NULL_TOKENS
- UNNEST_OFFSET_ALIAS_TOKENS
- SELECT_START_TOKENS
- COPY_INTO_VARLEN_OPTIONS
- STRICT_CAST
- PREFIXED_PIVOT_COLUMNS
- IDENTIFY_PIVOT_STRINGS
- TABLESAMPLE_CSV
- DEFAULT_SAMPLING_METHOD
- TRIM_PATTERN_FIRST
- MODIFIERS_ATTACHED_TO_SET_OP
- SET_OP_MODIFIERS
- JSON_ARROWS_REQUIRE_JSON_TYPE
- COLON_IS_VARIANT_EXTRACT
- VALUES_FOLLOWED_BY_PAREN
- SUPPORTS_IMPLICIT_UNNEST
- INTERVAL_SPANS
- SUPPORTS_PARTITION_SELECTION
- error_level
- error_message_context
- max_errors
- dialect
- reset
- parse
- parse_into
- check_errors
- raise_error
- expression
- validate_expression
- errors
- sql
782 class Generator(generator.Generator): 783 LIMIT_IS_TOP = True 784 QUERY_HINTS = False 785 RETURNING_END = False 786 NVL2_SUPPORTED = False 787 ALTER_TABLE_INCLUDE_COLUMN_KEYWORD = False 788 LIMIT_FETCH = "FETCH" 789 COMPUTED_COLUMN_WITH_TYPE = False 790 CTE_RECURSIVE_KEYWORD_REQUIRED = False 791 ENSURE_BOOLS = True 792 NULL_ORDERING_SUPPORTED = None 793 SUPPORTS_SINGLE_ARG_CONCAT = False 794 TABLESAMPLE_SEED_KEYWORD = "REPEATABLE" 795 SUPPORTS_SELECT_INTO = True 796 JSON_PATH_BRACKETED_KEY_SUPPORTED = False 797 SUPPORTS_TO_NUMBER = False 798 SET_OP_MODIFIERS = False 799 COPY_PARAMS_EQ_REQUIRED = True 800 PARSE_JSON_NAME = None 801 802 EXPRESSIONS_WITHOUT_NESTED_CTES = { 803 exp.Create, 804 exp.Delete, 805 exp.Insert, 806 exp.Intersect, 807 exp.Except, 808 exp.Merge, 809 exp.Select, 810 exp.Subquery, 811 exp.Union, 812 exp.Update, 813 } 814 815 SUPPORTED_JSON_PATH_PARTS = { 816 exp.JSONPathKey, 817 exp.JSONPathRoot, 818 exp.JSONPathSubscript, 819 } 820 821 TYPE_MAPPING = { 822 **generator.Generator.TYPE_MAPPING, 823 exp.DataType.Type.BOOLEAN: "BIT", 824 exp.DataType.Type.DECIMAL: "NUMERIC", 825 exp.DataType.Type.DATETIME: "DATETIME2", 826 exp.DataType.Type.DOUBLE: "FLOAT", 827 exp.DataType.Type.INT: "INTEGER", 828 exp.DataType.Type.ROWVERSION: "ROWVERSION", 829 exp.DataType.Type.TEXT: "VARCHAR(MAX)", 830 exp.DataType.Type.TIMESTAMP: "DATETIME2", 831 exp.DataType.Type.TIMESTAMPTZ: "DATETIMEOFFSET", 832 exp.DataType.Type.UTINYINT: "TINYINT", 833 exp.DataType.Type.VARIANT: "SQL_VARIANT", 834 } 835 836 TYPE_MAPPING.pop(exp.DataType.Type.NCHAR) 837 TYPE_MAPPING.pop(exp.DataType.Type.NVARCHAR) 838 839 TRANSFORMS = { 840 **generator.Generator.TRANSFORMS, 841 exp.AnyValue: any_value_to_max_sql, 842 exp.ArrayToString: rename_func("STRING_AGG"), 843 exp.AutoIncrementColumnConstraint: lambda *_: "IDENTITY", 844 exp.DateAdd: date_delta_sql("DATEADD"), 845 exp.DateDiff: date_delta_sql("DATEDIFF"), 846 exp.CTE: transforms.preprocess([qualify_derived_table_outputs]), 847 exp.CurrentDate: rename_func("GETDATE"), 848 exp.CurrentTimestamp: rename_func("GETDATE"), 849 exp.DateStrToDate: datestrtodate_sql, 850 exp.Extract: rename_func("DATEPART"), 851 exp.GeneratedAsIdentityColumnConstraint: generatedasidentitycolumnconstraint_sql, 852 exp.GroupConcat: _string_agg_sql, 853 exp.If: rename_func("IIF"), 854 exp.JSONExtract: _json_extract_sql, 855 exp.JSONExtractScalar: _json_extract_sql, 856 exp.LastDay: lambda self, e: self.func("EOMONTH", e.this), 857 exp.Max: max_or_greatest, 858 exp.MD5: lambda self, e: self.func("HASHBYTES", exp.Literal.string("MD5"), e.this), 859 exp.Min: min_or_least, 860 exp.NumberToStr: _format_sql, 861 exp.Repeat: rename_func("REPLICATE"), 862 exp.Select: transforms.preprocess( 863 [ 864 transforms.eliminate_distinct_on, 865 transforms.eliminate_semi_and_anti_joins, 866 transforms.eliminate_qualify, 867 ] 868 ), 869 exp.Stddev: rename_func("STDEV"), 870 exp.StrPosition: lambda self, e: self.func( 871 "CHARINDEX", e.args.get("substr"), e.this, e.args.get("position") 872 ), 873 exp.Subquery: transforms.preprocess([qualify_derived_table_outputs]), 874 exp.SHA: lambda self, e: self.func("HASHBYTES", exp.Literal.string("SHA1"), e.this), 875 exp.SHA2: lambda self, e: self.func( 876 "HASHBYTES", exp.Literal.string(f"SHA2_{e.args.get('length', 256)}"), e.this 877 ), 878 exp.TemporaryProperty: lambda self, e: "", 879 exp.TimeStrToTime: lambda self, e: