1"""sqlglot expressions query."""
   2
   3from __future__ import annotations
   4
   5import typing as t
   6
   7from sqlglot.errors import ParseError
   8from sqlglot.helper import trait, ensure_list
   9from sqlglot.expressions.core import (
  10    Aliases,
  11    Column,
  12    Condition,
  13    Distinct,
  14    Dot,
  15    DynamicIdentifier,
  16    Expr,
  17    Expression,
  18    Func,
  19    Hint,
  20    Identifier,
  21    In,
  22    _apply_builder,
  23    _apply_child_list_builder,
  24    _apply_list_builder,
  25    _apply_conjunction_builder,
  26    _apply_set_operation,
  27    ExpOrStr,
  28    QUERY_MODIFIERS,
  29    maybe_parse,
  30    maybe_copy,
  31    to_identifier,
  32    convert,
  33    and_,
  34    alias_,
  35    column,
  36)
  37
  38if t.TYPE_CHECKING:
  39    from sqlglot.dialects.dialect import DialectType
  40    from sqlglot.expressions.datatypes import DataType
  41    from sqlglot.expressions.constraints import ColumnConstraint
  42    from sqlglot.expressions.ddl import Create
  43    from sqlglot.expressions.array import Unnest
  44    from sqlglot._typing import E, ParserArgs, ParserNoDialectArgs
  45    from typing_extensions import Unpack
  46
  47    S = t.TypeVar("S", bound="SetOperation")
  48    Q = t.TypeVar("Q", bound="Query")
  49
  50
  51def _apply_cte_builder(
  52    instance: E,
  53    alias: ExpOrStr,
  54    as_: ExpOrStr,
  55    recursive: bool | None = None,
  56    materialized: bool | None = None,
  57    append: bool = True,
  58    dialect: DialectType = None,
  59    copy: bool = True,
  60    scalar: bool | None = None,
  61    **opts: Unpack[ParserNoDialectArgs],
  62) -> E:
  63    alias_expression = maybe_parse(alias, dialect=dialect, into=TableAlias, **opts)
  64    as_expression = maybe_parse(as_, dialect=dialect, copy=copy, **opts)
  65    if scalar and not isinstance(as_expression, Subquery):
  66        # scalar CTE must be wrapped in a subquery
  67        as_expression = Subquery(this=as_expression)
  68    cte = CTE(this=as_expression, alias=alias_expression, materialized=materialized, scalar=scalar)
  69    return _apply_child_list_builder(
  70        cte,
  71        instance=instance,
  72        arg="with_",
  73        append=append,
  74        copy=copy,
  75        into=With,
  76        properties={"recursive": recursive} if recursive else {},
  77    )
  78
  79
  80@trait
  81class Selectable(Expr):
  82    @property
  83    def selects(self) -> list[Expr]:
  84        raise NotImplementedError("Subclasses must implement selects")
  85
  86    @property
  87    def named_selects(self) -> list[str]:
  88        return _named_selects(self)
  89
  90
  91def _named_selects(self: Expr) -> list[str]:
  92    selectable = t.cast(Selectable, self)
  93    return [select.output_name for select in selectable.selects]
  94
  95
  96@trait
  97class DerivedTable(Selectable):
  98    @property
  99    def selects(self) -> list[Expr]:
 100        this = self.this
 101        return this.selects if isinstance(this, Query) else []
 102
 103
 104@trait
 105class UDTF(DerivedTable):
 106    @property
 107    def selects(self) -> list[Expr]:
 108        alias = self.args.get("alias")
 109        return alias.columns if alias else []
 110
 111
 112@trait
 113class Query(Selectable):
 114    """Trait for any SELECT/UNION/etc. query expression."""
 115
 116    @property
 117    def ctes(self) -> list[CTE]:
 118        with_ = self.args.get("with_")
 119        return with_.expressions if with_ else []
 120
 121    def select(
 122        self: Q,
 123        *expressions: ExpOrStr | None,
 124        append: bool = True,
 125        dialect: DialectType = None,
 126        copy: bool = True,
 127        **opts: Unpack[ParserNoDialectArgs],
 128    ) -> Q:
 129        raise NotImplementedError("Query objects must implement `select`")
 130
 131    def subquery(self, alias: ExpOrStr | None = None, copy: bool = True) -> Subquery:
 132        """
 133        Returns a `Subquery` that wraps around this query.
 134
 135        Example:
 136            >>> subquery = Select().select("x").from_("tbl").subquery()
 137            >>> Select().select("x").from_(subquery).sql()
 138            'SELECT x FROM (SELECT x FROM tbl)'
 139
 140        Args:
 141            alias: an optional alias for the subquery.
 142            copy: if `False`, modify this expression instance in-place.
 143        """
 144        instance = maybe_copy(self, copy)
 145        if not isinstance(alias, Expr):
 146            alias = TableAlias(this=to_identifier(alias)) if alias else None
 147
 148        return Subquery(this=instance, alias=alias)
 149
 150    def limit(
 151        self: Q,
 152        expression: ExpOrStr | int,
 153        dialect: DialectType = None,
 154        copy: bool = True,
 155        **opts: Unpack[ParserNoDialectArgs],
 156    ) -> Q:
 157        """
 158        Adds a LIMIT clause to this query.
 159
 160        Example:
 161            >>> Select().select("1").union(Select().select("1")).limit(1).sql()
 162            'SELECT 1 UNION SELECT 1 LIMIT 1'
 163
 164        Args:
 165            expression: the SQL code string to parse.
 166                This can also be an integer.
 167                If a `Limit` instance is passed, it will be used as-is.
 168                If another `Expr` instance is passed, it will be wrapped in a `Limit`.
 169            dialect: the dialect used to parse the input expression.
 170            copy: if `False`, modify this expression instance in-place.
 171            opts: other options to use to parse the input expressions.
 172
 173        Returns:
 174            A limited Select expression.
 175        """
 176        return _apply_builder(
 177            expression=expression,
 178            instance=self,
 179            arg="limit",
 180            into=Limit,
 181            prefix="LIMIT",
 182            dialect=dialect,
 183            copy=copy,
 184            into_arg="expression",
 185            **opts,
 186        )
 187
 188    def offset(
 189        self: Q,
 190        expression: ExpOrStr | int,
 191        dialect: DialectType = None,
 192        copy: bool = True,
 193        **opts: Unpack[ParserNoDialectArgs],
 194    ) -> Q:
 195        """
 196        Set the OFFSET expression.
 197
 198        Example:
 199            >>> Select().from_("tbl").select("x").offset(10).sql()
 200            'SELECT x FROM tbl OFFSET 10'
 201
 202        Args:
 203            expression: the SQL code string to parse.
 204                This can also be an integer.
 205                If a `Offset` instance is passed, this is used as-is.
 206                If another `Expr` instance is passed, it will be wrapped in a `Offset`.
 207            dialect: the dialect used to parse the input expression.
 208            copy: if `False`, modify this expression instance in-place.
 209            opts: other options to use to parse the input expressions.
 210
 211        Returns:
 212            The modified Select expression.
 213        """
 214        return _apply_builder(
 215            expression=expression,
 216            instance=self,
 217            arg="offset",
 218            into=Offset,
 219            prefix="OFFSET",
 220            dialect=dialect,
 221            copy=copy,
 222            into_arg="expression",
 223            **opts,
 224        )
 225
 226    def order_by(
 227        self: Q,
 228        *expressions: ExpOrStr | None,
 229        append: bool = True,
 230        dialect: DialectType = None,
 231        copy: bool = True,
 232        **opts: Unpack[ParserNoDialectArgs],
 233    ) -> Q:
 234        """
 235        Set the ORDER BY expression.
 236
 237        Example:
 238            >>> Select().from_("tbl").select("x").order_by("x DESC").sql()
 239            'SELECT x FROM tbl ORDER BY x DESC'
 240
 241        Args:
 242            *expressions: the SQL code strings to parse.
 243                If a `Group` instance is passed, this is used as-is.
 244                If another `Expr` instance is passed, it will be wrapped in a `Order`.
 245            append: if `True`, add to any existing expressions.
 246                Otherwise, this flattens all the `Order` expression into a single expression.
 247            dialect: the dialect used to parse the input expression.
 248            copy: if `False`, modify this expression instance in-place.
 249            opts: other options to use to parse the input expressions.
