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 textwrap
  20import typing as t
  21from collections import deque
  22from copy import deepcopy
  23from decimal import Decimal
  24from enum import auto
  25from functools import reduce
  26
  27from sqlglot.errors import ErrorLevel, ParseError
  28from sqlglot.helper import (
  29    AutoName,
  30    camel_to_snake_case,
  31    ensure_collection,
  32    ensure_list,
  33    seq_get,
  34    subclasses,
  35)
  36from sqlglot.tokens import Token, TokenError
  37
  38if t.TYPE_CHECKING:
  39    from sqlglot._typing import E, Lit
  40    from sqlglot.dialects.dialect import DialectType
  41
  42    Q = t.TypeVar("Q", bound="Query")
  43    S = t.TypeVar("S", bound="SetOperation")
  44
  45
  46class _Expression(type):
  47    def __new__(cls, clsname, bases, attrs):
  48        klass = super().__new__(cls, clsname, bases, attrs)
  49
  50        # When an Expression class is created, its key is automatically set to be
  51        # the lowercase version of the class' name.
  52        klass.key = clsname.lower()
  53
  54        # This is so that docstrings are not inherited in pdoc
  55        klass.__doc__ = klass.__doc__ or ""
  56
  57        return klass
  58
  59
  60SQLGLOT_META = "sqlglot.meta"
  61TABLE_PARTS = ("this", "db", "catalog")
  62COLUMN_PARTS = ("this", "table", "db", "catalog")
  63
  64
  65class Expression(metaclass=_Expression):
  66    """
  67    The base class for all expressions in a syntax tree. Each Expression encapsulates any necessary
  68    context, such as its child expressions, their names (arg keys), and whether a given child expression
  69    is optional or not.
  70
  71    Attributes:
  72        key: a unique key for each class in the Expression hierarchy. This is useful for hashing
  73            and representing expressions as strings.
  74        arg_types: determines the arguments (child nodes) supported by an expression. It maps
  75            arg keys to booleans that indicate whether the corresponding args are optional.
  76        parent: a reference to the parent expression (or None, in case of root expressions).
  77        arg_key: the arg key an expression is associated with, i.e. the name its parent expression
  78            uses to refer to it.
  79        index: the index of an expression if it is inside of a list argument in its parent.
  80        comments: a list of comments that are associated with a given expression. This is used in
  81            order to preserve comments when transpiling SQL code.
  82        type: the `sqlglot.expressions.DataType` type of an expression. This is inferred by the
  83            optimizer, in order to enable some transformations that require type information.
  84        meta: a dictionary that can be used to store useful metadata for a given expression.
  85
  86    Example:
  87        >>> class Foo(Expression):
  88        ...     arg_types = {"this": True, "expression": False}
  89
  90        The above definition informs us that Foo is an Expression that requires an argument called
  91        "this" and may also optionally receive an argument called "expression".
  92
  93    Args:
  94        args: a mapping used for retrieving the arguments of an expression, given their arg keys.
  95    """
  96
  97    key = "expression"
  98    arg_types = {"this": True}
  99    __slots__ = ("args", "parent", "arg_key", "index", "comments", "_type", "_meta", "_hash")
 100
 101    def __init__(self, **args: t.Any):
 102        self.args: t.Dict[str, t.Any] = args
 103        self.parent: t.Optional[Expression] = None
 104        self.arg_key: t.Optional[str] = None
 105        self.index: t.Optional[int] = None
 106        self.comments: t.Optional[t.List[str]] = None
 107        self._type: t.Optional[DataType] = None
 108        self._meta: t.Optional[t.Dict[str, t.Any]] = None
 109        self._hash: t.Optional[int] = None
 110
 111        for arg_key, value in self.args.items():
 112            self._set_parent(arg_key, value)
 113
 114    def __eq__(self, other) -> bool:
 115        return type(self) is type(other) and hash(self) == hash(other)
 116
 117    @property
 118    def hashable_args(self) -> t.Any:
 119        return frozenset(
 120            (k, tuple(_norm_arg(a) for a in v) if type(v) is list else _norm_arg(v))
 121            for k, v in self.args.items()
 122            if not (v is None or v is False or (type(v) is list and not v))
 123        )
 124
 125    def __hash__(self) -> int:
 126        if self._hash is not None:
 127            return self._hash
 128
 129        return hash((self.__class__, self.hashable_args))
 130
 131    @property
 132    def this(self) -> t.Any:
 133        """
 134        Retrieves the argument with key "this".
 135        """
 136        return self.args.get("this")
 137
 138    @property
 139    def expression(self) -> t.Any:
 140        """
 141        Retrieves the argument with key "expression".
 142        """
 143        return self.args.get("expression")
 144
 145    @property
 146    def expressions(self) -> t.List[t.Any]:
 147        """
 148        Retrieves the argument with key "expressions".
 149        """
 150        return self.args.get("expressions") or []
 151
 152    def text(self, key) -> str:
 153        """
 154        Returns a textual representation of the argument corresponding to "key". This can only be used
 155        for args that are strings or leaf Expression instances, such as identifiers and literals.
 156        """
 157        field = self.args.get(key)
 158        if isinstance(field, str):
 159            return field
 160        if isinstance(field, (Identifier, Literal, Var)):
 161            return field.this
 162        if isinstance(field, (Star, Null)):
 163            return field.name
 164        return ""
 165
 166    @property
 167    def is_string(self) -> bool:
 168        """
 169        Checks whether a Literal expression is a string.
 170        """
 171        return isinstance(self, Literal) and self.args["is_string"]
 172
 173    @property
 174    def is_number(self) -> bool:
 175        """
 176        Checks whether a Literal expression is a number.
 177        """
 178        return (isinstance(self, Literal) and not self.args["is_string"]) or (
 179            isinstance(self, Neg) and self.this.is_number
 180        )
 181
 182    def to_py(self) -> t.Any:
 183        """
 184        Returns a Python object equivalent of the SQL node.
 185        """
 186        raise ValueError(f"{self} cannot be converted to a Python object.")
 187
 188    @property
 189    def is_int(self) -> bool:
 190        """
 191        Checks whether an expression is an integer.
 192        """
 193        return self.is_number and isinstance(self.to_py(), int)
 194
 195    @property
 196    def is_star(self) -> bool:
 197        """Checks whether an expression is a star."""
 198        return isinstance(self, Star) or (isinstance(self, Column) and isinstance(self.this, Star))
 199
 200    @property
 201    def alias(self) -> str:
 202        """
 203        Returns the alias of the expression, or an empty string if it's not aliased.
 204        """
 205        if isinstance(self.args.get("alias"), TableAlias):
 206            return self.args["alias"].name
 207        return self.text("alias")
 208
 209    @property
 210    def alias_column_names(self) -> t.List[str]:
 211        table_alias = self.args.get("alias")
 212        if not table_alias:
 213            return []
 214        return [c.name for c in table_alias.args.get("columns") or []]
 215
 216    @property
 217    def name(self) -> str:
 218        return self.text("this")
 219
 220    @property
 221    def alias_or_name(self) -> str:
 222        return self.alias or self.name
 223
 224    @property
 225    def output_name(self) -> str:
 226        """
 227        Name of the output column if this expression is a selection.
 228
 229        If the Expression has no output name, an empty string is returned.
 230
 231        Example:
 232            >>> from sqlglot import parse_one
 233            >>> parse_one("SELECT a").expressions[0].output_name
 234            'a'
 235            >>> parse_one("SELECT b AS c").expressions[0].output_name
 236            'c'
 237            >>> parse_one("SELECT 1 + 2").expressions[0].output_name
 238            ''
 239        """
 240        return ""
 241
 242    @property
 243    def type(self) -> t.Optional[DataType]:
 244        return self._type
 245
 246    @type.setter
 247    def type(self, dtype: t.Optional[DataType | DataType.Type | str]) -> None:
 248        if dtype and not isinstance(dtype, DataType):
 249            dtype = DataType.build(dtype)
 250        self._type = dtype  # type: ignore
 251
 252    def is_type(self, *dtypes) -> bool:
 253        return self.type is not None and self.type.is_type(*dtypes)
 254
 255    def is_leaf(self) -> bool:
 256        return not any(isinstance(v, (Expression, list)) for v in self.args.values())
 257
 258    @property
 259    def meta(self) -> t.Dict[str, t.Any]:
 260        if self._meta is None:
 261            self._meta = {}
 262        return self._meta
 263
 264    def __deepcopy__(self, memo):
 265        root = self.__class__()
 266        stack = [(self, root)]
 267
 268        while stack:
 269            node, copy = stack.pop()
 270
 271            if node.comments is not None:
 272                copy.comments = deepcopy(node.comments)
 273            if node._type is not None:
 274                copy._type = deepcopy(node._type)
 275            if node._meta is not None:
 276                copy._meta = deepcopy(node._meta)
 277            if node._hash is not None:
 278                copy._hash = node._hash
 279
 280            for k, vs in node.args.items():
 281                if hasattr(vs, "parent"):
 282                    stack.append((vs, vs.__class__()))
 283                    copy.set(k, stack[-1][-1])
 284                elif type(vs) is list:
 285                    copy.args[k] = []
 286
 287                    for v in vs:
 288                        if hasattr(v, "parent"):
 289                            stack.append((v, v.__class__()))
 290                            copy.append(k, stack[-1][-1])
 291                        else:
 292                            copy.append(k, v)
 293                else:
 294                    copy.args[k] = vs
 295
 296        return root
 297
 298    def copy(self):
 299        """
 300        Returns a deep copy of the expression.
 301        """
 302        return deepcopy(self)
 303
 304    def add_comments(self, comments: t.Optional[t.List[str]] = None) -> None:
 305        if self.comments is None:
 306            self.comments = []
 307
 308        if comments:
 309            for comment in comments:
 310                _, *meta = comment.split(SQLGLOT_META)
 311                if meta:
 312                    for kv in "".join(meta).split(","):
 313                        k, *v = kv.split("=")
 314                        value = v[0].strip() if v else True
 315                        self.meta[k.strip()] = value
 316                self.comments.append(comment)
 317
 318    def pop_comments(self) -> t.List[str]:
 319        comments = self.comments or []
 320        self.comments = None
 321        return comments
 322
 323    def append(self, arg_key: str, value: t.Any) -> None:
 324        """
 325        Appends value to arg_key if it's a list or sets it as a new list.
 326
 327        Args:
 328            arg_key (str): name of the list expression arg
 329            value (Any): value to append to the list
 330        """
 331        if type(self.args.get(arg_key)) is not list:
 332            self.args[arg_key] = []
 333        self._set_parent(arg_key, value)
 334        values = self.args[arg_key]
 335        if hasattr(value, "parent"):
 336            value.index = len(values)
 337        values.append(value)
 338
 339    def set(
 340        self,
 341        arg_key: str,
 342        value: t.Any,
 343        index: t.Optional[int] = None,
 344        overwrite: bool = True,
 345    ) -> None:
 346        """
 347        Sets arg_key to value.
 348
 349        Args:
 350            arg_key: name of the expression arg.
 351            value: value to set the arg to.
 352            index: if the arg is a list, this specifies what position to add the value in it.
 353            overwrite: assuming an index is given, this determines whether to overwrite the
 354                list entry instead of only inserting a new value (i.e., like list.insert).
 355        """
 356        if index is not None:
 357            expressions = self.args.get(arg_key) or []
 358
 359            if seq_get(expressions, index) is None:
 360                return
 361            if value is None:
 362                expressions.pop(index)
 363                for v in expressions[index:]:
 364                    v.index = v.index - 1
 365                return
 366
 367            if isinstance(value, list):
 368                expressions.pop(index)
 369                expressions[index:index] = value
 370            elif overwrite:
 371                expressions[index] = value
 372            else:
 373                expressions.insert(index, value)
 374
 375            value = expressions
 376        elif value is None:
 377            self.args.pop(arg_key, None)
 378            return
 379
 380        self.args[arg_key] = value
 381        self._set_parent(arg_key, value, index)
 382
 383    def _set_parent(self, arg_key: str, value: t.Any, index: t.Optional[int] = None) -> None:
 384        if hasattr(value, "parent"):
 385            value.parent = self
 386            value.arg_key = arg_key
 387            value.index = index
 388        elif type(value) is list:
 389            for index, v in enumerate(value):
 390                if hasattr(v, "parent"):
 391                    v.parent = self
 392                    v.arg_key = arg_key
 393                    v.index = index
 394
 395    @property
 396    def depth(self) -> int:
 397        """
 398        Returns the depth of this tree.
 399        """
 400        if self.parent:
 401            return self.parent.depth + 1
 402        return 0
 403
 404    def iter_expressions(self, reverse: bool = False) -> t.Iterator[Expression]:
 405        """Yields the key and expression for all arguments, exploding list args."""
 406        # remove tuple when python 3.7 is deprecated
 407        for vs in reversed(tuple(self.args.values())) if reverse else self.args.values():
 408            if type(vs) is list:
 409                for v in reversed(vs) if reverse else vs:
 410                    if hasattr(v, "parent"):
 411                        yield v
 412            else:
 413                if hasattr(vs, "parent"):
 414                    yield vs
 415
 416    def find(self, *expression_types: t.Type[E], bfs: bool = True) -> t.Optional[E]:
 417        """
 418        Returns the first node in this tree which matches at least one of
 419        the specified types.
 420
 421        Args:
 422            expression_types: the expression type(s) to match.
 423            bfs: whether to search the AST using the BFS algorithm (DFS is used if false).
 424
 425        Returns:
 426            The node which matches the criteria or None if no such node was found.
 427        """
 428        return next(self.find_all(*expression_types, bfs=bfs), None)
 429
 430    def find_all(self, *expression_types: t.Type[E], bfs: bool = True) -> t.Iterator[E]:
 431        """
 432        Returns a generator object which visits all nodes in this tree and only
 433        yields those that match at least one of the specified expression types.
 434
 435        Args:
 436            expression_types: the expression type(s) to match.
 437            bfs: whether to search the AST using the BFS algorithm (DFS is used if false).
 438
 439        Returns:
 440            The generator object.
 441        """
 442        for expression in self.walk(bfs=bfs):
 443            if isinstance(expression, expression_types):
 444                yield expression
 445
 446    def find_ancestor(self, *expression_types: t.Type[E]) -> t.Optional[E]:
 447        """
 448        Returns a nearest parent matching expression_types.
 449
 450        Args:
 451            expression_types: the expression type(s) to match.
 452
 453        Returns:
 454            The parent node.
 455        """
 456        ancestor = self.parent
 457        while ancestor and not isinstance(ancestor, expression_types):
 458            ancestor = ancestor.parent
 459        return ancestor  # type: ignore
 460
 461    @property
 462    def parent_select(self) -> t.Optional[Select]:
 463        """
 464        Returns the parent select statement.
 465        """
 466        return self.find_ancestor(Select)
 467
 468    @property
 469    def same_parent(self) -> bool:
 470        """Returns if the parent is the same class as itself."""
 471        return type(self.parent) is self.__class__
 472
 473    def root(self) -> Expression:
 474        """
 475        Returns the root expression of this tree.
 476        """
 477        expression = self
 478        while expression.parent:
 479            expression = expression.parent
 480        return expression
 481
 482    def walk(
 483        self, bfs: bool = True, prune: t.Optional[t.Callable[[Expression], bool]] = None
 484    ) -> t.Iterator[Expression]:
 485        """
 486        Returns a generator object which visits all nodes in this tree.
 487
 488        Args:
 489            bfs: if set to True the BFS traversal order will be applied,
 490                otherwise the DFS traversal will be used instead.
 491            prune: callable that returns True if the generator should stop traversing
 492                this branch of the tree.
 493
 494        Returns:
 495            the generator object.
 496        """
 497        if bfs:
 498            yield from self.bfs(prune=prune)
 499        else:
 500            yield from self.dfs(prune=prune)
 501
 502    def dfs(
 503        self, prune: t.Optional[t.Callable[[Expression], bool]] = None
 504    ) -> t.Iterator[Expression]:
 505        """
 506        Returns a generator object which visits all nodes in this tree in
 507        the DFS (Depth-first) order.
 508
 509        Returns:
 510            The generator object.
 511        """
 512        stack = [self]
 513
 514        while stack:
 515            node = stack.pop()
 516
 517            yield node
 518
 519            if prune and prune(node):
 520                continue
 521
 522            for v in node.iter_expressions(reverse=True):
 523                stack.append(v)
 524
 525    def bfs(
 526        self, prune: t.Optional[t.Callable[[Expression], bool]] = None
 527    ) -> t.Iterator[Expression]:
 528        """
 529        Returns a generator object which visits all nodes in this tree in
 530        the BFS (Breadth-first) order.
 531
 532        Returns:
 533            The generator object.
 534        """
 535        queue = deque([self])
 536
 537        while queue:
 538            node = queue.popleft()
 539
 540            yield node
 541
 542            if prune and prune(node):
 543                continue
 544
 545            for v in node.iter_expressions():
 546                queue.append(v)
 547
 548    def unnest(self):
 549        """
 550        Returns the first non parenthesis child or self.
 551        """
 552        expression = self
 553        while type(expression) is Paren:
 554            expression = expression.this
 555        return expression
 556
 557    def unalias(self):
 558        """
 559        Returns the inner expression if this is an Alias.
 560        """
 561        if isinstance(self, Alias):
 562            return self.this
 563        return self
 564
 565    def unnest_operands(self):
 566        """
 567        Returns unnested operands as a tuple.
 568        """
 569        return tuple(arg.unnest() for arg in self.iter_expressions())
 570
 571    def flatten(self, unnest=True):
 572        """
 573        Returns a generator which yields child nodes whose parents are the same class.
 574
 575        A AND B AND C -> [A, B, C]
 576        """
 577        for node in self.dfs(prune=lambda n: n.parent and type(n) is not self.__class__):
 578            if type(node) is not self.__class__:
 579                yield node.unnest() if unnest and not isinstance(node, Subquery) else node
 580
 581    def __str__(self) -> str:
 582        return self.sql()
 583
 584    def __repr__(self) -> str:
 585        return _to_s(self)
 586
 587    def to_s(self) -> str:
 588        """
 589        Same as __repr__, but includes additional information which can be useful
 590        for debugging, like empty or missing args and the AST nodes' object IDs.
 591        """
 592        return _to_s(self, verbose=True)
 593
 594    def sql(self, dialect: DialectType = None, **opts) -> str:
 595        """
 596        Returns SQL string representation of this tree.
 597
 598        Args:
 599            dialect: the dialect of the output SQL string (eg. "spark", "hive", "presto", "mysql").
 600            opts: other `sqlglot.generator.Generator` options.
 601
 602        Returns:
 603            The SQL string.
 604        """
 605        from sqlglot.dialects import Dialect
 606
 607        return Dialect.get_or_raise(dialect).generate(self, **opts)
 608
 609    def transform(self, fun: t.Callable, *args: t.Any, copy: bool = True, **kwargs) -> Expression:
 610        """
 611        Visits all tree nodes (excluding already transformed ones)
 612        and applies the given transformation function to each node.
 613
 614        Args:
 615            fun: a function which takes a node as an argument and returns a
 616                new transformed node or the same node without modifications. If the function
 617                returns None, then the corresponding node will be removed from the syntax tree.
 618            copy: if set to True a new tree instance is constructed, otherwise the tree is
 619                modified in place.
 620
 621        Returns:
 622            The transformed tree.
 623        """
 624        root = None
 625        new_node = None
 626
 627        for node in (self.copy() if copy else self).dfs(prune=lambda n: n is not new_node):
 628            parent, arg_key, index = node.parent, node.arg_key, node.index
 629            new_node = fun(node, *args, **kwargs)
 630
 631            if not root:
 632                root = new_node
 633            elif new_node is not node:
 634                parent.set(arg_key, new_node, index)
 635
 636        assert root
 637        return root.assert_is(Expression)
 638
 639    @t.overload
 640    def replace(self, expression: E) -> E: ...
 641
 642    @t.overload
 643    def replace(self, expression: None) -> None: ...
 644
 645    def replace(self, expression):
 646        """
 647        Swap out this expression with a new expression.
 648
 649        For example::
 650
 651            >>> tree = Select().select("x").from_("tbl")
 652            >>> tree.find(Column).replace(column("y"))
 653            Column(
 654              this=Identifier(this=y, quoted=False))
 655            >>> tree.sql()
 656            'SELECT y FROM tbl'
 657
 658        Args:
 659            expression: new node
 660
 661        Returns:
 662            The new expression or expressions.
 663        """
 664        parent = self.parent
 665
 666        if not parent or parent is expression:
 667            return expression
 668
 669        key = self.arg_key
 670        value = parent.args.get(key)
 671
 672        if type(expression) is list and isinstance(value, Expression):
 673            # We are trying to replace an Expression with a list, so it's assumed that
 674            # the intention was to really replace the parent of this expression.
 675            value.parent.replace(expression)
 676        else:
 677            parent.set(key, expression, self.index)
 678
 679        if expression is not self:
 680            self.parent = None
 681            self.arg_key = None
 682            self.index = None
 683
 684        return expression
 685
 686    def pop(self: E) -> E:
 687        """
 688        Remove this expression from its AST.
 689
 690        Returns:
 691            The popped expression.
 692        """
 693        self.replace(None)
 694        return self
 695
 696    def assert_is(self, type_: t.Type[E]) -> E:
 697        """
 698        Assert that this `Expression` is an instance of `type_`.
 699
 700        If it is NOT an instance of `type_`, this raises an assertion error.
 701        Otherwise, this returns this expression.
 702
 703        Examples:
 704            This is useful for type security in chained expressions:
 705
 706            >>> import sqlglot
 707            >>> sqlglot.parse_one("SELECT x from y").assert_is(Select).select("z").sql()
 708            'SELECT x, z FROM y'
 709        """
 710        if not isinstance(self, type_):
 711            raise AssertionError(f"{self} is not {type_}.")
 712        return self
 713
 714    def error_messages(self, args: t.Optional[t.Sequence] = None) -> t.List[str]:
 715        """
 716        Checks if this expression is valid (e.g. all mandatory args are set).
 717
 718        Args:
 719            args: a sequence of values that were used to instantiate a Func expression. This is used
 720                to check that the provided arguments don't exceed the function argument limit.
 721
 722        Returns:
 723            A list of error messages for all possible errors that were found.
 724        """
 725        errors: t.List[str] = []
 726
 727        for k in self.args:
 728            if k not in self.arg_types:
 729                errors.append(f"Unexpected keyword: '{k}' for {self.__class__}")
 730        for k, mandatory in self.arg_types.items():
 731            v = self.args.get(k)
 732            if mandatory and (v is None or (isinstance(v, list) and not v)):
 733                errors.append(f"Required keyword: '{k}' missing for {self.__class__}")
 734
 735        if (
 736            args
 737            and isinstance(self, Func)
 738            and len(args) > len(self.arg_types)
 739            and not self.is_var_len_args
 740        ):
 741            errors.append(
 742                f"The number of provided arguments ({len(args)}) is greater than "
 743                f"the maximum number of supported arguments ({len(self.arg_types)})"
 744            )
 745
 746        return errors
 747
 748    def dump(self):
 749        """
 750        Dump this Expression to a JSON-serializable dict.
 751        """
 752        from sqlglot.serde import dump
 753
 754        return dump(self)
 755
 756    @classmethod
 757    def load(cls, obj):
 758        """
 759        Load a dict (as returned by `Expression.dump`) into an Expression instance.
 760        """
 761        from sqlglot.serde import load
 762
 763        return load(obj)
 764
 765    def and_(
 766        self,
 767        *expressions: t.Optional[ExpOrStr],
 768        dialect: DialectType = None,
 769        copy: bool = True,
 770        **opts,
 771    ) -> Condition:
 772        """
 773        AND this condition with one or multiple expressions.
 774
 775        Example:
 776            >>> condition("x=1").and_("y=1").sql()
 777            'x = 1 AND y = 1'
 778
 779        Args:
 780            *expressions: the SQL code strings to parse.