self.sql( 880 exp.cast(e.this, exp.DataType.Type.DATETIME) 881 ), 882 exp.TimeToStr: _format_sql, 883 exp.Trim: trim_sql, 884 exp.TsOrDsAdd: date_delta_sql("DATEADD", cast=True), 885 exp.TsOrDsDiff: date_delta_sql("DATEDIFF"), 886 } 887 888 TRANSFORMS.pop(exp.ReturnsProperty) 889 890 PROPERTIES_LOCATION = { 891 **generator.Generator.PROPERTIES_LOCATION, 892 exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED, 893 } 894 895 def scope_resolution(self, rhs: str, scope_name: str) -> str: 896 return f"{scope_name}::{rhs}" 897 898 def select_sql(self, expression: exp.Select) -> str: 899 if expression.args.get("offset"): 900 if not expression.args.get("order"): 901 # ORDER BY is required in order to use OFFSET in a query, so we use 902 # a noop order by, since we don't really care about the order. 903 # See: https://www.microsoftpressstore.com/articles/article.aspx?p=2314819 904 expression.order_by(exp.select(exp.null()).subquery(), copy=False) 905 906 limit = expression.args.get("limit") 907 if isinstance(limit, exp.Limit): 908 # TOP and OFFSET can't be combined, we need use FETCH instead of TOP 909 # we replace here because otherwise TOP would be generated in select_sql 910 limit.replace(exp.Fetch(direction="FIRST", count=limit.expression)) 911 912 return super().select_sql(expression) 913 914 def convert_sql(self, expression: exp.Convert) -> str: 915 name = "TRY_CONVERT" if expression.args.get("safe") else "CONVERT" 916 return self.func( 917 name, expression.this, expression.expression, expression.args.get("style") 918 ) 919 920 def queryoption_sql(self, expression: exp.QueryOption) -> str: 921 option = self.sql(expression, "this") 922 value = self.sql(expression, "expression") 923 if value: 924 optional_equal_sign = "= " if option in OPTIONS_THAT_REQUIRE_EQUAL else "" 925 return f"{option} {optional_equal_sign}{value}" 926 return option 927 928 def lateral_op(self, expression: exp.Lateral) -> str: 929 cross_apply = expression.args.get("cross_apply") 930 if cross_apply is True: 931 return "CROSS APPLY" 932 if cross_apply is False: 933 return "OUTER APPLY" 934 935 # TODO: perhaps we can check if the parent is a Join and transpile it appropriately 936 self.unsupported("LATERAL clause is not supported.") 937 return "LATERAL" 938 939 def timefromparts_sql(self, expression: exp.TimeFromParts) -> str: 940 nano = expression.args.get("nano") 941 if nano is not None: 942 nano.pop() 943 self.unsupported("Specifying nanoseconds is not supported in TIMEFROMPARTS.") 944 945 if expression.args.get("fractions") is None: 946 expression.set("fractions", exp.Literal.number(0)) 947 if expression.args.get("precision") is None: 948 expression.set("precision", exp.Literal.number(0)) 949 950 return rename_func("TIMEFROMPARTS")(self, expression) 951 952 def timestampfromparts_sql(self, expression: exp.TimestampFromParts) -> str: 953 zone = expression.args.get("zone") 954 if zone is not None: 955 zone.pop() 956 self.unsupported("Time zone is not supported in DATETIMEFROMPARTS.") 957 958 nano = expression.args.get("nano") 959 if nano is not None: 960 nano.pop() 961 self.unsupported("Specifying nanoseconds is not supported in DATETIMEFROMPARTS.") 962 963 if expression.args.get("milli") is None: 964 expression.set("milli", exp.Literal.number(0)) 965 966 return rename_func("DATETIMEFROMPARTS")(self, expression) 967 968 def setitem_sql(self, expression: exp.SetItem) -> str: 969 this = expression.this 970 if isinstance(this, exp.EQ) and not isinstance(this.left, exp.Parameter): 971 # T-SQL does not use '=' in SET command, except when the LHS is a variable. 