 250
 251        Returns:
 252            The modified Select expression.
 253        """
 254        return _apply_child_list_builder(
 255            *expressions,
 256            instance=self,
 257            arg="order",
 258            append=append,
 259            copy=copy,
 260            prefix="ORDER BY",
 261            into=Order,
 262            dialect=dialect,
 263            **opts,
 264        )
 265
 266    def where(
 267        self: Q,
 268        *expressions: ExpOrStr | None,
 269        append: bool = True,
 270        dialect: DialectType = None,
 271        copy: bool = True,
 272        **opts: Unpack[ParserNoDialectArgs],
 273    ) -> Q:
 274        """
 275        Append to or set the WHERE expressions.
 276
 277        Examples:
 278            >>> Select().select("x").from_("tbl").where("x = 'a' OR x < 'b'").sql()
 279            "SELECT x FROM tbl WHERE x = 'a' OR x < 'b'"
 280
 281        Args:
 282            *expressions: the SQL code strings to parse.
 283                If an `Expr` instance is passed, it will be used as-is.
 284                Multiple expressions are combined with an AND operator.
 285            append: if `True`, AND the new expressions to any existing expression.
 286                Otherwise, this resets the expression.
 287            dialect: the dialect used to parse the input expressions.
 288            copy: if `False`, modify this expression instance in-place.
 289            opts: other options to use to parse the input expressions.
 290
 291        Returns:
 292            The modified expression.
 293        """
 294        return _apply_conjunction_builder(
 295            *[expr.this if isinstance(expr, Where) else expr for expr in expressions],
 296            instance=self,
 297            arg="where",
 298            append=append,
 299            into=Where,
 300            dialect=dialect,
 301            copy=copy,
 302            **opts,
 303        )
 304
 305    def with_(
 306        self: Q,
 307        alias: ExpOrStr,
 308        as_: ExpOrStr,
 309        recursive: bool | None = None,
 310        materialized: bool | None = None,
 311        append: bool = True,
 312        dialect: DialectType = None,
 313        copy: bool = True,
 314        scalar: bool | None = None,
 315        **opts: Unpack[ParserNoDialectArgs],
 316    ) -> Q:
 317        """
 318        Append to or set the common table expressions.
 319
 320        Example:
 321            >>> Select().with_("tbl2", as_="SELECT * FROM tbl").select("x").from_("tbl2").sql()
 322            'WITH tbl2 AS (SELECT * FROM tbl) SELECT x FROM tbl2'
 323
 324        Args:
 325            alias: the SQL code string to parse as the table name.
 326                If an `Expr` instance is passed, this is used as-is.
 327            as_: the SQL code string to parse as the table expression.
 328                If an `Expr` instance is passed, it will be used as-is.
 329            recursive: set the RECURSIVE part of the expression. Defaults to `False`.
 330            materialized: set the MATERIALIZED part of the expression.
 331            append: if `True`, add to any existing expressions.
 332                Otherwise, this resets the expressions.
 333            dialect: the dialect used to parse the input expression.
 334            copy: if `False`, modify this expression instance in-place.
 335            scalar: if `True`, this is a scalar common table expression.
 336            opts: other options to use to parse the input expressions.
 337
 338        Returns:
 339            The modified expression.
 340        """
 341        return _apply_cte_builder(
 342            self,
 343            alias,
 344            as_,
 345            recursive=recursive,
 346            materialized=materialized,
 347            append=append,
 348            dialect=dialect,
 349            copy=copy,
 350            scalar=scalar,
 351            **opts,
 352        )
 353
 354    def union(
 355        self,
 356        *expressions: ExpOrStr,
 357        distinct: bool = True,
 358        dialect: DialectType = None,
 359        copy: bool = True,
 360        **opts: Unpack[ParserNoDialectArgs],
 361    ) -> Union:
 362        """
 363        Builds a UNION expression.
 364
 365        Example:
 366            >>> import sqlglot
 367            >>> sqlglot.parse_one("SELECT * FROM foo").union("SELECT * FROM bla").sql()
 368            'SELECT * FROM foo UNION SELECT * FROM bla'
 369
 370        Args:
 371            expressions: the SQL code strings.
 372                If `Expr` instances are passed, they will be used as-is.
 373            distinct: set the DISTINCT flag if and only if this is true.
 374            dialect: the dialect used to parse the input expression.
 375            opts: other options to use to parse the input expressions.
 376
 377        Returns:
 378            The new Union expression.
 379        """
 380        return union(self, *expressions, distinct=distinct, dialect=dialect, copy=copy, **opts)
 381
 382    def intersect(
 383        self,
 384        *expressions: ExpOrStr,
 385        distinct: bool = True,
 386        dialect: DialectType = None,
 387        copy: bool = True,
 388        **opts: Unpack[ParserNoDialectArgs],
 389    ) -> Intersect:
 390        """
 391        Builds an INTERSECT expression.
 392
 393        Example:
 394            >>> import sqlglot
 395            >>> sqlglot.parse_one("SELECT * FROM foo").intersect("SELECT * FROM bla").sql()
 396            'SELECT * FROM foo INTERSECT SELECT * FROM bla'
 397
 398        Args:
 399            expressions: the SQL code strings.
 400                If `Expr` instances are passed, they will be used as-is.
 401            distinct: set the DISTINCT flag if and only if this is true.
 402            dialect: the dialect used to parse the input expression.
 403            opts: other options to use to parse the input expressions.
 404
 405        Returns:
 406            The new Intersect expression.
 407        """
 408        return intersect(self, *expressions, distinct=distinct, dialect=dialect, copy=copy, **opts)
 409
 410    def except_(
 411        self,
 412        *expressions: ExpOrStr,
 413        distinct: bool = True,
 414        dialect: DialectType = None,
 415        copy: bool = True,
 416        **opts: Unpack[ParserNoDialectArgs],
 417    ) -> Except:
 418        """
 419        Builds an EXCEPT expression.
 420
 421        Example:
 422            >>> import sqlglot
 423            >>> sqlglot.parse_one("SELECT * FROM foo").except_("SELECT * FROM bla").sql()
 424            'SELECT * FROM foo EXCEPT SELECT * FROM bla'
 425
 426        Args:
 427            expressions: the SQL code strings.
 428                If `Expr` instance are passed, they will be used as-is.
 429            distinct: set the DISTINCT flag if and only if this is true.
 430            dialect: the dialect used to parse the input expression.
 431            opts: other options to use to parse the input expressions.
 432
 433        Returns:
 434            The new Except expression.
 435        """
 436        return except_(self, *expressions, distinct=distinct, dialect=dialect, copy=copy, **opts)
 437
 438
 439class QueryBand(Expression):
 440    arg_types = {"this": True, "scope": False, "update": False}
 441
 442
 443class RecursiveWithSearch(Expression):
 444    arg_types = {"kind": True, "this": True, "expression": True, "using": False}
 445
 446
 447class With(Expression):
 448    arg_types = {"expressions": False, "recursive": False, "search": False, "udfs": False}
 449
 450    @property
 451    def recursive(self) -> bool:
 452        return bool(self.args.get("recursive"))
 453
 454
 455class CTE(Expression, DerivedTable):
 456    arg_types = {
 457        "this": True,
 458        "alias": True,
 459        "scalar": False,
 460        "materialized": False,
 461        "key_expressions": False,
 462    }
 463
 464
 465class ProjectionDef(Expression):
 466    arg_types = {"this": True, "expression": True}
 467
 468
 469class TableAlias(Expression):
 470    arg_types = {"this": False, "columns": False}
 471
 472    @property
 473    def columns(self) -> list[t.Any]:
 474        return self.args.get("columns") or []
 475
 476
 477class BitString(Expression, Condition):
 478    is_primitive = True
 479
 480
 481class HexString(Expression, Condition):
 482    arg_types = {"this": True, "is_integer": False}
 483    is_primitive = True
 484
 485
 486class ByteString(Expression, Condition):
 487    arg_types = {"this": True, "is_bytes": False}
 488    is_primitive = True
 489
 490
 491class RawString(Expression, Condition):
 492    is_primitive = True
 493
 494
 495class UnicodeString(Expression, Condition):
 496    arg_types = {"this": True, "escape": False}
 497
 498
 499class ColumnPosition(Expression):
 500    arg_types = {"this": False, "position": True}
 501
 502
 503class ColumnDef(Expression):
 504    arg_types = {
 505        "this": True,
 506        "kind": False,
 507        "constraints": False,
 508        "exists": False,
 509        "position": False,
 510        "default": False,
 511        "output": False,
 512    }
 513
 514    @property
 515    def constraints(self) -> list[ColumnConstraint]:
 516        return self.args.get("constraints") or []
 517
 518    @property
 519    def kind(self) -> DataType | None:
 520        return self.args.get("kind")
 521
 522
 523class Changes(Expression):
 524    arg_types = {"information": True, "at_before": False, "end": False}
 525
 526
 527class Connect(Expression):
 528    arg_types = {"start": False, "connect": True, "nocycle": False}
 529
 530
 531class Prior(Expression):
 532    pass
 533
 534
 535class Into(Expression):
 536    arg_types = {
 537        "this": False,
 538        "temporary": False,
 539        "unlogged": False,
 540        "bulk_collect": False,
 541        "expressions": False,
 542    }
 543
 544
 545class From(Expression):
 546    @property
 547    def name(self) -> str:
 548        return self.this.name
 549
 550    @property
 551    def alias_or_name(self) -> str:
 552        return self.this.alias_or_name
 553
 554
 555class Having(Expression):
 556    pass
 557
 558
 559class Index(Expression):
 560    arg_types = {
 561        "this": False,
 562        "table": False,
 563        "unique": False,
 564        "primary": False,
 565        "amp": False,  # teradata
 566        "params": False,
 567    }
 568
 569
 570class ConditionalInsert(Expression):
 571    arg_types = {"this": True, "expression": False, "else_": False}
 572
 573
 574class MultitableInserts(Expression):
 575    arg_types = {"expressions": True, "kind": True, "source": True}
 576
 577
 578class OnCondition(Expression):
 579    arg_types = {"error": False, "empty": False, "null": False}
 580
 581
 582class Introducer(Expression):
 583    arg_types = {"this": True, "expression": True}
 584
 585
 586class National(Expression):
 587    is_primitive = True
 588
 589
 590class Partition(Expression):
 591    arg_types = {"expressions": True, "subpartition": False}
 592
 593
 594class PartitionRange(Expression):
 595    arg_types = {"this": True, "expression": False, "expressions": False}
 596
 597
 598class PartitionId(Expression):
 599    pass
 600
 601
 602class Fetch(Expression):
 603    arg_types = {
 604        "direction": False,
 605        "count": False,
 606        "limit_options": False,
 607    }
 608
 609
 610class Grant(Expression):
 611    arg_types = {
 612        "privileges": True,
 613        "kind": False,
 614        "securable": True,
 615        "principals": True,
 616        "grant_option": False,
 617    }
 618
 619
 620class Revoke(Expression):
 621    arg_types = {**Grant.arg_types, "cascade": False}
 622
 623
 624class Group(Expression):
 625    arg_types = {
 626        "expressions": False,
 627        "grouping_sets": False,
 628        "cube": False,
 629        "rollup": False,
 630        "totals": False,
 631        "all": False,
 632    }
 633
 634
 635class Cube(Expression):
 636    arg_types = {"expressions": False}
 637
 638
 639class Rollup(Expression):
 640    arg_types = {"expressions": False}
 641
 642
 643class GroupingSets(Expression):
 644    arg_types = {"expressions": True}
 645
 646
 647class Lambda(Expression):
 648    arg_types = {"this": True, "expressions": True, "colon": False}
 649
 650
 651class Limit(Expression):
 652    arg_types = {
 653        "this": False,
 654        "expression": True,
 655        "offset": False,
 656        "limit_options": False,
 657        "expressions": False,
 658    }
 659
 660
 661class LimitOptions(Expression):
 662    arg_types = {
 663        "percent": False,
 664        "rows": False,
 665        "with_ties": False,
 666    }
 667
 668
 669class Join(Expression):
 670    arg_types = {
 671        "this": True,
 672        "on": False,
 673        "side": False,
 674        "kind": False,
 675        "using": False,
 676        "method": False,
 677        "global_": False,
 678        "hint": False,
 679        "match_condition": False,  # Snowflake
 680        "directed": False,  # Snowflake
 681        "expressions": False,
 682        "pivots": False,
 683    }
 684
 685    @property
 686    def method(self) -> str:
 687        return self.text("method").upper()
 688
 689    @property
 690    def kind(self) -> str:
 691        return self.text("kind").upper()
 692
 693    @property
 694    def side(self) -> str:
 695        return self.text("side").upper()
 696
 697    @property
 698    def hint(self) -> str:
 699        return self.text("hint").upper()
 700
 701    @property
 702    def alias_or_name(self) -> str:
 703        return self.this.alias_or_name
 704
 705    @property
 706    def is_semi_or_anti_join(self) -> bool:
 707        return self.kind in ("SEMI", "ANTI")
 708
 709    def on(
 710        self,
 711        *expressions: ExpOrStr | None,
 712        append: bool = True,
 713        dialect: DialectType = None,
 714        copy: bool = True,
 715        **opts: Unpack[ParserNoDialectArgs],
 716    ) -> Join:
 717        """
 718        Append to or set the ON expressions.
 719
 720        Example:
 721            >>> import sqlglot
 722            >>> sqlglot.parse_one("JOIN x", into=Join).on("y = 1").sql()
 723            'JOIN x ON y = 1'
 724
 725        Args:
 726            *expressions: the SQL code strings to parse.
 727                If an `Expr` instance is passed, it will be used as-is.
 728                Multiple expressions are combined with an AND operator.
 729            append: if `True`, AND the new expressions to any existing expression.
 730                Otherwise, this resets the expression.
 731            dialect: the dialect used to parse the input expressions.
 732            copy: if `False`, modify this expression instance in-place.
 733            opts: other options to use to parse the input expressions.
 734
 735        Returns:
 736            The modified Join expression.
 737        """
 738        join = _apply_conjunction_builder(
 739            *expressions,
 740            instance=self,
 741            arg="on",
 742            append=append,
 743            dialect=dialect,
 744            copy=copy,
 745            **opts,
 746        )
 747
 748        if join.kind == "CROSS":
 749            join.set("kind", None)
 750
 751        return join
 752
 753    def using(
 754        self,
 755        *expressions: ExpOrStr | None,
 756        append: bool = True,
 757        dialect: DialectType = None,
 758        copy: bool = True,
 759        **opts: Unpack[ParserNoDialectArgs],
 760    ) -> Join:
 761        """
 762        Append to or set the USING expressions.
 763
 764        Example:
 765            >>> import sqlglot
 766            >>> sqlglot.parse_one("JOIN x", into=Join).using("foo", "bla").sql()
 767            'JOIN x USING (foo, bla)'
 768
 769        Args:
 770            *expressions: the SQL code strings to parse.
 771                If an `Expr` instance is passed, it will be used as-is.
 772            append: if `True`, concatenate the new expressions to the existing "using" list.
 773                Otherwise, this resets the expression.
 774            dialect: the dialect used to parse the input expressions.
 775            copy: if `False`, modify this expression instance in-place.
 776            opts: other options to use to parse the input expressions.
 777
 778        Returns:
 779            The modified Join expression.
 780        """
 781        join = _apply_list_builder(
 782            *expressions,
 783            instance=self,
 784            arg="using",
 785            append=append,
 786            dialect=dialect,
 787            copy=copy,
 788            **opts,
 789        )
 790
 791        if join.kind == "CROSS":
 792            join.set("kind", None)
 793
 794        return join
 795
 796
 797class Lateral(Expression, UDTF):
 798    arg_types = {
 799        "this": True,
 800        "view": False,
 801        "outer": False,
 802        "alias": False,
 803        "cross_apply": False,  # True -> CROSS APPLY, False -> OUTER APPLY
 804        "ordinality": False,
 805    }
 806
 807
 808class TableFromRows(Expression, UDTF):
 809    arg_types = {
 810        "this": True,
 811        "alias": False,
 812        "joins": False,
 813        "pivots": False,
 814        "sample": False,
 815    }
 816
 817
 818class MatchRecognizeMeasure(Expression):
 819    arg_types = {
 820        "this": True,
 821        "window_frame": False,
 822    }
 823
 824
 825class MatchRecognize(Expression):
 826    arg_types = {
 827        "partition_by": False,
 828        "order": False,
 829        "measures": False,
 830        "rows": False,
 831        "after": False,
 832        "pattern": False,
 833        "define": False,
 834        "alias": False,
 835    }
 836
 837
 838class Final(Expression):
 839    pass
 840
 841
 842class Offset(Expression):
 843    arg_types = {"this": False, "expression": True, "expressions": False}
 844
 845
 846class Order(Expression):
 847    arg_types = {"this": False, "expressions": True, "siblings": False}
 848
 849
 850class WithFill(Expression):
 851    arg_types = {
 852        "from_": False,
 853        "to": False,
 854        "step": False,
 855        "interpolate": False,
 856    }
 857
 858
 859class SkipJSONColumn(Expression):
 860    arg_types = {"regexp": False, "expression": True}
 861
 862
 863class Cluster(Expression):
 864    arg_types = {"expressions": True}
 865
 866
 867class Distribute(Order):
 868    pass
 869
 870
 871class Sort(Order):
 872    pass
 873
 874
 875class Qualify(Expression):
 876    pass
 877
 878
 879class InputOutputFormat(Expression):
 880    arg_types = {"input_format": False, "output_format": False}
 881
 882
 883class Return(Expression):
 884    pass
 885
 886
 887class Tuple(Expression):
 888    arg_types = {"expressions": False}
 889
 890    def isin(
 891        self,
 892        *expressions: t.Any,
 893        query: ExpOrStr | None = None,
 894        unnest: ExpOrStr | None | list[ExpOrStr] | tuple[ExpOrStr, ...] = None,
 895        copy: bool = True,
 896        **opts: Unpack[ParserArgs],
 897    ) -> In:
 898        return In(
 899            this=maybe_copy(self, copy),
 900            expressions=[convert(e, copy=copy) for e in expressions],
 901            query=maybe_parse(query, copy=copy, **opts) if query else None,
 902            unnest=(
 903                Unnest(
 904                    expressions=[
 905                        maybe_parse(e, copy=copy, **opts)
 906                        for e in t.cast(list[ExpOrStr], ensure_list(unnest))
 907                    ]
 908                )
 909                if unnest
 910                else None
 911            ),
 912        )
 913
 914
 915class QueryOption(Expression):
 916    arg_types = {"this": True, "expression": False}
 917
 918
 919# FOR { XML | JSON } query modifier; `kind` is the discriminant ("XML" or "JSON").
