sqlglot.dialects.tsql

  1from __future__ import annotations
  2
  3import datetime
  4import re
  5import typing as t
  6
  7from sqlglot import exp, generator, parser, tokens, transforms
  8from sqlglot.dialects.dialect import (
  9    Dialect,
 10    NormalizationStrategy,
 11    any_value_to_max_sql,
 12    date_delta_sql,
 13    generatedasidentitycolumnconstraint_sql,
 14    max_or_greatest,
 15    min_or_least,
 16    parse_date_delta,
 17    rename_func,
 18    timestrtotime_sql,
 19    trim_sql,
 20)
 21from sqlglot.expressions import DataType
 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
 67def _format_time_lambda(
 68    exp_class: t.Type[E], full_format_mapping: t.Optional[bool] = None
 69) -> t.Callable[[t.List], E]:
 70    def _format_time(args: t.List) -> E:
 71        assert len(args) == 2
 72
 73        return exp_class(
 74            this=exp.cast(args[1], "datetime"),
 75            format=exp.Literal.string(
 76                format_time(
 77                    args[0].name.lower(),
 78                    (
 79                        {**TSQL.TIME_MAPPING, **FULL_FORMAT_TIME_MAPPING}
 80                        if full_format_mapping
 81                        else TSQL.TIME_MAPPING
 82                    ),
 83                )
 84            ),
 85        )
 86
 87    return _format_time
 88
 89
 90def _parse_format(args: t.List) -> exp.Expression:
 91    this = seq_get(args, 0)
 92    fmt = seq_get(args, 1)
 93    culture = seq_get(args, 2)
 94
 95    number_fmt = fmt and (fmt.name in TRANSPILE_SAFE_NUMBER_FMT or not DATE_FMT_RE.search(fmt.name))
 96
 97    if number_fmt:
 98        return exp.NumberToStr(this=this, format=fmt, culture=culture)
 99
100    if fmt:
101        fmt = exp.Literal.string(
102            format_time(fmt.name, TSQL.FORMAT_TIME_MAPPING)
103            if len(fmt.name) == 1
104            else format_time(fmt.name, TSQL.TIME_MAPPING)
105        )
106
107    return exp.TimeToStr(this=this, format=fmt, culture=culture)
108
109
110def _parse_eomonth(args: t.List) -> exp.LastDay:
111    date = exp.TsOrDsToDate(this=seq_get(args, 0))
112    month_lag = seq_get(args, 1)
113
114    if month_lag is None:
115        this: exp.Expression = date
116    else:
117        unit = DATE_DELTA_INTERVAL.get("month")
118        this = exp.DateAdd(this=date, expression=month_lag, unit=unit and exp.var(unit))
119
120    return exp.LastDay(this=this)
121
122
123def _parse_hashbytes(args: t.List) -> exp.Expression:
124    kind, data = args
125    kind = kind.name.upper() if kind.is_string else ""
126
127    if kind == "MD5":
128        args.pop(0)
129        return exp.MD5(this=data)
130    if kind in ("SHA", "SHA1"):
131        args.pop(0)
132        return exp.SHA(this=data)
133    if kind == "SHA2_256":
134        return exp.SHA2(this=data, length=exp.Literal.number(256))
135    if kind == "SHA2_512":
136        return exp.SHA2(this=data, length=exp.Literal.number(512))
137
138    return exp.func("HASHBYTES", *args)
139
140
141DATEPART_ONLY_FORMATS = {"DW", "HOUR", "QUARTER"}
142
143
144def _format_sql(self: TSQL.Generator, expression: exp.NumberToStr | exp.TimeToStr) -> str:
145    fmt = expression.args["format"]
146
147    if not isinstance(expression, exp.NumberToStr):
148        if fmt.is_string:
149            mapped_fmt = format_time(fmt.name, TSQL.INVERSE_TIME_MAPPING)
150
151            name = (mapped_fmt or "").upper()
152            if name in DATEPART_ONLY_FORMATS:
153                return self.func("DATEPART", name, expression.this)
154
155            fmt_sql = self.sql(exp.Literal.string(mapped_fmt))
156        else:
157            fmt_sql = self.format_time(expression) or self.sql(fmt)
158    else:
159        fmt_sql = self.sql(fmt)
160
161    return self.func("FORMAT", expression.this, fmt_sql, expression.args.get("culture"))
162
163
164def _string_agg_sql(self: TSQL.Generator, expression: exp.GroupConcat) -> str:
165    this = expression.this
166    distinct = expression.find(exp.Distinct)
167    if distinct:
168        # exp.Distinct can appear below an exp.Order or an exp.GroupConcat expression
169        self.unsupported("T-SQL STRING_AGG doesn't support DISTINCT.")