972 return f"{self.sql(this.left)} {self.sql(this.right)}" 973 974 return super().setitem_sql(expression) 975 976 def boolean_sql(self, expression: exp.Boolean) -> str: 977 if type(expression.parent) in BIT_TYPES: 978 return "1" if expression.this else "0" 979 980 return "(1 = 1)" if expression.this else "(1 = 0)" 981 982 def is_sql(self, expression: exp.Is) -> str: 983 if isinstance(expression.expression, exp.Boolean): 984 return self.binary(expression, "=") 985 return self.binary(expression, "IS") 986 987 def createable_sql(self, expression: exp.Create, locations: t.DefaultDict) -> str: 988 sql = self.sql(expression, "this") 989 properties = expression.args.get("properties") 990 991 if sql[:1] != "#" and any( 992 isinstance(prop, exp.TemporaryProperty) 993 for prop in (properties.expressions if properties else []) 994 ): 995 sql = f"[#{sql[1:]}" if sql.startswith("[") else f"#{sql}" 996 997 return sql 998 999 def create_sql(self, expression: exp.Create) -> str: 1000 kind = expression.kind 1001 exists = expression.args.pop("exists", None) 1002 1003 like_property = expression.find(exp.LikeProperty) 1004 if like_property: 1005 ctas_expression = like_property.this 1006 else: 1007 ctas_expression = expression.expression 1008 1009 if kind == "VIEW": 1010 expression.this.set("catalog", None) 1011 with_ = expression.args.get("with") 1012 if ctas_expression and with_: 1013 # We've already preprocessed the Create expression to bubble up any nested CTEs, 1014 # but CREATE VIEW actually requires the WITH clause to come after it so we need 1015 # to amend the AST by moving the CTEs to the CREATE VIEW statement's query. 1016 ctas_expression.set("with", with_.pop()) 1017 1018 sql = super().create_sql(expression) 1019 1020 table = expression.find(exp.Table) 1021 1022 # Convert CTAS statement to SELECT .. INTO .. 1023 if kind == "TABLE" and ctas_expression: 1024 if isinstance(ctas_expression, exp.UNWRAPPED_QUERIES): 1025 ctas_expression = ctas_expression.subquery() 1026 1027 select_into = exp.select("*").from_(exp.alias_(ctas_expression, "temp", table=True)) 1028 select_into.set("into", exp.Into(this=table)) 1029 1030 if like_property: 1031 select_into.limit(0, copy=False) 1032 1033 sql = self.sql(select_into) 1034 1035 if exists: 1036 identifier = self.sql(exp.Literal.string(exp.table_name(table) if table else "")) 1037 sql = self.sql(exp.Literal.string(sql)) 1038 if kind == "SCHEMA": 1039 sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.schemata WHERE schema_name = {identifier}) EXEC({sql})""" 1040 elif kind == "TABLE": 1041 assert table 1042 where = exp.and_( 1043 exp.column("table_name").eq(table.name), 1044 exp.column("table_schema").eq(table.db) if table.db else None, 1045 exp.column("table_catalog").eq(table.catalog) if table.catalog else None, 1046 ) 1047 sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.tables WHERE {where}) EXEC({sql})""" 1048 elif kind == "INDEX": 1049 index = self.sql(exp.Literal.string(expression.this.text("this"))) 1050 sql = f"""IF NOT EXISTS (SELECT * FROM sys.indexes WHERE object_id = object_id({identifier}) AND name = {index}) EXEC({sql})""" 1051 elif expression.args.get("replace"): 1052 sql = sql.replace("CREATE OR REPLACE ", "CREATE OR ALTER ", 1) 1053 1054 return self.prepend_ctes(expression, sql) 1055 1056 def offset_sql(self, expression: exp.Offset) -> str: 1057 return f"{super().offset_sql(expression)} ROWS" 1058 1059 def version_sql(self, expression: exp.Version) -> str: 1060 name = "SYSTEM_TIME" if expression.name == "TIMESTAMP" else expression.name 1061 this = f"FOR {name}" 1062 expr = expression.expression 1063 kind = expression.text("kind") 1064 if kind in ("FROM", "BETWEEN"): 1065 args = expr.expressions 1066 sep = "TO" if kind == "FROM" else "AND" 1067 expr_sql = f"{self.sql(seq_get(args, 0))} {sep} {self.sql(seq_get(args, 1))}" 1068 else: 1069 expr_sql = self.sql(expr) 1070 1071 expr_sql = f" {expr_sql}" if expr_sql else "" 1072 return f"{this} {kind}{expr_sql}" 1073 1074 def returnsproperty_sql(self, expression: exp.ReturnsProperty) -> str: 1075 table = expression.args.get("table") 1076 table = f"{table} " if table else "" 1077 return f"RETURNS {table}{self.sql(expression, 'this')}" 1078 1079 def returning_sql(self, expression: exp.Returning) -> str: 1080 into = self.sql(expression, "into") 1081 into = self.seg(f"INTO {into}") if into else "" 1082 return f"{self.seg('OUTPUT')} {self.expressions(expression, flat=True)}{into}" 