 920class ForClause(Expression):
 921    arg_types = {"kind": True, "expressions": False}
 922
 923
 924class WithTableHint(Expression):
 925    arg_types = {"expressions": True}
 926
 927
 928class IndexTableHint(Expression):
 929    arg_types = {"this": True, "expressions": False, "target": False}
 930
 931
 932class HistoricalData(Expression):
 933    arg_types = {"this": True, "kind": True, "expression": True}
 934
 935
 936class Put(Expression):
 937    arg_types = {"this": True, "target": True, "properties": False}
 938
 939
 940class Get(Expression):
 941    arg_types = {"this": True, "target": True, "properties": False}
 942
 943
 944class Table(Expression, Selectable):
 945    arg_types = {
 946        "this": False,
 947        "alias": False,
 948        "db": False,
 949        "catalog": False,
 950        "laterals": False,
 951        "joins": False,
 952        "pivots": False,
 953        "hints": False,
 954        "system_time": False,
 955        "version": False,
 956        "format": False,
 957        "pattern": False,
 958        "ordinality": False,
 959        "when": False,
 960        "only": False,
 961        "partition": False,
 962        "changes": False,
 963        "rows_from": False,
 964        "sample": False,
 965        "indexed": False,
 966    }
 967
 968    @property
 969    def name(self) -> str:
 970        this = self.this
 971        if not this or (isinstance(this, Func) and not isinstance(this, DynamicIdentifier)):
 972            return ""
 973        return this.name
 974
 975    @property
 976    def db(self) -> str:
 977        return self.text("db")
 978
 979    @property
 980    def catalog(self) -> str:
 981        return self.text("catalog")
 982
 983    @property
 984    def selects(self) -> list[Expr]:
 985        return []
 986
 987    @property
 988    def named_selects(self) -> list[str]:
 989        return []
 990
 991    @property
 992    def parts(self) -> list[Expr]:
 993        """Return the parts of a table in order catalog, db, table."""
 994        parts: list[Expr] = []
 995
 996        for arg in ("catalog", "db", "this"):
 997            part = self.args.get(arg)
 998
 999            if isinstance(part, Dot):
1000                parts.extend(part.flatten())
1001            elif isinstance(part, Expr):
1002                parts.append(part)
1003
1004        return parts
1005
1006    def to_column(self, copy: bool = True) -> Expr:
1007        parts = self.parts
1008        last_part = parts[-1]
1009
1010        if isinstance(last_part, Identifier):
1011            col: Expr = column(*reversed(parts[0:4]), fields=parts[4:], copy=copy)  # type: ignore
1012        else:
1013            # This branch will be reached if a function or array is wrapped in a `Table`
1014            col = last_part
1015
1016        alias = self.args.get("alias")
1017        if alias:
1018            col = alias_(col, alias.this, copy=copy)
1019
1020        return col
1021
1022
1023def _is_star(expression: Expr) -> bool:
1024    stack = [expression]
1025    while stack:
1026        node = stack.pop()
1027        if isinstance(node, SetOperation):
1028            stack.append(node.this)
1029            stack.append(node.expression)
1030        elif isinstance(node, Subquery):
1031            stack.append(node.this)
1032        elif node.is_star:
1033            return True
1034    return False
1035
1036
1037class SetOperation(Expression, Query):
1038    arg_types = {
1039        "with_": False,
1040        "this": True,
1041        "expression": True,
1042        "distinct": False,
1043        "by_name": False,
1044        "side": False,
1045        "kind": False,
1046        "on": False,
1047        **QUERY_MODIFIERS,
1048    }
1049
1050    def select(
1051        self: S,
1052        *expressions: ExpOrStr | None,
1053        append: bool = True,
1054        dialect: DialectType = None,
1055        copy: bool = True,
1056        **opts: Unpack[ParserNoDialectArgs],
1057    ) -> S:
1058        this = maybe_copy(self, copy)
1059        this.this.unnest().select(*expressions, append=append, dialect=dialect, copy=False, **opts)
1060        this.expression.unnest().select(
1061            *expressions, append=append, dialect=dialect, copy=False, **opts
1062        )
1063        return this
1064
1065    @property
1066    def named_selects(self) -> list[str]:
1067        expr: Expr = self
1068        while isinstance(expr, SetOperation):
1069            if expr.args.get("by_name"):
1070                left = t.cast(Selectable, expr.this.unnest()).named_selects
1071                right = t.cast(Selectable, expr.expression.unnest()).named_selects
1072                return list(dict.fromkeys(left + right))
1073
1074            expr = expr.this.unnest()
1075        return _named_selects(expr)
1076
1077    @property
1078    def is_star(self) -> bool:
1079        return _is_star(self)
1080
1081    @property
1082    def selects(self) -> list[Expr]:
1083        expr: Expr = self
1084        while isinstance(expr, SetOperation):
1085            expr = expr.this.unnest()
1086        return getattr(expr, "selects", [])
1087
1088    @property
1089    def left(self) -> Query:
1090        return self.this
1091
1092    @property
1093    def right(self) -> Query:
1094        return self.expression
1095
1096    @property
1097    def kind(self) -> str:
1098        return self.text("kind").upper()
1099
1100    @property
1101    def side(self) -> str:
1102        return self.text("side").upper()
1103
1104
1105class Union(SetOperation):
1106    pass
1107
1108
1109class Except(SetOperation):
1110    pass
1111
1112
1113class Intersect(SetOperation):
1114    pass
1115
1116
1117class Values(Expression, UDTF):
1118    arg_types = {
1119        "expressions": True,
1120        "alias": False,
1121        "order": False,
1122        "limit": False,
1123        "offset": False,
1124    }
1125
1126
1127class Version(Expression):
1128    """
1129    Time travel, iceberg, bigquery etc
1130    https://trino.io/docs/current/connector/iceberg.html?highlight=snapshot#using-snapshots
1131    https://www.databricks.com/blog/2019/02/04/introducing-delta-time-travel-for-large-scale-data-lakes.html
1132    https://cloud.google.com/bigquery/docs/reference/standard-sql/query-syntax#for_system_time_as_of
1133    https://learn.microsoft.com/en-us/sql/relational-databases/tables/querying-data-in-a-system-versioned-temporal-table?view=sql-server-ver16
1134    this is either TIMESTAMP or VERSION
1135    kind is ("AS OF", "BETWEEN")
1136    """
1137
1138    arg_types = {"this": True, "kind": True, "expression": False}
1139
1140
1141class Schema(Expression):
1142    arg_types = {"this": False, "expressions": False}
1143
1144
1145class Lock(Expression):
1146    arg_types = {"update": True, "expressions": False, "wait": False, "key": False}
1147
1148
1149class Select(Expression, Query):
1150    arg_types = {
1151        "with_": False,
1152        "kind": False,
1153        "expressions": False,
1154        "hint": False,
1155        "distinct": False,
1156        "into": False,
1157        "from_": False,
1158        "operation_modifiers": False,
1159        "exclude": False,
1160        **QUERY_MODIFIERS,
1161    }
1162
1163    def from_(
1164        self,
1165        expression: ExpOrStr,
1166        dialect: DialectType = None,
1167        copy: bool = True,
1168        **opts: Unpack[ParserNoDialectArgs],
1169    ) -> Select:
1170        """
1171        Set the FROM expression.
1172
1173        Example:
1174            >>> Select().from_("tbl").select("x").sql()
1175            'SELECT x FROM tbl'
1176
1177        Args:
1178            expression : the SQL code strings to parse.
1179                If a `From` instance is passed, this is used as-is.
1180                If another `Expr` instance is passed, it will be wrapped in a `From`.
1181            dialect: the dialect used to parse the input expression.
