1"""
   2## Expressions
   3
   4Every AST node in SQLGlot is represented by a subclass of `Expression`.
   5
   6This module contains the implementation of all supported `Expression` types. Additionally,
   7it exposes a number of helper functions, which are mainly used to programmatically build
   8SQL expressions, such as `sqlglot.expressions.select`.
   9
  10----
  11"""
  12
  13from __future__ import annotations
  14
  15import datetime
  16import math
  17import numbers
  18import re
  19import typing as t
  20from collections import deque
  21from copy import deepcopy
  22from enum import auto
  23from functools import reduce
  24
  25from sqlglot._typing import E
  26from sqlglot.errors import ErrorLevel, ParseError
  27from sqlglot.helper import (
  28    AutoName,
  29    camel_to_snake_case,
  30    ensure_collection,
  31    ensure_list,
  32    seq_get,
  33    subclasses,
  34)
  35from sqlglot.tokens import Token
  36
  37if t.TYPE_CHECKING:
  38    from sqlglot.dialects.dialect import DialectType
  39
  40
  41class _Expression(type):
  42    def __new__(cls, clsname, bases, attrs):
  43        klass = super().__new__(cls, clsname, bases, attrs)
  44
  45        # When an Expression class is created, its key is automatically set to be
  46        # the lowercase version of the class' name.
  47        klass.key = clsname.lower()
  48
  49        # This is so that docstrings are not inherited in pdoc
  50        klass.__doc__ = klass.__doc__ or ""
  51
  52        return klass
  53
  54
  55SQLGLOT_META = "sqlglot.meta"
  56
  57
  58class Expression(metaclass=_Expression):
  59    """
  60    The base class for all expressions in a syntax tree. Each Expression encapsulates any necessary
  61    context, such as its child expressions, their names (arg keys), and whether a given child expression
  62    is optional or not.
  63
  64    Attributes:
  65        key: a unique key for each class in the Expression hierarchy. This is useful for hashing
  66            and representing expressions as strings.
  67        arg_types: determines what arguments (child nodes) are supported by an expression. It
  68            maps arg keys to booleans that indicate whether the corresponding args are optional.
  69        parent: a reference to the parent expression (or None, in case of root expressions).
  70        arg_key: the arg key an expression is associated with, i.e. the name its parent expression
  71            uses to refer to it.
  72        comments: a list of comments that are associated with a given expression. This is used in
  73            order to preserve comments when transpiling SQL code.
  74        type: the `sqlglot.expressions.DataType` type of an expression. This is inferred by the
  75            optimizer, in order to enable some transformations that require type information.
  76        meta: a dictionary that can be used to store useful metadata for a given expression.
  77
  78    Example:
  79        >>> class Foo(Expression):
  80        ...     arg_types = {"this": True, "expression": False}
  81
  82        The above definition informs us that Foo is an Expression that requires an argument called
  83        "this" and may also optionally receive an argument called "expression".
  84
  85    Args:
  86        args: a mapping used for retrieving the arguments of an expression, given their arg keys.
  87    """
  88
  89    key = "expression"
  90    arg_types = {"this": True}
  91    __slots__ = ("args", "parent", "arg_key", "comments", "_type", "_meta", "_hash")
  92
  93    def __init__(self, **args: t.Any):
  94        self.args: t.Dict[str, t.Any] = args
  95        self.parent: t.Optional[Expression] = None
  96        self.arg_key: t.Optional[str] = None
  97        self.comments: t.Optional[t.List[str]] = None
  98        self._type: t.Optional[DataType] = None
  99        self._meta: t.Optional[t.Dict[str, t.Any]] = None
 100        self._hash: t.Optional[int] = None
 101
 102        for arg_key, value in self.args.items():
 103            self._set_parent(arg_key, value)
 104
 105    def __eq__(self, other) -> bool:
 106        return type(self) is type(other) and hash(self) == hash(other)
 107
 108    @property
 109    def hashable_args(self) -> t.Any:
 110        return frozenset(
 111            (k, tuple(_norm_arg(a) for a in v) if type(v) is list else _norm_arg(v))
 112            for k, v in self.args.items()
 113            if not (v is None or v is False or (type(v) is list and not v))
 114        )
 115
 116    def __hash__(self) -> int:
 117        if self._hash is not None:
 118            return self._hash
 119
 120        return hash((self.__class__, self.hashable_args))
 121
 122    @property
 123    def this(self) -> t.Any:
 124        """
 125        Retrieves the argument with key "this".
 126        """
 127        return self.args.get("this")
 128
 129    @property
 130    def expression(self) -> t.Any:
 131        """
 132        Retrieves the argument with key "expression".
 133        """
 134        return self.args.get("expression")
 135
 136    @property
 137    def expressions(self) -> t.List[t.Any]:
 138        """
 139        Retrieves the argument with key "expressions".
 140        """
 141        return self.args.get("expressions") or []
 142
 143    def text(self, key) -> str:
 144        """
 145        Returns a textual representation of the argument corresponding to "key". This can only be used
 146        for args that are strings or leaf Expression instances, such as identifiers and literals.
 147        """
 148        field = self.args.get(key)
 149        if isinstance(field, str):
 150            return field
 151        if isinstance(field, (Identifier, Literal, Var)):
 152            return field.this
 153        if isinstance(field, (Star, Null)):
 154            return field.name
 155        return ""
 156
 157    @property
 158    def is_string(self) -> bool:
 159        """
 160        Checks whether a Literal expression is a string.
 161        """
 162        return isinstance(self, Literal) and self.args["is_string"]
 163
 164    @property
 165    def is_number(self) -> bool:
 166        """
 167        Checks whether a Literal expression is a number.
 168        """
 169        return isinstance(self, Literal) and not self.args["is_string"]
 170
 171    @property
 172    def is_int(self) -> bool:
 173        """
 174        Checks whether a Literal expression is an integer.
 175        """
 176        if self.is_number:
 177            try:
 178                int(self.name)
 179                return True
 180            except ValueError:
 181                pass
 182        return False
 183
 184    @property
 185    def is_star(self) -> bool:
 186        """Checks whether an expression is a star."""
 187        return isinstance(self, Star) or (isinstance(self, Column) and isinstance(self.this, Star))
 188
 189    @property
 190    def alias(self) -> str:
 191        """
 192        Returns the alias of the expression, or an empty string if it's not aliased.
 193        """
 194        if isinstance(self.args.get("alias"), TableAlias):
 195            return self.args["alias"].name
 196        return self.text("alias")
 197
 198    @property
 199    def alias_column_names(self) -> t.List[str]:
 200        table_alias = self.args.get("alias")
 201        if not table_alias:
 202            return []
 203        return [c.name for c in table_alias.args.get("columns") or []]
 204
 205    @property
 206    def name(self) -> str:
 207        return self.text("this")
 208
 209    @property
 210    def alias_or_name(self) -> str:
 211        return self.alias or self.name
 212
 213    @property
 214    def output_name(self) -> str:
 215        """
 216        Name of the output column if this expression is a selection.
 217
 218        If the Expression has no output name, an empty string is returned.
 219
 220        Example:
 221            >>> from sqlglot import parse_one
 222            >>> parse_one("SELECT a").expressions[0].output_name
 223            'a'
 224            >>> parse_one("SELECT b AS c").expressions[0].output_name
 225            'c'
 226            >>> parse_one("SELECT 1 + 2").expressions[0].output_name
 227            ''
 228        """
 229        return ""
 230
 231    @property
 232    def type(self) -> t.Optional[DataType]:
 233        return self._type
 234
 235    @type.setter
 236    def type(self, dtype: t.Optional[DataType | DataType.Type | str]) -> None:
 237        if dtype and not isinstance(dtype, DataType):
 238            dtype = DataType.build(dtype)
 239        self._type = dtype  # type: ignore
 240
 241    def is_type(self, *dtypes) -> bool:
 242        return self.type is not None and self.type.is_type(*dtypes)
 243
 244    @property
 245    def meta(self) -> t.Dict[str, t.Any]:
 246        if self._meta is None:
 247            self._meta = {}
 248        return self._meta
 249
 250    def __deepcopy__(self, memo):
 251        copy = self.__class__(**deepcopy(self.args))
 252        if self.comments is not None:
 253            copy.comments = deepcopy(self.comments)
 254
 255        if self._type is not None:
 256            copy._type = self._type.copy()
 257
 258        if self._meta is not None:
 259            copy._meta = deepcopy(self._meta)
 260
 261        return copy
 262
 263    def copy(self):
 264        """
 265        Returns a deep copy of the expression.
 266        """
 267        new = deepcopy(self)
 268        new.parent = self.parent
 269        return new
 270
 271    def add_comments(self, comments: t.Optional[t.List[str]]) -> None:
 272        if self.comments is None:
 273            self.comments = []
 274        if comments:
 275            for comment in comments:
 276                _, *meta = comment.split(SQLGLOT_META)
 277                if meta:
 278                    for kv in "".join(meta).split(","):
 279                        k, *v = kv.split("=")
 280                        value = v[0].strip() if v else True
 281                        self.meta[k.strip()] = value
 282                self.comments.append(comment)
 283
 284    def append(self, arg_key: str, value: t.Any) -> None:
 285        """
 286        Appends value to arg_key if it's a list or sets it as a new list.
 287
 288        Args:
 289            arg_key (str): name of the list expression arg
 290            value (Any): value to append to the list
 291        """
 292        if not isinstance(self.args.get(arg_key), list):
 293            self.args[arg_key] = []
 294        self.args[arg_key].append(value)
 295        self._set_parent(arg_key, value)
 296
 297    def set(self, arg_key: str, value: t.Any) -> None:
 298        """
 299        Sets arg_key to value.
 300
 301        Args:
 302            arg_key: name of the expression arg.
 303            value: value to set the arg to.
 304        """
 305        if value is None:
 306            self.args.pop(arg_key, None)
 307            return
 308
 309        self.args[arg_key] = value
 310        self._set_parent(arg_key, value)
 311
 312    def _set_parent(self, arg_key: str, value: t.Any) -> None:
 313        if hasattr(value, "parent"):
 314            value.parent = self
 315            value.arg_key = arg_key
 316        elif type(value) is list:
 317            for v in value:
 318                if hasattr(v, "parent"):
 319                    v.parent = self
 320                    v.arg_key = arg_key
 321
 322    @property
 323    def depth(self) -> int:
 324        """
 325        Returns the depth of this tree.
 326        """
 327        if self.parent:
 328            return self.parent.depth + 1
 329        return 0
 330
 331    def iter_expressions(self) -> t.Iterator[t.Tuple[str, Expression]]:
 332        """Yields the key and expression for all arguments, exploding list args."""
 333        for k, vs in self.args.items():
 334            if type(vs) is list:
 335                for v in vs:
 336                    if hasattr(v, "parent"):
 337                        yield k, v
 338            else:
 339                if hasattr(vs, "parent"):
 340                    yield k, vs
 341
 342    def find(self, *expression_types: t.Type[E], bfs: bool = True) -> t.Optional[E]:
 343        """
 344        Returns the first node in this tree which matches at least one of
 345        the specified types.
 346
 347        Args:
 348            expression_types: the expression type(s) to match.
 349            bfs: whether to search the AST using the BFS algorithm (DFS is used if false).
 350
 351        Returns:
 352            The node which matches the criteria or None if no such node was found.
 353        """
 354        return next(self.find_all(*expression_types, bfs=bfs), None)
 355
 356    def find_all(self, *expression_types: t.Type[E], bfs: bool = True) -> t.Iterator[E]:
 357        """
 358        Returns a generator object which visits all nodes in this tree and only
 359        yields those that match at least one of the specified expression types.
 360
 361        Args:
 362            expression_types: the expression type(s) to match.
 363            bfs: whether to search the AST using the BFS algorithm (DFS is used if false).
 364
 365        Returns:
 366            The generator object.
 367        """
 368        for expression, *_ in self.walk(bfs=bfs):
 369            if isinstance(expression, expression_types):
 370                yield expression
 371
 372    def find_ancestor(self, *expression_types: t.Type[E]) -> t.Optional[E]:
 373        """
 374        Returns a nearest parent matching expression_types.
 375
 376        Args:
 377            expression_types: the expression type(s) to match.
 378
 379        Returns:
 380            The parent node.
 381        """
 382        ancestor = self.parent
 383        while ancestor and not isinstance(ancestor, expression_types):
 384            ancestor = ancestor.parent
 385        return t.cast(E, ancestor)
 386
 387    @property
 388    def parent_select(self) -> t.Optional[Select]:
 389        """
 390        Returns the parent select statement.
 391        """
 392        return self.find_ancestor(Select)
 393
 394    @property
 395    def same_parent(self) -> bool:
 396        """Returns if the parent is the same class as itself."""
 397        return type(self.parent) is self.__class__
 398
 399    def root(self) -> Expression:
 400        """
 401        Returns the root expression of this tree.
 402        """
 403        expression = self
 404        while expression.parent:
 405            expression = expression.parent
 406        return expression
 407
 408    def walk(self, bfs=True, prune=None):
 409        """
 410        Returns a generator object which visits all nodes in this tree.
 411
 412        Args:
 413            bfs (bool): if set to True the BFS traversal order will be applied,
 414                otherwise the DFS traversal will be used instead.
 415            prune ((node, parent, arg_key) -> bool): callable that returns True if
 416                the generator should stop traversing this branch of the tree.
 417
 418        Returns:
 419            the generator object.
 420        """
 421        if bfs:
 422            yield from self.bfs(prune=prune)
 423        else:
 424            yield from self.dfs(prune=prune)
 425
 426    def dfs(self, parent=None, key=None, prune=None):
 427        """
 428        Returns a generator object which visits all nodes in this tree in
 429        the DFS (Depth-first) order.
 430
 431        Returns:
 432            The generator object.
 433        """
 434        parent = parent or self.parent
 435        yield self, parent, key
 436        if prune and prune(self, parent, key):
 437            return
 438
 439        for k, v in self.iter_expressions():
 440            yield from v.dfs(self, k, prune)
 441
 442    def bfs(self, prune=None):
 443        """
 444        Returns a generator object which visits all nodes in this tree in
 445        the BFS (Breadth-first) order.
 446
 447        Returns:
 448            The generator object.
 449        """
 450        queue = deque([(self, self.parent, None)])
 451
 452        while queue:
 453            item, parent, key = queue.popleft()
 454
 455            yield item, parent, key
 456            if prune and prune(item, parent, key):
 457                continue
 458
 459            for k, v in item.iter_expressions():
 460                queue.append((v, item, k))
 461
 462    def unnest(self):
 463        """
 464        Returns the first non parenthesis child or self.
 465        """
 466        expression = self
 467        while type(expression) is Paren:
 468            expression = expression.this
 469        return expression
 470
 471    def unalias(self):
 472        """
 473        Returns the inner expression if this is an Alias.
 474        """
 475        if isinstance(self, Alias):
 476            return self.this
 477        return self
 478
 479    def unnest_operands(self):
 480        """
 481        Returns unnested operands as a tuple.
 482        """
 483        return tuple(arg.unnest() for _, arg in self.iter_expressions())
 484
 485    def flatten(self, unnest=True):
 486        """
 487        Returns a generator which yields child nodes who's parents are the same class.
 488
 489        A AND B AND C -> [A, B, C]
 490        """
 491        for node, _, _ in self.dfs(prune=lambda n, p, *_: p and not type(n) is self.__class__):
 492            if not type(node) is self.__class__:
 493                yield node.unnest() if unnest and not isinstance(node, Subquery) else node
 494
 495    def __str__(self) -> str:
 496        return self.sql()
 497
 498    def __repr__(self) -> str:
 499        return self._to_s()
 500
 501    def sql(self, dialect: DialectType = None, **opts) -> str:
 502        """
 503        Returns SQL string representation of this tree.
 504
 505        Args:
 506            dialect: the dialect of the output SQL string (eg. "spark", "hive", "presto", "mysql").
 507            opts: other `sqlglot.generator.Generator` options.
 508
 509        Returns:
 510            The SQL string.
 511        """
 512        from sqlglot.dialects import Dialect
 513
 514        return Dialect.get_or_raise(dialect)().generate(self, **opts)
 515
 516    def _to_s(self, hide_missing: bool = True, level: int = 0) -> str:
 517        indent = "" if not level else "\n"
 518        indent += "".join(["  "] * level)
 519        left = f"({self.key.upper()} "
 520
 521        args: t.Dict[str, t.Any] = {
 522            k: ", ".join(
 523                v._to_s(hide_missing=hide_missing, level=level + 1)
 524                if hasattr(v, "_to_s")
 525                else str(v)
 526                for v in ensure_list(vs)
 527                if v is not None
 528            )
 529            for k, vs in self.args.items()
 530        }
 531        args["comments"] = self.comments
 532        args["type"] = self.type
 533        args = {k: v for k, v in args.items() if v or not hide_missing}
 534
 535        right = ", ".join(f"{k}: {v}" for k, v in args.items())
 536        right += ")"
 537
 538        return indent + left + right
 539
 540    def transform(self, fun, *args, copy=True, **kwargs):
 541        """
 542        Recursively visits all tree nodes (excluding already transformed ones)
 543        and applies the given transformation function to each node.
 544
 545        Args:
 546            fun (function): a function which takes a node as an argument and returns a
 547                new transformed node or the same node without modifications. If the function
 548                returns None, then the corresponding node will be removed from the syntax tree.
 549            copy (bool): if set to True a new tree instance is constructed, otherwise the tree is
 550                modified in place.
 551
 552        Returns:
 553            The transformed tree.
 554        """
 555        node = self.copy() if copy else self
 556        new_node = fun(node, *args, **kwargs)
 557
 558        if new_node is None or not isinstance(new_node, Expression):
 559            return new_node
 560        if new_node is not node:
 561            new_node.parent = node.parent
 562            return new_node
 563
 564        replace_children(new_node, lambda child: child.transform(fun, *args, copy=False, **kwargs))
 565        return new_node
 566
 567    @t.overload
 568    def replace(self, expression: E) -> E:
 569        ...
 570
 571    @t.overload
 572    def replace(self, expression: None) -> None:
 573        ...
 574
 575    def replace(self, expression):
 576        """
 577        Swap out this expression with a new expression.
 578
 579        For example::
 580
 581            >>> tree = Select().select("x").from_("tbl")
 582            >>> tree.find(Column).replace(Column(this="y"))
 583            (COLUMN this: y)
 584            >>> tree.sql()
 585            'SELECT y FROM tbl'
 586
 587        Args:
 588            expression: new node
 589
 590        Returns:
 591            The new expression or expressions.
 592        """
 593        if not self.parent:
 594            return expression
 595
 596        parent = self.parent
 597        self.parent = None
 598
 599        replace_children(parent, lambda child: expression if child is self else child)
 600        return expression
 601
 602    def pop(self: E) -> E:
 603        """
 604        Remove this expression from its AST.
 605
 606        Returns:
 607            The popped expression.
 608        """
 609        self.replace(None)
 610        return self
 611
 612    def assert_is(self, type_: t.Type[E]) -> E:
 613        """
 614        Assert that this `Expression` is an instance of `type_`.
 615
 616        If it is NOT an instance of `type_`, this raises an assertion error.
 617        Otherwise, this returns this expression.
 618
 619        Examples:
 620            This is useful for type security in chained expressions:
 621
 622            >>> import sqlglot
 623            >>> sqlglot.parse_one("SELECT x from y").assert_is(Select).select("z").sql()
 624            'SELECT x, z FROM y'
 625        """
 626        assert isinstance(self, type_)
 627        return self
 628
 629    def error_messages(self, args: t.Optional[t.Sequence] = None) -> t.List[str]:
 630        """
 631        Checks if this expression is valid (e.g. all mandatory args are set).
 632
 633        Args:
 634            args: a sequence of values that were used to instantiate a Func expression. This is used
 635                to check that the provided arguments don't exceed the function argument limit.
 636
 637        Returns:
 638            A list of error messages for all possible errors that were found.
 639        """
 640        errors: t.List[str] = []
 641
 642        for k in self.args:
 643            if k not in self.arg_types:
 644                errors.append(f"Unexpected keyword: '{k}' for {self.__class__}")
 645        for k, mandatory in self.arg_types.items():
 646            v = self.args.get(k)
 647            if mandatory and (v is None or (isinstance(v, list) and not v)):
 648                errors.append(f"Required keyword: '{k}' missing for {self.__class__}")
 649
 650        if (
 651            args
 652            and isinstance(self, Func)
 653            and len(args) > len(self.arg_types)
 654            and not self.is_var_len_args
 655        ):
 656            errors.append(
 657                f"The number of provided arguments ({len(args)}) is greater than "
 658                f"the maximum number of supported arguments ({len(self.arg_types)})"
 659            )
 660
 661        return errors
 662
 663    def dump(self):
 664        """
 665        Dump this Expression to a JSON-serializable dict.
 666        """
 667        from sqlglot.serde import dump
 668
 669        return dump(self)
 670
 671    @classmethod
 672    def load(cls, obj):
 673        """
 674        Load a dict (as returned by `Expression.dump`) into an Expression instance.
 675        """
 676        from sqlglot.serde import load
 677
 678        return load(obj)
 679
 680    def and_(
 681        self,
 682        *expressions: t.Optional[ExpOrStr],
 683        dialect: DialectType = None,
 684        copy: bool = True,
 685        **opts,
 686    ) -> Condition:
 687        """
 688        AND this condition with one or multiple expressions.
 689
 690        Example:
 691            >>> condition("x=1").and_("y=1").sql()
 692            'x = 1 AND y = 1'
 693
 694        Args:
 695            *expressions: the SQL code strings to parse.
 696                If an `Expression` instance is passed, it will be used as-is.
 697            dialect: the dialect used to parse the input expression.
 698            copy: whether or not to copy the involved expressions (only applies to Expressions).
 699            opts: other options to use to parse the input expressions.
 700
 701        Returns:
 702            The new And condition.
 703        """
 704        return and_(self, *expressions, dialect=dialect, copy=copy, **opts)
 705
 706    def or_(
 707        self,
 708        *expressions: t.Optional[ExpOrStr],
 709        dialect: DialectType = None,
 710        copy: bool = True,
 711        **opts,
 712    ) -> Condition:
 713        """
 714        OR this condition with one or multiple expressions.
 715
 716        Example:
 717            >>> condition("x=1").or_("y=1").sql()
 718            'x = 1 OR y = 1'
 719
 720        Args:
 721            *expressions: the SQL code strings to parse.
 722                If an `Expression` instance is passed, it will be used as-is.
 723            dialect: the dialect used to parse the input expression.
 724            copy: whether or not to copy the involved expressions (only applies to Expressions).
 725            opts: other options to use to parse the input expressions.
 726
 727        Returns:
 728            The new Or condition.
 729        """
 730        return or_(self, *expressions, dialect=dialect, copy=copy, **opts)
 731
 732    def not_(self, copy: bool = True):
 733        """
 734        Wrap this condition with NOT.
 735
 736        Example:
 737            >>> condition("x=1").not_().sql()
 738            'NOT x = 1'
 739
 740        Args:
 741            copy: whether or not to copy this object.
 742
 743        Returns:
 744            The new Not instance.
 745        """
 746        return not_(self, copy=copy)
 747
 748    def as_(
 749        self,
 750        alias: str | Identifier,
 751        quoted: t.Optional[bool] = None,
 752        dialect: DialectType = None,
 753        copy: bool = True,
 754        **opts,
 755    ) -> Alias:
 756        return alias_(self, alias, quoted=quoted, dialect=dialect, copy=copy, **opts)
 757
 758    def _binop(self, klass: t.Type[E], other: t.Any, reverse: bool = False) -> E:
 759        this = self.copy()
 760        other = convert(other, copy=True)
 761        if not isinstance(this, klass) and not isinstance(other, klass):
 762            this = _wrap(this, Binary)
 763            other = _wrap(other, Binary)
 764        if reverse:
 765            return klass(this=other, expression=this)
 766        return klass(this=this, expression=other)
 767
 768    def __getitem__(self, other: ExpOrStr | t.Tuple[ExpOrStr]) -> Bracket:
 769        return Bracket(
 770            this=self.copy(), expressions=[convert(e, copy=True) for e in ensure_list(other)]
 771        )
 772
 773    def __iter__(self) -> t.Iterator:
 774        if "expressions" in self.arg_types:
 775            return iter(self.args.get("expressions") or [])
 776        # We define this because __getitem__ converts Expression into an iterable, which is
 777        # problematic because one can hit infinite loops if they do "for x in some_expr: ..."