1083 1084 def transaction_sql(self, expression: exp.Transaction) -> str: 1085 this = self.sql(expression, "this") 1086 this = f" {this}" if this else "" 1087 mark = self.sql(expression, "mark") 1088 mark = f" WITH MARK {mark}" if mark else "" 1089 return f"BEGIN TRANSACTION{this}{mark}" 1090 1091 def commit_sql(self, expression: exp.Commit) -> str: 1092 this = self.sql(expression, "this") 1093 this = f" {this}" if this else "" 1094 durability = expression.args.get("durability") 1095 durability = ( 1096 f" WITH (DELAYED_DURABILITY = {'ON' if durability else 'OFF'})" 1097 if durability is not None 1098 else "" 1099 ) 1100 return f"COMMIT TRANSACTION{this}{durability}" 1101 1102 def rollback_sql(self, expression: exp.Rollback) -> str: 1103 this = self.sql(expression, "this") 1104 this = f" {this}" if this else "" 1105 return f"ROLLBACK TRANSACTION{this}" 1106 1107 def identifier_sql(self, expression: exp.Identifier) -> str: 1108 identifier = super().identifier_sql(expression) 1109 1110 if expression.args.get("global"): 1111 identifier = f"##{identifier}" 1112 elif expression.args.get("temporary"): 1113 identifier = f"#{identifier}" 1114 1115 return identifier 1116 1117 def constraint_sql(self, expression: exp.Constraint) -> str: 1118 this = self.sql(expression, "this") 1119 expressions = self.expressions(expression, flat=True, sep=" ") 1120 return f"CONSTRAINT {this} {expressions}" 1121 1122 def length_sql(self, expression: exp.Length) -> str: 1123 return self._uncast_text(expression, "LEN") 1124 1125 def right_sql(self, expression: exp.Right) -> str: 1126 return self._uncast_text(expression, "RIGHT") 1127 1128 def left_sql(self, expression: exp.Left) -> str: 1129 return self._uncast_text(expression, "LEFT") 1130 1131 def _uncast_text(self, expression: exp.Expression, name: str) -> str: 1132 this = expression.this 1133 if isinstance(this, exp.Cast) and this.is_type(exp.DataType.Type.TEXT): 1134 this_sql = self.sql(this, "this") 1135 else: 1136 this_sql = self.sql(this) 1137 expression_sql = self.sql(expression, "expression") 1138 return self.func(name, this_sql, expression_sql if expression_sql else None) 1139 1140 def partition_sql(self, expression: exp.Partition) -> str: 1141 return f"WITH (PARTITIONS({self.expressions(expression, flat=True)}))" 1142 1143 def altertable_sql(self, expression: exp.AlterTable) -> str: 1144 action = seq_get(expression.args.get("actions") or [], 0) 1145 if isinstance(action, exp.RenameTable): 1146 return f"EXEC sp_rename '{self.sql(expression.this)}', '{action.this.name}'" 1147 return super().altertable_sql(expression) 1148 1149 def drop_sql(self, expression: exp.Drop) -> str: 1150 if expression.args["kind"] == "VIEW": 1151 expression.this.set("catalog", None) 1152 return super().drop_sql(expression) 1153 1154 def declare_sql(self, expression: exp.Declare) -> str: 1155 return f"DECLARE {self.expressions(expression, flat=True)}" 1156 1157 def declareitem_sql(self, expression: exp.DeclareItem) -> str: 1158 variable = self.sql(expression, "this") 1159 default = self.sql(expression, "default") 1160 default = f" = {default}" if default else "" 1161 1162 kind = self.sql(expression, "kind") 1163 if isinstance(expression.args.get("kind"), exp.Schema): 1164 kind = f"TABLE {kind}" 1165 1166 return f"{variable} AS {kind}{default}" 1167 1168 def options_modifier(self, expression: exp.Expression) -> str: 1169 options = self.expressions(expression, key="options") 1170 return f" OPTION{self.wrap(options)}" if options else ""
Generator converts a given syntax tree to the corresponding SQL string.
Arguments:
- pretty: Whether to format the produced SQL string. Default: False.
- identify: Determines when an identifier should be quoted. Possible values are: False (default): Never quote, except in cases where it's mandatory by the dialect. True or 'always': Always quote. 'safe': Only quote identifiers that are case insensitive.
- normalize: Whether to normalize identifiers to lowercase. Default: False.
