sqlglot.dialects.tsql

   1from __future__ import annotations
   2
   3import datetime
   4import re
   5import typing as t
   6from functools import partial, reduce
   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    strposition_sql,
  21    trim_sql,
  22    timestrtotime_sql,
  23)
  24from sqlglot.helper import seq_get
  25from sqlglot.parser import build_coalesce
  26from sqlglot.time import format_time
  27from sqlglot.tokens import TokenType
  28
  29if t.TYPE_CHECKING:
  30    from sqlglot._typing import E
  31
  32FULL_FORMAT_TIME_MAPPING = {
  33    "weekday": "%A",
  34    "dw": "%A",
  35    "w": "%A",
  36    "month": "%B",
  37    "mm": "%B",
  38    "m": "%B",
  39}
  40
  41DATE_DELTA_INTERVAL = {
  42    "year": "year",
  43    "yyyy": "year",
  44    "yy": "year",
  45    "quarter": "quarter",
  46    "qq": "quarter",
  47    "q": "quarter",
  48    "month": "month",
  49    "mm": "month",
  50    "m": "month",
  51    "week": "week",
  52    "ww": "week",
  53    "wk": "week",
  54    "day": "day",
  55    "dd": "day",
  56    "d": "day",
  57}
  58
  59
  60DATE_FMT_RE = re.compile("([dD]{1,2})|([mM]{1,2})|([yY]{1,4})|([hH]{1,2})|([sS]{1,2})")
  61
  62# N = Numeric, C=Currency
  63TRANSPILE_SAFE_NUMBER_FMT = {"N", "C"}
  64
  65DEFAULT_START_DATE = datetime.date(1900, 1, 1)
  66
  67BIT_TYPES = {exp.EQ, exp.NEQ, exp.Is, exp.In, exp.Select, exp.Alias}
  68
  69# Unsupported options:
  70# - OPTIMIZE FOR ( @variable_name { UNKNOWN | = <literal_constant> } [ , ...n ] )
  71# - TABLE HINT
  72OPTIONS: parser.OPTIONS_TYPE = {
  73    **dict.fromkeys(
  74        (
  75            "DISABLE_OPTIMIZED_PLAN_FORCING",
  76            "FAST",
  77            "IGNORE_NONCLUSTERED_COLUMNSTORE_INDEX",
  78            "LABEL",
  79            "MAXDOP",
  80            "MAXRECURSION",
  81            "MAX_GRANT_PERCENT",
  82            "MIN_GRANT_PERCENT",
  83            "NO_PERFORMANCE_SPOOL",
  84            "QUERYTRACEON",
  85            "RECOMPILE",
  86        ),
  87        tuple(),
  88    ),
  89    "CONCAT": ("UNION",),
  90    "DISABLE": ("EXTERNALPUSHDOWN", "SCALEOUTEXECUTION"),
  91    "EXPAND": ("VIEWS",),
  92    "FORCE": ("EXTERNALPUSHDOWN", "ORDER", "SCALEOUTEXECUTION"),
  93    "HASH": ("GROUP", "JOIN", "UNION"),
  94    "KEEP": ("PLAN",),
  95    "KEEPFIXED": ("PLAN",),
  96    "LOOP": ("JOIN",),
  97    "MERGE": ("JOIN", "UNION"),
  98    "OPTIMIZE": (("FOR", "UNKNOWN"),),
  99    "ORDER": ("GROUP",),
 100    "PARAMETERIZATION": ("FORCED", "SIMPLE"),
 101    "ROBUST": ("PLAN",),
 102    "USE": ("PLAN",),
 103}
 104
 105OPTIONS_THAT_REQUIRE_EQUAL = ("MAX_GRANT_PERCENT", "MIN_GRANT_PERCENT", "LABEL")
 106
 107
 108def _build_formatted_time(
 109    exp_class: t.Type[E], full_format_mapping: t.Optional[bool] = None
 110) -> t.Callable[[t.List], E]:
 111    def _builder(args: t.List) -> E:
 112        fmt = seq_get(args, 0)
 113        if isinstance(fmt, exp.Expression):
 114            fmt = exp.Literal.string(
 115                format_time(
 116                    fmt.name.lower(),
 117                    (
 118                        {**TSQL.TIME_MAPPING, **FULL_FORMAT_TIME_MAPPING}
 119                        if full_format_mapping
 120                        else TSQL.TIME_MAPPING
 121                    ),
 122                )
 123            )
 124
 125        this = seq_get(args, 1)
 126        if isinstance(this, exp.Expression):
 127            this = exp.cast(this, exp.DataType.Type.DATETIME2)
 128
 129        return exp_class(this=this, format=fmt)
 130
 131    return _builder
 132
 133
 134def _build_format(args: t.List) -> exp.NumberToStr | exp.TimeToStr:
 135    this = seq_get(args, 0)
 136    fmt = seq_get(args, 1)
 137    culture = seq_get(args, 2)
 138
 139    number_fmt = fmt and (fmt.name in TRANSPILE_SAFE_NUMBER_FMT or not DATE_FMT_RE.search(fmt.name))
 140
 141    if number_fmt:
 142        return exp.NumberToStr(this=this, format=fmt, culture=culture)
 143
 144    if fmt:
 145        fmt = exp.Literal.string(
 146            format_time(fmt.name, TSQL.FORMAT_TIME_MAPPING)
 147            if len(fmt.name) == 1
 148            else format_time(fmt.name, TSQL.TIME_MAPPING)
 149        )
 150
 151    return exp.TimeToStr(this=this, format=fmt, culture=culture)
 152
 153
 154def _build_eomonth(args: t.List) -> exp.LastDay:
 155    date = exp.TsOrDsToDate(this=seq_get(args, 0))
 156    month_lag = seq_get(args, 1)
 157
 158    if month_lag is None:
 159        this: exp.Expression = date
 160    else:
 161        unit = DATE_DELTA_INTERVAL.get("month")
 162        this = exp.DateAdd(this=date, expression=month_lag, unit=unit and exp.var(unit))
 163
 164    return exp.LastDay(this=this)
 165
 166
 167def _build_hashbytes(args: t.List) -> exp.Expression:
 168    kind, data = args
 169    kind = kind.name.upper() if kind.is_string else ""
 170
 171    if kind == "MD5":
 172        args.pop(0)
 173        return exp.MD5(this=data)
 174    if kind in ("SHA", "SHA1"):
 175        args.pop(0)
 176        return exp.SHA(this=data)
 177    if kind == "SHA2_256":
 178        return exp.SHA2(this=data, length=exp.Literal.number(256))
 179    if kind == "SHA2_512":
 180        return exp.SHA2(this=data, length=exp.Literal.number(512))
 181
 182    return exp.func("HASHBYTES", *args)
 183
 184
 185DATEPART_ONLY_FORMATS = {"DW", "WK", "HOUR", "QUARTER"}
 186
 187
 188def _format_sql(self: TSQL.Generator, expression: exp.NumberToStr | exp.TimeToStr) -> str:
 189    fmt = expression.args["format"]
 190
 191    if not isinstance(expression, exp.NumberToStr):
 192        if fmt.is_string:
 193            mapped_fmt = format_time(fmt.name, TSQL.INVERSE_TIME_MAPPING)
 194
 195            name = (mapped_fmt or "").upper()
 196            if name in DATEPART_ONLY_FORMATS:
 197                return self.func("DATEPART", name, expression.this)
 198
 199            fmt_sql = self.sql(exp.Literal.string(mapped_fmt))
 200        else:
 201            fmt_sql = self.format_time(expression) or self.sql(fmt)
 202    else:
 203        fmt_sql = self.sql(fmt)
 204
 205    return self.func("FORMAT", expression.this, fmt_sql, expression.args.get("culture"))
 206
 207
 208def _string_agg_sql(self: TSQL.Generator, expression: exp.GroupConcat) -> str:
 209    this = expression.this
 210    distinct = expression.find(exp.Distinct)
 211    if distinct:
 212        # exp.Distinct can appear below an exp.Order or an exp.GroupConcat expression
 213        self.unsupported("T-SQL STRING_AGG doesn't support DISTINCT.")