 778        # See: https://peps.python.org/pep-0234/
 779        raise TypeError(f"'{self.__class__.__name__}' object is not iterable")
 780
 781    def isin(
 782        self,
 783        *expressions: t.Any,
 784        query: t.Optional[ExpOrStr] = None,
 785        unnest: t.Optional[ExpOrStr] | t.Collection[ExpOrStr] = None,
 786        copy: bool = True,
 787        **opts,
 788    ) -> In:
 789        return In(
 790            this=maybe_copy(self, copy),
 791            expressions=[convert(e, copy=copy) for e in expressions],
 792            query=maybe_parse(query, copy=copy, **opts) if query else None,
 793            unnest=Unnest(
 794                expressions=[
 795                    maybe_parse(t.cast(ExpOrStr, e), copy=copy, **opts) for e in ensure_list(unnest)
 796                ]
 797            )
 798            if unnest
 799            else None,
 800        )
 801
 802    def between(self, low: t.Any, high: t.Any, copy: bool = True, **opts) -> Between:
 803        return Between(
 804            this=maybe_copy(self, copy),
 805            low=convert(low, copy=copy, **opts),
 806            high=convert(high, copy=copy, **opts),
 807        )
 808
 809    def is_(self, other: ExpOrStr) -> Is:
 810        return self._binop(Is, other)
 811
 812    def like(self, other: ExpOrStr) -> Like:
 813        return self._binop(Like, other)
 814
 815    def ilike(self, other: ExpOrStr) -> ILike:
 816        return self._binop(ILike, other)
 817
 818    def eq(self, other: t.Any) -> EQ:
 819        return self._binop(EQ, other)
 820
 821    def neq(self, other: t.Any) -> NEQ:
 822        return self._binop(NEQ, other)
 823
 824    def rlike(self, other: ExpOrStr) -> RegexpLike:
 825        return self._binop(RegexpLike, other)
 826
 827    def div(self, other: ExpOrStr, typed: bool = False, safe: bool = False) -> Div:
 828        div = self._binop(Div, other)
 829        div.args["typed"] = typed
 830        div.args["safe"] = safe
 831        return div
 832
 833    def __lt__(self, other: t.Any) -> LT:
 834        return self._binop(LT, other)
 835
 836    def __le__(self, other: t.Any) -> LTE:
 837        return self._binop(LTE, other)
 838
 839    def __gt__(self, other: t.Any) -> GT:
 840        return self._binop(GT, other)
 841
 842    def __ge__(self, other: t.Any) -> GTE:
 843        return self._binop(GTE, other)
 844
 845    def __add__(self, other: t.Any) -> Add:
 846        return self._binop(Add, other)
 847
 848    def __radd__(self, other: t.Any) -> Add:
 849        return self._binop(Add, other, reverse=True)
 850
 851    def __sub__(self, other: t.Any) -> Sub:
 852        return self._binop(Sub, other)
 853
 854    def __rsub__(self, other: t.Any) -> Sub:
 855        return self._binop(Sub, other, reverse=True)
 856
 857    def __mul__(self, other: t.Any) -> Mul:
 858        return self._binop(Mul, other)
 859
 860    def __rmul__(self, other: t.Any) -> Mul:
 861        return self._binop(Mul, other, reverse=True)
 862
 863    def __truediv__(self, other: t.Any) -> Div:
 864        return self._binop(Div, other)
 865
 866    def __rtruediv__(self, other: t.Any) -> Div:
 867        return self._binop(Div, other, reverse=True)
 868
 869    def __floordiv__(self, other: t.Any) -> IntDiv:
 870        return self._binop(IntDiv, other)
 871
 872    def __rfloordiv__(self, other: t.Any) -> IntDiv:
 873        return self._binop(IntDiv, other, reverse=True)
 874
 875    def __mod__(self, other: t.Any) -> Mod:
 876        return self._binop(Mod, other)
 877
 878    def __rmod__(self, other: t.Any) -> Mod:
 879        return self._binop(Mod, other, reverse=True)
 880
 881    def __pow__(self, other: t.Any) -> Pow:
 882        return self._binop(Pow, other)
 883
 884    def __rpow__(self, other: t.Any) -> Pow:
 885        return self._binop(Pow, other, reverse=True)
 886
 887    def __and__(self, other: t.Any) -> And:
 888        return self._binop(And, other)
 889
 890    def __rand__(self, other: t.Any) -> And:
 891        return self._binop(And, other, reverse=True)
 892
 893    def __or__(self, other: t.Any) -> Or:
 894        return self._binop(Or, other)
 895
 896    def __ror__(self, other: t.Any) -> Or:
 897        return self._binop(Or, other, reverse=True)
 898
 899    def __neg__(self) -> Neg:
 900        return Neg(this=_wrap(self.copy(), Binary))
 901
 902    def __invert__(self) -> Not:
 903        return not_(self.copy())
 904
 905
 906IntoType = t.Union[
 907    str,
 908    t.Type[Expression],
 909    t.Collection[t.Union[str, t.Type[Expression]]],
 910]
 911ExpOrStr = t.Union[str, Expression]
 912
 913
 914class Condition(Expression):
 915    """Logical conditions like x AND y, or simply x"""
 916
 917
 918class Predicate(Condition):
 919    """Relationships like x = y, x > 1, x >= y."""
 920
 921
 922class DerivedTable(Expression):
 923    @property
 924    def selects(self) -> t.List[Expression]:
 925        return self.this.selects if isinstance(self.this, Subqueryable) else []
 926
 927    @property
 928    def named_selects(self) -> t.List[str]:
 929        return [select.output_name for select in self.selects]
 930
 931
 932class Unionable(Expression):
 933    def union(
 934        self, expression: ExpOrStr, distinct: bool = True, dialect: DialectType = None, **opts
 935    ) -> Unionable:
 936        """
 937        Builds a UNION expression.
 938
 939        Example:
 940            >>> import sqlglot
 941            >>> sqlglot.parse_one("SELECT * FROM foo").union("SELECT * FROM bla").sql()
 942            'SELECT * FROM foo UNION SELECT * FROM bla'
 943
 944        Args:
 945            expression: the SQL code string.
 946                If an `Expression` instance is passed, it will be used as-is.
 947            distinct: set the DISTINCT flag if and only if this is true.
 948            dialect: the dialect used to parse the input expression.
 949            opts: other options to use to parse the input expressions.
 950
 951        Returns:
 952            The new Union expression.
 953        """
 954        return union(left=self, right=expression, distinct=distinct, dialect=dialect, **opts)
 955
 956    def intersect(
 957        self, expression: ExpOrStr, distinct: bool = True, dialect: DialectType = None, **opts
 958    ) -> Unionable:
 959        """
 960        Builds an INTERSECT expression.
 961
 962        Example:
 963            >>> import sqlglot
 964            >>> sqlglot.parse_one("SELECT * FROM foo").intersect("SELECT * FROM bla").sql()
 965            'SELECT * FROM foo INTERSECT SELECT * FROM bla'
 966
 967        Args:
 968            expression: the SQL code string.
 969                If an `Expression` instance is passed, it will be used as-is.
 970            distinct: set the DISTINCT flag if and only if this is true.
 971            dialect: the dialect used to parse the input expression.
 972            opts: other options to use to parse the input expressions.
 973
 974        Returns:
 975            The new Intersect expression.
 976        """
 977        return intersect(left=self, right=expression, distinct=distinct, dialect=dialect, **opts)
 978
 979    def except_(
 980        self, expression: ExpOrStr, distinct: bool = True, dialect: DialectType = None, **opts
 981    ) -> Unionable:
 982        """
 983        Builds an EXCEPT expression.
 984
 985        Example:
 986            >>> import sqlglot
 987            >>> sqlglot.parse_one("SELECT * FROM foo").except_("SELECT * FROM bla").sql()
 988            'SELECT * FROM foo EXCEPT SELECT * FROM bla'
 989
 990        Args:
 991            expression: the SQL code string.
 992                If an `Expression` instance is passed, it will be used as-is.
 993            distinct: set the DISTINCT flag if and only if this is true.
 994            dialect: the dialect used to parse the input expression.
 995            opts: other options to use to parse the input expressions.
 996
 997        Returns:
 998            The new Except expression.
 999        """
1000        return except_(left=self, right=expression, distinct=distinct, dialect=dialect, **opts)
1001
1002
1003class UDTF(DerivedTable, Unionable):
1004    @property
1005    def selects(self) -> t.List[Expression]:
1006        alias = self.args.get("alias")
1007        return alias.columns if alias else []
1008
1009
1010class Cache(Expression):
1011    arg_types = {
1012        "with": False,
1013        "this": True,
1014        "lazy": False,
1015        "options": False,
1016        "expression": False,
1017    }
1018
1019
1020class Uncache(Expression):
1021    arg_types = {"this": True, "exists": False}
1022
1023
1024class Refresh(Expression):
1025    pass
1026
1027
1028class DDL(Expression):
1029    @property
1030    def ctes(self):
1031        with_ = self.args.get("with")
1032        if not with_:
1033            return []
1034        return with_.expressions
1035
1036    @property
1037    def named_selects(self) -> t.List[str]:
1038        if isinstance(self.expression, Subqueryable):
1039            return self.expression.named_selects
1040        return []
1041
1042    @property
1043    def selects(self) -> t.List[Expression]:
1044        if isinstance(self.expression, Subqueryable):
1045            return self.expression.selects
1046        return []
1047
1048
1049class Create(DDL):
1050    arg_types = {
1051        "with": False,
1052        "this": True,
1053        "kind": True,
1054        "expression": False,
1055        "exists": False,
1056        "properties": False,
1057        "replace": False,
1058        "unique": False,
1059        "indexes": False,
1060        "no_schema_binding": False,
1061        "begin": False,
1062        "end": False,
1063        "clone": False,
1064    }
1065
1066
1067# https://docs.snowflake.com/en/sql-reference/sql/create-clone
1068# https://cloud.google.com/bigquery/docs/reference/standard-sql/data-definition-language#create_table_clone_statement
1069# https://cloud.google.com/bigquery/docs/reference/standard-sql/data-definition-language#create_table_copy
1070class Clone(Expression):
1071    arg_types = {
1072        "this": True,
1073        "when": False,
1074        "kind": False,
1075        "shallow": False,
1076        "expression": False,
1077        "copy": False,
1078    }
1079
1080
1081class Describe(Expression):
1082    arg_types = {"this": True, "kind": False, "expressions": False}
1083
1084
1085class Kill(Expression):
1086    arg_types = {"this": True, "kind": False}
1087
1088
1089class Pragma(Expression):
1090    pass
1091
1092
1093class Set(Expression):
1094    arg_types = {"expressions": False, "unset": False, "tag": False}
1095
1096
1097class SetItem(Expression):
1098    arg_types = {
1099        "this": False,
1100        "expressions": False,
1101        "kind": False,
1102        "collate": False,  # MySQL SET NAMES statement
1103        "global": False,
1104    }
1105
1106
1107class Show(Expression):
1108    arg_types = {
1109        "this": True,
1110        "target": False,
1111        "offset": False,
1112        "limit": False,
1113        "like": False,
1114        "where": False,
1115        "db": False,
1116        "scope": False,
1117        "scope_kind": False,
1118        "full": False,
1119        "mutex": False,
1120        "query": False,
1121        "channel": False,
1122        "global": False,
1123        "log": False,
1124        "position": False,
1125        "types": False,
1126    }
1127
1128
1129class UserDefinedFunction(Expression):
1130    arg_types = {"this": True, "expressions": False, "wrapped": False}
1131
1132
1133class CharacterSet(Expression):
1134    arg_types = {"this": True, "default": False}
1135
1136
1137class With(Expression):
1138    arg_types = {"expressions": True, "recursive": False}
1139
1140    @property
1141    def recursive(self) -> bool:
1142        return bool(self.args.get("recursive"))
1143
1144
1145class WithinGroup(Expression):
1146    arg_types = {"this": True, "expression": False}
1147
1148
1149class CTE(DerivedTable):
1150    arg_types = {"this": True, "alias": True}
1151
1152
1153class TableAlias(Expression):
1154    arg_types = {"this": False, "columns": False}
1155
1156    @property
1157    def columns(self):
1158        return self.args.get("columns") or []
1159
1160
1161class BitString(Condition):
1162    pass
1163
1164
1165class HexString(Condition):
1166    pass
1167
1168
1169class ByteString(Condition):
1170    pass
1171
1172
1173class RawString(Condition):
1174    pass
1175
1176
1177class Column(Condition):
1178    arg_types = {"this": True, "table": False, "db": False, "catalog": False, "join_mark": False}
1179
1180    @property
1181    def table(self) -> str:
1182        return self.text("table")
1183
1184    @property
1185    def db(self) -> str:
1186        return self.text("db")
1187
1188    @property
1189    def catalog(self) -> str:
1190        return self.text("catalog")
1191
1192    @property
1193    def output_name(self) -> str:
1194        return self.name
1195
1196    @property
1197    def parts(self) -> t.List[Identifier]:
1198        """Return the parts of a column in order catalog, db, table, name."""
1199        return [
1200            t.cast(Identifier, self.args[part])
1201            for part in ("catalog", "db", "table", "this")
1202            if self.args.get(part)
1203        ]
1204
1205    def to_dot(self) -> Dot | Identifier:
1206        """Converts the column into a dot expression."""
1207        parts = self.parts
1208        parent = self.parent
1209
1210        while parent:
1211            if isinstance(parent, Dot):
1212                parts.append(parent.expression)
1213            parent = parent.parent
1214
1215        return Dot.build(deepcopy(parts)) if len(parts) > 1 else parts[0]
1216
1217
1218class ColumnPosition(Expression):
1219    arg_types = {"this": False, "position": True}
1220
1221
1222class ColumnDef(Expression):
1223    arg_types = {
1224        "this": True,
1225        "kind": False,
1226        "constraints": False,
1227        "exists": False,
1228        "position": False,
1229    }
1230
1231    @property
1232    def constraints(self) -> t.List[ColumnConstraint]:
1233        return self.args.get("constraints") or []
1234
1235
1236class AlterColumn(Expression):
1237    arg_types = {
1238        "this": True,
1239        "dtype": False,
1240        "collate": False,
1241        "using": False,
1242        "default": False,
1243        "drop": False,
1244    }
1245
1246
1247class RenameTable(Expression):
1248    pass
1249
1250
1251class SwapTable(Expression):
1252    pass
1253
1254
1255class Comment(Expression):
1256    arg_types = {"this": True, "kind": True, "expression": True, "exists": False}
1257
1258
1259class Comprehension(Expression):
1260    arg_types = {"this": True, "expression": True, "iterator": True, "condition": False}
1261
1262
1263# https://clickhouse.com/docs/en/engines/table-engines/mergetree-family/mergetree#mergetree-table-ttl
1264class MergeTreeTTLAction(Expression):
1265    arg_types = {
1266        "this": True,
1267        "delete": False,
1268        "recompress": False,
1269        "to_disk": False,
1270        "to_volume": False,
1271    }
1272
1273
1274# https://clickhouse.com/docs/en/engines/table-engines/mergetree-family/mergetree#mergetree-table-ttl
1275class MergeTreeTTL(Expression):
1276    arg_types = {
1277        "expressions": True,
1278        "where": False,
1279        "group": False,
1280        "aggregates": False,
1281    }
1282
1283
1284# https://dev.mysql.com/doc/refman/8.0/en/create-table.html
1285class IndexConstraintOption(Expression):
1286    arg_types = {
1287        "key_block_size": False,
1288        "using": False,
1289        "parser": False,
1290        "comment": False,
1291        "visible": False,
1292        "engine_attr": False,
1293        "secondary_engine_attr": False,
1294    }
1295
1296
1297class ColumnConstraint(Expression):
1298    arg_types = {"this": False, "kind": True}
1299
1300    @property
1301    def kind(self) -> ColumnConstraintKind:
1302        return self.args["kind"]
1303
1304
1305class ColumnConstraintKind(Expression):
1306    pass
1307
1308
1309class AutoIncrementColumnConstraint(ColumnConstraintKind):
1310    pass
1311
1312
1313class PeriodForSystemTimeConstraint(ColumnConstraintKind):
1314    arg_types = {"this": True, "expression": True}
1315
1316
1317class CaseSpecificColumnConstraint(ColumnConstraintKind):
1318    arg_types = {"not_": True}
1319
1320
1321class CharacterSetColumnConstraint(ColumnConstraintKind):
1322    arg_types = {"this": True}
1323
1324
1325class CheckColumnConstraint(ColumnConstraintKind):
1326    pass
1327
1328
1329class ClusteredColumnConstraint(ColumnConstraintKind):
1330    pass
1331
1332
1333class CollateColumnConstraint(ColumnConstraintKind):
1334    pass
1335
1336
1337class CommentColumnConstraint(ColumnConstraintKind):
1338    pass
1339
1340
1341class CompressColumnConstraint(ColumnConstraintKind):
1342    pass
1343
1344
1345class DateFormatColumnConstraint(ColumnConstraintKind):
1346    arg_types = {"this": True}
1347
1348
1349class DefaultColumnConstraint(ColumnConstraintKind):
1350    pass
1351
1352
1353class EncodeColumnConstraint(ColumnConstraintKind):
1354    pass
1355
1356
1357class GeneratedAsIdentityColumnConstraint(ColumnConstraintKind):
1358    # this: True -> ALWAYS, this: False -> BY DEFAULT
1359    arg_types = {
1360        "this": False,
1361        "expression": False,
1362        "on_null": False,
1363        "start": False,
1364        "increment": False,
1365        "minvalue": False,
1366        "maxvalue": False,
1367        "cycle": False,
1368    }
1369
1370
1371class GeneratedAsRowColumnConstraint(ColumnConstraintKind):
1372    arg_types = {"start": True, "hidden": False}
1373
1374
1375# https://dev.mysql.com/doc/refman/8.0/en/create-table.html
1376class IndexColumnConstraint(ColumnConstraintKind):
1377    arg_types = {
1378        "this": False,
1379        "schema": True,
1380        "kind": False,
1381        "index_type": False,
1382        "options": False,
1383    }
1384
1385
1386class InlineLengthColumnConstraint(ColumnConstraintKind):
1387    pass
1388
1389
1390class NonClusteredColumnConstraint(ColumnConstraintKind):
1391    pass
1392
1393
1394class NotForReplicationColumnConstraint(ColumnConstraintKind):
1395    arg_types = {}
1396
1397
1398class NotNullColumnConstraint(ColumnConstraintKind):
1399    arg_types = {"allow_null": False}
1400
1401
1402# https://dev.mysql.com/doc/refman/5.7/en/timestamp-initialization.html
1403class OnUpdateColumnConstraint(ColumnConstraintKind):
1404    pass
1405
1406
1407class PrimaryKeyColumnConstraint(ColumnConstraintKind):
1408    arg_types = {"desc": False}
1409
1410
1411class TitleColumnConstraint(ColumnConstraintKind):
1412    pass
1413
1414
1415class UniqueColumnConstraint(ColumnConstraintKind):
1416    arg_types = {"this": False, "index_type": False}
1417
1418
1419class UppercaseColumnConstraint(ColumnConstraintKind):
1420    arg_types: t.Dict[str, t.Any] = {}
1421
1422
1423class PathColumnConstraint(ColumnConstraintKind):
1424    pass
1425
1426
1427# computed column expression
1428# https://learn.microsoft.com/en-us/sql/t-sql/statements/create-table-transact-sql?view=sql-server-ver16
1429class ComputedColumnConstraint(ColumnConstraintKind):
1430    arg_types = {"this": True, "persisted": False, "not_null": False}
1431
1432
1433class Constraint(Expression):
1434    arg_types = {"this": True, "expressions": True}
1435
1436
1437class Delete(Expression):
1438    arg_types = {
1439        "with": False,
1440        "this": False,
1441        "using": False,
1442        "where": False,
1443        "returning": False,
1444        "limit": False,
1445        "tables": False,  # Multiple-Table Syntax (MySQL)
1446    }
1447
1448    def delete(
1449        self,
1450        table: ExpOrStr,
1451        dialect: DialectType = None,
1452        copy: bool = True,
1453        **opts,
1454    ) -> Delete:
1455        """
1456        Create a DELETE expression or replace the table on an existing DELETE expression.
1457
1458        Example:
1459            >>> delete("tbl").sql()
1460            'DELETE FROM tbl'
1461
1462        Args:
1463            table: the table from which to delete.
1464            dialect: the dialect used to parse the input expression.
1465            copy: if `False`, modify this expression instance in-place.
1466            opts: other options to use to parse the input expressions.
1467
1468        Returns:
1469            Delete: the modified expression.
1470        """
1471        return _apply_builder(
1472            expression=table,
1473            instance=self,
1474            arg="this",
1475            dialect=dialect,
1476            into=Table,
1477            copy=copy,
1478            **opts,
1479        )
1480
1481    def where(
1482        self,
1483        *expressions: t.Optional[ExpOrStr],
1484        append: bool = True,
1485        dialect: DialectType = None,
1486        copy: bool = True,
1487        **opts,
1488    ) -> Delete:
1489        """
1490        Append to or set the WHERE expressions.
1491
1492        Example:
1493            >>> delete("tbl").where("x = 'a' OR x < 'b'").sql()
1494            "DELETE FROM tbl WHERE x = 'a' OR x < 'b'"
1495
1496        Args:
1497            *expressions: the SQL code strings to parse.
1498                If an `Expression` instance is passed, it will be used as-is.
1499                Multiple expressions are combined with an AND operator.
1500            append: if `True`, AND the new expressions to any existing expression.
1501                Otherwise, this resets the expression.
1502            dialect: the dialect used to parse the input expressions.
1503            copy: if `False`, modify this expression instance in-place.
1504            opts: other options to use to parse the input expressions.
1505
1506        Returns:
1507            Delete: the modified expression.
1508        """
1509        return _apply_conjunction_builder(
1510            *expressions,
1511            instance=self,
1512            arg="where",
1513            append=append,
1514            into=Where,
1515            dialect=dialect,
1516            copy=copy,
1517            **opts,
1518        )
1519
1520    def returning(
1521        self,
1522        expression: ExpOrStr,
1523        dialect: DialectType = None,
1524        copy: bool = True,
1525        **opts,
1526    ) -> Delete:
1527        """
1528        Set the RETURNING expression. Not supported by all dialects.
1529
1530        Example:
1531            >>> delete("tbl").returning("*", dialect="postgres").sql()
1532            'DELETE FROM tbl RETURNING *'
1533
1534        Args:
1535            expression: the SQL code strings to parse.
1536                If an `Expression` instance is passed, it will be used as-is.
1537            dialect: the dialect used to parse the input expressions.
1538            copy: if `False`, modify this expression instance in-place.
1539            opts: other options to use to parse the input expressions.
1540
1541        Returns:
1542            Delete: the modified expression.
1543        """
1544        return _apply_builder(
1545            expression=expression,
1546            instance=self,
1547            arg="returning",
1548            prefix="RETURNING",
1549            dialect=dialect,
1550            copy=copy,
1551            into=Returning,
1552            **opts,
1553        )
1554
1555
1556class Drop(Expression):
1557    arg_types = {
1558        "this": False,
1559        "kind": False,
1560        "exists": False,
1561        "temporary": False,
1562        "materialized": False,
1563        "cascade": False,
1564        "constraints": False,
1565        "purge": False,
1566    }
1567
1568
1569class Filter(Expression):
1570    arg_types = {"this": True, "expression": True}
1571
1572
1573class Check(Expression):
1574    pass
1575
1576
1577# https://docs.snowflake.com/en/sql-reference/constructs/connect-by
1578class Connect(Expression):
1579    arg_types = {"start": False, "connect": True}
1580
1581
1582class Prior(Expression):
1583    pass
1584
1585
1586class Directory(Expression):
1587    # https://spark.apache.org/docs/3.0.0-preview/sql-ref-syntax-dml-insert-overwrite-directory-hive.html
1588    arg_types = {"this": True, "local": False, "row_format": False}
1589
1590
1591class ForeignKey(Expression):
1592    arg_types = {
1593        "expressions": True,
1594        "reference": False,
1595        "delete": False,
1596        "update": False,
1597    }
1598
1599
1600class ColumnPrefix(Expression):
1601    arg_types = {"this": True, "expression": True}
1602
1603
1604class PrimaryKey(Expression):
1605    arg_types = {"expressions": True, "options": False}
1606
1607
1608# https://www.postgresql.org/docs/9.1/sql-selectinto.html
1609# https://docs.aws.amazon.com/redshift/latest/dg/r_SELECT_INTO.html#r_SELECT_INTO-examples
1610class Into(Expression):
1611    arg_types = {"this": True, "temporary": False, "unlogged": False}
1612
1613
1614class From(Expression):
1615    @property
1616    def name(self) -> str:
1617        return self.this.name
1618
1619    @property
1620    def alias_or_name(self) -> str:
1621        return self.this.alias_or_name
1622
1623
1624class Having(Expression):
1625    pass
1626
1627
1628class Hint(Expression):
1629    arg_types = {"expressions": True}
1630
1631
1632class JoinHint(Expression):
1633    arg_types = {"this": True, "expressions": True}
1634
1635
1636class Identifier(Expression):
1637    arg_types = {"this": True, "quoted": False, "global": False, "temporary": False}
1638
1639    @property
1640    def quoted(self) -> bool:
1641        return bool(self.args.get("quoted"))
1642
1643    @property
1644    def hashable_args(self) -> t.Any:
1645        return (self.this, self.quoted)
1646
1647    @property
1648    def output_name(self) -> str:
1649        return self.name
1650
1651
1652# https://www.postgresql.org/docs/current/indexes-opclass.html
1653class Opclass(Expression):
1654    arg_types = {"this": True, "expression": True}
1655
1656
1657class Index(Expression):
1658    arg_types = {
1659        "this": False,
1660        "table": False,
1661        "using": False,
1662        "where": False,
1663        "columns": False,
1664        "unique": False,
1665        "primary": False,
1666        "amp": False,  # teradata
1667        "partition_by": False,  # teradata
1668        "where": False,  # postgres partial indexes
1669    }
1670
1671
1672class Insert(DDL):
1673    arg_types = {
1674        "with": False,
1675        "this": True,
1676        "expression": False,
1677        "conflict": False,
1678        "returning": False,
1679        "overwrite": False,
1680        "exists": False,
1681        "partition": False,
1682        "alternative": False,
1683        "where": False,
1684        "ignore": False,
1685        "by_name": False,
1686    }
1687
1688    def with_(
1689        self,
1690        alias: ExpOrStr,
1691        as_: ExpOrStr,
1692        recursive: t.Optional[bool] = None,
1693        append: bool = True,
1694        dialect: DialectType = None,
1695        copy: bool = True,
1696        **opts,
1697    ) -> Insert:
1698        """
1699        Append to or set the common table expressions.
1700
1701        Example:
1702            >>> insert("SELECT x FROM cte", "t").with_("cte", as_="SELECT * FROM tbl").sql()
1703            'WITH cte AS (SELECT * FROM tbl) INSERT INTO t SELECT x FROM cte'
1704
1705        Args:
1706            alias: the SQL code string to parse as the table name.
