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 _json_extract_sql( 326 self: TSQL.Generator, expression: exp.JSONExtract | exp.JSONExtractScalar 327) -> str: 328 json_query = self.func("JSON_QUERY", expression.this, expression.expression) 329 json_value = self.func("JSON_VALUE", expression.this, expression.expression) 330 return self.func("ISNULL", json_query, json_value) 331 332 333class TSQL(Dialect): 334 NORMALIZATION_STRATEGY = NormalizationStrategy.CASE_INSENSITIVE 335 TIME_FORMAT = "'yyyy-mm-dd hh:mm:ss'" 336 SUPPORTS_SEMI_ANTI_JOIN = False 337 LOG_BASE_FIRST = False 338 TYPED_DIVISION = True 339 CONCAT_COALESCE = True 340 341 TIME_MAPPING = { 342 "year": "%Y", 343 "dayofyear": "%j", 344 "day": "%d", 345 "dy": "%d", 346 "y": "%Y", 347 "week": "%W", 348 "ww": "%W", 349 "wk": "%W", 350 "hour": "%h", 351 "hh": "%I", 352 "minute": "%M", 353 "mi": "%M", 354 "n": "%M", 355 "second": "%S", 356 "ss": "%S", 357 "s": "%-S", 358 "millisecond": "%f", 359 "ms": "%f", 360 "weekday": "%W", 361 "dw": "%W", 362 "month": "%m", 363 "mm": "%M", 364 "m": "%-M", 365 "Y": "%Y", 366 "YYYY": "%Y", 367 "YY": "%y", 368 "MMMM": "%B", 369 "MMM": "%b", 370 "MM": "%m", 371 "M": "%-m", 372 "dddd": "%A", 373 "dd": "%d", 374 "d": "%-d", 375 "HH": "%H", 376 "H": "%-H", 377 "h": "%-I", 378 "S": "%f", 379 "yyyy": "%Y", 380 "yy": "%y", 381 } 382 383 CONVERT_FORMAT_MAPPING = { 384 "0": "%b %d %Y %-I:%M%p", 385 "1": "%m/%d/%y", 386 "2": "%y.%m.%d", 387 "3": "%d/%m/%y", 388 "4": "%d.%m.%y", 389 "5": "%d-%m-%y", 390 "6": "%d %b %y", 391 "7": "%b %d, %y", 392 "8": "%H:%M:%S", 393 "9": "%b %d %Y %-I:%M:%S:%f%p", 394 "10": "mm-dd-yy", 395 "11": "yy/mm/dd", 396 "12": "yymmdd", 397 "13": "%d %b %Y %H:%M:ss:%f", 398 "14": "%H:%M:%S:%f", 399 "20": "%Y-%m-%d %H:%M:%S", 400 "21": "%Y-%m-%d %H:%M:%S.%f", 401 "22": "%m/%d/%y %-I:%M:%S %p", 402 "23": "%Y-%m-%d", 403 "24": "%H:%M:%S", 404 "25": "%Y-%m-%d %H:%M:%S.%f", 405 "100": "%b %d %Y %-I:%M%p", 406 "101": "%m/%d/%Y", 407 "102": "%Y.%m.%d", 408 "103": "%d/%m/%Y", 409 "104": "%d.%m.%Y", 410 "105": "%d-%m-%Y", 411 "106": "%d %b %Y", 412 "107": "%b %d, %Y", 413 "108": "%H:%M:%S", 414 "109": "%b %d %Y %-I:%M:%S:%f%p", 415 "110": "%m-%d-%Y", 416 "111": "%Y/%m/%d", 417 "112": "%Y%m%d", 418 "113": "%d %b %Y %H:%M:%S:%f", 419 "114": "%H:%M:%S:%f", 420 "120": "%Y-%m-%d %H:%M:%S", 421 "121": "%Y-%m-%d %H:%M:%S.%f", 422 } 423 424 FORMAT_TIME_MAPPING = { 425 "y": "%B %Y", 426 "d": "%m/%d/%Y", 427 "H": "%-H", 428 "h": "%-I", 429 "s": "%Y-%m-%d %H:%M:%S", 430 "D": "%A,%B,%Y", 431 "f": "%A,%B,%Y %-I:%M %p", 432 "F": "%A,%B,%Y %-I:%M:%S %p", 433 "g": "%m/%d/%Y %-I:%M %p", 434 "G": "%m/%d/%Y %-I:%M:%S %p", 435 "M": "%B %-d", 436 "m": "%B %-d", 437 "O": "%Y-%m-%dT%H:%M:%S", 438 "u": "%Y-%M-%D %H:%M:%S%z", 439 "U": "%A, %B %D, %Y %H:%M:%S%z", 440 "T": "%-I:%M:%S %p", 441 "t": "%-I:%M", 442 "Y": "%a %Y", 443 } 444 445 class Tokenizer(tokens.Tokenizer): 446 IDENTIFIERS = [("[", "]"), '"'] 447 QUOTES = ["'", '"'] 448 HEX_STRINGS = [("0x", ""), ("0X", "")] 449 VAR_SINGLE_TOKENS = {"@", "$", "#"} 450 451 KEYWORDS = { 452 **tokens.Tokenizer.KEYWORDS, 453 "CLUSTERED INDEX": TokenType.INDEX, 454 "DATETIME2": TokenType.DATETIME, 455 "DATETIMEOFFSET": TokenType.TIMESTAMPTZ, 456 "DECLARE": TokenType.DECLARE, 457 "EXEC": TokenType.COMMAND, 458 "FOR SYSTEM_TIME": TokenType.TIMESTAMP_SNAPSHOT, 459 "IMAGE": TokenType.IMAGE, 460 "MONEY": TokenType.MONEY, 461 "NONCLUSTERED INDEX": TokenType.INDEX, 462 "NTEXT": TokenType.TEXT, 463 "OPTION": TokenType.OPTION, 464 "OUTPUT": TokenType.RETURNING, 465 "PRINT": TokenType.COMMAND, 466 "PROC": TokenType.PROCEDURE, 467 "REAL": TokenType.FLOAT, 468 "ROWVERSION": TokenType.ROWVERSION, 469 "SMALLDATETIME": TokenType.DATETIME, 470 "SMALLMONEY": TokenType.SMALLMONEY, 471 "SQL_VARIANT": TokenType.VARIANT, 472 "SYSTEM_USER": TokenType.CURRENT_USER, 473 "TOP": TokenType.TOP, 474 "TIMESTAMP": TokenType.ROWVERSION, 475 "TINYINT": TokenType.UTINYINT, 476 "UNIQUEIDENTIFIER": TokenType.UNIQUEIDENTIFIER, 477 "UPDATE STATISTICS": TokenType.COMMAND, 478 "XML": TokenType.XML, 479 } 480 481 COMMANDS = {*tokens.Tokenizer.COMMANDS, TokenType.END} 482 483 class Parser(parser.Parser): 484 SET_REQUIRES_ASSIGNMENT_DELIMITER = False 485 LOG_DEFAULTS_TO_LN = True 486 ALTER_TABLE_ADD_REQUIRED_FOR_EACH_COLUMN = False 487 STRING_ALIASES = True 488 NO_PAREN_IF_COMMANDS = False 489 490 QUERY_MODIFIER_PARSERS = { 491 **parser.Parser.QUERY_MODIFIER_PARSERS, 492 TokenType.OPTION: lambda self: ("options", self._parse_options()), 493 } 494 495 FUNCTIONS = { 496 **parser.Parser.FUNCTIONS, 497 "CHARINDEX": lambda args: exp.StrPosition( 498 this=seq_get(args, 1), 499 substr=seq_get(args, 0), 500 position=seq_get(args, 2), 501 ), 502 "DATEADD": build_date_delta(exp.DateAdd, unit_mapping=DATE_DELTA_INTERVAL), 503 "DATEDIFF": _build_date_delta(exp.DateDiff, unit_mapping=DATE_DELTA_INTERVAL), 504 "DATENAME": _build_formatted_time(exp.TimeToStr, full_format_mapping=True), 505 "DATEPART": _build_formatted_time(exp.TimeToStr), 506 "DATETIMEFROMPARTS": _build_datetimefromparts, 507 "EOMONTH": _build_eomonth, 508 "FORMAT": _build_format, 509 "GETDATE": exp.CurrentTimestamp.from_arg_list, 510 "HASHBYTES": _build_hashbytes, 511 "ISNULL": exp.Coalesce.from_arg_list, 512 "JSON_QUERY": parser.build_extract_json_with_path(exp.JSONExtract), 513 "JSON_VALUE": parser.build_extract_json_with_path(exp.JSONExtractScalar), 514 "LEN": _build_with_arg_as_text(exp.Length), 515 "LEFT": _build_with_arg_as_text(exp.Left), 516 "RIGHT": _build_with_arg_as_text(exp.Right), 517 "REPLICATE": exp.Repeat.from_arg_list, 518 "SQUARE": lambda args: exp.Pow(this=seq_get(args, 0), expression=exp.Literal.number(2)), 519 "SYSDATETIME": exp.CurrentTimestamp.from_arg_list, 520 "SUSER_NAME": exp.CurrentUser.from_arg_list, 521 "SUSER_SNAME": exp.CurrentUser.from_arg_list, 522 "SYSTEM_USER": exp.CurrentUser.from_arg_list, 523 "TIMEFROMPARTS": _build_timefromparts, 524 } 525 526 JOIN_HINTS = {"LOOP", "HASH", "MERGE", "REMOTE"} 527 528 RETURNS_TABLE_TOKENS = parser.Parser.ID_VAR_TOKENS - { 529 TokenType.TABLE, 530 *parser.Parser.TYPE_TOKENS, 531 } 532 533 STATEMENT_PARSERS = { 534 **parser.Parser.STATEMENT_PARSERS, 535 TokenType.DECLARE: lambda self: self._parse_declare(), 536 } 537 538 def _parse_options(self) -> t.Optional[t.List[exp.Expression]]: 539 if not self._match(TokenType.OPTION): 540 return None 541 542 def _parse_option() -> t.Optional[exp.Expression]: 543 option = self._parse_var_from_options(OPTIONS) 544 if not option: 545 return None 546 547 self._match(TokenType.EQ) 548 return self.expression( 549 exp.QueryOption, this=option, expression=self._parse_primary_or_var() 550 ) 551 552 return self._parse_wrapped_csv(_parse_option) 553 554 def _parse_projections(self) -> t.List[exp.Expression]: 555 """ 556 T-SQL supports the syntax alias = expression in the SELECT's projection list, 557 so we transform all parsed Selects to convert their EQ projections into Aliases. 558 559 See: https://learn.microsoft.com/en-us/sql/t-sql/queries/select-clause-transact-sql?view=sql-server-ver16#syntax 560 """ 561 return [ 562 ( 563 exp.alias_(projection.expression, projection.this.this, copy=False) 564 if isinstance(projection, exp.EQ) and isinstance(projection.this, exp.Column) 565 else projection 566 ) 567 for projection in super()._parse_projections() 568 ] 569 570 def _parse_commit_or_rollback(self) -> exp.Commit | exp.Rollback: 571 """Applies to SQL Server and Azure SQL Database 572 COMMIT [ { TRAN | TRANSACTION } 573 [ transaction_name | @tran_name_variable ] ] 574 [ WITH ( DELAYED_DURABILITY = { OFF | ON } ) ] 575 576 ROLLBACK { TRAN | TRANSACTION } 577 [ transaction_name | @tran_name_variable 578 | savepoint_name | @savepoint_variable ] 579 """ 580 rollback = self._prev.token_type == TokenType.ROLLBACK 581 582 self._match_texts(("TRAN", "TRANSACTION")) 583 this = self._parse_id_var() 584 585 if rollback: 586 return self.expression(exp.Rollback, this=this) 587 588 durability = None 589 if self._match_pair(TokenType.WITH, TokenType.L_PAREN): 590 self._match_text_seq("DELAYED_DURABILITY") 591 self._match(TokenType.EQ) 592 593 if self._match_text_seq("OFF"): 594 durability = False 595 else: 596 self._match(TokenType.ON) 597 durability = True 598 599 self._match_r_paren() 600 601 return self.expression(exp.Commit, this=this, durability=durability) 602 603 def _parse_transaction(self) -> exp.Transaction | exp.Command: 604 """Applies to SQL Server and Azure SQL Database 605 BEGIN { TRAN | TRANSACTION } 606 [ { transaction_name | @tran_name_variable } 607 [ WITH MARK [ 'description' ] ] 608 ] 609 """ 610 if self._match_texts(("TRAN", "TRANSACTION")): 611 transaction = self.expression(exp.Transaction, this=self._parse_id_var()) 612 if self._match_text_seq("WITH", "MARK"): 613 transaction.set("mark", self._parse_string()) 614 615 return transaction 616 617 return self._parse_as_command(self._prev) 618 619 def _parse_returns(self) -> exp.ReturnsProperty: 620 table = self._parse_id_var(any_token=False, tokens=self.RETURNS_TABLE_TOKENS) 621 returns = super()._parse_returns() 622 returns.set("table", table) 623 return returns 624 625 def _parse_convert( 626 self, strict: bool, safe: t.Optional[bool] = None 627 ) -> t.Optional[exp.Expression]: 628 this = self._parse_types() 629 self._match(TokenType.COMMA) 630 args = [this, *self._parse_csv(self._parse_assignment)] 631 convert = exp.Convert.from_arg_list(args) 632 convert.set("safe", safe) 633 convert.set("strict", strict) 634 return convert 635 636 def _parse_user_defined_function( 637 self, kind: t.Optional[TokenType] = None 638 ) -> t.Optional[exp.Expression]: 639 this = super()._parse_user_defined_function(kind=kind) 640 641 if ( 642 kind == TokenType.FUNCTION 643 or isinstance(this, exp.UserDefinedFunction) 644 or self._match(TokenType.ALIAS, advance=False) 645 ): 646 return this 647 648 expressions = self._parse_csv(self._parse_function_parameter) 649 return self.expression(exp.UserDefinedFunction, this=this, expressions=expressions) 650 651 def _parse_id_var( 652 self, 653 any_token: bool = True, 654 tokens: t.Optional[t.Collection[TokenType]] = None, 655 ) -> t.Optional[exp.Expression]: 656 is_temporary = self._match(TokenType.HASH) 657 is_global = is_temporary and self._match(TokenType.HASH) 658 659 this = super()._parse_id_var(any_token=any_token, tokens=tokens) 660 if this: 661 if is_global: 662 this.set("global", True) 663 elif is_temporary: 664 this.set("temporary", True) 665 666 return this 667 668 def _parse_create(self) -> exp.Create | exp.Command: 669 create = super()._parse_create() 670 671 if isinstance(create, exp.Create): 672 table = create.this.this if isinstance(create.this, exp.Schema) else create.this 673 if isinstance(table, exp.Table) and table.this.args.get("temporary"): 674 if not create.args.get("properties"): 675 create.set("properties", exp.Properties(expressions=[])) 676 677 create.args["properties"].append("expressions", exp.TemporaryProperty()) 678 679 return create 