 214        this = distinct.pop().expressions[0]
 215
 216    order = ""
 217    if isinstance(expression.this, exp.Order):
 218        if expression.this.this:
 219            this = expression.this.this.pop()
 220        # Order has a leading space
 221        order = f" WITHIN GROUP ({self.sql(expression.this)[1:]})"
 222
 223    separator = expression.args.get("separator") or exp.Literal.string(",")
 224    return f"STRING_AGG({self.format_args(this, separator)}){order}"
 225
 226
 227def _build_date_delta(
 228    exp_class: t.Type[E], unit_mapping: t.Optional[t.Dict[str, str]] = None
 229) -> t.Callable[[t.List], E]:
 230    def _builder(args: t.List) -> E:
 231        unit = seq_get(args, 0)
 232        if unit and unit_mapping:
 233            unit = exp.var(unit_mapping.get(unit.name.lower(), unit.name))
 234
 235        start_date = seq_get(args, 1)
 236        if start_date and start_date.is_number:
 237            # Numeric types are valid DATETIME values
 238            if start_date.is_int:
 239                adds = DEFAULT_START_DATE + datetime.timedelta(days=start_date.to_py())
 240                start_date = exp.Literal.string(adds.strftime("%F"))
 241            else:
 242                # We currently don't handle float values, i.e. they're not converted to equivalent DATETIMEs.
 243                # This is not a problem when generating T-SQL code, it is when transpiling to other dialects.
 244                return exp_class(this=seq_get(args, 2), expression=start_date, unit=unit)
 245
 246        return exp_class(
 247            this=exp.TimeStrToTime(this=seq_get(args, 2)),
 248            expression=exp.TimeStrToTime(this=start_date),
 249            unit=unit,
 250        )
 251
 252    return _builder
 253
 254
 255def qualify_derived_table_outputs(expression: exp.Expression) -> exp.Expression:
 256    """Ensures all (unnamed) output columns are aliased for CTEs and Subqueries."""
 257    alias = expression.args.get("alias")
 258
 259    if (
 260        isinstance(expression, (exp.CTE, exp.Subquery))
 261        and isinstance(alias, exp.TableAlias)
 262        and not alias.columns
 263    ):
 264        from sqlglot.optimizer.qualify_columns import qualify_outputs
 265
 266        # We keep track of the unaliased column projection indexes instead of the expressions
 267        # themselves, because the latter are going to be replaced by new nodes when the aliases
 268        # are added and hence we won't be able to reach these newly added Alias parents
 269        query = expression.this
 270        unaliased_column_indexes = (
 271            i for i, c in enumerate(query.selects) if isinstance(c, exp.Column) and not c.alias
 272        )
 273
 274        qualify_outputs(query)
 275
 276        # Preserve the quoting information of columns for newly added Alias nodes
 277        query_selects = query.selects
 278        for select_index in unaliased_column_indexes:
 279            alias = query_selects[select_index]
 280            column = alias.this
 281            if isinstance(column.this, exp.Identifier):
 282                alias.args["alias"].set("quoted", column.this.quoted)
 283
 284    return expression
 285
 286
 287# https://learn.microsoft.com/en-us/sql/t-sql/functions/datetimefromparts-transact-sql?view=sql-server-ver16#syntax
 288def _build_datetimefromparts(args: t.List) -> exp.TimestampFromParts:
 289    return exp.TimestampFromParts(
 290        year=seq_get(args, 0),
 291        month=seq_get(args, 1),
 292        day=seq_get(args, 2),
 293        hour=seq_get(args, 3),
 294        min=seq_get(args, 4),
 295        sec=seq_get(args, 5),
 296        milli=seq_get(args, 6),
 297    )
 298
 299
 300# https://learn.microsoft.com/en-us/sql/t-sql/functions/timefromparts-transact-sql?view=sql-server-ver16#syntax
 301def _build_timefromparts(args: t.List) -> exp.TimeFromParts:
 302    return exp.TimeFromParts(
 303        hour=seq_get(args, 0),
 304        min=seq_get(args, 1),
 305        sec=seq_get(args, 2),
 306        fractions=seq_get(args, 3),
 307        precision=seq_get(args, 4),
 308    )
 309
 310
 311def _build_with_arg_as_text(
 312    klass: t.Type[exp.Expression],
 313) -> t.Callable[[t.List[exp.Expression]], exp.Expression]:
 314    def _parse(args: t.List[exp.Expression]) -> exp.Expression:
 315        this = seq_get(args, 0)
 316
 317        if this and not this.is_string:
 318            this = exp.cast(this, exp.DataType.Type.TEXT)
 319
 320        expression = seq_get(args, 1)
 321        kwargs = {"this": this}
 322
 323        if expression:
 324            kwargs["expression"] = expression
 325
 326        return klass(**kwargs)
 327
 328    return _parse
 329
 330
 331# https://learn.microsoft.com/en-us/sql/t-sql/functions/parsename-transact-sql?view=sql-server-ver16
 332def _build_parsename(args: t.List) -> exp.SplitPart | exp.Anonymous:
 333    # PARSENAME(...) will be stored into exp.SplitPart if:
 334    # - All args are literals
 335    # - The part index (2nd arg) is <= 4 (max valid value, otherwise TSQL returns NULL)
 336    if len(args) == 2 and all(isinstance(arg, exp.Literal) for arg in args):
 337        this = args[0]
 338        part_index = args[1]
 339        split_count = len(this.name.split("."))
 340        if split_count <= 4:
 341            return exp.SplitPart(
 342                this=this,
 343                delimiter=exp.Literal.string("."),
 344                part_index=exp.Literal.number(split_count + 1 - part_index.to_py()),
 345            )
 346
 347    return exp.Anonymous(this="PARSENAME", expressions=args)
 348
 349
 350def _build_json_query(args: t.List, dialect: Dialect) -> exp.JSONExtract:
 351    if len(args) == 1:
 352        # The default value for path is '$'. As a result, if you don't provide a
 353        # value for path, JSON_QUERY returns the input expression.