1707                If an `Expression` instance is passed, this is used as-is.
1708            as_: the SQL code string to parse as the table expression.
1709                If an `Expression` instance is passed, it will be used as-is.
1710            recursive: set the RECURSIVE part of the expression. Defaults to `False`.
1711            append: if `True`, add to any existing expressions.
1712                Otherwise, this resets the expressions.
1713            dialect: the dialect used to parse the input expression.
1714            copy: if `False`, modify this expression instance in-place.
1715            opts: other options to use to parse the input expressions.
1716
1717        Returns:
1718            The modified expression.
1719        """
1720        return _apply_cte_builder(
1721            self, alias, as_, recursive=recursive, append=append, dialect=dialect, copy=copy, **opts
1722        )
1723
1724
1725class OnConflict(Expression):
1726    arg_types = {
1727        "duplicate": False,
1728        "expressions": False,
1729        "nothing": False,
1730        "key": False,
1731        "constraint": False,
1732    }
1733
1734
1735class Returning(Expression):
1736    arg_types = {"expressions": True, "into": False}
1737
1738
1739# https://dev.mysql.com/doc/refman/8.0/en/charset-introducer.html
1740class Introducer(Expression):
1741    arg_types = {"this": True, "expression": True}
1742
1743
1744# national char, like n'utf8'
1745class National(Expression):
1746    pass
1747
1748
1749class LoadData(Expression):
1750    arg_types = {
1751        "this": True,
1752        "local": False,
1753        "overwrite": False,
1754        "inpath": True,
1755        "partition": False,
1756        "input_format": False,
1757        "serde": False,
1758    }
1759
1760
1761class Partition(Expression):
1762    arg_types = {"expressions": True}
1763
1764
1765class Fetch(Expression):
1766    arg_types = {
1767        "direction": False,
1768        "count": False,
1769        "percent": False,
1770        "with_ties": False,
1771    }
1772
1773
1774class Group(Expression):
1775    arg_types = {
1776        "expressions": False,
1777        "grouping_sets": False,
1778        "cube": False,
1779        "rollup": False,
1780        "totals": False,
1781        "all": False,
1782    }
1783
1784
1785class Lambda(Expression):
1786    arg_types = {"this": True, "expressions": True}
1787
1788
1789class Limit(Expression):
1790    arg_types = {"this": False, "expression": True, "offset": False}
1791
1792
1793class Literal(Condition):
1794    arg_types = {"this": True, "is_string": True}
1795
1796    @property
1797    def hashable_args(self) -> t.Any:
1798        return (self.this, self.args.get("is_string"))
1799
1800    @classmethod
1801    def number(cls, number) -> Literal:
1802        return cls(this=str(number), is_string=False)
1803
1804    @classmethod
1805    def string(cls, string) -> Literal:
1806        return cls(this=str(string), is_string=True)
1807
1808    @property
1809    def output_name(self) -> str:
1810        return self.name
1811
1812
1813class Join(Expression):
1814    arg_types = {
1815        "this": True,
1816        "on": False,
1817        "side": False,
1818        "kind": False,
1819        "using": False,
1820        "method": False,
1821        "global": False,
1822        "hint": False,
1823    }
1824
1825    @property
1826    def method(self) -> str:
1827        return self.text("method").upper()
1828
1829    @property
1830    def kind(self) -> str:
1831        return self.text("kind").upper()
1832
1833    @property
1834    def side(self) -> str:
1835        return self.text("side").upper()
1836
1837    @property
1838    def hint(self) -> str:
1839        return self.text("hint").upper()
1840
1841    @property
1842    def alias_or_name(self) -> str:
1843        return self.this.alias_or_name
1844
1845    def on(
1846        self,
1847        *expressions: t.Optional[ExpOrStr],
1848        append: bool = True,
1849        dialect: DialectType = None,
1850        copy: bool = True,
1851        **opts,
1852    ) -> Join:
1853        """
1854        Append to or set the ON expressions.
1855
1856        Example:
1857            >>> import sqlglot
1858            >>> sqlglot.parse_one("JOIN x", into=Join).on("y = 1").sql()
1859            'JOIN x ON y = 1'
1860
1861        Args:
1862            *expressions: the SQL code strings to parse.
1863                If an `Expression` instance is passed, it will be used as-is.
1864                Multiple expressions are combined with an AND operator.
1865            append: if `True`, AND the new expressions to any existing expression.
1866                Otherwise, this resets the expression.
1867            dialect: the dialect used to parse the input expressions.
1868            copy: if `False`, modify this expression instance in-place.
1869            opts: other options to use to parse the input expressions.
1870
1871        Returns:
1872            The modified Join expression.
1873        """
1874        join = _apply_conjunction_builder(
1875            *expressions,
1876            instance=self,
1877            arg="on",
1878            append=append,
1879            dialect=dialect,
1880            copy=copy,
1881            **opts,
1882        )
1883
1884        if join.kind == "CROSS":
1885            join.set("kind", None)
1886
1887        return join
1888
1889    def using(
1890        self,
1891        *expressions: t.Optional[ExpOrStr],
1892        append: bool = True,
1893        dialect: DialectType = None,
1894        copy: bool = True,
1895        **opts,
1896    ) -> Join:
1897        """
1898        Append to or set the USING expressions.
1899
1900        Example:
1901            >>> import sqlglot
1902            >>> sqlglot.parse_one("JOIN x", into=Join).using("foo", "bla").sql()
1903            'JOIN x USING (foo, bla)'
1904
1905        Args:
1906            *expressions: the SQL code strings to parse.
1907                If an `Expression` instance is passed, it will be used as-is.
1908            append: if `True`, concatenate the new expressions to the existing "using" list.
1909                Otherwise, this resets the expression.
1910            dialect: the dialect used to parse the input expressions.
1911            copy: if `False`, modify this expression instance in-place.
1912            opts: other options to use to parse the input expressions.
1913
1914        Returns:
1915            The modified Join expression.
1916        """
1917        join = _apply_list_builder(
1918            *expressions,
1919            instance=self,
1920            arg="using",
1921            append=append,
1922            dialect=dialect,
1923            copy=copy,
1924            **opts,
1925        )
1926
1927        if join.kind == "CROSS":
1928            join.set("kind", None)
1929
1930        return join
1931
1932
1933class Lateral(UDTF):
1934    arg_types = {"this": True, "view": False, "outer": False, "alias": False}
1935
1936
1937class MatchRecognize(Expression):
1938    arg_types = {
1939        "partition_by": False,
1940        "order": False,
1941        "measures": False,
1942        "rows": False,
1943        "after": False,
1944        "pattern": False,
1945        "define": False,
1946        "alias": False,
1947    }
1948
1949
1950# Clickhouse FROM FINAL modifier
1951# https://clickhouse.com/docs/en/sql-reference/statements/select/from/#final-modifier
1952class Final(Expression):
1953    pass
1954
1955
1956class Offset(Expression):
1957    arg_types = {"this": False, "expression": True}
1958
1959
1960class Order(Expression):
1961    arg_types = {"this": False, "expressions": True}
1962
1963
1964# hive specific sorts
1965# https://cwiki.apache.org/confluence/display/Hive/LanguageManual+SortBy
1966class Cluster(Order):
1967    pass
1968
1969
1970class Distribute(Order):
1971    pass
1972
1973
1974class Sort(Order):
1975    pass
1976
1977
1978class Ordered(Expression):
1979    arg_types = {"this": True, "desc": False, "nulls_first": True}
1980
1981
1982class Property(Expression):
1983    arg_types = {"this": True, "value": True}
1984
1985
1986class AlgorithmProperty(Property):
1987    arg_types = {"this": True}
1988
1989
1990class AutoIncrementProperty(Property):
1991    arg_types = {"this": True}
1992
1993
1994class BlockCompressionProperty(Property):
1995    arg_types = {"autotemp": False, "always": False, "default": True, "manual": True, "never": True}
1996
1997
1998class CharacterSetProperty(Property):
1999    arg_types = {"this": True, "default": True}
2000
2001
2002class ChecksumProperty(Property):
2003    arg_types = {"on": False, "default": False}
2004
2005
2006class CollateProperty(Property):
2007    arg_types = {"this": True, "default": False}
2008
2009
2010class CopyGrantsProperty(Property):
2011    arg_types = {}
2012
2013
2014class DataBlocksizeProperty(Property):
2015    arg_types = {
2016        "size": False,
2017        "units": False,
2018        "minimum": False,
2019        "maximum": False,
2020        "default": False,
2021    }
2022
2023
2024class DefinerProperty(Property):
2025    arg_types = {"this": True}
2026
2027
2028class DistKeyProperty(Property):
2029    arg_types = {"this": True}
2030
2031
2032class DistStyleProperty(Property):
2033    arg_types = {"this": True}
2034
2035
2036class EngineProperty(Property):
2037    arg_types = {"this": True}
2038
2039
2040class HeapProperty(Property):
2041    arg_types = {}
2042
2043
2044class ToTableProperty(Property):
2045    arg_types = {"this": True}
2046
2047
2048class ExecuteAsProperty(Property):
2049    arg_types = {"this": True}
2050
2051
2052class ExternalProperty(Property):
2053    arg_types = {"this": False}
2054
2055
2056class FallbackProperty(Property):
2057    arg_types = {"no": True, "protection": False}
2058
2059
2060class FileFormatProperty(Property):
2061    arg_types = {"this": True}
2062
2063
2064class FreespaceProperty(Property):
2065    arg_types = {"this": True, "percent": False}
2066
2067
2068class InputModelProperty(Property):
2069    arg_types = {"this": True}
2070
2071
2072class OutputModelProperty(Property):
2073    arg_types = {"this": True}
2074
2075
2076class IsolatedLoadingProperty(Property):
2077    arg_types = {
2078        "no": True,
2079        "concurrent": True,
2080        "for_all": True,
2081        "for_insert": True,
2082        "for_none": True,
2083    }
2084
2085
2086class JournalProperty(Property):
2087    arg_types = {
2088        "no": False,
2089        "dual": False,
2090        "before": False,
2091        "local": False,
2092        "after": False,
2093    }
2094
2095
2096class LanguageProperty(Property):
2097    arg_types = {"this": True}
2098
2099
2100# spark ddl
2101class ClusteredByProperty(Property):
2102    arg_types = {"expressions": True, "sorted_by": False, "buckets": True}
2103
2104
2105class DictProperty(Property):
2106    arg_types = {"this": True, "kind": True, "settings": False}
2107
2108
2109class DictSubProperty(Property):
2110    pass
2111
2112
2113class DictRange(Property):
2114    arg_types = {"this": True, "min": True, "max": True}
2115
2116
2117# Clickhouse CREATE ... ON CLUSTER modifier
2118# https://clickhouse.com/docs/en/sql-reference/distributed-ddl
2119class OnCluster(Property):
2120    arg_types = {"this": True}
2121
2122
2123class LikeProperty(Property):
2124    arg_types = {"this": True, "expressions": False}
2125
2126
2127class LocationProperty(Property):
2128    arg_types = {"this": True}
2129
2130
2131class LockingProperty(Property):
2132    arg_types = {
2133        "this": False,
2134        "kind": True,
2135        "for_or_in": False,
2136        "lock_type": True,
2137        "override": False,
2138    }
2139
2140
2141class LogProperty(Property):
2142    arg_types = {"no": True}
2143
2144
2145class MaterializedProperty(Property):
2146    arg_types = {"this": False}
2147
2148
2149class MergeBlockRatioProperty(Property):
2150    arg_types = {"this": False, "no": False, "default": False, "percent": False}
2151
2152
2153class NoPrimaryIndexProperty(Property):
2154    arg_types = {}
2155
2156
2157class OnProperty(Property):
2158    arg_types = {"this": True}
2159
2160
2161class OnCommitProperty(Property):
2162    arg_types = {"delete": False}
2163
2164
2165class PartitionedByProperty(Property):
2166    arg_types = {"this": True}
2167
2168
2169# https://www.postgresql.org/docs/current/sql-createtable.html
2170class PartitionBoundSpec(Expression):
2171    # this -> IN / MODULUS, expression -> REMAINDER, from_expressions -> FROM (...), to_expressions -> TO (...)
2172    arg_types = {
2173        "this": False,
2174        "expression": False,
2175        "from_expressions": False,
2176        "to_expressions": False,
2177    }
2178
2179
2180class PartitionedOfProperty(Property):
2181    # this -> parent_table (schema), expression -> FOR VALUES ... / DEFAULT
2182    arg_types = {"this": True, "expression": True}
2183
2184
2185class RemoteWithConnectionModelProperty(Property):
2186    arg_types = {"this": True}
2187
2188
2189class ReturnsProperty(Property):
2190    arg_types = {"this": True, "is_table": False, "table": False}
2191
2192
2193class RowFormatProperty(Property):
2194    arg_types = {"this": True}
2195
2196
2197class RowFormatDelimitedProperty(Property):
2198    # https://cwiki.apache.org/confluence/display/hive/languagemanual+dml
2199    arg_types = {
2200        "fields": False,
2201        "escaped": False,
2202        "collection_items": False,
2203        "map_keys": False,
2204        "lines": False,
2205        "null": False,
2206        "serde": False,
2207    }
2208
2209
2210class RowFormatSerdeProperty(Property):
2211    arg_types = {"this": True, "serde_properties": False}
2212
2213
2214# https://spark.apache.org/docs/3.1.2/sql-ref-syntax-qry-select-transform.html
2215class QueryTransform(Expression):
2216    arg_types = {
2217        "expressions": True,
2218        "command_script": True,
2219        "schema": False,
2220        "row_format_before": False,
2221        "record_writer": False,
2222        "row_format_after": False,
2223        "record_reader": False,
2224    }
2225
2226
2227class SampleProperty(Property):
2228    arg_types = {"this": True}
2229
2230
2231class SchemaCommentProperty(Property):
2232    arg_types = {"this": True}
2233
2234
2235class SerdeProperties(Property):
2236    arg_types = {"expressions": True}
2237
2238
2239class SetProperty(Property):
2240    arg_types = {"multi": True}
2241
2242
2243class SettingsProperty(Property):
2244    arg_types = {"expressions": True}
2245
2246
2247class SortKeyProperty(Property):
2248    arg_types = {"this": True, "compound": False}
2249
2250
2251class SqlSecurityProperty(Property):
2252    arg_types = {"definer": True}
2253
2254
2255class StabilityProperty(Property):
2256    arg_types = {"this": True}
2257
2258
2259class TemporaryProperty(Property):
2260    arg_types = {}
2261
2262
2263class TransformModelProperty(Property):
2264    arg_types = {"expressions": True}
2265
2266
2267class TransientProperty(Property):
2268    arg_types = {"this": False}
2269
2270
2271class VolatileProperty(Property):
2272    arg_types = {"this": False}
2273
2274
2275class WithDataProperty(Property):
2276    arg_types = {"no": True, "statistics": False}
2277
2278
2279class WithJournalTableProperty(Property):
2280    arg_types = {"this": True}
2281
2282
2283class WithSystemVersioningProperty(Property):
2284    # this -> history table name, expression -> data consistency check
2285    arg_types = {"this": False, "expression": False}
2286
2287
2288class Properties(Expression):
2289    arg_types = {"expressions": True}
2290
2291    NAME_TO_PROPERTY = {
2292        "ALGORITHM": AlgorithmProperty,
2293        "AUTO_INCREMENT": AutoIncrementProperty,
2294        "CHARACTER SET": CharacterSetProperty,
2295        "CLUSTERED_BY": ClusteredByProperty,
2296        "COLLATE": CollateProperty,
2297        "COMMENT": SchemaCommentProperty,
2298        "DEFINER": DefinerProperty,
2299        "DISTKEY": DistKeyProperty,
2300        "DISTSTYLE": DistStyleProperty,
2301        "ENGINE": EngineProperty,
2302        "EXECUTE AS": ExecuteAsProperty,
2303        "FORMAT": FileFormatProperty,
2304        "LANGUAGE": LanguageProperty,
2305        "LOCATION": LocationProperty,
2306        "PARTITIONED_BY": PartitionedByProperty,
2307        "RETURNS": ReturnsProperty,
2308        "ROW_FORMAT": RowFormatProperty,
2309        "SORTKEY": SortKeyProperty,
2310    }
2311
2312    PROPERTY_TO_NAME = {v: k for k, v in NAME_TO_PROPERTY.items()}
2313
2314    # CREATE property locations
2315    # Form: schema specified
2316    #   create [POST_CREATE]
2317    #     table a [POST_NAME]
2318    #     (b int) [POST_SCHEMA]
2319    #     with ([POST_WITH])
2320    #     index (b) [POST_INDEX]
2321    #
2322    # Form: alias selection
2323    #   create [POST_CREATE]
2324    #     table a [POST_NAME]
2325    #     as [POST_ALIAS] (select * from b) [POST_EXPRESSION]
2326    #     index (c) [POST_INDEX]
2327    class Location(AutoName):
2328        POST_CREATE = auto()
2329        POST_NAME = auto()
2330        POST_SCHEMA = auto()
2331        POST_WITH = auto()
2332        POST_ALIAS = auto()
2333        POST_EXPRESSION = auto()
2334        POST_INDEX = auto()
2335        UNSUPPORTED = auto()
2336
2337    @classmethod
2338    def from_dict(cls, properties_dict: t.Dict) -> Properties:
2339        expressions = []
2340        for key, value in properties_dict.items():
2341            property_cls = cls.NAME_TO_PROPERTY.get(key.upper())
2342            if property_cls:
2343                expressions.append(property_cls(this=convert(value)))
2344            else:
2345                expressions.append(Property(this=Literal.string(key), value=convert(value)))
2346
2347        return cls(expressions=expressions)
2348
2349
2350class Qualify(Expression):
2351    pass
2352
2353
2354class InputOutputFormat(Expression):
2355    arg_types = {"input_format": False, "output_format": False}
2356
2357
2358# https://www.ibm.com/docs/en/ias?topic=procedures-return-statement-in-sql
2359class Return(Expression):
2360    pass
2361
2362
2363class Reference(Expression):
2364    arg_types = {"this": True, "expressions": False, "options": False}
2365
2366
2367class Tuple(Expression):
2368    arg_types = {"expressions": False}
2369
2370    def isin(
2371        self,
2372        *expressions: t.Any,
2373        query: t.Optional[ExpOrStr] = None,
2374        unnest: t.Optional[ExpOrStr] | t.Collection[ExpOrStr] = None,
2375        copy: bool = True,
2376        **opts,
2377    ) -> In:
2378        return In(
2379            this=maybe_copy(self, copy),
2380            expressions=[convert(e, copy=copy) for e in expressions],
2381            query=maybe_parse(query, copy=copy, **opts) if query else None,
2382            unnest=Unnest(
2383                expressions=[
2384                    maybe_parse(t.cast(ExpOrStr, e), copy=copy, **opts) for e in ensure_list(unnest)
2385                ]
2386            )
2387            if unnest
2388            else None,
2389        )
2390
2391
2392class Subqueryable(Unionable):
2393    def subquery(self, alias: t.Optional[ExpOrStr] = None, copy: bool = True) -> Subquery:
2394        """
2395        Convert this expression to an aliased expression that can be used as a Subquery.
2396
2397        Example:
2398            >>> subquery = Select().select("x").from_("tbl").subquery()
2399            >>> Select().select("x").from_(subquery).sql()
2400            'SELECT x FROM (SELECT x FROM tbl)'
2401
2402        Args:
2403            alias (str | Identifier): an optional alias for the subquery
2404            copy (bool): if `False`, modify this expression instance in-place.
2405
2406        Returns:
2407            Alias: the subquery
2408        """
2409        instance = maybe_copy(self, copy)
2410        if not isinstance(alias, Expression):
2411            alias = TableAlias(this=to_identifier(alias)) if alias else None
2412
2413        return Subquery(this=instance, alias=alias)
2414
2415    def limit(
2416        self, expression: ExpOrStr | int, dialect: DialectType = None, copy: bool = True, **opts
2417    ) -> Select:
2418        raise NotImplementedError
2419
2420    @property
2421    def ctes(self):
2422        with_ = self.args.get("with")
2423        if not with_:
2424            return []
2425        return with_.expressions
2426
2427    @property
2428    def selects(self) -> t.List[Expression]:
2429        raise NotImplementedError("Subqueryable objects must implement `selects`")
2430
2431    @property
2432    def named_selects(self) -> t.List[str]:
2433        raise NotImplementedError("Subqueryable objects must implement `named_selects`")
2434
2435    def select(
2436        self,
2437        *expressions: t.Optional[ExpOrStr],
2438        append: bool = True,
2439        dialect: DialectType = None,
2440        copy: bool = True,
2441        **opts,
2442    ) -> Subqueryable:
2443        raise NotImplementedError("Subqueryable objects must implement `select`")
2444
2445    def with_(
2446        self,
2447        alias: ExpOrStr,
2448        as_: ExpOrStr,
2449        recursive: t.Optional[bool] = None,
2450        append: bool = True,
2451        dialect: DialectType = None,
2452        copy: bool = True,
2453        **opts,
2454    ) -> Subqueryable:
2455        """
2456        Append to or set the common table expressions.
2457
2458        Example:
2459            >>> Select().with_("tbl2", as_="SELECT * FROM tbl").select("x").from_("tbl2").sql()
2460            'WITH tbl2 AS (SELECT * FROM tbl) SELECT x FROM tbl2'
2461
2462        Args:
2463            alias: the SQL code string to parse as the table name.
2464                If an `Expression` instance is passed, this is used as-is.
2465            as_: the SQL code string to parse as the table expression.
2466                If an `Expression` instance is passed, it will be used as-is.
2467            recursive: set the RECURSIVE part of the expression. Defaults to `False`.
2468            append: if `True`, add to any existing expressions.
2469                Otherwise, this resets the expressions.
2470            dialect: the dialect used to parse the input expression.
2471            copy: if `False`, modify this expression instance in-place.
2472            opts: other options to use to parse the input expressions.