680 681 def _parse_if(self) -> t.Optional[exp.Expression]: 682 index = self._index 683 684 if self._match_text_seq("OBJECT_ID"): 685 self._parse_wrapped_csv(self._parse_string) 686 if self._match_text_seq("IS", "NOT", "NULL") and self._match(TokenType.DROP): 687 return self._parse_drop(exists=True) 688 self._retreat(index) 689 690 return super()._parse_if() 691 692 def _parse_unique(self) -> exp.UniqueColumnConstraint: 693 if self._match_texts(("CLUSTERED", "NONCLUSTERED")): 694 this = self.CONSTRAINT_PARSERS[self._prev.text.upper()](self) 695 else: 696 this = self._parse_schema(self._parse_id_var(any_token=False)) 697 698 return self.expression(exp.UniqueColumnConstraint, this=this) 699 700 def _parse_partition(self) -> t.Optional[exp.Partition]: 701 if not self._match_text_seq("WITH", "(", "PARTITIONS"): 702 return None 703 704 def parse_range(): 705 low = self._parse_bitwise() 706 high = self._parse_bitwise() if self._match_text_seq("TO") else None 707 708 return ( 709 self.expression(exp.PartitionRange, this=low, expression=high) if high else low 710 ) 711 712 partition = self.expression( 713 exp.Partition, expressions=self._parse_wrapped_csv(parse_range) 714 ) 715 716 self._match_r_paren() 717 718 return partition 719 720 def _parse_declare(self) -> exp.Declare | exp.Command: 721 index = self._index 722 expressions = self._try_parse(partial(self._parse_csv, self._parse_declareitem)) 723 724 if not expressions or self._curr: 725 self._retreat(index) 726 return self._parse_as_command(self._prev) 727 728 return self.expression(exp.Declare, expressions=expressions) 729 730 def _parse_declareitem(self) -> t.Optional[exp.DeclareItem]: 731 var = self._parse_id_var() 732 if not var: 733 return None 734 735 value = None 736 self._match(TokenType.ALIAS) 737 if self._match(TokenType.TABLE): 738 data_type = self._parse_schema() 739 else: 740 data_type = self._parse_types() 741 if self._match(TokenType.EQ): 742 value = self._parse_bitwise() 743 744 return self.expression(exp.DeclareItem, this=var, kind=data_type, default=value) 745 746 class Generator(generator.Generator): 747 LIMIT_IS_TOP = True 748 QUERY_HINTS = False 749 RETURNING_END = False 750 NVL2_SUPPORTED = False 751 ALTER_TABLE_INCLUDE_COLUMN_KEYWORD = False 752 LIMIT_FETCH = "FETCH" 753 COMPUTED_COLUMN_WITH_TYPE = False 754 CTE_RECURSIVE_KEYWORD_REQUIRED = False 755 ENSURE_BOOLS = True 756 NULL_ORDERING_SUPPORTED = None 757 SUPPORTS_SINGLE_ARG_CONCAT = False 758 TABLESAMPLE_SEED_KEYWORD = "REPEATABLE" 759 SUPPORTS_SELECT_INTO = True 760 JSON_PATH_BRACKETED_KEY_SUPPORTED = False 761 SUPPORTS_TO_NUMBER = False 762 SET_OP_MODIFIERS = False 763 COPY_PARAMS_EQ_REQUIRED = True 764 765 EXPRESSIONS_WITHOUT_NESTED_CTES = { 766 exp.Delete, 767 exp.Insert, 768 exp.Intersect, 769 exp.Except, 770 exp.Merge, 771 exp.Select, 772 exp.Subquery, 773 exp.Union, 774 exp.Update, 775 } 776 777 SUPPORTED_JSON_PATH_PARTS = { 778 exp.JSONPathKey, 779 exp.JSONPathRoot, 780 exp.JSONPathSubscript, 781 } 782 783 TYPE_MAPPING = { 784 **generator.Generator.TYPE_MAPPING, 785 exp.DataType.Type.BOOLEAN: "BIT", 786 exp.DataType.Type.DECIMAL: "NUMERIC", 787 exp.DataType.Type.DATETIME: "DATETIME2", 788 exp.DataType.Type.DOUBLE: "FLOAT", 789 exp.DataType.Type.INT: "INTEGER", 790 exp.DataType.Type.ROWVERSION: "ROWVERSION", 791 exp.DataType.Type.TEXT: "VARCHAR(MAX)", 792 exp.DataType.Type.TIMESTAMP: "DATETIME2", 793 exp.DataType.Type.TIMESTAMPTZ: "DATETIMEOFFSET", 794 exp.DataType.Type.UTINYINT: "TINYINT", 795 exp.DataType.Type.VARIANT: "SQL_VARIANT", 796 } 797 798 TYPE_MAPPING.pop(exp.DataType.Type.NCHAR) 799 TYPE_MAPPING.pop(exp.DataType.Type.NVARCHAR) 800 801 TRANSFORMS = { 802 **generator.Generator.TRANSFORMS, 803 exp.AnyValue: any_value_to_max_sql, 804 exp.ArrayToString: rename_func("STRING_AGG"), 805 exp.AutoIncrementColumnConstraint: lambda *_: "IDENTITY", 806 exp.DateAdd: date_delta_sql("DATEADD"), 807 exp.DateDiff: date_delta_sql("DATEDIFF"), 808 exp.CTE: transforms.preprocess([qualify_derived_table_outputs]), 809 exp.CurrentDate: rename_func("GETDATE"), 810 exp.CurrentTimestamp: rename_func("GETDATE"), 811 exp.DateStrToDate: datestrtodate_sql, 812 exp.Extract: rename_func("DATEPART"), 813 exp.GeneratedAsIdentityColumnConstraint: generatedasidentitycolumnconstraint_sql, 814 exp.GroupConcat: _string_agg_sql, 815 exp.If: rename_func("IIF"), 816 exp.JSONExtract: _json_extract_sql, 817 exp.JSONExtractScalar: _json_extract_sql, 818 exp.LastDay: lambda self, e: self.func("EOMONTH", e.this), 819 exp.Max: max_or_greatest, 820 exp.MD5: lambda self, e: self.func("HASHBYTES", exp.Literal.string("MD5"), e.this), 821 exp.Min: min_or_least, 822 exp.NumberToStr: _format_sql, 823 exp.ParseJSON: lambda self, e: self.sql(e, "this"), 824 exp.Repeat: rename_func("REPLICATE"), 825 exp.Select: transforms.preprocess( 826 [ 827 transforms.eliminate_distinct_on, 828 transforms.eliminate_semi_and_anti_joins, 829 transforms.eliminate_qualify, 830 ] 831 ), 832 exp.StrPosition: lambda self, e: self.func( 833 "CHARINDEX", e.args.get("substr"), e.this, e.args.get("position") 834 ), 835 exp.Subquery: transforms.preprocess([qualify_derived_table_outputs]), 836 exp.SHA: lambda self, e: self.func("HASHBYTES", exp.Literal.string("SHA1"), e.this), 837 exp.SHA2: lambda self, e: self.func( 838 "HASHBYTES", exp.Literal.string(f"SHA2_{e.args.get('length', 256)}"), e.this 839 ), 840 exp.TemporaryProperty: lambda self, e: "", 841 exp.TimeStrToTime: lambda self, e: self.sql( 842 exp.cast(e.this, exp.DataType.Type.DATETIME) 843 ), 844 exp.TimeToStr: _format_sql, 845 exp.Trim: trim_sql, 846 exp.TsOrDsAdd: date_delta_sql("DATEADD", cast=True), 847 exp.TsOrDsDiff: date_delta_sql("DATEDIFF"), 848 } 849 850 TRANSFORMS.pop(exp.ReturnsProperty) 851 852 PROPERTIES_LOCATION = { 853 **generator.Generator.PROPERTIES_LOCATION, 854 exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED, 855 } 856 857 def select_sql(self, expression: exp.Select) -> str: 858 if expression.args.get("offset"): 859 if not expression.args.get("order"): 860 # ORDER BY is required in order to use OFFSET in a query, so we use 861 # a noop order by, since we don't really care about the order. 862 # See: https://www.microsoftpressstore.com/articles/article.aspx?p=2314819 863 expression.order_by(exp.select(exp.null()).subquery(), copy=False) 864 865 limit = expression.args.get("limit") 866 if isinstance(limit, exp.Limit): 867 # TOP and OFFSET can't be combined, we need use FETCH instead of TOP 868 # we replace here because otherwise TOP would be generated in select_sql 869 limit.replace(exp.Fetch(direction="FIRST", count=limit.expression)) 870 871 return super().select_sql(expression) 872 873 def convert_sql(self, expression: exp.Convert) -> str: 874 name = "TRY_CONVERT" if expression.args.get("safe") else "CONVERT" 875 return self.func( 876 name, expression.this, expression.expression, expression.args.get("style") 877 ) 878 879 def queryoption_sql(self, expression: exp.QueryOption) -> str: 880 option = self.sql(expression, "this") 881 value = self.sql(expression, "expression") 882 if value: 883 optional_equal_sign = "= " if option in OPTIONS_THAT_REQUIRE_EQUAL else "" 884 return f"{option} {optional_equal_sign}{value}" 885 return option 886 887 def lateral_op(self, expression: exp.Lateral) -> str: 888 cross_apply = expression.args.get("cross_apply") 889 if cross_apply is True: 890 return "CROSS APPLY" 891 if cross_apply is False: 892 return "OUTER APPLY" 893 894 # TODO: perhaps we can check if the parent is a Join and transpile it appropriately 895 self.unsupported("LATERAL clause is not supported.") 896 return "LATERAL" 897 898 def timefromparts_sql(self, expression: exp.TimeFromParts) -> str: 899 nano = expression.args.get("nano") 900 if nano is not None: 901 nano.pop() 902 self.unsupported("Specifying nanoseconds is not supported in TIMEFROMPARTS.") 903 904 if expression.args.get("fractions") is None: 905 expression.set("fractions", exp.Literal.number(0)) 906 if expression.args.get("precision") is None: 907 expression.set("precision", exp.Literal.number(0)) 908 909 return rename_func("TIMEFROMPARTS")(self, expression) 910 911 def timestampfromparts_sql(self, expression: exp.TimestampFromParts) -> str: 912 zone = expression.args.get("zone") 913 if zone is not None: 914 zone.pop() 915 self.unsupported("Time zone is not supported in DATETIMEFROMPARTS.") 916 917 nano = expression.args.get("nano") 918 if nano is not None: 919 nano.pop() 920 self.unsupported("Specifying nanoseconds is not supported in DATETIMEFROMPARTS.") 921 922 if expression.args.get("milli") is None: 923 expression.set("milli", exp.Literal.number(0)) 924 925 return rename_func("DATETIMEFROMPARTS")(self, expression) 926 927 def setitem_sql(self, expression: exp.SetItem) -> str: 928 this = expression.this 929 if isinstance(this, exp.EQ) and not isinstance(this.left, exp.Parameter): 930 # T-SQL does not use '=' in SET command, except when the LHS is a variable. 931 return f"{self.sql(this.left)} {self.sql(this.right)}" 932 933 return super().setitem_sql(expression) 934 935 def boolean_sql(self, expression: exp.Boolean) -> str: 936 if type(expression.parent) in BIT_TYPES: 937 return "1" if expression.this else "0" 938 939 return "(1 = 1)" if expression.this else "(1 = 0)" 940 941 def is_sql(self, expression: exp.Is) -> str: 942 if isinstance(expression.expression, exp.Boolean): 943 return self.binary(expression, "=") 944 return self.binary(expression, "IS") 945 946 def createable_sql(self, expression: exp.Create, locations: t.DefaultDict) -> str: 947 sql = self.sql(expression, "this") 948 properties = expression.args.get("properties") 949 950 if sql[:1] != "#" and any( 951 isinstance(prop, exp.TemporaryProperty) 952 for prop in (properties.expressions if properties else []) 953 ): 954 sql = f"[#{sql[1:]}" if sql.startswith("[") else f"#{sql}" 955 956 return sql 957 958 def create_sql(self, expression: exp.Create) -> str: 959 kind = expression.kind 960 exists = expression.args.pop("exists", None) 961 962 if kind == "VIEW": 963 expression.this.set("catalog", None) 964 965 sql = super().create_sql(expression) 966 967 like_property = expression.find(exp.LikeProperty) 968 if like_property: 969 ctas_expression = like_property.this 970 else: 971 ctas_expression = expression.expression 972 973 table = expression.find(exp.Table) 974 975 # Convert CTAS statement to SELECT .. INTO .. 