 781                If an `Expression` instance is passed, it will be used as-is.
 782            dialect: the dialect used to parse the input expression.
 783            copy: whether to copy the involved expressions (only applies to Expressions).
 784            opts: other options to use to parse the input expressions.
 785
 786        Returns:
 787            The new And condition.
 788        """
 789        return and_(self, *expressions, dialect=dialect, copy=copy, **opts)
 790
 791    def or_(
 792        self,
 793        *expressions: t.Optional[ExpOrStr],
 794        dialect: DialectType = None,
 795        copy: bool = True,
 796        **opts,
 797    ) -> Condition:
 798        """
 799        OR this condition with one or multiple expressions.
 800
 801        Example:
 802            >>> condition("x=1").or_("y=1").sql()
 803            'x = 1 OR y = 1'
 804
 805        Args:
 806            *expressions: the SQL code strings to parse.
 807                If an `Expression` instance is passed, it will be used as-is.
 808            dialect: the dialect used to parse the input expression.
 809            copy: whether to copy the involved expressions (only applies to Expressions).
 810            opts: other options to use to parse the input expressions.
 811
 812        Returns:
 813            The new Or condition.
 814        """
 815        return or_(self, *expressions, dialect=dialect, copy=copy, **opts)
 816
 817    def not_(self, copy: bool = True):
 818        """
 819        Wrap this condition with NOT.
 820
 821        Example:
 822            >>> condition("x=1").not_().sql()
 823            'NOT x = 1'
 824
 825        Args:
 826            copy: whether to copy this object.
 827
 828        Returns:
 829            The new Not instance.
 830        """
 831        return not_(self, copy=copy)
 832
 833    def as_(
 834        self,
 835        alias: str | Identifier,
 836        quoted: t.Optional[bool] = None,
 837        dialect: DialectType = None,
 838        copy: bool = True,
 839        **opts,
 840    ) -> Alias:
 841        return alias_(self, alias, quoted=quoted, dialect=dialect, copy=copy, **opts)
 842
 843    def _binop(self, klass: t.Type[E], other: t.Any, reverse: bool = False) -> E:
 844        this = self.copy()
 845        other = convert(other, copy=True)
 846        if not isinstance(this, klass) and not isinstance(other, klass):
 847            this = _wrap(this, Binary)
 848            other = _wrap(other, Binary)
 849        if reverse:
 850            return klass(this=other, expression=this)
 851        return klass(this=this, expression=other)
 852
 853    def __getitem__(self, other: ExpOrStr | t.Tuple[ExpOrStr]) -> Bracket:
 854        return Bracket(
 855            this=self.copy(), expressions=[convert(e, copy=True) for e in ensure_list(other)]
 856        )
 857
 858    def __iter__(self) -> t.Iterator:
 859        if "expressions" in self.arg_types:
 860            return iter(self.args.get("expressions") or [])
 861        # We define this because __getitem__ converts Expression into an iterable, which is
 862        # problematic because one can hit infinite loops if they do "for x in some_expr: ..."
 863        # See: https://peps.python.org/pep-0234/
 864        raise TypeError(f"'{self.__class__.__name__}' object is not iterable")
 865
 866    def isin(
 867        self,
 868        *expressions: t.Any,
 869        query: t.Optional[ExpOrStr] = None,
 870        unnest: t.Optional[ExpOrStr] | t.Collection[ExpOrStr] = None,
 871        copy: bool = True,
 872        **opts,
 873    ) -> In:
 874        subquery = maybe_parse(query, copy=copy, **opts) if query else None
 875        if subquery and not isinstance(subquery, Subquery):
 876            subquery = subquery.subquery(copy=False)
 877
 878        return In(
 879            this=maybe_copy(self, copy),
 880            expressions=[convert(e, copy=copy) for e in expressions],
 881            query=subquery,
 882            unnest=(
 883                Unnest(
 884                    expressions=[
 885                        maybe_parse(t.cast(ExpOrStr, e), copy=copy, **opts)
 886                        for e in ensure_list(unnest)
 887                    ]
 888                )
 889                if unnest
 890                else None
 891            ),
 892        )
 893
 894    def between(self, low: t.Any, high: t.Any, copy: bool = True, **opts) -> Between:
 895        return Between(
 896            this=maybe_copy(self, copy),
 897            low=convert(low, copy=copy, **opts),
 898            high=convert(high, copy=copy, **opts),
 899        )
 900
 901    def is_(self, other: ExpOrStr) -> Is:
 902        return self._binop(Is, other)
 903
 904    def like(self, other: ExpOrStr) -> Like:
 905        return self._binop(Like, other)
 906
 907    def ilike(self, other: ExpOrStr) -> ILike:
 908        return self._binop(ILike, other)
 909
 910    def eq(self, other: t.Any) -> EQ:
 911        return self._binop(EQ, other)
 912
 913    def neq(self, other: t.Any) -> NEQ:
 914        return self._binop(NEQ, other)
 915
 916    def rlike(self, other: ExpOrStr) -> RegexpLike:
 917        return self._binop(RegexpLike, other)
 918
 919    def div(self, other: ExpOrStr, typed: bool = False, safe: bool = False) -> Div:
 920        div = self._binop(Div, other)
 921        div.args["typed"] = typed
 922        div.args["safe"] = safe
 923        return div
 924
 925    def asc(self, nulls_first: bool = True) -> Ordered:
 926        return Ordered(this=self.copy(), nulls_first=nulls_first)
 927
 928    def desc(self, nulls_first: bool = False) -> Ordered:
 929        return Ordered(this=self.copy(), desc=True, nulls_first=nulls_first)
 930
 931    def __lt__(self, other: t.Any) -> LT:
 932        return self._binop(LT, other)
 933
 934    def __le__(self, other: t.Any) -> LTE:
 935        return self._binop(LTE, other)
 936
 937    def __gt__(self, other: t.Any) -> GT:
 938        return self._binop(GT, other)
 939
 940    def __ge__(self, other: t.Any) -> GTE:
 941        return self._binop(GTE, other)
 942
 943    def __add__(self, other: t.Any) -> Add:
 944        return self._binop(Add, other)
 945
 946    def __radd__(self, other: t.Any) -> Add:
 947        return self._binop(Add, other, reverse=True)
 948
 949    def __sub__(self, other: t.Any) -> Sub:
 950        return self._binop(Sub, other)
 951
 952    def __rsub__(self, other: t.Any) -> Sub:
 953        return self._binop(Sub, other, reverse=True)
 954
 955    def __mul__(self, other: t.Any) -> Mul:
 956        return self._binop(Mul, other)
 957
 958    def __rmul__(self, other: t.Any) -> Mul:
 959        return self._binop(Mul, other, reverse=True)
 960
 961    def __truediv__(self, other: t.Any) -> Div:
 962        return self._binop(Div, other)
 963
 964    def __rtruediv__(self, other: t.Any) -> Div:
 965        return self._binop(Div, other, reverse=True)
 966
 967    def __floordiv__(self, other: t.Any) -> IntDiv:
 968        return self._binop(IntDiv, other)
 969
 970    def __rfloordiv__(self, other: t.Any) -> IntDiv:
 971        return self._binop(IntDiv, other, reverse=True)
 972
 973    def __mod__(self, other: t.Any) -> Mod:
 974        return self._binop(Mod, other)
 975
 976    def __rmod__(self, other: t.Any) -> Mod:
 977        return self._binop(Mod, other, reverse=True)
 978
 979    def __pow__(self, other: t.Any) -> Pow:
 980        return self._binop(Pow, other)
 981
 982    def __rpow__(self, other: t.Any) -> Pow:
 983        return self._binop(Pow, other, reverse=True)
 984
 985    def __and__(self, other: t.Any) -> And:
 986        return self._binop(And, other)
 987
 988    def __rand__(self, other: t.Any) -> And:
 989        return self._binop(And, other, reverse=True)
 990
 991    def __or__(self, other: t.Any) -> Or:
 992        return self._binop(Or, other)
 993
 994    def __ror__(self, other: t.Any) -> Or:
 995        return self._binop(Or, other, reverse=True)
 996
 997    def __neg__(self) -> Neg:
 998        return Neg(this=_wrap(self.copy(), Binary))
 999
1000    def __invert__(self) -> Not:
1001        return not_(self.copy())
1002
1003
1004IntoType = t.Union[
1005    str,
1006    t.Type[Expression],
1007    t.Collection[t.Union[str, t.Type[Expression]]],
1008]
1009ExpOrStr = t.Union[str, Expression]
1010
1011
1012class Condition(Expression):
1013    """Logical conditions like x AND y, or simply x"""
1014
1015
1016class Predicate(Condition):
1017    """Relationships like x = y, x > 1, x >= y."""
1018
1019
1020class DerivedTable(Expression):
1021    @property
1022    def selects(self) -> t.List[Expression]:
1023        return self.this.selects if isinstance(self.this, Query) else []
1024
1025    @property
1026    def named_selects(self) -> t.List[str]:
1027        return [select.output_name for select in self.selects]
1028
1029
1030class Query(Expression):
1031    def subquery(self, alias: t.Optional[ExpOrStr] = None, copy: bool = True) -> Subquery:
1032        """
1033        Returns a `Subquery` that wraps around this query.
1034
1035        Example:
1036            >>> subquery = Select().select("x").from_("tbl").subquery()
1037            >>> Select().select("x").from_(subquery).sql()
1038            'SELECT x FROM (SELECT x FROM tbl)'
1039
1040        Args:
1041            alias: an optional alias for the subquery.
1042            copy: if `False`, modify this expression instance in-place.
1043        """
1044        instance = maybe_copy(self, copy)
1045        if not isinstance(alias, Expression):
1046            alias = TableAlias(this=to_identifier(alias)) if alias else None
1047
1048        return Subquery(this=instance, alias=alias)
1049
1050    def limit(
1051        self: Q, expression: ExpOrStr | int, dialect: DialectType = None, copy: bool = True, **opts
1052    ) -> Q:
1053        """
1054        Adds a LIMIT clause to this query.
1055
1056        Example:
1057            >>> select("1").union(select("1")).limit(1).sql()
1058            'SELECT 1 UNION SELECT 1 LIMIT 1'
1059
1060        Args:
1061            expression: the SQL code string to parse.
1062                This can also be an integer.
1063                If a `Limit` instance is passed, it will be used as-is.
1064                If another `Expression` instance is passed, it will be wrapped in a `Limit`.
1065            dialect: the dialect used to parse the input expression.
1066            copy: if `False`, modify this expression instance in-place.
1067            opts: other options to use to parse the input expressions.
1068
1069        Returns:
1070            A limited Select expression.
1071        """
1072        return _apply_builder(
1073            expression=expression,
1074            instance=self,
1075            arg="limit",
1076            into=Limit,
1077            prefix="LIMIT",
1078            dialect=dialect,
1079            copy=copy,
1080            into_arg="expression",
1081            **opts,
1082        )
1083
1084    def offset(
1085        self: Q, expression: ExpOrStr | int, dialect: DialectType = None, copy: bool = True, **opts
1086    ) -> Q:
1087        """
1088        Set the OFFSET expression.
1089
1090        Example:
1091            >>> Select().from_("tbl").select("x").offset(10).sql()
1092            'SELECT x FROM tbl OFFSET 10'
1093
1094        Args:
1095            expression: the SQL code string to parse.
1096                This can also be an integer.
1097                If a `Offset` instance is passed, this is used as-is.
1098                If another `Expression` instance is passed, it will be wrapped in a `Offset`.
1099            dialect: the dialect used to parse the input expression.
1100            copy: if `False`, modify this expression instance in-place.
1101            opts: other options to use to parse the input expressions.
1102
1103        Returns:
1104            The modified Select expression.
1105        """
1106        return _apply_builder(
1107            expression=expression,
1108            instance=self,
1109            arg="offset",
1110            into=Offset,
1111            prefix="OFFSET",
1112            dialect=dialect,
1113            copy=copy,
1114            into_arg="expression",
1115            **opts,
1116        )
1117
1118    def order_by(
1119        self: Q,
1120        *expressions: t.Optional[ExpOrStr],
1121        append: bool = True,
1122        dialect: DialectType = None,
1123        copy: bool = True,
1124        **opts,
1125    ) -> Q:
1126        """
1127        Set the ORDER BY expression.
1128
1129        Example:
1130            >>> Select().from_("tbl").select("x").order_by("x DESC").sql()
1131            'SELECT x FROM tbl ORDER BY x DESC'
1132
1133        Args:
1134            *expressions: the SQL code strings to parse.
1135                If a `Group` instance is passed, this is used as-is.
1136                If another `Expression` instance is passed, it will be wrapped in a `Order`.
1137            append: if `True`, add to any existing expressions.
1138                Otherwise, this flattens all the `Order` expression into a single expression.
1139            dialect: the dialect used to parse the input expression.
1140            copy: if `False`, modify this expression instance in-place.
1141            opts: other options to use to parse the input expressions.
1142
1143        Returns:
1144            The modified Select expression.
1145        """
1146        return _apply_child_list_builder(
1147            *expressions,
1148            instance=self,
1149            arg="order",
1150            append=append,
1151            copy=copy,
1152            prefix="ORDER BY",
1153            into=Order,
1154            dialect=dialect,
1155            **opts,
1156        )
1157
1158    @property
1159    def ctes(self) -> t.List[CTE]:
1160        """Returns a list of all the CTEs attached to this query."""
1161        with_ = self.args.get("with")
1162        return with_.expressions if with_ else []
1163
1164    @property
1165    def selects(self) -> t.List[Expression]:
1166        """Returns the query's projections."""
1167        raise NotImplementedError("Query objects must implement `selects`")
1168
1169    @property
1170    def named_selects(self) -> t.List[str]:
1171        """Returns the output names of the query's projections."""
1172        raise NotImplementedError("Query objects must implement `named_selects`")
1173
1174    def select(
1175        self: Q,
1176        *expressions: t.Optional[ExpOrStr],
1177        append: bool = True,
1178        dialect: DialectType = None,
1179        copy: bool = True,
1180        **opts,
1181    ) -> Q:
1182        """
1183        Append to or set the SELECT expressions.
1184
1185        Example:
1186            >>> Select().select("x", "y").sql()
1187            'SELECT x, y'
1188
1189        Args:
1190            *expressions: the SQL code strings to parse.
1191                If an `Expression` instance is passed, it will be used as-is.
1192            append: if `True`, add to any existing expressions.
1193                Otherwise, this resets the expressions.
1194            dialect: the dialect used to parse the input expressions.
1195            copy: if `False`, modify this expression instance in-place.
1196            opts: other options to use to parse the input expressions.
1197
1198        Returns:
1199            The modified Query expression.
1200        """
1201        raise NotImplementedError("Query objects must implement `select`")
1202
1203    def with_(
1204        self: Q,
1205        alias: ExpOrStr,
1206        as_: ExpOrStr,
1207        recursive: t.Optional[bool] = None,
1208        materialized: t.Optional[bool] = None,
1209        append: bool = True,
1210        dialect: DialectType = None,
1211        copy: bool = True,
1212        **opts,
1213    ) -> Q:
1214        """
1215        Append to or set the common table expressions.
1216
1217        Example:
1218            >>> Select().with_("tbl2", as_="SELECT * FROM tbl").select("x").from_("tbl2").sql()
1219            'WITH tbl2 AS (SELECT * FROM tbl) SELECT x FROM tbl2'
1220
1221        Args:
1222            alias: the SQL code string to parse as the table name.
1223                If an `Expression` instance is passed, this is used as-is.
1224            as_: the SQL code string to parse as the table expression.
1225                If an `Expression` instance is passed, it will be used as-is.
1226            recursive: set the RECURSIVE part of the expression. Defaults to `False`.
1227            materialized: set the MATERIALIZED part of the expression.
1228            append: if `True`, add to any existing expressions.
1229                Otherwise, this resets the expressions.
1230            dialect: the dialect used to parse the input expression.
1231            copy: if `False`, modify this expression instance in-place.
1232            opts: other options to use to parse the input expressions.
1233
1234        Returns:
1235            The modified expression.
1236        """
1237        return _apply_cte_builder(
1238            self,
1239            alias,
1240            as_,
1241            recursive=recursive,
1242            materialized=materialized,
1243            append=append,
1244            dialect=dialect,
1245            copy=copy,
1246            **opts,
1247        )
1248
1249    def union(
1250        self, expression: ExpOrStr, distinct: bool = True, dialect: DialectType = None, **opts
1251    ) -> Union:
1252        """
1253        Builds a UNION expression.
1254
1255        Example:
1256            >>> import sqlglot
1257            >>> sqlglot.parse_one("SELECT * FROM foo").union("SELECT * FROM bla").sql()
1258            'SELECT * FROM foo UNION SELECT * FROM bla'
1259
1260        Args:
1261            expression: the SQL code string.
1262                If an `Expression` instance is passed, it will be used as-is.
1263            distinct: set the DISTINCT flag if and only if this is true.
1264            dialect: the dialect used to parse the input expression.
1265            opts: other options to use to parse the input expressions.
1266
1267        Returns:
1268            The new Union expression.
1269        """
1270        return union(left=self, right=expression, distinct=distinct, dialect=dialect, **opts)
1271
1272    def intersect(
1273        self, expression: ExpOrStr, distinct: bool = True, dialect: DialectType = None, **opts
1274    ) -> Intersect:
1275        """
1276        Builds an INTERSECT expression.
1277
1278        Example:
1279            >>> import sqlglot
1280            >>> sqlglot.parse_one("SELECT * FROM foo").intersect("SELECT * FROM bla").sql()
1281            'SELECT * FROM foo INTERSECT SELECT * FROM bla'
1282
1283        Args:
1284            expression: the SQL code string.
1285                If an `Expression` instance is passed, it will be used as-is.
1286            distinct: set the DISTINCT flag if and only if this is true.
1287            dialect: the dialect used to parse the input expression.
1288            opts: other options to use to parse the input expressions.
1289
1290        Returns:
1291            The new Intersect expression.
1292        """
1293        return intersect(left=self, right=expression, distinct=distinct, dialect=dialect, **opts)
1294
1295    def except_(
1296        self, expression: ExpOrStr, distinct: bool = True, dialect: DialectType = None, **opts
1297    ) -> Except:
1298        """
1299        Builds an EXCEPT expression.
1300
1301        Example:
1302            >>> import sqlglot
1303            >>> sqlglot.parse_one("SELECT * FROM foo").except_("SELECT * FROM bla").sql()
1304            'SELECT * FROM foo EXCEPT SELECT * FROM bla'
1305
1306        Args:
1307            expression: the SQL code string.
1308                If an `Expression` instance is passed, it will be used as-is.
1309            distinct: set the DISTINCT flag if and only if this is true.
1310            dialect: the dialect used to parse the input expression.
1311            opts: other options to use to parse the input expressions.
1312
1313        Returns:
1314            The new Except expression.
1315        """
1316        return except_(left=self, right=expression, distinct=distinct, dialect=dialect, **opts)
1317
1318
1319class UDTF(DerivedTable):
1320    @property
1321    def selects(self) -> t.List[Expression]:
1322        alias = self.args.get("alias")
1323        return alias.columns if alias else []
1324
1325
1326class Cache(Expression):
1327    arg_types = {
1328        "this": True,
1329        "lazy": False,
1330        "options": False,
1331        "expression": False,
1332    }
1333
1334
1335class Uncache(Expression):
1336    arg_types = {"this": True, "exists": False}
1337
1338
1339class Refresh(Expression):
1340    pass
1341
1342
1343class DDL(Expression):
1344    @property
1345    def ctes(self) -> t.List[CTE]:
1346        """Returns a list of all the CTEs attached to this statement."""
1347        with_ = self.args.get("with")
1348        return with_.expressions if with_ else []
1349
1350    @property
1351    def selects(self) -> t.List[Expression]:
1352        """If this statement contains a query (e.g. a CTAS), this returns the query's projections."""
1353        return self.expression.selects if isinstance(self.expression, Query) else []
1354
1355    @property
1356    def named_selects(self) -> t.List[str]:
1357        """
1358        If this statement contains a query (e.g. a CTAS), this returns the output
1359        names of the query's projections.
1360        """
1361        return self.expression.named_selects if isinstance(self.expression, Query) else []
1362
1363
1364class DML(Expression):
1365    def returning(
1366        self,
1367        expression: ExpOrStr,
1368        dialect: DialectType = None,
1369        copy: bool = True,
1370        **opts,
1371    ) -> DML:
1372        """
1373        Set the RETURNING expression. Not supported by all dialects.
1374
1375        Example:
1376            >>> delete("tbl").returning("*", dialect="postgres").sql()
1377            'DELETE FROM tbl RETURNING *'
1378
1379        Args:
1380            expression: the SQL code strings to parse.
1381                If an `Expression` instance is passed, it will be used as-is.
1382            dialect: the dialect used to parse the input expressions.
1383            copy: if `False`, modify this expression instance in-place.
1384            opts: other options to use to parse the input expressions.