2473
2474        Returns:
2475            The modified expression.
2476        """
2477        return _apply_cte_builder(
2478            self, alias, as_, recursive=recursive, append=append, dialect=dialect, copy=copy, **opts
2479        )
2480
2481
2482QUERY_MODIFIERS = {
2483    "match": False,
2484    "laterals": False,
2485    "joins": False,
2486    "connect": False,
2487    "pivots": False,
2488    "where": False,
2489    "group": False,
2490    "having": False,
2491    "qualify": False,
2492    "windows": False,
2493    "distribute": False,
2494    "sort": False,
2495    "cluster": False,
2496    "order": False,
2497    "limit": False,
2498    "offset": False,
2499    "locks": False,
2500    "sample": False,
2501    "settings": False,
2502    "format": False,
2503}
2504
2505
2506# https://learn.microsoft.com/en-us/sql/t-sql/queries/hints-transact-sql-table?view=sql-server-ver16
2507class WithTableHint(Expression):
2508    arg_types = {"expressions": True}
2509
2510
2511# https://dev.mysql.com/doc/refman/8.0/en/index-hints.html
2512class IndexTableHint(Expression):
2513    arg_types = {"this": True, "expressions": False, "target": False}
2514
2515
2516class Table(Expression):
2517    arg_types = {
2518        "this": True,
2519        "alias": False,
2520        "db": False,
2521        "catalog": False,
2522        "laterals": False,
2523        "joins": False,
2524        "pivots": False,
2525        "hints": False,
2526        "system_time": False,
2527        "version": False,
2528        "format": False,
2529        "pattern": False,
2530        "index": False,
2531        "ordinality": False,
2532    }
2533
2534    @property
2535    def name(self) -> str:
2536        if isinstance(self.this, Func):
2537            return ""
2538        return self.this.name
2539
2540    @property
2541    def db(self) -> str:
2542        return self.text("db")
2543
2544    @property
2545    def catalog(self) -> str:
2546        return self.text("catalog")
2547
2548    @property
2549    def selects(self) -> t.List[Expression]:
2550        return []
2551
2552    @property
2553    def named_selects(self) -> t.List[str]:
2554        return []
2555
2556    @property
2557    def parts(self) -> t.List[Expression]:
2558        """Return the parts of a table in order catalog, db, table."""
2559        parts: t.List[Expression] = []
2560
2561        for arg in ("catalog", "db", "this"):
2562            part = self.args.get(arg)
2563
2564            if isinstance(part, Dot):
2565                parts.extend(part.flatten())
2566            elif isinstance(part, Expression):
2567                parts.append(part)
2568
2569        return parts
2570
2571
2572class Union(Subqueryable):
2573    arg_types = {
2574        "with": False,
2575        "this": True,
2576        "expression": True,
2577        "distinct": False,
2578        "by_name": False,
2579        **QUERY_MODIFIERS,
2580    }
2581
2582    def limit(
2583        self, expression: ExpOrStr | int, dialect: DialectType = None, copy: bool = True, **opts
2584    ) -> Select:
2585        """
2586        Set the LIMIT expression.
2587
2588        Example:
2589            >>> select("1").union(select("1")).limit(1).sql()
2590            'SELECT * FROM (SELECT 1 UNION SELECT 1) AS _l_0 LIMIT 1'
2591
2592        Args:
2593            expression: the SQL code string to parse.
2594                This can also be an integer.
2595                If a `Limit` instance is passed, this is used as-is.
2596                If another `Expression` instance is passed, it will be wrapped in a `Limit`.
2597            dialect: the dialect used to parse the input expression.
2598            copy: if `False`, modify this expression instance in-place.
2599            opts: other options to use to parse the input expressions.
2600
2601        Returns:
2602            The limited subqueryable.
2603        """
2604        return (
2605            select("*")
2606            .from_(self.subquery(alias="_l_0", copy=copy))
2607            .limit(expression, dialect=dialect, copy=False, **opts)
2608        )
2609
2610    def select(
2611        self,
2612        *expressions: t.Optional[ExpOrStr],
2613        append: bool = True,
2614        dialect: DialectType = None,
2615        copy: bool = True,
2616        **opts,
2617    ) -> Union:
2618        """Append to or set the SELECT of the union recursively.
2619
2620        Example:
2621            >>> from sqlglot import parse_one
2622            >>> parse_one("select a from x union select a from y union select a from z").select("b").sql()
2623            'SELECT a, b FROM x UNION SELECT a, b FROM y UNION SELECT a, b FROM z'
2624
2625        Args:
2626            *expressions: the SQL code strings to parse.
2627                If an `Expression` instance is passed, it will be used as-is.
2628            append: if `True`, add to any existing expressions.
2629                Otherwise, this resets the expressions.
2630            dialect: the dialect used to parse the input expressions.
2631            copy: if `False`, modify this expression instance in-place.
2632            opts: other options to use to parse the input expressions.
2633
2634        Returns:
2635            Union: the modified expression.
2636        """
2637        this = self.copy() if copy else self
2638        this.this.unnest().select(*expressions, append=append, dialect=dialect, copy=False, **opts)
2639        this.expression.unnest().select(
2640            *expressions, append=append, dialect=dialect, copy=False, **opts
2641        )
2642        return this
2643
2644    @property
2645    def named_selects(self) -> t.List[str]:
2646        return self.this.unnest().named_selects
2647
2648    @property
2649    def is_star(self) -> bool:
2650        return self.this.is_star or self.expression.is_star
2651
2652    @property
2653    def selects(self) -> t.List[Expression]:
2654        return self.this.unnest().selects
2655
2656    @property
2657    def left(self) -> Expression:
2658        return self.this
2659
2660    @property
2661    def right(self) -> Expression:
2662        return self.expression
2663
2664
2665class Except(Union):
2666    pass
2667
2668
2669class Intersect(Union):
2670    pass
2671
2672
2673class Unnest(UDTF):
2674    arg_types = {
2675        "expressions": True,
2676        "alias": False,
2677        "offset": False,
2678    }
2679
2680
2681class Update(Expression):
2682    arg_types = {
2683        "with": False,
2684        "this": False,
2685        "expressions": True,
2686        "from": False,
2687        "where": False,
2688        "returning": False,
2689        "order": False,
2690        "limit": False,
2691    }
2692
2693
2694class Values(UDTF):
2695    arg_types = {"expressions": True, "alias": False}
2696
2697
2698class Var(Expression):
2699    pass
2700
2701
2702class Version(Expression):
2703    """
2704    Time travel, iceberg, bigquery etc
2705    https://trino.io/docs/current/connector/iceberg.html?highlight=snapshot#using-snapshots
2706    https://www.databricks.com/blog/2019/02/04/introducing-delta-time-travel-for-large-scale-data-lakes.html
2707    https://cloud.google.com/bigquery/docs/reference/standard-sql/query-syntax#for_system_time_as_of
2708    https://learn.microsoft.com/en-us/sql/relational-databases/tables/querying-data-in-a-system-versioned-temporal-table?view=sql-server-ver16
2709    this is either TIMESTAMP or VERSION
2710    kind is ("AS OF", "BETWEEN")
2711    """
2712
2713    arg_types = {"this": True, "kind": True, "expression": False}
2714
2715
2716class Schema(Expression):
2717    arg_types = {"this": False, "expressions": False}
2718
2719
2720# https://dev.mysql.com/doc/refman/8.0/en/select.html
2721# https://docs.oracle.com/en/database/oracle/oracle-database/19/sqlrf/SELECT.html
2722class Lock(Expression):
2723    arg_types = {"update": True, "expressions": False, "wait": False}
2724
2725
2726class Select(Subqueryable):
2727    arg_types = {
2728        "with": False,
2729        "kind": False,
2730        "expressions": False,
2731        "hint": False,
2732        "distinct": False,
2733        "into": False,
2734        "from": False,
2735        **QUERY_MODIFIERS,
2736    }
2737
2738    def from_(
2739        self, expression: ExpOrStr, dialect: DialectType = None, copy: bool = True, **opts
2740    ) -> Select:
2741        """
2742        Set the FROM expression.
2743
2744        Example:
2745            >>> Select().from_("tbl").select("x").sql()
2746            'SELECT x FROM tbl'
2747
2748        Args:
2749            expression : the SQL code strings to parse.
2750                If a `From` instance is passed, this is used as-is.
2751                If another `Expression` instance is passed, it will be wrapped in a `From`.
2752            dialect: the dialect used to parse the input expression.
2753            copy: if `False`, modify this expression instance in-place.
2754            opts: other options to use to parse the input expressions.
2755
2756        Returns:
2757            The modified Select expression.
2758        """
2759        return _apply_builder(
2760            expression=expression,
2761            instance=self,
2762            arg="from",
2763            into=From,
2764            prefix="FROM",
2765            dialect=dialect,
2766            copy=copy,
2767            **opts,
2768        )
2769
2770    def group_by(
2771        self,
2772        *expressions: t.Optional[ExpOrStr],
2773        append: bool = True,
2774        dialect: DialectType = None,
2775        copy: bool = True,
2776        **opts,
2777    ) -> Select:
2778        """
2779        Set the GROUP BY expression.
2780
2781        Example:
2782            >>> Select().from_("tbl").select("x", "COUNT(1)").group_by("x").sql()
2783            'SELECT x, COUNT(1) FROM tbl GROUP BY x'
2784
2785        Args:
2786            *expressions: the SQL code strings to parse.
2787                If a `Group` instance is passed, this is used as-is.
2788                If another `Expression` instance is passed, it will be wrapped in a `Group`.
2789                If nothing is passed in then a group by is not applied to the expression
2790            append: if `True`, add to any existing expressions.
2791                Otherwise, this flattens all the `Group` expression into a single expression.
2792            dialect: the dialect used to parse the input expression.
2793            copy: if `False`, modify this expression instance in-place.
2794            opts: other options to use to parse the input expressions.
2795
2796        Returns:
2797            The modified Select expression.
2798        """
2799        if not expressions:
2800            return self if not copy else self.copy()
2801
2802        return _apply_child_list_builder(
2803            *expressions,
2804            instance=self,
2805            arg="group",
2806            append=append,
2807            copy=copy,
2808            prefix="GROUP BY",
2809            into=Group,
2810            dialect=dialect,
2811            **opts,
2812        )
2813
2814    def order_by(
2815        self,
2816        *expressions: t.Optional[ExpOrStr],
2817        append: bool = True,
2818        dialect: DialectType = None,
2819        copy: bool = True,
2820        **opts,
2821    ) -> Select:
2822        """
2823        Set the ORDER BY expression.
2824
2825        Example:
2826            >>> Select().from_("tbl").select("x").order_by("x DESC").sql()
2827            'SELECT x FROM tbl ORDER BY x DESC'
2828
2829        Args:
2830            *expressions: the SQL code strings to parse.
2831                If a `Group` instance is passed, this is used as-is.
2832                If another `Expression` instance is passed, it will be wrapped in a `Order`.
2833            append: if `True`, add to any existing expressions.
2834                Otherwise, this flattens all the `Order` expression into a single expression.
2835            dialect: the dialect used to parse the input expression.
2836            copy: if `False`, modify this expression instance in-place.
2837            opts: other options to use to parse the input expressions.
2838
2839        Returns:
2840            The modified Select expression.
2841        """
2842        return _apply_child_list_builder(
2843            *expressions,
2844            instance=self,
2845            arg="order",
2846            append=append,
2847            copy=copy,
2848            prefix="ORDER BY",
2849            into=Order,
2850            dialect=dialect,
2851            **opts,
2852        )
2853
2854    def sort_by(
2855        self,
2856        *expressions: t.Optional[ExpOrStr],
2857        append: bool = True,
2858        dialect: DialectType = None,
2859        copy: bool = True,
2860        **opts,
2861    ) -> Select:
2862        """
2863        Set the SORT BY expression.
2864
2865        Example:
2866            >>> Select().from_("tbl").select("x").sort_by("x DESC").sql(dialect="hive")
2867            'SELECT x FROM tbl SORT BY x DESC'
2868
2869        Args:
2870            *expressions: the SQL code strings to parse.
2871                If a `Group` instance is passed, this is used as-is.
2872                If another `Expression` instance is passed, it will be wrapped in a `SORT`.
2873            append: if `True`, add to any existing expressions.
2874                Otherwise, this flattens all the `Order` expression into a single expression.
2875            dialect: the dialect used to parse the input expression.
2876            copy: if `False`, modify this expression instance in-place.
2877            opts: other options to use to parse the input expressions.
2878
2879        Returns:
2880            The modified Select expression.
2881        """
2882        return _apply_child_list_builder(
2883            *expressions,
2884            instance=self,
2885            arg="sort",
2886            append=append,
2887            copy=copy,
2888            prefix="SORT BY",
2889            into=Sort,
2890            dialect=dialect,
2891            **opts,
2892        )
2893
2894    def cluster_by(
2895        self,
2896        *expressions: t.Optional[ExpOrStr],
2897        append: bool = True,
2898        dialect: DialectType = None,
2899        copy: bool = True,
2900        **opts,
2901    ) -> Select:
2902        """
2903        Set the CLUSTER BY expression.
2904
2905        Example:
2906            >>> Select().from_("tbl").select("x").cluster_by("x DESC").sql(dialect="hive")
2907            'SELECT x FROM tbl CLUSTER BY x DESC'
2908
2909        Args:
2910            *expressions: the SQL code strings to parse.
2911                If a `Group` instance is passed, this is used as-is.
2912                If another `Expression` instance is passed, it will be wrapped in a `Cluster`.
2913            append: if `True`, add to any existing expressions.
2914                Otherwise, this flattens all the `Order` expression into a single expression.
2915            dialect: the dialect used to parse the input expression.
2916            copy: if `False`, modify this expression instance in-place.
2917            opts: other options to use to parse the input expressions.
2918
2919        Returns:
2920            The modified Select expression.
2921        """
2922        return _apply_child_list_builder(
2923            *expressions,
2924            instance=self,
2925            arg="cluster",
2926            append=append,
2927            copy=copy,
2928            prefix="CLUSTER BY",
2929            into=Cluster,
2930            dialect=dialect,
2931            **opts,
2932        )
2933
2934    def limit(
2935        self, expression: ExpOrStr | int, dialect: DialectType = None, copy: bool = True, **opts
2936    ) -> Select:
2937        """
2938        Set the LIMIT expression.
2939
2940        Example:
2941            >>> Select().from_("tbl").select("x").limit(10).sql()
2942            'SELECT x FROM tbl LIMIT 10'
2943
2944        Args:
2945            expression: the SQL code string to parse.
2946                This can also be an integer.
2947                If a `Limit` instance is passed, this is used as-is.
2948                If another `Expression` instance is passed, it will be wrapped in a `Limit`.
2949            dialect: the dialect used to parse the input expression.
2950            copy: if `False`, modify this expression instance in-place.
2951            opts: other options to use to parse the input expressions.
2952
2953        Returns:
2954            Select: the modified expression.
2955        """
2956        return _apply_builder(
2957            expression=expression,
2958            instance=self,
2959            arg="limit",
2960            into=Limit,
2961            prefix="LIMIT",
2962            dialect=dialect,
2963            copy=copy,
2964            into_arg="expression",
2965            **opts,
2966        )
2967
2968    def offset(
2969        self, expression: ExpOrStr | int, dialect: DialectType = None, copy: bool = True, **opts
2970    ) -> Select:
2971        """
2972        Set the OFFSET expression.
2973
2974        Example:
2975            >>> Select().from_("tbl").select("x").offset(10).sql()
2976            'SELECT x FROM tbl OFFSET 10'
2977
2978        Args:
2979            expression: the SQL code string to parse.
2980                This can also be an integer.
2981                If a `Offset` instance is passed, this is used as-is.
2982                If another `Expression` instance is passed, it will be wrapped in a `Offset`.
2983            dialect: the dialect used to parse the input expression.
2984            copy: if `False`, modify this expression instance in-place.
2985            opts: other options to use to parse the input expressions.
2986
2987        Returns:
2988            The modified Select expression.
2989        """
2990        return _apply_builder(
2991            expression=expression,
2992            instance=self,
2993            arg="offset",
2994            into=Offset,
2995            prefix="OFFSET",
2996            dialect=dialect,
2997            copy=copy,
2998            into_arg="expression",
2999            **opts,
3000        )
3001
3002    def select(
3003        self,
3004        *expressions: t.Optional[ExpOrStr],
3005        append: bool = True,
3006        dialect: DialectType = None,
3007        copy: bool = True,
3008        **opts,
3009    ) -> Select:
3010        """
3011        Append to or set the SELECT expressions.
3012
3013        Example:
3014            >>> Select().select("x", "y").sql()
3015            'SELECT x, y'
3016
3017        Args:
3018            *expressions: the SQL code strings to parse.
3019                If an `Expression` instance is passed, it will be used as-is.
3020            append: if `True`, add to any existing expressions.
3021                Otherwise, this resets the expressions.
3022            dialect: the dialect used to parse the input expressions.
3023            copy: if `False`, modify this expression instance in-place.
3024            opts: other options to use to parse the input expressions.
3025
3026        Returns:
3027            The modified Select expression.
3028        """
3029        return _apply_list_builder(
3030            *expressions,
3031            instance=self,
3032            arg="expressions",
3033            append=append,
3034            dialect=dialect,
3035            copy=copy,
3036            **opts,
3037        )
3038
3039    def lateral(
3040        self,
3041        *expressions: t.Optional[ExpOrStr],
3042        append: bool = True,
3043        dialect: DialectType = None,
3044        copy: bool = True,
3045        **opts,
3046    ) -> Select:
3047        """
3048        Append to or set the LATERAL expressions.
3049
3050        Example:
3051            >>> Select().select("x").lateral("OUTER explode(y) tbl2 AS z").from_("tbl").sql()
3052            'SELECT x FROM tbl LATERAL VIEW OUTER EXPLODE(y) tbl2 AS z'
3053
3054        Args:
3055            *expressions: the SQL code strings to parse.
3056                If an `Expression` instance is passed, it will be used as-is.
3057            append: if `True`, add to any existing expressions.
3058                Otherwise, this resets the expressions.
3059            dialect: the dialect used to parse the input expressions.
3060            copy: if `False`, modify this expression instance in-place.
3061            opts: other options to use to parse the input expressions.
3062
3063        Returns:
3064            The modified Select expression.
3065        """
3066        return _apply_list_builder(
3067            *expressions,
3068            instance=self,
3069            arg="laterals",
3070            append=append,
3071            into=Lateral,
3072            prefix="LATERAL VIEW",
3073            dialect=dialect,
3074            copy=copy,
3075            **opts,
3076        )
3077
3078    def join(
3079        self,
3080        expression: ExpOrStr,
3081        on: t.Optional[ExpOrStr] = None,
3082        using: t.Optional[ExpOrStr | t.Collection[ExpOrStr]] = None,
3083        append: bool = True,
3084        join_type: t.Optional[str] = None,
3085        join_alias: t.Optional[Identifier | str] = None,
3086        dialect: DialectType = None,
3087        copy: bool = True,
3088        **opts,
3089    ) -> Select:
3090        """
3091        Append to or set the JOIN expressions.
3092
3093        Example:
3094            >>> Select().select("*").from_("tbl").join("tbl2", on="tbl1.y = tbl2.y").sql()
3095            'SELECT * FROM tbl JOIN tbl2 ON tbl1.y = tbl2.y'
3096
3097            >>> Select().select("1").from_("a").join("b", using=["x", "y", "z"]).sql()
3098            'SELECT 1 FROM a JOIN b USING (x, y, z)'
3099
3100            Use `join_type` to change the type of join:
3101
3102            >>> Select().select("*").from_("tbl").join("tbl2", on="tbl1.y = tbl2.y", join_type="left outer").sql()
3103            'SELECT * FROM tbl LEFT OUTER JOIN tbl2 ON tbl1.y = tbl2.y'
3104
3105        Args:
3106            expression: the SQL code string to parse.
3107                If an `Expression` instance is passed, it will be used as-is.
3108            on: optionally specify the join "on" criteria as a SQL string.
3109                If an `Expression` instance is passed, it will be used as-is.
3110            using: optionally specify the join "using" criteria as a SQL string.
3111                If an `Expression` instance is passed, it will be used as-is.
3112            append: if `True`, add to any existing expressions.
3113                Otherwise, this resets the expressions.
3114            join_type: if set, alter the parsed join type.
3115            join_alias: an optional alias for the joined source.
3116            dialect: the dialect used to parse the input expressions.
3117            copy: if `False`, modify this expression instance in-place.
3118            opts: other options to use to parse the input expressions.
3119
3120        Returns:
3121            Select: the modified expression.
3122        """
3123        parse_args: t.Dict[str, t.Any] = {"dialect": dialect, **opts}
3124
3125        try:
3126            expression = maybe_parse(expression, into=Join, prefix="JOIN", **parse_args)
3127        except ParseError:
3128            expression = maybe_parse(expression, into=(Join, Expression), **parse_args)
3129
3130        join = expression if isinstance(expression, Join) else Join(this=expression)
3131
3132        if isinstance(join.this, Select):
3133            join.this.replace(join.this.subquery())
3134
3135        if join_type:
3136            method: t.Optional[Token]
3137            side: t.Optional[Token]
3138            kind: t.Optional[Token]
3139
3140            method, side, kind = maybe_parse(join_type, into="JOIN_TYPE", **parse_args)  # type: ignore
3141
3142            if method:
3143                join.set("method", method.text)
3144            if side:
3145                join.set("side", side.text)
3146            if kind:
3147                join.set("kind", kind.text)
3148
3149        if on:
3150            on = and_(*ensure_list(on), dialect=dialect, copy=copy, **opts)
3151            join.set("on", on)
3152
3153        if using:
3154            join = _apply_list_builder(
3155                *ensure_list(using),
3156                instance=join,
3157                arg="using",
3158                append=append,
3159                copy=copy,
3160                into=Identifier,
3161                **opts,
3162            )
3163
3164        if join_alias:
3165            join.set("this", alias_(join.this, join_alias, table=True))
3166
3167        return _apply_list_builder(
3168            join,
3169            instance=self,
3170            arg="joins",
3171            append=append,
3172            copy=copy,
3173            **opts,
3174        )
3175
3176    def where(
3177        self,
3178        *expressions: t.Optional[ExpOrStr],
3179        append: bool = True,
3180        dialect: DialectType = None,
3181        copy: bool = True,
3182        **opts,
3183    ) -> Select:
3184        """
3185        Append to or set the WHERE expressions.
3186
3187        Example:
3188            >>> Select().select("x").from_("tbl").where("x = 'a' OR x < 'b'").sql()
3189            "SELECT x FROM tbl WHERE x = 'a' OR x < 'b'"
3190
3191        Args:
3192            *expressions: the SQL code strings to parse.
3193                If an `Expression` instance is passed, it will be used as-is.
3194                Multiple expressions are combined with an AND operator.
3195            append: if `True`, AND the new expressions to any existing expression.
3196                Otherwise, this resets the expression.
3197            dialect: the dialect used to parse the input expressions.
3198            copy: if `False`, modify this expression instance in-place.
3199            opts: other options to use to parse the input expressions.
3200
3201        Returns:
3202            Select: the modified expression.
3203        """
3204        return _apply_conjunction_builder(
3205            *expressions,
3206            instance=self,
3207            arg="where",
3208            append=append,
3209            into=Where,
3210            dialect=dialect,
3211            copy=copy,
3212            **opts,
3213        )
3214
3215    def having(
3216        self,
3217        *expressions: t.Optional[ExpOrStr],
3218        append: bool = True,
3219        dialect: DialectType = None,
3220        copy: bool = True,
3221        **opts,
3222    ) -> Select:
3223        """
3224        Append to or set the HAVING expressions.
3225
3226        Example:
3227            >>> Select().select("x", "COUNT(y)").from_("tbl").group_by("x").having("COUNT(y) > 3").sql()
3228            'SELECT x, COUNT(y) FROM tbl GROUP BY x HAVING COUNT(y) > 3'
3229
3230        Args:
3231            *expressions: the SQL code strings to parse.
3232                If an `Expression` instance is passed, it will be used as-is.
3233                Multiple expressions are combined with an AND operator.
3234            append: if `True`, AND the new expressions to any existing expression.
3235                Otherwise, this resets the expression.
3236            dialect: the dialect used to parse the input expressions.
3237            copy: if `False`, modify this expression instance in-place.
3238            opts: other options to use to parse the input expressions.
3239
3240        Returns:
3241            The modified Select expression.
3242        """
3243        return _apply_conjunction_builder(
3244            *expressions,
3245            instance=self,
3246            arg="having",
3247            append=append,
3248            into=Having,
3249            dialect=dialect,
3250            copy=copy,
3251            **opts,
3252        )
3253
3254    def window(
3255        self,
3256        *expressions: t.Optional[ExpOrStr],
3257        append: bool = True,
3258        dialect: DialectType = None,
3259        copy: bool = True,
3260        **opts,
3261    ) -> Select:
3262        return _apply_list_builder(
3263            *expressions,
3264            instance=self,
3265            arg="windows",
3266            append=append,
3267            into=Window,
3268            dialect=dialect,
3269            copy=copy,
3270            **opts,
3271        )
3272
3273    def qualify(
3274        self,
3275        *expressions: t.Optional[ExpOrStr],
3276        append: bool = True,
3277        dialect: DialectType = None,
3278        copy: bool = True,
3279        **opts,
3280    ) -> Select:
3281        return _apply_conjunction_builder(
3282            *expressions,
3283            instance=self,
3284            arg="qualify",
3285            append=append,
3286            into=Qualify,
3287            dialect=dialect,
3288            copy=copy,
3289            **opts,
3290        )
3291
3292    def distinct(
3293        self, *ons: t.Optional[ExpOrStr], distinct: bool = True, copy: bool = True
3294    ) -> Select:
3295        """
3296        Set the OFFSET expression.
3297
3298        Example:
3299            >>> Select().from_("tbl").select("x").distinct().sql()
3300            'SELECT DISTINCT x FROM tbl'
3301
3302        Args:
3303            ons: the expressions to distinct on
3304            distinct: whether the Select should be distinct
3305            copy: if `False`, modify this expression instance in-place.