976 if kind == "TABLE" and ctas_expression: 977 ctas_with = ctas_expression.args.get("with") 978 if ctas_with: 979 ctas_with = ctas_with.pop() 980 981 if isinstance(ctas_expression, exp.UNWRAPPED_QUERIES): 982 ctas_expression = ctas_expression.subquery() 983 984 select_into = exp.select("*").from_(exp.alias_(ctas_expression, "temp", table=True)) 985 select_into.set("into", exp.Into(this=table)) 986 select_into.set("with", ctas_with) 987 988 if like_property: 989 select_into.limit(0, copy=False) 990 991 sql = self.sql(select_into) 992 993 if exists: 994 identifier = self.sql(exp.Literal.string(exp.table_name(table) if table else "")) 995 sql = self.sql(exp.Literal.string(sql)) 996 if kind == "SCHEMA": 997 sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.schemata WHERE schema_name = {identifier}) EXEC({sql})""" 998 elif kind == "TABLE": 999 assert table 1000 where = exp.and_( 1001 exp.column("table_name").eq(table.name), 1002 exp.column("table_schema").eq(table.db) if table.db else None, 1003 exp.column("table_catalog").eq(table.catalog) if table.catalog else None, 1004 ) 1005 sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.tables WHERE {where}) EXEC({sql})""" 1006 elif kind == "INDEX": 1007 index = self.sql(exp.Literal.string(expression.this.text("this"))) 1008 sql = f"""IF NOT EXISTS (SELECT * FROM sys.indexes WHERE object_id = object_id({identifier}) AND name = {index}) EXEC({sql})""" 1009 elif expression.args.get("replace"): 1010 sql = sql.replace("CREATE OR REPLACE ", "CREATE OR ALTER ", 1) 1011 1012 return self.prepend_ctes(expression, sql) 1013 1014 def offset_sql(self, expression: exp.Offset) -> str: 1015 return f"{super().offset_sql(expression)} ROWS" 1016 1017 def version_sql(self, expression: exp.Version) -> str: 1018 name = "SYSTEM_TIME" if expression.name == "TIMESTAMP" else expression.name 1019 this = f"FOR {name}" 1020 expr = expression.expression 1021 kind = expression.text("kind") 1022 if kind in ("FROM", "BETWEEN"): 1023 args = expr.expressions 1024 sep = "TO" if kind == "FROM" else "AND" 1025 expr_sql = f"{self.sql(seq_get(args, 0))} {sep} {self.sql(seq_get(args, 1))}" 1026 else: 1027 expr_sql = self.sql(expr) 1028 1029 expr_sql = f" {expr_sql}" if expr_sql else "" 1030 return f"{this} {kind}{expr_sql}" 1031 1032 def returnsproperty_sql(self, expression: exp.ReturnsProperty) -> str: 1033 table = expression.args.get("table") 1034 table = f"{table} " if table else "" 1035 return f"RETURNS {table}{self.sql(expression, 'this')}" 1036 1037 def returning_sql(self, expression: exp.Returning) -> str: 1038 into = self.sql(expression, "into") 1039 into = self.seg(f"INTO {into}") if into else "" 1040 return f"{self.seg('OUTPUT')} {self.expressions(expression, flat=True)}{into}" 1041 1042 def transaction_sql(self, expression: exp.Transaction) -> str: 1043 this = self.sql(expression, "this") 1044 this = f" {this}" if this else "" 1045 mark = self.sql(expression, "mark") 1046 mark = f" WITH MARK {mark}" if mark else "" 1047 return f"BEGIN TRANSACTION{this}{mark}" 1048 1049 def commit_sql(self, expression: exp.Commit) -> str: 1050 this = self.sql(expression, "this") 1051 this = f" {this}" if this else "" 1052 durability = expression.args.get("durability") 1053 durability = ( 1054 f" WITH (DELAYED_DURABILITY = {'ON' if durability else 'OFF'})" 1055 if durability is not None 1056 else "" 1057 ) 1058 return f"COMMIT TRANSACTION{this}{durability}" 1059 1060 def rollback_sql(self, expression: exp.Rollback) -> str: 1061 this = self.sql(expression, "this") 1062 this = f" {this}" if this else "" 1063 return f"ROLLBACK TRANSACTION{this}" 1064 1065 def identifier_sql(self, expression: exp.Identifier) -> str: 1066 identifier = super().identifier_sql(expression) 1067 1068 if expression.args.get("global"): 1069 identifier = f"##{identifier}" 1070 elif expression.args.get("temporary"): 1071 identifier = f"#{identifier}" 1072 1073 return identifier 1074 1075 def constraint_sql(self, expression: exp.Constraint) -> str: 1076 this = self.sql(expression, "this") 1077 expressions = self.expressions(expression, flat=True, sep=" ") 1078 return f"CONSTRAINT {this} {expressions}" 1079 1080 def length_sql(self, expression: exp.Length) -> str: 1081 return self._uncast_text(expression, "LEN") 1082 1083 def right_sql(self, expression: exp.Right) -> str: 1084 return self._uncast_text(expression, "RIGHT") 1085 1086 def left_sql(self, expression: exp.Left) -> str: 1087 return self._uncast_text(expression, "LEFT") 1088 1089 def _uncast_text(self, expression: exp.Expression, name: str) -> str: 1090 this = expression.this 1091 if isinstance(this, exp.Cast) and this.is_type(exp.DataType.Type.TEXT): 1092 this_sql = self.sql(this, "this") 1093 else: 1094 this_sql = self.sql(this) 1095 expression_sql = self.sql(expression, "expression") 1096 return self.func(name, this_sql, expression_sql if expression_sql else None) 1097 1098 def partition_sql(self, expression: exp.Partition) -> str: 1099 return f"WITH (PARTITIONS({self.expressions(expression, flat=True)}))" 1100 1101 def altertable_sql(self, expression: exp.AlterTable) -> str: 1102 action = seq_get(expression.args.get("actions") or [], 0) 1103 if isinstance(action, exp.RenameTable): 1104 return f"EXEC sp_rename '{self.sql(expression.this)}', '{action.this.name}'" 1105 return super().altertable_sql(expression) 1106 1107 def drop_sql(self, expression: exp.Drop) -> str: 1108 if expression.args["kind"] == "VIEW": 1109 expression.this.set("catalog", None) 1110 return super().drop_sql(expression) 1111 1112 def declare_sql(self, expression: exp.Declare) -> str: 1113 return f"DECLARE {self.expressions(expression, flat=True)}" 1114 1115 def declareitem_sql(self, expression: exp.DeclareItem) -> str: 1116 variable = self.sql(expression, "this") 1117 default = self.sql(expression, "default") 1118 default = f" = {default}" if default else "" 1119 1120 kind = self.sql(expression, "kind") 1121 if isinstance(expression.args.get("kind"), exp.Schema): 1122 kind = f"TABLE {kind}" 1123 1124 return f"{variable} AS {kind}{default}" 1125 1126 def options_modifier(self, expression: exp.Expression) -> str: 1127 options = self.expressions(expression, key="options") 1128 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.
334class TSQL(Dialect): 335 NORMALIZATION_STRATEGY = NormalizationStrategy.CASE_INSENSITIVE 336 TIME_FORMAT = "'yyyy-mm-dd hh:mm:ss'" 337 SUPPORTS_SEMI_ANTI_JOIN = False 338 LOG_BASE_FIRST = False 339 TYPED_DIVISION = True 340 CONCAT_COALESCE = True 341 342 TIME_MAPPING = { 343 "year": "%Y", 344 "dayofyear": "%j", 345 "day": "%d", 346 "dy": "%d", 347 "y": "%Y", 348 "week": "%W", 349 "ww": "%W", 350 "wk": "%W", 351 "hour": "%h", 352 "hh": "%I", 353 "minute": "%M", 354 "mi": "%M", 355 "n": "%M", 356 "second": "%S", 357 "ss": "%S", 358 "s": "%-S", 359 "millisecond": "%f", 360 "ms": "%f", 361 "weekday": "%W", 362 "dw": "%W", 363 "month": "%m", 364 "mm": "%M", 365 "m": "%-M", 366 "Y": "%Y", 367 "YYYY": "%Y", 368 "YY": "%y", 369 "MMMM": "%B", 370 "MMM": "%b", 371 "MM": "%m", 372 "M": "%-m", 373 "dddd": "%A", 374 "dd": "%d", 375 "d": "%-d", 376 "HH": "%H", 377 "H": "%-H", 378 "h": "%-I", 379 "S": "%f", 380 "yyyy": "%Y", 381 "yy": "%y", 382 } 383 384 CONVERT_FORMAT_MAPPING = { 385 "0": "%b %d %Y %-I:%M%p", 386 "1": "%m/%d/%y", 387 "2": "%y.%m.%d", 388 "3": "%d/%m/%y", 389 "4": "%d.%m.%y", 390 "5": "%d-%m-%y", 391 "6": "%d %b %y", 392 "7": "%b %d, %y", 393 "8": "%H:%M:%S", 394 "9": "%b %d %Y %-I:%M:%S:%f%p", 395 "10": "mm-dd-yy", 396 "11": "yy/mm/dd", 397 "12": "yymmdd", 398 "13": "%d %b %Y %H:%M:ss:%f", 399 "14": "%H:%M:%S:%f", 400 "20": "%Y-%m-%d %H:%M:%S", 401 "21": "%Y-%m-%d %H:%M:%S.%f", 402 "22": "%m/%d/%y %-I:%M:%S %p", 403 "23": "%Y-%m-%d", 404 "24": "%H:%M:%S", 405 "25": "%Y-%m-%d %H:%M:%S.%f", 406 "100": "%b %d %Y %-I:%M%p", 407 "101": "%m/%d/%Y", 408 "102": "%Y.%m.%d", 409 "103": "%d/%m/%Y", 410 "104": "%d.%m.%Y", 411 "105": "%d-%m-%Y", 412 "106": "%d %b %Y", 413 "107": "%b %d, %Y", 414 "108": "%H:%M:%S", 415 "109": "%b %d %Y %-I:%M:%S:%f%p", 416 "110": "%m-%d-%Y", 417 "111": "%Y/%m/%d", 418 "112": "%Y%m%d", 419 "113": "%d %b %Y %H:%M:%S:%f", 420 "114": "%H:%M:%S:%f", 421 "120": "%Y-%m-%d %H:%M:%S", 422 "121": "%Y-%m-%d %H:%M:%S.%f", 423 } 424 425 FORMAT_TIME_MAPPING = { 426 "y": "%B %Y", 427 "d": "%m/%d/%Y", 428 "H": "%-H", 429 "h": "%-I", 430 "s": "%Y-%m-%d %H:%M:%S", 431 "D": "%A,%B,%Y", 432 "f": "%A,%B,%Y %-I:%M %p", 433 "F": "%A,%B,%Y %-I:%M:%S %p", 434 "g": "%m/%d/%Y %-I:%M %p", 435 "G": "%m/%d/%Y %-I:%M:%S %p", 436 "M": "%B %-d", 437 "m": "%B %-d", 438 "O": "%Y-%m-%dT%H:%M:%S", 439 "u": "%Y-%M-%D %H:%M:%S%z", 440 "U": "%A, %B %D, %Y %H:%M:%S%z", 441 "T": "%-I:%M:%S %p", 442 "t": "%-I:%M", 443 "Y": "%a %Y", 444 } 445 446 class Tokenizer(tokens.Tokenizer): 447 IDENTIFIERS = [("[", "]"), '"'] 448 QUOTES = ["'", '"'] 449 HEX_STRINGS = [("0x", ""), ("0X", "")] 450 VAR_SINGLE_TOKENS = {"@", "$", "#"} 451 452 KEYWORDS = { 453 **tokens.Tokenizer.KEYWORDS, 454 "CLUSTERED INDEX": TokenType.INDEX, 455 "DATETIME2": TokenType.DATETIME, 456 "DATETIMEOFFSET": TokenType.TIMESTAMPTZ, 457 "DECLARE": TokenType.DECLARE, 458 "EXEC": TokenType.COMMAND, 459 "FOR SYSTEM_TIME": TokenType.TIMESTAMP_SNAPSHOT, 460 "IMAGE": TokenType.IMAGE, 461 "MONEY": TokenType.MONEY, 462 "NONCLUSTERED INDEX": TokenType.INDEX, 463 "NTEXT": TokenType.TEXT, 464 "OPTION": TokenType.OPTION, 465 "OUTPUT": TokenType.RETURNING, 466 "PRINT": TokenType.COMMAND, 467 "PROC": TokenType.PROCEDURE, 468 "REAL": TokenType.FLOAT, 469 "ROWVERSION": TokenType.ROWVERSION, 470 "SMALLDATETIME": TokenType.DATETIME, 471 "SMALLMONEY": TokenType.SMALLMONEY, 472 "SQL_VARIANT": TokenType.VARIANT, 473 "SYSTEM_USER": TokenType.CURRENT_USER, 474 "TOP": TokenType.TOP, 475 "TIMESTAMP": TokenType.ROWVERSION, 476 "TINYINT": TokenType.UTINYINT, 477 "UNIQUEIDENTIFIER": TokenType.UNIQUEIDENTIFIER, 478 "UPDATE STATISTICS": TokenType.COMMAND, 479 "XML": TokenType.XML, 480 } 481 482 COMMANDS = {*tokens.Tokenizer.COMMANDS, TokenType.END} 483 484 class Parser(parser.Parser): 485 SET_REQUIRES_ASSIGNMENT_DELIMITER = False 486 LOG_DEFAULTS_TO_LN = True 487 ALTER_TABLE_ADD_REQUIRED_FOR_EACH_COLUMN = False 488 STRING_ALIASES = True 489 NO_PAREN_IF_COMMANDS = False 490 491 QUERY_MODIFIER_PARSERS = { 492 **parser.Parser.QUERY_MODIFIER_PARSERS, 493 TokenType.OPTION: lambda self: ("options", self._parse_options()), 494 } 495 496 FUNCTIONS = { 497 **parser.Parser.FUNCTIONS, 498 "CHARINDEX": lambda args: exp.StrPosition( 499 this=seq_get(args, 1), 500 substr=seq_get(args, 0), 501 position=seq_get(args, 2), 502 ), 503 "DATEADD": build_date_delta(exp.DateAdd, unit_mapping=DATE_DELTA_INTERVAL), 504 "DATEDIFF": _build_date_delta(exp.DateDiff, unit_mapping=DATE_DELTA_INTERVAL), 505 "DATENAME": _build_formatted_time(exp.TimeToStr, full_format_mapping=True), 506 "DATEPART": _build_formatted_time(exp.TimeToStr), 507 "DATETIMEFROMPARTS": _build_datetimefromparts, 508 "EOMONTH": _build_eomonth, 509 "FORMAT": _build_format, 510 "GETDATE": exp.CurrentTimestamp.from_arg_list, 511 "HASHBYTES": _build_hashbytes, 512 "ISNULL": exp.Coalesce.from_arg_list, 513 "JSON_QUERY": parser.build_extract_json_with_path(exp.JSONExtract), 514 "JSON_VALUE": parser.build_extract_json_with_path(exp.JSONExtractScalar), 515 "LEN": _build_with_arg_as_text(exp.Length), 516 "LEFT": _build_with_arg_as_text(exp.Left), 517 "RIGHT": _build_with_arg_as_text(exp.Right), 518 "REPLICATE": exp.Repeat.from_arg_list, 519 "SQUARE": lambda args: exp.Pow(this=seq_get(args, 0), expression=exp.Literal.number(2)), 520 "SYSDATETIME": exp.CurrentTimestamp.from_arg_list, 521 "SUSER_NAME": exp.CurrentUser.from_arg_list, 522 "SUSER_SNAME": exp.CurrentUser.from_arg_list, 523 "SYSTEM_USER": exp.CurrentUser.from_arg_list, 524 "TIMEFROMPARTS": _build_timefromparts, 525 } 526 527 JOIN_HINTS = {"LOOP", "HASH", "MERGE", "REMOTE"} 528 529 RETURNS_TABLE_TOKENS = parser.Parser.ID_VAR_TOKENS - { 530 TokenType.TABLE, 531 *parser.Parser.TYPE_TOKENS, 532 } 533 534 STATEMENT_PARSERS = { 535 **parser.Parser.STATEMENT_PARSERS, 536 TokenType.DECLARE: lambda self: self._parse_declare(), 537 } 538 539 def _parse_options(self) -> t.Optional[t.List[exp.Expression]]: 540 if not self._match(TokenType.OPTION): 541 return None 542 543 def _parse_option() -> t.Optional[exp.Expression]: 544 option = self._parse_var_from_options(OPTIONS) 545 if not option: 546 return None 547 548 self._match(TokenType.EQ) 549 return self.expression( 550 exp.QueryOption, this=option, expression=self._parse_primary_or_var() 551 ) 552 553 return self._parse_wrapped_csv(_parse_option) 554 555 def _parse_projections(self) -> t.List[exp.Expression]: 556 """ 557 T-SQL supports the syntax alias = expression in the SELECT's projection list, 558 so we transform all parsed Selects to convert their EQ projections into Aliases. 