- pad: The pad size in a formatted string. For example, this affects the indentation of a projection in a query, relative to its nesting level. Default: 2.
- indent: The indentation size in a formatted string. For example, this affects the
indentation of subqueries and filters under a
WHERE
clause. Default: 2. - normalize_functions: How to normalize function names. Possible values are: "upper" or True (default): Convert names to uppercase. "lower": Convert names to lowercase. False: Disables function name normalization.
- unsupported_level: Determines the generator's behavior when it encounters unsupported expressions. Default ErrorLevel.WARN.
- max_unsupported: Maximum number of unsupported messages to include in a raised UnsupportedError. This is only relevant if unsupported_level is ErrorLevel.RAISE. Default: 3
- leading_comma: Whether the comma is leading or trailing in select expressions. This is only relevant when generating in pretty mode. Default: False
- max_text_width: The max number of characters in a segment before creating new lines in pretty mode. The default is on the smaller end because the length only represents a segment and not the true line length. Default: 80
- comments: Whether to preserve comments in the output SQL code. Default: True
898 def select_sql(self, expression: exp.Select) -> str: 899 if expression.args.get("offset"): 900 if not expression.args.get("order"): 901 # ORDER BY is required in order to use OFFSET in a query, so we use 902 # a noop order by, since we don't really care about the order. 903 # See: https://www.microsoftpressstore.com/articles/article.aspx?p=2314819 904 expression.order_by(exp.select(exp.null()).subquery(), copy=False) 905 906 limit = expression.args.get("limit") 907 if isinstance(limit, exp.Limit): 908 # TOP and OFFSET can't be combined, we need use FETCH instead of TOP 909 # we replace here because otherwise TOP would be generated in select_sql 910 limit.replace(exp.Fetch(direction="FIRST", count=limit.expression)) 911 912 return super().select_sql(expression)
920 def queryoption_sql(self, expression: exp.QueryOption) -> str: 921 option = self.sql(expression, "this") 922 value = self.sql(expression, "expression") 923 if value: 924 optional_equal_sign = "= " if option in OPTIONS_THAT_REQUIRE_EQUAL else "" 925 return f"{option} {optional_equal_sign}{value}" 926 return option
928 def lateral_op(self, expression: exp.Lateral) -> str: 929 cross_apply = expression.args.get("cross_apply") 930 if cross_apply is True: 931 return "CROSS APPLY" 932 if cross_apply is False: 933 return "OUTER APPLY" 934 935 # TODO: perhaps we can check if the parent is a Join and transpile it appropriately 936 self.unsupported("LATERAL clause is not supported.") 937 return "LATERAL"
939 def timefromparts_sql(self, expression: exp.TimeFromParts) -> str: 940 nano = expression.args.get("nano") 941 if nano is not None: 942 nano.pop() 943 self.unsupported("Specifying nanoseconds is not supported in TIMEFROMPARTS.") 944 945 if expression.args.get("fractions") is None: 946 expression.set("fractions", exp.Literal.number(0)) 947 if expression.args.get("precision") is None: 948 expression.set("precision", exp.Literal.number(0)) 949 950 return rename_func("TIMEFROMPARTS")(self, expression)
952 def timestampfromparts_sql(self, expression: exp.TimestampFromParts) -> str: 953 zone = expression.args.get("zone") 954 if zone is not None: 955 zone.pop() 956 self.unsupported("Time zone is not supported in DATETIMEFROMPARTS.") 957 958 nano = expression.args.get("nano") 959 if nano is not None: 960 nano.pop() 961 self.unsupported("Specifying nanoseconds is not supported in DATETIMEFROMPARTS.") 962 963 if expression.args.get("milli") is None: 964 expression.set("milli", exp.Literal.number(0)) 965 966 return rename_func("DATETIMEFROMPARTS")(self, expression)