 354        args.append(exp.Literal.string("$"))
 355
 356    return parser.build_extract_json_with_path(exp.JSONExtract)(args, dialect)
 357
 358
 359def _json_extract_sql(
 360    self: TSQL.Generator, expression: exp.JSONExtract | exp.JSONExtractScalar
 361) -> str:
 362    json_query = self.func("JSON_QUERY", expression.this, expression.expression)
 363    json_value = self.func("JSON_VALUE", expression.this, expression.expression)
 364    return self.func("ISNULL", json_query, json_value)
 365
 366
 367def _timestrtotime_sql(self: TSQL.Generator, expression: exp.TimeStrToTime):
 368    sql = timestrtotime_sql(self, expression)
 369    if expression.args.get("zone"):
 370        # If there is a timezone, produce an expression like:
 371        # CAST('2020-01-01 12:13:14-08:00' AS DATETIMEOFFSET) AT TIME ZONE 'UTC'
 372        # If you dont have AT TIME ZONE 'UTC', wrapping that expression in another cast back to DATETIME2 just drops the timezone information
 373        return self.sql(exp.AtTimeZone(this=sql, zone=exp.Literal.string("UTC")))
 374    return sql
 375
 376
 377def _build_datetrunc(args: t.List) -> exp.TimestampTrunc:
 378    unit = seq_get(args, 0)
 379    this = seq_get(args, 1)
 380
 381    if this and this.is_string:
 382        this = exp.cast(this, exp.DataType.Type.DATETIME2)
 383
 384    return exp.TimestampTrunc(this=this, unit=unit)
 385
 386
 387class TSQL(Dialect):
 388    SUPPORTS_SEMI_ANTI_JOIN = False
 389    LOG_BASE_FIRST = False
 390    TYPED_DIVISION = True
 391    CONCAT_COALESCE = True
 392    NORMALIZATION_STRATEGY = NormalizationStrategy.CASE_INSENSITIVE
 393
 394    TIME_FORMAT = "'yyyy-mm-dd hh:mm:ss'"
 395
 396    TIME_MAPPING = {
 397        "year": "%Y",
 398        "dayofyear": "%j",
 399        "day": "%d",
 400        "dy": "%d",
 401        "y": "%Y",
 402        "week": "%W",
 403        "ww": "%W",
 404        "wk": "%W",
 405        "hour": "%h",
 406        "hh": "%I",
 407        "minute": "%M",
 408        "mi": "%M",
 409        "n": "%M",
 410        "second": "%S",
 411        "ss": "%S",
 412        "s": "%-S",
 413        "millisecond": "%f",
 414        "ms": "%f",
 415        "weekday": "%w",
 416        "dw": "%w",
 417        "month": "%m",
 418        "mm": "%M",
 419        "m": "%-M",
 420        "Y": "%Y",
 421        "YYYY": "%Y",
 422        "YY": "%y",
 423        "MMMM": "%B",
 424        "MMM": "%b",
 425        "MM": "%m",
 426        "M": "%-m",
 427        "dddd": "%A",
 428        "dd": "%d",
 429        "d": "%-d",
 430        "HH": "%H",
 431        "H": "%-H",
 432        "h": "%-I",
 433        "ffffff": "%f",
 434        "yyyy": "%Y",
 435        "yy": "%y",
 436    }
 437
 438    CONVERT_FORMAT_MAPPING = {
 439        "0": "%b %d %Y %-I:%M%p",
 440        "1": "%m/%d/%y",
 441        "2": "%y.%m.%d",
 442        "3": "%d/%m/%y",
 443        "4": "%d.%m.%y",
 444        "5": "%d-%m-%y",
 445        "6": "%d %b %y",
 446        "7": "%b %d, %y",
 447        "8": "%H:%M:%S",
 448        "9": "%b %d %Y %-I:%M:%S:%f%p",
 449        "10": "mm-dd-yy",
 450        "11": "yy/mm/dd",
 451        "12": "yymmdd",
 452        "13": "%d %b %Y %H:%M:ss:%f",
 453        "14": "%H:%M:%S:%f",
 454        "20": "%Y-%m-%d %H:%M:%S",
 455        "21": "%Y-%m-%d %H:%M:%S.%f",
 456        "22": "%m/%d/%y %-I:%M:%S %p",
 457        "23": "%Y-%m-%d",
 458        "24": "%H:%M:%S",
 459        "25": "%Y-%m-%d %H:%M:%S.%f",
 460        "100": "%b %d %Y %-I:%M%p",
 461        "101": "%m/%d/%Y",
 462        "102": "%Y.%m.%d",
 463        "103": "%d/%m/%Y",
 464        "104": "%d.%m.%Y",
 465        "105": "%d-%m-%Y",
 466        "106": "%d %b %Y",
 467        "107": "%b %d, %Y",
 468        "108": "%H:%M:%S",
 469        "109": "%b %d %Y %-I:%M:%S:%f%p",
 470        "110": "%m-%d-%Y",
 471        "111": "%Y/%m/%d",
 472        "112": "%Y%m%d",
 473        "113": "%d %b %Y %H:%M:%S:%f",
 474        "114": "%H:%M:%S:%f",
 475        "120": "%Y-%m-%d %H:%M:%S",
 476        "121": "%Y-%m-%d %H:%M:%S.%f",
 477    }
 478
 479    FORMAT_TIME_MAPPING = {
 480        "y": "%B %Y",
 481        "d": "%m/%d/%Y",
 482        "H": "%-H",
 483        "h": "%-I",
 484        "s": "%Y-%m-%d %H:%M:%S",
 485        "D": "%A,%B,%Y",
 486        "f": "%A,%B,%Y %-I:%M %p",
 487        "F": "%A,%B,%Y %-I:%M:%S %p",
 488        "g": "%m/%d/%Y %-I:%M %p",
 489        "G": "%m/%d/%Y %-I:%M:%S %p",
 490        "M": "%B %-d",
 491        "m": "%B %-d",
 492        "O": "%Y-%m-%dT%H:%M:%S",
 493        "u": "%Y-%M-%D %H:%M:%S%z",
 494        "U": "%A, %B %D, %Y %H:%M:%S%z",
 495        "T": "%-I:%M:%S %p",
 496        "t": "%-I:%M",
 497        "Y": "%a %Y",
 498    }
 499
 500    class Tokenizer(tokens.Tokenizer):
 501        IDENTIFIERS = [("[", "]"), '"']
 502        QUOTES = ["'", '"']
 503        HEX_STRINGS = [("0x", ""), ("0X", "")]
 504        VAR_SINGLE_TOKENS = {"@", "$", "#"}
 505
 506        KEYWORDS = {
 507            **tokens.Tokenizer.KEYWORDS,
 508            "CLUSTERED INDEX": TokenType.INDEX,
 509            "DATETIME2": TokenType.DATETIME2,
 510            "DATETIMEOFFSET": TokenType.TIMESTAMPTZ,
 511            "DECLARE": TokenType.DECLARE,
 512            "EXEC": TokenType.COMMAND,
 513            "FOR SYSTEM_TIME": TokenType.TIMESTAMP_SNAPSHOT,
 514            "GO": TokenType.COMMAND,
 515            "IMAGE": TokenType.IMAGE,
 516            "MONEY": TokenType.MONEY,
 517            "NONCLUSTERED INDEX": TokenType.INDEX,
 518            "NTEXT": TokenType.TEXT,
 519            "OPTION": TokenType.OPTION,
 520            "OUTPUT": TokenType.RETURNING,
 521            "PRINT": TokenType.COMMAND,
 522            "PROC": TokenType.PROCEDURE,
 523            "REAL": TokenType.FLOAT,
 524            "ROWVERSION": TokenType.ROWVERSION,
 525            "SMALLDATETIME": TokenType.SMALLDATETIME,
 526            "SMALLMONEY": TokenType.SMALLMONEY,
 527            "SQL_VARIANT": TokenType.VARIANT,
 528            "SYSTEM_USER": TokenType.CURRENT_USER,
 529            "TOP": TokenType.TOP,
 530            "TIMESTAMP": TokenType.ROWVERSION,
 531            "TINYINT": TokenType.UTINYINT,
 532            "UNIQUEIDENTIFIER": TokenType.UUID,
 533            "UPDATE STATISTICS": TokenType.COMMAND,
 534            "XML": TokenType.XML,
 535        }
 536        KEYWORDS.pop("/*+")
 537
 538        COMMANDS = {*tokens.Tokenizer.COMMANDS, TokenType.END}
 539
 540    class Parser(parser.Parser):
 541        SET_REQUIRES_ASSIGNMENT_DELIMITER = False
 542        LOG_DEFAULTS_TO_LN = True