1385
1386        Returns:
1387            Delete: the modified expression.
1388        """
1389        return _apply_builder(
1390            expression=expression,
1391            instance=self,
1392            arg="returning",
1393            prefix="RETURNING",
1394            dialect=dialect,
1395            copy=copy,
1396            into=Returning,
1397            **opts,
1398        )
1399
1400
1401class Create(DDL):
1402    arg_types = {
1403        "with": False,
1404        "this": True,
1405        "kind": True,
1406        "expression": False,
1407        "exists": False,
1408        "properties": False,
1409        "replace": False,
1410        "refresh": False,
1411        "unique": False,
1412        "indexes": False,
1413        "no_schema_binding": False,
1414        "begin": False,
1415        "end": False,
1416        "clone": False,
1417        "concurrently": False,
1418        "clustered": False,
1419    }
1420
1421    @property
1422    def kind(self) -> t.Optional[str]:
1423        kind = self.args.get("kind")
1424        return kind and kind.upper()
1425
1426
1427class SequenceProperties(Expression):
1428    arg_types = {
1429        "increment": False,
1430        "minvalue": False,
1431        "maxvalue": False,
1432        "cache": False,
1433        "start": False,
1434        "owned": False,
1435        "options": False,
1436    }
1437
1438
1439class TruncateTable(Expression):
1440    arg_types = {
1441        "expressions": True,
1442        "is_database": False,
1443        "exists": False,
1444        "only": False,
1445        "cluster": False,
1446        "identity": False,
1447        "option": False,
1448        "partition": False,
1449    }
1450
1451
1452# https://docs.snowflake.com/en/sql-reference/sql/create-clone
1453# https://cloud.google.com/bigquery/docs/reference/standard-sql/data-definition-language#create_table_clone_statement
1454# https://cloud.google.com/bigquery/docs/reference/standard-sql/data-definition-language#create_table_copy
1455class Clone(Expression):
1456    arg_types = {"this": True, "shallow": False, "copy": False}
1457
1458
1459class Describe(Expression):
1460    arg_types = {
1461        "this": True,
1462        "style": False,
1463        "kind": False,
1464        "expressions": False,
1465        "partition": False,
1466    }
1467
1468
1469# https://duckdb.org/docs/guides/meta/summarize.html
1470class Summarize(Expression):
1471    arg_types = {"this": True, "table": False}
1472
1473
1474class Kill(Expression):
1475    arg_types = {"this": True, "kind": False}
1476
1477
1478class Pragma(Expression):
1479    pass
1480
1481
1482class Declare(Expression):
1483    arg_types = {"expressions": True}
1484
1485
1486class DeclareItem(Expression):
1487    arg_types = {"this": True, "kind": True, "default": False}
1488
1489
1490class Set(Expression):
1491    arg_types = {"expressions": False, "unset": False, "tag": False}
1492
1493
1494class Heredoc(Expression):
1495    arg_types = {"this": True, "tag": False}
1496
1497
1498class SetItem(Expression):
1499    arg_types = {
1500        "this": False,
1501        "expressions": False,
1502        "kind": False,
1503        "collate": False,  # MySQL SET NAMES statement
1504        "global": False,
1505    }
1506
1507
1508class Show(Expression):
1509    arg_types = {
1510        "this": True,
1511        "history": False,
1512        "terse": False,
1513        "target": False,
1514        "offset": False,
1515        "starts_with": False,
1516        "limit": False,
1517        "from": False,
1518        "like": False,
1519        "where": False,
1520        "db": False,
1521        "scope": False,
1522        "scope_kind": False,
1523        "full": False,
1524        "mutex": False,
1525        "query": False,
1526        "channel": False,
1527        "global": False,
1528        "log": False,
1529        "position": False,
1530        "types": False,
1531    }
1532
1533
1534class UserDefinedFunction(Expression):
1535    arg_types = {"this": True, "expressions": False, "wrapped": False}
1536
1537
1538class CharacterSet(Expression):
1539    arg_types = {"this": True, "default": False}
1540
1541
1542class With(Expression):
1543    arg_types = {"expressions": True, "recursive": False}
1544
1545    @property
1546    def recursive(self) -> bool:
1547        return bool(self.args.get("recursive"))
1548
1549
1550class WithinGroup(Expression):
1551    arg_types = {"this": True, "expression": False}
1552
1553
1554# clickhouse supports scalar ctes
1555# https://clickhouse.com/docs/en/sql-reference/statements/select/with
1556class CTE(DerivedTable):
1557    arg_types = {
1558        "this": True,
1559        "alias": True,
1560        "scalar": False,
1561        "materialized": False,
1562    }
1563
1564
1565class ProjectionDef(Expression):
1566    arg_types = {"this": True, "expression": True}
1567
1568
1569class TableAlias(Expression):
1570    arg_types = {"this": False, "columns": False}
1571
1572    @property
1573    def columns(self):
1574        return self.args.get("columns") or []
1575
1576
1577class BitString(Condition):
1578    pass
1579
1580
1581class HexString(Condition):
1582    pass
1583
1584
1585class ByteString(Condition):
1586    pass
1587
1588
1589class RawString(Condition):
1590    pass
1591
1592
1593class UnicodeString(Condition):
1594    arg_types = {"this": True, "escape": False}
1595
1596
1597class Column(Condition):
1598    arg_types = {"this": True, "table": False, "db": False, "catalog": False, "join_mark": False}
1599
1600    @property
1601    def table(self) -> str:
1602        return self.text("table")
1603
1604    @property
1605    def db(self) -> str:
1606        return self.text("db")
1607
1608    @property
1609    def catalog(self) -> str:
1610        return self.text("catalog")
1611
1612    @property
1613    def output_name(self) -> str:
1614        return self.name
1615
1616    @property
1617    def parts(self) -> t.List[Identifier]:
1618        """Return the parts of a column in order catalog, db, table, name."""
1619        return [
1620            t.cast(Identifier, self.args[part])
1621            for part in ("catalog", "db", "table", "this")
1622            if self.args.get(part)
1623        ]
1624
1625    def to_dot(self) -> Dot | Identifier:
1626        """Converts the column into a dot expression."""
1627        parts = self.parts
1628        parent = self.parent
1629
1630        while parent:
1631            if isinstance(parent, Dot):
1632                parts.append(parent.expression)
1633            parent = parent.parent
1634
1635        return Dot.build(deepcopy(parts)) if len(parts) > 1 else parts[0]
1636
1637
1638class ColumnPosition(Expression):
1639    arg_types = {"this": False, "position": True}
1640
1641
1642class ColumnDef(Expression):
1643    arg_types = {
1644        "this": True,
1645        "kind": False,
1646        "constraints": False,
1647        "exists": False,
1648        "position": False,
1649    }
1650
1651    @property
1652    def constraints(self) -> t.List[ColumnConstraint]:
1653        return self.args.get("constraints") or []
1654
1655    @property
1656    def kind(self) -> t.Optional[DataType]:
1657        return self.args.get("kind")
1658
1659
1660class AlterColumn(Expression):
1661    arg_types = {
1662        "this": True,
1663        "dtype": False,
1664        "collate": False,
1665        "using": False,
1666        "default": False,
1667        "drop": False,
1668        "comment": False,
1669        "allow_null": False,
1670    }
1671
1672
1673# https://docs.aws.amazon.com/redshift/latest/dg/r_ALTER_TABLE.html
1674class AlterDistStyle(Expression):
1675    pass
1676
1677
1678class AlterSortKey(Expression):
1679    arg_types = {"this": False, "expressions": False, "compound": False}
1680
1681
1682class AlterSet(Expression):
1683    arg_types = {
1684        "expressions": False,
1685        "option": False,
1686        "tablespace": False,
1687        "access_method": False,
1688        "file_format": False,
1689        "copy_options": False,
1690        "tag": False,
1691        "location": False,
1692        "serde": False,
1693    }
1694
1695
1696class RenameColumn(Expression):
1697    arg_types = {"this": True, "to": True, "exists": False}
1698
1699
1700class RenameTable(Expression):
1701    pass
1702
1703
1704class SwapTable(Expression):
1705    pass
1706
1707
1708class Comment(Expression):
1709    arg_types = {
1710        "this": True,
1711        "kind": True,
1712        "expression": True,
1713        "exists": False,
1714        "materialized": False,
1715    }
1716
1717
1718class Comprehension(Expression):
1719    arg_types = {"this": True, "expression": True, "iterator": True, "condition": False}
1720
1721
1722# https://clickhouse.com/docs/en/engines/table-engines/mergetree-family/mergetree#mergetree-table-ttl
1723class MergeTreeTTLAction(Expression):
1724    arg_types = {
1725        "this": True,
1726        "delete": False,
1727        "recompress": False,
1728        "to_disk": False,
1729        "to_volume": False,
1730    }
1731
1732
1733# https://clickhouse.com/docs/en/engines/table-engines/mergetree-family/mergetree#mergetree-table-ttl
1734class MergeTreeTTL(Expression):
1735    arg_types = {
1736        "expressions": True,
1737        "where": False,
1738        "group": False,
1739        "aggregates": False,
1740    }
1741
1742
1743# https://dev.mysql.com/doc/refman/8.0/en/create-table.html
1744class IndexConstraintOption(Expression):
1745    arg_types = {
1746        "key_block_size": False,
1747        "using": False,
1748        "parser": False,
1749        "comment": False,
1750        "visible": False,
1751        "engine_attr": False,
1752        "secondary_engine_attr": False,
1753    }
1754
1755
1756class ColumnConstraint(Expression):
1757    arg_types = {"this": False, "kind": True}
1758
1759    @property
1760    def kind(self) -> ColumnConstraintKind:
1761        return self.args["kind"]
1762
1763
1764class ColumnConstraintKind(Expression):
1765    pass
1766
1767
1768class AutoIncrementColumnConstraint(ColumnConstraintKind):
1769    pass
1770
1771
1772class PeriodForSystemTimeConstraint(ColumnConstraintKind):
1773    arg_types = {"this": True, "expression": True}
1774
1775
1776class CaseSpecificColumnConstraint(ColumnConstraintKind):
1777    arg_types = {"not_": True}
1778
1779
1780class CharacterSetColumnConstraint(ColumnConstraintKind):
1781    arg_types = {"this": True}
1782
1783
1784class CheckColumnConstraint(ColumnConstraintKind):
1785    arg_types = {"this": True, "enforced": False}
1786
1787
1788class ClusteredColumnConstraint(ColumnConstraintKind):
1789    pass
1790
1791
1792class CollateColumnConstraint(ColumnConstraintKind):
1793    pass
1794
1795
1796class CommentColumnConstraint(ColumnConstraintKind):
1797    pass
1798
1799
1800class CompressColumnConstraint(ColumnConstraintKind):
1801    arg_types = {"this": False}
1802
1803
1804class DateFormatColumnConstraint(ColumnConstraintKind):
1805    arg_types = {"this": True}
1806
1807
1808class DefaultColumnConstraint(ColumnConstraintKind):
1809    pass
1810
1811
1812class EncodeColumnConstraint(ColumnConstraintKind):
1813    pass
1814
1815
1816# https://www.postgresql.org/docs/current/sql-createtable.html#SQL-CREATETABLE-EXCLUDE
1817class ExcludeColumnConstraint(ColumnConstraintKind):
1818    pass
1819
1820
1821class EphemeralColumnConstraint(ColumnConstraintKind):
1822    arg_types = {"this": False}
1823
1824
1825class WithOperator(Expression):
1826    arg_types = {"this": True, "op": True}
1827
1828
1829class GeneratedAsIdentityColumnConstraint(ColumnConstraintKind):
1830    # this: True -> ALWAYS, this: False -> BY DEFAULT
1831    arg_types = {
1832        "this": False,
1833        "expression": False,
1834        "on_null": False,
1835        "start": False,
1836        "increment": False,
1837        "minvalue": False,
1838        "maxvalue": False,
1839        "cycle": False,
1840    }
1841
1842
1843class GeneratedAsRowColumnConstraint(ColumnConstraintKind):
1844    arg_types = {"start": False, "hidden": False}
1845
1846
1847# https://dev.mysql.com/doc/refman/8.0/en/create-table.html
1848# https://github.com/ClickHouse/ClickHouse/blob/master/src/Parsers/ParserCreateQuery.h#L646
1849class IndexColumnConstraint(ColumnConstraintKind):
1850    arg_types = {
1851        "this": False,
1852        "expressions": False,
1853        "kind": False,
1854        "index_type": False,
1855        "options": False,
1856        "expression": False,  # Clickhouse
1857        "granularity": False,
1858    }
1859
1860
1861class InlineLengthColumnConstraint(ColumnConstraintKind):
1862    pass
1863
1864
1865class NonClusteredColumnConstraint(ColumnConstraintKind):
1866    pass
1867
1868
1869class NotForReplicationColumnConstraint(ColumnConstraintKind):
1870    arg_types = {}
1871
1872
1873# https://docs.snowflake.com/en/sql-reference/sql/create-table
1874class MaskingPolicyColumnConstraint(ColumnConstraintKind):
1875    arg_types = {"this": True, "expressions": False}
1876
1877
1878class NotNullColumnConstraint(ColumnConstraintKind):
1879    arg_types = {"allow_null": False}
1880
1881
1882# https://dev.mysql.com/doc/refman/5.7/en/timestamp-initialization.html
1883class OnUpdateColumnConstraint(ColumnConstraintKind):
1884    pass
1885
1886
1887# https://docs.snowflake.com/en/sql-reference/sql/create-table
1888class TagColumnConstraint(ColumnConstraintKind):
1889    arg_types = {"expressions": True}
1890
1891
1892# https://docs.snowflake.com/en/sql-reference/sql/create-external-table#optional-parameters
1893class TransformColumnConstraint(ColumnConstraintKind):
1894    pass
1895
1896
1897class PrimaryKeyColumnConstraint(ColumnConstraintKind):
1898    arg_types = {"desc": False}
1899
1900
1901class TitleColumnConstraint(ColumnConstraintKind):
1902    pass
1903
1904
1905class UniqueColumnConstraint(ColumnConstraintKind):
1906    arg_types = {"this": False, "index_type": False, "on_conflict": False, "nulls": False}
1907
1908
1909class UppercaseColumnConstraint(ColumnConstraintKind):
1910    arg_types: t.Dict[str, t.Any] = {}
1911
1912
1913class PathColumnConstraint(ColumnConstraintKind):
1914    pass
1915
1916
1917# https://docs.snowflake.com/en/sql-reference/sql/create-table
1918class ProjectionPolicyColumnConstraint(ColumnConstraintKind):
1919    pass
1920
1921
1922# computed column expression
1923# https://learn.microsoft.com/en-us/sql/t-sql/statements/create-table-transact-sql?view=sql-server-ver16
1924class ComputedColumnConstraint(ColumnConstraintKind):
1925    arg_types = {"this": True, "persisted": False, "not_null": False}
1926
1927
1928class Constraint(Expression):
1929    arg_types = {"this": True, "expressions": True}
1930
1931
1932class Delete(DML):
1933    arg_types = {
1934        "with": False,
1935        "this": False,
1936        "using": False,
1937        "where": False,
1938        "returning": False,
1939        "limit": False,
1940        "tables": False,  # Multiple-Table Syntax (MySQL)
1941    }
1942
1943    def delete(
1944        self,
1945        table: ExpOrStr,
1946        dialect: DialectType = None,
1947        copy: bool = True,
1948        **opts,
1949    ) -> Delete:
1950        """
1951        Create a DELETE expression or replace the table on an existing DELETE expression.
1952
1953        Example:
1954            >>> delete("tbl").sql()
1955            'DELETE FROM tbl'
1956
1957        Args:
1958            table: the table from which to delete.
1959            dialect: the dialect used to parse the input expression.
1960            copy: if `False`, modify this expression instance in-place.
1961            opts: other options to use to parse the input expressions.
1962
1963        Returns:
1964            Delete: the modified expression.
1965        """
1966        return _apply_builder(
1967            expression=table,
1968            instance=self,
1969            arg="this",
1970            dialect=dialect,
1971            into=Table,
1972            copy=copy,
1973            **opts,
1974        )
1975
1976    def where(
1977        self,
1978        *expressions: t.Optional[ExpOrStr],
1979        append: bool = True,
1980        dialect: DialectType = None,
1981        copy: bool = True,
1982        **opts,
1983    ) -> Delete:
1984        """
1985        Append to or set the WHERE expressions.
1986
1987        Example:
1988            >>> delete("tbl").where("x = 'a' OR x < 'b'").sql()
1989            "DELETE FROM tbl WHERE x = 'a' OR x < 'b'"
1990
1991        Args:
1992            *expressions: the SQL code strings to parse.
1993                If an `Expression` instance is passed, it will be used as-is.
1994                Multiple expressions are combined with an AND operator.
1995            append: if `True`, AND the new expressions to any existing expression.
1996                Otherwise, this resets the expression.
1997            dialect: the dialect used to parse the input expressions.
1998            copy: if `False`, modify this expression instance in-place.
1999            opts: other options to use to parse the input expressions.
2000
2001        Returns:
2002            Delete: the modified expression.
2003        """
2004        return _apply_conjunction_builder(
2005            *expressions,
2006            instance=self,
2007            arg="where",
2008            append=append,
2009            into=Where,
2010            dialect=dialect,
2011            copy=copy,
2012            **opts,
2013        )
2014
2015
2016class Drop(Expression):
2017    arg_types = {
2018        "this": False,
2019        "kind": False,
2020        "expressions": False,
2021        "exists": False,
2022        "temporary": False,
2023        "materialized": False,
2024        "cascade": False,
2025        "constraints": False,
2026        "purge": False,
2027        "cluster": False,
2028        "concurrently": False,
2029    }
2030
2031    @property
2032    def kind(self) -> t.Optional[str]:
2033        kind = self.args.get("kind")
2034        return kind and kind.upper()
2035
2036
2037class Filter(Expression):
2038    arg_types = {"this": True, "expression": True}
2039
2040
2041class Check(Expression):
2042    pass
2043
2044
2045class Changes(Expression):
2046    arg_types = {"information": True, "at_before": False, "end": False}
2047
2048
2049# https://docs.snowflake.com/en/sql-reference/constructs/connect-by
2050class Connect(Expression):
2051    arg_types = {"start": False, "connect": True, "nocycle": False}
2052
2053
2054class CopyParameter(Expression):
2055    arg_types = {"this": True, "expression": False, "expressions": False}
2056
2057
2058class Copy(DML):
2059    arg_types = {
2060        "this": True,
2061        "kind": True,
2062        "files": True,
2063        "credentials": False,
2064        "format": False,
2065        "params": False,
2066    }
2067
2068
2069class Credentials(Expression):
2070    arg_types = {
2071        "credentials": False,
2072        "encryption": False,
2073        "storage": False,
2074        "iam_role": False,
2075        "region": False,
2076    }
2077
2078
2079class Prior(Expression):
2080    pass
2081
2082
2083class Directory(Expression):
2084    # https://spark.apache.org/docs/3.0.0-preview/sql-ref-syntax-dml-insert-overwrite-directory-hive.html
2085    arg_types = {"this": True, "local": False, "row_format": False}
2086
2087
2088class ForeignKey(Expression):
2089    arg_types = {
2090        "expressions": True,
2091        "reference": False,
2092        "delete": False,
2093        "update": False,
2094    }
2095
2096
2097class ColumnPrefix(Expression):
2098    arg_types = {"this": True, "expression": True}
2099
2100
2101class PrimaryKey(Expression):
2102    arg_types = {"expressions": True, "options": False}
2103
2104
2105# https://www.postgresql.org/docs/9.1/sql-selectinto.html
2106# https://docs.aws.amazon.com/redshift/latest/dg/r_SELECT_INTO.html#r_SELECT_INTO-examples
2107class Into(Expression):
2108    arg_types = {"this": True, "temporary": False, "unlogged": False}
2109
2110
2111class From(Expression):
2112    @property
2113    def name(self) -> str:
2114        return self.this.name
2115
2116    @property
2117    def alias_or_name(self) -> str:
2118        return self.this.alias_or_name
2119
2120
2121class Having(Expression):
2122    pass
2123
2124
2125class Hint(Expression):
2126    arg_types = {"expressions": True}
2127
2128
2129class JoinHint(Expression):
2130    arg_types = {"this": True, "expressions": True}
2131
2132
2133class Identifier(Expression):
2134    arg_types = {"this": True, "quoted": False, "global": False, "temporary": False}
2135
2136    @property
2137    def quoted(self) -> bool:
2138        return bool(self.args.get("quoted"))
2139
2140    @property
2141    def hashable_args(self) -> t.Any:
2142        return (self.this, self.quoted)
2143
2144    @property
2145    def output_name(self) -> str:
2146        return self.name
2147
2148
2149# https://www.postgresql.org/docs/current/indexes-opclass.html
2150class Opclass(Expression):
2151    arg_types = {"this": True, "expression": True}
2152
2153
2154class Index(Expression):
2155    arg_types = {
2156        "this": False,
2157        "table": False,
2158        "unique": False,
2159        "primary": False,
2160        "amp": False,  # teradata
2161        "params": False,
2162    }
2163
2164
2165class IndexParameters(Expression):
2166    arg_types = {
2167        "using": False,
2168        "include": False,
2169        "columns": False,
2170        "with_storage": False,
2171        "partition_by": False,
2172        "tablespace": False,
2173        "where": False,
2174        "on": False,
2175    }
2176
2177
2178class Insert(DDL, DML):
2179    arg_types = {
2180        "hint": False,
2181        "with": False,
2182        "is_function": False,
2183        "this": False,
2184        "expression": False,
2185        "conflict": False,
2186        "returning": False,
2187        "overwrite": False,
2188        "exists": False,
2189        "alternative": False,
2190        "where": False,
2191        "ignore": False,
2192        "by_name": False,
2193        "stored": False,
2194        "partition": False,
2195        "settings": False,
2196        "source": False,
2197    }
2198
2199    def with_(
2200        self,
2201        alias: ExpOrStr,
2202        as_: ExpOrStr,
2203        recursive: t.Optional[bool] = None,
2204        materialized: t.Optional[bool] = None,
2205        append: bool = True,
2206        dialect: DialectType = None,
2207        copy: bool = True,
2208        **opts,
2209    ) -> Insert:
2210        """
2211        Append to or set the common table expressions.
2212
2213        Example:
2214            >>> insert("SELECT x FROM cte", "t").with_("cte", as_="SELECT * FROM tbl").sql()
2215            'WITH cte AS (SELECT * FROM tbl) INSERT INTO t SELECT x FROM cte'
2216
2217        Args:
2218            alias: the SQL code string to parse as the table name.
2219                If an `Expression` instance is passed, this is used as-is.
2220            as_: the SQL code string to parse as the table expression.
2221                If an `Expression` instance is passed, it will be used as-is.
2222            recursive: set the RECURSIVE part of the expression. Defaults to `False`.
2223            materialized: set the MATERIALIZED part of the expression.
2224            append: if `True`, add to any existing expressions.
2225                Otherwise, this resets the expressions.
2226            dialect: the dialect used to parse the input expression.
2227            copy: if `False`, modify this expression instance in-place.
2228            opts: other options to use to parse the input expressions.
2229
2230        Returns:
2231            The modified expression.
2232        """
2233        return _apply_cte_builder(
2234            self,
2235            alias,
2236            as_,
2237            recursive=recursive,
2238            materialized=materialized,
2239            append=append,
2240            dialect=dialect,
2241            copy=copy,
2242            **opts,
2243        )
2244
2245
2246class ConditionalInsert(Expression):
2247    arg_types = {"this": True, "expression": False, "else_": False}
2248
2249
2250class MultitableInserts(Expression):
2251    arg_types = {"expressions": True, "kind": True, "source": True}
2252
2253
2254class OnConflict(Expression):
2255    arg_types = {
2256        "duplicate": False,
2257        "expressions": False,
2258        "action": False,
2259        "conflict_keys": False,
2260        "constraint": False,
2261    }
2262
2263
2264class OnCondition(Expression):
2265    arg_types = {"error": False, "empty": False, "null": False}
2266
2267
2268class Returning(Expression):
2269    arg_types = {"expressions": True, "into": False}
2270
2271
2272# https://dev.mysql.com/doc/refman/8.0/en/charset-introducer.html
2273class Introducer(Expression):
2274    arg_types = {"this": True, "expression": True}
2275
2276
2277# national char, like n'utf8'
2278class National(Expression):
2279    pass
2280
2281
2282class LoadData(Expression):
2283    arg_types = {
2284        "this": True,
2285        "local": False,
2286        "overwrite": False,
2287        "inpath": True,
2288        "partition": False,
2289        "input_format": False,
2290        "serde": False,
2291    }
2292
2293
2294class Partition(Expression):
2295    arg_types = {"expressions": True}
2296
2297
2298class PartitionRange(Expression):
2299    arg_types = {"this": True, "expression": True}
2300
2301
2302# https://clickhouse.com/docs/en/sql-reference/statements/alter/partition#how-to-set-partition-expression
2303class PartitionId(Expression):
2304    pass
2305
2306
2307class Fetch(Expression):
2308    arg_types = {
2309        "direction": False,
2310        "count": False,
2311        "percent": False,
2312        "with_ties": False,
2313    }
2314
2315
2316class Group(Expression):
2317    arg_types = {
2318        "expressions": False,
2319        "grouping_sets": False,
2320        "cube": False,
2321        "rollup": False,
2322        "totals": False,
2323        "all": False,
2324    }
2325
2326
2327class Cube(Expression):
2328    arg_types = {"expressions": False}
2329
2330
2331class Rollup(Expression):
2332    arg_types = {"expressions": False}
2333
2334
2335class GroupingSets(Expression):
2336    arg_types = {"expressions": True}
2337
2338
2339class Lambda(Expression):
2340    arg_types = {"this": True, "expressions": True}
2341
2342
2343class Limit(Expression):
2344    arg_types = {"this": False, "expression": True, "offset": False, "expressions": False}
2345
2346
2347class Literal(Condition):
2348    arg_types = {"this": True, "is_string": True}
2349
2350    @property
2351    def hashable_args(self) -> t.Any:
2352        return (self.this, self.args.get("is_string"))
2353
2354    @classmethod
2355    def number(cls, number) -> Literal:
2356        return cls(this=str(number), is_string=False)
2357
2358    @classmethod
2359    def string(cls, string) -> Literal:
2360        return cls(this=str(string), is_string=True)
2361
2362    @property
2363    def output_name(self) -> str:
2364        return self.name
2365
2366    def to_py(self) -> int | str | Decimal:
2367        if self.is_number:
2368            try:
2369                return int(self.this)
2370            except ValueError:
2371                return Decimal(self.this)
2372        return self.this
2373
2374
2375class Join(Expression):
2376    arg_types = {
2377        "this": True,
2378        "on": False,
2379        "side": False,
2380        "kind": False,
2381        "using": False,
2382        "method": False,
2383        "global": False,
2384        "hint": False,
2385        "match_condition": False,  # Snowflake
2386    }
2387
2388    @property
2389    def method(self) -> str:
2390        return self.text("method").upper()
2391
2392    @property
2393    def kind(self) -> str:
2394        return self.text("kind").upper()
2395
2396    @property
2397    def side(self) -> str:
2398        return self.text("side").upper()
2399
2400    @property
2401    def hint(self) -> str:
2402        return self.text("hint").upper()
2403
2404    @property
2405    def alias_or_name(self) -> str:
2406        return self.this.alias_or_name
2407
2408    def on(
2409        self,
2410        *expressions: t.Optional[ExpOrStr],
2411        append: bool = True,
2412        dialect: DialectType = None,
2413        copy: bool = True,
2414        **opts,
2415    ) -> Join:
2416        """
2417        Append to or set the ON expressions.
2418
2419        Example:
2420            >>> import sqlglot
2421            >>> sqlglot.parse_one("JOIN x", into=Join).on("y = 1").sql()
2422            'JOIN x ON y = 1'
2423
2424        Args:
2425            *expressions: the SQL code strings to parse.