559 560 See: https://learn.microsoft.com/en-us/sql/t-sql/queries/select-clause-transact-sql?view=sql-server-ver16#syntax 561 """ 562 return [ 563 ( 564 exp.alias_(projection.expression, projection.this.this, copy=False) 565 if isinstance(projection, exp.EQ) and isinstance(projection.this, exp.Column) 566 else projection 567 ) 568 for projection in super()._parse_projections() 569 ] 570 571 def _parse_commit_or_rollback(self) -> exp.Commit | exp.Rollback: 572 """Applies to SQL Server and Azure SQL Database 573 COMMIT [ { TRAN | TRANSACTION } 574 [ transaction_name | @tran_name_variable ] ] 575 [ WITH ( DELAYED_DURABILITY = { OFF | ON } ) ] 576 577 ROLLBACK { TRAN | TRANSACTION } 578 [ transaction_name | @tran_name_variable 579 | savepoint_name | @savepoint_variable ] 580 """ 581 rollback = self._prev.token_type == TokenType.ROLLBACK 582 583 self._match_texts(("TRAN", "TRANSACTION")) 584 this = self._parse_id_var() 585 586 if rollback: 587 return self.expression(exp.Rollback, this=this) 588 589 durability = None 590 if self._match_pair(TokenType.WITH, TokenType.L_PAREN): 591 self._match_text_seq("DELAYED_DURABILITY") 592 self._match(TokenType.EQ) 593 594 if self._match_text_seq("OFF"): 595 durability = False 596 else: 597 self._match(TokenType.ON) 598 durability = True 599 600 self._match_r_paren() 601 602 return self.expression(exp.Commit, this=this, durability=durability) 603 604 def _parse_transaction(self) -> exp.Transaction | exp.Command: 605 """Applies to SQL Server and Azure SQL Database 606 BEGIN { TRAN | TRANSACTION } 607 [ { transaction_name | @tran_name_variable } 608 [ WITH MARK [ 'description' ] ] 609 ] 610 """ 611 if self._match_texts(("TRAN", "TRANSACTION")): 612 transaction = self.expression(exp.Transaction, this=self._parse_id_var()) 613 if self._match_text_seq("WITH", "MARK"): 614 transaction.set("mark", self._parse_string()) 615 616 return transaction 617 618 return self._parse_as_command(self._prev) 619 620 def _parse_returns(self) -> exp.ReturnsProperty: 621 table = self._parse_id_var(any_token=False, tokens=self.RETURNS_TABLE_TOKENS) 622 returns = super()._parse_returns() 623 returns.set("table", table) 624 return returns 625 626 def _parse_convert( 627 self, strict: bool, safe: t.Optional[bool] = None 628 ) -> t.Optional[exp.Expression]: 629 this = self._parse_types() 630 self._match(TokenType.COMMA) 631 args = [this, *self._parse_csv(self._parse_assignment)] 632 convert = exp.Convert.from_arg_list(args) 633 convert.set("safe", safe) 634 convert.set("strict", strict) 635 return convert 636 637 def _parse_user_defined_function( 638 self, kind: t.Optional[TokenType] = None 639 ) -> t.Optional[exp.Expression]: 640 this = super()._parse_user_defined_function(kind=kind) 641 642 if ( 643 kind == TokenType.FUNCTION 644 or isinstance(this, exp.UserDefinedFunction) 645 or self._match(TokenType.ALIAS, advance=False) 646 ): 647 return this 648 649 expressions = self._parse_csv(self._parse_function_parameter) 650 return self.expression(exp.UserDefinedFunction, this=this, expressions=expressions) 651 652 def _parse_id_var( 653 self, 654 any_token: bool = True, 655 tokens: t.Optional[t.Collection[TokenType]] = None, 656 ) -> t.Optional[exp.Expression]: 657 is_temporary = self._match(TokenType.HASH) 658 is_global = is_temporary and self._match(TokenType.HASH) 659 660 this = super()._parse_id_var(any_token=any_token, tokens=tokens) 661 if this: 662 if is_global: 663 this.set("global", True) 664 elif is_temporary: 665 this.set("temporary", True) 666 667 return this 668 669 def _parse_create(self) -> exp.Create | exp.Command: 670 create = super()._parse_create() 671 672 if isinstance(create, exp.Create): 673 table = create.this.this if isinstance(create.this, exp.Schema) else create.this 674 if isinstance(table, exp.Table) and table.this.args.get("temporary"): 675 if not create.args.get("properties"): 676 create.set("properties", exp.Properties(expressions=[])) 677 678 create.args["properties"].append("expressions", exp.TemporaryProperty()) 679 680 return create 681 682 def _parse_if(self) -> t.Optional[exp.Expression]: 683 index = self._index 684 685 if self._match_text_seq("OBJECT_ID"): 686 self._parse_wrapped_csv(self._parse_string) 687 if self._match_text_seq("IS", "NOT", "NULL") and self._match(TokenType.DROP): 688 return self._parse_drop(exists=True) 689 self._retreat(index) 690 691 return super()._parse_if() 692 693 def _parse_unique(self) -> exp.UniqueColumnConstraint: 694 if self._match_texts(("CLUSTERED", "NONCLUSTERED")): 695 this = self.CONSTRAINT_PARSERS[self._prev.text.upper()](self) 696 else: 697 this = self._parse_schema(self._parse_id_var(any_token=False)) 698 699 return self.expression(exp.UniqueColumnConstraint, this=this) 700 701 def _parse_partition(self) -> t.Optional[exp.Partition]: 702 if not self._match_text_seq("WITH", "(", "PARTITIONS"): 703 return None 704 705 def parse_range(): 706 low = self._parse_bitwise() 707 high = self._parse_bitwise() if self._match_text_seq("TO") else None 708 709 return ( 710 self.expression(exp.PartitionRange, this=low, expression=high) if high else low 711 ) 712 713 partition = self.expression( 714 exp.Partition, expressions=self._parse_wrapped_csv(parse_range) 715 ) 716 717 self._match_r_paren() 718 719 return partition 720 721 def _parse_declare(self) -> exp.Declare | exp.Command: 722 index = self._index 723 expressions = self._try_parse(partial(self._parse_csv, self._parse_declareitem)) 724 725 if not expressions or self._curr: 726 self._retreat(index) 727 return self._parse_as_command(self._prev) 728 729 return self.expression(exp.Declare, expressions=expressions) 730 731 def _parse_declareitem(self) -> t.Optional[exp.DeclareItem]: 732 var = self._parse_id_var() 733 if not var: 734 return None 735 736 value = None 737 self._match(TokenType.ALIAS) 738 if self._match(TokenType.TABLE): 739 data_type = self._parse_schema() 740 else: 741 data_type = self._parse_types() 742 if self._match(TokenType.EQ): 743 value = self._parse_bitwise() 744 745 return self.expression(exp.DeclareItem, this=var, kind=data_type, default=value) 746 747 class Generator(generator.Generator): 748 LIMIT_IS_TOP = True 749 QUERY_HINTS = False 750 RETURNING_END = False 751 NVL2_SUPPORTED = False 752 ALTER_TABLE_INCLUDE_COLUMN_KEYWORD = False 753 LIMIT_FETCH = "FETCH" 754 COMPUTED_COLUMN_WITH_TYPE = False 755 CTE_RECURSIVE_KEYWORD_REQUIRED = False 756 ENSURE_BOOLS = True 757 NULL_ORDERING_SUPPORTED = None 758 SUPPORTS_SINGLE_ARG_CONCAT = False 759 TABLESAMPLE_SEED_KEYWORD = "REPEATABLE" 760 SUPPORTS_SELECT_INTO = True 761 JSON_PATH_BRACKETED_KEY_SUPPORTED = False 762 SUPPORTS_TO_NUMBER = False 763 SET_OP_MODIFIERS = False 764 COPY_PARAMS_EQ_REQUIRED = True 765 766 EXPRESSIONS_WITHOUT_NESTED_CTES = { 767 exp.Delete, 768 exp.Insert, 769 exp.Intersect, 770 exp.Except, 771 exp.Merge, 772 exp.Select, 773 exp.Subquery, 774 exp.Union, 775 exp.Update, 776 } 777 778 SUPPORTED_JSON_PATH_PARTS = { 779 exp.JSONPathKey, 780 exp.JSONPathRoot, 781 exp.JSONPathSubscript, 782 } 783 784 TYPE_MAPPING = { 785 **generator.Generator.TYPE_MAPPING, 786 exp.DataType.Type.BOOLEAN: "BIT", 787 exp.DataType.Type.DECIMAL: "NUMERIC", 788 exp.DataType.Type.DATETIME: "DATETIME2", 789 exp.DataType.Type.DOUBLE: "FLOAT", 790 exp.DataType.Type.INT: "INTEGER", 791 exp.DataType.Type.ROWVERSION: "ROWVERSION", 792 exp.DataType.Type.TEXT: "VARCHAR(MAX)", 793 exp.DataType.Type.TIMESTAMP: "DATETIME2", 794 exp.DataType.Type.TIMESTAMPTZ: "DATETIMEOFFSET", 795 exp.DataType.Type.UTINYINT: "TINYINT", 796 exp.DataType.Type.VARIANT: "SQL_VARIANT", 797 } 798 799 TYPE_MAPPING.pop(exp.DataType.Type.NCHAR) 800 TYPE_MAPPING.pop(exp.DataType.Type.NVARCHAR) 801 802 TRANSFORMS = { 803 **generator.Generator.TRANSFORMS, 804 exp.AnyValue: any_value_to_max_sql, 805 exp.ArrayToString: rename_func("STRING_AGG"), 806 exp.AutoIncrementColumnConstraint: lambda *_: "IDENTITY", 807 exp.DateAdd: date_delta_sql("DATEADD"), 808 exp.DateDiff: date_delta_sql("DATEDIFF"), 809 exp.CTE: transforms.preprocess([qualify_derived_table_outputs]), 810 exp.CurrentDate: rename_func("GETDATE"), 811 exp.CurrentTimestamp: rename_func("GETDATE"), 812 exp.DateStrToDate: datestrtodate_sql, 813 exp.Extract: rename_func("DATEPART"), 814 exp.GeneratedAsIdentityColumnConstraint: generatedasidentitycolumnconstraint_sql, 815 exp.GroupConcat: _string_agg_sql, 816 exp.If: rename_func("IIF"), 817 exp.JSONExtract: _json_extract_sql, 818 exp.JSONExtractScalar: _json_extract_sql, 819 exp.LastDay: lambda self, e: self.func("EOMONTH", e.this), 820 exp.Max: max_or_greatest, 821 exp.MD5: lambda self, e: self.func("HASHBYTES", exp.Literal.string("MD5"), e.this), 822 exp.Min: min_or_least, 823 exp.NumberToStr: _format_sql, 824 exp.ParseJSON: lambda self, e: self.sql(e, "this"), 825 exp.Repeat: rename_func("REPLICATE"), 826 exp.Select: transforms.preprocess( 827 [ 828 transforms.eliminate_distinct_on, 829 transforms.eliminate_semi_and_anti_joins, 830 transforms.eliminate_qualify, 831 ] 832 ), 833 exp.StrPosition: lambda self, e: self.func( 834 "CHARINDEX", e.args.get("substr"), e.this, e.args.get("position") 835 ), 836 exp.Subquery: transforms.preprocess([qualify_derived_table_outputs]), 837 exp.SHA: lambda self, e: self.func("HASHBYTES", exp.Literal.string("SHA1"), e.this), 838 exp.SHA2: lambda self, e: self.func( 839 "HASHBYTES", exp.Literal.string(f"SHA2_{e.args.get('length', 256)}"), e.this 840 ), 841 exp.TemporaryProperty: lambda self, e: "", 842 exp.TimeStrToTime: lambda self, e: self.sql( 843 exp.cast(e.this, exp.DataType.Type.DATETIME) 844 ), 845 exp.TimeToStr: _format_sql, 846 exp.Trim: trim_sql, 847 exp.TsOrDsAdd: date_delta_sql("DATEADD", cast=True), 848 exp.TsOrDsDiff: date_delta_sql("DATEDIFF"), 849 } 850 851 TRANSFORMS.pop(exp.ReturnsProperty) 852 853 PROPERTIES_LOCATION = { 854 **generator.Generator.PROPERTIES_LOCATION, 855 exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED, 856 } 857 858 def select_sql(self, expression: exp.Select) -> str: 859 if expression.args.get("offset"): 860 if not expression.args.get("order"): 861 # ORDER BY is required in order to use OFFSET in a query, so we use 862 # a noop order by, since we don't really care about the order. 