3306
3307        Returns:
3308            Select: the modified expression.
3309        """
3310        instance = maybe_copy(self, copy)
3311        on = Tuple(expressions=[maybe_parse(on, copy=copy) for on in ons if on]) if ons else None
3312        instance.set("distinct", Distinct(on=on) if distinct else None)
3313        return instance
3314
3315    def ctas(
3316        self,
3317        table: ExpOrStr,
3318        properties: t.Optional[t.Dict] = None,
3319        dialect: DialectType = None,
3320        copy: bool = True,
3321        **opts,
3322    ) -> Create:
3323        """
3324        Convert this expression to a CREATE TABLE AS statement.
3325
3326        Example:
3327            >>> Select().select("*").from_("tbl").ctas("x").sql()
3328            'CREATE TABLE x AS SELECT * FROM tbl'
3329
3330        Args:
3331            table: the SQL code string to parse as the table name.
3332                If another `Expression` instance is passed, it will be used as-is.
3333            properties: an optional mapping of table properties
3334            dialect: the dialect used to parse the input table.
3335            copy: if `False`, modify this expression instance in-place.
3336            opts: other options to use to parse the input table.
3337
3338        Returns:
3339            The new Create expression.
3340        """
3341        instance = maybe_copy(self, copy)
3342        table_expression = maybe_parse(
3343            table,
3344            into=Table,
3345            dialect=dialect,
3346            **opts,
3347        )
3348        properties_expression = None
3349        if properties:
3350            properties_expression = Properties.from_dict(properties)
3351
3352        return Create(
3353            this=table_expression,
3354            kind="table",
3355            expression=instance,
3356            properties=properties_expression,
3357        )
3358
3359    def lock(self, update: bool = True, copy: bool = True) -> Select:
3360        """
3361        Set the locking read mode for this expression.
3362
3363        Examples:
3364            >>> Select().select("x").from_("tbl").where("x = 'a'").lock().sql("mysql")
3365            "SELECT x FROM tbl WHERE x = 'a' FOR UPDATE"
3366
3367            >>> Select().select("x").from_("tbl").where("x = 'a'").lock(update=False).sql("mysql")
3368            "SELECT x FROM tbl WHERE x = 'a' FOR SHARE"
3369
3370        Args:
3371            update: if `True`, the locking type will be `FOR UPDATE`, else it will be `FOR SHARE`.
3372            copy: if `False`, modify this expression instance in-place.
3373
3374        Returns:
3375            The modified expression.
3376        """
3377        inst = maybe_copy(self, copy)
3378        inst.set("locks", [Lock(update=update)])
3379
3380        return inst
3381
3382    def hint(self, *hints: ExpOrStr, dialect: DialectType = None, copy: bool = True) -> Select:
3383        """
3384        Set hints for this expression.
3385
3386        Examples:
3387            >>> Select().select("x").from_("tbl").hint("BROADCAST(y)").sql(dialect="spark")
3388            'SELECT /*+ BROADCAST(y) */ x FROM tbl'
3389
3390        Args:
3391            hints: The SQL code strings to parse as the hints.
3392                If an `Expression` instance is passed, it will be used as-is.
3393            dialect: The dialect used to parse the hints.
3394            copy: If `False`, modify this expression instance in-place.
3395
3396        Returns:
3397            The modified expression.
3398        """
3399        inst = maybe_copy(self, copy)
3400        inst.set(
3401            "hint", Hint(expressions=[maybe_parse(h, copy=copy, dialect=dialect) for h in hints])
3402        )
3403
3404        return inst
3405
3406    @property
3407    def named_selects(self) -> t.List[str]:
3408        return [e.output_name for e in self.expressions if e.alias_or_name]
3409
3410    @property
3411    def is_star(self) -> bool:
3412        return any(expression.is_star for expression in self.expressions)
3413
3414    @property
3415    def selects(self) -> t.List[Expression]:
3416        return self.expressions
3417
3418
3419class Subquery(DerivedTable, Unionable):
3420    arg_types = {
3421        "this": True,
3422        "alias": False,
3423        "with": False,
3424        **QUERY_MODIFIERS,
3425    }
3426
3427    def unnest(self):
3428        """
3429        Returns the first non subquery.
3430        """
3431        expression = self
3432        while isinstance(expression, Subquery):
3433            expression = expression.this
3434        return expression
3435
3436    def unwrap(self) -> Subquery:
3437        expression = self
3438        while expression.same_parent and expression.is_wrapper:
3439            expression = t.cast(Subquery, expression.parent)
3440        return expression
3441
3442    @property
3443    def is_wrapper(self) -> bool:
3444        """
3445        Whether this Subquery acts as a simple wrapper around another expression.
3446
3447        SELECT * FROM (((SELECT * FROM t)))
3448                      ^
3449                      This corresponds to a "wrapper" Subquery node
3450        """
3451        return all(v is None for k, v in self.args.items() if k != "this")
3452
3453    @property
3454    def is_star(self) -> bool:
3455        return self.this.is_star
3456
3457    @property
3458    def output_name(self) -> str:
3459        return self.alias
3460
3461
3462class TableSample(Expression):
3463    arg_types = {
3464        "this": False,
3465        "expressions": False,
3466        "method": False,
3467        "bucket_numerator": False,
3468        "bucket_denominator": False,
3469        "bucket_field": False,
3470        "percent": False,
3471        "rows": False,
3472        "size": False,
3473        "seed": False,
3474        "kind": False,
3475    }
3476
3477
3478class Tag(Expression):
3479    """Tags are used for generating arbitrary sql like SELECT <span>x</span>."""
3480
3481    arg_types = {
3482        "this": False,
3483        "prefix": False,
3484        "postfix": False,
3485    }
3486
3487
3488# Represents both the standard SQL PIVOT operator and DuckDB's "simplified" PIVOT syntax
3489# https://duckdb.org/docs/sql/statements/pivot
3490class Pivot(Expression):
3491    arg_types = {
3492        "this": False,
3493        "alias": False,
3494        "expressions": False,
3495        "field": False,
3496        "unpivot": False,
3497        "using": False,
3498        "group": False,
3499        "columns": False,
3500        "include_nulls": False,
3501    }
3502
3503
3504class Window(Condition):
3505    arg_types = {
3506        "this": True,
3507        "partition_by": False,
3508        "order": False,
3509        "spec": False,
3510        "alias": False,
3511        "over": False,
3512        "first": False,
3513    }
3514
3515
3516class WindowSpec(Expression):
3517    arg_types = {
3518        "kind": False,
3519        "start": False,
3520        "start_side": False,
3521        "end": False,
3522        "end_side": False,
3523    }
3524
3525
3526class Where(Expression):
3527    pass
3528
3529
3530class Star(Expression):
3531    arg_types = {"except": False, "replace": False}
3532
3533    @property
3534    def name(self) -> str:
3535        return "*"
3536
3537    @property
3538    def output_name(self) -> str:
3539        return self.name
3540
3541
3542class Parameter(Condition):
3543    arg_types = {"this": True, "expression": False}
3544
3545
3546class SessionParameter(Condition):
3547    arg_types = {"this": True, "kind": False}
3548
3549
3550class Placeholder(Condition):
3551    arg_types = {"this": False, "kind": False}
3552
3553
3554class Null(Condition):
3555    arg_types: t.Dict[str, t.Any] = {}
3556
3557    @property
3558    def name(self) -> str:
3559        return "NULL"
3560
3561
3562class Boolean(Condition):
3563    pass
3564
3565
3566class DataTypeParam(Expression):
3567    arg_types = {"this": True, "expression": False}
3568
3569
3570class DataType(Expression):
3571    arg_types = {
3572        "this": True,
3573        "expressions": False,
3574        "nested": False,
3575        "values": False,
3576        "prefix": False,
3577        "kind": False,
3578    }
3579
3580    class Type(AutoName):
3581        ARRAY = auto()
3582        BIGDECIMAL = auto()
3583        BIGINT = auto()
3584        BIGSERIAL = auto()
3585        BINARY = auto()
3586        BIT = auto()
3587        BOOLEAN = auto()
3588        CHAR = auto()
3589        DATE = auto()
3590        DATEMULTIRANGE = auto()
3591        DATERANGE = auto()
3592        DATETIME = auto()
3593        DATETIME64 = auto()
3594        DECIMAL = auto()
3595        DOUBLE = auto()
3596        ENUM = auto()
3597        ENUM8 = auto()
3598        ENUM16 = auto()
3599        FIXEDSTRING = auto()
3600        FLOAT = auto()
3601        GEOGRAPHY = auto()
3602        GEOMETRY = auto()
3603        HLLSKETCH = auto()
3604        HSTORE = auto()
3605        IMAGE = auto()
3606        INET = auto()
3607        INT = auto()
3608        INT128 = auto()
3609        INT256 = auto()
3610        INT4MULTIRANGE = auto()
3611        INT4RANGE = auto()
3612        INT8MULTIRANGE = auto()
3613        INT8RANGE = auto()
3614        INTERVAL = auto()
3615        IPADDRESS = auto()
3616        IPPREFIX = auto()
3617        JSON = auto()
3618        JSONB = auto()
3619        LONGBLOB = auto()
3620        LONGTEXT = auto()
3621        LOWCARDINALITY = auto()
3622        MAP = auto()
3623        MEDIUMBLOB = auto()
3624        MEDIUMINT = auto()
3625        MEDIUMTEXT = auto()
3626        MONEY = auto()
3627        NCHAR = auto()
3628        NESTED = auto()
3629        NULL = auto()
3630        NULLABLE = auto()
3631        NUMMULTIRANGE = auto()
3632        NUMRANGE = auto()
3633        NVARCHAR = auto()
3634        OBJECT = auto()
3635        ROWVERSION = auto()
3636        SERIAL = auto()
3637        SET = auto()
3638        SMALLINT = auto()
3639        SMALLMONEY = auto()
3640        SMALLSERIAL = auto()
3641        STRUCT = auto()
3642        SUPER = auto()
3643        TEXT = auto()
3644        TINYBLOB = auto()
3645        TINYTEXT = auto()
3646        TIME = auto()
3647        TIMETZ = auto()
3648        TIMESTAMP = auto()
3649        TIMESTAMPLTZ = auto()
3650        TIMESTAMPTZ = auto()
3651        TIMESTAMP_S = auto()
3652        TIMESTAMP_MS = auto()
3653        TIMESTAMP_NS = auto()
3654        TINYINT = auto()
3655        TSMULTIRANGE = auto()
3656        TSRANGE = auto()
3657        TSTZMULTIRANGE = auto()
3658        TSTZRANGE = auto()
3659        UBIGINT = auto()
3660        UINT = auto()
3661        UINT128 = auto()
3662        UINT256 = auto()
3663        UMEDIUMINT = auto()
3664        UDECIMAL = auto()
3665        UNIQUEIDENTIFIER = auto()
3666        UNKNOWN = auto()  # Sentinel value, useful for type annotation
3667        USERDEFINED = "USER-DEFINED"
3668        USMALLINT = auto()
3669        UTINYINT = auto()
3670        UUID = auto()
3671        VARBINARY = auto()
3672        VARCHAR = auto()
3673        VARIANT = auto()
3674        XML = auto()
3675        YEAR = auto()
3676
3677    TEXT_TYPES = {
3678        Type.CHAR,
3679        Type.NCHAR,
3680        Type.VARCHAR,
3681        Type.NVARCHAR,
3682        Type.TEXT,
3683    }
3684
3685    INTEGER_TYPES = {
3686        Type.INT,
3687        Type.TINYINT,
3688        Type.SMALLINT,
3689        Type.BIGINT,
3690        Type.INT128,
3691        Type.INT256,
3692        Type.BIT,
3693    }
3694
3695    FLOAT_TYPES = {
3696        Type.FLOAT,
3697        Type.DOUBLE,
3698    }
3699
3700    NUMERIC_TYPES = {
3701        *INTEGER_TYPES,
3702        *FLOAT_TYPES,
3703    }
3704
3705    TEMPORAL_TYPES = {
3706        Type.TIME,
3707        Type.TIMETZ,
3708        Type.TIMESTAMP,
3709        Type.TIMESTAMPTZ,
3710        Type.TIMESTAMPLTZ,
3711        Type.TIMESTAMP_S,
3712        Type.TIMESTAMP_MS,
3713        Type.TIMESTAMP_NS,
3714        Type.DATE,
3715        Type.DATETIME,
3716        Type.DATETIME64,
3717    }
3718
3719    @classmethod
3720    def build(
3721        cls,
3722        dtype: str | DataType | DataType.Type,
3723        dialect: DialectType = None,
3724        udt: bool = False,
3725        **kwargs,
3726    ) -> DataType:
3727        """
3728        Constructs a DataType object.
3729
3730        Args:
3731            dtype: the data type of interest.
3732            dialect: the dialect to use for parsing `dtype`, in case it's a string.
3733            udt: when set to True, `dtype` will be used as-is if it can't be parsed into a
3734                DataType, thus creating a user-defined type.
3735            kawrgs: additional arguments to pass in the constructor of DataType.
3736
3737        Returns:
3738            The constructed DataType object.
3739        """
3740        from sqlglot import parse_one
3741
3742        if isinstance(dtype, str):
3743            if dtype.upper() == "UNKNOWN":
3744                return DataType(this=DataType.Type.UNKNOWN, **kwargs)
3745
3746            try:
3747                data_type_exp = parse_one(
3748                    dtype, read=dialect, into=DataType, error_level=ErrorLevel.IGNORE
3749                )
3750            except ParseError:
3751                if udt:
3752                    return DataType(this=DataType.Type.USERDEFINED, kind=dtype, **kwargs)
3753                raise
3754        elif isinstance(dtype, DataType.Type):
3755            data_type_exp = DataType(this=dtype)
3756        elif isinstance(dtype, DataType):
3757            return dtype
3758        else:
3759            raise ValueError(f"Invalid data type: {type(dtype)}. Expected str or DataType.Type")
3760
3761        return DataType(**{**data_type_exp.args, **kwargs})
3762
3763    def is_type(self, *dtypes: str | DataType | DataType.Type) -> bool:
3764        """
3765        Checks whether this DataType matches one of the provided data types. Nested types or precision
3766        will be compared using "structural equivalence" semantics, so e.g. array<int> != array<float>.
3767
3768        Args:
3769            dtypes: the data types to compare this DataType to.
3770
3771        Returns:
3772            True, if and only if there is a type in `dtypes` which is equal to this DataType.
3773        """
3774        for dtype in dtypes:
3775            other = DataType.build(dtype, udt=True)
3776
3777            if (
3778                other.expressions
3779                or self.this == DataType.Type.USERDEFINED
3780                or other.this == DataType.Type.USERDEFINED
3781            ):
3782                matches = self == other
3783            else:
3784                matches = self.this == other.this
3785
3786            if matches:
3787                return True
3788        return False
3789
3790
3791# https://www.postgresql.org/docs/15/datatype-pseudo.html
3792class PseudoType(DataType):
3793    arg_types = {"this": True}
3794
3795
3796# https://www.postgresql.org/docs/15/datatype-oid.html
3797class ObjectIdentifier(DataType):
3798    arg_types = {"this": True}
3799
3800
3801# WHERE x <OP> EXISTS|ALL|ANY|SOME(SELECT ...)
3802class SubqueryPredicate(Predicate):
3803    pass
3804
3805
3806class All(SubqueryPredicate):
3807    pass
3808
3809
3810class Any(SubqueryPredicate):
3811    pass
3812
3813
3814class Exists(SubqueryPredicate):
3815    pass
3816
3817
3818# Commands to interact with the databases or engines. For most of the command
3819# expressions we parse whatever comes after the command's name as a string.
3820class Command(Expression):
3821    arg_types = {"this": True, "expression": False}
3822
3823
3824class Transaction(Expression):
3825    arg_types = {"this": False, "modes": False, "mark": False}
3826
3827
3828class Commit(Expression):
3829    arg_types = {"chain": False, "this": False, "durability": False}
3830
3831
3832class Rollback(Expression):
3833    arg_types = {"savepoint": False, "this": False}
3834
3835
3836class AlterTable(Expression):
3837    arg_types = {"this": True, "actions": True, "exists": False, "only": False}
3838
3839
3840class AddConstraint(Expression):
3841    arg_types = {"this": False, "expression": False, "enforced": False}
3842
3843
3844class DropPartition(Expression):
3845    arg_types = {"expressions": True, "exists": False}
3846
3847
3848# Binary expressions like (ADD a b)
3849class Binary(Condition):
3850    arg_types = {"this": True, "expression": True}
3851
3852    @property
3853    def left(self) -> Expression:
3854        return self.this
3855
3856    @property
3857    def right(self) -> Expression:
3858        return self.expression
3859
3860
3861class Add(Binary):
3862    pass
3863
3864
3865class Connector(Binary):
3866    pass
3867
3868
3869class And(Connector):
3870    pass
3871
3872
3873class Or(Connector):
3874    pass
3875
3876
3877class BitwiseAnd(Binary):
3878    pass
3879
3880
3881class BitwiseLeftShift(Binary):
3882    pass
3883
3884
3885class BitwiseOr(Binary):
3886    pass
3887
3888
3889class BitwiseRightShift(Binary):
3890    pass
3891
3892
3893class BitwiseXor(Binary):
3894    pass
3895
3896
3897class Div(Binary):
3898    arg_types = {"this": True, "expression": True, "typed": False, "safe": False}
3899
3900
3901class Overlaps(Binary):
3902    pass
3903
3904
3905class Dot(Binary):
3906    @property
3907    def name(self) -> str:
3908        return self.expression.name
3909
3910    @property
3911    def output_name(self) -> str:
3912        return self.name
3913
3914    @classmethod
3915    def build(self, expressions: t.Sequence[Expression]) -> Dot:
3916        """Build a Dot object with a sequence of expressions."""
3917        if len(expressions) < 2:
3918            raise ValueError(f"Dot requires >= 2 expressions.")
3919
3920        return t.cast(Dot, reduce(lambda x, y: Dot(this=x, expression=y), expressions))
3921
3922
3923class DPipe(Binary):
3924    pass
3925
3926
3927class SafeDPipe(DPipe):
3928    pass
3929
3930
3931class EQ(Binary, Predicate):
3932    pass
3933
3934
3935class NullSafeEQ(Binary, Predicate):
3936    pass
3937
3938
3939class NullSafeNEQ(Binary, Predicate):
3940    pass
3941
3942
3943# Represents e.g. := in DuckDB which is mostly used for setting parameters
3944class PropertyEQ(Binary):
3945    pass
3946
3947
3948class Distance(Binary):
3949    pass
3950
3951
3952class Escape(Binary):
3953    pass
3954
3955
3956class Glob(Binary, Predicate):
3957    pass
3958
3959
3960class GT(Binary, Predicate):
3961    pass
3962
3963
3964class GTE(Binary, Predicate):
3965    pass
3966
3967
3968class ILike(Binary, Predicate):
3969    pass
3970
3971
3972class ILikeAny(Binary, Predicate):
3973    pass
3974
3975
3976class IntDiv(Binary):
3977    pass
3978
3979
3980class Is(Binary, Predicate):
3981    pass
3982
3983
3984class Kwarg(Binary):
3985    """Kwarg in special functions like func(kwarg => y)."""
3986
3987
3988class Like(Binary, Predicate):
3989    pass
3990
3991
3992class LikeAny(Binary, Predicate):
3993    pass
3994
3995
3996class LT(Binary, Predicate):
3997    pass
3998
3999
4000class LTE(Binary, Predicate):
4001    pass
4002
4003
4004class Mod(Binary):
4005    pass
4006
4007
4008class Mul(Binary):
4009    pass
4010
4011
4012class NEQ(Binary, Predicate):
4013    pass
4014
4015
4016class SimilarTo(Binary, Predicate):
4017    pass
4018
4019
4020class Slice(Binary):
4021    arg_types = {"this": False, "expression": False}
4022
4023
4024class Sub(Binary):
4025    pass
4026
4027
4028class ArrayOverlaps(Binary):
4029    pass
4030
4031
4032# Unary Expressions
4033# (NOT a)
4034class Unary(Condition):
4035    pass
4036
4037
4038class BitwiseNot(Unary):
4039    pass
4040
4041
4042class Not(Unary):
4043    pass
4044
4045
4046class Paren(Unary):
4047    arg_types = {"this": True, "with": False}
4048
4049    @property
4050    def output_name(self) -> str:
4051        return self.this.name
4052
4053
4054class Neg(Unary):
4055    pass
4056
4057
4058class Alias(Expression):
4059    arg_types = {"this": True, "alias": False}
4060
4061    @property
4062    def output_name(self) -> str:
4063        return self.alias
4064
4065
4066class Aliases(Expression):
4067    arg_types = {"this": True, "expressions": True}
4068
4069    @property
4070    def aliases(self):
4071        return self.expressions
4072
4073
4074class AtTimeZone(Expression):
4075    arg_types = {"this": True, "zone": True}
4076
4077
4078class Between(Predicate):
4079    arg_types = {"this": True, "low": True, "high": True}
4080
4081
4082class Bracket(Condition):
4083    arg_types = {"this": True, "expressions": True}
4084
4085    @property
4086    def output_name(self) -> str:
4087        if len(self.expressions) == 1:
4088            return self.expressions[0].output_name
4089
4090        return super().output_name
4091
4092
4093class SafeBracket(Bracket):
4094    """Represents array lookup where OOB index yields NULL instead of causing a failure."""
4095
4096
4097class Distinct(Expression):
4098    arg_types = {"expressions": False, "on": False}
4099
4100
4101class In(Predicate):
4102    arg_types = {
4103        "this": True,
4104        "expressions": False,
4105        "query": False,
4106        "unnest": False,
4107        "field": False,
4108        "is_global": False,
4109    }
4110
4111
4112# https://cloud.google.com/bigquery/docs/reference/standard-sql/procedural-language#for-in
4113class ForIn(Expression):
4114    arg_types = {"this": True, "expression": True}
4115
4116
4117class TimeUnit(Expression):
4118    """Automatically converts unit arg into a var."""
4119
4120    arg_types = {"unit": False}
4121
4122    UNABBREVIATED_UNIT_NAME = {
4123        "d": "day",
4124        "h": "hour",
4125        "m": "minute",
4126        "ms": "millisecond",
4127        "ns": "nanosecond",
4128        "q": "quarter",
4129        "s": "second",
4130        "us": "microsecond",
4131        "w": "week",
4132        "y": "year",
4133    }
4134
4135    VAR_LIKE = (Column, Literal, Var)
4136
4137    def __init__(self, **args):
4138        unit = args.get("unit")
4139        if isinstance(unit, self.VAR_LIKE):
4140            args["unit"] = Var(this=self.UNABBREVIATED_UNIT_NAME.get(unit.name) or unit.name)
4141        elif isinstance(unit, Week):
4142            unit.set("this", Var(this=unit.this.name))
4143
4144        super().__init__(**args)
4145
4146    @property
4147    def unit(self) -> t.Optional[Var]:
4148        return self.args.get("unit")
4149
4150
4151class IntervalOp(TimeUnit):
4152    arg_types = {"unit": True, "expression": True}
4153
4154    def interval(self):
4155        return Interval(
4156            this=self.expression.copy(),
4157            unit=self.unit.copy(),
4158        )
4159
4160
4161# https://www.oracletutorial.com/oracle-basics/oracle-interval/
4162# https://trino.io/docs/current/language/types.html#interval-day-to-second
4163# https://docs.databricks.com/en/sql/language-manual/data-types/interval-type.html
4164class IntervalSpan(DataType):
4165    arg_types = {"this": True, "expression": True}
4166
4167
4168class Interval(TimeUnit):
4169    arg_types = {"this": False, "unit": False}
4170
4171
4172class IgnoreNulls(Expression):
4173    pass
4174
4175
4176class RespectNulls(Expression):
4177    pass
4178
4179
4180# Functions
4181class Func(Condition):
4182    """
4183    The base class for all function expressions.
4184
4185    Attributes:
4186        is_var_len_args (bool): if set to True the last argument defined in arg_types will be
4187            treated as a variable length argument and the argument's value will be stored as a list.