2426                If an `Expression` instance is passed, it will be used as-is.
2427                Multiple expressions are combined with an AND operator.
2428            append: if `True`, AND the new expressions to any existing expression.
2429                Otherwise, this resets the expression.
2430            dialect: the dialect used to parse the input expressions.
2431            copy: if `False`, modify this expression instance in-place.
2432            opts: other options to use to parse the input expressions.
2433
2434        Returns:
2435            The modified Join expression.
2436        """
2437        join = _apply_conjunction_builder(
2438            *expressions,
2439            instance=self,
2440            arg="on",
2441            append=append,
2442            dialect=dialect,
2443            copy=copy,
2444            **opts,
2445        )
2446
2447        if join.kind == "CROSS":
2448            join.set("kind", None)
2449
2450        return join
2451
2452    def using(
2453        self,
2454        *expressions: t.Optional[ExpOrStr],
2455        append: bool = True,
2456        dialect: DialectType = None,
2457        copy: bool = True,
2458        **opts,
2459    ) -> Join:
2460        """
2461        Append to or set the USING expressions.
2462
2463        Example:
2464            >>> import sqlglot
2465            >>> sqlglot.parse_one("JOIN x", into=Join).using("foo", "bla").sql()
2466            'JOIN x USING (foo, bla)'
2467
2468        Args:
2469            *expressions: the SQL code strings to parse.
2470                If an `Expression` instance is passed, it will be used as-is.
2471            append: if `True`, concatenate the new expressions to the existing "using" list.
2472                Otherwise, this resets the expression.
2473            dialect: the dialect used to parse the input expressions.
2474            copy: if `False`, modify this expression instance in-place.
2475            opts: other options to use to parse the input expressions.
2476
2477        Returns:
2478            The modified Join expression.
2479        """
2480        join = _apply_list_builder(
2481            *expressions,
2482            instance=self,
2483            arg="using",
2484            append=append,
2485            dialect=dialect,
2486            copy=copy,
2487            **opts,
2488        )
2489
2490        if join.kind == "CROSS":
2491            join.set("kind", None)
2492
2493        return join
2494
2495
2496class Lateral(UDTF):
2497    arg_types = {
2498        "this": True,
2499        "view": False,
2500        "outer": False,
2501        "alias": False,
2502        "cross_apply": False,  # True -> CROSS APPLY, False -> OUTER APPLY
2503    }
2504
2505
2506class MatchRecognizeMeasure(Expression):
2507    arg_types = {
2508        "this": True,
2509        "window_frame": False,
2510    }
2511
2512
2513class MatchRecognize(Expression):
2514    arg_types = {
2515        "partition_by": False,
2516        "order": False,
2517        "measures": False,
2518        "rows": False,
2519        "after": False,
2520        "pattern": False,
2521        "define": False,
2522        "alias": False,
2523    }
2524
2525
2526# Clickhouse FROM FINAL modifier
2527# https://clickhouse.com/docs/en/sql-reference/statements/select/from/#final-modifier
2528class Final(Expression):
2529    pass
2530
2531
2532class Offset(Expression):
2533    arg_types = {"this": False, "expression": True, "expressions": False}
2534
2535
2536class Order(Expression):
2537    arg_types = {"this": False, "expressions": True, "siblings": False}
2538
2539
2540# https://clickhouse.com/docs/en/sql-reference/statements/select/order-by#order-by-expr-with-fill-modifier
2541class WithFill(Expression):
2542    arg_types = {
2543        "from": False,
2544        "to": False,
2545        "step": False,
2546        "interpolate": False,
2547    }
2548
2549
2550# hive specific sorts
2551# https://cwiki.apache.org/confluence/display/Hive/LanguageManual+SortBy
2552class Cluster(Order):
2553    pass
2554
2555
2556class Distribute(Order):
2557    pass
2558
2559
2560class Sort(Order):
2561    pass
2562
2563
2564class Ordered(Expression):
2565    arg_types = {"this": True, "desc": False, "nulls_first": True, "with_fill": False}
2566
2567
2568class Property(Expression):
2569    arg_types = {"this": True, "value": True}
2570
2571
2572class AllowedValuesProperty(Expression):
2573    arg_types = {"expressions": True}
2574
2575
2576class AlgorithmProperty(Property):
2577    arg_types = {"this": True}
2578
2579
2580class AutoIncrementProperty(Property):
2581    arg_types = {"this": True}
2582
2583
2584# https://docs.aws.amazon.com/prescriptive-guidance/latest/materialized-views-redshift/refreshing-materialized-views.html
2585class AutoRefreshProperty(Property):
2586    arg_types = {"this": True}
2587
2588
2589class BackupProperty(Property):
2590    arg_types = {"this": True}
2591
2592
2593class BlockCompressionProperty(Property):
2594    arg_types = {
2595        "autotemp": False,
2596        "always": False,
2597        "default": False,
2598        "manual": False,
2599        "never": False,
2600    }
2601
2602
2603class CharacterSetProperty(Property):
2604    arg_types = {"this": True, "default": True}
2605
2606
2607class ChecksumProperty(Property):
2608    arg_types = {"on": False, "default": False}
2609
2610
2611class CollateProperty(Property):
2612    arg_types = {"this": True, "default": False}
2613
2614
2615class CopyGrantsProperty(Property):
2616    arg_types = {}
2617
2618
2619class DataBlocksizeProperty(Property):
2620    arg_types = {
2621        "size": False,
2622        "units": False,
2623        "minimum": False,
2624        "maximum": False,
2625        "default": False,
2626    }
2627
2628
2629class DataDeletionProperty(Property):
2630    arg_types = {"on": True, "filter_col": False, "retention_period": False}
2631
2632
2633class DefinerProperty(Property):
2634    arg_types = {"this": True}
2635
2636
2637class DistKeyProperty(Property):
2638    arg_types = {"this": True}
2639
2640
2641# https://docs.starrocks.io/docs/sql-reference/sql-statements/data-definition/CREATE_TABLE/#distribution_desc
2642# https://doris.apache.org/docs/sql-manual/sql-statements/Data-Definition-Statements/Create/CREATE-TABLE?_highlight=create&_highlight=table#distribution_desc
2643class DistributedByProperty(Property):
2644    arg_types = {"expressions": False, "kind": True, "buckets": False, "order": False}
2645
2646
2647class DistStyleProperty(Property):
2648    arg_types = {"this": True}
2649
2650
2651class DuplicateKeyProperty(Property):
2652    arg_types = {"expressions": True}
2653
2654
2655class EngineProperty(Property):
2656    arg_types = {"this": True}
2657
2658
2659class HeapProperty(Property):
2660    arg_types = {}
2661
2662
2663class ToTableProperty(Property):
2664    arg_types = {"this": True}
2665
2666
2667class ExecuteAsProperty(Property):
2668    arg_types = {"this": True}
2669
2670
2671class ExternalProperty(Property):
2672    arg_types = {"this": False}
2673
2674
2675class FallbackProperty(Property):
2676    arg_types = {"no": True, "protection": False}
2677
2678
2679class FileFormatProperty(Property):
2680    arg_types = {"this": True}
2681
2682
2683class FreespaceProperty(Property):
2684    arg_types = {"this": True, "percent": False}
2685
2686
2687class GlobalProperty(Property):
2688    arg_types = {}
2689
2690
2691class IcebergProperty(Property):
2692    arg_types = {}
2693
2694
2695class InheritsProperty(Property):
2696    arg_types = {"expressions": True}
2697
2698
2699class InputModelProperty(Property):
2700    arg_types = {"this": True}
2701
2702
2703class OutputModelProperty(Property):
2704    arg_types = {"this": True}
2705
2706
2707class IsolatedLoadingProperty(Property):
2708    arg_types = {"no": False, "concurrent": False, "target": False}
2709
2710
2711class JournalProperty(Property):
2712    arg_types = {
2713        "no": False,
2714        "dual": False,
2715        "before": False,
2716        "local": False,
2717        "after": False,
2718    }
2719
2720
2721class LanguageProperty(Property):
2722    arg_types = {"this": True}
2723
2724
2725# spark ddl
2726class ClusteredByProperty(Property):
2727    arg_types = {"expressions": True, "sorted_by": False, "buckets": True}
2728
2729
2730class DictProperty(Property):
2731    arg_types = {"this": True, "kind": True, "settings": False}
2732
2733
2734class DictSubProperty(Property):
2735    pass
2736
2737
2738class DictRange(Property):
2739    arg_types = {"this": True, "min": True, "max": True}
2740
2741
2742class DynamicProperty(Property):
2743    arg_types = {}
2744
2745
2746# Clickhouse CREATE ... ON CLUSTER modifier
2747# https://clickhouse.com/docs/en/sql-reference/distributed-ddl
2748class OnCluster(Property):
2749    arg_types = {"this": True}
2750
2751
2752# Clickhouse EMPTY table "property"
2753class EmptyProperty(Property):
2754    arg_types = {}
2755
2756
2757class LikeProperty(Property):
2758    arg_types = {"this": True, "expressions": False}
2759
2760
2761class LocationProperty(Property):
2762    arg_types = {"this": True}
2763
2764
2765class LockProperty(Property):
2766    arg_types = {"this": True}
2767
2768
2769class LockingProperty(Property):
2770    arg_types = {
2771        "this": False,
2772        "kind": True,
2773        "for_or_in": False,
2774        "lock_type": True,
2775        "override": False,
2776    }
2777
2778
2779class LogProperty(Property):
2780    arg_types = {"no": True}
2781
2782
2783class MaterializedProperty(Property):
2784    arg_types = {"this": False}
2785
2786
2787class MergeBlockRatioProperty(Property):
2788    arg_types = {"this": False, "no": False, "default": False, "percent": False}
2789
2790
2791class NoPrimaryIndexProperty(Property):
2792    arg_types = {}
2793
2794
2795class OnProperty(Property):
2796    arg_types = {"this": True}
2797
2798
2799class OnCommitProperty(Property):
2800    arg_types = {"delete": False}
2801
2802
2803class PartitionedByProperty(Property):
2804    arg_types = {"this": True}
2805
2806
2807# https://www.postgresql.org/docs/current/sql-createtable.html
2808class PartitionBoundSpec(Expression):
2809    # this -> IN / MODULUS, expression -> REMAINDER, from_expressions -> FROM (...), to_expressions -> TO (...)
2810    arg_types = {
2811        "this": False,
2812        "expression": False,
2813        "from_expressions": False,
2814        "to_expressions": False,
2815    }
2816
2817
2818class PartitionedOfProperty(Property):
2819    # this -> parent_table (schema), expression -> FOR VALUES ... / DEFAULT
2820    arg_types = {"this": True, "expression": True}
2821
2822
2823class StreamingTableProperty(Property):
2824    arg_types = {}
2825
2826
2827class RemoteWithConnectionModelProperty(Property):
2828    arg_types = {"this": True}
2829
2830
2831class ReturnsProperty(Property):
2832    arg_types = {"this": False, "is_table": False, "table": False, "null": False}
2833
2834
2835class StrictProperty(Property):
2836    arg_types = {}
2837
2838
2839class RowFormatProperty(Property):
2840    arg_types = {"this": True}
2841
2842
2843class RowFormatDelimitedProperty(Property):
2844    # https://cwiki.apache.org/confluence/display/hive/languagemanual+dml
2845    arg_types = {
2846        "fields": False,
2847        "escaped": False,
2848        "collection_items": False,
2849        "map_keys": False,
2850        "lines": False,
2851        "null": False,
2852        "serde": False,
2853    }
2854
2855
2856class RowFormatSerdeProperty(Property):
2857    arg_types = {"this": True, "serde_properties": False}
2858
2859
2860# https://spark.apache.org/docs/3.1.2/sql-ref-syntax-qry-select-transform.html
2861class QueryTransform(Expression):
2862    arg_types = {
2863        "expressions": True,
2864        "command_script": True,
2865        "schema": False,
2866        "row_format_before": False,
2867        "record_writer": False,
2868        "row_format_after": False,
2869        "record_reader": False,
2870    }
2871
2872
2873class SampleProperty(Property):
2874    arg_types = {"this": True}
2875
2876
2877# https://prestodb.io/docs/current/sql/create-view.html#synopsis
2878class SecurityProperty(Property):
2879    arg_types = {"this": True}
2880
2881
2882class SchemaCommentProperty(Property):
2883    arg_types = {"this": True}
2884
2885
2886class SerdeProperties(Property):
2887    arg_types = {"expressions": True, "with": False}
2888
2889
2890class SetProperty(Property):
2891    arg_types = {"multi": True}
2892
2893
2894class SharingProperty(Property):
2895    arg_types = {"this": False}
2896
2897
2898class SetConfigProperty(Property):
2899    arg_types = {"this": True}
2900
2901
2902class SettingsProperty(Property):
2903    arg_types = {"expressions": True}
2904
2905
2906class SortKeyProperty(Property):
2907    arg_types = {"this": True, "compound": False}
2908
2909
2910class SqlReadWriteProperty(Property):
2911    arg_types = {"this": True}
2912
2913
2914class SqlSecurityProperty(Property):
2915    arg_types = {"definer": True}
2916
2917
2918class StabilityProperty(Property):
2919    arg_types = {"this": True}
2920
2921
2922class TemporaryProperty(Property):
2923    arg_types = {"this": False}
2924
2925
2926class SecureProperty(Property):
2927    arg_types = {}
2928
2929
2930class TransformModelProperty(Property):
2931    arg_types = {"expressions": True}
2932
2933
2934class TransientProperty(Property):
2935    arg_types = {"this": False}
2936
2937
2938class UnloggedProperty(Property):
2939    arg_types = {}
2940
2941
2942# https://learn.microsoft.com/en-us/sql/t-sql/statements/create-view-transact-sql?view=sql-server-ver16
2943class ViewAttributeProperty(Property):
2944    arg_types = {"this": True}
2945
2946
2947class VolatileProperty(Property):
2948    arg_types = {"this": False}
2949
2950
2951class WithDataProperty(Property):
2952    arg_types = {"no": True, "statistics": False}
2953
2954
2955class WithJournalTableProperty(Property):
2956    arg_types = {"this": True}
2957
2958
2959class WithSchemaBindingProperty(Property):
2960    arg_types = {"this": True}
2961
2962
2963class WithSystemVersioningProperty(Property):
2964    arg_types = {
2965        "on": False,
2966        "this": False,
2967        "data_consistency": False,
2968        "retention_period": False,
2969        "with": True,
2970    }
2971
2972
2973class Properties(Expression):
2974    arg_types = {"expressions": True}
2975
2976    NAME_TO_PROPERTY = {
2977        "ALGORITHM": AlgorithmProperty,
2978        "AUTO_INCREMENT": AutoIncrementProperty,
2979        "CHARACTER SET": CharacterSetProperty,
2980        "CLUSTERED_BY": ClusteredByProperty,
2981        "COLLATE": CollateProperty,
2982        "COMMENT": SchemaCommentProperty,
2983        "DEFINER": DefinerProperty,
2984        "DISTKEY": DistKeyProperty,
2985        "DISTRIBUTED_BY": DistributedByProperty,
2986        "DISTSTYLE": DistStyleProperty,
2987        "ENGINE": EngineProperty,
2988        "EXECUTE AS": ExecuteAsProperty,
2989        "FORMAT": FileFormatProperty,
2990        "LANGUAGE": LanguageProperty,
2991        "LOCATION": LocationProperty,
2992        "LOCK": LockProperty,
2993        "PARTITIONED_BY": PartitionedByProperty,
2994        "RETURNS": ReturnsProperty,
2995        "ROW_FORMAT": RowFormatProperty,
2996        "SORTKEY": SortKeyProperty,
2997    }
2998
2999    PROPERTY_TO_NAME = {v: k for k, v in NAME_TO_PROPERTY.items()}
3000
3001    # CREATE property locations
3002    # Form: schema specified
3003    #   create [POST_CREATE]
3004    #     table a [POST_NAME]
3005    #     (b int) [POST_SCHEMA]
3006    #     with ([POST_WITH])
3007    #     index (b) [POST_INDEX]
3008    #
3009    # Form: alias selection
3010    #   create [POST_CREATE]
3011    #     table a [POST_NAME]
3012    #     as [POST_ALIAS] (select * from b) [POST_EXPRESSION]
3013    #     index (c) [POST_INDEX]
3014    class Location(AutoName):
3015        POST_CREATE = auto()
3016        POST_NAME = auto()
3017        POST_SCHEMA = auto()
3018        POST_WITH = auto()
3019        POST_ALIAS = auto()
3020        POST_EXPRESSION = auto()
3021        POST_INDEX = auto()
3022        UNSUPPORTED = auto()
3023
3024    @classmethod
3025    def from_dict(cls, properties_dict: t.Dict) -> Properties:
3026        expressions = []
3027        for key, value in properties_dict.items():
3028            property_cls = cls.NAME_TO_PROPERTY.get(key.upper())
3029            if property_cls:
3030                expressions.append(property_cls(this=convert(value)))
3031            else:
3032                expressions.append(Property(this=Literal.string(key), value=convert(value)))
3033
3034        return cls(expressions=expressions)
3035
3036
3037class Qualify(Expression):
3038    pass
3039
3040
3041class InputOutputFormat(Expression):
3042    arg_types = {"input_format": False, "output_format": False}
3043
3044
3045# https://www.ibm.com/docs/en/ias?topic=procedures-return-statement-in-sql
3046class Return(Expression):
3047    pass
3048
3049
3050class Reference(Expression):
3051    arg_types = {"this": True, "expressions": False, "options": False}
3052
3053
3054class Tuple(Expression):
3055    arg_types = {"expressions": False}
3056
3057    def isin(
3058        self,
3059        *expressions: t.Any,
3060        query: t.Optional[ExpOrStr] = None,
3061        unnest: t.Optional[ExpOrStr] | t.Collection[ExpOrStr] = None,
3062        copy: bool = True,
3063        **opts,
3064    ) -> In:
3065        return In(
3066            this=maybe_copy(self, copy),
3067            expressions=[convert(e, copy=copy) for e in expressions],
3068            query=maybe_parse(query, copy=copy, **opts) if query else None,
3069            unnest=(
3070                Unnest(
3071                    expressions=[
3072                        maybe_parse(t.cast(ExpOrStr, e), copy=copy, **opts)
3073                        for e in ensure_list(unnest)
3074                    ]
3075                )
3076                if unnest
3077                else None
3078            ),
3079        )
3080
3081
3082QUERY_MODIFIERS = {
3083    "match": False,
3084    "laterals": False,
3085    "joins": False,
3086    "connect": False,
3087    "pivots": False,
3088    "prewhere": False,
3089    "where": False,
3090    "group": False,
3091    "having": False,
3092    "qualify": False,
3093    "windows": False,
3094    "distribute": False,
3095    "sort": False,
3096    "cluster": False,
3097    "order": False,
3098    "limit": False,
3099    "offset": False,
3100    "locks": False,
3101    "sample": False,
3102    "settings": False,
3103    "format": False,
3104    "options": False,
3105}
3106
3107
3108# https://learn.microsoft.com/en-us/sql/t-sql/queries/option-clause-transact-sql?view=sql-server-ver16
3109# https://learn.microsoft.com/en-us/sql/t-sql/queries/hints-transact-sql-query?view=sql-server-ver16
3110class QueryOption(Expression):
3111    arg_types = {"this": True, "expression": False}
3112
3113
3114# https://learn.microsoft.com/en-us/sql/t-sql/queries/hints-transact-sql-table?view=sql-server-ver16
3115class WithTableHint(Expression):
3116    arg_types = {"expressions": True}
3117
3118
3119# https://dev.mysql.com/doc/refman/8.0/en/index-hints.html
3120class IndexTableHint(Expression):
3121    arg_types = {"this": True, "expressions": False, "target": False}
3122
3123
3124# https://docs.snowflake.com/en/sql-reference/constructs/at-before
3125class HistoricalData(Expression):
3126    arg_types = {"this": True, "kind": True, "expression": True}
3127
3128
3129class Table(Expression):
3130    arg_types = {
3131        "this": False,
3132        "alias": False,
3133        "db": False,
3134        "catalog": False,
3135        "laterals": False,
3136        "joins": False,
3137        "pivots": False,
3138        "hints": False,
3139        "system_time": False,
3140        "version": False,
3141        "format": False,
3142        "pattern": False,
3143        "ordinality": False,
3144        "when": False,
3145        "only": False,
3146        "partition": False,
3147        "changes": False,
3148        "rows_from": False,
3149        "sample": False,
3150    }
3151
3152    @property
3153    def name(self) -> str:
3154        if isinstance(self.this, Func):
3155            return ""
3156        return self.this.name
3157
3158    @property
3159    def db(self) -> str:
3160        return self.text("db")
3161
3162    @property
3163    def catalog(self) -> str:
3164        return self.text("catalog")
3165
3166    @property
3167    def selects(self) -> t.List[Expression]:
3168        return []
3169
3170    @property
3171    def named_selects(self) -> t.List[str]:
3172        return []
3173
3174    @property
3175    def parts(self) -> t.List[Expression]:
3176        """Return the parts of a table in order catalog, db, table."""