863 # See: https://www.microsoftpressstore.com/articles/article.aspx?p=2314819 864 expression.order_by(exp.select(exp.null()).subquery(), copy=False) 865 866 limit = expression.args.get("limit") 867 if isinstance(limit, exp.Limit): 868 # TOP and OFFSET can't be combined, we need use FETCH instead of TOP 869 # we replace here because otherwise TOP would be generated in select_sql 870 limit.replace(exp.Fetch(direction="FIRST", count=limit.expression)) 871 872 return super().select_sql(expression) 873 874 def convert_sql(self, expression: exp.Convert) -> str: 875 name = "TRY_CONVERT" if expression.args.get("safe") else "CONVERT" 876 return self.func( 877 name, expression.this, expression.expression, expression.args.get("style") 878 ) 879 880 def queryoption_sql(self, expression: exp.QueryOption) -> str: 881 option = self.sql(expression, "this") 882 value = self.sql(expression, "expression") 883 if value: 884 optional_equal_sign = "= " if option in OPTIONS_THAT_REQUIRE_EQUAL else "" 885 return f"{option} {optional_equal_sign}{value}" 886 return option 887 888 def lateral_op(self, expression: exp.Lateral) -> str: 889 cross_apply = expression.args.get("cross_apply") 890 if cross_apply is True: 891 return "CROSS APPLY" 892 if cross_apply is False: 893 return "OUTER APPLY" 894 895 # TODO: perhaps we can check if the parent is a Join and transpile it appropriately 896 self.unsupported("LATERAL clause is not supported.") 897 return "LATERAL" 898 899 def timefromparts_sql(self, expression: exp.TimeFromParts) -> str: 900 nano = expression.args.get("nano") 901 if nano is not None: 902 nano.pop() 903 self.unsupported("Specifying nanoseconds is not supported in TIMEFROMPARTS.") 904 905 if expression.args.get("fractions") is None: 906 expression.set("fractions", exp.Literal.number(0)) 907 if expression.args.get("precision") is None: 908 expression.set("precision", exp.Literal.number(0)) 909 910 return rename_func("TIMEFROMPARTS")(self, expression) 911 912 def timestampfromparts_sql(self, expression: exp.TimestampFromParts) -> str: 913 zone = expression.args.get("zone") 914 if zone is not None: 915 zone.pop() 916 self.unsupported("Time zone is not supported in DATETIMEFROMPARTS.") 917 918 nano = expression.args.get("nano") 919 if nano is not None: 920 nano.pop() 921 self.unsupported("Specifying nanoseconds is not supported in DATETIMEFROMPARTS.") 922 923 if expression.args.get("milli") is None: 924 expression.set("milli", exp.Literal.number(0)) 925 926 return rename_func("DATETIMEFROMPARTS")(self, expression) 927 928 def setitem_sql(self, expression: exp.SetItem) -> str: 929 this = expression.this 930 if isinstance(this, exp.EQ) and not isinstance(this.left, exp.Parameter): 931 # T-SQL does not use '=' in SET command, except when the LHS is a variable. 932 return f"{self.sql(this.left)} {self.sql(this.right)}" 933 934 return super().setitem_sql(expression) 935 936 def boolean_sql(self, expression: exp.Boolean) -> str: 937 if type(expression.parent) in BIT_TYPES: 938 return "1" if expression.this else "0" 939 940 return "(1 = 1)" if expression.this else "(1 = 0)" 941 942 def is_sql(self, expression: exp.Is) -> str: 943 if isinstance(expression.expression, exp.Boolean): 944 return self.binary(expression, "=") 945 return self.binary(expression, "IS") 946 947 def createable_sql(self, expression: exp.Create, locations: t.DefaultDict) -> str: 948 sql = self.sql(expression, "this") 949 properties = expression.args.get("properties") 950 951 if sql[:1] != "#" and any( 952 isinstance(prop, exp.TemporaryProperty) 953 for prop in (properties.expressions if properties else []) 954 ): 955 sql = f"[#{sql[1:]}" if sql.startswith("[") else f"#{sql}" 956 957 return sql 958 959 def create_sql(self, expression: exp.Create) -> str: 960 kind = expression.kind 961 exists = expression.args.pop("exists", None) 962 963 if kind == "VIEW": 964 expression.this.set("catalog", None) 965 966 sql = super().create_sql(expression) 967 968 like_property = expression.find(exp.LikeProperty) 969 if like_property: 970 ctas_expression = like_property.this 971 else: 972 ctas_expression = expression.expression 973 974 table = expression.find(exp.Table) 975 976 # Convert CTAS statement to SELECT .. INTO .. 977 if kind == "TABLE" and ctas_expression: 978 ctas_with = ctas_expression.args.get("with") 979 if ctas_with: 980 ctas_with = ctas_with.pop() 981 982 if isinstance(ctas_expression, exp.UNWRAPPED_QUERIES): 983 ctas_expression = ctas_expression.subquery() 984 985 select_into = exp.select("*").from_(exp.alias_(ctas_expression, "temp", table=True)) 986 select_into.set("into", exp.Into(this=table)) 987 select_into.set("with", ctas_with) 988 989 if like_property: 990 select_into.limit(0, copy=False) 991 992 sql = self.sql(select_into) 993 994 if exists: 995 identifier = self.sql(exp.Literal.string(exp.table_name(table) if table else "")) 996 sql = self.sql(exp.Literal.string(sql)) 997 if kind == "SCHEMA": 998 sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.schemata WHERE schema_name = {identifier}) EXEC({sql})""" 999 elif kind == "TABLE": 1000 assert table 1001 where = exp.and_( 1002 exp.column("table_name").eq(table.name), 1003 exp.column("table_schema").eq(table.db) if table.db else None, 1004 exp.column("table_catalog").eq(table.catalog) if table.catalog else None, 1005 ) 1006 sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.tables WHERE {where}) EXEC({sql})""" 1007 elif kind == "INDEX": 1008 index = self.sql(exp.Literal.string(expression.this.text("this"))) 1009 sql = f"""IF NOT EXISTS (SELECT * FROM sys.indexes WHERE object_id = object_id({identifier}) AND name = {index}) EXEC({sql})""" 1010 elif expression.args.get("replace"): 1011 sql = sql.replace("CREATE OR REPLACE ", "CREATE OR ALTER ", 1) 1012 1013 return self.prepend_ctes(expression, sql) 1014 1015 def offset_sql(self, expression: exp.Offset) -> str: 1016 return f"{super().offset_sql(expression)} ROWS" 1017 1018 def version_sql(self, expression: exp.Version) -> str: 1019 name = "SYSTEM_TIME" if expression.name == "TIMESTAMP" else expression.name 1020 this = f"FOR {name}" 1021 expr = expression.expression 1022 kind = expression.text("kind") 1023 if kind in ("FROM", "BETWEEN"): 1024 args = expr.expressions 1025 sep = "TO" if kind == "FROM" else "AND" 1026 expr_sql = f"{self.sql(seq_get(args, 0))} {sep} {self.sql(seq_get(args, 1))}" 1027 else: 1028 expr_sql = self.sql(expr) 1029 1030 expr_sql = f" {expr_sql}" if expr_sql else "" 1031 return f"{this} {kind}{expr_sql}" 1032 1033 def returnsproperty_sql(self, expression: exp.ReturnsProperty) -> str: 1034 table = expression.args.get("table") 1035 table = f"{table} " if table else "" 1036 return f"RETURNS {table}{self.sql(expression, 'this')}" 1037 1038 def returning_sql(self, expression: exp.Returning) -> str: 1039 into = self.sql(expression, "into") 1040 into = self.seg(f"INTO {into}") if into else "" 1041 return f"{self.seg('OUTPUT')} {self.expressions(expression, flat=True)}{into}" 1042 1043 def transaction_sql(self, expression: exp.Transaction) -> str: 1044 this = self.sql(expression, "this") 1045 this = f" {this}" if this else "" 1046 mark = self.sql(expression, "mark") 1047 mark = f" WITH MARK {mark}" if mark else "" 1048 return f"BEGIN TRANSACTION{this}{mark}" 1049 1050 def commit_sql(self, expression: exp.Commit) -> str: 1051 this = self.sql(expression, "this") 1052 this = f" {this}" if this else "" 1053 durability = expression.args.get("durability") 1054 durability = ( 1055 f" WITH (DELAYED_DURABILITY = {'ON' if durability else 'OFF'})" 1056 if durability is not None 1057 else "" 1058 ) 1059 return f"COMMIT TRANSACTION{this}{durability}" 1060 1061 def rollback_sql(self, expression: exp.Rollback) -> str: 1062 this = self.sql(expression, "this") 1063 this = f" {this}" if this else "" 1064 return f"ROLLBACK TRANSACTION{this}" 1065 1066 def identifier_sql(self, expression: exp.Identifier) -> str: 1067 identifier = super().identifier_sql(expression) 1068 1069 if expression.args.get("global"): 1070 identifier = f"##{identifier}" 1071 elif expression.args.get("temporary"): 1072 identifier = f"#{identifier}" 1073 1074 return identifier 1075 1076 def constraint_sql(self, expression: exp.Constraint) -> str: 1077 this = self.sql(expression, "this") 1078 expressions = self.expressions(expression, flat=True, sep=" ") 1079 return f"CONSTRAINT {this} {expressions}" 1080 1081 def length_sql(self, expression: exp.Length) -> str: 1082 return self._uncast_text(expression, "LEN") 1083 1084 def right_sql(self, expression: exp.Right) -> str: 1085 return self._uncast_text(expression, "RIGHT") 1086 1087 def left_sql(self, expression: exp.Left) -> str: 1088 return self._uncast_text(expression, "LEFT") 1089 1090 def _uncast_text(self, expression: exp.Expression, name: str) -> str: 1091 this = expression.this 1092 if isinstance(this, exp.Cast) and this.is_type(exp.DataType.Type.TEXT): 1093 this_sql = self.sql(this, "this") 1094 else: 1095 this_sql = self.sql(this) 1096 expression_sql = self.sql(expression, "expression") 1097 return self.func(name, this_sql, expression_sql if expression_sql else None) 1098 1099 def partition_sql(self, expression: exp.Partition) -> str: 1100 return f"WITH (PARTITIONS({self.expressions(expression, flat=True)}))" 1101 1102 def altertable_sql(self, expression: exp.AlterTable) -> str: 1103 action = seq_get(expression.args.get("actions") or [], 0) 1104 if isinstance(action, exp.RenameTable): 1105 return f"EXEC sp_rename '{self.sql(expression.this)}', '{action.this.name}'" 1106 return super().altertable_sql(expression) 1107 1108 def drop_sql(self, expression: exp.Drop) -> str: 1109 if expression.args["kind"] == "VIEW": 1110 expression.this.set("catalog", None) 1111 return super().drop_sql(expression) 1112 1113 def declare_sql(self, expression: exp.Declare) -> str: 1114 return f"DECLARE {self.expressions(expression, flat=True)}" 1115 1116 def declareitem_sql(self, expression: exp.DeclareItem) -> str: 1117 variable = self.sql(expression, "this") 1118 default = self.sql(expression, "default") 1119 default = f" = {default}" if default else "" 1120 1121 kind = self.sql(expression, "kind") 1122 if isinstance(expression.args.get("kind"), exp.Schema): 1123 kind = f"TABLE {kind}" 1124 1125 return f"{variable} AS {kind}{default}" 1126 1127 def options_modifier(self, expression: exp.Expression) -> str: 1128 options = self.expressions(expression, key="options") 1129 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
- DATE_PART_MAPPING
- 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