 543        ALTER_TABLE_ADD_REQUIRED_FOR_EACH_COLUMN = False
 544        STRING_ALIASES = True
 545        NO_PAREN_IF_COMMANDS = False
 546
 547        QUERY_MODIFIER_PARSERS = {
 548            **parser.Parser.QUERY_MODIFIER_PARSERS,
 549            TokenType.OPTION: lambda self: ("options", self._parse_options()),
 550        }
 551
 552        # T-SQL does not allow BEGIN to be used as an identifier
 553        ID_VAR_TOKENS = parser.Parser.ID_VAR_TOKENS - {TokenType.BEGIN}
 554        ALIAS_TOKENS = parser.Parser.ALIAS_TOKENS - {TokenType.BEGIN}
 555        TABLE_ALIAS_TOKENS = parser.Parser.TABLE_ALIAS_TOKENS - {TokenType.BEGIN}
 556        COMMENT_TABLE_ALIAS_TOKENS = parser.Parser.COMMENT_TABLE_ALIAS_TOKENS - {TokenType.BEGIN}
 557        UPDATE_ALIAS_TOKENS = parser.Parser.UPDATE_ALIAS_TOKENS - {TokenType.BEGIN}
 558
 559        FUNCTIONS = {
 560            **parser.Parser.FUNCTIONS,
 561            "CHARINDEX": lambda args: exp.StrPosition(
 562                this=seq_get(args, 1),
 563                substr=seq_get(args, 0),
 564                position=seq_get(args, 2),
 565            ),
 566            "COUNT": lambda args: exp.Count(
 567                this=seq_get(args, 0), expressions=args[1:], big_int=False
 568            ),
 569            "COUNT_BIG": lambda args: exp.Count(
 570                this=seq_get(args, 0), expressions=args[1:], big_int=True
 571            ),
 572            "DATEADD": build_date_delta(exp.DateAdd, unit_mapping=DATE_DELTA_INTERVAL),
 573            "DATEDIFF": _build_date_delta(exp.DateDiff, unit_mapping=DATE_DELTA_INTERVAL),
 574            "DATENAME": _build_formatted_time(exp.TimeToStr, full_format_mapping=True),
 575            "DATEPART": _build_formatted_time(exp.TimeToStr),
 576            "DATETIMEFROMPARTS": _build_datetimefromparts,
 577            "EOMONTH": _build_eomonth,
 578            "FORMAT": _build_format,
 579            "GETDATE": exp.CurrentTimestamp.from_arg_list,
 580            "HASHBYTES": _build_hashbytes,
 581            "ISNULL": build_coalesce,
 582            "JSON_QUERY": _build_json_query,
 583            "JSON_VALUE": parser.build_extract_json_with_path(exp.JSONExtractScalar),
 584            "LEN": _build_with_arg_as_text(exp.Length),
 585            "LEFT": _build_with_arg_as_text(exp.Left),
 586            "NEWID": exp.Uuid.from_arg_list,
 587            "RIGHT": _build_with_arg_as_text(exp.Right),
 588            "PARSENAME": _build_parsename,
 589            "REPLICATE": exp.Repeat.from_arg_list,
 590            "SCHEMA_NAME": exp.CurrentSchema.from_arg_list,
 591            "SQUARE": lambda args: exp.Pow(this=seq_get(args, 0), expression=exp.Literal.number(2)),
 592            "SYSDATETIME": exp.CurrentTimestamp.from_arg_list,
 593            "SUSER_NAME": exp.CurrentUser.from_arg_list,
 594            "SUSER_SNAME": exp.CurrentUser.from_arg_list,
 595            "SYSTEM_USER": exp.CurrentUser.from_arg_list,
 596            "TIMEFROMPARTS": _build_timefromparts,
 597            "DATETRUNC": _build_datetrunc,
 598        }
 599
 600        JOIN_HINTS = {"LOOP", "HASH", "MERGE", "REMOTE"}
 601
 602        PROCEDURE_OPTIONS = dict.fromkeys(
 603            ("ENCRYPTION", "RECOMPILE", "SCHEMABINDING", "NATIVE_COMPILATION", "EXECUTE"), tuple()
 604        )
 605
 606        COLUMN_DEFINITION_MODES = {"OUT", "OUTPUT", "READ_ONLY"}
 607
 608        RETURNS_TABLE_TOKENS = parser.Parser.ID_VAR_TOKENS - {
 609            TokenType.TABLE,
 610            *parser.Parser.TYPE_TOKENS,
 611        }
 612
 613        STATEMENT_PARSERS = {
 614            **parser.Parser.STATEMENT_PARSERS,
 615            TokenType.DECLARE: lambda self: self._parse_declare(),
 616        }
 617
 618        RANGE_PARSERS = {
 619            **parser.Parser.RANGE_PARSERS,
 620            TokenType.DCOLON: lambda self, this: self.expression(
 621                exp.ScopeResolution,
 622                this=this,
 623                expression=self._parse_function() or self._parse_var(any_token=True),
 624            ),
 625        }
 626
 627        NO_PAREN_FUNCTION_PARSERS = {
 628            **parser.Parser.NO_PAREN_FUNCTION_PARSERS,
 629            "NEXT": lambda self: self._parse_next_value_for(),
 630        }
 631
 632        # The DCOLON (::) operator serves as a scope resolution (exp.ScopeResolution) operator in T-SQL
 633        COLUMN_OPERATORS = {
 634            **parser.Parser.COLUMN_OPERATORS,
 635            TokenType.DCOLON: lambda self, this, to: self.expression(exp.Cast, this=this, to=to)
 636            if isinstance(to, exp.DataType) and to.this != exp.DataType.Type.USERDEFINED
 637            else self.expression(exp.ScopeResolution, this=this, expression=to),
 638        }
 639
 640        def _parse_wrapped_select(self, table: bool = False) -> t.Optional[exp.Expression]:
 641            if self._match(TokenType.MERGE):
 642                comments = self._prev_comments
 643                merge = self._parse_merge()
 644                merge.add_comments(comments, prepend=True)
 645                return merge
 646
 647            return super()._parse_wrapped_select(table=table)
 648
 649        def _parse_dcolon(self) -> t.Optional[exp.Expression]:
 650            # We want to use _parse_types() if the first token after :: is a known type,
 651            # otherwise we could parse something like x::varchar(max) into a function
 652            if self._match_set(self.TYPE_TOKENS, advance=False):
 653                return self._parse_types()
 654
 655            return self._parse_function() or self._parse_types()
 656
 657        def _parse_options(self) -> t.Optional[t.List[exp.Expression]]:
 658            if not self._match(TokenType.OPTION):
 659                return None
 660
 661            def _parse_option() -> t.Optional[exp.Expression]:
 662                option = self._parse_var_from_options(OPTIONS)
 663                if not option:
 664                    return None
 665
 666                self._match(TokenType.EQ)
 667                return self.expression(
 668                    exp.QueryOption, this=option, expression=self._parse_primary_or_var()
 669                )
 670
 671            return self._parse_wrapped_csv(_parse_option)
 672
 673        def _parse_projections(self) -> t.List[exp.Expression]:
 674            """
 675            T-SQL supports the syntax alias = expression in the SELECT's projection list,
 676            so we transform all parsed Selects to convert their EQ projections into Aliases.