3177        parts: t.List[Expression] = []
3178
3179        for arg in ("catalog", "db", "this"):
3180            part = self.args.get(arg)
3181
3182            if isinstance(part, Dot):
3183                parts.extend(part.flatten())
3184            elif isinstance(part, Expression):
3185                parts.append(part)
3186
3187        return parts
3188
3189    def to_column(self, copy: bool = True) -> Alias | Column | Dot:
3190        parts = self.parts
3191        col = column(*reversed(parts[0:4]), fields=parts[4:], copy=copy)  # type: ignore
3192        alias = self.args.get("alias")
3193        if alias:
3194            col = alias_(col, alias.this, copy=copy)
3195        return col
3196
3197
3198class SetOperation(Query):
3199    arg_types = {
3200        "with": False,
3201        "this": True,
3202        "expression": True,
3203        "distinct": False,
3204        "by_name": False,
3205        **QUERY_MODIFIERS,
3206    }
3207
3208    def select(
3209        self: S,
3210        *expressions: t.Optional[ExpOrStr],
3211        append: bool = True,
3212        dialect: DialectType = None,
3213        copy: bool = True,
3214        **opts,
3215    ) -> S:
3216        this = maybe_copy(self, copy)
3217        this.this.unnest().select(*expressions, append=append, dialect=dialect, copy=False, **opts)
3218        this.expression.unnest().select(
3219            *expressions, append=append, dialect=dialect, copy=False, **opts
3220        )
3221        return this
3222
3223    @property
3224    def named_selects(self) -> t.List[str]:
3225        return self.this.unnest().named_selects
3226
3227    @property
3228    def is_star(self) -> bool:
3229        return self.this.is_star or self.expression.is_star
3230
3231    @property
3232    def selects(self) -> t.List[Expression]:
3233        return self.this.unnest().selects
3234
3235    @property
3236    def left(self) -> Query:
3237        return self.this
3238
3239    @property
3240    def right(self) -> Query:
3241        return self.expression
3242
3243
3244class Union(SetOperation):
3245    pass
3246
3247
3248class Except(SetOperation):
3249    pass
3250
3251
3252class Intersect(SetOperation):
3253    pass
3254
3255
3256class Update(Expression):
3257    arg_types = {
3258        "with": False,
3259        "this": False,
3260        "expressions": True,
3261        "from": False,
3262        "where": False,
3263        "returning": False,
3264        "order": False,
3265        "limit": False,
3266    }
3267
3268
3269class Values(UDTF):
3270    arg_types = {"expressions": True, "alias": False}
3271
3272
3273class Var(Expression):
3274    pass
3275
3276
3277class Version(Expression):
3278    """
3279    Time travel, iceberg, bigquery etc
3280    https://trino.io/docs/current/connector/iceberg.html?highlight=snapshot#using-snapshots
3281    https://www.databricks.com/blog/2019/02/04/introducing-delta-time-travel-for-large-scale-data-lakes.html
3282    https://cloud.google.com/bigquery/docs/reference/standard-sql/query-syntax#for_system_time_as_of
3283    https://learn.microsoft.com/en-us/sql/relational-databases/tables/querying-data-in-a-system-versioned-temporal-table?view=sql-server-ver16
3284    this is either TIMESTAMP or VERSION
3285    kind is ("AS OF", "BETWEEN")
3286    """
3287
3288    arg_types = {"this": True, "kind": True, "expression": False}
3289
3290
3291class Schema(Expression):
3292    arg_types = {"this": False, "expressions": False}
3293
3294
3295# https://dev.mysql.com/doc/refman/8.0/en/select.html
3296# https://docs.oracle.com/en/database/oracle/oracle-database/19/sqlrf/SELECT.html
3297class Lock(Expression):
3298    arg_types = {"update": True, "expressions": False, "wait": False}
3299
3300
3301class Select(Query):
3302    arg_types = {
3303        "with": False,
3304        "kind": False,
3305        "expressions": False,
3306        "hint": False,
3307        "distinct": False,
3308        "into": False,
3309        "from": False,
3310        **QUERY_MODIFIERS,
3311    }
3312
3313    def from_(
3314        self, expression: ExpOrStr, dialect: DialectType = None, copy: bool = True, **opts
3315    ) -> Select:
3316        """
3317        Set the FROM expression.
3318
3319        Example:
3320            >>> Select().from_("tbl").select("x").sql()
3321            'SELECT x FROM tbl'
3322
3323        Args:
3324            expression : the SQL code strings to parse.
3325                If a `From` instance is passed, this is used as-is.
3326                If another `Expression` instance is passed, it will be wrapped in a `From`.
3327            dialect: the dialect used to parse the input expression.
3328            copy: if `False`, modify this expression instance in-place.
3329            opts: other options to use to parse the input expressions.
3330
3331        Returns:
3332            The modified Select expression.
3333        """
3334        return _apply_builder(
3335            expression=expression,
3336            instance=self,
3337            arg="from",
3338            into=From,
3339            prefix="FROM",
3340            dialect=dialect,
3341            copy=copy,
3342            **opts,
3343        )
3344
3345    def group_by(
3346        self,
3347        *expressions: t.Optional[ExpOrStr],
3348        append: bool = True,
3349        dialect: DialectType = None,
3350        copy: bool = True,
3351        **opts,
3352    ) -> Select:
3353        """
3354        Set the GROUP BY expression.
3355
3356        Example:
3357            >>> Select().from_("tbl").select("x", "COUNT(1)").group_by("x").sql()
3358            'SELECT x, COUNT(1) FROM tbl GROUP BY x'
3359
3360        Args:
3361            *expressions: the SQL code strings to parse.
3362                If a `Group` instance is passed, this is used as-is.
3363                If another `Expression` instance is passed, it will be wrapped in a `Group`.
3364                If nothing is passed in then a group by is not applied to the expression
3365            append: if `True`, add to any existing expressions.
3366                Otherwise, this flattens all the `Group` expression into a single expression.
3367            dialect: the dialect used to parse the input expression.
3368            copy: if `False`, modify this expression instance in-place.
3369            opts: other options to use to parse the input expressions.
3370
3371        Returns:
3372            The modified Select expression.
3373        """
3374        if not expressions:
3375            return self if not copy else self.copy()
3376
3377        return _apply_child_list_builder(
3378            *expressions,
3379            instance=self,
3380            arg="group",
3381            append=append,
3382            copy=copy,
3383            prefix="GROUP BY",
3384            into=Group,
3385            dialect=dialect,
3386            **opts,
3387        )
3388
3389    def sort_by(
3390        self,
3391        *expressions: t.Optional[ExpOrStr],
3392        append: bool = True,
3393        dialect: DialectType = None,
3394        copy: bool = True,
3395        **opts,
3396    ) -> Select:
3397        """
3398        Set the SORT BY expression.
3399
3400        Example:
3401            >>> Select().from_("tbl").select("x").sort_by("x DESC").sql(dialect="hive")
3402            'SELECT x FROM tbl SORT BY x DESC'
3403
3404        Args:
3405            *expressions: the SQL code strings to parse.
3406                If a `Group` instance is passed, this is used as-is.
3407                If another `Expression` instance is passed, it will be wrapped in a `SORT`.
3408            append: if `True`, add to any existing expressions.
3409                Otherwise, this flattens all the `Order` expression into a single expression.
3410            dialect: the dialect used to parse the input expression.
3411            copy: if `False`, modify this expression instance in-place.
3412            opts: other options to use to parse the input expressions.
3413
3414        Returns:
3415            The modified Select expression.
3416        """
3417        return _apply_child_list_builder(
3418            *expressions,
3419            instance=self,
3420            arg="sort",
3421            append=append,
3422            copy=copy,
3423            prefix="SORT BY",
3424            into=Sort,
3425            dialect=dialect,
3426            **opts,
3427        )
3428
3429    def cluster_by(
3430        self,
3431        *expressions: t.Optional[ExpOrStr],
3432        append: bool = True,
3433        dialect: DialectType = None,
3434        copy: bool = True,
3435        **opts,
3436    ) -> Select:
3437        """
3438        Set the CLUSTER BY expression.
3439
3440        Example:
3441            >>> Select().from_("tbl").select("x").cluster_by("x DESC").sql(dialect="hive")
3442            'SELECT x FROM tbl CLUSTER BY x DESC'
3443
3444        Args:
3445            *expressions: the SQL code strings to parse.
3446                If a `Group` instance is passed, this is used as-is.
3447                If another `Expression` instance is passed, it will be wrapped in a `Cluster`.
3448            append: if `True`, add to any existing expressions.
3449                Otherwise, this flattens all the `Order` expression into a single expression.
3450            dialect: the dialect used to parse the input expression.
3451            copy: if `False`, modify this expression instance in-place.
3452            opts: other options to use to parse the input expressions.
3453
3454        Returns:
3455            The modified Select expression.
3456        """
3457        return _apply_child_list_builder(
3458            *expressions,
3459            instance=self,
3460            arg="cluster",
3461            append=append,
3462            copy=copy,
3463            prefix="CLUSTER BY",
3464            into=Cluster,
3465            dialect=dialect,
3466            **opts,
3467        )
3468
3469    def select(
3470        self,
3471        *expressions: t.Optional[ExpOrStr],
3472        append: bool = True,
3473        dialect: DialectType = None,
3474        copy: bool = True,
3475        **opts,
3476    ) -> Select:
3477        return _apply_list_builder(
3478            *expressions,
3479            instance=self,
3480            arg="expressions",
3481            append=append,
3482            dialect=dialect,
3483            into=Expression,
3484            copy=copy,
3485            **opts,
3486        )
3487
3488    def lateral(
3489        self,
3490        *expressions: t.Optional[ExpOrStr],
3491        append: bool = True,
3492        dialect: DialectType = None,
3493        copy: bool = True,
3494        **opts,
3495    ) -> Select:
3496        """
3497        Append to or set the LATERAL expressions.
3498
3499        Example:
3500            >>> Select().select("x").lateral("OUTER explode(y) tbl2 AS z").from_("tbl").sql()
3501            'SELECT x FROM tbl LATERAL VIEW OUTER EXPLODE(y) tbl2 AS z'
3502
3503        Args:
3504            *expressions: the SQL code strings to parse.
3505                If an `Expression` instance is passed, it will be used as-is.
3506            append: if `True`, add to any existing expressions.
3507                Otherwise, this resets the expressions.
3508            dialect: the dialect used to parse the input expressions.
3509            copy: if `False`, modify this expression instance in-place.
3510            opts: other options to use to parse the input expressions.
3511
3512        Returns:
3513            The modified Select expression.
3514        """
3515        return _apply_list_builder(
3516            *expressions,
3517            instance=self,
3518            arg="laterals",
3519            append=append,
3520            into=Lateral,
3521            prefix="LATERAL VIEW",
3522            dialect=dialect,
3523            copy=copy,
3524            **opts,
3525        )
3526
3527    def join(
3528        self,
3529        expression: ExpOrStr,
3530        on: t.Optional[ExpOrStr] = None,
3531        using: t.Optional[ExpOrStr | t.Collection[ExpOrStr]] = None,
3532        append: bool = True,
3533        join_type: t.Optional[str] = None,
3534        join_alias: t.Optional[Identifier | str] = None,
3535        dialect: DialectType = None,
3536        copy: bool = True,
3537        **opts,
3538    ) -> Select:
3539        """
3540        Append to or set the JOIN expressions.
3541
3542        Example:
3543            >>> Select().select("*").from_("tbl").join("tbl2", on="tbl1.y = tbl2.y").sql()
3544            'SELECT * FROM tbl JOIN tbl2 ON tbl1.y = tbl2.y'
3545
3546            >>> Select().select("1").from_("a").join("b", using=["x", "y", "z"]).sql()
3547            'SELECT 1 FROM a JOIN b USING (x, y, z)'
3548
3549            Use `join_type` to change the type of join:
3550
3551            >>> Select().select("*").from_("tbl").join("tbl2", on="tbl1.y = tbl2.y", join_type="left outer").sql()
3552            'SELECT * FROM tbl LEFT OUTER JOIN tbl2 ON tbl1.y = tbl2.y'
3553
3554        Args:
3555            expression: the SQL code string to parse.
3556                If an `Expression` instance is passed, it will be used as-is.
3557            on: optionally specify the join "on" criteria as a SQL string.
3558                If an `Expression` instance is passed, it will be used as-is.
3559            using: optionally specify the join "using" criteria as a SQL string.
3560                If an `Expression` instance is passed, it will be used as-is.
3561            append: if `True`, add to any existing expressions.
3562                Otherwise, this resets the expressions.
3563            join_type: if set, alter the parsed join type.
3564            join_alias: an optional alias for the joined source.
3565            dialect: the dialect used to parse the input expressions.
3566            copy: if `False`, modify this expression instance in-place.
3567            opts: other options to use to parse the input expressions.
3568
3569        Returns:
3570            Select: the modified expression.
3571        """
3572        parse_args: t.Dict[str, t.Any] = {"dialect": dialect, **opts}
3573
3574        try:
3575            expression = maybe_parse(expression, into=Join, prefix="JOIN", **parse_args)
3576        except ParseError:
3577            expression = maybe_parse(expression, into=(Join, Expression), **parse_args)
3578
3579        join = expression if isinstance(expression, Join) else Join(this=expression)
3580
3581        if isinstance(join.this, Select):
3582            join.this.replace(join.this.subquery())
3583
3584        if join_type:
3585            method: t.Optional[Token]
3586            side: t.Optional[Token]
3587            kind: t.Optional[Token]
3588
3589            method, side, kind = maybe_parse(join_type, into="JOIN_TYPE", **parse_args)  # type: ignore
3590
3591            if method:
3592                join.set("method", method.text)
3593            if side:
3594                join.set("side", side.text)
3595            if kind:
3596                join.set("kind", kind.text)
3597
3598        if on:
3599            on = and_(*ensure_list(on), dialect=dialect, copy=copy, **opts)
3600            join.set("on", on)
3601
3602        if using:
3603            join = _apply_list_builder(
3604                *ensure_list(using),
3605                instance=join,
3606                arg="using",
3607                append=append,
3608                copy=copy,
3609                into=Identifier,
3610                **opts,
3611            )
3612
3613        if join_alias:
3614            join.set("this", alias_(join.this, join_alias, table=True))
3615
3616        return _apply_list_builder(
3617            join,
3618            instance=self,
3619            arg="joins",
3620            append=append,
3621            copy=copy,
3622            **opts,
3623        )
3624
3625    def where(
3626        self,
3627        *expressions: t.Optional[ExpOrStr],
3628        append: bool = True,
3629        dialect: DialectType = None,
3630        copy: bool = True,
3631        **opts,
3632    ) -> Select:
3633        """
3634        Append to or set the WHERE expressions.
3635
3636        Example:
3637            >>> Select().select("x").from_("tbl").where("x = 'a' OR x < 'b'").sql()
3638            "SELECT x FROM tbl WHERE x = 'a' OR x < 'b'"
3639
3640        Args:
3641            *expressions: the SQL code strings to parse.
3642                If an `Expression` instance is passed, it will be used as-is.
3643                Multiple expressions are combined with an AND operator.
3644            append: if `True`, AND the new expressions to any existing expression.
3645                Otherwise, this resets the expression.
3646            dialect: the dialect used to parse the input expressions.
3647            copy: if `False`, modify this expression instance in-place.
3648            opts: other options to use to parse the input expressions.
3649
3650        Returns:
3651            Select: the modified expression.
3652        """
3653        return _apply_conjunction_builder(
3654            *expressions,
3655            instance=self,
3656            arg="where",
3657            append=append,
3658            into=Where,
3659            dialect=dialect,
3660            copy=copy,
3661            **opts,
3662        )
3663
3664    def having(
3665        self,
3666        *expressions: t.Optional[ExpOrStr],
3667        append: bool = True,
3668        dialect: DialectType = None,
3669        copy: bool = True,
3670        **opts,
3671    ) -> Select:
3672        """
3673        Append to or set the HAVING expressions.
3674
3675        Example:
3676            >>> Select().select("x", "COUNT(y)").from_("tbl").group_by("x").having("COUNT(y) > 3").sql()
3677            'SELECT x, COUNT(y) FROM tbl GROUP BY x HAVING COUNT(y) > 3'
3678
3679        Args:
3680            *expressions: the SQL code strings to parse.
3681                If an `Expression` instance is passed, it will be used as-is.
3682                Multiple expressions are combined with an AND operator.
3683            append: if `True`, AND the new expressions to any existing expression.
3684                Otherwise, this resets the expression.
3685            dialect: the dialect used to parse the input expressions.
3686            copy: if `False`, modify this expression instance in-place.
3687            opts: other options to use to parse the input expressions.
3688
3689        Returns:
3690            The modified Select expression.
3691        """
3692        return _apply_conjunction_builder(
3693            *expressions,
3694            instance=self,
3695            arg="having",
3696            append=append,
3697            into=Having,
3698            dialect=dialect,
3699            copy=copy,
3700            **opts,
3701        )
3702
3703    def window(
3704        self,
3705        *expressions: t.Optional[ExpOrStr],
3706        append: bool = True,
3707        dialect: DialectType = None,
3708        copy: bool = True,
3709        **opts,
3710    ) -> Select:
3711        return _apply_list_builder(
3712            *expressions,
3713            instance=self,
3714            arg="windows",
3715            append=append,
3716            into=Window,
3717            dialect=dialect,
3718            copy=copy,
3719            **opts,
3720        )
3721
3722    def qualify(
3723        self,
3724        *expressions: t.Optional[ExpOrStr],
3725        append: bool = True,
3726        dialect: DialectType = None,
3727        copy: bool = True,
3728        **opts,
3729    ) -> Select:
3730        return _apply_conjunction_builder(
3731            *expressions,
3732            instance=self,
3733            arg="qualify",
3734            append=append,
3735            into=Qualify,
3736            dialect=dialect,
3737            copy=copy,
3738            **opts,
3739        )
3740
3741    def distinct(
3742        self, *ons: t.Optional[ExpOrStr], distinct: bool = True, copy: bool = True
3743    ) -> Select:
3744        """
3745        Set the OFFSET expression.
3746
3747        Example:
3748            >>> Select().from_("tbl").select("x").distinct().sql()
3749            'SELECT DISTINCT x FROM tbl'
3750
3751        Args:
3752            ons: the expressions to distinct on
3753            distinct: whether the Select should be distinct
3754            copy: if `False`, modify this expression instance in-place.
3755
3756        Returns:
3757            Select: the modified expression.
3758        """
3759        instance = maybe_copy(self, copy)
3760        on = Tuple(expressions=[maybe_parse(on, copy=copy) for on in ons if on]) if ons else None
3761        instance.set("distinct", Distinct(on=on) if distinct else None)
3762        return instance
3763
3764    def ctas(
3765        self,
3766        table: ExpOrStr,
3767        properties: t.Optional[t.Dict] = None,
3768        dialect: DialectType = None,
3769        copy: bool = True,
3770        **opts,
3771    ) -> Create:
3772        """
3773        Convert this expression to a CREATE TABLE AS statement.
3774
3775        Example:
3776            >>> Select().select("*").from_("tbl").ctas("x").sql()
3777            'CREATE TABLE x AS SELECT * FROM tbl'
3778
3779        Args:
3780            table: the SQL code string to parse as the table name.
3781                If another `Expression` instance is passed, it will be used as-is.
3782            properties: an optional mapping of table properties
3783            dialect: the dialect used to parse the input table.
3784            copy: if `False`, modify this expression instance in-place.
3785            opts: other options to use to parse the input table.
3786
3787        Returns:
3788            The new Create expression.
3789        """
3790        instance = maybe_copy(self, copy)
3791        table_expression = maybe_parse(table, into=Table, dialect=dialect, **opts)
3792
3793        properties_expression = None
3794        if properties:
3795            properties_expression = Properties.from_dict(properties)
3796
3797        return Create(
3798            this=table_expression,
3799            kind="TABLE",
3800            expression=instance,
3801            properties=properties_expression,
3802        )
3803
3804    def lock(self, update: bool = True, copy: bool = True) -> Select:
3805        """
3806        Set the locking read mode for this expression.
3807
3808        Examples:
3809            >>> Select().select("x").from_("tbl").where("x = 'a'").lock().sql("mysql")
3810            "SELECT x FROM tbl WHERE x = 'a' FOR UPDATE"
3811
3812            >>> Select().select("x").from_("tbl").where("x = 'a'").lock(update=False).sql("mysql")
3813            "SELECT x FROM tbl WHERE x = 'a' FOR SHARE"
3814
3815        Args:
3816            update: if `True`, the locking type will be `FOR UPDATE`, else it will be `FOR SHARE`.
3817            copy: if `False`, modify this expression instance in-place.
3818
3819        Returns:
3820            The modified expression.
3821        """
3822        inst = maybe_copy(self, copy)
3823        inst.set("locks", [Lock(update=update)])
3824
3825        return inst
3826
3827    def hint(self, *hints: ExpOrStr, dialect: DialectType = None, copy: bool = True) -> Select:
3828        """
3829        Set hints for this expression.
3830
3831        Examples:
3832            >>> Select().select("x").from_("tbl").hint("BROADCAST(y)").sql(dialect="spark")
3833            'SELECT /*+ BROADCAST(y) */ x FROM tbl'
3834
3835        Args:
3836            hints: The SQL code strings to parse as the hints.
3837                If an `Expression` instance is passed, it will be used as-is.
3838            dialect: The dialect used to parse the hints.
3839            copy: If `False`, modify this expression instance in-place.
3840
3841        Returns:
3842            The modified expression.
3843        """
3844        inst = maybe_copy(self, copy)
3845        inst.set(
3846            "hint", Hint(expressions=[maybe_parse(h, copy=copy, dialect=dialect) for h in hints])
3847        )
3848
3849        return inst
3850
3851    @property
3852    def named_selects(self) -> t.List[str]:
3853        return [e.output_name for e in self.expressions if e.alias_or_name]
3854
3855    @property
3856    def is_star(self) -> bool:
3857        return any(expression.is_star for expression in self.expressions)
3858
3859    @property
3860    def selects(self) -> t.List[Expression]:
3861        return self.expressions
3862
3863
3864UNWRAPPED_QUERIES = (Select, SetOperation)
3865
3866
3867class Subquery(DerivedTable, Query):
3868    arg_types = {
3869        "this": True,
3870        "alias": False,
3871        "with": False,
3872        **QUERY_MODIFIERS,
3873    }
3874
3875    def unnest(self):
3876        """Returns the first non subquery."""
3877        expression = self
3878        while isinstance(expression, Subquery):
3879            expression = expression.this
3880        return expression
3881
3882    def unwrap(self) -> Subquery:
3883        expression = self
3884        while expression.same_parent and expression.is_wrapper:
3885            expression = t.cast(Subquery, expression.parent)
3886        return expression
3887
3888    def select(
3889        self,
3890        *expressions: t.Optional[ExpOrStr],
3891        append: bool = True,
3892        dialect: DialectType = None,
3893        copy: bool = True,
3894        **opts,
3895    ) -> Subquery:
3896        this = maybe_copy(self, copy)
3897        this.unnest().select(*expressions, append=append, dialect=dialect, copy=False, **opts)
3898        return this
3899
3900    @property
3901    def is_wrapper(self) -> bool:
3902        """
3903        Whether this Subquery acts as a simple wrapper around another expression.
3904
3905        SELECT * FROM (((SELECT * FROM t)))
3906                      ^
3907                      This corresponds to a "wrapper" Subquery node
3908        """
3909        return all(v is None for k, v in self.args.items() if k != "this")
3910
3911    @property
3912    def is_star(self) -> bool:
3913        return self.this.is_star
3914
3915    @property
3916    def output_name(self) -> str:
3917        return self.alias
3918
3919
3920class TableSample(Expression):
3921    arg_types = {
3922        "expressions": False,
3923        "method": False,
3924        "bucket_numerator": False,
3925        "bucket_denominator": False,
3926        "bucket_field": False,
3927        "percent": False,
3928        "rows": False,
3929        "size": False,
3930        "seed": False,
3931    }
3932
3933
3934class Tag(Expression):
3935    """Tags are used for generating arbitrary sql like SELECT <span>x</span>."""