968 def setitem_sql(self, expression: exp.SetItem) -> str: 969 this = expression.this 970 if isinstance(this, exp.EQ) and not isinstance(this.left, exp.Parameter): 971 # T-SQL does not use '=' in SET command, except when the LHS is a variable. 972 return f"{self.sql(this.left)} {self.sql(this.right)}" 973 974 return super().setitem_sql(expression)
987 def createable_sql(self, expression: exp.Create, locations: t.DefaultDict) -> str: 988 sql = self.sql(expression, "this") 989 properties = expression.args.get("properties") 990 991 if sql[:1] != "#" and any( 992 isinstance(prop, exp.TemporaryProperty) 993 for prop in (properties.expressions if properties else []) 994 ): 995 sql = f"[#{sql[1:]}" if sql.startswith("[") else f"#{sql}" 996 997 return sql
999 def create_sql(self, expression: exp.Create) -> str: 1000 kind = expression.kind 1001 exists = expression.args.pop("exists", None) 1002 1003 like_property = expression.find(exp.LikeProperty) 1004 if like_property: 1005 ctas_expression = like_property.this 1006 else: 1007 ctas_expression = expression.expression 1008 1009 if kind == "VIEW": 1010 expression.this.set("catalog", None) 1011 with_ = expression.args.get("with") 1012 if ctas_expression and with_: 1013 # We've already preprocessed the Create expression to bubble up any nested CTEs, 1014 # but CREATE VIEW actually requires the WITH clause to come after it so we need 1015 # to amend the AST by moving the CTEs to the CREATE VIEW statement's query. 1016 ctas_expression.set("with", with_.pop()) 1017 1018 sql = super().create_sql(expression) 1019 1020 table = expression.find(exp.Table) 1021 1022 # Convert CTAS statement to SELECT .. INTO .. 1023 if kind == "TABLE" and ctas_expression: 1024 if isinstance(ctas_expression, exp.UNWRAPPED_QUERIES): 1025 ctas_expression = ctas_expression.subquery() 1026 1027 select_into = exp.select("*").from_(exp.alias_(ctas_expression, "temp", table=True)) 1028 select_into.set("into", exp.Into(this=table)) 1029 1030 if like_property: 1031 select_into.limit(0, copy=False) 1032 1033 sql = self.sql(select_into) 1034 1035 if exists: 1036 identifier = self.sql(exp.Literal.string(exp.table_name(table) if table else "")) 1037 sql = self.sql(exp.Literal.string(sql)) 1038 if kind == "SCHEMA": 1039 sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.schemata WHERE schema_name = {identifier}) EXEC({sql})""" 1040 elif kind == "TABLE": 1041 assert table 1042 where = exp.and_( 1043 exp.column("table_name").eq(table.name), 1044 exp.column("table_schema").eq(table.db) if table.db else None, 1045 exp.column("table_catalog").eq(table.catalog) if table.catalog else None, 1046 ) 1047 sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.tables WHERE {where}) EXEC({sql})""" 1048 elif kind == "INDEX": 1049 index = self.sql(exp.Literal.string(expression.this.text("this"))) 1050 sql = f"""IF NOT EXISTS (SELECT * FROM sys.indexes WHERE object_id = object_id({identifier}) AND name = {index}) EXEC({sql})""" 1051 elif expression.args.get("replace"): 1052 sql = sql.replace("CREATE OR REPLACE ", "CREATE OR ALTER ", 1) 1053 1054 return self.prepend_ctes(expression, sql)
1059 def version_sql(self, expression: exp.Version) -> str: 1060 name = "SYSTEM_TIME" if expression.name == "TIMESTAMP" else expression.name 1061 this = f"FOR {name}" 1062 expr = expression.expression 1063 kind = expression.text("kind") 1064 if kind in ("FROM", "BETWEEN"): 1065 args = expr.expressions 1066 sep = "TO" if kind == "FROM" else "AND" 1067 expr_sql = f"{self.sql(seq_get(args, 0))} {sep} {self.sql(seq_get(args, 1))}" 1068 else: 1069 expr_sql = self.sql(expr) 1070 1071 expr_sql = f" {expr_sql}" if expr_sql else "" 1072 return f"{this} {kind}{expr_sql}"
1091 def commit_sql(self, expression: exp.Commit) -> str: 1092 this = self.sql(expression, "this") 1093 this = f" {this}" if this else "" 1094 durability = expression.args.get("durability") 1095 durability = ( 1096 f" WITH (DELAYED_DURABILITY = {'ON' if durability else 'OFF'})" 1097 if durability is not None 1098 else "" 1099 ) 1100 return f"COMMIT TRANSACTION{this}{durability}"