1182            copy: if `False`, modify this expression instance in-place.
1183            opts: other options to use to parse the input expressions.
1184
1185        Returns:
1186            The modified Select expression.
1187        """
1188        return _apply_builder(
1189            expression=expression,
1190            instance=self,
1191            arg="from_",
1192            into=From,
1193            prefix="FROM",
1194            dialect=dialect,
1195            copy=copy,
1196            **opts,
1197        )
1198
1199    def group_by(
1200        self,
1201        *expressions: ExpOrStr | None,
1202        append: bool = True,
1203        dialect: DialectType = None,
1204        copy: bool = True,
1205        **opts: Unpack[ParserNoDialectArgs],
1206    ) -> Select:
1207        """
1208        Set the GROUP BY expression.
1209
1210        Example:
1211            >>> Select().from_("tbl").select("x", "COUNT(1)").group_by("x").sql()
1212            'SELECT x, COUNT(1) FROM tbl GROUP BY x'
1213
1214        Args:
1215            *expressions: the SQL code strings to parse.
1216                If a `Group` instance is passed, this is used as-is.
1217                If another `Expr` instance is passed, it will be wrapped in a `Group`.
1218                If nothing is passed in then a group by is not applied to the expression
1219            append: if `True`, add to any existing expressions.
1220                Otherwise, this flattens all the `Group` expression into a single expression.
1221            dialect: the dialect used to parse the input expression.
1222            copy: if `False`, modify this expression instance in-place.
1223            opts: other options to use to parse the input expressions.
1224
1225        Returns:
1226            The modified Select expression.
1227        """
1228        if not expressions:
1229            return self if not copy else self.copy()
1230
1231        return _apply_child_list_builder(
1232            *expressions,
1233            instance=self,
1234            arg="group",
1235            append=append,
1236            copy=copy,
1237            prefix="GROUP BY",
1238            into=Group,
1239            dialect=dialect,
1240            **opts,
1241        )
1242
1243    def sort_by(
1244        self,
1245        *expressions: ExpOrStr | None,
1246        append: bool = True,
1247        dialect: DialectType = None,
1248        copy: bool = True,
1249        **opts: Unpack[ParserNoDialectArgs],
1250    ) -> Select:
1251        """
1252        Set the SORT BY expression.
1253
1254        Example:
1255            >>> Select().from_("tbl").select("x").sort_by("x DESC").sql(dialect="hive")
1256            'SELECT x FROM tbl SORT BY x DESC'
1257
1258        Args:
1259            *expressions: the SQL code strings to parse.
1260                If a `Group` instance is passed, this is used as-is.
1261                If another `Expr` instance is passed, it will be wrapped in a `SORT`.
1262            append: if `True`, add to any existing expressions.
1263                Otherwise, this flattens all the `Order` expression into a single expression.
1264            dialect: the dialect used to parse the input expression.
1265            copy: if `False`, modify this expression instance in-place.
1266            opts: other options to use to parse the input expressions.
1267
1268        Returns:
1269            The modified Select expression.
1270        """
1271        return _apply_child_list_builder(
1272            *expressions,
1273            instance=self,
1274            arg="sort",
1275            append=append,
1276            copy=copy,
1277            prefix="SORT BY",
1278            into=Sort,
1279            dialect=dialect,
1280            **opts,
1281        )
1282
1283    def cluster_by(
1284        self,
1285        *expressions: ExpOrStr | None,
1286        append: bool = True,
1287        dialect: DialectType = None,
1288        copy: bool = True,
1289        **opts: Unpack[ParserNoDialectArgs],
1290    ) -> Select:
1291        """
1292        Set the CLUSTER BY expression.
1293
1294        Example:
1295            >>> Select().from_("tbl").select("x").cluster_by("x").sql(dialect="hive")
1296            'SELECT x FROM tbl CLUSTER BY x'
1297
1298        Args:
1299            *expressions: the SQL code strings to parse.
1300                If a `Group` instance is passed, this is used as-is.
1301                If another `Expr` instance is passed, it will be wrapped in a `Cluster`.
1302            append: if `True`, add to any existing expressions.
1303                Otherwise, this flattens all the `Order` expression into a single expression.
1304            dialect: the dialect used to parse the input expression.
1305            copy: if `False`, modify this expression instance in-place.
1306            opts: other options to use to parse the input expressions.
1307
1308        Returns:
1309            The modified Select expression.
1310        """
1311        return _apply_child_list_builder(
1312            *expressions,
1313            instance=self,
1314            arg="cluster",
1315            append=append,
1316            copy=copy,
1317            prefix="CLUSTER BY",
1318            into=Cluster,
1319            dialect=dialect,
1320            **opts,
1321        )
1322
1323    def select(
1324        self,
1325        *expressions: ExpOrStr | None,
1326        append: bool = True,
1327        dialect: DialectType = None,
1328        copy: bool = True,
1329        **opts: Unpack[ParserNoDialectArgs],
1330    ) -> Select:
1331        return _apply_list_builder(
1332            *expressions,
1333            instance=self,
1334            arg="expressions",
1335            append=append,
1336            dialect=dialect,
1337            into=Expr,
1338            copy=copy,
1339            **opts,
1340        )
1341
1342    def lateral(
1343        self,
1344        *expressions: ExpOrStr | None,
1345        append: bool = True,
1346        dialect: DialectType = None,
1347        copy: bool = True,
1348        **opts: Unpack[ParserNoDialectArgs],
1349    ) -> Select:
1350        """
1351        Append to or set the LATERAL expressions.
1352
1353        Example:
1354            >>> Select().select("x").lateral("OUTER explode(y) tbl2 AS z").from_("tbl").sql()
1355            'SELECT x FROM tbl LATERAL VIEW OUTER EXPLODE(y) tbl2 AS z'
1356
1357        Args:
1358            *expressions: the SQL code strings to parse.
1359                If an `Expr` instance is passed, it will be used as-is.
1360            append: if `True`, add to any existing expressions.
1361                Otherwise, this resets the expressions.
1362            dialect: the dialect used to parse the input expressions.
1363            copy: if `False`, modify this expression instance in-place.
1364            opts: other options to use to parse the input expressions.
1365
1366        Returns:
1367            The modified Select expression.
1368        """
1369        return _apply_list_builder(
1370            *expressions,
1371            instance=self,
1372            arg="laterals",
1373            append=append,
1374            into=Lateral,
1375            prefix="LATERAL VIEW",
1376            dialect=dialect,
1377            copy=copy,
1378            **opts,
1379        )
1380
1381    def join(
1382        self,
1383        expression: ExpOrStr,
1384        on: ExpOrStr | list[ExpOrStr] | tuple[ExpOrStr, ...] | None = None,
1385        using: ExpOrStr | list[ExpOrStr] | tuple[ExpOrStr, ...] | None = None,
1386        append: bool = True,
1387        join_type: str | None = None,
1388        join_alias: Identifier | str | None = None,
1389        dialect: DialectType = None,
1390        copy: bool = True,
1391        **opts: Unpack[ParserNoDialectArgs],
1392    ) -> Select:
1393        """
1394        Append to or set the JOIN expressions.
1395
1396        Example:
1397            >>> Select().select("*").from_("tbl").join("tbl2", on="tbl1.y = tbl2.y").sql()
1398            'SELECT * FROM tbl JOIN tbl2 ON tbl1.y = tbl2.y'
1399
1400            >>> Select().select("1").from_("a").join("b", using=["x", "y", "z"]).sql()
1401            'SELECT 1 FROM a JOIN b USING (x, y, z)'
1402
1403            Use `join_type` to change the type of join:
1404
1405            >>> Select().select("*").from_("tbl").join("tbl2", on="tbl1.y = tbl2.y", join_type="left outer").sql()
1406            'SELECT * FROM tbl LEFT OUTER JOIN tbl2 ON tbl1.y = tbl2.y'
1407
1408        Args:
1409            expression: the SQL code string to parse.
1410                If an `Expr` instance is passed, it will be used as-is.
1411            on: optionally specify the join "on" criteria as a SQL string.
1412                If an `Expr` instance is passed, it will be used as-is.
1413            using: optionally specify the join "using" criteria as a SQL string.
1414                If an `Expr` instance is passed, it will be used as-is.
1415            append: if `True`, add to any existing expressions.
1416                Otherwise, this resets the expressions.
1417            join_type: if set, alter the parsed join type.
1418            join_alias: an optional alias for the joined source.
1419            dialect: the dialect used to parse the input expressions.
1420            copy: if `False`, modify this expression instance in-place.
1421            opts: other options to use to parse the input expressions.