 677
 678            See: https://learn.microsoft.com/en-us/sql/t-sql/queries/select-clause-transact-sql?view=sql-server-ver16#syntax
 679            """
 680            return [
 681                (
 682                    exp.alias_(projection.expression, projection.this.this, copy=False)
 683                    if isinstance(projection, exp.EQ) and isinstance(projection.this, exp.Column)
 684                    else projection
 685                )
 686                for projection in super()._parse_projections()
 687            ]
 688
 689        def _parse_commit_or_rollback(self) -> exp.Commit | exp.Rollback:
 690            """Applies to SQL Server and Azure SQL Database
 691            COMMIT [ { TRAN | TRANSACTION }
 692                [ transaction_name | @tran_name_variable ] ]
 693                [ WITH ( DELAYED_DURABILITY = { OFF | ON } ) ]
 694
 695            ROLLBACK { TRAN | TRANSACTION }
 696                [ transaction_name | @tran_name_variable
 697                | savepoint_name | @savepoint_variable ]
 698            """
 699            rollback = self._prev.token_type == TokenType.ROLLBACK
 700
 701            self._match_texts(("TRAN", "TRANSACTION"))
 702            this = self._parse_id_var()
 703
 704            if rollback:
 705                return self.expression(exp.Rollback, this=this)
 706
 707            durability = None
 708            if self._match_pair(TokenType.WITH, TokenType.L_PAREN):
 709                self._match_text_seq("DELAYED_DURABILITY")
 710                self._match(TokenType.EQ)
 711
 712                if self._match_text_seq("OFF"):
 713                    durability = False
 714                else:
 715                    self._match(TokenType.ON)
 716                    durability = True
 717
 718                self._match_r_paren()
 719
 720            return self.expression(exp.Commit, this=this, durability=durability)
 721
 722        def _parse_transaction(self) -> exp.Transaction | exp.Command:
 723            """Applies to SQL Server and Azure SQL Database
 724            BEGIN { TRAN | TRANSACTION }
 725            [ { transaction_name | @tran_name_variable }
 726            [ WITH MARK [ 'description' ] ]
 727            ]
 728            """
 729            if self._match_texts(("TRAN", "TRANSACTION")):
 730                transaction = self.expression(exp.Transaction, this=self._parse_id_var())
 731                if self._match_text_seq("WITH", "MARK"):
 732                    transaction.set("mark", self._parse_string())
 733
 734                return transaction
 735
 736            return self._parse_as_command(self._prev)
 737
 738        def _parse_returns(self) -> exp.ReturnsProperty:
 739            table = self._parse_id_var(any_token=False, tokens=self.RETURNS_TABLE_TOKENS)
 740            returns = super()._parse_returns()
 741            returns.set("table", table)
 742            return returns
 743
 744        def _parse_convert(
 745            self, strict: bool, safe: t.Optional[bool] = None
 746        ) -> t.Optional[exp.Expression]:
 747            this = self._parse_types()
 748            self._match(TokenType.COMMA)
 749            args = [this, *self._parse_csv(self._parse_assignment)]
 750            convert = exp.Convert.from_arg_list(args)
 751            convert.set("safe", safe)
 752            convert.set("strict", strict)
 753            return convert
 754
 755        def _parse_column_def(
 756            self, this: t.Optional[exp.Expression], computed_column: bool = True
 757        ) -> t.Optional[exp.Expression]:
 758            this = super()._parse_column_def(this=this, computed_column=computed_column)
 759            if not this:
 760                return None
 761            if self._match(TokenType.EQ):
 762                this.set("default", self._parse_disjunction())
 763            if self._match_texts(self.COLUMN_DEFINITION_MODES):
 764                this.set("output", self._prev.text)
 765            return this
 766
 767        def _parse_user_defined_function(
 768            self, kind: t.Optional[TokenType] = None
 769        ) -> t.Optional[exp.Expression]:
 770            this = super()._parse_user_defined_function(kind=kind)
 771
 772            if (
 773                kind == TokenType.FUNCTION
 774                or isinstance(this, exp.UserDefinedFunction)
 775                or self._match(TokenType.ALIAS, advance=False)
 776            ):
 777                return this
 778
 779            if not self._match(TokenType.WITH, advance=False):
 780                expressions = self._parse_csv(self._parse_function_parameter)
 781            else:
 782                expressions = None
 783
 784            return self.expression(exp.UserDefinedFunction, this=this, expressions=expressions)
 785
 786        def _parse_into(self) -> t.Optional[exp.Into]:
 787            into = super()._parse_into()
 788
 789            table = isinstance(into, exp.Into) and into.find(exp.Table)
 790            if isinstance(table, exp.Table):
 791                table_identifier = table.this
 792                if table_identifier.args.get("temporary"):
 793                    # Promote the temporary property from the Identifier to the Into expression
 794                    t.cast(exp.Into, into).set("temporary", True)
 795
 796            return into
 797
 798        def _parse_id_var(
 799            self,
 800            any_token: bool = True,
 801            tokens: t.Optional[t.Collection[TokenType]] = None,
 802        ) -> t.Optional[exp.Expression]:
 803            is_temporary = self._match(TokenType.HASH)
 804            is_global = is_temporary and self._match(TokenType.HASH)
 805
 806            this = super()._parse_id_var(any_token=any_token, tokens=tokens)
 807            if this:
 808                if is_global:
 809                    this.set("global", True)
 810                elif is_temporary:
 811                    this.set("temporary", True)
 812
 813            return this
 814
 815        def _parse_create(self) -> exp.Create | exp.Command:
 816            create = super()._parse_create()
 817
 818            if isinstance(create, exp.Create):
 819                table = create.this.this if isinstance(create.this, exp.Schema) else create.this
 820                if isinstance(table, exp.Table) and table.this and table.this.args.get("temporary"):
 821                    if not create.args.get("properties"):
 822                        create.set("properties", exp.Properties(expressions=[]))
 823
 824                    create.args["properties"].append("expressions", exp.TemporaryProperty())
 825
 826            return create
 827
 828        def _parse_if(self) -> t.Optional[exp.Expression]:
 829            index = self._index
 830
 831            if self._match_text_seq("OBJECT_ID"):
 832                self._parse_wrapped_csv(self._parse_string)
 833                if self._match_text_seq("IS", "NOT", "NULL") and self._match(TokenType.DROP):
 834                    return self._parse_drop(exists=True)
 835                self._retreat(index)
 836
 837            return super()._parse_if()
 838
 839        def _parse_unique(self) -> exp.UniqueColumnConstraint:
 840            if self._match_texts(("CLUSTERED", "NONCLUSTERED")):
 841                this = self.CONSTRAINT_PARSERS[self._prev.text.upper()](self)
 842            else:
 843                this = self._parse_schema(self._parse_id_var(any_token=False))
 844
 845            return self.expression(exp.UniqueColumnConstraint, this=this)
 846
 847        def _parse_partition(self) -> t.Optional[exp.Partition]:
 848            if not self._match_text_seq("WITH", "(", "PARTITIONS"):
 849                return None
 850
 851            def parse_range():
 852                low = self._parse_bitwise()
 853                high = self._parse_bitwise() if self._match_text_seq("TO") else None
 854
 855                return (