1107 def identifier_sql(self, expression: exp.Identifier) -> str: 1108 identifier = super().identifier_sql(expression) 1109 1110 if expression.args.get("global"): 1111 identifier = f"##{identifier}" 1112 elif expression.args.get("temporary"): 1113 identifier = f"#{identifier}" 1114 1115 return identifier
1143 def altertable_sql(self, expression: exp.AlterTable) -> str: 1144 action = seq_get(expression.args.get("actions") or [], 0) 1145 if isinstance(action, exp.RenameTable): 1146 return f"EXEC sp_rename '{self.sql(expression.this)}', '{action.this.name}'" 1147 return super().altertable_sql(expression)
1157 def declareitem_sql(self, expression: exp.DeclareItem) -> str: 1158 variable = self.sql(expression, "this") 1159 default = self.sql(expression, "default") 1160 default = f" = {default}" if default else "" 1161 1162 kind = self.sql(expression, "kind") 1163 if isinstance(expression.args.get("kind"), exp.Schema): 1164 kind = f"TABLE {kind}" 1165 1166 return f"{variable} AS {kind}{default}"
Inherited Members
- sqlglot.generator.Generator
- Generator
- IGNORE_NULLS_IN_FUNC
- LOCKING_READS_SUPPORTED
- EXPLICIT_SET_OP
- WRAP_DERIVED_VALUES
- CREATE_FUNCTION_RETURN_AS
- MATCHED_BY_SOURCE
- SINGLE_STRING_INTERVAL
- INTERVAL_ALLOWS_PLURAL_FORM
- LIMIT_ONLY_LITERALS
- RENAME_TABLE_WITH_DB
- GROUPINGS_SEP
- INDEX_ON
- JOIN_HINTS
- TABLE_HINTS
- QUERY_HINT_SEP
- IS_BOOL_ALLOWED
- DUPLICATE_KEY_UPDATE_WITH_SET
- EXTRACT_ALLOWS_QUOTES
- TZ_TO_WITH_TIME_ZONE
- VALUES_AS_TABLE
- UNNEST_WITH_ORDINALITY
- AGGREGATE_FILTER_SUPPORTED
- SEMI_ANTI_JOIN_WITH_SIDE
- SUPPORTS_TABLE_COPY
- TABLESAMPLE_REQUIRES_PARENS
- TABLESAMPLE_SIZE_IS_ROWS
- TABLESAMPLE_KEYWORDS
- TABLESAMPLE_WITH_METHOD
- COLLATE_IS_FUNC
- DATA_TYPE_SPECIFIERS_ALLOWED
- LAST_DAY_SUPPORTS_DATE_PART
- SUPPORTS_TABLE_ALIAS_COLUMNS
- UNPIVOT_ALIASES_ARE_IDENTIFIERS
- JSON_KEY_VALUE_PAIR_SEP
- INSERT_OVERWRITE
- SUPPORTS_UNLOGGED_TABLES
- SUPPORTS_CREATE_TABLE_LIKE
- LIKE_PROPERTY_INSIDE_SCHEMA
- MULTI_ARG_DISTINCT
- JSON_TYPE_REQUIRED_FOR_EXTRACTION
- JSON_PATH_SINGLE_QUOTE_ESCAPE
- CAN_IMPLEMENT_ARRAY_ANY
- COPY_PARAMS_ARE_WRAPPED
- COPY_HAS_INTO_KEYWORD
- STAR_EXCEPT
- HEX_FUNC
- WITH_PROPERTIES_PREFIX
- QUOTE_JSON_PATH
- PAD_FILL_PATTERN_IS_REQUIRED
- TIME_PART_SINGULARS
- TOKEN_MAPPING
- STRUCT_DELIMITER
- PARAMETER_TOKEN
- NAMED_PLACEHOLDER_TOKEN
- RESERVED_KEYWORDS
- WITH_SEPARATED_COMMENTS
- EXCLUDE_COMMENTS
- UNWRAPPED_INTERVAL_VALUES
- PARAMETERIZABLE_TEXT_TYPES
- SENTINEL_LINE_BREAK
- pretty
- identify
- normalize
- pad
- unsupported_level
- max_unsupported
- leading_comma
- max_text_width
- comments
- dialect
- normalize_functions
- unsupported_messages
- generate
- preprocess
- unsupported
- sep
- seg
- pad_comment
- maybe_comment
- wrap
- no_identify
- normalize_func
- indent
- sql
- uncache_sql
- cache_sql
- characterset_sql
- column_parts
- column_sql
- columnposition_sql
- columndef_sql
- columnconstraint_sql
- computedcolumnconstraint_sql
- autoincrementcolumnconstraint_sql
- compresscolumnconstraint_sql
- generatedasidentitycolumnconstraint_sql
- generatedasrowcolumnconstraint_sql
- periodforsystemtimeconstraint_sql
- notnullcolumnconstraint_sql
- transformcolumnconstraint_sql
- primarykeycolumnconstraint_sql
- uniquecolumnconstraint_sql
- sequenceproperties_sql
- clone_sql
- describe_sql
- heredoc_sql
- prepend_ctes
- with_sql
- cte_sql
- tablealias_sql
- bitstring_sql
- hexstring_sql
- bytestring_sql
- unicodestring_sql
- rawstring_sql
- datatypeparam_sql
- datatype_sql
- directory_sql
- delete_sql
- except_sql
- except_op
- fetch_sql
- filter_sql
- hint_sql
- indexparameters_sql
- index_sql
- hex_sql
- lowerhex_sql
- inputoutputformat_sql
- national_sql
- properties_sql
- root_properties
- properties
- with_properties
- locate_properties
- property_name
- property_sql
- likeproperty_sql
- fallbackproperty_sql
- journalproperty_sql