1422
1423        Returns:
1424            Select: the modified expression.
1425        """
1426        parse_args: ParserArgs = {"dialect": dialect, **opts}
1427        try:
1428            expression = maybe_parse(expression, into=Join, prefix="JOIN", **parse_args)
1429        except ParseError:
1430            expression = maybe_parse(expression, into=(Join, Expr), **parse_args)
1431
1432        join = expression if isinstance(expression, Join) else Join(this=expression)
1433
1434        if isinstance(join.this, Select):
1435            join.this.replace(join.this.subquery())
1436
1437        if join_type:
1438            new_join: Join = maybe_parse(f"FROM _ {join_type} JOIN _", **parse_args).find(Join)
1439            method = new_join.method
1440            side = new_join.side
1441            kind = new_join.kind
1442
1443            if method:
1444                join.set("method", method)
1445            if side:
1446                join.set("side", side)
1447            if kind:
1448                join.set("kind", kind)
1449
1450        if on:
1451            on_exprs: list[ExpOrStr] = ensure_list(on)
1452            on = and_(*on_exprs, dialect=dialect, copy=copy, **opts)
1453            join.set("on", on)
1454
1455        if using:
1456            using_exprs: list[ExpOrStr] = ensure_list(using)
1457            join = _apply_list_builder(
1458                *using_exprs,
1459                instance=join,
1460                arg="using",
1461                append=append,
1462                copy=copy,
1463                into=Identifier,
1464                **opts,
1465            )
1466
1467        if join_alias:
1468            join.set("this", alias_(join.this, join_alias, table=True))
1469
1470        return _apply_list_builder(
1471            join,
1472            instance=self,
1473            arg="joins",
1474            append=append,
1475            copy=copy,
1476            **opts,
1477        )
1478
1479    def having(
1480        self,
1481        *expressions: ExpOrStr | None,
1482        append: bool = True,
1483        dialect: DialectType = None,
1484        copy: bool = True,
1485        **opts: Unpack[ParserNoDialectArgs],
1486    ) -> Select:
1487        """
1488        Append to or set the HAVING expressions.
1489
1490        Example:
1491            >>> Select().select("x", "COUNT(y)").from_("tbl").group_by("x").having("COUNT(y) > 3").sql()
1492            'SELECT x, COUNT(y) FROM tbl GROUP BY x HAVING COUNT(y) > 3'
1493
1494        Args:
1495            *expressions: the SQL code strings to parse.
1496                If an `Expr` instance is passed, it will be used as-is.
1497                Multiple expressions are combined with an AND operator.
1498            append: if `True`, AND the new expressions to any existing expression.
1499                Otherwise, this resets the expression.
1500            dialect: the dialect used to parse the input expressions.
1501            copy: if `False`, modify this expression instance in-place.
1502            opts: other options to use to parse the input expressions.
1503
1504        Returns:
1505            The modified Select expression.
1506        """
1507        return _apply_conjunction_builder(
1508            *expressions,
1509            instance=self,
1510            arg="having",
1511            append=append,
1512            into=Having,
1513            dialect=dialect,
1514            copy=copy,
1515            **opts,
1516        )
1517
1518    def window(
1519        self,
1520        *expressions: ExpOrStr | None,
1521        append: bool = True,
1522        dialect: DialectType = None,
1523        copy: bool = True,
1524        **opts: Unpack[ParserNoDialectArgs],
1525    ) -> Select:
1526        return _apply_list_builder(
1527            *expressions,
1528            instance=self,
1529            arg="windows",
1530            append=append,
1531            into=Window,
1532            dialect=dialect,
1533            copy=copy,
1534            **opts,
1535        )
1536
1537    def qualify(
1538        self,
1539        *expressions: ExpOrStr | None,
1540        append: bool = True,
1541        dialect: DialectType = None,
1542        copy: bool = True,
1543        **opts: Unpack[ParserNoDialectArgs],
1544    ) -> Select:
1545        return _apply_conjunction_builder(
1546            *expressions,
1547            instance=self,
1548            arg="qualify",
1549            append=append,
1550            into=Qualify,
1551            dialect=dialect,
1552            copy=copy,
1553            **opts,
1554        )
1555
1556    def distinct(self, *ons: ExpOrStr | None, distinct: bool = True, copy: bool = True) -> Select:
1557        """
1558        Set the OFFSET expression.
1559
1560        Example:
1561            >>> Select().from_("tbl").select("x").distinct().sql()
1562            'SELECT DISTINCT x FROM tbl'
1563
1564        Args:
1565            ons: the expressions to distinct on
1566            distinct: whether the Select should be distinct
1567            copy: if `False`, modify this expression instance in-place.
1568
1569        Returns:
1570            Select: the modified expression.
1571        """
1572        instance = maybe_copy(self, copy)
1573        on = Tuple(expressions=[maybe_parse(on, copy=copy) for on in ons if on]) if ons else None
1574        instance.set("distinct", Distinct(on=on) if distinct else None)
1575        return instance
1576
1577    def ctas(
1578        self,
1579        table: ExpOrStr,
1580        properties: dict | None = None,
1581        dialect: DialectType = None,
1582        copy: bool = True,
1583        **opts: Unpack[ParserNoDialectArgs],
1584    ) -> Create:
1585        """
1586        Convert this expression to a CREATE TABLE AS statement.
1587
1588        Example:
1589            >>> Select().select("*").from_("tbl").ctas("x").sql()
1590            'CREATE TABLE x AS SELECT * FROM tbl'
1591
1592        Args:
1593            table: the SQL code string to parse as the table name.
1594                If another `Expr` instance is passed, it will be used as-is.
1595            properties: an optional mapping of table properties
1596            dialect: the dialect used to parse the input table.
1597            copy: if `False`, modify this expression instance in-place.
1598            opts: other options to use to parse the input table.
1599
1600        Returns:
1601            The new Create expression.
1602        """
1603        instance = maybe_copy(self, copy)
1604        table_expression = maybe_parse(table, into=Table, dialect=dialect, **opts)
1605
1606        properties_expression = None
1607        if properties:
1608            from sqlglot.expressions.properties import Properties as _Properties
1609
1610            properties_expression = _Properties.from_dict(properties)
1611
1612        from sqlglot.expressions.ddl import Create as _Create
1613
1614        return _Create(
1615            this=table_expression,
1616            kind="TABLE",
1617            expression=instance,
1618            properties=properties_expression,
1619        )
1620
1621    def lock(self, update: bool = True, copy: bool = True) -> Select:
1622        """
1623        Set the locking read mode for this expression.
1624
1625        Examples:
1626            >>> Select().select("x").from_("tbl").where("x = 'a'").lock().sql("mysql")
1627            "SELECT x FROM tbl WHERE x = 'a' FOR UPDATE"
1628
1629            >>> Select().select("x").from_("tbl").where("x = 'a'").lock(update=False).sql("mysql")
1630            "SELECT x FROM tbl WHERE x = 'a' FOR SHARE"
1631
1632        Args:
1633            update: if `True`, the locking type will be `FOR UPDATE`, else it will be `FOR SHARE`.
1634            copy: if `False`, modify this expression instance in-place.
1635
1636        Returns:
1637            The modified expression.
1638        """
1639        inst = maybe_copy(self, copy)
1640        inst.set("locks", [Lock(update=update)])
1641
1642        return inst
1643
1644    def hint(self, *hints: ExpOrStr, dialect: DialectType = None, copy: bool = True) -> Select:
1645        """
1646        Set hints for this expression.
1647
1648        Examples:
1649            >>> Select().select("x").from_("tbl").hint("BROADCAST(y)").sql(dialect="spark")
1650            'SELECT /*+ BROADCAST(y) */ x FROM tbl'
1651
1652        Args:
1653            hints: The SQL code strings to parse as the hints.
1654                If an `Expr` instance is passed, it will be used as-is.
1655            dialect: The dialect used to parse the hints.
1656            copy: If `False`, modify this expression instance in-place.
1657
1658        Returns:
1659            The modified expression.
1660        """
1661        inst = maybe_copy(self, copy)
1662        inst.set(
1663            "hint", Hint(expressions=[maybe_parse(h, copy=copy, dialect=dialect) for h in hints])
1664        )
1665
1666        return inst
1667
1668    @property
1669    def named_selects(self) -> list[str]:
1670        selects = []
1671
1672        for e in self.expressions:
1673            if e.alias_or_name:
1674                selects.append(e.output_name)
1675            elif isinstance(e, Aliases):
1676                selects.extend([a.name for a in e.aliases])
1677        return selects
1678
1679    @property
1680    def is_star(self) -> bool:
1681        return any(expression.is_star for expression in self.expressions)
1682
1683    @property
1684    def selects(self) -> list[Expr]:
1685        return self.expressions
1686
1687
1688class Subquery(Expression, DerivedTable, Query):
1689    is_subquery: t.ClassVar[bool] = True
1690    arg_types = {
1691        "this": True,
1692        "alias": False,
1693        "with_": False,
1694        **QUERY_MODIFIERS,
1695    }
1696
1697    def unnest(self) -> Expr:
1698        """Returns the first non subquery."""