170        this = distinct.pop().expressions[0]
171
172    order = ""
173    if isinstance(expression.this, exp.Order):
174        if expression.this.this:
175            this = expression.this.this.pop()
176        order = f" WITHIN GROUP ({self.sql(expression.this)[1:]})"  # Order has a leading space
177
178    separator = expression.args.get("separator") or exp.Literal.string(",")
179    return f"STRING_AGG({self.format_args(this, separator)}){order}"
180
181
182def _parse_date_delta(
183    exp_class: t.Type[E], unit_mapping: t.Optional[t.Dict[str, str]] = None
184) -> t.Callable[[t.List], E]:
185    def inner_func(args: t.List) -> E:
186        unit = seq_get(args, 0)
187        if unit and unit_mapping:
188            unit = exp.var(unit_mapping.get(unit.name.lower(), unit.name))
189
190        start_date = seq_get(args, 1)
191        if start_date and start_date.is_number:
192            # Numeric types are valid DATETIME values
193            if start_date.is_int:
194                adds = DEFAULT_START_DATE + datetime.timedelta(days=int(start_date.this))
195                start_date = exp.Literal.string(adds.strftime("%F"))
196            else:
197                # We currently don't handle float values, i.e. they're not converted to equivalent DATETIMEs.
198                # This is not a problem when generating T-SQL code, it is when transpiling to other dialects.
199                return exp_class(this=seq_get(args, 2), expression=start_date, unit=unit)
200
201        return exp_class(
202            this=exp.TimeStrToTime(this=seq_get(args, 2)),
203            expression=exp.TimeStrToTime(this=start_date),
204            unit=unit,
205        )
206
207    return inner_func
208
209
210def qualify_derived_table_outputs(expression: exp.Expression) -> exp.Expression:
211    """Ensures all (unnamed) output columns are aliased for CTEs and Subqueries."""
212    alias = expression.args.get("alias")
213
214    if (
215        isinstance(expression, (exp.CTE, exp.Subquery))
216        and isinstance(alias, exp.TableAlias)
217        and not alias.columns
218    ):
219        from sqlglot.optimizer.qualify_columns import qualify_outputs
220
221        # We keep track of the unaliased column projection indexes instead of the expressions
222        # themselves, because the latter are going to be replaced by new nodes when the aliases
223        # are added and hence we won't be able to reach these newly added Alias parents
224        subqueryable = expression.this
225        unaliased_column_indexes = (
226            i
227            for i, c in enumerate(subqueryable.selects)
228            if isinstance(c, exp.Column) and not c.alias
229        )
230
231        qualify_outputs(subqueryable)
232
233        # Preserve the quoting information of columns for newly added Alias nodes
234        subqueryable_selects = subqueryable.selects
235        for select_index in unaliased_column_indexes:
236            alias = subqueryable_selects[select_index]
237            column = alias.this
238            if isinstance(column.this, exp.Identifier):
239                alias.args["alias"].set("quoted", column.this.quoted)
240
241    return expression
242
243
244# https://learn.microsoft.com/en-us/sql/t-sql/functions/datetimefromparts-transact-sql?view=sql-server-ver16#syntax
245def _parse_datetimefromparts(args: t.List) -> exp.TimestampFromParts:
246    return exp.TimestampFromParts(
247        year=seq_get(args, 0),
248        month=seq_get(args, 1),
249        day=seq_get(args, 2),
250        hour=seq_get(args, 3),
251        min=seq_get(args, 4),
252        sec=seq_get(args, 5),
253        milli=seq_get(args, 6),
254    )
255
256
257# https://learn.microsoft.com/en-us/sql/t-sql/functions/timefromparts-transact-sql?view=sql-server-ver16#syntax
258def _parse_timefromparts(args: t.List) -> exp.TimeFromParts:
259    return exp.TimeFromParts(
260        hour=seq_get(args, 0),
261        min=seq_get(args, 1),
262        sec=seq_get(args, 2),
263        fractions=seq_get(args, 3),
264        precision=seq_get(args, 4),
265    )
266
267
268def _parse_as_text(
269    klass: t.Type[exp.Expression],