4188        _sql_names (list): determines the SQL name (1st item in the list) and aliases (subsequent items)
4189            for this function expression. These values are used to map this node to a name during parsing
4190            as well as to provide the function's name during SQL string generation. By default the SQL
4191            name is set to the expression's class name transformed to snake case.
4192    """
4193
4194    is_var_len_args = False
4195
4196    @classmethod
4197    def from_arg_list(cls, args):
4198        if cls.is_var_len_args:
4199            all_arg_keys = list(cls.arg_types)
4200            # If this function supports variable length argument treat the last argument as such.
4201            non_var_len_arg_keys = all_arg_keys[:-1] if cls.is_var_len_args else all_arg_keys
4202            num_non_var = len(non_var_len_arg_keys)
4203
4204            args_dict = {arg_key: arg for arg, arg_key in zip(args, non_var_len_arg_keys)}
4205            args_dict[all_arg_keys[-1]] = args[num_non_var:]
4206        else:
4207            args_dict = {arg_key: arg for arg, arg_key in zip(args, cls.arg_types)}
4208
4209        return cls(**args_dict)
4210
4211    @classmethod
4212    def sql_names(cls):
4213        if cls is Func:
4214            raise NotImplementedError(
4215                "SQL name is only supported by concrete function implementations"
4216            )
4217        if "_sql_names" not in cls.__dict__:
4218            cls._sql_names = [camel_to_snake_case(cls.__name__)]
4219        return cls._sql_names
4220
4221    @classmethod
4222    def sql_name(cls):
4223        return cls.sql_names()[0]
4224
4225    @classmethod
4226    def default_parser_mappings(cls):
4227        return {name: cls.from_arg_list for name in cls.sql_names()}
4228
4229
4230class AggFunc(Func):
4231    pass
4232
4233
4234class ParameterizedAgg(AggFunc):
4235    arg_types = {"this": True, "expressions": True, "params": True}
4236
4237
4238class Abs(Func):
4239    pass
4240
4241
4242class ArgMax(AggFunc):
4243    arg_types = {"this": True, "expression": True, "count": False}
4244    _sql_names = ["ARG_MAX", "ARGMAX", "MAX_BY"]
4245
4246
4247class ArgMin(AggFunc):
4248    arg_types = {"this": True, "expression": True, "count": False}
4249    _sql_names = ["ARG_MIN", "ARGMIN", "MIN_BY"]
4250
4251
4252class ApproxTopK(AggFunc):
4253    arg_types = {"this": True, "expression": False, "counters": False}
4254
4255
4256class Flatten(Func):
4257    pass
4258
4259
4260# https://spark.apache.org/docs/latest/api/sql/index.html#transform
4261class Transform(Func):
4262    arg_types = {"this": True, "expression": True}
4263
4264
4265class Anonymous(Func):
4266    arg_types = {"this": True, "expressions": False}
4267    is_var_len_args = True
4268
4269
4270# https://docs.snowflake.com/en/sql-reference/functions/hll
4271# https://docs.aws.amazon.com/redshift/latest/dg/r_HLL_function.html
4272class Hll(AggFunc):
4273    arg_types = {"this": True, "expressions": False}
4274    is_var_len_args = True
4275
4276
4277class ApproxDistinct(AggFunc):
4278    arg_types = {"this": True, "accuracy": False}
4279    _sql_names = ["APPROX_DISTINCT", "APPROX_COUNT_DISTINCT"]
4280
4281
4282class Array(Func):
4283    arg_types = {"expressions": False}
4284    is_var_len_args = True
4285
4286
4287# https://docs.snowflake.com/en/sql-reference/functions/to_char
4288class ToChar(Func):
4289    arg_types = {"this": True, "format": False}
4290
4291
4292class GenerateSeries(Func):
4293    arg_types = {"start": True, "end": True, "step": False}
4294
4295
4296class ArrayAgg(AggFunc):
4297    pass
4298
4299
4300class ArrayAll(Func):
4301    arg_types = {"this": True, "expression": True}
4302
4303
4304class ArrayAny(Func):
4305    arg_types = {"this": True, "expression": True}
4306
4307
4308class ArrayConcat(Func):
4309    _sql_names = ["ARRAY_CONCAT", "ARRAY_CAT"]
4310    arg_types = {"this": True, "expressions": False}
4311    is_var_len_args = True
4312
4313
4314class ArrayContains(Binary, Func):
4315    pass
4316
4317
4318class ArrayContained(Binary):
4319    pass
4320
4321
4322class ArrayFilter(Func):
4323    arg_types = {"this": True, "expression": True}
4324    _sql_names = ["FILTER", "ARRAY_FILTER"]
4325
4326
4327class ArrayJoin(Func):
4328    arg_types = {"this": True, "expression": True, "null": False}
4329
4330
4331class ArraySize(Func):
4332    arg_types = {"this": True, "expression": False}
4333
4334
4335class ArraySort(Func):
4336    arg_types = {"this": True, "expression": False}
4337
4338
4339class ArraySum(Func):
4340    pass
4341
4342
4343class ArrayUnionAgg(AggFunc):
4344    pass
4345
4346
4347class Avg(AggFunc):
4348    pass
4349
4350
4351class AnyValue(AggFunc):
4352    arg_types = {"this": True, "having": False, "max": False, "ignore_nulls": False}
4353
4354
4355class First(Func):
4356    arg_types = {"this": True, "ignore_nulls": False}
4357
4358
4359class Last(Func):
4360    arg_types = {"this": True, "ignore_nulls": False}
4361
4362
4363class Case(Func):
4364    arg_types = {"this": False, "ifs": True, "default": False}
4365
4366    def when(self, condition: ExpOrStr, then: ExpOrStr, copy: bool = True, **opts) -> Case:
4367        instance = maybe_copy(self, copy)
4368        instance.append(
4369            "ifs",
4370            If(
4371                this=maybe_parse(condition, copy=copy, **opts),
4372                true=maybe_parse(then, copy=copy, **opts),
4373            ),
4374        )
4375        return instance
4376
4377    def else_(self, condition: ExpOrStr, copy: bool = True, **opts) -> Case:
4378        instance = maybe_copy(self, copy)
4379        instance.set("default", maybe_parse(condition, copy=copy, **opts))
4380        return instance
4381
4382
4383class Cast(Func):
4384    arg_types = {"this": True, "to": True, "format": False, "safe": False}
4385
4386    @property
4387    def name(self) -> str:
4388        return self.this.name
4389
4390    @property
4391    def to(self) -> DataType:
4392        return self.args["to"]
4393
4394    @property
4395    def output_name(self) -> str:
4396        return self.name
4397
4398    def is_type(self, *dtypes: str | DataType | DataType.Type) -> bool:
4399        """
4400        Checks whether this Cast's DataType matches one of the provided data types. Nested types
4401        like arrays or structs will be compared using "structural equivalence" semantics, so e.g.
4402        array<int> != array<float>.
4403
4404        Args:
4405            dtypes: the data types to compare this Cast's DataType to.
4406
4407        Returns:
4408            True, if and only if there is a type in `dtypes` which is equal to this Cast's DataType.
4409        """
4410        return self.to.is_type(*dtypes)
4411
4412
4413class TryCast(Cast):
4414    pass
4415
4416
4417class CastToStrType(Func):
4418    arg_types = {"this": True, "to": True}
4419
4420
4421class Collate(Binary, Func):
4422    pass
4423
4424
4425class Ceil(Func):
4426    arg_types = {"this": True, "decimals": False}
4427    _sql_names = ["CEIL", "CEILING"]
4428
4429
4430class Coalesce(Func):
4431    arg_types = {"this": True, "expressions": False}
4432    is_var_len_args = True
4433    _sql_names = ["COALESCE", "IFNULL", "NVL"]
4434
4435
4436class Chr(Func):
4437    arg_types = {"this": True, "charset": False, "expressions": False}
4438    is_var_len_args = True
4439    _sql_names = ["CHR", "CHAR"]
4440
4441
4442class Concat(Func):
4443    arg_types = {"expressions": True}
4444    is_var_len_args = True
4445
4446
4447class SafeConcat(Concat):
4448    pass
4449
4450
4451class ConcatWs(Concat):
4452    _sql_names = ["CONCAT_WS"]
4453
4454
4455class Count(AggFunc):
4456    arg_types = {"this": False, "expressions": False}
4457    is_var_len_args = True
4458
4459
4460class CountIf(AggFunc):
4461    pass
4462
4463
4464class CurrentDate(Func):
4465    arg_types = {"this": False}
4466
4467
4468class CurrentDatetime(Func):
4469    arg_types = {"this": False}
4470
4471
4472class CurrentTime(Func):
4473    arg_types = {"this": False}
4474
4475
4476class CurrentTimestamp(Func):
4477    arg_types = {"this": False}
4478
4479
4480class CurrentUser(Func):
4481    arg_types = {"this": False}
4482
4483
4484class DateAdd(Func, IntervalOp):
4485    arg_types = {"this": True, "expression": True, "unit": False}
4486
4487
4488class DateSub(Func, IntervalOp):
4489    arg_types = {"this": True, "expression": True, "unit": False}
4490
4491
4492class DateDiff(Func, TimeUnit):
4493    _sql_names = ["DATEDIFF", "DATE_DIFF"]
4494    arg_types = {"this": True, "expression": True, "unit": False}
4495
4496
4497class DateTrunc(Func):
4498    arg_types = {"unit": True, "this": True, "zone": False}
4499
4500    @property
4501    def unit(self) -> Expression:
4502        return self.args["unit"]
4503
4504
4505class DatetimeAdd(Func, IntervalOp):
4506    arg_types = {"this": True, "expression": True, "unit": False}
4507
4508
4509class DatetimeSub(Func, IntervalOp):
4510    arg_types = {"this": True, "expression": True, "unit": False}
4511
4512
4513class DatetimeDiff(Func, TimeUnit):
4514    arg_types = {"this": True, "expression": True, "unit": False}
4515
4516
4517class DatetimeTrunc(Func, TimeUnit):
4518    arg_types = {"this": True, "unit": True, "zone": False}
4519
4520
4521class DayOfWeek(Func):
4522    _sql_names = ["DAY_OF_WEEK", "DAYOFWEEK"]
4523
4524
4525class DayOfMonth(Func):
4526    _sql_names = ["DAY_OF_MONTH", "DAYOFMONTH"]
4527
4528
4529class DayOfYear(Func):
4530    _sql_names = ["DAY_OF_YEAR", "DAYOFYEAR"]
4531
4532
4533class ToDays(Func):
4534    pass
4535
4536
4537class WeekOfYear(Func):
4538    _sql_names = ["WEEK_OF_YEAR", "WEEKOFYEAR"]
4539
4540
4541class MonthsBetween(Func):
4542    arg_types = {"this": True, "expression": True, "roundoff": False}
4543
4544
4545class LastDateOfMonth(Func):
4546    pass
4547
4548
4549class Extract(Func):
4550    arg_types = {"this": True, "expression": True}
4551
4552
4553class Timestamp(Func):
4554    arg_types = {"this": False, "expression": False}
4555
4556
4557class TimestampAdd(Func, TimeUnit):
4558    arg_types = {"this": True, "expression": True, "unit": False}
4559
4560
4561class TimestampSub(Func, TimeUnit):
4562    arg_types = {"this": True, "expression": True, "unit": False}
4563
4564
4565class TimestampDiff(Func, TimeUnit):
4566    arg_types = {"this": True, "expression": True, "unit": False}
4567
4568
4569class TimestampTrunc(Func, TimeUnit):
4570    arg_types = {"this": True, "unit": True, "zone": False}
4571
4572
4573class TimeAdd(Func, TimeUnit):
4574    arg_types = {"this": True, "expression": True, "unit": False}
4575
4576
4577class TimeSub(Func, TimeUnit):
4578    arg_types = {"this": True, "expression": True, "unit": False}
4579
4580
4581class TimeDiff(Func, TimeUnit):
4582    arg_types = {"this": True, "expression": True, "unit": False}
4583
4584
4585class TimeTrunc(Func, TimeUnit):
4586    arg_types = {"this": True, "unit": True, "zone": False}
4587
4588
4589class DateFromParts(Func):
4590    _sql_names = ["DATEFROMPARTS"]
4591    arg_types = {"year": True, "month": True, "day": True}
4592
4593
4594class DateStrToDate(Func):
4595    pass
4596
4597
4598class DateToDateStr(Func):
4599    pass
4600
4601
4602class DateToDi(Func):
4603    pass
4604
4605
4606# https://cloud.google.com/bigquery/docs/reference/standard-sql/date_functions#date
4607class Date(Func):
4608    arg_types = {"this": False, "zone": False, "expressions": False}
4609    is_var_len_args = True
4610
4611
4612class Day(Func):
4613    pass
4614
4615
4616class Decode(Func):
4617    arg_types = {"this": True, "charset": True, "replace": False}
4618
4619
4620class DiToDate(Func):
4621    pass
4622
4623
4624class Encode(Func):
4625    arg_types = {"this": True, "charset": True}
4626
4627
4628class Exp(Func):
4629    pass
4630
4631
4632# https://docs.snowflake.com/en/sql-reference/functions/flatten
4633class Explode(Func):
4634    arg_types = {"this": True, "expressions": False}
4635    is_var_len_args = True
4636
4637
4638class ExplodeOuter(Explode):
4639    pass
4640
4641
4642class Posexplode(Explode):
4643    pass
4644
4645
4646class PosexplodeOuter(Posexplode):
4647    pass
4648
4649
4650class Floor(Func):
4651    arg_types = {"this": True, "decimals": False}
4652
4653
4654class FromBase64(Func):
4655    pass
4656
4657
4658class ToBase64(Func):
4659    pass
4660
4661
4662class Greatest(Func):
4663    arg_types = {"this": True, "expressions": False}
4664    is_var_len_args = True
4665
4666
4667class GroupConcat(AggFunc):
4668    arg_types = {"this": True, "separator": False}
4669
4670
4671class Hex(Func):
4672    pass
4673
4674
4675class Xor(Connector, Func):
4676    arg_types = {"this": False, "expression": False, "expressions": False}
4677
4678
4679class If(Func):
4680    arg_types = {"this": True, "true": True, "false": False}
4681
4682
4683class Nullif(Func):
4684    arg_types = {"this": True, "expression": True}
4685
4686
4687class Initcap(Func):
4688    arg_types = {"this": True, "expression": False}
4689
4690
4691class IsNan(Func):
4692    _sql_names = ["IS_NAN", "ISNAN"]
4693
4694
4695class FormatJson(Expression):
4696    pass
4697
4698
4699class JSONKeyValue(Expression):
4700    arg_types = {"this": True, "expression": True}
4701
4702
4703class JSONObject(Func):
4704    arg_types = {
4705        "expressions": False,
4706        "null_handling": False,
4707        "unique_keys": False,
4708        "return_type": False,
4709        "encoding": False,
4710    }
4711
4712
4713# https://docs.oracle.com/en/database/oracle/oracle-database/19/sqlrf/JSON_ARRAY.html
4714class JSONArray(Func):
4715    arg_types = {
4716        "expressions": True,
4717        "null_handling": False,
4718        "return_type": False,
4719        "strict": False,
4720    }
4721
4722
4723# https://docs.oracle.com/en/database/oracle/oracle-database/19/sqlrf/JSON_ARRAYAGG.html
4724class JSONArrayAgg(Func):
4725    arg_types = {
4726        "this": True,
4727        "order": False,
4728        "null_handling": False,
4729        "return_type": False,
4730        "strict": False,
4731    }
4732
4733
4734# https://docs.oracle.com/en/database/oracle/oracle-database/19/sqlrf/JSON_TABLE.html
4735# Note: parsing of JSON column definitions is currently incomplete.
4736class JSONColumnDef(Expression):
4737    arg_types = {"this": False, "kind": False, "path": False, "nested_schema": False}
4738
4739
4740class JSONSchema(Expression):
4741    arg_types = {"expressions": True}
4742
4743
4744# # https://docs.oracle.com/en/database/oracle/oracle-database/19/sqlrf/JSON_TABLE.html
4745class JSONTable(Func):
4746    arg_types = {
4747        "this": True,
4748        "schema": True,
4749        "path": False,
4750        "error_handling": False,
4751        "empty_handling": False,
4752    }
4753
4754
4755class OpenJSONColumnDef(Expression):
4756    arg_types = {"this": True, "kind": True, "path": False, "as_json": False}
4757
4758
4759class OpenJSON(Func):
4760    arg_types = {"this": True, "path": False, "expressions": False}
4761
4762
4763class JSONBContains(Binary):
4764    _sql_names = ["JSONB_CONTAINS"]
4765
4766
4767class JSONExtract(Binary, Func):
4768    _sql_names = ["JSON_EXTRACT"]
4769
4770
4771class JSONExtractScalar(JSONExtract):
4772    _sql_names = ["JSON_EXTRACT_SCALAR"]
4773
4774
4775class JSONBExtract(JSONExtract):
4776    _sql_names = ["JSONB_EXTRACT"]
4777
4778
4779class JSONBExtractScalar(JSONExtract):
4780    _sql_names = ["JSONB_EXTRACT_SCALAR"]
4781
4782
4783class JSONFormat(Func):
4784    arg_types = {"this": False, "options": False}
4785    _sql_names = ["JSON_FORMAT"]
4786
4787
4788# https://dev.mysql.com/doc/refman/8.0/en/json-search-functions.html#operator_member-of
4789class JSONArrayContains(Binary, Predicate, Func):
4790    _sql_names = ["JSON_ARRAY_CONTAINS"]
4791
4792
4793class ParseJSON(Func):
4794    # BigQuery, Snowflake have PARSE_JSON, Presto has JSON_PARSE
4795    _sql_names = ["PARSE_JSON", "JSON_PARSE"]
4796    arg_types = {"this": True, "expressions": False}
4797    is_var_len_args = True
4798
4799
4800class Least(Func):
4801    arg_types = {"this": True, "expressions": False}
4802    is_var_len_args = True
4803
4804
4805class Left(Func):
4806    arg_types = {"this": True, "expression": True}
4807
4808
4809class Right(Func):
4810    arg_types = {"this": True, "expression": True}
4811
4812
4813class Length(Func):
4814    _sql_names = ["LENGTH", "LEN"]
4815
4816
4817class Levenshtein(Func):
4818    arg_types = {
4819        "this": True,
4820        "expression": False,
4821        "ins_cost": False,
4822        "del_cost": False,
4823        "sub_cost": False,
4824    }
4825
4826
4827class Ln(Func):
4828    pass
4829
4830
4831class Log(Func):
4832    arg_types = {"this": True, "expression": False}
4833
4834
4835class Log2(Func):
4836    pass
4837
4838
4839class Log10(Func):
4840    pass
4841
4842
4843class LogicalOr(AggFunc):
4844    _sql_names = ["LOGICAL_OR", "BOOL_OR", "BOOLOR_AGG"]
4845
4846
4847class LogicalAnd(AggFunc):
4848    _sql_names = ["LOGICAL_AND", "BOOL_AND", "BOOLAND_AGG"]
4849
4850
4851class Lower(Func):
4852    _sql_names = ["LOWER", "LCASE"]
4853
4854
4855class Map(Func):
4856    arg_types = {"keys": False, "values": False}
4857
4858    @property
4859    def keys(self) -> t.List[Expression]:
4860        keys = self.args.get("keys")
4861        return keys.expressions if keys else []
4862
4863    @property
4864    def values(self) -> t.List[Expression]:
4865        values = self.args.get("values")
4866        return values.expressions if values else []
4867
4868
4869class MapFromEntries(Func):
4870    pass
4871
4872
4873class StarMap(Func):
4874    pass
4875
4876
4877class VarMap(Func):
4878    arg_types = {"keys": True, "values": True}
4879    is_var_len_args = True
4880
4881    @property
4882    def keys(self) -> t.List[Expression]:
4883        return self.args["keys"].expressions
4884
4885    @property
4886    def values(self) -> t.List[Expression]:
4887        return self.args["values"].expressions
4888
4889
4890# https://dev.mysql.com/doc/refman/8.0/en/fulltext-search.html
4891class MatchAgainst(Func):
4892    arg_types = {"this": True, "expressions": True, "modifier": False}
4893
4894
4895class Max(AggFunc):
4896    arg_types = {"this": True, "expressions": False}
4897    is_var_len_args = True
4898
4899
4900class MD5(Func):
4901    _sql_names = ["MD5"]
4902
4903
4904# Represents the variant of the MD5 function that returns a binary value
4905class MD5Digest(Func):
4906    _sql_names = ["MD5_DIGEST"]
4907
4908
4909class Min(AggFunc):
4910    arg_types = {"this": True, "expressions": False}
4911    is_var_len_args = True
4912
4913
4914class Month(Func):
4915    pass
4916
4917
4918class Nvl2(Func):
4919    arg_types = {"this": True, "true": True, "false": False}
4920
4921
4922# https://cloud.google.com/bigquery/docs/reference/standard-sql/bigqueryml-syntax-predict#mlpredict_function
4923class Predict(Func):
4924    arg_types = {"this": True, "expression": True, "params_struct": False}
4925
4926
4927class Pow(Binary, Func):
4928    _sql_names = ["POWER", "POW"]
4929
4930
4931class PercentileCont(AggFunc):
4932    arg_types = {"this": True, "expression": False}
4933
4934
4935class PercentileDisc(AggFunc):
4936    arg_types = {"this": True, "expression": False}
4937
4938
4939class Quantile(AggFunc):
4940    arg_types = {"this": True, "quantile": True}
4941
4942
4943class ApproxQuantile(Quantile):
4944    arg_types = {"this": True, "quantile": True, "accuracy": False, "weight": False}
4945
4946
4947class RangeN(Func):
4948    arg_types = {"this": True, "expressions": True, "each": False}
4949
4950
4951class ReadCSV(Func):
4952    _sql_names = ["READ_CSV"]
4953    is_var_len_args = True
4954    arg_types = {"this": True, "expressions": False}
4955
4956
4957class Reduce(Func):
4958    arg_types = {"this": True, "initial": True, "merge": True, "finish": False}
4959
4960
4961class RegexpExtract(Func):
4962    arg_types = {
4963        "this": True,
4964        "expression": True,
4965        "position": False,
4966        "occurrence": False,
4967        "parameters": False,
4968        "group": False,
4969    }
4970
4971
4972class RegexpReplace(Func):
4973    arg_types = {
4974        "this": True,
4975        "expression": True,
4976        "replacement": True,
4977        "position": False,
4978        "occurrence": False,
4979        "parameters": False,
4980        "modifiers": False,
4981    }
4982
4983
4984class RegexpLike(Binary, Func):
4985    arg_types = {"this": True, "expression": True, "flag": False}
4986
4987
4988class RegexpILike(Binary, Func):
4989    arg_types = {"this": True, "expression": True, "flag": False}
4990
4991
4992# https://spark.apache.org/docs/latest/api/python/reference/pyspark.sql/api/pyspark.sql.functions.split.html
4993# limit is the number of times a pattern is applied
4994class RegexpSplit(Func):
4995    arg_types = {"this": True, "expression": True, "limit": False}
4996
4997
4998class Repeat(Func):
4999    arg_types = {"this": True, "times": True}
5000
5001
5002class Round(Func):
5003    arg_types = {"this": True, "decimals": False}
5004
5005
5006class RowNumber(Func):
5007    arg_types: t.Dict[str, t.Any] = {}
5008
5009
5010class SafeDivide(Func):
5011    arg_types = {"this": True, "expression": True}
5012
5013
5014class SetAgg(AggFunc):
5015    pass
5016
5017
5018class SHA(Func):
5019    _sql_names = ["SHA", "SHA1"]
5020
5021
5022class SHA2(Func):
5023    _sql_names = ["SHA2"]
5024    arg_types = {"this": True, "length": False}
5025
5026
5027class SortArray(Func):
5028    arg_types = {"this": True, "asc": False}
5029
5030
5031class Split(Func):
5032    arg_types = {"this": True, "expression": True, "limit": False}
5033
5034
5035# Start may be omitted in the case of postgres
5036# https://www.postgresql.org/docs/9.1/functions-string.html @ Table 9-6
5037class Substring(Func):
5038    arg_types = {"this": True, "start": False, "length": False}
5039
5040
5041class StandardHash(Func):
5042    arg_types = {"this": True, "expression": False}
5043
5044
5045class StartsWith(Func):
5046    _sql_names = ["STARTS_WITH", "STARTSWITH"]
5047    arg_types = {"this": True, "expression": True}
5048
5049
5050class StrPosition(Func):
5051    arg_types = {
5052        "this": True,
5053        "substr": True,
5054        "position": False,
5055        "instance": False,
5056    }
5057
5058
5059class StrToDate(Func):
5060    arg_types = {"this": True, "format": True}
5061
5062
5063class StrToTime(Func):
5064    arg_types = {"this": True, "format": True, "zone": False}
5065
5066
5067# Spark allows unix_timestamp()
5068# https://spark.apache.org/docs/3.1.3/api/python/reference/api/pyspark.sql.functions.unix_timestamp.html
5069class StrToUnix(Func):
5070    arg_types = {"this": False, "format": False}
5071
5072
5073# https://prestodb.io/docs/current/functions/string.html
5074# https://spark.apache.org/docs/latest/api/sql/index.html#str_to_map
5075class StrToMap(Func):
5076    arg_types = {
5077        "this": True,
5078        "pair_delim": False,
5079        "key_value_delim": False,
5080        "duplicate_resolution_callback": False,
5081    }
5082
5083
5084class NumberToStr(Func):
5085    arg_types = {"this": True, "format": True, "culture": False}
5086
5087
5088class FromBase(Func):
5089    arg_types = {"this": True, "expression": True}
5090
5091
5092class Struct(Func):
5093    arg_types = {"expressions": False}
5094    is_var_len_args = True
5095
5096
5097class StructExtract(Func):
5098    arg_types = {"this": True, "expression": True}
5099
5100
5101# https://learn.microsoft.com/en-us/sql/t-sql/functions/stuff-transact-sql?view=sql-server-ver16
5102# https://docs.snowflake.com/en/sql-reference/functions/insert
5103class Stuff(Func):
5104    _sql_names = ["STUFF", "INSERT"]
5105    arg_types = {"this": True, "start": True, "length": True, "expression": True}
5106
5107
5108class Sum(AggFunc):
5109    pass
5110
5111
5112class Sqrt(Func):
5113    pass
5114
5115
5116class Stddev(AggFunc):
5117    pass
5118
5119
5120class StddevPop(AggFunc):
5121    pass
5122
5123
5124class StddevSamp(AggFunc):
5125    pass
5126
5127
5128class TimeToStr(Func):
5129    arg_types = {"this": True, "format": True, "culture": False}
5130
5131
5132class TimeToTimeStr(Func):
5133    pass
5134
5135
5136class TimeToUnix(Func):
5137    pass
5138
5139
5140class TimeStrToDate(Func):
5141    pass
5142
5143
5144class TimeStrToTime(Func):
5145    pass
5146
5147
5148class TimeStrToUnix(Func):
5149    pass
5150
5151
5152class Trim(Func):
5153    arg_types = {
5154        "this": True,
5155        "expression": False,
5156        "position": False,
5157        "collation": False,
5158    }
5159
5160
5161class TsOrDsAdd(Func, TimeUnit):
5162    arg_types = {"this": True, "expression": True, "unit": False}
5163
5164
5165class TsOrDsToDateStr(Func):
5166    pass
5167
5168
5169class TsOrDsToDate(Func):
5170    arg_types = {"this": True, "format": False}
5171
5172
5173class TsOrDiToDi(Func):
5174    pass
5175
5176
5177class Unhex(Func):
5178    pass
5179
5180
5181class UnixToStr(Func):
5182    arg_types = {"this": True, "format": False}
5183
5184
5185# https://prestodb.io/docs/current/functions/datetime.html
5186# presto has weird zone/hours/minutes
5187class UnixToTime(Func):
5188    arg_types = {"this": True, "scale": False, "zone": False, "hours": False, "minutes": False}
5189
5190    SECONDS = Literal.string("seconds")
5191    MILLIS = Literal.string("millis")
5192    MICROS = Literal.string("micros")
5193
5194
5195class UnixToTimeStr(Func):
5196    pass
5197
5198
5199class Upper(Func):
5200    _sql_names = ["UPPER", "UCASE"]
5201
5202
5203class Variance(AggFunc):
5204    _sql_names = ["VARIANCE", "VARIANCE_SAMP", "VAR_SAMP"]
5205
5206
5207class VariancePop(AggFunc):
5208    _sql_names = ["VARIANCE_POP", "VAR_POP"]
5209
5210
5211class Week(Func):
5212    arg_types = {"this": True, "mode": False}
5213
5214
5215class XMLTable(Func):
5216    arg_types = {"this": True, "passing": False, "columns": False, "by_ref": False}
5217
5218
5219class Year(Func):
5220    pass
5221
5222
5223class Use(Expression):
5224    arg_types = {"this": True, "kind": False}
5225
5226
5227class Merge(Expression):
5228    arg_types = {"this": True, "using": True, "on": True, "expressions": True, "with": False}
5229
5230
5231class When(Func):
5232    arg_types = {"matched": True, "source": False, "condition": False, "then": True}
5233
5234
5235# https://docs.oracle.com/javadb/10.8.3.0/ref/rrefsqljnextvaluefor.html
5236# https://learn.microsoft.com/en-us/sql/t-sql/functions/next-value-for-transact-sql?view=sql-server-ver16
5237class NextValueFor(Func):
5238    arg_types = {"this": True, "order": False}
5239
5240
5241def _norm_arg(arg):
5242    return arg.lower() if type(arg) is str else arg
5243
5244
5245ALL_FUNCTIONS = subclasses(__name__, Func, (AggFunc, Anonymous, Func))
5246
5247
5248# Helpers
5249@t.overload
5250def maybe_parse(
5251    sql_or_expression: ExpOrStr,
5252    *,
5253    into: t.Type[E],
5254    dialect: DialectType = None,
5255    prefix: t.Optional[str] = None,
5256    copy: bool = False,
5257    **opts,
5258) -> E:
5259    ...