3936
3937    arg_types = {
3938        "this": False,
3939        "prefix": False,
3940        "postfix": False,
3941    }
3942
3943
3944# Represents both the standard SQL PIVOT operator and DuckDB's "simplified" PIVOT syntax
3945# https://duckdb.org/docs/sql/statements/pivot
3946class Pivot(Expression):
3947    arg_types = {
3948        "this": False,
3949        "alias": False,
3950        "expressions": False,
3951        "field": False,
3952        "unpivot": False,
3953        "using": False,
3954        "group": False,
3955        "columns": False,
3956        "include_nulls": False,
3957        "default_on_null": False,
3958    }
3959
3960    @property
3961    def unpivot(self) -> bool:
3962        return bool(self.args.get("unpivot"))
3963
3964
3965class Window(Condition):
3966    arg_types = {
3967        "this": True,
3968        "partition_by": False,
3969        "order": False,
3970        "spec": False,
3971        "alias": False,
3972        "over": False,
3973        "first": False,
3974    }
3975
3976
3977class WindowSpec(Expression):
3978    arg_types = {
3979        "kind": False,
3980        "start": False,
3981        "start_side": False,
3982        "end": False,
3983        "end_side": False,
3984    }
3985
3986
3987class PreWhere(Expression):
3988    pass
3989
3990
3991class Where(Expression):
3992    pass
3993
3994
3995class Star(Expression):
3996    arg_types = {"except": False, "replace": False, "rename": False}
3997
3998    @property
3999    def name(self) -> str:
4000        return "*"
4001
4002    @property
4003    def output_name(self) -> str:
4004        return self.name
4005
4006
4007class Parameter(Condition):
4008    arg_types = {"this": True, "expression": False}
4009
4010
4011class SessionParameter(Condition):
4012    arg_types = {"this": True, "kind": False}
4013
4014
4015class Placeholder(Condition):
4016    arg_types = {"this": False, "kind": False}
4017
4018    @property
4019    def name(self) -> str:
4020        return self.this or "?"
4021
4022
4023class Null(Condition):
4024    arg_types: t.Dict[str, t.Any] = {}
4025
4026    @property
4027    def name(self) -> str:
4028        return "NULL"
4029
4030    def to_py(self) -> Lit[None]:
4031        return None
4032
4033
4034class Boolean(Condition):
4035    def to_py(self) -> bool:
4036        return self.this
4037
4038
4039class DataTypeParam(Expression):
4040    arg_types = {"this": True, "expression": False}
4041
4042    @property
4043    def name(self) -> str:
4044        return self.this.name
4045
4046
4047# The `nullable` arg is helpful when transpiling types from other dialects to ClickHouse, which
4048# assumes non-nullable types by default. Values `None` and `True` mean the type is nullable.
4049class DataType(Expression):
4050    arg_types = {
4051        "this": True,
4052        "expressions": False,
4053        "nested": False,
4054        "values": False,
4055        "prefix": False,
4056        "kind": False,
4057        "nullable": False,
4058    }
4059
4060    class Type(AutoName):
4061        ARRAY = auto()
4062        AGGREGATEFUNCTION = auto()
4063        SIMPLEAGGREGATEFUNCTION = auto()
4064        BIGDECIMAL = auto()
4065        BIGINT = auto()
4066        BIGSERIAL = auto()
4067        BINARY = auto()
4068        BIT = auto()
4069        BOOLEAN = auto()
4070        BPCHAR = auto()
4071        CHAR = auto()
4072        DATE = auto()
4073        DATE32 = auto()
4074        DATEMULTIRANGE = auto()
4075        DATERANGE = auto()
4076        DATETIME = auto()
4077        DATETIME64 = auto()
4078        DECIMAL = auto()
4079        DECIMAL32 = auto()
4080        DECIMAL64 = auto()
4081        DECIMAL128 = auto()
4082        DOUBLE = auto()
4083        ENUM = auto()
4084        ENUM8 = auto()
4085        ENUM16 = auto()
4086        FIXEDSTRING = auto()
4087        FLOAT = auto()
4088        GEOGRAPHY = auto()
4089        GEOMETRY = auto()
4090        HLLSKETCH = auto()
4091        HSTORE = auto()
4092        IMAGE = auto()
4093        INET = auto()
4094        INT = auto()
4095        INT128 = auto()
4096        INT256 = auto()
4097        INT4MULTIRANGE = auto()
4098        INT4RANGE = auto()
4099        INT8MULTIRANGE = auto()
4100        INT8RANGE = auto()
4101        INTERVAL = auto()
4102        IPADDRESS = auto()
4103        IPPREFIX = auto()
4104        IPV4 = auto()
4105        IPV6 = auto()
4106        JSON = auto()
4107        JSONB = auto()
4108        LIST = auto()
4109        LONGBLOB = auto()
4110        LONGTEXT = auto()
4111        LOWCARDINALITY = auto()
4112        MAP = auto()
4113        MEDIUMBLOB = auto()
4114        MEDIUMINT = auto()
4115        MEDIUMTEXT = auto()
4116        MONEY = auto()
4117        NAME = auto()
4118        NCHAR = auto()
4119        NESTED = auto()
4120        NULL = auto()
4121        NULLABLE = auto()
4122        NUMMULTIRANGE = auto()
4123        NUMRANGE = auto()
4124        NVARCHAR = auto()
4125        OBJECT = auto()
4126        ROWVERSION = auto()
4127        SERIAL = auto()
4128        SET = auto()
4129        SMALLINT = auto()
4130        SMALLMONEY = auto()
4131        SMALLSERIAL = auto()
4132        STRUCT = auto()
4133        SUPER = auto()
4134        TEXT = auto()
4135        TINYBLOB = auto()
4136        TINYTEXT = auto()
4137        TIME = auto()
4138        TIMETZ = auto()
4139        TIMESTAMP = auto()
4140        TIMESTAMPNTZ = auto()
4141        TIMESTAMPLTZ = auto()
4142        TIMESTAMPTZ = auto()
4143        TIMESTAMP_S = auto()
4144        TIMESTAMP_MS = auto()
4145        TIMESTAMP_NS = auto()
4146        TINYINT = auto()
4147        TSMULTIRANGE = auto()
4148        TSRANGE = auto()
4149        TSTZMULTIRANGE = auto()
4150        TSTZRANGE = auto()
4151        UBIGINT = auto()
4152        UINT = auto()
4153        UINT128 = auto()
4154        UINT256 = auto()
4155        UMEDIUMINT = auto()
4156        UDECIMAL = auto()
4157        UNIQUEIDENTIFIER = auto()
4158        UNKNOWN = auto()  # Sentinel value, useful for type annotation
4159        USERDEFINED = "USER-DEFINED"
4160        USMALLINT = auto()
4161        UTINYINT = auto()
4162        UUID = auto()
4163        VARBINARY = auto()
4164        VARCHAR = auto()
4165        VARIANT = auto()
4166        VECTOR = auto()
4167        XML = auto()
4168        YEAR = auto()
4169        TDIGEST = auto()
4170
4171    STRUCT_TYPES = {
4172        Type.NESTED,
4173        Type.OBJECT,
4174        Type.STRUCT,
4175    }
4176
4177    NESTED_TYPES = {
4178        *STRUCT_TYPES,
4179        Type.ARRAY,
4180        Type.MAP,
4181    }
4182
4183    TEXT_TYPES = {
4184        Type.CHAR,
4185        Type.NCHAR,
4186        Type.NVARCHAR,
4187        Type.TEXT,
4188        Type.VARCHAR,
4189        Type.NAME,
4190    }
4191
4192    SIGNED_INTEGER_TYPES = {
4193        Type.BIGINT,
4194        Type.INT,
4195        Type.INT128,
4196        Type.INT256,
4197        Type.MEDIUMINT,
4198        Type.SMALLINT,
4199        Type.TINYINT,
4200    }
4201
4202    UNSIGNED_INTEGER_TYPES = {
4203        Type.UBIGINT,
4204        Type.UINT,
4205        Type.UINT128,
4206        Type.UINT256,
4207        Type.UMEDIUMINT,
4208        Type.USMALLINT,
4209        Type.UTINYINT,
4210    }
4211
4212    INTEGER_TYPES = {
4213        *SIGNED_INTEGER_TYPES,
4214        *UNSIGNED_INTEGER_TYPES,
4215        Type.BIT,
4216    }
4217
4218    FLOAT_TYPES = {
4219        Type.DOUBLE,
4220        Type.FLOAT,
4221    }
4222
4223    REAL_TYPES = {
4224        *FLOAT_TYPES,
4225        Type.BIGDECIMAL,
4226        Type.DECIMAL,
4227        Type.DECIMAL32,
4228        Type.DECIMAL64,
4229        Type.DECIMAL128,
4230        Type.MONEY,
4231        Type.SMALLMONEY,
4232        Type.UDECIMAL,
4233    }
4234
4235    NUMERIC_TYPES = {
4236        *INTEGER_TYPES,
4237        *REAL_TYPES,
4238    }
4239
4240    TEMPORAL_TYPES = {
4241        Type.DATE,
4242        Type.DATE32,
4243        Type.DATETIME,
4244        Type.DATETIME64,
4245        Type.TIME,
4246        Type.TIMESTAMP,
4247        Type.TIMESTAMPNTZ,
4248        Type.TIMESTAMPLTZ,
4249        Type.TIMESTAMPTZ,
4250        Type.TIMESTAMP_MS,
4251        Type.TIMESTAMP_NS,
4252        Type.TIMESTAMP_S,
4253        Type.TIMETZ,
4254    }
4255
4256    @classmethod
4257    def build(
4258        cls,
4259        dtype: DATA_TYPE,
4260        dialect: DialectType = None,
4261        udt: bool = False,
4262        copy: bool = True,
4263        **kwargs,
4264    ) -> DataType:
4265        """
4266        Constructs a DataType object.
4267
4268        Args:
4269            dtype: the data type of interest.
4270            dialect: the dialect to use for parsing `dtype`, in case it's a string.
4271            udt: when set to True, `dtype` will be used as-is if it can't be parsed into a
4272                DataType, thus creating a user-defined type.
4273            copy: whether to copy the data type.
4274            kwargs: additional arguments to pass in the constructor of DataType.
4275
4276        Returns:
4277            The constructed DataType object.
4278        """
4279        from sqlglot import parse_one
4280
4281        if isinstance(dtype, str):
4282            if dtype.upper() == "UNKNOWN":
4283                return DataType(this=DataType.Type.UNKNOWN, **kwargs)
4284
4285            try:
4286                data_type_exp = parse_one(
4287                    dtype, read=dialect, into=DataType, error_level=ErrorLevel.IGNORE
4288                )
4289            except ParseError:
4290                if udt:
4291                    return DataType(this=DataType.Type.USERDEFINED, kind=dtype, **kwargs)
4292                raise
4293        elif isinstance(dtype, DataType.Type):
4294            data_type_exp = DataType(this=dtype)
4295        elif isinstance(dtype, DataType):
4296            return maybe_copy(dtype, copy)
4297        else:
4298            raise ValueError(f"Invalid data type: {type(dtype)}. Expected str or DataType.Type")
4299
4300        return DataType(**{**data_type_exp.args, **kwargs})
4301
4302    def is_type(self, *dtypes: DATA_TYPE, check_nullable: bool = False) -> bool:
4303        """
4304        Checks whether this DataType matches one of the provided data types. Nested types or precision
4305        will be compared using "structural equivalence" semantics, so e.g. array<int> != array<float>.
4306
4307        Args:
4308            dtypes: the data types to compare this DataType to.
4309            check_nullable: whether to take the NULLABLE type constructor into account for the comparison.
4310                If false, it means that NULLABLE<INT> is equivalent to INT.
4311
4312        Returns:
4313            True, if and only if there is a type in `dtypes` which is equal to this DataType.
4314        """
4315        if (
4316            not check_nullable
4317            and self.this == DataType.Type.NULLABLE
4318            and len(self.expressions) == 1
4319        ):
4320            this_type = self.expressions[0]
4321        else:
4322            this_type = self
4323
4324        for dtype in dtypes:
4325            other_type = DataType.build(dtype, copy=False, udt=True)
4326            if (
4327                not check_nullable
4328                and other_type.this == DataType.Type.NULLABLE
4329                and len(other_type.expressions) == 1
4330            ):
4331                other_type = other_type.expressions[0]
4332
4333            if (
4334                other_type.expressions
4335                or this_type.this == DataType.Type.USERDEFINED
4336                or other_type.this == DataType.Type.USERDEFINED
4337            ):
4338                matches = this_type == other_type
4339            else:
4340                matches = this_type.this == other_type.this
4341
4342            if matches:
4343                return True
4344        return False
4345
4346
4347DATA_TYPE = t.Union[str, DataType, DataType.Type]
4348
4349
4350# https://www.postgresql.org/docs/15/datatype-pseudo.html
4351class PseudoType(DataType):
4352    arg_types = {"this": True}
4353
4354
4355# https://www.postgresql.org/docs/15/datatype-oid.html
4356class ObjectIdentifier(DataType):
4357    arg_types = {"this": True}
4358
4359
4360# WHERE x <OP> EXISTS|ALL|ANY|SOME(SELECT ...)
4361class SubqueryPredicate(Predicate):
4362    pass
4363
4364
4365class All(SubqueryPredicate):
4366    pass
4367
4368
4369class Any(SubqueryPredicate):
4370    pass
4371
4372
4373class Exists(SubqueryPredicate):
4374    pass
4375
4376
4377# Commands to interact with the databases or engines. For most of the command
4378# expressions we parse whatever comes after the command's name as a string.
4379class Command(Expression):
4380    arg_types = {"this": True, "expression": False}
4381
4382
4383class Transaction(Expression):
4384    arg_types = {"this": False, "modes": False, "mark": False}
4385
4386
4387class Commit(Expression):
4388    arg_types = {"chain": False, "this": False, "durability": False}
4389
4390
4391class Rollback(Expression):
4392    arg_types = {"savepoint": False, "this": False}
4393
4394
4395class Alter(Expression):
4396    arg_types = {
4397        "this": True,
4398        "kind": True,
4399        "actions": True,
4400        "exists": False,
4401        "only": False,
4402        "options": False,
4403        "cluster": False,
4404        "not_valid": False,
4405    }
4406
4407
4408class AddConstraint(Expression):
4409    arg_types = {"expressions": True}
4410
4411
4412class DropPartition(Expression):
4413    arg_types = {"expressions": True, "exists": False}
4414
4415
4416# https://clickhouse.com/docs/en/sql-reference/statements/alter/partition#replace-partition
4417class ReplacePartition(Expression):
4418    arg_types = {"expression": True, "source": True}
4419
4420
4421# Binary expressions like (ADD a b)
4422class Binary(Condition):
4423    arg_types = {"this": True, "expression": True}
4424
4425    @property
4426    def left(self) -> Expression:
4427        return self.this
4428
4429    @property
4430    def right(self) -> Expression:
4431        return self.expression
4432
4433
4434class Add(Binary):
4435    pass
4436
4437
4438class Connector(Binary):
4439    pass
4440
4441
4442class And(Connector):
4443    pass
4444
4445
4446class Or(Connector):
4447    pass
4448
4449
4450class BitwiseAnd(Binary):
4451    pass
4452
4453
4454class BitwiseLeftShift(Binary):
4455    pass
4456
4457
4458class BitwiseOr(Binary):
4459    pass
4460
4461
4462class BitwiseRightShift(Binary):
4463    pass
4464
4465
4466class BitwiseXor(Binary):
4467    pass
4468
4469
4470class Div(Binary):
4471    arg_types = {"this": True, "expression": True, "typed": False, "safe": False}
4472
4473
4474class Overlaps(Binary):
4475    pass
4476
4477
4478class Dot(Binary):
4479    @property
4480    def is_star(self) -> bool:
4481        return self.expression.is_star
4482
4483    @property
4484    def name(self) -> str:
4485        return self.expression.name
4486
4487    @property
4488    def output_name(self) -> str:
4489        return self.name
4490
4491    @classmethod
4492    def build(self, expressions: t.Sequence[Expression]) -> Dot:
4493        """Build a Dot object with a sequence of expressions."""
4494        if len(expressions) < 2:
4495            raise ValueError("Dot requires >= 2 expressions.")
4496
4497        return t.cast(Dot, reduce(lambda x, y: Dot(this=x, expression=y), expressions))
4498
4499    @property
4500    def parts(self) -> t.List[Expression]:
4501        """Return the parts of a table / column in order catalog, db, table."""
4502        this, *parts = self.flatten()
4503
4504        parts.reverse()
4505
4506        for arg in COLUMN_PARTS:
4507            part = this.args.get(arg)
4508
4509            if isinstance(part, Expression):
4510                parts.append(part)
4511
4512        parts.reverse()
4513        return parts
4514
4515
4516class DPipe(Binary):
4517    arg_types = {"this": True, "expression": True, "safe": False}
4518
4519
4520class EQ(Binary, Predicate):
4521    pass
4522
4523
4524class NullSafeEQ(Binary, Predicate):
4525    pass
4526
4527
4528class NullSafeNEQ(Binary, Predicate):
4529    pass
4530
4531
4532# Represents e.g. := in DuckDB which is mostly used for setting parameters
4533class PropertyEQ(Binary):
4534    pass
4535
4536
4537class Distance(Binary):
4538    pass
4539
4540
4541class Escape(Binary):
4542    pass
4543
4544
4545class Glob(Binary, Predicate):
4546    pass
4547
4548
4549class GT(Binary, Predicate):
4550    pass
4551
4552
4553class GTE(Binary, Predicate):
4554    pass
4555
4556
4557class ILike(Binary, Predicate):
4558    pass
4559
4560
4561class ILikeAny(Binary, Predicate):
4562    pass
4563
4564
4565class IntDiv(Binary):
4566    pass
4567
4568
4569class Is(Binary, Predicate):
4570    pass
4571
4572
4573class Kwarg(Binary):
4574    """Kwarg in special functions like func(kwarg => y)."""
4575
4576
4577class Like(Binary, Predicate):
4578    pass
4579
4580
4581class LikeAny(Binary, Predicate):
4582    pass
4583
4584
4585class LT(Binary, Predicate):
4586    pass
4587
4588
4589class LTE(Binary, Predicate):
4590    pass
4591
4592
4593class Mod(Binary):
4594    pass
4595
4596
4597class Mul(Binary):
4598    pass
4599
4600
4601class NEQ(Binary, Predicate):
4602    pass
4603
4604
4605# https://www.postgresql.org/docs/current/ddl-schemas.html#DDL-SCHEMAS-PATH
4606class Operator(Binary):
4607    arg_types = {"this": True, "operator": True, "expression": True}
4608
4609
4610class SimilarTo(Binary, Predicate):
4611    pass
4612
4613
4614class Slice(Binary):
4615    arg_types = {"this": False, "expression": False}
4616
4617
4618class Sub(Binary):
4619    pass
4620
4621
4622# Unary Expressions
4623# (NOT a)
4624class Unary(Condition):
4625    pass
4626
4627
4628class BitwiseNot(Unary):
4629    pass
4630
4631
4632class Not(Unary):
4633    pass
4634
4635
4636class Paren(Unary):
4637    @property
4638    def output_name(self) -> str:
4639        return self.this.name
4640
4641
4642class Neg(Unary):
4643    def to_py(self) -> int | Decimal:
4644        if self.is_number:
4645            return self.this.to_py() * -1
4646        return super().to_py()
4647
4648
4649class Alias(Expression):
4650    arg_types = {"this": True, "alias": False}
4651
4652    @property
4653    def output_name(self) -> str:
4654        return self.alias
4655
4656
4657# BigQuery requires the UNPIVOT column list aliases to be either strings or ints, but
4658# other dialects require identifiers. This enables us to transpile between them easily.
4659class PivotAlias(Alias):
4660    pass
4661
4662
4663# Represents Snowflake's ANY [ ORDER BY ... ] syntax
4664# https://docs.snowflake.com/en/sql-reference/constructs/pivot
4665class PivotAny(Expression):
4666    arg_types = {"this": False}
4667
4668
4669class Aliases(Expression):
4670    arg_types = {"this": True, "expressions": True}
4671
4672    @property
4673    def aliases(self):
4674        return self.expressions
4675
4676
4677# https://docs.aws.amazon.com/redshift/latest/dg/query-super.html
4678class AtIndex(Expression):
4679    arg_types = {"this": True, "expression": True}
4680
4681
4682class AtTimeZone(Expression):
4683    arg_types = {"this": True, "zone": True}
4684
4685
4686class FromTimeZone(Expression):
4687    arg_types = {"this": True, "zone": True}
4688
4689
4690class Between(Predicate):
4691    arg_types = {"this": True, "low": True, "high": True}
4692
4693
4694class Bracket(Condition):
4695    # https://cloud.google.com/bigquery/docs/reference/standard-sql/operators#array_subscript_operator
4696    arg_types = {
4697        "this": True,
4698        "expressions": True,
4699        "offset": False,
4700        "safe": False,
4701        "returns_list_for_maps": False,
4702    }
4703
4704    @property
4705    def output_name(self) -> str:
4706        if len(self.expressions) == 1:
4707            return self.expressions[0].output_name
4708
4709        return super().output_name
4710
4711
4712class Distinct(Expression):
4713    arg_types = {"expressions": False, "on": False}
4714
4715
4716class In(Predicate):
4717    arg_types = {
4718        "this": True,
4719        "expressions": False,
4720        "query": False,
4721        "unnest": False,
4722        "field": False,
4723        "is_global": False,
4724    }
4725
4726
4727# https://cloud.google.com/bigquery/docs/reference/standard-sql/procedural-language#for-in
4728class ForIn(Expression):
4729    arg_types = {"this": True, "expression": True}
4730
4731
4732class TimeUnit(Expression):
4733    """Automatically converts unit arg into a var."""
4734
4735    arg_types = {"unit": False}
4736
4737    UNABBREVIATED_UNIT_NAME = {
4738        "D": "DAY",
4739        "H": "HOUR",
4740        "M": "MINUTE",
4741        "MS": "MILLISECOND",
4742        "NS": "NANOSECOND",
4743        "Q": "QUARTER",
4744        "S": "SECOND",
4745        "US": "MICROSECOND",
4746        "W": "WEEK",
4747        "Y": "YEAR",
4748    }
4749
4750    VAR_LIKE = (Column, Literal, Var)
4751
4752    def __init__(self, **args):
4753        unit = args.get("unit")
4754        if isinstance(unit, self.VAR_LIKE):
4755            args["unit"] = Var(
4756                this=(self.UNABBREVIATED_UNIT_NAME.get(unit.name) or unit.name).upper()
4757            )
4758        elif isinstance(unit, Week):
4759            unit.set("this", Var(this=unit.this.name.upper()))
4760
4761        super().__init__(**args)
4762
4763    @property
4764    def unit(self) -> t.Optional[Var | IntervalSpan]:
4765        return self.args.get("unit")
4766
4767
4768class IntervalOp(TimeUnit):
4769    arg_types = {"unit": True, "expression": True}
4770
4771    def interval(self):
4772        return Interval(
4773            this=self.expression.copy(),
4774            unit=self.unit.copy(),
4775        )
4776
4777
4778# https://www.oracletutorial.com/oracle-basics/oracle-interval/
4779# https://trino.io/docs/current/language/types.html#interval-day-to-second
4780# https://docs.databricks.com/en/sql/language-manual/data-types/interval-type.html
4781class IntervalSpan(DataType):
4782    arg_types = {"this": True, "expression": True}
4783
4784
4785class Interval(TimeUnit):
4786    arg_types = {"this": False, "unit": False}
4787
4788
4789class IgnoreNulls(Expression):
4790    pass
4791
4792
4793class RespectNulls(Expression):
4794    pass
4795
4796
4797# https://cloud.google.com/bigquery/docs/reference/standard-sql/aggregate-function-calls#max_min_clause
4798class HavingMax(Expression):
4799    arg_types = {"this": True, "expression": True, "max": True}
4800
4801
4802# Functions
4803class Func(Condition):
4804    """
4805    The base class for all function expressions.
4806
4807    Attributes:
4808        is_var_len_args (bool): if set to True the last argument defined in arg_types will be
4809            treated as a variable length argument and the argument's value will be stored as a list.