270) -> t.Callable[[t.List[exp.Expression]], exp.Expression]:
271    def _parse(args: t.List[exp.Expression]) -> exp.Expression:
272        this = seq_get(args, 0)
273
274        if this and not this.is_string:
275            this = exp.cast(this, exp.DataType.Type.TEXT)
276
277        expression = seq_get(args, 1)
278        kwargs = {"this": this}
279
280        if expression:
281            kwargs["expression"] = expression
282
283        return klass(**kwargs)
284
285    return _parse
286
287
288def _json_extract_sql(
289    self: TSQL.Generator, expression: exp.JSONExtract | exp.JSONExtractScalar
290) -> str:
291    json_query = rename_func("JSON_QUERY")(self, expression)
292    json_value = rename_func("JSON_VALUE")(self, expression)
293    return self.func("ISNULL", json_query, json_value)
294
295
296class TSQL(Dialect):
297    NORMALIZATION_STRATEGY = NormalizationStrategy.CASE_INSENSITIVE
298    TIME_FORMAT = "'yyyy-mm-dd hh:mm:ss'"
299    SUPPORTS_SEMI_ANTI_JOIN = False
300    LOG_BASE_FIRST = False
301    TYPED_DIVISION = True
302    CONCAT_COALESCE = True
303
304    TIME_MAPPING = {
305        "year": "%Y",
306        "dayofyear": "%j",
307        "day": "%d",
308        "dy": "%d",
309        "y": "%Y",
310        "week": "%W",
311        "ww": "%W",
312        "wk": "%W",
313        "hour": "%h",
314        "hh": "%I",
315        "minute": "%M",
316        "mi": "%M",
317        "n": "%M",
318        "second": "%S",
319        "ss": "%S",
320        "s": "%-S",
321        "millisecond": "%f",
322        "ms": "%f",
323        "weekday": "%W",
324        "dw": "%W",
325        "month": "%m",
326        "mm": "%M",
327        "m": "%-M",
328        "Y": "%Y",
329        "YYYY": "%Y",
330        "YY": "%y",
331        "MMMM": "%B",
332        "MMM": "%b",
333        "MM": "%m",
334        "M": "%-m",
335        "dddd": "%A",
336        "dd": "%d",
337        "d": "%-d",
338        "HH": "%H",
339        "H": "%-H",
340        "h": "%-I",
341        "S": "%f",
342        "yyyy": "%Y",
343        "yy": "%y",
344    }
345
346    CONVERT_FORMAT_MAPPING = {
347        "0": "%b %d %Y %-I:%M%p",
348        "1": "%m/%d/%y",
349        "2": "%y.%m.%d",
350        "3": "%d/%m/%y",
351        "4": "%d.%m.%y",
352        "5": "%d-%m-%y",
353        "6": "%d %b %y",
354        "7": "%b %d, %y",
355        "8": "%H:%M:%S",
356        "9": "%b %d %Y %-I:%M:%S:%f%p",
357        "10": "mm-dd-yy",
358        "11": "yy/mm/dd",
359        "12": "yymmdd",
360        "13": "%d %b %Y %H:%M:ss:%f",
361        "14": "%H:%M:%S:%f",
362        "20": "%Y-%m-%d %H:%M:%S",
363        "21": "%Y-%m-%d %H:%M:%S.%f",
364        "22": "%m/%d/%y %-I:%M:%S %p",
365        "23": "%Y-%m-%d",
366        "24": "%H:%M:%S",
367        "25": "%Y-%m-%d %H:%M:%S.%f",
368        "100": "%b %d %Y %-I:%M%p",
369        "101": "%m/%d/%Y",
370        "102": "%Y.%m.%d",
371        "103": "%d/%m/%Y",
372        "104": "%d.%m.%Y",
373        "105": "%d-%m-%Y",
374        "106": "%d %b %Y",
375        "107": "%b %d, %Y",
376        "108": "%H:%M:%S",
377        "109": "%b %d %Y %-I:%M:%S:%f%p",
378        "110": "%m-%d-%Y",
379        "111": "%Y/%m/%d",
380        "112": "%Y%m%d",
381        "113": "%d %b %Y %H:%M:%S:%f",
382        "114": "%H:%M:%S:%f",
383        "120": "%Y-%m-%d %H:%M:%S",
384        "121": "%Y-%m-%d %H:%M:%S.%f",
385    }
386
387    FORMAT_TIME_MAPPING = {
388        "y": "%B %Y",
389        "d": "%m/%d/%Y",
390        "H": "%-H",
391        "h": "%-I",
392        "s": "%Y-%m-%d %H:%M:%S",
393        "D": "%A,%B,%Y",
394        "f": "%A,%B,%Y %-I:%M %p",
395        "F": "%A,%B,%Y %-I:%M:%S %p",
396        "g": "%m/%d/%Y %-I:%M %p",
397        "G": "%m/%d/%Y %-I:%M:%S %p",
398        "M": "%B %-d",
399        "m": "%B %-d",
400        "O": "%Y-%m-%dT%H:%M:%S",
401        "u": "%Y-%M-%D %H:%M:%S%z",
402        "U": "%A, %B %D, %Y %H:%M:%S%z",
403        "T": "%-I:%M:%S %p",
404        "t": "%-I:%M",
405        "Y": "%a %Y",
406    }
407
408    class Tokenizer(tokens.Tokenizer):
409        IDENTIFIERS = [("[", "]"), '"']
410        QUOTES = ["'", '"']
411        HEX_STRINGS = [("0x", ""), ("0X", "")]