5260
5261
5262@t.overload
5263def maybe_parse(
5264    sql_or_expression: str | E,
5265    *,
5266    into: t.Optional[IntoType] = None,
5267    dialect: DialectType = None,
5268    prefix: t.Optional[str] = None,
5269    copy: bool = False,
5270    **opts,
5271) -> E:
5272    ...
5273
5274
5275def maybe_parse(
5276    sql_or_expression: ExpOrStr,
5277    *,
5278    into: t.Optional[IntoType] = None,
5279    dialect: DialectType = None,
5280    prefix: t.Optional[str] = None,
5281    copy: bool = False,
5282    **opts,
5283) -> Expression:
5284    """Gracefully handle a possible string or expression.
5285
5286    Example:
5287        >>> maybe_parse("1")
5288        (LITERAL this: 1, is_string: False)
5289        >>> maybe_parse(to_identifier("x"))
5290        (IDENTIFIER this: x, quoted: False)
5291
5292    Args:
5293        sql_or_expression: the SQL code string or an expression
5294        into: the SQLGlot Expression to parse into
5295        dialect: the dialect used to parse the input expressions (in the case that an
5296            input expression is a SQL string).
5297        prefix: a string to prefix the sql with before it gets parsed
5298            (automatically includes a space)
5299        copy: whether or not to copy the expression.
5300        **opts: other options to use to parse the input expressions (again, in the case
5301            that an input expression is a SQL string).
5302
5303    Returns:
5304        Expression: the parsed or given expression.
5305    """
5306    if isinstance(sql_or_expression, Expression):
5307        if copy:
5308            return sql_or_expression.copy()
5309        return sql_or_expression
5310
5311    if sql_or_expression is None:
5312        raise ParseError(f"SQL cannot be None")
5313
5314    import sqlglot
5315
5316    sql = str(sql_or_expression)
5317    if prefix:
5318        sql = f"{prefix} {sql}"
5319
5320    return sqlglot.parse_one(sql, read=dialect, into=into, **opts)
5321
5322
5323@t.overload
5324def maybe_copy(instance: None, copy: bool = True) -> None:
5325    ...
5326
5327
5328@t.overload
5329def maybe_copy(instance: E, copy: bool = True) -> E:
5330    ...
5331
5332
5333def maybe_copy(instance, copy=True):
5334    return instance.copy() if copy and instance else instance
5335
5336
5337def _is_wrong_expression(expression, into):
5338    return isinstance(expression, Expression) and not isinstance(expression, into)
5339
5340
5341def _apply_builder(
5342    expression,
5343    instance,
5344    arg,
5345    copy=True,
5346    prefix=None,
5347    into=None,
5348    dialect=None,
5349    into_arg="this",
5350    **opts,
5351):
5352    if _is_wrong_expression(expression, into):
5353        expression = into(**{into_arg: expression})
5354    instance = maybe_copy(instance, copy)
5355    expression = maybe_parse(
5356        sql_or_expression=expression,
5357        prefix=prefix,
5358        into=into,
5359        dialect=dialect,
5360        **opts,
5361    )
5362    instance.set(arg, expression)
5363    return instance
5364
5365
5366def _apply_child_list_builder(
5367    *expressions,
5368    instance,
5369    arg,
5370    append=True,
5371    copy=True,
5372    prefix=None,
5373    into=None,
5374    dialect=None,
5375    properties=None,
5376    **opts,
5377):
5378    instance = maybe_copy(instance, copy)
5379    parsed = []
5380    for expression in expressions:
5381        if expression is not None:
5382            if _is_wrong_expression(expression, into):
5383                expression = into(expressions=[expression])
5384
5385            expression = maybe_parse(
5386                expression,
5387                into=into,
5388                dialect=dialect,
5389                prefix=prefix,
5390                **opts,
5391            )
5392            parsed.extend(expression.expressions)
5393
5394    existing = instance.args.get(arg)
5395    if append and existing:
5396        parsed = existing.expressions + parsed
5397
5398    child = into(expressions=parsed)
5399    for k, v in (properties or {}).items():
5400        child.set(k, v)
5401    instance.set(arg, child)
5402
5403    return instance
5404
5405
5406def _apply_list_builder(
5407    *expressions,
5408    instance,
5409    arg,
5410    append=True,
5411    copy=True,
5412    prefix=None,
5413    into=None,
5414    dialect=None,
5415    **opts,
5416):
5417    inst = maybe_copy(instance, copy)
5418
5419    expressions = [
5420        maybe_parse(
5421            sql_or_expression=expression,
5422            into=into,
5423            prefix=prefix,
5424            dialect=dialect,
5425            **opts,
5426        )
5427        for expression in expressions
5428        if expression is not None
5429    ]
5430
5431    existing_expressions = inst.args.get(arg)
5432    if append and existing_expressions:
5433        expressions = existing_expressions + expressions
5434
5435    inst.set(arg, expressions)
5436    return inst
5437
5438
5439def _apply_conjunction_builder(
5440    *expressions,
5441    instance,
5442    arg,
5443    into=None,
5444    append=True,
5445    copy=True,
5446    dialect=None,
5447    **opts,
5448):
5449    expressions = [exp for exp in expressions if exp is not None and exp != ""]
5450    if not expressions:
5451        return instance
5452
5453    inst = maybe_copy(instance, copy)
5454
5455    existing = inst.args.get(arg)
5456    if append and existing is not None:
5457        expressions = [existing.this if into else existing] + list(expressions)
5458
5459    node = and_(*expressions, dialect=dialect, copy=copy, **opts)
5460
5461    inst.set(arg, into(this=node) if into else node)
5462    return inst
5463
5464
5465def _apply_cte_builder(
5466    instance: E,
5467    alias: ExpOrStr,
5468    as_: ExpOrStr,
5469    recursive: t.Optional[bool] = None,
5470    append: bool = True,
5471    dialect: DialectType = None,
5472    copy: bool = True,
5473    **opts,
5474) -> E:
5475    alias_expression = maybe_parse(alias, dialect=dialect, into=TableAlias, **opts)
5476    as_expression = maybe_parse(as_, dialect=dialect, **opts)
5477    cte = CTE(this=as_expression, alias=alias_expression)
5478    return _apply_child_list_builder(
5479        cte,
5480        instance=instance,
5481        arg="with",
5482        append=append,
5483        copy=copy,
5484        into=With,
5485        properties={"recursive": recursive or False},
5486    )
5487
5488
5489def _combine(
5490    expressions: t.Sequence[t.Optional[ExpOrStr]],
5491    operator: t.Type[Connector],
5492    dialect: DialectType = None,
5493    copy: bool = True,
5494    **opts,
5495) -> Expression:
5496    conditions = [
5497        condition(expression, dialect=dialect, copy=copy, **opts)
5498        for expression in expressions
5499        if expression is not None
5500    ]
5501
5502    this, *rest = conditions
5503    if rest:
5504        this = _wrap(this, Connector)
5505    for expression in rest:
5506        this = operator(this=this, expression=_wrap(expression, Connector))
5507
5508    return this
5509
5510
5511def _wrap(expression: E, kind: t.Type[Expression]) -> E | Paren:
5512    return Paren(this=expression) if isinstance(expression, kind) else expression
5513
5514
5515def union(
5516    left: ExpOrStr,
5517    right: ExpOrStr,
5518    distinct: bool = True,
5519    dialect: DialectType = None,
5520    copy: bool = True,
5521    **opts,
5522) -> Union:
5523    """
5524    Initializes a syntax tree from one UNION expression.
5525
5526    Example:
5527        >>> union("SELECT * FROM foo", "SELECT * FROM bla").sql()
5528        'SELECT * FROM foo UNION SELECT * FROM bla'
5529
5530    Args:
5531        left: the SQL code string corresponding to the left-hand side.
5532            If an `Expression` instance is passed, it will be used as-is.
5533        right: the SQL code string corresponding to the right-hand side.
5534            If an `Expression` instance is passed, it will be used as-is.
5535        distinct: set the DISTINCT flag if and only if this is true.
5536        dialect: the dialect used to parse the input expression.
5537        copy: whether or not to copy the expression.
5538        opts: other options to use to parse the input expressions.
5539
5540    Returns:
5541        The new Union instance.
5542    """
5543    left = maybe_parse(sql_or_expression=left, dialect=dialect, copy=copy, **opts)
5544    right = maybe_parse(sql_or_expression=right, dialect=dialect, copy=copy, **opts)
5545
5546    return Union(this=left, expression=right, distinct=distinct)
5547
5548
5549def intersect(
5550    left: ExpOrStr,
5551    right: ExpOrStr,
5552    distinct: bool = True,
5553    dialect: DialectType = None,
5554    copy: bool = True,
5555    **opts,
5556) -> Intersect:
5557    """
5558    Initializes a syntax tree from one INTERSECT expression.
5559
5560    Example:
5561        >>> intersect("SELECT * FROM foo", "SELECT * FROM bla").sql()
5562        'SELECT * FROM foo INTERSECT SELECT * FROM bla'
5563
5564    Args:
5565        left: the SQL code string corresponding to the left-hand side.
5566            If an `Expression` instance is passed, it will be used as-is.
5567        right: the SQL code string corresponding to the right-hand side.
5568            If an `Expression` instance is passed, it will be used as-is.
5569        distinct: set the DISTINCT flag if and only if this is true.
5570        dialect: the dialect used to parse the input expression.
5571        copy: whether or not to copy the expression.
5572        opts: other options to use to parse the input expressions.
5573
5574    Returns:
5575        The new Intersect instance.
5576    """
5577    left = maybe_parse(sql_or_expression=left, dialect=dialect, copy=copy, **opts)
5578    right = maybe_parse(sql_or_expression=right, dialect=dialect, copy=copy, **opts)
5579
5580    return Intersect(this=left, expression=right, distinct=distinct)
5581
5582
5583def except_(
5584    left: ExpOrStr,
5585    right: ExpOrStr,
5586    distinct: bool = True,
5587    dialect: DialectType = None,
5588    copy: bool = True,
5589    **opts,
5590) -> Except:
5591    """
5592    Initializes a syntax tree from one EXCEPT expression.
5593
5594    Example:
5595        >>> except_("SELECT * FROM foo", "SELECT * FROM bla").sql()
5596        'SELECT * FROM foo EXCEPT SELECT * FROM bla'
5597
5598    Args:
5599        left: the SQL code string corresponding to the left-hand side.
5600            If an `Expression` instance is passed, it will be used as-is.
5601        right: the SQL code string corresponding to the right-hand side.
5602            If an `Expression` instance is passed, it will be used as-is.
5603        distinct: set the DISTINCT flag if and only if this is true.
5604        dialect: the dialect used to parse the input expression.
5605        copy: whether or not to copy the expression.
5606        opts: other options to use to parse the input expressions.
5607
5608    Returns:
5609        The new Except instance.
5610    """
5611    left = maybe_parse(sql_or_expression=left, dialect=dialect, copy=copy, **opts)
5612    right = maybe_parse(sql_or_expression=right, dialect=dialect, copy=copy, **opts)
5613
5614    return Except(this=left, expression=right, distinct=distinct)
5615
5616
5617def select(*expressions: ExpOrStr, dialect: DialectType = None, **opts) -> Select:
5618    """
5619    Initializes a syntax tree from one or multiple SELECT expressions.
5620
5621    Example:
5622        >>> select("col1", "col2").from_("tbl").sql()
5623        'SELECT col1, col2 FROM tbl'
5624
5625    Args:
5626        *expressions: the SQL code string to parse as the expressions of a
5627            SELECT statement. If an Expression instance is passed, this is used as-is.
5628        dialect: the dialect used to parse the input expressions (in the case that an
5629            input expression is a SQL string).
5630        **opts: other options to use to parse the input expressions (again, in the case
5631            that an input expression is a SQL string).
5632
5633    Returns:
5634        Select: the syntax tree for the SELECT statement.
5635    """
5636    return Select().select(*expressions, dialect=dialect, **opts)
5637
5638
5639def from_(expression: ExpOrStr, dialect: DialectType = None, **opts) -> Select:
5640    """
5641    Initializes a syntax tree from a FROM expression.
5642
5643    Example:
5644        >>> from_("tbl").select("col1", "col2").sql()
5645        'SELECT col1, col2 FROM tbl'
5646
5647    Args:
5648        *expression: the SQL code string to parse as the FROM expressions of a
5649            SELECT statement. If an Expression instance is passed, this is used as-is.
5650        dialect: the dialect used to parse the input expression (in the case that the
5651            input expression is a SQL string).
5652        **opts: other options to use to parse the input expressions (again, in the case
5653            that the input expression is a SQL string).
5654
5655    Returns:
5656        Select: the syntax tree for the SELECT statement.
5657    """
5658    return Select().from_(expression, dialect=dialect, **opts)
5659
5660
5661def update(
5662    table: str | Table,
5663    properties: dict,
5664    where: t.Optional[ExpOrStr] = None,
5665    from_: t.Optional[ExpOrStr] = None,
5666    dialect: DialectType = None,
5667    **opts,
5668) -> Update:
5669    """
5670    Creates an update statement.
5671
5672    Example:
5673        >>> update("my_table", {"x": 1, "y": "2", "z": None}, from_="baz", where="id > 1").sql()
5674        "UPDATE my_table SET x = 1, y = '2', z = NULL FROM baz WHERE id > 1"
5675
5676    Args:
5677        *properties: dictionary of properties to set which are
5678            auto converted to sql objects eg None -> NULL
5679        where: sql conditional parsed into a WHERE statement
5680        from_: sql statement parsed into a FROM statement
5681        dialect: the dialect used to parse the input expressions.
5682        **opts: other options to use to parse the input expressions.
5683
5684    Returns:
5685        Update: the syntax tree for the UPDATE statement.
5686    """
5687    update_expr = Update(this=maybe_parse(table, into=Table, dialect=dialect))
5688    update_expr.set(
5689        "expressions",
5690        [
5691            EQ(this=maybe_parse(k, dialect=dialect, **opts), expression=convert(v))
5692            for k, v in properties.items()
5693        ],
5694    )
5695    if from_:
5696        update_expr.set(
5697            "from",
5698            maybe_parse(from_, into=From, dialect=dialect, prefix="FROM", **opts),
5699        )
5700    if isinstance(where, Condition):
5701        where = Where(this=where)
5702    if where:
5703        update_expr.set(
5704            "where",
5705            maybe_parse(where, into=Where, dialect=dialect, prefix="WHERE", **opts),
5706        )
5707    return update_expr
5708
5709
5710def delete(
5711    table: ExpOrStr,
5712    where: t.Optional[ExpOrStr] = None,
5713    returning: t.Optional[ExpOrStr] = None,
5714    dialect: DialectType = None,
5715    **opts,
5716) -> Delete:
5717    """
5718    Builds a delete statement.
5719
5720    Example:
5721        >>> delete("my_table", where="id > 1").sql()
5722        'DELETE FROM my_table WHERE id > 1'
5723
5724    Args:
5725        where: sql conditional parsed into a WHERE statement
5726        returning: sql conditional parsed into a RETURNING statement
5727        dialect: the dialect used to parse the input expressions.
5728        **opts: other options to use to parse the input expressions.
5729
5730    Returns:
5731        Delete: the syntax tree for the DELETE statement.
5732    """
5733    delete_expr = Delete().delete(table, dialect=dialect, copy=False, **opts)
5734    if where:
5735        delete_expr = delete_expr.where(where, dialect=dialect, copy=False, **opts)
5736    if returning:
5737        delete_expr = delete_expr.returning(returning, dialect=dialect, copy=False, **opts)
5738    return delete_expr
5739
5740
5741def insert(
5742    expression: ExpOrStr,
5743    into: ExpOrStr,
5744    columns: t.Optional[t.Sequence[ExpOrStr]] = None,
5745    overwrite: t.Optional[bool] = None,
5746    dialect: DialectType = None,
5747    copy: bool = True,
5748    **opts,
5749) -> Insert:
5750    """
5751    Builds an INSERT statement.
5752
5753    Example:
5754        >>> insert("VALUES (1, 2, 3)", "tbl").sql()
5755        'INSERT INTO tbl VALUES (1, 2, 3)'
5756
5757    Args:
5758        expression: the sql string or expression of the INSERT statement
5759        into: the tbl to insert data to.
5760        columns: optionally the table's column names.
5761        overwrite: whether to INSERT OVERWRITE or not.
5762        dialect: the dialect used to parse the input expressions.
5763        copy: whether or not to copy the expression.
5764        **opts: other options to use to parse the input expressions.
5765
5766    Returns:
5767        Insert: the syntax tree for the INSERT statement.
5768    """
5769    expr = maybe_parse(expression, dialect=dialect, copy=copy, **opts)
5770    this: Table | Schema = maybe_parse(into, into=Table, dialect=dialect, copy=copy, **opts)
5771
5772    if columns:
5773        this = _apply_list_builder(
5774            *columns,
5775            instance=Schema(this=this),
5776            arg="expressions",
5777            into=Identifier,
5778            copy=False,
5779            dialect=dialect,
5780            **opts,
5781        )
5782
5783    return Insert(this=this, expression=expr, overwrite=overwrite)
5784
5785
5786def condition(
5787    expression: ExpOrStr, dialect: DialectType = None, copy: bool = True, **opts
5788) -> Condition:
5789    """
5790    Initialize a logical condition expression.
5791
5792    Example:
5793        >>> condition("x=1").sql()
5794        'x = 1'
5795
5796        This is helpful for composing larger logical syntax trees:
5797        >>> where = condition("x=1")
5798        >>> where = where.and_("y=1")
5799        >>> Select().from_("tbl").select("*").where(where).sql()
5800        'SELECT * FROM tbl WHERE x = 1 AND y = 1'
5801
5802    Args:
5803        *expression: the SQL code string to parse.
5804            If an Expression instance is passed, this is used as-is.
5805        dialect: the dialect used to parse the input expression (in the case that the
5806            input expression is a SQL string).
5807        copy: Whether or not to copy `expression` (only applies to expressions).
5808        **opts: other options to use to parse the input expressions (again, in the case
5809            that the input expression is a SQL string).
5810
5811    Returns:
5812        The new Condition instance
5813    """
5814    return maybe_parse(
5815        expression,
5816        into=Condition,
5817        dialect=dialect,
5818        copy=copy,
5819        **opts,
5820    )
5821
5822
5823def and_(
5824    *expressions: t.Optional[ExpOrStr], dialect: DialectType = None, copy: bool = True, **opts
5825) -> Condition:
5826    """
5827    Combine multiple conditions with an AND logical operator.
5828
5829    Example:
5830        >>> and_("x=1", and_("y=1", "z=1")).sql()
5831        'x = 1 AND (y = 1 AND z = 1)'
5832
5833    Args:
5834        *expressions: the SQL code strings to parse.
5835            If an Expression instance is passed, this is used as-is.
5836        dialect: the dialect used to parse the input expression.
5837        copy: whether or not to copy `expressions` (only applies to Expressions).
5838        **opts: other options to use to parse the input expressions.