446 class Tokenizer(tokens.Tokenizer): 447 IDENTIFIERS = [("[", "]"), '"'] 448 QUOTES = ["'", '"'] 449 HEX_STRINGS = [("0x", ""), ("0X", "")] 450 VAR_SINGLE_TOKENS = {"@", "$", "#"} 451 452 KEYWORDS = { 453 **tokens.Tokenizer.KEYWORDS, 454 "CLUSTERED INDEX": TokenType.INDEX, 455 "DATETIME2": TokenType.DATETIME, 456 "DATETIMEOFFSET": TokenType.TIMESTAMPTZ, 457 "DECLARE": TokenType.DECLARE, 458 "EXEC": TokenType.COMMAND, 459 "FOR SYSTEM_TIME": TokenType.TIMESTAMP_SNAPSHOT, 460 "IMAGE": TokenType.IMAGE, 461 "MONEY": TokenType.MONEY, 462 "NONCLUSTERED INDEX": TokenType.INDEX, 463 "NTEXT": TokenType.TEXT, 464 "OPTION": TokenType.OPTION, 465 "OUTPUT": TokenType.RETURNING, 466 "PRINT": TokenType.COMMAND, 467 "PROC": TokenType.PROCEDURE, 468 "REAL": TokenType.FLOAT, 469 "ROWVERSION": TokenType.ROWVERSION, 470 "SMALLDATETIME": TokenType.DATETIME, 471 "SMALLMONEY": TokenType.SMALLMONEY, 472 "SQL_VARIANT": TokenType.VARIANT, 473 "SYSTEM_USER": TokenType.CURRENT_USER, 474 "TOP": TokenType.TOP, 475 "TIMESTAMP": TokenType.ROWVERSION, 476 "TINYINT": TokenType.UTINYINT, 477 "UNIQUEIDENTIFIER": TokenType.UNIQUEIDENTIFIER, 478 "UPDATE STATISTICS": TokenType.COMMAND, 479 "XML": TokenType.XML, 480 } 481 482 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
- WHITE_SPACE
- COMMAND_PREFIX_TOKENS
- NUMERIC_LITERALS
- COMMENTS
- dialect
- reset
- tokenize
- tokenize_rs
- size
- sql
- tokens
484 class Parser(parser.Parser): 485 SET_REQUIRES_ASSIGNMENT_DELIMITER = False 486 LOG_DEFAULTS_TO_LN = True 487 ALTER_TABLE_ADD_REQUIRED_FOR_EACH_COLUMN = False 488 STRING_ALIASES = True 489 NO_PAREN_IF_COMMANDS = False 490 491 QUERY_MODIFIER_PARSERS = { 492 **parser.Parser.QUERY_MODIFIER_PARSERS, 493 TokenType.OPTION: lambda self: ("options", self._parse_options()), 494 } 495 496 FUNCTIONS = { 497 **parser.Parser.FUNCTIONS, 498 "CHARINDEX": lambda args: exp.StrPosition( 499 this=seq_get(args, 1), 500 substr=seq_get(args, 0), 501 position=seq_get(args, 2), 502 ), 503 "DATEADD": build_date_delta(exp.DateAdd, unit_mapping=DATE_DELTA_INTERVAL), 504 "DATEDIFF": _build_date_delta(exp.DateDiff, unit_mapping=DATE_DELTA_INTERVAL), 505 "DATENAME": _build_formatted_time(exp.TimeToStr, full_format_mapping=True), 506 "DATEPART": _build_formatted_time(exp.TimeToStr), 507 "DATETIMEFROMPARTS": _build_datetimefromparts, 508 "EOMONTH": _build_eomonth, 509 "FORMAT": _build_format, 510 "GETDATE": exp.CurrentTimestamp.from_arg_list, 511 "HASHBYTES": _build_hashbytes, 512 "ISNULL": exp.Coalesce.from_arg_list, 513 "JSON_QUERY": parser.build_extract_json_with_path(exp.JSONExtract), 514 "JSON_VALUE": parser.build_extract_json_with_path(exp.JSONExtractScalar), 515 "LEN": _build_with_arg_as_text(exp.Length), 516 "LEFT": _build_with_arg_as_text(exp.Left), 517 "RIGHT": _build_with_arg_as_text(exp.Right), 518 "REPLICATE": exp.Repeat.from_arg_list, 519 "SQUARE": lambda args: exp.Pow(this=seq_get(args, 0), expression=exp.Literal.number(2)), 520 "SYSDATETIME": exp.CurrentTimestamp.from_arg_list, 521 "SUSER_NAME": exp.CurrentUser.from_arg_list, 522 "SUSER_SNAME": exp.CurrentUser.from_arg_list, 523 "SYSTEM_USER": exp.CurrentUser.from_arg_list, 524 "TIMEFROMPARTS": _build_timefromparts, 525 } 526 527 JOIN_HINTS = {"LOOP", "HASH", "MERGE", "REMOTE"} 528 529 RETURNS_TABLE_TOKENS = parser.Parser.ID_VAR_TOKENS - { 530 TokenType.TABLE, 531 *parser.Parser.TYPE_TOKENS, 532 } 533 534 STATEMENT_PARSERS = { 535 **parser.Parser.STATEMENT_PARSERS, 536 TokenType.DECLARE: lambda self: self._parse_declare(), 537 } 538 539 def _parse_options(self) -> t.Optional[t.List[exp.Expression]]: 540 if not self._match(TokenType.OPTION): 541 return None 542 543 def _parse_option() -> t.Optional[exp.Expression]: 544 option = self._parse_var_from_options(OPTIONS) 545 if not option: 546 return None 547 548 self._match(TokenType.EQ) 549 return self.expression( 550 exp.QueryOption, this=option, expression=self._parse_primary_or_var() 551 ) 552 553 return self._parse_wrapped_csv(_parse_option) 554 555 def _parse_projections(self) -> t.List[exp.Expression]: 556 """ 557 T-SQL supports the syntax alias = expression in the SELECT's projection list, 558 so we transform all parsed Selects to convert their EQ projections into Aliases. 559 560 See: https://learn.microsoft.com/en-us/sql/t-sql/queries/select-clause-transact-sql?view=sql-server-ver16#syntax 561 """ 562 return [ 563 ( 564 exp.alias_(projection.expression, projection.this.this, copy=False) 565 if isinstance(projection, exp.EQ) and isinstance(projection.this, exp.Column) 566 else projection 567 ) 568 for projection in super()._parse_projections() 569 ] 570 571 def _parse_commit_or_rollback(self) -> exp.Commit | exp.Rollback: 572 """Applies to SQL Server and Azure SQL Database 573 COMMIT [ { TRAN | TRANSACTION } 574 [ transaction_name | @tran_name_variable ] ] 575 [ WITH ( DELAYED_DURABILITY = { OFF | ON } ) ] 576 577 ROLLBACK { TRAN | TRANSACTION } 578 [ transaction_name | @tran_name_variable 579 | savepoint_name | @savepoint_variable ] 580 """ 581 rollback = self._prev.token_type == TokenType.ROLLBACK 582 583 self._match_texts(("TRAN", "TRANSACTION")) 584 this = self._parse_id_var() 585 586 if rollback: 587 return self.expression(exp.Rollback, this=this) 588 589 durability = None 590 if self._match_pair(TokenType.WITH, TokenType.L_PAREN): 591 self._match_text_seq("DELAYED_DURABILITY") 592 self._match(TokenType.EQ) 593 594 if self._match_text_seq("OFF"): 595 durability = False 596 else: 597 self._match(TokenType.ON) 598 durability = True 599 600 self._match_r_paren() 601 602 return self.expression(exp.Commit, this=this, durability=durability) 603 604 def _parse_transaction(self) -> exp.Transaction | exp.Command: 605 """Applies to SQL Server and Azure SQL Database 606 BEGIN { TRAN | TRANSACTION } 607 [ { transaction_name | @tran_name_variable } 608 [ WITH MARK [ 'description' ] ] 609 ] 610 """ 611 if self._match_texts(("TRAN", "TRANSACTION")): 612 transaction = self.expression(exp.Transaction, this=self._parse_id_var()) 613 if self._match_text_seq("WITH", "MARK"): 614 transaction.set("mark", self._parse_string()) 615 616 return transaction 617 618 return self._parse_as_command(self._prev) 619 620 def _parse_returns(self) -> exp.ReturnsProperty: 621 table = self._parse_id_var(any_token=False, tokens=self.RETURNS_TABLE_TOKENS) 622 returns = super()._parse_returns() 623 returns.set("table", table) 624 return returns 625 626 def _parse_convert( 627 self, strict: bool, safe: t.Optional[bool] = None 628 ) -> t.Optional[exp.Expression]: 629 this = self._parse_types() 630 self._match(TokenType.COMMA) 631 args = [this, *self._parse_csv(self._parse_assignment)] 632 convert = exp.Convert.from_arg_list(args) 633 convert.set("safe", safe) 634 convert.set("strict", strict) 635 return convert 636 637 def _parse_user_defined_function( 638 self, kind: t.Optional[TokenType] = None 639 ) -> t.Optional[exp.Expression]: 640 this = super()._parse_user_defined_function(kind=kind) 641 642 if ( 643 kind == TokenType.FUNCTION 644 or isinstance(this, exp.UserDefinedFunction) 645 or self._match(TokenType.ALIAS, advance=False) 646 ): 647 return this 648 649 expressions = self._parse_csv(self._parse_function_parameter) 650 return self.expression(exp.UserDefinedFunction, this=this, expressions=expressions) 651 652 def _parse_id_var( 653 self, 654 any_token: bool = True, 655 tokens: t.Optional[t.Collection[TokenType]] = None, 656 ) -> t.Optional[exp.Expression]: 657 is_temporary = self._match(TokenType.HASH) 658 is_global = is_temporary and self._match(TokenType.HASH) 659 660 this = super()._parse_id_var(any_token=any_token, tokens=tokens) 661 if this: 662 if is_global: 663 this.set("global", True) 664 elif is_temporary: 665 this.set("temporary", True) 666 667 return this 668 669 def _parse_create(self) -> exp.Create | exp.Command: 670 create = super()._parse_create() 671 672 if isinstance(create, exp.Create): 673 table = create.this.this if isinstance(create.this, exp.Schema) else create.this 674 if isinstance(table, exp.Table) and table.this.args.get("temporary"): 675 if not create.args.get("properties"): 676 create.set("properties", exp.Properties(expressions=[])) 677 678 create.args["properties"].append("expressions", exp.TemporaryProperty()) 679 680 return create 681 682 def _parse_if(self) -> t.Optional[exp.Expression]: 683 index = self._index 684 685 if self._match_text_seq("OBJECT_ID"): 686 self._parse_wrapped_csv(self._parse_string) 687 if self._match_text_seq("IS", "NOT", "NULL") and self._match(TokenType.DROP): 688 return self._parse_drop(exists=True) 689 self._retreat(index) 690 691 return super()._parse_if() 692 693 def _parse_unique(self) -> exp.UniqueColumnConstraint: 694 if self._match_texts(("CLUSTERED", "NONCLUSTERED")): 695 this = self.CONSTRAINT_PARSERS[self._prev.text.upper()](self) 696 else: 697 this = self._parse_schema(self._parse_id_var(any_token=False)) 698 699 return self.expression(exp.UniqueColumnConstraint, this=this) 700 701 def _parse_partition(self) -> t.Optional[exp.Partition]: 702 if not self._match_text_seq("WITH", "(", "PARTITIONS"): 703 return None 704 705 def parse_range(): 706 low = self._parse_bitwise() 707 high = self._parse_bitwise() if self._match_text_seq("TO") else None 708 709 return ( 710 self.expression(exp.PartitionRange, this=low, expression=high) if high else low 711 ) 712 713 partition = self.expression( 714 exp.Partition, expressions=self._parse_wrapped_csv(parse_range) 715 ) 716 717 self._match_r_paren() 718 719 return partition 720 721 def _parse_declare(self) -> exp.Declare | exp.Command: 722 index = self._index 723 expressions = self._try_parse(partial(self._parse_csv, self._parse_declareitem)) 724 725 if not expressions or self._curr: 726 self._retreat(index) 727 return self._parse_as_command(self._prev) 728 729 return self.expression(exp.Declare, expressions=expressions) 730 731 def _parse_declareitem(self) -> t.Optional[exp.DeclareItem]: 732 var = self._parse_id_var() 733 if not var: 734 return None 735 736 value = None 737 self._match(TokenType.ALIAS) 738 if self._match(TokenType.TABLE): 739 data_type = self._parse_schema() 740 else: 741 data_type = self._parse_types() 742 if self._match(TokenType.EQ): 743 value = self._parse_bitwise() 744 745 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
- COLUMN_OPERATORS
- EXPRESSION_PARSERS
- UNARY_PARSERS
- STRING_PARSERS
- NUMERIC_PARSERS
- PRIMARY_PARSERS
- PLACEHOLDER_PARSERS
- RANGE_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
- INSERT_ALTERNATIVES
- CLONE_KEYWORDS
- 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_JSON_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