412        VAR_SINGLE_TOKENS = {"@", "$", "#"}
413
414        KEYWORDS = {
415            **tokens.Tokenizer.KEYWORDS,
416            "DATETIME2": TokenType.DATETIME,
417            "DATETIMEOFFSET": TokenType.TIMESTAMPTZ,
418            "DECLARE": TokenType.COMMAND,
419            "EXEC": TokenType.COMMAND,
420            "IMAGE": TokenType.IMAGE,
421            "MONEY": TokenType.MONEY,
422            "NTEXT": TokenType.TEXT,
423            "NVARCHAR(MAX)": TokenType.TEXT,
424            "PRINT": TokenType.COMMAND,
425            "PROC": TokenType.PROCEDURE,
426            "REAL": TokenType.FLOAT,
427            "ROWVERSION": TokenType.ROWVERSION,
428            "SMALLDATETIME": TokenType.DATETIME,
429            "SMALLMONEY": TokenType.SMALLMONEY,
430            "SQL_VARIANT": TokenType.VARIANT,
431            "TOP": TokenType.TOP,
432            "UNIQUEIDENTIFIER": TokenType.UNIQUEIDENTIFIER,
433            "UPDATE STATISTICS": TokenType.COMMAND,
434            "VARCHAR(MAX)": TokenType.TEXT,
435            "XML": TokenType.XML,
436            "OUTPUT": TokenType.RETURNING,
437            "SYSTEM_USER": TokenType.CURRENT_USER,
438            "FOR SYSTEM_TIME": TokenType.TIMESTAMP_SNAPSHOT,
439        }
440
441    class Parser(parser.Parser):
442        SET_REQUIRES_ASSIGNMENT_DELIMITER = False
443
444        FUNCTIONS = {
445            **parser.Parser.FUNCTIONS,
446            "CHARINDEX": lambda args: exp.StrPosition(
447                this=seq_get(args, 1),
448                substr=seq_get(args, 0),
449                position=seq_get(args, 2),
450            ),
451            "DATEADD": parse_date_delta(exp.DateAdd, unit_mapping=DATE_DELTA_INTERVAL),
452            "DATEDIFF": _parse_date_delta(exp.DateDiff, unit_mapping=DATE_DELTA_INTERVAL),
453            "DATENAME": _format_time_lambda(exp.TimeToStr, full_format_mapping=True),
454            "DATEPART": _format_time_lambda(exp.TimeToStr),
455            "DATETIMEFROMPARTS": _parse_datetimefromparts,
456            "EOMONTH": _parse_eomonth,
457            "FORMAT": _parse_format,
458            "GETDATE": exp.CurrentTimestamp.from_arg_list,
459            "HASHBYTES": _parse_hashbytes,
460            "ISNULL": exp.Coalesce.from_arg_list,
461            "JSON_QUERY": parser.parse_extract_json_with_path(exp.JSONExtract),
462            "JSON_VALUE": parser.parse_extract_json_with_path(exp.JSONExtractScalar),
463            "LEN": _parse_as_text(exp.Length),
464            "LEFT": _parse_as_text(exp.Left),
465            "RIGHT": _parse_as_text(exp.Right),
466            "REPLICATE": exp.Repeat.from_arg_list,
467            "SQUARE": lambda args: exp.Pow(this=seq_get(args, 0), expression=exp.Literal.number(2)),
468            "SYSDATETIME": exp.CurrentTimestamp.from_arg_list,
469            "SUSER_NAME": exp.CurrentUser.from_arg_list,
470            "SUSER_SNAME": exp.CurrentUser.from_arg_list,
471            "SYSTEM_USER": exp.CurrentUser.from_arg_list,
472            "TIMEFROMPARTS": _parse_timefromparts,
473        }
474
475        JOIN_HINTS = {
476            "LOOP",
477            "HASH",
478            "MERGE",
479            "REMOTE",
480        }
481
482        VAR_LENGTH_DATATYPES = {
483            DataType.Type.NVARCHAR,
484            DataType.Type.VARCHAR,
485            DataType.Type.CHAR,
486            DataType.Type.NCHAR,
487        }
488
489        RETURNS_TABLE_TOKENS = parser.Parser.ID_VAR_TOKENS - {
490            TokenType.TABLE,
491            *parser.Parser.TYPE_TOKENS,
492        }
493
494        STATEMENT_PARSERS = {
495            **parser.Parser.STATEMENT_PARSERS,
496            TokenType.END: lambda self: self._parse_command(),
497        }
498
499        LOG_DEFAULTS_TO_LN = True
500
501        ALTER_TABLE_ADD_REQUIRED_FOR_EACH_COLUMN = False
502        STRING_ALIASES = True
503        NO_PAREN_IF_COMMANDS = False
504
505        def _parse_projections(self) -> t.List[exp.Expression]:
506            """
507            T-SQL supports the syntax alias = expression in the SELECT's projection list,
508            so we transform all parsed Selects to convert their EQ projections into Aliases.