1699        expression: Expr = self
1700        while isinstance(expression, Subquery):
1701            expression = expression.this
1702        return expression
1703
1704    def unwrap(self) -> Subquery:
1705        expression = self
1706        while expression.same_parent and expression.is_wrapper:
1707            expression = t.cast(Subquery, expression.parent)
1708        return expression
1709
1710    def select(
1711        self,
1712        *expressions: ExpOrStr | None,
1713        append: bool = True,
1714        dialect: DialectType = None,
1715        copy: bool = True,
1716        **opts: Unpack[ParserNoDialectArgs],
1717    ) -> Subquery:
1718        this = maybe_copy(self, copy)
1719        inner = this.unnest()
1720        if hasattr(inner, "select"):
1721            inner.select(*expressions, append=append, dialect=dialect, copy=False, **opts)
1722        return this
1723
1724    @property
1725    def is_wrapper(self) -> bool:
1726        """
1727        Whether this Subquery acts as a simple wrapper around another expression.
1728
1729        SELECT * FROM (((SELECT * FROM t)))
1730                      ^
1731                      This corresponds to a "wrapper" Subquery node
1732        """
1733        return all(v is None for k, v in self.args.items() if k != "this")
1734
1735    @property
1736    def is_star(self) -> bool:
1737        return _is_star(self)
1738
1739    @property
1740    def output_name(self) -> str:
1741        return self.alias
1742
1743
1744class TableSample(Expression):
1745    arg_types = {
1746        "expressions": False,
1747        "method": False,
1748        "bucket_numerator": False,
1749        "bucket_denominator": False,
1750        "bucket_field": False,
1751        "percent": False,
1752        "rows": False,
1753        "size": False,
1754        "seed": False,
1755    }
1756
1757
1758class Tag(Expression):
1759    """Tags are used for generating arbitrary sql like SELECT <span>x</span>."""
1760
1761    arg_types = {
1762        "this": False,
1763        "prefix": False,
1764        "postfix": False,
1765    }
1766
1767
1768class Pivot(Expression):
1769    arg_types = {
1770        "this": False,
1771        "alias": False,
1772        "expressions": False,
1773        "fields": False,
1774        "unpivot": False,
1775        "using": False,
1776        "group": False,
1777        "columns": False,
1778        "include_nulls": False,
1779        "default_on_null": False,
1780        "into": False,
1781        "with_": False,
1782        "identify_pivot_strings": False,
1783        "prefixed_pivot_columns": False,
1784        "pivot_column_naming": False,
1785        "value_columns_first": False,
1786    }
1787
1788    @property
1789    def unpivot(self) -> bool:
1790        return bool(self.args.get("unpivot"))
1791
1792    @property
1793    def fields(self) -> list[Expr]:
1794        return self.args.get("fields", [])
1795
1796    def output_columns(self, pre_pivot_columns: t.Iterable[str]) -> dict[str, str]:
1797        """
1798        Returns an ordered map of post-rename output column name -> pre-rename
1799        source-side name, in the order the (UN)PIVOT produces them.
1800
1801        For callers that just want the names, iterate the dict (or call .keys()):
1802            >>> from sqlglot import parse_one, exp
1803            >>> piv = parse_one("SELECT * FROM t UNPIVOT(val FOR name IN (a, b))").find(exp.Pivot)
1804            >>> list(piv.output_columns(["a", "b", "c"]))
1805            ['c', 'name', 'val']
1806
1807        AST shape:
1808            PIVOT(SUM(val) FOR name IN ('a', 'b')):
1809                expressions: aggregate(s), e.g. [Sum(this=Column(val))]
1810                fields:      [In(this=Column(name), expressions=[Literal('a'), Literal('b')])]
1811                columns:     optional explicit output identifiers (e.g. set by Snowflake)
1812
1813            UNPIVOT(val FOR name IN (a, b)):
1814                expressions: value Identifier(s), or Tuple(Identifiers) for multi-value
1815                fields:      [In(this=Identifier(name), expressions=[Column(a), Column(b)])]
1816                             For literal-aliased entries (`a AS 'x'`) the IN expressions
1817                             are wrapped in PivotAlias(this=Column, alias=Literal).
1818
1819        Args:
1820            pre_pivot_columns: Columns visible to the operator before it runs
1821                (e.g. the source table or subquery's projections).
1822        """
1823        if self.unpivot:
1824            excluded: set[str] = set()
1825            name_columns: list[Identifier] = []
1826            for field in self.fields:
1827                if not isinstance(field, In):
1828                    continue
1829                if isinstance(field.this, Identifier):
1830                    name_columns.append(field.this)
1831                for e in field.expressions:
1832                    excluded.update(c.output_name for c in e.find_all(Column))
1833            value_columns = [
1834                ident
1835                for e in self.expressions
1836                for ident in (e.expressions if isinstance(e, Tuple) else [e])
1837                if isinstance(ident, Identifier)
1838            ]
1839            # T-SQL emits the value column(s) ahead of the name column, everyone else emits them after it
1840            ordered = (
1841                value_columns + name_columns
1842                if self.args.get("value_columns_first")
1843                else name_columns + value_columns
1844            )
1845            outputs = [i.name for i in ordered]
1846        else:
1847            excluded = {c.output_name for c in self.find_all(Column)}
1848            outputs = [c.output_name for c in self.args.get("columns") or []]
1849            if not outputs:
1850                outputs = [c.alias_or_name for c in self.expressions]
1851
1852        if not excluded or not outputs:
1853            return {}
1854
1855        pre_rename = [c for c in pre_pivot_columns if c not in excluded] + outputs
1856
1857        alias = self.args.get("alias")
1858        renames = alias.args.get("columns") if alias else None
1859
1860        # `PIVOT(...) AS alias(c1, c2, ...)` renames the operator's output columns
1861        # positionally from the front (DuckDB, Snowflake): the user's names cover
1862        # the leading N output columns, remaining columns keep their auto names.