747 class Generator(generator.Generator): 748 LIMIT_IS_TOP = True 749 QUERY_HINTS = False 750 RETURNING_END = False 751 NVL2_SUPPORTED = False 752 ALTER_TABLE_INCLUDE_COLUMN_KEYWORD = False 753 LIMIT_FETCH = "FETCH" 754 COMPUTED_COLUMN_WITH_TYPE = False 755 CTE_RECURSIVE_KEYWORD_REQUIRED = False 756 ENSURE_BOOLS = True 757 NULL_ORDERING_SUPPORTED = None 758 SUPPORTS_SINGLE_ARG_CONCAT = False 759 TABLESAMPLE_SEED_KEYWORD = "REPEATABLE" 760 SUPPORTS_SELECT_INTO = True 761 JSON_PATH_BRACKETED_KEY_SUPPORTED = False 762 SUPPORTS_TO_NUMBER = False 763 SET_OP_MODIFIERS = False 764 COPY_PARAMS_EQ_REQUIRED = True 765 766 EXPRESSIONS_WITHOUT_NESTED_CTES = { 767 exp.Delete, 768 exp.Insert, 769 exp.Intersect, 770 exp.Except, 771 exp.Merge, 772 exp.Select, 773 exp.Subquery, 774 exp.Union, 775 exp.Update, 776 } 777 778 SUPPORTED_JSON_PATH_PARTS = { 779 exp.JSONPathKey, 780 exp.JSONPathRoot, 781 exp.JSONPathSubscript, 782 } 783 784 TYPE_MAPPING = { 785 **generator.Generator.TYPE_MAPPING, 786 exp.DataType.Type.BOOLEAN: "BIT", 787 exp.DataType.Type.DECIMAL: "NUMERIC", 788 exp.DataType.Type.DATETIME: "DATETIME2", 789 exp.DataType.Type.DOUBLE: "FLOAT", 790 exp.DataType.Type.INT: "INTEGER", 791 exp.DataType.Type.ROWVERSION: "ROWVERSION", 792 exp.DataType.Type.TEXT: "VARCHAR(MAX)", 793 exp.DataType.Type.TIMESTAMP: "DATETIME2", 794 exp.DataType.Type.TIMESTAMPTZ: "DATETIMEOFFSET", 795 exp.DataType.Type.UTINYINT: "TINYINT", 796 exp.DataType.Type.VARIANT: "SQL_VARIANT", 797 } 798 799 TYPE_MAPPING.pop(exp.DataType.Type.NCHAR) 800 TYPE_MAPPING.pop(exp.DataType.Type.NVARCHAR) 801 802 TRANSFORMS = { 803 **generator.Generator.TRANSFORMS, 804 exp.AnyValue: any_value_to_max_sql, 805 exp.ArrayToString: rename_func("STRING_AGG"), 806 exp.AutoIncrementColumnConstraint: lambda *_: "IDENTITY", 807 exp.DateAdd: date_delta_sql("DATEADD"), 808 exp.DateDiff: date_delta_sql("DATEDIFF"), 809 exp.CTE: transforms.preprocess([qualify_derived_table_outputs]), 810 exp.CurrentDate: rename_func("GETDATE"), 811 exp.CurrentTimestamp: rename_func("GETDATE"), 812 exp.DateStrToDate: datestrtodate_sql, 813 exp.Extract: rename_func("DATEPART"), 814 exp.GeneratedAsIdentityColumnConstraint: generatedasidentitycolumnconstraint_sql, 815 exp.GroupConcat: _string_agg_sql, 816 exp.If: rename_func("IIF"), 817 exp.JSONExtract: _json_extract_sql, 818 exp.JSONExtractScalar: _json_extract_sql, 819 exp.LastDay: lambda self, e: self.func("EOMONTH", e.this), 820 exp.Max: max_or_greatest, 821 exp.MD5: lambda self, e: self.func("HASHBYTES", exp.Literal.string("MD5"), e.this), 822 exp.Min: min_or_least, 823 exp.NumberToStr: _format_sql, 824 exp.ParseJSON: lambda self, e: self.sql(e, "this"), 825 exp.Repeat: rename_func("REPLICATE"), 826 exp.Select: transforms.preprocess( 827 [ 828 transforms.eliminate_distinct_on, 829 transforms.eliminate_semi_and_anti_joins, 830 transforms.eliminate_qualify, 831 ] 832 ), 833 exp.StrPosition: lambda self, e: self.func( 834 "CHARINDEX", e.args.get("substr"), e.this, e.args.get("position") 835 ), 836 exp.Subquery: transforms.preprocess([qualify_derived_table_outputs]), 837 exp.SHA: lambda self, e: self.func("HASHBYTES", exp.Literal.string("SHA1"), e.this), 838 exp.SHA2: lambda self, e: self.func( 839 "HASHBYTES", exp.Literal.string(f"SHA2_{e.args.get('length', 256)}"), e.this 840 ), 841 exp.TemporaryProperty: lambda self, e: "", 842 exp.TimeStrToTime: lambda self, e: self.sql( 843 exp.cast(e.this, exp.DataType.Type.DATETIME) 844 ), 845 exp.TimeToStr: _format_sql, 846 exp.Trim: trim_sql, 847 exp.TsOrDsAdd: date_delta_sql("DATEADD", cast=True), 848 exp.TsOrDsDiff: date_delta_sql("DATEDIFF"), 849 } 850 851 TRANSFORMS.pop(exp.ReturnsProperty) 852 853 PROPERTIES_LOCATION = { 854 **generator.Generator.PROPERTIES_LOCATION, 855 exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED, 856 } 857 858 def select_sql(self, expression: exp.Select) -> str: 859 if expression.args.get("offset"): 860 if not expression.args.get("order"): 861 # ORDER BY is required in order to use OFFSET in a query, so we use 862 # a noop order by, since we don't really care about the order. 863 # See: https://www.microsoftpressstore.com/articles/article.aspx?p=2314819 864 expression.order_by(exp.select(exp.null()).subquery(), copy=False) 865 866 limit = expression.args.get("limit") 867 if isinstance(limit, exp.Limit): 868 # TOP and OFFSET can't be combined, we need use FETCH instead of TOP 869 # we replace here because otherwise TOP would be generated in select_sql 870 limit.replace(exp.Fetch(direction="FIRST", count=limit.expression)) 871 872 return super().select_sql(expression) 873 874 def convert_sql(self, expression: exp.Convert) -> str: 875 name = "TRY_CONVERT" if expression.args.get("safe") else "CONVERT" 876 return self.func( 877 name, expression.this, expression.expression, expression.args.get("style") 878 ) 879 880 def queryoption_sql(self, expression: exp.QueryOption) -> str: 881 option = self.sql(expression, "this") 882 value = self.sql(expression, "expression") 883 if value: 884 optional_equal_sign = "= " if option in OPTIONS_THAT_REQUIRE_EQUAL else "" 885 return f"{option} {optional_equal_sign}{value}" 886 return option 887 888 def lateral_op(self, expression: exp.Lateral) -> str: 889 cross_apply = expression.args.get("cross_apply") 890 if cross_apply is True: 891 return "CROSS APPLY" 892 if cross_apply is False: 893 return "OUTER APPLY" 894 895 # TODO: perhaps we can check if the parent is a Join and transpile it appropriately 896 self.unsupported("LATERAL clause is not supported.") 897 return "LATERAL" 898 899 def timefromparts_sql(self, expression: exp.TimeFromParts) -> str: 900 nano = expression.args.get("nano") 901 if nano is not None: 902 nano.pop() 903 self.unsupported("Specifying nanoseconds is not supported in TIMEFROMPARTS.") 904 905 if expression.args.get("fractions") is None: 906 expression.set("fractions", exp.Literal.number(0)) 907 if expression.args.get("precision") is None: 908 expression.set("precision", exp.Literal.number(0)) 909 910 return rename_func("TIMEFROMPARTS")(self, expression) 911 912 def timestampfromparts_sql(self, expression: exp.TimestampFromParts) -> str: 913 zone = expression.args.get("zone") 914 if zone is not None: 915 zone.pop() 916 self.unsupported("Time zone is not supported in DATETIMEFROMPARTS.") 917 918 nano = expression.args.get("nano") 919 if nano is not None: 920 nano.pop() 921 self.unsupported("Specifying nanoseconds is not supported in DATETIMEFROMPARTS.") 922 923 if expression.args.get("milli") is None: 924 expression.set("milli", exp.Literal.number(0)) 925 926 return rename_func("DATETIMEFROMPARTS")(self, expression) 927 928 def setitem_sql(self, expression: exp.SetItem) -> str: 929 this = expression.this 930 if isinstance(this, exp.EQ) and not isinstance(this.left, exp.Parameter): 931 # T-SQL does not use '=' in SET command, except when the LHS is a variable. 932 return f"{self.sql(this.left)} {self.sql(this.right)}" 933 934 return super().setitem_sql(expression) 935 936 def boolean_sql(self, expression: exp.Boolean) -> str: 937 if type(expression.parent) in BIT_TYPES: 938 return "1" if expression.this else "0" 939 940 return "(1 = 1)" if expression.this else "(1 = 0)" 941 942 def is_sql(self, expression: exp.Is) -> str: 943 if isinstance(expression.expression, exp.Boolean): 944 return self.binary(expression, "=") 945 return self.binary(expression, "IS") 946 947 def createable_sql(self, expression: exp.Create, locations: t.DefaultDict) -> str: 948 sql = self.sql(expression, "this") 949 properties = expression.args.get("properties") 950 951 if sql[:1] != "#" and any( 952 isinstance(prop, exp.TemporaryProperty) 953 for prop in (properties.expressions if properties else []) 954 ): 955 sql = f"[#{sql[1:]}" if sql.startswith("[") else f"#{sql}" 956 957 return sql 958 959 def create_sql(self, expression: exp.Create) -> str: 960 kind = expression.kind 961 exists = expression.args.pop("exists", None) 962 963 if kind == "VIEW": 964 expression.this.set("catalog", None) 965 966 sql = super().create_sql(expression) 967 968 like_property = expression.find(exp.LikeProperty) 969 if like_property: 970 ctas_expression = like_property.this 971 else: 972 ctas_expression = expression.expression 973 974 table = expression.find(exp.Table) 975 976 # Convert CTAS statement to SELECT .. INTO .. 977 if kind == "TABLE" and ctas_expression: 978 ctas_with = ctas_expression.args.get("with") 979 if ctas_with: 980 ctas_with = ctas_with.pop() 981 982 if isinstance(ctas_expression, exp.UNWRAPPED_QUERIES): 983 ctas_expression = ctas_expression.subquery() 984 985 select_into = exp.select("*").from_(exp.alias_(ctas_expression, "temp", table=True)) 986 select_into.set("into", exp.Into(this=table)) 987 select_into.set("with", ctas_with) 988 989 if like_property: 990 select_into.limit(0, copy=False) 991 992 sql = self.sql(select_into) 993 994 if exists: 995 identifier = self.sql(exp.Literal.string(exp.table_name(table) if table else "")) 996 sql = self.sql(exp.Literal.string(sql)) 997 if kind == "SCHEMA": 998 sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.schemata WHERE schema_name = {identifier}) EXEC({sql})""" 999 elif kind == "TABLE": 1000 assert table 1001 where = exp.and_( 1002 exp.column("table_name").eq(table.name), 1003 exp.column("table_schema").eq(table.db) if table.db else None, 1004 exp.column("table_catalog").eq(table.catalog) if table.catalog else None, 1005 ) 1006 sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.tables WHERE {where}) EXEC({sql})""" 1007 elif kind == "INDEX": 1008 index = self.sql(exp.Literal.string(expression.this.text("this"))) 1009 sql = f"""IF NOT EXISTS (SELECT * FROM sys.indexes WHERE object_id = object_id({identifier}) AND name = {index}) EXEC({sql})""" 1010 elif expression.args.get("replace"): 1011 sql = sql.replace("CREATE OR REPLACE ", "CREATE OR ALTER ", 1) 1012 1013 return self.prepend_ctes(expression, sql) 1014 1015 def offset_sql(self, expression: exp.Offset) -> str: 1016 return f"{super().offset_sql(expression)} ROWS" 1017 1018 def version_sql(self, expression: exp.Version) -> str: 1019 name = "SYSTEM_TIME" if expression.name == "TIMESTAMP" else expression.name 1020 this = f"FOR {name}" 1021 expr = expression.expression 1022 kind = expression.text("kind") 1023 if kind in ("FROM", "BETWEEN"): 1024 args = expr.expressions 1025 sep = "TO" if kind == "FROM" else "AND" 1026 expr_sql = f"{self.sql(seq_get(args, 0))} {sep} {self.sql(seq_get(args, 1))}" 1027 else: 1028 expr_sql = self.sql(expr) 1029 1030 expr_sql = f" {expr_sql}" if expr_sql else "" 1031 return f"{this} {kind}{expr_sql}" 1032 1033 def returnsproperty_sql(self, expression: exp.ReturnsProperty) -> str: 1034 table = expression.args.get("table") 1035 table = f"{table} " if table else "" 1036 return f"RETURNS {table}{self.sql(expression, 'this')}" 1037 1038 def returning_sql(self, expression: exp.Returning) -> str: 1039 into = self.sql(expression, "into") 1040 into = self.seg(f"INTO {into}") if into else "" 1041 return f"{self.seg('OUTPUT')} {self.expressions(expression, flat=True)}{into}" 1042 1043 def transaction_sql(self, expression: exp.Transaction) -> str: 1044 this = self.sql(expression, "this") 1045 this = f" {this}" if this else "" 1046 mark = self.sql(expression, "mark") 1047 mark = f" WITH MARK {mark}" if mark else "" 1048 return