509
510            See: https://learn.microsoft.com/en-us/sql/t-sql/queries/select-clause-transact-sql?view=sql-server-ver16#syntax
511            """
512            return [
513                (
514                    exp.alias_(projection.expression, projection.this.this, copy=False)
515                    if isinstance(projection, exp.EQ) and isinstance(projection.this, exp.Column)
516                    else projection
517                )
518                for projection in super()._parse_projections()
519            ]
520
521        def _parse_commit_or_rollback(self) -> exp.Commit | exp.Rollback:
522            """Applies to SQL Server and Azure SQL Database
523            COMMIT [ { TRAN | TRANSACTION }
524                [ transaction_name | @tran_name_variable ] ]
525                [ WITH ( DELAYED_DURABILITY = { OFF | ON } ) ]
526
527            ROLLBACK { TRAN | TRANSACTION }
528                [ transaction_name | @tran_name_variable
529                | savepoint_name | @savepoint_variable ]
530            """
531            rollback = self._prev.token_type == TokenType.ROLLBACK
532
533            self._match_texts(("TRAN", "TRANSACTION"))
534            this = self._parse_id_var()
535
536            if rollback:
537                return self.expression(exp.Rollback, this=this)
538
539            durability = None
540            if self._match_pair(TokenType.WITH, TokenType.L_PAREN):
541                self._match_text_seq("DELAYED_DURABILITY")
542                self._match(TokenType.EQ)
543
544                if self._match_text_seq("OFF"):
545                    durability = False
546                else:
547                    self._match(TokenType.ON)
548                    durability = True
549
550                self._match_r_paren()
551
552            return self.expression(exp.Commit, this=this, durability=durability)
553
554        def _parse_transaction(self) -> exp.Transaction | exp.Command:
555            """Applies to SQL Server and Azure SQL Database
556            BEGIN { TRAN | TRANSACTION }
557            [ { transaction_name | @tran_name_variable }
558            [ WITH MARK [ 'description' ] ]
559            ]
560            """
561            if self._match_texts(("TRAN", "TRANSACTION")):
562                transaction = self.expression(exp.Transaction, this=self._parse_id_var())
563                if self._match_text_seq("WITH", "MARK"):
564                    transaction.set("mark", self._parse_string())
565
566                return transaction
567
568            return self._parse_as_command(self._prev)
569
570        def _parse_returns(self) -> exp.ReturnsProperty:
571            table = self._parse_id_var(any_token=False, tokens=self.RETURNS_TABLE_TOKENS)
572            returns = super()._parse_returns()
573            returns.set("table", table)
574            return returns
575
576        def _parse_convert(
577            self, strict: bool, safe: t.Optional[bool] = None
578        ) -> t.Optional[exp.Expression]:
579            to = self._parse_types()
580            self._match(TokenType.COMMA)
581            this = self._parse_conjunction()
582
583            if not to or not this:
584                return None
585
586            # Retrieve length of datatype and override to default if not specified
587            if seq_get(to.expressions, 0) is None and to.this in self.VAR_LENGTH_DATATYPES:
588                to = exp.DataType.build(to.this, expressions=[exp.Literal.number(30)], nested=False)
589
590            # Check whether a conversion with format is applicable
591            if self._match(TokenType.COMMA):
592                format_val = self._parse_number()
593                format_val_name = format_val.name if format_val else ""
594
595                if format_val_name not in TSQL.CONVERT_FORMAT_MAPPING:
596                    raise ValueError(
597                        f"CONVERT function at T-SQL does not support format style {format_val_name}"
598                    )
599
600                format_norm = exp.Literal.string(TSQL.CONVERT_FORMAT_MAPPING[format_val_name])
601
602                # Check whether the convert entails a string to date format
603                if to.this == DataType.Type.DATE:
604                    return self.expression(exp.StrToDate, this=this, format=format_norm)
605                # Check whether the convert entails a string to datetime format
606                elif to.this == DataType.Type.DATETIME:
607                    return self.expression(exp.StrToTime, this=this, format=format_norm)
608                # Check whether the convert entails a date to string format
609                elif to.this in self.VAR_LENGTH_DATATYPES:
610                    return self.expression(
611                        exp.Cast if strict else exp.TryCast,
612                        to=to,
613                        this=self.expression(exp.TimeToStr, this=this, format=format_norm),
614                        safe=safe,
615                    )
616                elif to.this == DataType.Type.TEXT:
617                    return self.expression(exp.TimeToStr, this=this, format=format_norm)
618
619            # Entails a simple cast without any format requirement
620            return self.expression(exp.Cast if strict else exp.TryCast, this=this, to=to, safe=safe)
621
622        def _parse_user_defined_function(
623            self, kind: t.Optional[TokenType] = None
624        ) -> t.Optional[exp.Expression]:
625            this = super()._parse_user_defined_function(kind=kind)
626
627            if (
628                kind == TokenType.FUNCTION
629                or isinstance(this, exp.UserDefinedFunction)
630                or self._match(TokenType.ALIAS, advance=False)
631            ):
632                return this
633
634            expressions = self._parse_csv(self._parse_function_parameter)
635            return self.expression(exp.UserDefinedFunction, this=this, expressions=expressions)
636
637        def _parse_id_var(
638            self,
639            any_token: bool = True,
640            tokens: t.Optional[t.Collection[TokenType]] = None,
641        ) -> t.Optional[exp.Expression]:
642            is_temporary = self._match(TokenType.HASH)
643            is_global = is_temporary and self._match(TokenType.HASH)