1863        if renames:
1864            rename_names = [r.name for r in renames]
1865            post_rename = rename_names + pre_rename[len(rename_names) :]
1866        else:
1867            post_rename = pre_rename
1868
1869        return dict(zip(post_rename, pre_rename))
1870
1871
1872class UnpivotColumns(Expression):
1873    arg_types = {"this": True, "expressions": True}
1874
1875
1876class Window(Expression, Condition):
1877    arg_types = {
1878        "this": True,
1879        "partition_by": False,
1880        "order": False,
1881        "spec": False,
1882        "alias": False,
1883        "over": False,
1884        "first": False,
1885    }
1886
1887
1888class WindowSpec(Expression):
1889    arg_types = {
1890        "kind": False,
1891        "start": False,
1892        "start_side": False,
1893        "end": False,
1894        "end_side": False,
1895        "exclude": False,
1896    }
1897
1898
1899class PreWhere(Expression):
1900    pass
1901
1902
1903class Where(Expression):
1904    pass
1905
1906
1907class Analyze(Expression):
1908    arg_types = {
1909        "kind": False,
1910        "tables": False,
1911        "options": False,
1912        "mode": False,
1913        "partition": False,
1914        "expression": False,
1915        "properties": False,
1916    }
1917
1918
1919class AnalyzeStatistics(Expression):
1920    arg_types = {
1921        "kind": True,
1922        "option": False,
1923        "this": False,
1924        "expressions": False,
1925    }
1926
1927
1928class AnalyzeHistogram(Expression):
1929    arg_types = {
1930        "this": True,
1931        "expressions": True,
1932        "expression": False,
1933        "update_options": False,
1934    }
1935
1936
1937class AnalyzeSample(Expression):
1938    arg_types = {"kind": True, "sample": True}
1939
1940
1941class AnalyzeListChainedRows(Expression):
1942    arg_types = {"expression": False}
1943
1944
1945class AnalyzeDelete(Expression):
1946    arg_types = {"kind": False}
1947
1948
1949class AnalyzeWith(Expression):
1950    arg_types = {"expressions": True}
1951
1952
1953class AnalyzeValidate(Expression):
1954    arg_types = {
1955        "kind": True,
1956        "this": False,
1957        "expression": False,
1958    }
1959
1960
1961class AnalyzeColumns(Expression):
1962    pass
1963
1964
1965class UsingData(Expression):
1966    pass
1967
1968
1969class AddPartition(Expression):
1970    arg_types = {"this": True, "exists": False, "location": False}
1971
1972
1973class AttachOption(Expression):
1974    arg_types = {"this": True, "expression": False}
1975
1976
1977class DropPartition(Expression):
1978    arg_types = {"expressions": True, "exists": False}
1979
1980
1981class ReplacePartition(Expression):
1982    arg_types = {"expression": True, "source": True}
1983
1984
1985class TranslateCharacters(Expression):
1986    arg_types = {"this": True, "expression": True, "with_error": False}
1987
1988
1989class OverflowTruncateBehavior(Expression):
1990    arg_types = {"this": False, "with_count": True}
1991
1992
1993class JSON(Expression):
1994    arg_types = {"this": False, "with_": False, "unique": False}
1995
1996
1997class JSONPath(Expression):
1998    arg_types = {"expressions": True}
1999
2000    @property
2001    def output_name(self) -> str:
2002        last_segment = self.expressions[-1].this
2003        return last_segment if isinstance(last_segment, str) else ""
2004
2005
2006class JSONPathPart(Expression):
2007    arg_types = {}
2008
2009
2010class JSONPathFilter(JSONPathPart):
2011    arg_types = {"this": True}
2012
2013
2014class JSONPathKey(JSONPathPart):
2015    arg_types = {"this": True, "quoted": False}
2016
2017
2018class JSONPathRecursive(JSONPathPart):
2019    arg_types = {"this": False}
2020
2021
2022class JSONPathRoot(JSONPathPart):
2023    pass
2024
2025
2026class JSONPathScript(JSONPathPart):
2027    arg_types = {"this": True}
2028
2029
2030class JSONPathSlice(JSONPathPart):
2031    arg_types = {"start": False, "end": False, "step": False}
2032
2033
2034class JSONPathSelector(JSONPathPart):
2035    arg_types = {"this": True}
2036
2037
2038class JSONPathSubscript(JSONPathPart):
2039    arg_types = {"this": True}
2040
2041
2042class JSONPathUnion(JSONPathPart):
2043    arg_types = {"expressions": True}
2044
2045
2046class JSONPathWildcard(JSONPathPart):
2047    pass
2048
2049
2050class FormatJson(Expression):
2051    pass
2052
2053
2054class JSONKeyValue(Expression):
2055    arg_types = {"this": True, "expression": True}
2056
2057
2058class JSONColumnDef(Expression):
2059    arg_types = {
2060        "this": False,
2061        "kind": False,
2062        "path": False,
2063        "nested_schema": False,
2064        "ordinality": False,
2065        "format_json": False,
2066    }
2067
2068
2069class JSONSchema(Expression):
2070    arg_types = {"expressions": True}
2071
2072
2073class JSONValue(Expression):
2074    arg_types = {
2075        "this": True,
2076        "path": True,
2077        "returning": False,
2078        "on_condition": False,
2079    }
2080
2081
2082class JSONValueArray(Expression, Func):
2083    arg_types = {"this": True, "expression": False}
2084
2085
2086class OpenJSONColumnDef(Expression):
2087    arg_types = {"this": True, "kind": True, "path": False, "as_json": False}
2088
2089
2090class JSONExtractQuote(Expression):
2091    arg_types = {
2092        "option": True,
2093        "scalar": False,
2094    }
2095
2096
2097class ScopeResolution(Expression):
2098    arg_types = {"this": False, "expression": True}
2099
2100
2101class Stream(Expression):
2102    pass
2103
2104
2105class ModelAttribute(Expression):
2106    arg_types = {"this": True, "expression": True}
2107
2108
2109class XMLNamespace(Expression):
2110    pass
2111
2112
2113class XMLKeyValueOption(Expression):
2114    arg_types = {"this": True, "expression": False}
2115
2116
2117class Semicolon(Expression):
2118    arg_types = {}
2119
2120
2121class TableColumn(Expression):
2122    @property
2123    def output_name(self) -> str:
2124        return self.name
2125
2126
2127class Variadic(Expression):
2128    pass
2129
2130
2131class StoredProcedure(Expression):
2132    arg_types = {"this": True, "expressions": False, "wrapped": False}
2133
2134
2135class Block(Expression):
2136    arg_types = {"expressions": True, "begin": False}
2137
2138
2139class IfBlock(Expression):
2140    arg_types = {"this": True, "true": True, "false": False}
2141
2142
2143class CaseStatement(Expression):
2144    arg_types = {"this": False, "ifs": True, "default": False}
2145
2146
2147class WhileBlock(Expression):
2148    arg_types = {"this": True, "body": True, "label": False}
2149
2150
2151class LoopBlock(Expression):
2152    arg_types = {"body": True, "label": False}
2153
2154
2155class RepeatBlock(Expression):
2156    arg_types = {"body": True, "until": True, "label": False}
2157
2158
2159class Leave(Expression):
2160    pass
2161
2162
2163class Iterate(Expression):
2164    pass
2165
2166
2167class EndStatement(Expression):
2168    arg_types = {}
2169
2170
2171# https://trino.io/docs/current/udf.html
2172class FunctionSpecification(Expression):
2173    arg_types = {
2174        "this": True,
2175        "characteristics": False,
2176        "properties": False,
2177        "expression": True,
2178    }
2179
2180
2181UNWRAPPED_QUERIES = (Select, SetOperation)
2182
2183
2184def union(
2185    *expressions: ExpOrStr,
2186    distinct: bool = True,
2187    dialect: DialectType = None,
2188    copy: bool = True,
2189    **opts: Unpack[ParserNoDialectArgs],
2190) -> Union:
2191    """
2192    Initializes a syntax tree for the `UNION` operation.
2193
2194    Example:
2195        >>> union("SELECT * FROM foo", "SELECT * FROM bla").sql()
2196        'SELECT * FROM foo UNION SELECT * FROM bla'
2197
2198    Args:
2199        expressions: the SQL code strings, corresponding to the `UNION`'s operands.
2200            If `Expr` instances are passed, they will be used as-is.
2201        distinct: set the DISTINCT flag if and only if this is true.
2202        dialect: the dialect used to parse the input expression.
2203        copy: whether to copy the expression.
2204        opts: other options to use to parse the input expressions.
2205
2206    Returns:
2207        The new Union instance.
2208    """
2209    assert len(expressions) >= 2, "At least two expressions are required by `union`."
2210    return _apply_set_operation(
2211        *expressions, set_operation=Union, distinct=distinct, dialect=dialect, copy=copy, **opts
2212    )
2213
2214
2215def intersect(
2216    *expressions: ExpOrStr,
2217    distinct: bool = True,
2218    dialect: DialectType = None,
2219    copy: bool = True,
2220    **opts: Unpack[ParserNoDialectArgs],
2221) -> Intersect:
2222    """
2223    Initializes a syntax tree for the `INTERSECT` operation.
2224
2225    Example:
2226        >>> intersect("SELECT * FROM foo", "SELECT * FROM bla").sql()
2227        'SELECT * FROM foo INTERSECT SELECT * FROM bla'
2228
2229    Args:
2230        expressions: the SQL code strings, corresponding to the `INTERSECT`'s operands.
2231            If `Expr` instances are passed, they will be used as-is.
2232        distinct: set the DISTINCT flag if and only if this is true.
2233        dialect: the dialect used to parse the input expression.
2234        copy: whether to copy the expression.
2235        opts: other options to use to parse the input expressions.
2236
2237    Returns:
2238        The new Intersect instance.
2239    """
2240    assert len(expressions) >= 2, "At least two expressions are required by `intersect`."
2241    return _apply_set_operation(
2242        *expressions, set_operation=Intersect, distinct=distinct, dialect=dialect, copy=copy, **opts
2243    )
2244
2245
2246def except_(
2247    *expressions: ExpOrStr,
2248    distinct: bool = True,
2249    dialect: DialectType = None,
2250    copy: bool = True,
2251    **opts: Unpack[ParserNoDialectArgs],
2252) -> Except:
2253    """
2254    Initializes a syntax tree for the `EXCEPT` operation.
2255
2256    Example:
2257        >>> except_("SELECT * FROM foo", "SELECT * FROM bla").sql()
2258        'SELECT * FROM foo EXCEPT SELECT * FROM bla'
2259
2260    Args:
2261        expressions: the SQL code strings, corresponding to the `EXCEPT`'s operands.
2262            If `Expr` instances are passed, they will be used as-is.
2263        distinct: set the DISTINCT flag if and only if this is true.
2264        dialect: the dialect used to parse the input expression.
2265        copy: whether to copy the expression.
2266        opts: other options to use to parse the input expressions.
2267
2268    Returns:
2269        The new Except instance.
2270    """
2271    assert len(expressions) >= 2, "At least two expressions are required by `except_`."
2272    return _apply_set_operation(
2273        *expressions, set_operation=Except, distinct=distinct, dialect=dialect, copy=copy, **opts
2274    )