 856                    self.expression(exp.PartitionRange, this=low, expression=high) if high else low
 857                )
 858
 859            partition = self.expression(
 860                exp.Partition, expressions=self._parse_wrapped_csv(parse_range)
 861            )
 862
 863            self._match_r_paren()
 864
 865            return partition
 866
 867        def _parse_declare(self) -> exp.Declare | exp.Command:
 868            index = self._index
 869            expressions = self._try_parse(partial(self._parse_csv, self._parse_declareitem))
 870
 871            if not expressions or self._curr:
 872                self._retreat(index)
 873                return self._parse_as_command(self._prev)
 874
 875            return self.expression(exp.Declare, expressions=expressions)
 876
 877        def _parse_declareitem(self) -> t.Optional[exp.DeclareItem]:
 878            var = self._parse_id_var()
 879            if not var:
 880                return None
 881
 882            value = None
 883            self._match(TokenType.ALIAS)
 884            if self._match(TokenType.TABLE):
 885                data_type = self._parse_schema()
 886            else:
 887                data_type = self._parse_types()
 888                if self._match(TokenType.EQ):
 889                    value = self._parse_bitwise()
 890
 891            return self.expression(exp.DeclareItem, this=var, kind=data_type, default=value)
 892
 893    class Generator(generator.Generator):
 894        LIMIT_IS_TOP = True
 895        QUERY_HINTS = False
 896        RETURNING_END = False
 897        NVL2_SUPPORTED = False
 898        ALTER_TABLE_INCLUDE_COLUMN_KEYWORD = False
 899        LIMIT_FETCH = "FETCH"
 900        COMPUTED_COLUMN_WITH_TYPE = False
 901        CTE_RECURSIVE_KEYWORD_REQUIRED = False
 902        ENSURE_BOOLS = True
 903        NULL_ORDERING_SUPPORTED = None
 904        SUPPORTS_SINGLE_ARG_CONCAT = False
 905        TABLESAMPLE_SEED_KEYWORD = "REPEATABLE"
 906        SUPPORTS_SELECT_INTO = True
 907        JSON_PATH_BRACKETED_KEY_SUPPORTED = False
 908        SUPPORTS_TO_NUMBER = False
 909        SET_OP_MODIFIERS = False
 910        COPY_PARAMS_EQ_REQUIRED = True
 911        PARSE_JSON_NAME = None
 912        EXCEPT_INTERSECT_SUPPORT_ALL_CLAUSE = False
 913
 914        EXPRESSIONS_WITHOUT_NESTED_CTES = {
 915            exp.Create,
 916            exp.Delete,
 917            exp.Insert,
 918            exp.Intersect,
 919            exp.Except,
 920            exp.Merge,
 921            exp.Select,
 922            exp.Subquery,
 923            exp.Union,
 924            exp.Update,
 925        }
 926
 927        SUPPORTED_JSON_PATH_PARTS = {
 928            exp.JSONPathKey,
 929            exp.JSONPathRoot,
 930            exp.JSONPathSubscript,
 931        }
 932
 933        TYPE_MAPPING = {
 934            **generator.Generator.TYPE_MAPPING,
 935            exp.DataType.Type.BOOLEAN: "BIT",
 936            exp.DataType.Type.DATETIME2: "DATETIME2",
 937            exp.DataType.Type.DECIMAL: "NUMERIC",
 938            exp.DataType.Type.DOUBLE: "FLOAT",
 939            exp.DataType.Type.INT: "INTEGER",
 940            exp.DataType.Type.ROWVERSION: "ROWVERSION",
 941            exp.DataType.Type.TEXT: "VARCHAR(MAX)",
 942            exp.DataType.Type.TIMESTAMP: "DATETIME2",
 943            exp.DataType.Type.TIMESTAMPNTZ: "DATETIME2",
 944            exp.DataType.Type.TIMESTAMPTZ: "DATETIMEOFFSET",
 945            exp.DataType.Type.SMALLDATETIME: "SMALLDATETIME",
 946            exp.DataType.Type.UTINYINT: "TINYINT",
 947            exp.DataType.Type.VARIANT: "SQL_VARIANT",
 948            exp.DataType.Type.UUID: "UNIQUEIDENTIFIER",
 949        }
 950
 951        TYPE_MAPPING.pop(exp.DataType.Type.NCHAR)
 952        TYPE_MAPPING.pop(exp.DataType.Type.NVARCHAR)
 953
 954        TRANSFORMS = {
 955            **generator.Generator.TRANSFORMS,
 956            exp.AnyValue: any_value_to_max_sql,
 957            exp.ArrayToString: rename_func("STRING_AGG"),
 958            exp.AutoIncrementColumnConstraint: lambda *_: "IDENTITY",
 959            exp.Chr: rename_func("CHAR"),
 960            exp.DateAdd: date_delta_sql("DATEADD"),
 961            exp.DateDiff: date_delta_sql("DATEDIFF"),
 962            exp.CTE: transforms.preprocess([qualify_derived_table_outputs]),
 963            exp.CurrentDate: rename_func("GETDATE"),
 964            exp.CurrentTimestamp: rename_func("GETDATE"),
 965            exp.DateStrToDate: datestrtodate_sql,
 966            exp.Extract: rename_func("DATEPART"),
 967            exp.GeneratedAsIdentityColumnConstraint: generatedasidentitycolumnconstraint_sql,
 968            exp.GroupConcat: _string_agg_sql,
 969            exp.If: rename_func("IIF"),
 970            exp.JSONExtract: _json_extract_sql,
 971            exp.JSONExtractScalar: _json_extract_sql,
 972            exp.LastDay: lambda self, e: self.func("EOMONTH", e.this),
 973            exp.Ln: rename_func("LOG"),
 974            exp.Max: max_or_greatest,
 975            exp.MD5: lambda self, e: self.func("HASHBYTES", exp.Literal.string("MD5"), e.this),
 976            exp.Min: min_or_least,
 977            exp.NumberToStr: _format_sql,
 978            exp.Repeat: rename_func("REPLICATE"),
 979            exp.CurrentSchema: rename_func("SCHEMA_NAME"),
 980            exp.Select: transforms.preprocess(
 981                [
 982                    transforms.eliminate_distinct_on,
 983                    transforms.eliminate_semi_and_anti_joins,
 984                    transforms.eliminate_qualify,
 985                    transforms.unnest_generate_date_array_using_recursive_cte,
 986                ]
 987            ),
 988            exp.Stddev: rename_func("STDEV"),
 989            exp.StrPosition: lambda self, e: strposition_sql(
 990                self, e, func_name="CHARINDEX", supports_position=True
 991            ),
 992            exp.Subquery: transforms.preprocess([qualify_derived_table_outputs]),
 993            exp.SHA: lambda self, e: self.func("HASHBYTES", exp.Literal.string("SHA1"), e.this),
 994            exp.SHA2: lambda self, e: self.func(
 995                "HASHBYTES", exp.Literal.string(f"SHA2_{e.args.get('length', 256)}"), e.this
 996            ),
 997            exp.TemporaryProperty: lambda self, e: "",
 998            exp.TimeStrToTime: _timestrtotime_sql,
 999            exp.TimeToStr: _format_sql,
1000            exp.Trim: trim_sql,
1001            exp.TsOrDsAdd: date_delta_sql("DATEADD", cast=True),
1002            exp.TsOrDsDiff: date_delta_sql("DATEDIFF"),
1003            exp.TimestampTrunc: lambda self, e: self.func("DATETRUNC", e.unit, e.this),
1004            exp.Uuid: lambda *_: "NEWID()",
1005            exp.DateFromParts: rename_func("DATEFROMPARTS"),
1006        }
1007
1008        TRANSFORMS.pop(exp.ReturnsProperty)
1009
1010        PROPERTIES_LOCATION = {
1011            **generator.Generator.PROPERTIES_LOCATION,
1012            exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED,
1013        }
1014
1015        def scope_resolution(self, rhs: str, scope_name: str) -> str:
1016            return f"{scope_name}::{rhs}"
1017
1018        def select_sql(self, expression: exp.Select) -> str:
1019            limit = expression.args.get("limit")
1020            offset = expression.args.get("offset")
1021
1022            if isinstance(limit, exp.Fetch) and not offset:
1023                # Dialects like Oracle can FETCH directly from a row set but
1024                # T-SQL requires an ORDER BY + OFFSET clause in order to FETCH
1025                offset = exp.Offset(expression=exp.Literal.number(0))
1026                expression.set("offset", offset)
1027
1028            if offset:
1029                if not expression.args.get("order"):
1030                    # ORDER BY is required in order to use OFFSET in a query, so we use
1031                    # a noop order by, since we don't really care about the order.