644
645            this = super()._parse_id_var(any_token=any_token, tokens=tokens)
646            if this:
647                if is_global:
648                    this.set("global", True)
649                elif is_temporary:
650                    this.set("temporary", True)
651
652            return this
653
654        def _parse_create(self) -> exp.Create | exp.Command:
655            create = super()._parse_create()
656
657            if isinstance(create, exp.Create):
658                table = create.this.this if isinstance(create.this, exp.Schema) else create.this
659                if isinstance(table, exp.Table) and table.this.args.get("temporary"):
660                    if not create.args.get("properties"):
661                        create.set("properties", exp.Properties(expressions=[]))
662
663                    create.args["properties"].append("expressions", exp.TemporaryProperty())
664
665            return create
666
667        def _parse_if(self) -> t.Optional[exp.Expression]:
668            index = self._index
669
670            if self._match_text_seq("OBJECT_ID"):
671                self._parse_wrapped_csv(self._parse_string)
672                if self._match_text_seq("IS", "NOT", "NULL") and self._match(TokenType.DROP):
673                    return self._parse_drop(exists=True)
674                self._retreat(index)
675
676            return super()._parse_if()
677
678        def _parse_unique(self) -> exp.UniqueColumnConstraint:
679            if self._match_texts(("CLUSTERED", "NONCLUSTERED")):
680                this = self.CONSTRAINT_PARSERS[self._prev.text.upper()](self)
681            else:
682                this = self._parse_schema(self._parse_id_var(any_token=False))
683
684            return self.expression(exp.UniqueColumnConstraint, this=this)
685
686    class Generator(generator.Generator):
687        LIMIT_IS_TOP = True
688        QUERY_HINTS = False
689        RETURNING_END = False
690        NVL2_SUPPORTED = False
691        ALTER_TABLE_INCLUDE_COLUMN_KEYWORD = False
692        LIMIT_FETCH = "FETCH"
693        COMPUTED_COLUMN_WITH_TYPE = False
694        CTE_RECURSIVE_KEYWORD_REQUIRED = False
695        ENSURE_BOOLS = True
696        NULL_ORDERING_SUPPORTED = None
697        SUPPORTS_SINGLE_ARG_CONCAT = False
698        TABLESAMPLE_SEED_KEYWORD = "REPEATABLE"
699        SUPPORTS_SELECT_INTO = True
700        JSON_PATH_BRACKETED_KEY_SUPPORTED = False
701
702        EXPRESSIONS_WITHOUT_NESTED_CTES = {
703            exp.Delete,
704            exp.Insert,
705            exp.Merge,
706            exp.Select,
707            exp.Subquery,
708            exp.Union,
709            exp.Update,
710        }
711
712        SUPPORTED_JSON_PATH_PARTS = {
713            exp.JSONPathKey,
714            exp.JSONPathRoot,
715            exp.JSONPathSubscript,
716        }
717
718        TYPE_MAPPING = {
719            **generator.Generator.TYPE_MAPPING,
720            exp.DataType.Type.BOOLEAN: "BIT",
721            exp.DataType.Type.DECIMAL: "NUMERIC",
722            exp.DataType.Type.DATETIME: "DATETIME2",
723            exp.DataType.Type.DOUBLE: "FLOAT",
724            exp.DataType.Type.INT: "INTEGER",
725            exp.DataType.Type.TEXT: "VARCHAR(MAX)",
726            exp.DataType.Type.TIMESTAMP: "DATETIME2",
727            exp.DataType.Type.TIMESTAMPTZ: "DATETIMEOFFSET",
728            exp.DataType.Type.VARIANT: "SQL_VARIANT",
729        }
730
731        TRANSFORMS = {
732            **generator.Generator.TRANSFORMS,
733            exp.AnyValue: any_value_to_max_sql,
734            exp.AutoIncrementColumnConstraint: lambda *_: "IDENTITY",
735            exp.DateAdd: date_delta_sql("DATEADD"),
736            exp.DateDiff: date_delta_sql("DATEDIFF"),
737            exp.CTE: transforms.preprocess([qualify_derived_table_outputs]),
738            exp.CurrentDate: rename_func("GETDATE"),
739            exp.CurrentTimestamp: rename_func("GETDATE"),
740            exp.Extract: rename_func("DATEPART"),
741            exp.GeneratedAsIdentityColumnConstraint: generatedasidentitycolumnconstraint_sql,
742            exp.GroupConcat: _string_agg_sql,
743            exp.If: rename_func("IIF"),
744            exp.JSONExtract: _json_extract_sql,
745            exp.JSONExtractScalar: _json_extract_sql,
746            exp.LastDay: lambda self, e: self.func("EOMONTH", e.this),
747            exp.Max: max_or_greatest,
748            exp.MD5: lambda self, e: self.func("HASHBYTES", exp.Literal.string("MD5"), e.this),
749            exp.Min: min_or_least,
750            exp.NumberToStr: _format_sql,
751            exp.ParseJSON: lambda self, e: self.sql(e, "this"),
752            exp.Select: transforms.preprocess(
753                [
754                    transforms.eliminate_distinct_on,
755                    transforms.eliminate_semi_and_anti_joins,
756                    transforms.eliminate_qualify,
757                ]
758            ),
759            exp.Subquery: transforms.preprocess([qualify_derived_table_outputs]),
760            exp.SHA: lambda self, e: self.func("HASHBYTES", exp.Literal.string("SHA1"), e.this),
761            exp.SHA2: lambda self, e: self.func(
762                "HASHBYTES", exp.Literal.string(f"SHA2_{e.args.get('length', 256)}"), e.this
763            ),
764            exp.TemporaryProperty: lambda self, e: "",
765            exp.TimeStrToTime: timestrtotime_sql,
766            exp.TimeToStr: _format_sql,
767            exp.Trim: trim_sql,
768            exp.TsOrDsAdd: date_delta_sql("DATEADD", cast=True),
769            exp.TsOrDsDiff: date_delta_sql("DATEDIFF"),
770        }
771
772        TRANSFORMS.pop(exp.ReturnsProperty)
773
774        PROPERTIES_LOCATION = {
775            **generator.Generator.PROPERTIES_LOCATION,
776            exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED,
777        }
778
779        def lateral_op(self, expression: exp.Lateral) -> str:
780            cross_apply = expression.args.get("cross_apply")
781            if cross_apply is True:
782                return "CROSS APPLY"
783            if cross_apply is False:
784                return "OUTER APPLY"
785
786            # TODO: perhaps we can check if the parent is a Join and transpile it appropriately
787            self.unsupported("LATERAL clause is not supported.")