5839
5840    Returns:
5841        And: the new condition
5842    """
5843    return t.cast(Condition, _combine(expressions, And, dialect, copy=copy, **opts))
5844
5845
5846def or_(
5847    *expressions: t.Optional[ExpOrStr], dialect: DialectType = None, copy: bool = True, **opts
5848) -> Condition:
5849    """
5850    Combine multiple conditions with an OR logical operator.
5851
5852    Example:
5853        >>> or_("x=1", or_("y=1", "z=1")).sql()
5854        'x = 1 OR (y = 1 OR z = 1)'
5855
5856    Args:
5857        *expressions: the SQL code strings to parse.
5858            If an Expression instance is passed, this is used as-is.
5859        dialect: the dialect used to parse the input expression.
5860        copy: whether or not to copy `expressions` (only applies to Expressions).
5861        **opts: other options to use to parse the input expressions.
5862
5863    Returns:
5864        Or: the new condition
5865    """
5866    return t.cast(Condition, _combine(expressions, Or, dialect, copy=copy, **opts))
5867
5868
5869def not_(expression: ExpOrStr, dialect: DialectType = None, copy: bool = True, **opts) -> Not:
5870    """
5871    Wrap a condition with a NOT operator.
5872
5873    Example:
5874        >>> not_("this_suit='black'").sql()
5875        "NOT this_suit = 'black'"
5876
5877    Args:
5878        expression: the SQL code string to parse.
5879            If an Expression instance is passed, this is used as-is.
5880        dialect: the dialect used to parse the input expression.
5881        copy: whether to copy the expression or not.
5882        **opts: other options to use to parse the input expressions.
5883
5884    Returns:
5885        The new condition.
5886    """
5887    this = condition(
5888        expression,
5889        dialect=dialect,
5890        copy=copy,
5891        **opts,
5892    )
5893    return Not(this=_wrap(this, Connector))
5894
5895
5896def paren(expression: ExpOrStr, copy: bool = True) -> Paren:
5897    """
5898    Wrap an expression in parentheses.
5899
5900    Example:
5901        >>> paren("5 + 3").sql()
5902        '(5 + 3)'
5903
5904    Args:
5905        expression: the SQL code string to parse.
5906            If an Expression instance is passed, this is used as-is.
5907        copy: whether to copy the expression or not.
5908
5909    Returns:
5910        The wrapped expression.
5911    """
5912    return Paren(this=maybe_parse(expression, copy=copy))
5913
5914
5915SAFE_IDENTIFIER_RE = re.compile(r"^[_a-zA-Z][\w]*$")
5916
5917
5918@t.overload
5919def to_identifier(name: None, quoted: t.Optional[bool] = None, copy: bool = True) -> None:
5920    ...
5921
5922
5923@t.overload
5924def to_identifier(
5925    name: str | Identifier, quoted: t.Optional[bool] = None, copy: bool = True
5926) -> Identifier:
5927    ...
5928
5929
5930def to_identifier(name, quoted=None, copy=True):
5931    """Builds an identifier.
5932
5933    Args:
5934        name: The name to turn into an identifier.
5935        quoted: Whether or not force quote the identifier.
5936        copy: Whether or not to copy name if it's an Identifier.
5937
5938    Returns:
5939        The identifier ast node.
5940    """
5941
5942    if name is None:
5943        return None
5944
5945    if isinstance(name, Identifier):
5946        identifier = maybe_copy(name, copy)
5947    elif isinstance(name, str):
5948        identifier = Identifier(
5949            this=name,
5950            quoted=not SAFE_IDENTIFIER_RE.match(name) if quoted is None else quoted,
5951        )
5952    else:
5953        raise ValueError(f"Name needs to be a string or an Identifier, got: {name.__class__}")
5954    return identifier
5955
5956
5957def parse_identifier(name: str | Identifier, dialect: DialectType = None) -> Identifier:
5958    """
5959    Parses a given string into an identifier.
5960
5961    Args:
5962        name: The name to parse into an identifier.
5963        dialect: The dialect to parse against.
5964
5965    Returns:
5966        The identifier ast node.
5967    """
5968    try:
5969        expression = maybe_parse(name, dialect=dialect, into=Identifier)
5970    except ParseError:
5971        expression = to_identifier(name)
5972
5973    return expression
5974
5975
5976INTERVAL_STRING_RE = re.compile(r"\s*([0-9]+)\s*([a-zA-Z]+)\s*")
5977
5978
5979def to_interval(interval: str | Literal) -> Interval:
5980    """Builds an interval expression from a string like '1 day' or '5 months'."""
5981    if isinstance(interval, Literal):
5982        if not interval.is_string:
5983            raise ValueError("Invalid interval string.")
5984
5985        interval = interval.this
5986
5987    interval_parts = INTERVAL_STRING_RE.match(interval)  # type: ignore
5988
5989    if not interval_parts:
5990        raise ValueError("Invalid interval string.")
5991
5992    return Interval(
5993        this=Literal.string(interval_parts.group(1)),
5994        unit=Var(this=interval_parts.group(2)),
5995    )
5996
5997
5998@t.overload
5999def to_table(sql_path: str | Table, **kwargs) -> Table:
6000    ...
6001
6002
6003@t.overload
6004def to_table(sql_path: None, **kwargs) -> None:
6005    ...
6006
6007
6008def to_table(
6009    sql_path: t.Optional[str | Table], dialect: DialectType = None, **kwargs
6010) -> t.Optional[Table]:
6011    """
6012    Create a table expression from a `[catalog].[schema].[table]` sql path. Catalog and schema are optional.
6013    If a table is passed in then that table is returned.
6014
6015    Args:
6016        sql_path: a `[catalog].[schema].[table]` string.
6017        dialect: the source dialect according to which the table name will be parsed.
6018        kwargs: the kwargs to instantiate the resulting `Table` expression with.
6019
6020    Returns:
6021        A table expression.
6022    """
6023    if sql_path is None or isinstance(sql_path, Table):
6024        return sql_path
6025    if not isinstance(sql_path, str):
6026        raise ValueError(f"Invalid type provided for a table: {type(sql_path)}")
6027
6028    table = maybe_parse(sql_path, into=Table, dialect=dialect)
6029    if table:
6030        for k, v in kwargs.items():
6031            table.set(k, v)
6032
6033    return table
6034
6035
6036def to_column(sql_path: str | Column, **kwargs) -> Column:
6037    """
6038    Create a column from a `[table].[column]` sql path. Schema is optional.
6039
6040    If a column is passed in then that column is returned.
6041
6042    Args:
6043        sql_path: `[table].[column]` string
6044    Returns:
6045        Table: A column expression
6046    """
6047    if sql_path is None or isinstance(sql_path, Column):
6048        return sql_path
6049    if not isinstance(sql_path, str):
6050        raise ValueError(f"Invalid type provided for column: {type(sql_path)}")
6051    return column(*reversed(sql_path.split(".")), **kwargs)  # type: ignore
6052
6053
6054def alias_(
6055    expression: ExpOrStr,
6056    alias: str | Identifier,
6057    table: bool | t.Sequence[str | Identifier] = False,
6058    quoted: t.Optional[bool] = None,
6059    dialect: DialectType = None,
6060    copy: bool = True,
6061    **opts,
6062):
6063    """Create an Alias expression.
6064
6065    Example:
6066        >>> alias_('foo', 'bar').sql()
6067        'foo AS bar'
6068
6069        >>> alias_('(select 1, 2)', 'bar', table=['a', 'b']).sql()
6070        '(SELECT 1, 2) AS bar(a, b)'
6071
6072    Args:
6073        expression: the SQL code strings to parse.
6074            If an Expression instance is passed, this is used as-is.
6075        alias: the alias name to use. If the name has
6076            special characters it is quoted.
6077        table: Whether or not to create a table alias, can also be a list of columns.
6078        quoted: whether or not to quote the alias
6079        dialect: the dialect used to parse the input expression.
6080        copy: Whether or not to copy the expression.
6081        **opts: other options to use to parse the input expressions.
6082
6083    Returns:
6084        Alias: the aliased expression
6085    """
6086    exp = maybe_parse(expression, dialect=dialect, copy=copy, **opts)
6087    alias = to_identifier(alias, quoted=quoted)
6088
6089    if table:
6090        table_alias = TableAlias(this=alias)
6091        exp.set("alias", table_alias)
6092
6093        if not isinstance(table, bool):
6094            for column in table:
6095                table_alias.append("columns", to_identifier(column, quoted=quoted))
6096
6097        return exp
6098
6099    # We don't set the "alias" arg for Window expressions, because that would add an IDENTIFIER node in
6100    # the AST, representing a "named_window" [1] construct (eg. bigquery). What we want is an ALIAS node
6101    # for the complete Window expression.
6102    #
6103    # [1]: https://cloud.google.com/bigquery/docs/reference/standard-sql/window-function-calls
6104
6105    if "alias" in exp.arg_types and not isinstance(exp, Window):
6106        exp.set("alias", alias)
6107        return exp
6108    return Alias(this=exp, alias=alias)
6109
6110
6111def subquery(
6112    expression: ExpOrStr,
6113    alias: t.Optional[Identifier | str] = None,
6114    dialect: DialectType = None,
6115    **opts,
6116) -> Select:
6117    """
6118    Build a subquery expression.
6119
6120    Example:
6121        >>> subquery('select x from tbl', 'bar').select('x').sql()
6122        'SELECT x FROM (SELECT x FROM tbl) AS bar'
6123
6124    Args:
6125        expression: the SQL code strings to parse.
6126            If an Expression instance is passed, this is used as-is.
6127        alias: the alias name to use.
6128        dialect: the dialect used to parse the input expression.
6129        **opts: other options to use to parse the input expressions.
6130
6131    Returns:
6132        A new Select instance with the subquery expression included.
6133    """
6134
6135    expression = maybe_parse(expression, dialect=dialect, **opts).subquery(alias)
6136    return Select().from_(expression, dialect=dialect, **opts)
6137
6138
6139def column(
6140    col: str | Identifier,
6141    table: t.Optional[str | Identifier] = None,
6142    db: t.Optional[str | Identifier] = None,
6143    catalog: t.Optional[str | Identifier] = None,
6144    quoted: t.Optional[bool] = None,
6145) -> Column:
6146    """
6147    Build a Column.
6148
6149    Args:
6150        col: Column name.
6151        table: Table name.
6152        db: Database name.
6153        catalog: Catalog name.
6154        quoted: Whether to force quotes on the column's identifiers.
6155
6156    Returns:
6157        The new Column instance.
6158    """
6159    return Column(
6160        this=to_identifier(col, quoted=quoted),
6161        table=to_identifier(table, quoted=quoted),
6162        db=to_identifier(db, quoted=quoted),
6163        catalog=to_identifier(catalog, quoted=quoted),
6164    )
6165
6166
6167def cast(expression: ExpOrStr, to: str | DataType | DataType.Type, **opts) -> Cast:
6168    """Cast an expression to a data type.
6169
6170    Example:
6171        >>> cast('x + 1', 'int').sql()
6172        'CAST(x + 1 AS INT)'
6173
6174    Args:
6175        expression: The expression to cast.
6176        to: The datatype to cast to.
6177
6178    Returns:
6179        The new Cast instance.
6180    """
6181    expression = maybe_parse(expression, **opts)
6182    data_type = DataType.build(to, **opts)
6183    expression = Cast(this=expression, to=data_type)
6184    expression.type = data_type
6185    return expression
6186
6187
6188def table_(
6189    table: Identifier | str,
6190    db: t.Optional[Identifier | str] = None,
6191    catalog: t.Optional[Identifier | str] = None,
6192    quoted: t.Optional[bool] = None,
6193    alias: t.Optional[Identifier | str] = None,
6194) -> Table:
6195    """Build a Table.
6196
6197    Args:
6198        table: Table name.
6199        db: Database name.
6200        catalog: Catalog name.
6201        quote: Whether to force quotes on the table's identifiers.
6202        alias: Table's alias.
6203
6204    Returns:
6205        The new Table instance.
6206    """
6207    return Table(
6208        this=to_identifier(table, quoted=quoted) if table else None,
6209        db=to_identifier(db, quoted=quoted) if db else None,
6210        catalog=to_identifier(catalog, quoted=quoted) if catalog else None,
6211        alias=TableAlias(this=to_identifier(alias)) if alias else None,
6212    )
6213
6214
6215def values(
6216    values: t.Iterable[t.Tuple[t.Any, ...]],
6217    alias: t.Optional[str] = None,
6218    columns: t.Optional[t.Iterable[str] | t.Dict[str, DataType]] = None,
6219) -> Values:
6220    """Build VALUES statement.
6221
6222    Example:
6223        >>> values([(1, '2')]).sql()
6224        "VALUES (1, '2')"
6225
6226    Args:
6227        values: values statements that will be converted to SQL
6228        alias: optional alias
6229        columns: Optional list of ordered column names or ordered dictionary of column names to types.
6230         If either are provided then an alias is also required.
6231
6232    Returns:
6233        Values: the Values expression object
6234    """
6235    if columns and not alias:
6236        raise ValueError("Alias is required when providing columns")
6237
6238    return Values(
6239        expressions=[convert(tup) for tup in values],
6240        alias=(
6241            TableAlias(this=to_identifier(alias), columns=[to_identifier(x) for x in columns])
6242            if columns
6243            else (TableAlias(this=to_identifier(alias)) if alias else None)
6244        ),
6245    )
6246
6247
6248def var(name: t.Optional[ExpOrStr]) -> Var:
6249    """Build a SQL variable.
6250
6251    Example:
6252        >>> repr(var('x'))
6253        '(VAR this: x)'
6254
6255        >>> repr(var(column('x', table='y')))
6256        '(VAR this: x)'
6257
6258    Args:
6259        name: The name of the var or an expression who's name will become the var.
6260
6261    Returns:
6262        The new variable node.
6263    """
6264    if not name:
6265        raise ValueError("Cannot convert empty name into var.")
6266
6267    if isinstance(name, Expression):
6268        name = name.name
6269    return Var(this=name)
6270
6271
6272def rename_table(old_name: str | Table, new_name: str | Table) -> AlterTable:
6273    """Build ALTER TABLE... RENAME... expression
6274
6275    Args:
6276        old_name: The old name of the table
6277        new_name: The new name of the table
6278
6279    Returns:
6280        Alter table expression
6281    """
6282    old_table = to_table(old_name)
6283    new_table = to_table(new_name)
6284    return AlterTable(
6285        this=old_table,
6286        actions=[
6287            RenameTable(this=new_table),
6288        ],
6289    )
6290
6291
6292def convert(value: t.Any, copy: bool = False) -> Expression:
6293    """Convert a python value into an expression object.
6294
6295    Raises an error if a conversion is not possible.
6296
6297    Args:
6298        value: A python object.
6299        copy: Whether or not to copy `value` (only applies to Expressions and collections).
6300
6301    Returns:
6302        Expression: the equivalent expression object.
6303    """
6304    if isinstance(value, Expression):
6305        return maybe_copy(value, copy)
6306    if isinstance(value, str):
6307        return Literal.string(value)
6308    if isinstance(value, bool):
6309        return Boolean(this=value)
6310    if value is None or (isinstance(value, float) and math.isnan(value)):
6311        return NULL
6312    if isinstance(value, numbers.Number):
6313        return Literal.number(value)
6314    if isinstance(value, datetime.datetime):
6315        datetime_literal = Literal.string(
6316            (value if value.tzinfo else value.replace(tzinfo=datetime.timezone.utc)).isoformat()
6317        )
6318        return TimeStrToTime(this=datetime_literal)
6319    if isinstance(value, datetime.date):
6320        date_literal = Literal.string(value.strftime("%Y-%m-%d"))
6321        return DateStrToDate(this=date_literal)
6322    if isinstance(value, tuple):
6323        return Tuple(expressions=[convert(v, copy=copy) for v in value])
6324    if isinstance(value, list):
6325        return Array(expressions=[convert(v, copy=copy) for v in value])
6326    if isinstance(value, dict):
6327        return Map(
6328            keys=Array(expressions=[convert(k, copy=copy) for k in value]),
6329            values=Array(expressions=[convert(v, copy=copy) for v in value.values()]),
6330        )
6331    raise ValueError(f"Cannot convert {value}")
6332
6333
6334def replace_children(expression: Expression, fun: t.Callable, *args, **kwargs) -> None:
6335    """
6336    Replace children of an expression with the result of a lambda fun(child) -> exp.
6337    """
6338    for k, v in expression.args.items():
6339        is_list_arg = type(v) is list
6340
6341        child_nodes = v if is_list_arg else [v]
6342        new_child_nodes = []
6343
6344        for cn in child_nodes:
6345            if isinstance(cn, Expression):
6346                for child_node in ensure_collection(fun(cn, *args, **kwargs)):
6347                    new_child_nodes.append(child_node)
6348                    child_node.parent = expression
6349                    child_node.arg_key = k
6350            else:
6351                new_child_nodes.append(cn)
6352
6353        expression.args[k] = new_child_nodes if is_list_arg else seq_get(new_child_nodes, 0)
6354
6355
6356def column_table_names(expression: Expression, exclude: str = "") -> t.Set[str]:
6357    """
6358    Return all table names referenced through columns in an expression.
6359
6360    Example:
6361        >>> import sqlglot
6362        >>> sorted(column_table_names(sqlglot.parse_one("a.b AND c.d AND c.e")))
6363        ['a', 'c']
6364
6365    Args:
6366        expression: expression to find table names.
6367        exclude: a table name to exclude
6368
6369    Returns:
6370        A list of unique names.
6371    """
6372    return {
6373        table
6374        for table in (column.table for column in expression.find_all(Column))
6375        if table and table != exclude
6376    }
6377
6378
6379def table_name(table: Table | str, dialect: DialectType = None) -> str:
6380    """Get the full name of a table as a string.
6381
6382    Args:
6383        table: Table expression node or string.
6384        dialect: The dialect to generate the table name for.
6385
6386    Examples:
6387        >>> from sqlglot import exp, parse_one
6388        >>> table_name(parse_one("select * from a.b.c").find(exp.Table))
6389        'a.b.c'
6390
6391    Returns:
6392        The table name.
6393    """
6394
6395    table = maybe_parse(table, into=Table, dialect=dialect)
6396
6397    if not table:
6398        raise ValueError(f"Cannot parse {table}")
6399
6400    return ".".join(
6401        part.sql(dialect=dialect, identify=True)
6402        if not SAFE_IDENTIFIER_RE.match(part.name)
6403        else part.name
6404        for part in table.parts
6405    )
6406
6407
6408def replace_tables(expression: E, mapping: t.Dict[str, str], copy: bool = True) -> E:
6409    """Replace all tables in expression according to the mapping.
6410
6411    Args:
6412        expression: expression node to be transformed and replaced.
6413        mapping: mapping of table names.
6414        copy: whether or not to copy the expression.
6415
6416    Examples:
6417        >>> from sqlglot import exp, parse_one
6418        >>> replace_tables(parse_one("select * from a.b"), {"a.b": "c"}).sql()
6419        'SELECT * FROM c'
6420
6421    Returns:
6422        The mapped expression.
6423    """
6424
6425    def _replace_tables(node: Expression) -> Expression:
6426        if isinstance(node, Table):
6427            new_name = mapping.get(table_name(node))
6428            if new_name:
6429                return to_table(
6430                    new_name,
6431                    **{k: v for k, v in node.args.items() if k not in ("this", "db", "catalog")},
6432                )
6433        return node
6434
6435    return expression.transform(_replace_tables, copy=copy)
6436
6437
6438def replace_placeholders(expression: Expression, *args, **kwargs) -> Expression:
6439    """Replace placeholders in an expression.
6440
6441    Args:
6442        expression: expression node to be transformed and replaced.
6443        args: positional names that will substitute unnamed placeholders in the given order.
6444        kwargs: keyword arguments that will substitute named placeholders.
6445
6446    Examples:
6447        >>> from sqlglot import exp, parse_one
6448        >>> replace_placeholders(
6449        ...     parse_one("select * from :tbl where ? = ?"),
6450        ...     exp.to_identifier("str_col"), "b", tbl=exp.to_identifier("foo")
6451        ... ).sql()
6452        "SELECT * FROM foo WHERE str_col = 'b'"
6453
6454    Returns:
6455        The mapped expression.
6456    """
6457
6458    def _replace_placeholders(node: Expression, args, **kwargs) -> Expression:
6459        if isinstance(node, Placeholder):
6460            if node.name:
6461                new_name = kwargs.get(node.name)
6462                if new_name:
6463                    return convert(new_name)
6464            else:
6465                try:
6466                    return convert(next(args))
6467                except StopIteration:
6468                    pass
6469        return node
6470
6471    return expression.transform(_replace_placeholders, iter(args), **kwargs)
6472
6473
6474def expand(
6475    expression: Expression, sources: t.Dict[str, Subqueryable], copy: bool = True
6476) -> Expression:
6477    """Transforms an expression by expanding all referenced sources into subqueries.
6478
6479    Examples:
6480        >>> from sqlglot import parse_one
6481        >>> expand(parse_one("select * from x AS z"), {"x": parse_one("select * from y")}).sql()
6482        'SELECT * FROM (SELECT * FROM y) AS z /* source: x */'
6483
6484        >>> expand(parse_one("select * from x AS z"), {"x": parse_one("select * from y"), "y": parse_one("select * from z")}).sql()
6485        'SELECT * FROM (SELECT * FROM (SELECT * FROM z) AS y /* source: y */) AS z /* source: x */'
6486
6487    Args:
6488        expression: The expression to expand.
6489        sources: A dictionary of name to Subqueryables.
6490        copy: Whether or not to copy the expression during transformation. Defaults to True.
6491
6492    Returns:
6493        The transformed expression.
6494    """
6495
6496    def _expand(node: Expression):
6497        if isinstance(node, Table):
6498            name = table_name(node)
6499            source = sources.get(name)
6500            if source:
6501                subquery = source.subquery(node.alias or name)
6502                subquery.comments = [f"source: {name}"]
6503                return subquery.transform(_expand, copy=False)
6504        return node
6505
6506    return expression.transform(_expand, copy=copy)
6507
6508
6509def func(name: str, *args, dialect: DialectType = None, **kwargs) -> Func:
6510    """
6511    Returns a Func expression.
6512
6513    Examples:
6514        >>> func("abs", 5).sql()
6515        'ABS(5)'
6516
6517        >>> func("cast", this=5, to=DataType.build("DOUBLE")).sql()
6518        'CAST(5 AS DOUBLE)'
6519
6520    Args:
6521        name: the name of the function to build.
6522        args: the args used to instantiate the function of interest.
6523        dialect: the source dialect.
6524        kwargs: the kwargs used to instantiate the function of interest.
6525
6526    Note:
6527        The arguments `args` and `kwargs` are mutually exclusive.
6528
6529    Returns:
6530        An instance of the function of interest, or an anonymous function, if `name` doesn't
6531        correspond to an existing `sqlglot.expressions.Func` class.
6532    """
6533    if args and kwargs:
6534        raise ValueError("Can't use both args and kwargs to instantiate a function.")
6535
6536    from sqlglot.dialects.dialect import Dialect
6537
6538    converted: t.List[Expression] = [maybe_parse(arg, dialect=dialect) for arg in args]
6539    kwargs = {key: maybe_parse(value, dialect=dialect) for key, value in kwargs.items()}
6540
6541    parser = Dialect.get_or_raise(dialect)().parser()
6542    from_args_list = parser.FUNCTIONS.get(name.upper())
6543
6544    if from_args_list:
6545        function = from_args_list(converted) if converted else from_args_list.__self__(**kwargs)  # type: ignore
6546    else:
6547        kwargs = kwargs or {"expressions": converted}
6548        function = Anonymous(this=name, **kwargs)
6549
6550    for error_message in function.error_messages(converted):
6551        raise ValueError(error_message)
6552
6553    return function
6554
6555
6556def case(
6557    expression: t.Optional[ExpOrStr] = None,
6558    **opts,
6559) -> Case:
6560    """
6561    Initialize a CASE statement.
6562
6563    Example:
6564        case().when("a = 1", "foo").else_("bar")
6565
6566    Args:
6567        expression: Optionally, the input expression (not all dialects support this)
6568        **opts: Extra keyword arguments for parsing `expression`
6569    """
6570    if expression is not None:
6571        this = maybe_parse(expression, **opts)
6572    else:
6573        this = None
6574    return Case(this=this, ifs=[])
6575
6576
6577def true() -> Boolean:
6578    """
6579    Returns a true Boolean expression.
6580    """
6581    return Boolean(this=True)
6582
6583
6584def false() -> Boolean:
6585    """
6586    Returns a false Boolean expression.
6587    """
6588    return Boolean(this=False)
6589
6590
6591def null() -> Null:
6592    """
6593    Returns a Null expression.
6594    """
6595    return Null()
6596
6597
6598# TODO: deprecate this
6599TRUE = Boolean(this=True)
6600FALSE = Boolean(this=False)
6601NULL = Null()