1032                    # See: https://www.microsoftpressstore.com/articles/article.aspx?p=2314819
1033                    expression.order_by(exp.select(exp.null()).subquery(), copy=False)
1034
1035                if isinstance(limit, exp.Limit):
1036                    # TOP and OFFSET can't be combined, we need use FETCH instead of TOP
1037                    # we replace here because otherwise TOP would be generated in select_sql
1038                    limit.replace(exp.Fetch(direction="FIRST", count=limit.expression))
1039
1040            return super().select_sql(expression)
1041
1042        def convert_sql(self, expression: exp.Convert) -> str:
1043            name = "TRY_CONVERT" if expression.args.get("safe") else "CONVERT"
1044            return self.func(
1045                name, expression.this, expression.expression, expression.args.get("style")
1046            )
1047
1048        def queryoption_sql(self, expression: exp.QueryOption) -> str:
1049            option = self.sql(expression, "this")
1050            value = self.sql(expression, "expression")
1051            if value:
1052                optional_equal_sign = "= " if option in OPTIONS_THAT_REQUIRE_EQUAL else ""
1053                return f"{option} {optional_equal_sign}{value}"
1054            return option
1055
1056        def lateral_op(self, expression: exp.Lateral) -> str:
1057            cross_apply = expression.args.get("cross_apply")
1058            if cross_apply is True:
1059                return "CROSS APPLY"
1060            if cross_apply is False:
1061                return "OUTER APPLY"
1062
1063            # TODO: perhaps we can check if the parent is a Join and transpile it appropriately
1064            self.unsupported("LATERAL clause is not supported.")
1065            return "LATERAL"
1066
1067        def splitpart_sql(self: TSQL.Generator, expression: exp.SplitPart) -> str:
1068            this = expression.this
1069            split_count = len(this.name.split("."))
1070            delimiter = expression.args.get("delimiter")
1071            part_index = expression.args.get("part_index")
1072
1073            if (
1074                not all(isinstance(arg, exp.Literal) for arg in (this, delimiter, part_index))
1075                or (delimiter and delimiter.name != ".")
1076                or not part_index
1077                or split_count > 4
1078            ):
1079                self.unsupported(
1080                    "SPLIT_PART can be transpiled to PARSENAME only for '.' delimiter and literal values"
1081                )
1082                return ""
1083
1084            return self.func(
1085                "PARSENAME", this, exp.Literal.number(split_count + 1 - part_index.to_py())
1086            )
1087
1088        def timefromparts_sql(self, expression: exp.TimeFromParts) -> str:
1089            nano = expression.args.get("nano")
1090            if nano is not None:
1091                nano.pop()
1092                self.unsupported("Specifying nanoseconds is not supported in TIMEFROMPARTS.")
1093
1094            if expression.args.get("fractions") is None:
1095                expression.set("fractions", exp.Literal.number(0))
1096            if expression.args.get("precision") is None:
1097                expression.set("precision", exp.Literal.number(0))
1098
1099            return rename_func("TIMEFROMPARTS")(self, expression)
1100
1101        def timestampfromparts_sql(self, expression: exp.TimestampFromParts) -> str:
1102            zone = expression.args.get("zone")
1103            if zone is not None:
1104                zone.pop()
1105                self.unsupported("Time zone is not supported in DATETIMEFROMPARTS.")
1106
1107            nano = expression.args.get("nano")
1108            if nano is not None:
1109                nano.pop()
1110                self.unsupported("Specifying nanoseconds is not supported in DATETIMEFROMPARTS.")
1111
1112            if expression.args.get("milli") is None:
1113                expression.set("milli", exp.Literal.number(0))
1114
1115            return rename_func("DATETIMEFROMPARTS")(self, expression)
1116
1117        def setitem_sql(self, expression: exp.SetItem) -> str:
1118            this = expression.this
1119            if isinstance(this, exp.EQ) and not isinstance(this.left, exp.Parameter):
1120                # T-SQL does not use '=' in SET command, except when the LHS is a variable.
1121                return f"{self.sql(this.left)} {self.sql(this.right)}"
1122
1123            return super().setitem_sql(expression)
1124
1125        def boolean_sql(self, expression: exp.Boolean) -> str:
1126            if type(expression.parent) in BIT_TYPES or isinstance(
1127                expression.find_ancestor(exp.Values, exp.Select), exp.Values
1128            ):
1129                return "1" if expression.this else "0"
1130
1131            return "(1 = 1)" if expression.this else "(1 = 0)"
1132
1133        def is_sql(self, expression: exp.Is) -> str:
1134            if isinstance(expression.expression, exp.Boolean):
1135                return self.binary(expression, "=")
1136            return self.binary(expression, "IS")
1137
1138        def createable_sql(self, expression: exp.Create, locations: t.DefaultDict) -> str:
1139            sql = self.sql(expression, "this")
1140            properties = expression.args.get("properties")
1141
1142            if sql[:1] != "#" and any(
1143                isinstance(prop, exp.TemporaryProperty)
1144                for prop in (properties.expressions if properties else [])
1145            ):
1146                sql = f"[#{sql[1:]}" if sql.startswith("[") else f"#{sql}"
1147
1148            return sql
1149
1150        def create_sql(self, expression: exp.Create) -> str:
1151            kind = expression.kind
1152            exists = expression.args.pop("exists", None)
1153
1154            like_property = expression.find(exp.LikeProperty)
1155            if like_property:
1156                ctas_expression = like_property.this
1157            else:
1158                ctas_expression = expression.expression
1159
1160            if kind == "VIEW":
1161                expression.this.set("catalog", None)
1162                with_ = expression.args.get("with")
1163                if ctas_expression and with_:
1164                    # We've already preprocessed the Create expression to bubble up any nested CTEs,
1165                    # but CREATE VIEW actually requires the WITH clause to come after it so we need
1166                    # to amend the AST by moving the CTEs to the CREATE VIEW statement's query.
1167                    ctas_expression.set("with", with_.pop())
1168
1169            sql = super().create_sql(expression)
1170
1171            table = expression.find(exp.Table)
1172
1173            # Convert CTAS statement to SELECT .. INTO ..