788            return "LATERAL"
789
790        def timefromparts_sql(self, expression: exp.TimeFromParts) -> str:
791            nano = expression.args.get("nano")
792            if nano is not None:
793                nano.pop()
794                self.unsupported("Specifying nanoseconds is not supported in TIMEFROMPARTS.")
795
796            if expression.args.get("fractions") is None:
797                expression.set("fractions", exp.Literal.number(0))
798            if expression.args.get("precision") is None:
799                expression.set("precision", exp.Literal.number(0))
800
801            return rename_func("TIMEFROMPARTS")(self, expression)
802
803        def timestampfromparts_sql(self, expression: exp.TimestampFromParts) -> str:
804            zone = expression.args.get("zone")
805            if zone is not None:
806                zone.pop()
807                self.unsupported("Time zone is not supported in DATETIMEFROMPARTS.")
808
809            nano = expression.args.get("nano")
810            if nano is not None:
811                nano.pop()
812                self.unsupported("Specifying nanoseconds is not supported in DATETIMEFROMPARTS.")
813
814            if expression.args.get("milli") is None:
815                expression.set("milli", exp.Literal.number(0))
816
817            return rename_func("DATETIMEFROMPARTS")(self, expression)
818
819        def set_operation(self, expression: exp.Union, op: str) -> str:
820            limit = expression.args.get("limit")
821            if limit:
822                return self.sql(expression.limit(limit.pop(), copy=False))
823
824            return super().set_operation(expression, op)
825
826        def setitem_sql(self, expression: exp.SetItem) -> str:
827            this = expression.this
828            if isinstance(this, exp.EQ) and not isinstance(this.left, exp.Parameter):
829                # T-SQL does not use '=' in SET command, except when the LHS is a variable.
830                return f"{self.sql(this.left)} {self.sql(this.right)}"
831
832            return super().setitem_sql(expression)
833
834        def boolean_sql(self, expression: exp.Boolean) -> str:
835            if type(expression.parent) in BIT_TYPES:
836                return "1" if expression.this else "0"
837
838            return "(1 = 1)" if expression.this else "(1 = 0)"
839
840        def is_sql(self, expression: exp.Is) -> str:
841            if isinstance(expression.expression, exp.Boolean):
842                return self.binary(expression, "=")
843            return self.binary(expression, "IS")
844
845        def createable_sql(self, expression: exp.Create, locations: t.DefaultDict) -> str:
846            sql = self.sql(expression, "this")
847            properties = expression.args.get("properties")
848
849            if sql[:1] != "#" and any(
850                isinstance(prop, exp.TemporaryProperty)
851                for prop in (properties.expressions if properties else [])
852            ):
853                sql = f"#{sql}"
854
855            return sql
856
857        def create_sql(self, expression: exp.Create) -> str:
858            kind = self.sql(expression, "kind").upper()
859            exists = expression.args.pop("exists", None)
860            sql = super().create_sql(expression)
861
862            like_property = expression.find(exp.LikeProperty)
863            if like_property:
864                ctas_expression = like_property.this
865            else:
866                ctas_expression = expression.expression
867
868            table = expression.find(exp.Table)
869
870            # Convert CTAS statement to SELECT .. INTO ..