4810        _sql_names (list): the SQL name (1st item in the list) and aliases (subsequent items) for this
4811            function expression. These values are used to map this node to a name during parsing as
4812            well as to provide the function's name during SQL string generation. By default the SQL
4813            name is set to the expression's class name transformed to snake case.
4814    """
4815
4816    is_var_len_args = False
4817
4818    @classmethod
4819    def from_arg_list(cls, args):
4820        if cls.is_var_len_args:
4821            all_arg_keys = list(cls.arg_types)
4822            # If this function supports variable length argument treat the last argument as such.
4823            non_var_len_arg_keys = all_arg_keys[:-1] if cls.is_var_len_args else all_arg_keys
4824            num_non_var = len(non_var_len_arg_keys)
4825
4826            args_dict = {arg_key: arg for arg, arg_key in zip(args, non_var_len_arg_keys)}
4827            args_dict[all_arg_keys[-1]] = args[num_non_var:]
4828        else:
4829            args_dict = {arg_key: arg for arg, arg_key in zip(args, cls.arg_types)}
4830
4831        return cls(**args_dict)
4832
4833    @classmethod
4834    def sql_names(cls):
4835        if cls is Func:
4836            raise NotImplementedError(
4837                "SQL name is only supported by concrete function implementations"
4838            )
4839        if "_sql_names" not in cls.__dict__:
4840            cls._sql_names = [camel_to_snake_case(cls.__name__)]
4841        return cls._sql_names
4842
4843    @classmethod
4844    def sql_name(cls):
4845        return cls.sql_names()[0]
4846
4847    @classmethod
4848    def default_parser_mappings(cls):
4849        return {name: cls.from_arg_list for name in cls.sql_names()}
4850
4851
4852class AggFunc(Func):
4853    pass
4854
4855
4856class ParameterizedAgg(AggFunc):
4857    arg_types = {"this": True, "expressions": True, "params": True}
4858
4859
4860class Abs(Func):
4861    pass
4862
4863
4864class ArgMax(AggFunc):
4865    arg_types = {"this": True, "expression": True, "count": False}
4866    _sql_names = ["ARG_MAX", "ARGMAX", "MAX_BY"]
4867
4868
4869class ArgMin(AggFunc):
4870    arg_types = {"this": True, "expression": True, "count": False}
4871    _sql_names = ["ARG_MIN", "ARGMIN", "MIN_BY"]
4872
4873
4874class ApproxTopK(AggFunc):
4875    arg_types = {"this": True, "expression": False, "counters": False}
4876
4877
4878class Flatten(Func):
4879    pass
4880
4881
4882# https://spark.apache.org/docs/latest/api/sql/index.html#transform
4883class Transform(Func):
4884    arg_types = {"this": True, "expression": True}
4885
4886
4887class Anonymous(Func):
4888    arg_types = {"this": True, "expressions": False}
4889    is_var_len_args = True
4890
4891    @property
4892    def name(self) -> str:
4893        return self.this if isinstance(self.this, str) else self.this.name
4894
4895
4896class AnonymousAggFunc(AggFunc):
4897    arg_types = {"this": True, "expressions": False}
4898    is_var_len_args = True
4899
4900
4901# https://clickhouse.com/docs/en/sql-reference/aggregate-functions/combinators
4902class CombinedAggFunc(AnonymousAggFunc):
4903    arg_types = {"this": True, "expressions": False, "parts": True}
4904
4905
4906class CombinedParameterizedAgg(ParameterizedAgg):
4907    arg_types = {"this": True, "expressions": True, "params": True, "parts": True}
4908
4909
4910# https://docs.snowflake.com/en/sql-reference/functions/hll
4911# https://docs.aws.amazon.com/redshift/latest/dg/r_HLL_function.html
4912class Hll(AggFunc):
4913    arg_types = {"this": True, "expressions": False}
4914    is_var_len_args = True
4915
4916
4917class ApproxDistinct(AggFunc):
4918    arg_types = {"this": True, "accuracy": False}
4919    _sql_names = ["APPROX_DISTINCT", "APPROX_COUNT_DISTINCT"]
4920
4921
4922class Array(Func):
4923    arg_types = {"expressions": False, "bracket_notation": False}
4924    is_var_len_args = True
4925
4926
4927# https://docs.snowflake.com/en/sql-reference/functions/to_array
4928class ToArray(Func):
4929    pass
4930
4931
4932# https://materialize.com/docs/sql/types/list/
4933class List(Func):
4934    arg_types = {"expressions": False}
4935    is_var_len_args = True
4936
4937
4938# String pad, kind True -> LPAD, False -> RPAD
4939class Pad(Func):
4940    arg_types = {"this": True, "expression": True, "fill_pattern": False, "is_left": True}
4941
4942
4943# https://docs.snowflake.com/en/sql-reference/functions/to_char
4944# https://docs.oracle.com/en/database/oracle/oracle-database/23/sqlrf/TO_CHAR-number.html
4945class ToChar(Func):
4946    arg_types = {"this": True, "format": False, "nlsparam": False}
4947
4948
4949# https://docs.snowflake.com/en/sql-reference/functions/to_decimal
4950# https://docs.oracle.com/en/database/oracle/oracle-database/23/sqlrf/TO_NUMBER.html
4951class ToNumber(Func):
4952    arg_types = {
4953        "this": True,
4954        "format": False,
4955        "nlsparam": False,
4956        "precision": False,
4957        "scale": False,
4958    }
4959
4960
4961# https://learn.microsoft.com/en-us/sql/t-sql/functions/cast-and-convert-transact-sql?view=sql-server-ver16#syntax
4962class Convert(Func):
4963    arg_types = {"this": True, "expression": True, "style": False}
4964
4965
4966class ConvertTimezone(Func):
4967    arg_types = {"source_tz": False, "target_tz": True, "timestamp": True}
4968
4969
4970class GenerateSeries(Func):
4971    arg_types = {"start": True, "end": True, "step": False, "is_end_exclusive": False}
4972
4973
4974# Postgres' GENERATE_SERIES function returns a row set, i.e. it implicitly explodes when it's
4975# used in a projection, so this expression is a helper that facilitates transpilation to other
4976# dialects. For example, we'd generate UNNEST(GENERATE_SERIES(...)) in DuckDB
4977class ExplodingGenerateSeries(GenerateSeries):
4978    pass
4979
4980
4981class ArrayAgg(AggFunc):
4982    arg_types = {"this": True, "nulls_excluded": False}
4983
4984
4985class ArrayUniqueAgg(AggFunc):
4986    pass
4987
4988
4989class ArrayAll(Func):
4990    arg_types = {"this": True, "expression": True}
4991
4992
4993# Represents Python's `any(f(x) for x in array)`, where `array` is `this` and `f` is `expression`
4994class ArrayAny(Func):
4995    arg_types = {"this": True, "expression": True}
4996
4997
4998class ArrayConcat(Func):
4999    _sql_names = ["ARRAY_CONCAT", "ARRAY_CAT"]
5000    arg_types = {"this": True, "expressions": False}
5001    is_var_len_args = True
5002
5003
5004class ArrayConstructCompact(Func):
5005    arg_types = {"expressions": True}
5006    is_var_len_args = True
5007
5008
5009class ArrayContains(Binary, Func):
5010    _sql_names = ["ARRAY_CONTAINS", "ARRAY_HAS"]
5011
5012
5013class ArrayContainsAll(Binary, Func):
5014    _sql_names = ["ARRAY_CONTAINS_ALL", "ARRAY_HAS_ALL"]
5015
5016
5017class ArrayFilter(Func):
5018    arg_types = {"this": True, "expression": True}
5019    _sql_names = ["FILTER", "ARRAY_FILTER"]
5020
5021
5022class ArrayToString(Func):
5023    arg_types = {"this": True, "expression": True, "null": False}
5024    _sql_names = ["ARRAY_TO_STRING", "ARRAY_JOIN"]
5025
5026
5027class StringToArray(Func):
5028    arg_types = {"this": True, "expression": True, "null": False}
5029    _sql_names = ["STRING_TO_ARRAY", "SPLIT_BY_STRING"]
5030
5031
5032class ArrayOverlaps(Binary, Func):
5033    pass
5034
5035
5036class ArraySize(Func):
5037    arg_types = {"this": True, "expression": False}
5038    _sql_names = ["ARRAY_SIZE", "ARRAY_LENGTH"]
5039
5040
5041class ArraySort(Func):
5042    arg_types = {"this": True, "expression": False}
5043
5044
5045class ArraySum(Func):
5046    arg_types = {"this": True, "expression": False}
5047
5048
5049class ArrayUnionAgg(AggFunc):
5050    pass
5051
5052
5053class Avg(AggFunc):
5054    pass
5055
5056
5057class AnyValue(AggFunc):
5058    pass
5059
5060
5061class Lag(AggFunc):
5062    arg_types = {"this": True, "offset": False, "default": False}
5063
5064
5065class Lead(AggFunc):
5066    arg_types = {"this": True, "offset": False, "default": False}
5067
5068
5069# some dialects have a distinction between first and first_value, usually first is an aggregate func
5070# and first_value is a window func
5071class First(AggFunc):
5072    pass
5073
5074
5075class Last(AggFunc):
5076    pass
5077
5078
5079class FirstValue(AggFunc):
5080    pass
5081
5082
5083class LastValue(AggFunc):
5084    pass
5085
5086
5087class NthValue(AggFunc):
5088    arg_types = {"this": True, "offset": True}
5089
5090
5091class Case(Func):
5092    arg_types = {"this": False, "ifs": True, "default": False}
5093
5094    def when(self, condition: ExpOrStr, then: ExpOrStr, copy: bool = True, **opts) -> Case:
5095        instance = maybe_copy(self, copy)
5096        instance.append(
5097            "ifs",
5098            If(
5099                this=maybe_parse(condition, copy=copy, **opts),
5100                true=maybe_parse(then, copy=copy, **opts),
5101            ),
5102        )
5103        return instance
5104
5105    def else_(self, condition: ExpOrStr, copy: bool = True, **opts) -> Case:
5106        instance = maybe_copy(self, copy)
5107        instance.set("default", maybe_parse(condition, copy=copy, **opts))
5108        return instance
5109
5110
5111class Cast(Func):
5112    arg_types = {
5113        "this": True,
5114        "to": True,
5115        "format": False,
5116        "safe": False,
5117        "action": False,
5118    }
5119
5120    @property
5121    def name(self) -> str:
5122        return self.this.name
5123
5124    @property
5125    def to(self) -> DataType:
5126        return self.args["to"]
5127
5128    @property
5129    def output_name(self) -> str:
5130        return self.name
5131
5132    def is_type(self, *dtypes: DATA_TYPE) -> bool:
5133        """
5134        Checks whether this Cast's DataType matches one of the provided data types. Nested types
5135        like arrays or structs will be compared using "structural equivalence" semantics, so e.g.
5136        array<int> != array<float>.
5137
5138        Args:
5139            dtypes: the data types to compare this Cast's DataType to.
5140
5141        Returns:
5142            True, if and only if there is a type in `dtypes` which is equal to this Cast's DataType.
5143        """
5144        return self.to.is_type(*dtypes)
5145
5146
5147class TryCast(Cast):
5148    pass
5149
5150
5151class Try(Func):
5152    pass
5153
5154
5155class CastToStrType(Func):
5156    arg_types = {"this": True, "to": True}
5157
5158
5159class Collate(Binary, Func):
5160    pass
5161
5162
5163class Ceil(Func):
5164    arg_types = {"this": True, "decimals": False}
5165    _sql_names = ["CEIL", "CEILING"]
5166
5167
5168class Coalesce(Func):
5169    arg_types = {"this": True, "expressions": False, "is_nvl": False}
5170    is_var_len_args = True
5171    _sql_names = ["COALESCE", "IFNULL", "NVL"]
5172
5173
5174class Chr(Func):
5175    arg_types = {"this": True, "charset": False, "expressions": False}
5176    is_var_len_args = True
5177    _sql_names = ["CHR", "CHAR"]
5178
5179
5180class Concat(Func):
5181    arg_types = {"expressions": True, "safe": False, "coalesce": False}
5182    is_var_len_args = True
5183
5184
5185class ConcatWs(Concat):
5186    _sql_names = ["CONCAT_WS"]
5187
5188
5189# https://docs.oracle.com/cd/B13789_01/server.101/b10759/operators004.htm#i1035022
5190class ConnectByRoot(Func):
5191    pass
5192
5193
5194class Count(AggFunc):
5195    arg_types = {"this": False, "expressions": False, "big_int": False}
5196    is_var_len_args = True
5197
5198
5199class CountIf(AggFunc):
5200    _sql_names = ["COUNT_IF", "COUNTIF"]
5201
5202
5203# cube root
5204class Cbrt(Func):
5205    pass
5206
5207
5208class CurrentDate(Func):
5209    arg_types = {"this": False}
5210
5211
5212class CurrentDatetime(Func):
5213    arg_types = {"this": False}
5214
5215
5216class CurrentTime(Func):
5217    arg_types = {"this": False}
5218
5219
5220class CurrentTimestamp(Func):
5221    arg_types = {"this": False, "sysdate": False}
5222
5223
5224class CurrentUser(Func):
5225    arg_types = {"this": False}
5226
5227
5228class DateAdd(Func, IntervalOp):
5229    arg_types = {"this": True, "expression": True, "unit": False}
5230
5231
5232class DateSub(Func, IntervalOp):
5233    arg_types = {"this": True, "expression": True, "unit": False}
5234
5235
5236class DateDiff(Func, TimeUnit):
5237    _sql_names = ["DATEDIFF", "DATE_DIFF"]
5238    arg_types = {"this": True, "expression": True, "unit": False}
5239
5240
5241class DateTrunc(Func):
5242    arg_types = {"unit": True, "this": True, "zone": False}
5243
5244    def __init__(self, **args):
5245        unit = args.get("unit")
5246        if isinstance(unit, TimeUnit.VAR_LIKE):
5247            args["unit"] = Literal.string(
5248                (TimeUnit.UNABBREVIATED_UNIT_NAME.get(unit.name) or unit.name).upper()
5249            )
5250        elif isinstance(unit, Week):
5251            unit.set("this", Literal.string(unit.this.name.upper()))
5252
5253        super().__init__(**args)
5254
5255    @property
5256    def unit(self) -> Expression:
5257        return self.args["unit"]
5258
5259
5260# https://cloud.google.com/bigquery/docs/reference/standard-sql/datetime_functions#datetime
5261# expression can either be time_expr or time_zone
5262class Datetime(Func):
5263    arg_types = {"this": True, "expression": False}
5264
5265
5266class DatetimeAdd(Func, IntervalOp):
5267    arg_types = {"this": True, "expression": True, "unit": False}
5268
5269
5270class DatetimeSub(Func, IntervalOp):
5271    arg_types = {"this": True, "expression": True, "unit": False}
5272
5273
5274class DatetimeDiff(Func, TimeUnit):
5275    arg_types = {"this": True, "expression": True, "unit": False}
5276
5277
5278class DatetimeTrunc(Func, TimeUnit):
5279    arg_types = {"this": True, "unit": True, "zone": False}
5280
5281
5282class DayOfWeek(Func):
5283    _sql_names = ["DAY_OF_WEEK", "DAYOFWEEK"]
5284
5285
5286# https://duckdb.org/docs/sql/functions/datepart.html#part-specifiers-only-usable-as-date-part-specifiers
5287# ISO day of week function in duckdb is ISODOW
5288class DayOfWeekIso(Func):
5289    _sql_names = ["DAYOFWEEK_ISO", "ISODOW"]
5290
5291
5292class DayOfMonth(Func):
5293    _sql_names = ["DAY_OF_MONTH", "DAYOFMONTH"]
5294
5295
5296class DayOfYear(Func):
5297    _sql_names = ["DAY_OF_YEAR", "DAYOFYEAR"]
5298
5299
5300class ToDays(Func):
5301    pass
5302
5303
5304class WeekOfYear(Func):
5305    _sql_names = ["WEEK_OF_YEAR", "WEEKOFYEAR"]
5306
5307
5308class MonthsBetween(Func):
5309    arg_types = {"this": True, "expression": True, "roundoff": False}
5310
5311
5312class LastDay(Func, TimeUnit):
5313    _sql_names = ["LAST_DAY", "LAST_DAY_OF_MONTH"]
5314    arg_types = {"this": True, "unit": False}
5315
5316
5317class Extract(Func):
5318    arg_types = {"this": True, "expression": True}
5319
5320
5321class Timestamp(Func):
5322    arg_types = {"this": False, "zone": False, "with_tz": False}
5323
5324
5325class TimestampAdd(Func, TimeUnit):
5326    arg_types = {"this": True, "expression": True, "unit": False}
5327
5328
5329class TimestampSub(Func, TimeUnit):
5330    arg_types = {"this": True, "expression": True, "unit": False}
5331
5332
5333class TimestampDiff(Func, TimeUnit):
5334    _sql_names = ["TIMESTAMPDIFF", "TIMESTAMP_DIFF"]
5335    arg_types = {"this": True, "expression": True, "unit": False}
5336
5337
5338class TimestampTrunc(Func, TimeUnit):
5339    arg_types = {"this": True, "unit": True, "zone": False}
5340
5341
5342class TimeAdd(Func, TimeUnit):
5343    arg_types = {"this": True, "expression": True, "unit": False}
5344
5345
5346class TimeSub(Func, TimeUnit):
5347    arg_types = {"this": True, "expression": True, "unit": False}
5348
5349
5350class TimeDiff(Func, TimeUnit):
5351    arg_types = {"this": True, "expression": True, "unit": False}
5352
5353
5354class TimeTrunc(Func, TimeUnit):
5355    arg_types = {"this": True, "unit": True, "zone": False}
5356
5357
5358class DateFromParts(Func):
5359    _sql_names = ["DATE_FROM_PARTS", "DATEFROMPARTS"]
5360    arg_types = {"year": True, "month": True, "day": True}
5361
5362
5363class TimeFromParts(Func):
5364    _sql_names = ["TIME_FROM_PARTS", "TIMEFROMPARTS"]
5365    arg_types = {
5366        "hour": True,
5367        "min": True,
5368        "sec": True,
5369        "nano": False,
5370        "fractions": False,
5371        "precision": False,
5372    }
5373
5374
5375class DateStrToDate(Func):
5376    pass
5377
5378
5379class DateToDateStr(Func):
5380    pass
5381
5382
5383class DateToDi(Func):
5384    pass
5385
5386
5387# https://cloud.google.com/bigquery/docs/reference/standard-sql/date_functions#date
5388class Date(Func):
5389    arg_types = {"this": False, "zone": False, "expressions": False}
5390    is_var_len_args = True
5391
5392
5393class Day(Func):
5394    pass
5395
5396
5397class Decode(Func):
5398    arg_types = {"this": True, "charset": True, "replace": False}
5399
5400
5401class DiToDate(Func):
5402    pass
5403
5404
5405class Encode(Func):
5406    arg_types = {"this": True, "charset": True}
5407
5408
5409class Exp(Func):
5410    pass
5411
5412
5413# https://docs.snowflake.com/en/sql-reference/functions/flatten
5414class Explode(Func):
5415    arg_types = {"this": True, "expressions": False}
5416    is_var_len_args = True
5417
5418
5419# https://spark.apache.org/docs/latest/api/sql/#inline
5420class Inline(Func):
5421    pass
5422
5423
5424class ExplodeOuter(Explode):
5425    pass
5426
5427
5428class Posexplode(Explode):
5429    pass
5430
5431
5432class PosexplodeOuter(Posexplode, ExplodeOuter):
5433    pass
5434
5435
5436class Unnest(Func, UDTF):
5437    arg_types = {
5438        "expressions": True,
5439        "alias": False,
5440        "offset": False,
5441        "explode_array": False,
5442    }
5443
5444    @property
5445    def selects(self) -> t.List[Expression]:
5446        columns = super().selects
5447        offset = self.args.get("offset")
5448        if offset:
5449            columns = columns + [to_identifier("offset") if offset is True else offset]
5450        return columns
5451
5452
5453class Floor(Func):
5454    arg_types = {"this": True, "decimals": False}
5455
5456
5457class FromBase64(Func):
5458    pass
5459
5460
5461class ToBase64(Func):
5462    pass
5463
5464
5465# https://trino.io/docs/current/functions/datetime.html#from_iso8601_timestamp
5466class FromISO8601Timestamp(Func):
5467    _sql_names = ["FROM_ISO8601_TIMESTAMP"]
5468
5469
5470class GapFill(Func):
5471    arg_types = {
5472        "this": True,
5473        "ts_column": True,
5474        "bucket_width": True,
5475        "partitioning_columns": False,
5476        "value_columns": False,
5477        "origin": False,
5478        "ignore_nulls": False,
5479    }
5480
5481
5482# https://cloud.google.com/bigquery/docs/reference/standard-sql/array_functions#generate_date_array
5483class GenerateDateArray(Func):
5484    arg_types = {"start": True, "end": True, "step": False}
5485
5486
5487# https://cloud.google.com/bigquery/docs/reference/standard-sql/array_functions#generate_timestamp_array
5488class GenerateTimestampArray(Func):
5489    arg_types = {"start": True, "end": True, "step": True}
5490
5491
5492class Greatest(Func):
5493    arg_types = {"this": True, "expressions": False}
5494    is_var_len_args = True
5495
5496
5497class GroupConcat(AggFunc):
5498    arg_types = {"this": True, "separator": False}
5499
5500
5501class Hex(Func):
5502    pass
5503
5504
5505class LowerHex(Hex):
5506    pass
5507
5508
5509class Xor(Connector, Func):
5510    arg_types = {"this": False, "expression": False, "expressions": False}
5511
5512
5513class If(Func):
5514    arg_types = {"this": True, "true": True, "false": False}
5515    _sql_names = ["IF", "IIF"]
5516
5517
5518class Nullif(Func):
5519    arg_types = {"this": True, "expression": True}
5520
5521
5522class Initcap(Func):
5523    arg_types = {"this": True, "expression": False}
5524
5525
5526class IsNan(Func):
5527    _sql_names = ["IS_NAN", "ISNAN"]
5528
5529
5530class IsInf(Func):
5531    _sql_names = ["IS_INF", "ISINF"]
5532
5533
5534# https://www.postgresql.org/docs/current/functions-json.html
5535class JSON(Expression):
5536    arg_types = {"this": False, "with": False, "unique": False}
5537
5538
5539class JSONPath(Expression):
5540    arg_types = {"expressions": True}
5541
5542    @property
5543    def output_name(self) -> str:
5544        last_segment = self.expressions[-1].this
5545        return last_segment if isinstance(last_segment, str) else ""
5546
5547
5548class JSONPathPart(Expression):
5549    arg_types = {}
5550
5551
5552class JSONPathFilter(JSONPathPart):
5553    arg_types = {"this": True}
5554
5555
5556class JSONPathKey(JSONPathPart):
5557    arg_types = {"this": True}
5558
5559
5560class JSONPathRecursive(JSONPathPart):
5561    arg_types = {"this": False}
5562
5563
5564class JSONPathRoot(JSONPathPart):
5565    pass
5566
5567
5568class JSONPathScript(JSONPathPart):
5569    arg_types = {"this": True}
5570
5571
5572class JSONPathSlice(JSONPathPart):
5573    arg_types = {"start": False, "end": False, "step": False}
5574
5575
5576class JSONPathSelector(JSONPathPart):
5577    arg_types = {"this": True}
5578
5579
5580class JSONPathSubscript(JSONPathPart):
5581    arg_types = {"this": True}
5582
5583
5584class JSONPathUnion(JSONPathPart):
5585    arg_types = {"expressions": True}
5586
5587
5588class JSONPathWildcard(JSONPathPart):
5589    pass
5590
5591
5592class FormatJson(Expression):
5593    pass
5594
5595
5596class JSONKeyValue(Expression):
5597    arg_types = {"this": True, "expression": True}
5598
5599
5600class JSONObject(Func):
5601    arg_types = {
5602        "expressions": False,
5603        "null_handling": False,
5604        "unique_keys": False,
5605        "return_type": False,
5606        "encoding": False,
5607    }
5608
5609
5610class JSONObjectAgg(AggFunc):
5611    arg_types = {
5612        "expressions": False,
5613        "null_handling": False,
5614        "unique_keys": False,
5615        "return_type": False,
5616        "encoding": False,
5617    }
5618
5619
5620# https://docs.oracle.com/en/database/oracle/oracle-database/19/sqlrf/JSON_ARRAY.html
5621class JSONArray(Func):
5622    arg_types = {
5623        "expressions": True,
5624        "null_handling": False,
5625        "return_type": False,
5626        "strict": False,
5627    }
5628
5629
5630# https://docs.oracle.com/en/database/oracle/oracle-database/19/sqlrf/JSON_ARRAYAGG.html
5631class JSONArrayAgg(Func):
5632    arg_types = {
5633        "this": True,
5634        "order": False,
5635        "null_handling": False,
5636        "return_type": False,
5637        "strict": False,
5638    }
5639
5640
5641class JSONExists(Func):
5642    arg_types = {"this": True, "path": True, "passing": False, "on_condition": False}
5643
5644
5645# https://docs.oracle.com/en/database/oracle/oracle-database/19/sqlrf/JSON_TABLE.html
5646# Note: parsing of JSON column definitions is currently incomplete.