- freespaceproperty_sql
- checksumproperty_sql
- mergeblockratioproperty_sql
- datablocksizeproperty_sql
- blockcompressionproperty_sql
- isolatedloadingproperty_sql
- partitionboundspec_sql
- partitionedofproperty_sql
- lockingproperty_sql
- withdataproperty_sql
- withsystemversioningproperty_sql
- insert_sql
- intersect_sql
- intersect_op
- introducer_sql
- kill_sql
- pseudotype_sql
- objectidentifier_sql
- onconflict_sql
- rowformatdelimitedproperty_sql
- withtablehint_sql
- indextablehint_sql
- historicaldata_sql
- table_parts
- table_sql
- tablesample_sql
- pivot_sql
- tuple_sql
- update_sql
- values_sql
- var_sql
- into_sql
- from_sql
- group_sql
- having_sql
- connect_sql
- prior_sql
- join_sql
- lambda_sql
- lateral_sql
- limit_sql
- set_sql
- pragma_sql
- lock_sql
- literal_sql
- escape_str
- loaddata_sql
- null_sql
- order_sql
- withfill_sql
- cluster_sql
- distribute_sql
- sort_sql
- ordered_sql
- matchrecognizemeasure_sql
- matchrecognize_sql
- query_modifiers
- offset_limit_modifiers
- after_limit_modifiers
- schema_sql
- schema_columns_sql
- star_sql
- parameter_sql
- sessionparameter_sql
- placeholder_sql
- subquery_sql
- qualify_sql
- set_operations
- union_sql
- union_op
- unnest_sql
- prewhere_sql
- where_sql
- window_sql
- partition_by_sql
- windowspec_sql
- withingroup_sql
- between_sql
- bracket_offset_expressions
- bracket_sql
- all_sql
- any_sql
- exists_sql
- case_sql
- nextvaluefor_sql
- extract_sql
- trim_sql
- convert_concat_args
- concat_sql
- concatws_sql
- check_sql
- foreignkey_sql
- primarykey_sql
- if_sql
- matchagainst_sql
- jsonkeyvalue_sql
- jsonpath_sql
- json_path_part
- formatjson_sql
- jsonobject_sql
- jsonobjectagg_sql
- jsonarray_sql
- jsonarrayagg_sql
- jsoncolumndef_sql
- jsonschema_sql
- jsontable_sql
- openjsoncolumndef_sql
- openjson_sql
- in_sql
- in_unnest_op
- interval_sql
- return_sql
- reference_sql
- anonymous_sql
- paren_sql
- neg_sql
- not_sql
- alias_sql
- pivotalias_sql
- aliases_sql
- atindex_sql
- attimezone_sql
- fromtimezone_sql
- add_sql
- and_sql
- or_sql
- xor_sql
- connector_sql
- bitwiseand_sql
- bitwiseleftshift_sql
- bitwisenot_sql
- bitwiseor_sql
- bitwiserightshift_sql
- bitwisexor_sql
- cast_sql
- currentdate_sql
- collate_sql
- command_sql
- comment_sql
- mergetreettlaction_sql
- mergetreettl_sql
- altercolumn_sql
- alterdiststyle_sql
- altersortkey_sql
- renametable_sql
- renamecolumn_sql
- alterset_sql
- add_column_sql
- droppartition_sql
- addconstraint_sql
- distinct_sql
- ignorenulls_sql
- respectnulls_sql
- havingmax_sql
- intdiv_sql
- dpipe_sql
- div_sql
- overlaps_sql
- distance_sql
- dot_sql
- eq_sql
- propertyeq_sql
- escape_sql
- glob_sql
- gt_sql
- gte_sql
- ilike_sql
- ilikeany_sql
- like_sql
- likeany_sql
- similarto_sql
- lt_sql
- lte_sql
- mod_sql
- mul_sql
- neq_sql
- nullsafeeq_sql
- nullsafeneq_sql
- slice_sql
- sub_sql
- trycast_sql
- try_sql
- log_sql
- use_sql
- binary
- function_fallback_sql
- func
- format_args
- too_wide
- format_time
- expressions
- op_expressions
- naked_property
- tag_sql
- token_sql
- userdefinedfunction_sql
- joinhint_sql
- kwarg_sql
- when_sql
- merge_sql
- tochar_sql
- tonumber_sql
- dictproperty_sql
- dictrange_sql
- dictsubproperty_sql
- oncluster_sql
- clusteredbyproperty_sql
- anyvalue_sql
- querytransform_sql
- indexconstraintoption_sql
- checkcolumnconstraint_sql
- indexcolumnconstraint_sql
- nvl2_sql
- comprehension_sql
- columnprefix_sql
- opclass_sql
- predict_sql
- forin_sql
- refresh_sql
- operator_sql
- toarray_sql
- tsordstotime_sql
- tsordstotimestamp_sql
- tsordstodate_sql
- unixdate_sql
- lastday_sql
- dateadd_sql
- arrayany_sql
- struct_sql
- partitionrange_sql
- truncatetable_sql
- copyparameter_sql
- credentials_sql
- copy_sql
- semicolon_sql
- datadeletionproperty_sql
- maskingpolicycolumnconstraint_sql
- gapfill_sql
- scoperesolution_sql
- parsejson_sql
- rand_sql
- changes_sql
- pad_sql
- summarize_sql