871            if kind == "TABLE" and ctas_expression:
872                ctas_with = ctas_expression.args.get("with")
873                if ctas_with:
874                    ctas_with = ctas_with.pop()
875
876                subquery = ctas_expression
877                if isinstance(subquery, exp.Subqueryable):
878                    subquery = subquery.subquery()
879
880                select_into = exp.select("*").from_(exp.alias_(subquery, "temp", table=True))
881                select_into.set("into", exp.Into(this=table))
882                select_into.set("with", ctas_with)
883
884                if like_property:
885                    select_into.limit(0, copy=False)
886
887                sql = self.sql(select_into)
888
889            if exists:
890                identifier = self.sql(exp.Literal.string(exp.table_name(table) if table else ""))
891                sql = self.sql(exp.Literal.string(sql))
892                if kind == "SCHEMA":
893                    sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.schemata WHERE schema_name = {identifier}) EXEC({sql})"""
894                elif kind == "TABLE":
895                    assert table
896                    where = exp.and_(
897                        exp.column("table_name").eq(table.name),
898                        exp.column("table_schema").eq(table.db) if table.db else None,
899                        exp.column("table_catalog").eq(table.catalog) if table.catalog else None,
900                    )
901                    sql = f"""IF NOT EXISTS (SELECT * FROM information_schema.tables WHERE {where}) EXEC({sql})"""
902                elif kind == "INDEX":
903                    index = self.sql(exp.Literal.string(expression.this.text("this")))
904                    sql = f"""IF NOT EXISTS (SELECT * FROM sys.indexes WHERE object_id = object_id({identifier}) AND name = {index}) EXEC({sql})"""
905            elif expression.args.get("replace"):
906                sql = sql.replace("CREATE OR REPLACE ", "CREATE OR ALTER ", 1)
907
908            return self.prepend_ctes(expression, sql)
909
910        def offset_sql(self, expression: exp.Offset) -> str:
911            return f"{super().offset_sql(expression)} ROWS"
912
913        def version_sql(self, expression: exp.Version) -> str:
914            name = "SYSTEM_TIME" if expression.name == "TIMESTAMP" else expression.name
915            this = f"FOR {name}"
916            expr = expression.expression
917            kind = expression.text("kind")
918            if kind in ("FROM", "BETWEEN"):
919                args = expr.expressions
920                sep = "TO" if kind == "FROM" else "AND"
921                expr_sql = f"{self.sql(seq_get(args, 0))} {sep} {self.sql(seq_get(args, 1))}"
922            else:
923                expr_sql = self.sql(expr)
924
925            expr_sql = f" {expr_sql}" if expr_sql else ""
926            return f"{this} {kind}{expr_sql}"
927
928        def returnsproperty_sql(self, expression: exp.ReturnsProperty) -> str:
929            table = expression.args.get("table")
930            table = f"{table} " if table else ""
931            return f"RETURNS {table}{self.sql(expression, 'this')}"
932
933        def returning_sql(self, expression: exp.Returning) -> str:
934            into = self.sql(expression, "into")
935            into = self.seg(f"INTO {into}") if into else ""
936            return f"{self.seg('OUTPUT')} {self.expressions(expression, flat=True)}{into}"
937
938        def transaction_sql(self, expression: exp.Transaction) -> str:
939            this = self.sql(expression, "this")
940            this = f" {this}" if this else ""
941            mark = self.sql(expression, "mark")
942            mark = f" WITH MARK {mark}" if mark else ""
943            return f"BEGIN TRANSACTION{this}{mark}"
944
945        def commit_sql(self, expression: exp.Commit) -> str:
946            this = self.sql(expression, "this")
947            this = f" {this}" if this else ""
948            durability = expression.args.get("durability")
949            durability = (
950                f" WITH (DELAYED_DURABILITY = {'ON' if durability else 'OFF'})"
951                if durability is not None
952                else ""
953            )
954            return f"COMMIT TRANSACTION{this}{durability}"
955
956        def rollback_sql(self, expression: exp.Rollback) -> str:
957            this = self.sql(expression, "this")
958            this = f" {this}" if this else ""
959            return f"ROLLBACK TRANSACTION{this}"
960
961        def identifier_sql(self, expression: exp.Identifier) -> str:
962            identifier = super().identifier_sql(expression)
963
964            if expression.args.get("global"):
965                identifier = f"##{identifier}"
966            elif expression.args.get("temporary"):
967                identifier = f"#{identifier}"
968
969            return identifier
970
971        def constraint_sql(self, expression: exp.Constraint) -> str:
972            this = self.sql(expression, "this")
973            expressions = self.expressions(expression, flat=True, sep=" ")
974            return f"CONSTRAINT {this} {expressions}"
975
976        def length_sql(self, expression: exp.Length) -> str:
977            return self._uncast_text(expression, "LEN")
978
979        def right_sql(self, expression: exp.Right) -> str:
980            return self._uncast_text(expression, "RIGHT")
981
982        def left_sql(self, expression: exp.Left) -> str:
983            return self._uncast_text(expression, "LEFT")
984
985        def _uncast_text(self, expression: exp.Expression, name: str) -> str:
986            this = expression.this
987            if isinstance(this, exp.Cast) and this.is_type(exp.DataType.Type.TEXT):
988                this_sql = self.sql(this, "this")
989            else:
990                this_sql = self.sql(this)
991            expression_sql = self.sql(expression, "expression")
992            return self.func(name, this_sql, expression_sql if expression_sql else None)