f"BEGIN TRANSACTION{this}{mark}" 1049 1050 def commit_sql(self, expression: exp.Commit) -> str: 1051 this = self.sql(expression, "this") 1052 this = f" {this}" if this else "" 1053 durability = expression.args.get("durability") 1054 durability = ( 1055 f" WITH (DELAYED_DURABILITY = {'ON' if durability else 'OFF'})" 1056 if durability is not None 1057 else "" 1058 ) 1059 return f"COMMIT TRANSACTION{this}{durability}" 1060 1061 def rollback_sql(self, expression: exp.Rollback) -> str: 1062 this = self.sql(expression, "this") 1063 this = f" {this}" if this else "" 1064 return f"ROLLBACK TRANSACTION{this}" 1065 1066 def identifier_sql(self, expression: exp.Identifier) -> str: 1067 identifier = super().identifier_sql(expression) 1068 1069 if expression.args.get("global"): 1070 identifier = f"##{identifier}" 1071 elif expression.args.get("temporary"): 1072 identifier = f"#{identifier}" 1073 1074 return identifier 1075 1076 def constraint_sql(self, expression: exp.Constraint) -> str: 1077 this = self.sql(expression, "this") 1078 expressions = self.expressions(expression, flat=True, sep=" ") 1079 return f"CONSTRAINT {this} {expressions}" 1080 1081 def length_sql(self, expression: exp.Length) -> str: 1082 return self._uncast_text(expression, "LEN") 1083 1084 def right_sql(self, expression: exp.Right) -> str: 1085 return self._uncast_text(expression, "RIGHT") 1086 1087 def left_sql(self, expression: exp.Left) -> str: 1088 return self._uncast_text(expression, "LEFT") 1089 1090 def _uncast_text(self, expression: exp.Expression, name: str) -> str: 1091 this = expression.this 1092 if isinstance(this, exp.Cast) and this.is_type(exp.DataType.Type.TEXT): 1093 this_sql = self.sql(this, "this") 1094 else: 1095 this_sql = self.sql(this) 1096 expression_sql = self.sql(expression, "expression") 1097 return self.func(name, this_sql, expression_sql if expression_sql else None) 1098 1099 def partition_sql(self, expression: exp.Partition) -> str: 1100 return f"WITH (PARTITIONS({self.expressions(expression, flat=True)}))" 1101 1102 def altertable_sql(self, expression: exp.AlterTable) -> str: 1103 action = seq_get(expression.args.get("actions") or [], 0) 1104 if isinstance(action, exp.RenameTable): 1105 return f"EXEC sp_rename '{self.sql(expression.this)}', '{action.this.name}'" 1106 return super().altertable_sql(expression) 1107 1108 def drop_sql(self, expression: exp.Drop) -> str: 1109 if expression.args["kind"] == "VIEW": 1110 expression.this.set("catalog", None) 1111 return super().drop_sql(expression) 1112 1113 def declare_sql(self, expression: exp.Declare) -> str: 1114 return f"DECLARE {self.expressions(expression, flat=True)}" 1115 1116 def declareitem_sql(self, expression: exp.DeclareItem) -> str: 1117 variable = self.sql(expression, "this") 1118 default = self.sql(expression, "default") 1119 default = f" = {default}" if default else "" 1120 1121 kind = self.sql(expression, "kind") 1122 if isinstance(expression.args.get("kind"), exp.Schema): 1123 kind = f"TABLE {kind}" 1124 1125 return f"{variable} AS {kind}{default}" 1126 1127 def options_modifier(self, expression: exp.Expression) -> str: 1128 options = self.expressions(expression, key="options") 1129 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
858 def select_sql(self, expression: exp.Select) -> str: 859 if expression.args.get("offset"): 860 if not expression.args.get("order"): 861 # ORDER BY is required in order to use OFFSET in a query, so we use 862 # a noop order by, since we don't really care about the order. 863 # See: https://www.microsoftpressstore.com/articles/article.aspx?p=2314819 864 expression.order_by(exp.select(exp.null()).subquery(), copy=False) 865 866 limit = expression.args.get("limit") 867 if isinstance(limit, exp.Limit): 868 # TOP and OFFSET can't be combined, we need use FETCH instead of TOP 869 # we replace here because otherwise TOP would be generated in select_sql 870 limit.replace(exp.Fetch(direction="FIRST", count=limit.expression)) 871 872 return super().select_sql(expression)
880 def queryoption_sql(self, expression: exp.QueryOption) -> str: 881 option = self.sql(expression, "this") 882 value = self.sql(expression, "expression") 883 if value: 884 optional_equal_sign = "= " if option in OPTIONS_THAT_REQUIRE_EQUAL else "" 885 return f"{option} {optional_equal_sign}{value}" 886 return option
888 def lateral_op(self, expression: exp.Lateral) -> str: 889 cross_apply = expression.args.get("cross_apply") 890 if cross_apply is True: 891 return "CROSS APPLY" 892 if cross_apply is False: 893 return "OUTER APPLY" 894 895 # TODO: perhaps we can check if the parent is a Join and transpile it appropriately 896 self.unsupported("LATERAL clause is not supported.") 897 return "LATERAL"
899 def timefromparts_sql(self, expression: exp.TimeFromParts) -> str: 900 nano = expression.args.get("nano") 901 if nano is not None: 902 nano.pop() 903 self.unsupported("Specifying nanoseconds is not supported in TIMEFROMPARTS.") 904 905 if expression.args.get("fractions") is None: 906 expression.set("fractions", exp.Literal.number(0)) 907 if expression.args.get("precision") is None: 908 expression.set("precision", exp.Literal.number(0)) 909 910 return rename_func("TIMEFROMPARTS")(self, expression)
912 def timestampfromparts_sql(self, expression: exp.TimestampFromParts) -> str: 913 zone = expression.args.get("zone") 914 if zone is not None: 915 zone.pop() 916 self.unsupported("Time zone is not supported in DATETIMEFROMPARTS.") 917 918 nano = expression.args.get("nano") 919 if nano is not None: 920 nano.pop() 921 self.unsupported("Specifying nanoseconds is not supported in DATETIMEFROMPARTS.") 922 923 if expression.args.get("milli") is None: 924 expression.set("milli", exp.Literal.number(0)) 925 926 return rename_func("DATETIMEFROMPARTS")(self, expression)
928 def setitem_sql(self, expression: exp.SetItem) -> str: 929 this = expression.this 930 if isinstance(this, exp.EQ) and not isinstance(this.left, exp.Parameter): 931 # T-SQL does not use '=' in SET command, except when the LHS is a variable. 932 return f"{self.sql(this.left)} {self.sql(this.right)}" 933 934 return super().setitem_sql(expression)
947 def createable_sql(self, expression: exp.Create, locations: t.DefaultDict) -> str: 948 sql = self.sql(expression, "this") 949 properties = expression.args.get("properties") 950 951 if sql[:1] != "#" and any( 952 isinstance(prop, exp.TemporaryProperty) 953 for prop in (properties.expressions if properties else []) 954 ): 955 sql = f"[#{sql[1:]}" if sql.startswith("[") else f"#{sql}" 956 957 return sql
959 def create_sql(self, expression: exp.Create) -> str: 960 kind = expression.kind 961 exists = expression.args.pop("exists", None) 962 963 if kind == "VIEW": 964 expression.this.set("catalog", None) 965 966 sql = super().create_sql(expression) 967 968 like_property = expression.find(exp.LikeProperty) 969 if like_property: 970 ctas_expression = like_property.this 971 else: 972 ctas_expression = expression.expression 973 974 table = expression.find(exp.Table) 975 976 # Convert CTAS statement to SELECT .. INTO .. 977 if kind == "TABLE" and ctas_expression: 978 ctas_with = ctas_expression.args.get("with") 979 if ctas_with: 980 ctas_with = ctas_with.pop() 981 982 if isinstance(ctas_expression, exp.UNWRAPPED_QUERIES): 983 ctas_expression = ctas_expression.subquery() 984 985 select_into = exp.select("*").from_(exp.alias_(ctas_expression, "temp", table=True)) 986 select_into.set("into", exp.Into(this=table)) 987 select_into.set("with", ctas_with) 988 989 if like_property: 990 select_into.limit(0, copy=False) 991 992 sql = self.sql(select_into) 993 994 if exists: 995 identifier = self.sql(exp.Literal.string(exp.table_name(table) if table else "")) 996 sql = self.sql(exp.Literal.string(sql)) 997 if kind == "SCHEMA": 998 sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.schemata WHERE schema_name = {identifier}) EXEC({sql})""" 999 elif kind == "TABLE": 1000 assert table 1001 where = exp.and_( 1002 exp.column("table_name").eq(table.name), 1003 exp.column("table_schema").eq(table.db) if table.db else None, 1004 exp.column("table_catalog").eq(table.catalog) if table.catalog else None, 1005 ) 1006 sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.tables WHERE {where}) EXEC({sql})""" 1007 elif kind == "INDEX": 1008 index = self.sql(exp.Literal.string(expression.this.text("this"))) 1009 sql = f"""IF NOT EXISTS (SELECT * FROM sys.indexes WHERE object_id = object_id({identifier}) AND name = {index}) EXEC({sql})""" 1010 elif expression.args.get("replace"): 1011 sql = sql.replace("CREATE OR REPLACE ", "CREATE OR ALTER ", 1) 1012 1013 return self.prepend_ctes(expression, sql)
1018 def version_sql(self, expression: exp.Version) -> str: 1019 name = "SYSTEM_TIME" if expression.name == "TIMESTAMP" else expression.name 1020 this = f"FOR {name}" 1021 expr = expression.expression 1022 kind = expression.text("kind") 1023 if kind in ("FROM", "BETWEEN"): 1024 args = expr.expressions 1025 sep = "TO" if kind == "FROM" else "AND" 1026 expr_sql = f"{self.sql(seq_get(args, 0))} {sep} {self.sql(seq_get(args, 1))}" 1027 else: 1028 expr_sql = self.sql(expr) 1029 1030 expr_sql = f" {expr_sql}" if expr_sql else "" 1031 return f"{this} {kind}{expr_sql}"
1050 def commit_sql(self, expression: exp.Commit) -> str: 1051 this = self.sql(expression, "this") 1052 this = f" {this}" if this else "" 1053 durability = expression.args.get("durability") 1054 durability = ( 1055 f" WITH (DELAYED_DURABILITY = {'ON' if durability else 'OFF'})" 1056 if durability is not None 1057 else "" 1058 ) 1059 return f"COMMIT TRANSACTION{this}{durability}"
1066 def identifier_sql(self, expression: exp.Identifier) -> str: 1067 identifier = super().identifier_sql(expression) 1068 1069 if expression.args.get("global"): 1070 identifier = f"##{identifier}" 1071 elif expression.args.get("temporary"): 1072 identifier = f"#{identifier}" 1073 1074 return identifier
1102 def altertable_sql(self, expression: exp.AlterTable) -> str: 1103 action = seq_get(expression.args.get("actions") or [], 0) 1104 if isinstance(action, exp.RenameTable): 1105 return f"EXEC sp_rename '{self.sql(expression.this)}', '{action.this.name}'" 1106 return super().altertable_sql(expression)
1116 def declareitem_sql(self, expression: exp.DeclareItem) -> str: 1117 variable = self.sql(expression, "this") 1118 default = self.sql(expression, "default") 1119 default = f" = {default}" if default else "" 1120 1121 kind = self.sql(expression, "kind") 1122 if isinstance(expression.args.get("kind"), exp.Schema): 1123 kind = f"TABLE {kind}" 1124 1125 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
- 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
- currenttimestamp_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
- generateseries_sql
- struct_sql
- partitionrange_sql
- truncatetable_sql
- copyparameter_sql
- credentials_sql
- copy_sql
- semicolon_sql
- datadeletionproperty_sql
- maskingpolicycolumnconstraint_sql
- gapfill_sql