1174            if kind == "TABLE" and ctas_expression:
1175                if isinstance(ctas_expression, exp.UNWRAPPED_QUERIES):
1176                    ctas_expression = ctas_expression.subquery()
1177
1178                properties = expression.args.get("properties") or exp.Properties()
1179                is_temp = any(isinstance(p, exp.TemporaryProperty) for p in properties.expressions)
1180
1181                select_into = exp.select("*").from_(exp.alias_(ctas_expression, "temp", table=True))
1182                select_into.set("into", exp.Into(this=table, temporary=is_temp))
1183
1184                if like_property:
1185                    select_into.limit(0, copy=False)
1186
1187                sql = self.sql(select_into)
1188
1189            if exists:
1190                identifier = self.sql(exp.Literal.string(exp.table_name(table) if table else ""))
1191                sql_with_ctes = self.prepend_ctes(expression, sql)
1192                sql_literal = self.sql(exp.Literal.string(sql_with_ctes))
1193                if kind == "SCHEMA":
1194                    return f"""IF NOT EXISTS (SELECT * FROM information_schema.schemata WHERE schema_name = {identifier}) EXEC({sql_literal})"""
1195                elif kind == "TABLE":
1196                    assert table
1197                    where = exp.and_(
1198                        exp.column("table_name").eq(table.name),
1199                        exp.column("table_schema").eq(table.db) if table.db else None,
1200                        exp.column("table_catalog").eq(table.catalog) if table.catalog else None,
1201                    )
1202                    return f"""IF NOT EXISTS (SELECT * FROM information_schema.tables WHERE {where}) EXEC({sql_literal})"""
1203                elif kind == "INDEX":
1204                    index = self.sql(exp.Literal.string(expression.this.text("this")))
1205                    return f"""IF NOT EXISTS (SELECT * FROM sys.indexes WHERE object_id = object_id({identifier}) AND name = {index}) EXEC({sql_literal})"""
1206            elif expression.args.get("replace"):
1207                sql = sql.replace("CREATE OR REPLACE ", "CREATE OR ALTER ", 1)
1208
1209            return self.prepend_ctes(expression, sql)
1210
1211        @generator.unsupported_args("unlogged", "expressions")
1212        def into_sql(self, expression: exp.Into) -> str:
1213            if expression.args.get("temporary"):
1214                # If the Into expression has a temporary property, push this down to the Identifier
1215                table = expression.find(exp.Table)
1216                if table and isinstance(table.this, exp.Identifier):
1217                    table.this.set("temporary", True)
1218
1219            return f"{self.seg('INTO')} {self.sql(expression, 'this')}"
1220
1221        def count_sql(self, expression: exp.Count) -> str:
1222            func_name = "COUNT_BIG" if expression.args.get("big_int") else "COUNT"
1223            return rename_func(func_name)(self, expression)
1224
1225        def offset_sql(self, expression: exp.Offset) -> str:
1226            return f"{super().offset_sql(expression)} ROWS"
1227
1228        def version_sql(self, expression: exp.Version) -> str:
1229            name = "SYSTEM_TIME" if expression.name == "TIMESTAMP" else expression.name
1230            this = f"FOR {name}"
1231            expr = expression.expression
1232            kind = expression.text("kind")
1233            if kind in ("FROM", "BETWEEN"):
1234                args = expr.expressions
1235                sep = "TO" if kind == "FROM" else "AND"
1236                expr_sql = f"{self.sql(seq_get(args, 0))} {sep} {self.sql(seq_get(args, 1))}"
1237            else:
1238                expr_sql = self.sql(expr)
1239
1240            expr_sql = f" {expr_sql}" if expr_sql else ""
1241            return f"{this} {kind}{expr_sql}"
1242
1243        def returnsproperty_sql(self, expression: exp.ReturnsProperty) -> str:
1244            table = expression.args.get("table")
1245            table = f"{table} " if table else ""
1246            return f"RETURNS {table}{self.sql(expression, 'this')}"
1247
1248        def returning_sql(self, expression: exp.Returning) -> str:
1249            into = self.sql(expression, "into")
1250            into = self.seg(f"INTO {into}") if into else ""
1251            return f"{self.seg('OUTPUT')} {self.expressions(expression, flat=True)}{into}"
1252
1253        def transaction_sql(self, expression: exp.Transaction) -> str:
1254            this = self.sql(expression, "this")
1255            this = f" {this}" if this else ""
1256            mark = self.sql(expression, "mark")
1257            mark = f" WITH MARK {mark}" if mark else ""
1258            return f"BEGIN TRANSACTION{this}{mark}"
1259
1260        def commit_sql(self, expression: exp.Commit) -> str:
1261            this = self.sql(expression, "this")
1262            this = f" {this}" if this else ""
1263            durability = expression.args.get("durability")
1264            durability = (
1265                f" WITH (DELAYED_DURABILITY = {'ON' if durability else 'OFF'})"
1266                if durability is not None
1267                else ""
1268            )
1269            return f"COMMIT TRANSACTION{this}{durability}"
1270
1271        def rollback_sql(self, expression: exp.Rollback) -> str:
1272            this = self.sql(expression, "this")
1273            this = f" {this}" if this else ""
1274            return f"ROLLBACK TRANSACTION{this}"
1275
1276        def identifier_sql(self, expression: exp.Identifier) -> str:
1277            identifier = super().identifier_sql(expression)
1278
1279            if expression.args.get("global"):
1280                identifier = f"##{identifier}"
1281            elif expression.args.get("temporary"):
1282                identifier = f"#{identifier}"
1283
1284            return identifier
1285
1286        def constraint_sql(self, expression: exp.Constraint) -> str:
1287            this = self.sql(expression, "this")
1288            expressions = self.expressions(expression, flat=True, sep=" ")
1289            return f"CONSTRAINT {this} {expressions}"
1290
1291        def length_sql(self, expression: exp.Length) -> str:
1292            return self._uncast_text(expression, "LEN")
1293
1294        def right_sql(self, expression: exp.Right) -> str:
1295            return self._uncast_text(expression, "RIGHT")
1296
1297        def left_sql(self, expression: exp.Left) -> str:
1298            return self._uncast_text(expression, "LEFT")
1299
1300        def _uncast_text(self, expression: exp.Expression, name: str) -> str:
1301            this = expression.this
1302            if isinstance(this, exp.Cast) and this.is_type(exp.DataType.Type.TEXT):
1303                this_sql = self.sql(this, "this")
1304            else:
1305                this_sql = self.sql(this)
1306            expression_sql = self.sql(expression, "expression")
1307            return self.func(name, this_sql, expression_sql if expression_sql else None)
1308
1309        def partition_sql(self, expression: exp.Partition) -> str:
1310            return f"WITH (PARTITIONS({self.expressions(expression, flat=True)}))"
1311
1312        def alter_sql(self, expression: exp.Alter) -> str:
1313            action = seq_get(expression.args.get("actions") or [], 0)
1314            if isinstance(action, exp.AlterRename):
1315                return f"EXEC sp_rename '{self.sql(expression.this)}', '{action.this.name}'"
1316            return super().alter_sql(expression)
1317
1318        def drop_sql(self, expression: exp.Drop) -> str:
1319            if expression.args["kind"] == "VIEW":
1320                expression.this.set("catalog", None)
1321            return super().drop_sql(expression)
1322
1323        def options_modifier(self, expression: exp.Expression) -> str:
1324            options = self.expressions(expression, key="options")
1325            return f" OPTION{self.wrap(options)}" if options else ""
1326
1327        def dpipe_sql(self, expression: exp.DPipe) -> str:
1328            return self.sql(
1329                reduce(lambda x, y: exp.Add(this=x, expression=y), expression.flatten())
1330            )
1331
1332        def isascii_sql(self, expression: exp.IsAscii) -> str:
1333            return f"(PATINDEX(CONVERT(VARCHAR(MAX), 0x255b5e002d7f5d25) COLLATE Latin1_General_BIN, {self.sql(expression.this)}) = 0)"
1334
1335        def columndef_sql(self, expression: exp.ColumnDef, sep: str = " ") -> str:
1336            this = super().columndef_sql(expression, sep)
1337            default = self.sql(expression, "default")
1338            default = f" = {default}" if default else ""
1339            output = self.sql(expression, "output")
1340            output = f" {output}" if output else ""
1341            return f"{this}{default}{output}"