5647class JSONColumnDef(Expression):
5648    arg_types = {"this": False, "kind": False, "path": False, "nested_schema": False}
5649
5650
5651class JSONSchema(Expression):
5652    arg_types = {"expressions": True}
5653
5654
5655# https://dev.mysql.com/doc/refman/8.4/en/json-search-functions.html#function_json-value
5656class JSONValue(Expression):
5657    arg_types = {
5658        "this": True,
5659        "path": True,
5660        "returning": False,
5661        "on_condition": False,
5662    }
5663
5664
5665# # https://docs.oracle.com/en/database/oracle/oracle-database/19/sqlrf/JSON_TABLE.html
5666class JSONTable(Func):
5667    arg_types = {
5668        "this": True,
5669        "schema": True,
5670        "path": False,
5671        "error_handling": False,
5672        "empty_handling": False,
5673    }
5674
5675
5676# https://docs.snowflake.com/en/sql-reference/functions/object_insert
5677class ObjectInsert(Func):
5678    arg_types = {
5679        "this": True,
5680        "key": True,
5681        "value": True,
5682        "update_flag": False,
5683    }
5684
5685
5686class OpenJSONColumnDef(Expression):
5687    arg_types = {"this": True, "kind": True, "path": False, "as_json": False}
5688
5689
5690class OpenJSON(Func):
5691    arg_types = {"this": True, "path": False, "expressions": False}
5692
5693
5694class JSONBContains(Binary, Func):
5695    _sql_names = ["JSONB_CONTAINS"]
5696
5697
5698class JSONExtract(Binary, Func):
5699    arg_types = {
5700        "this": True,
5701        "expression": True,
5702        "only_json_types": False,
5703        "expressions": False,
5704        "variant_extract": False,
5705    }
5706    _sql_names = ["JSON_EXTRACT"]
5707    is_var_len_args = True
5708
5709    @property
5710    def output_name(self) -> str:
5711        return self.expression.output_name if not self.expressions else ""
5712
5713
5714class JSONExtractScalar(Binary, Func):
5715    arg_types = {"this": True, "expression": True, "only_json_types": False, "expressions": False}
5716    _sql_names = ["JSON_EXTRACT_SCALAR"]
5717    is_var_len_args = True
5718
5719    @property
5720    def output_name(self) -> str:
5721        return self.expression.output_name
5722
5723
5724class JSONBExtract(Binary, Func):
5725    _sql_names = ["JSONB_EXTRACT"]
5726
5727
5728class JSONBExtractScalar(Binary, Func):
5729    _sql_names = ["JSONB_EXTRACT_SCALAR"]
5730
5731
5732class JSONFormat(Func):
5733    arg_types = {"this": False, "options": False}
5734    _sql_names = ["JSON_FORMAT"]
5735
5736
5737# https://dev.mysql.com/doc/refman/8.0/en/json-search-functions.html#operator_member-of
5738class JSONArrayContains(Binary, Predicate, Func):
5739    _sql_names = ["JSON_ARRAY_CONTAINS"]
5740
5741
5742class ParseJSON(Func):
5743    # BigQuery, Snowflake have PARSE_JSON, Presto has JSON_PARSE
5744    # Snowflake also has TRY_PARSE_JSON, which is represented using `safe`
5745    _sql_names = ["PARSE_JSON", "JSON_PARSE"]
5746    arg_types = {"this": True, "expression": False, "safe": False}
5747
5748
5749class Least(Func):
5750    arg_types = {"this": True, "expressions": False}
5751    is_var_len_args = True
5752
5753
5754class Left(Func):
5755    arg_types = {"this": True, "expression": True}
5756
5757
5758class Right(Func):
5759    arg_types = {"this": True, "expression": True}
5760
5761
5762class Length(Func):
5763    arg_types = {"this": True, "binary": False}
5764    _sql_names = ["LENGTH", "LEN"]
5765
5766
5767class Levenshtein(Func):
5768    arg_types = {
5769        "this": True,
5770        "expression": False,
5771        "ins_cost": False,
5772        "del_cost": False,
5773        "sub_cost": False,
5774    }
5775
5776
5777class Ln(Func):
5778    pass
5779
5780
5781class Log(Func):
5782    arg_types = {"this": True, "expression": False}
5783
5784
5785class LogicalOr(AggFunc):
5786    _sql_names = ["LOGICAL_OR", "BOOL_OR", "BOOLOR_AGG"]
5787
5788
5789class LogicalAnd(AggFunc):
5790    _sql_names = ["LOGICAL_AND", "BOOL_AND", "BOOLAND_AGG"]
5791
5792
5793class Lower(Func):
5794    _sql_names = ["LOWER", "LCASE"]
5795
5796
5797class Map(Func):
5798    arg_types = {"keys": False, "values": False}
5799
5800    @property
5801    def keys(self) -> t.List[Expression]:
5802        keys = self.args.get("keys")
5803        return keys.expressions if keys else []
5804
5805    @property
5806    def values(self) -> t.List[Expression]:
5807        values = self.args.get("values")
5808        return values.expressions if values else []
5809
5810
5811# Represents the MAP {...} syntax in DuckDB - basically convert a struct to a MAP
5812class ToMap(Func):
5813    pass
5814
5815
5816class MapFromEntries(Func):
5817    pass
5818
5819
5820# https://learn.microsoft.com/en-us/sql/t-sql/language-elements/scope-resolution-operator-transact-sql?view=sql-server-ver16
5821class ScopeResolution(Expression):
5822    arg_types = {"this": False, "expression": True}
5823
5824
5825class Stream(Expression):
5826    pass
5827
5828
5829class StarMap(Func):
5830    pass
5831
5832
5833class VarMap(Func):
5834    arg_types = {"keys": True, "values": True}
5835    is_var_len_args = True
5836
5837    @property
5838    def keys(self) -> t.List[Expression]:
5839        return self.args["keys"].expressions
5840
5841    @property
5842    def values(self) -> t.List[Expression]:
5843        return self.args["values"].expressions
5844
5845
5846# https://dev.mysql.com/doc/refman/8.0/en/fulltext-search.html
5847class MatchAgainst(Func):
5848    arg_types = {"this": True, "expressions": True, "modifier": False}
5849
5850
5851class Max(AggFunc):
5852    arg_types = {"this": True, "expressions": False}
5853    is_var_len_args = True
5854
5855
5856class MD5(Func):
5857    _sql_names = ["MD5"]
5858
5859
5860# Represents the variant of the MD5 function that returns a binary value
5861class MD5Digest(Func):
5862    _sql_names = ["MD5_DIGEST"]
5863
5864
5865class Min(AggFunc):
5866    arg_types = {"this": True, "expressions": False}
5867    is_var_len_args = True
5868
5869
5870class Month(Func):
5871    pass
5872
5873
5874class AddMonths(Func):
5875    arg_types = {"this": True, "expression": True}
5876
5877
5878class Nvl2(Func):
5879    arg_types = {"this": True, "true": True, "false": False}
5880
5881
5882class Normalize(Func):
5883    arg_types = {"this": True, "form": False}
5884
5885
5886# https://cloud.google.com/bigquery/docs/reference/standard-sql/bigqueryml-syntax-predict#mlpredict_function
5887class Predict(Func):
5888    arg_types = {"this": True, "expression": True, "params_struct": False}
5889
5890
5891class Pow(Binary, Func):
5892    _sql_names = ["POWER", "POW"]
5893
5894
5895class PercentileCont(AggFunc):
5896    arg_types = {"this": True, "expression": False}
5897
5898
5899class PercentileDisc(AggFunc):
5900    arg_types = {"this": True, "expression": False}
5901
5902
5903class Quantile(AggFunc):
5904    arg_types = {"this": True, "quantile": True}
5905
5906
5907class ApproxQuantile(Quantile):
5908    arg_types = {"this": True, "quantile": True, "accuracy": False, "weight": False}
5909
5910
5911class Quarter(Func):
5912    pass
5913
5914
5915# https://docs.teradata.com/r/Enterprise_IntelliFlex_VMware/SQL-Functions-Expressions-and-Predicates/Arithmetic-Trigonometric-Hyperbolic-Operators/Functions/RANDOM/RANDOM-Function-Syntax
5916# teradata lower and upper bounds
5917class Rand(Func):
5918    _sql_names = ["RAND", "RANDOM"]
5919    arg_types = {"this": False, "lower": False, "upper": False}
5920
5921
5922class Randn(Func):
5923    arg_types = {"this": False}
5924
5925
5926class RangeN(Func):
5927    arg_types = {"this": True, "expressions": True, "each": False}
5928
5929
5930class ReadCSV(Func):
5931    _sql_names = ["READ_CSV"]
5932    is_var_len_args = True
5933    arg_types = {"this": True, "expressions": False}
5934
5935
5936class Reduce(Func):
5937    arg_types = {"this": True, "initial": True, "merge": True, "finish": False}
5938
5939
5940class RegexpExtract(Func):
5941    arg_types = {
5942        "this": True,
5943        "expression": True,
5944        "position": False,
5945        "occurrence": False,
5946        "parameters": False,
5947        "group": False,
5948    }
5949
5950
5951class RegexpReplace(Func):
5952    arg_types = {
5953        "this": True,
5954        "expression": True,
5955        "replacement": False,
5956        "position": False,
5957        "occurrence": False,
5958        "modifiers": False,
5959    }
5960
5961
5962class RegexpLike(Binary, Func):
5963    arg_types = {"this": True, "expression": True, "flag": False}
5964
5965
5966class RegexpILike(Binary, Func):
5967    arg_types = {"this": True, "expression": True, "flag": False}
5968
5969
5970# https://spark.apache.org/docs/latest/api/python/reference/pyspark.sql/api/pyspark.sql.functions.split.html
5971# limit is the number of times a pattern is applied
5972class RegexpSplit(Func):
5973    arg_types = {"this": True, "expression": True, "limit": False}
5974
5975
5976class Repeat(Func):
5977    arg_types = {"this": True, "times": True}
5978
5979
5980# https://learn.microsoft.com/en-us/sql/t-sql/functions/round-transact-sql?view=sql-server-ver16
5981# tsql third argument function == trunctaion if not 0
5982class Round(Func):
5983    arg_types = {"this": True, "decimals": False, "truncate": False}
5984
5985
5986class RowNumber(Func):
5987    arg_types: t.Dict[str, t.Any] = {}
5988
5989
5990class SafeDivide(Func):
5991    arg_types = {"this": True, "expression": True}
5992
5993
5994class SHA(Func):
5995    _sql_names = ["SHA", "SHA1"]
5996
5997
5998class SHA2(Func):
5999    _sql_names = ["SHA2"]
6000    arg_types = {"this": True, "length": False}
6001
6002
6003class Sign(Func):
6004    _sql_names = ["SIGN", "SIGNUM"]
6005
6006
6007class SortArray(Func):
6008    arg_types = {"this": True, "asc": False}
6009
6010
6011class Split(Func):
6012    arg_types = {"this": True, "expression": True, "limit": False}
6013
6014
6015# Start may be omitted in the case of postgres
6016# https://www.postgresql.org/docs/9.1/functions-string.html @ Table 9-6
6017class Substring(Func):
6018    arg_types = {"this": True, "start": False, "length": False}
6019
6020
6021class StandardHash(Func):
6022    arg_types = {"this": True, "expression": False}
6023
6024
6025class StartsWith(Func):
6026    _sql_names = ["STARTS_WITH", "STARTSWITH"]
6027    arg_types = {"this": True, "expression": True}
6028
6029
6030class StrPosition(Func):
6031    arg_types = {
6032        "this": True,
6033        "substr": True,
6034        "position": False,
6035        "instance": False,
6036    }
6037
6038
6039class StrToDate(Func):
6040    arg_types = {"this": True, "format": False, "safe": False}
6041
6042
6043class StrToTime(Func):
6044    arg_types = {"this": True, "format": True, "zone": False, "safe": False}
6045
6046
6047# Spark allows unix_timestamp()
6048# https://spark.apache.org/docs/3.1.3/api/python/reference/api/pyspark.sql.functions.unix_timestamp.html
6049class StrToUnix(Func):
6050    arg_types = {"this": False, "format": False}
6051
6052
6053# https://prestodb.io/docs/current/functions/string.html
6054# https://spark.apache.org/docs/latest/api/sql/index.html#str_to_map
6055class StrToMap(Func):
6056    arg_types = {
6057        "this": True,
6058        "pair_delim": False,
6059        "key_value_delim": False,
6060        "duplicate_resolution_callback": False,
6061    }
6062
6063
6064class NumberToStr(Func):
6065    arg_types = {"this": True, "format": True, "culture": False}
6066
6067
6068class FromBase(Func):
6069    arg_types = {"this": True, "expression": True}
6070
6071
6072class Struct(Func):
6073    arg_types = {"expressions": False}
6074    is_var_len_args = True
6075
6076
6077class StructExtract(Func):
6078    arg_types = {"this": True, "expression": True}
6079
6080
6081# https://learn.microsoft.com/en-us/sql/t-sql/functions/stuff-transact-sql?view=sql-server-ver16
6082# https://docs.snowflake.com/en/sql-reference/functions/insert
6083class Stuff(Func):
6084    _sql_names = ["STUFF", "INSERT"]
6085    arg_types = {"this": True, "start": True, "length": True, "expression": True}
6086
6087
6088class Sum(AggFunc):
6089    pass
6090
6091
6092class Sqrt(Func):
6093    pass
6094
6095
6096class Stddev(AggFunc):
6097    _sql_names = ["STDDEV", "STDEV"]
6098
6099
6100class StddevPop(AggFunc):
6101    pass
6102
6103
6104class StddevSamp(AggFunc):
6105    pass
6106
6107
6108# https://cloud.google.com/bigquery/docs/reference/standard-sql/time_functions#time
6109class Time(Func):
6110    arg_types = {"this": False, "zone": False}
6111
6112
6113class TimeToStr(Func):
6114    arg_types = {"this": True, "format": True, "culture": False, "zone": False}
6115
6116
6117class TimeToTimeStr(Func):
6118    pass
6119
6120
6121class TimeToUnix(Func):
6122    pass
6123
6124
6125class TimeStrToDate(Func):
6126    pass
6127
6128
6129class TimeStrToTime(Func):
6130    arg_types = {"this": True, "zone": False}
6131
6132
6133class TimeStrToUnix(Func):
6134    pass
6135
6136
6137class Trim(Func):
6138    arg_types = {
6139        "this": True,
6140        "expression": False,
6141        "position": False,
6142        "collation": False,
6143    }
6144
6145
6146class TsOrDsAdd(Func, TimeUnit):
6147    # return_type is used to correctly cast the arguments of this expression when transpiling it
6148    arg_types = {"this": True, "expression": True, "unit": False, "return_type": False}
6149
6150    @property
6151    def return_type(self) -> DataType:
6152        return DataType.build(self.args.get("return_type") or DataType.Type.DATE)
6153
6154
6155class TsOrDsDiff(Func, TimeUnit):
6156    arg_types = {"this": True, "expression": True, "unit": False}
6157
6158
6159class TsOrDsToDateStr(Func):
6160    pass
6161
6162
6163class TsOrDsToDate(Func):
6164    arg_types = {"this": True, "format": False, "safe": False}
6165
6166
6167class TsOrDsToTime(Func):
6168    pass
6169
6170
6171class TsOrDsToTimestamp(Func):
6172    pass
6173
6174
6175class TsOrDiToDi(Func):
6176    pass
6177
6178
6179class Unhex(Func):
6180    pass
6181
6182
6183# https://cloud.google.com/bigquery/docs/reference/standard-sql/date_functions#unix_date
6184class UnixDate(Func):
6185    pass
6186
6187
6188class UnixToStr(Func):
6189    arg_types = {"this": True, "format": False}
6190
6191
6192# https://prestodb.io/docs/current/functions/datetime.html
6193# presto has weird zone/hours/minutes
6194class UnixToTime(Func):
6195    arg_types = {
6196        "this": True,
6197        "scale": False,
6198        "zone": False,
6199        "hours": False,
6200        "minutes": False,
6201        "format": False,
6202    }
6203
6204    SECONDS = Literal.number(0)
6205    DECIS = Literal.number(1)
6206    CENTIS = Literal.number(2)
6207    MILLIS = Literal.number(3)
6208    DECIMILLIS = Literal.number(4)
6209    CENTIMILLIS = Literal.number(5)
6210    MICROS = Literal.number(6)
6211    DECIMICROS = Literal.number(7)
6212    CENTIMICROS = Literal.number(8)
6213    NANOS = Literal.number(9)
6214
6215
6216class UnixToTimeStr(Func):
6217    pass
6218
6219
6220class TimestampFromParts(Func):
6221    _sql_names = ["TIMESTAMP_FROM_PARTS", "TIMESTAMPFROMPARTS"]
6222    arg_types = {
6223        "year": True,
6224        "month": True,
6225        "day": True,
6226        "hour": True,
6227        "min": True,
6228        "sec": True,
6229        "nano": False,
6230        "zone": False,
6231        "milli": False,
6232    }
6233
6234
6235class Upper(Func):
6236    _sql_names = ["UPPER", "UCASE"]
6237
6238
6239class Corr(Binary, AggFunc):
6240    pass
6241
6242
6243class Variance(AggFunc):
6244    _sql_names = ["VARIANCE", "VARIANCE_SAMP", "VAR_SAMP"]
6245
6246
6247class VariancePop(AggFunc):
6248    _sql_names = ["VARIANCE_POP", "VAR_POP"]
6249
6250
6251class CovarSamp(Binary, AggFunc):
6252    pass
6253
6254
6255class CovarPop(Binary, AggFunc):
6256    pass
6257
6258
6259class Week(Func):
6260    arg_types = {"this": True, "mode": False}
6261
6262
6263class XMLTable(Func):
6264    arg_types = {"this": True, "passing": False, "columns": False, "by_ref": False}
6265
6266
6267class Year(Func):
6268    pass
6269
6270
6271class Use(Expression):
6272    arg_types = {"this": True, "kind": False}
6273
6274
6275class Merge(Expression):
6276    arg_types = {
6277        "this": True,
6278        "using": True,
6279        "on": True,
6280        "expressions": True,
6281        "with": False,
6282    }
6283
6284
6285class When(Func):
6286    arg_types = {"matched": True, "source": False, "condition": False, "then": True}
6287
6288
6289# https://docs.oracle.com/javadb/10.8.3.0/ref/rrefsqljnextvaluefor.html
6290# https://learn.microsoft.com/en-us/sql/t-sql/functions/next-value-for-transact-sql?view=sql-server-ver16
6291class NextValueFor(Func):
6292    arg_types = {"this": True, "order": False}
6293
6294
6295# Refers to a trailing semi-colon. This is only used to preserve trailing comments
6296# select 1; -- my comment
6297class Semicolon(Expression):
6298    arg_types = {}
6299
6300
6301def _norm_arg(arg):
6302    return arg.lower() if type(arg) is str else arg
6303
6304
6305ALL_FUNCTIONS = subclasses(__name__, Func, (AggFunc, Anonymous, Func))
6306FUNCTION_BY_NAME = {name: func for func in ALL_FUNCTIONS for name in func.sql_names()}
6307
6308JSON_PATH_PARTS = subclasses(__name__, JSONPathPart, (JSONPathPart,))
6309
6310PERCENTILES = (PercentileCont, PercentileDisc)
6311
6312
6313# Helpers
6314@t.overload
6315def maybe_parse(
6316    sql_or_expression: ExpOrStr,
6317    *,
6318    into: t.Type[E],
6319    dialect: DialectType = None,
6320    prefix: t.Optional[str] = None,
6321    copy: bool = False,
6322    **opts,
6323) -> E: ...
6324
6325
6326@t.overload
6327def maybe_parse(
6328    sql_or_expression: str | E,
6329    *,
6330    into: t.Optional[IntoType] = None,
6331    dialect: DialectType = None,
6332    prefix: t.Optional[str] = None,
6333    copy: bool = False,
6334    **opts,
6335) -> E: ...
6336
6337
6338def maybe_parse(
6339    sql_or_expression: ExpOrStr,
6340    *,
6341    into: t.Optional[IntoType] = None,
6342    dialect: DialectType = None,
6343    prefix: t.Optional[str] = None,
6344    copy: bool = False,
6345    **opts,
6346) -> Expression:
6347    """Gracefully handle a possible string or expression.
6348
6349    Example:
6350        >>> maybe_parse("1")
6351        Literal(this=1, is_string=False)
6352        >>> maybe_parse(to_identifier("x"))
6353        Identifier(this=x, quoted=False)
6354
6355    Args:
6356        sql_or_expression: the SQL code string or an expression
6357        into: the SQLGlot Expression to parse into
6358        dialect: the dialect used to parse the input expressions (in the case that an
6359            input expression is a SQL string).
6360        prefix: a string to prefix the sql with before it gets parsed
6361            (automatically includes a space)
6362        copy: whether to copy the expression.
6363        **opts: other options to use to parse the input expressions (again, in the case
6364            that an input expression is a SQL string).
6365
6366    Returns:
6367        Expression: the parsed or given expression.
6368    """
6369    if isinstance(sql_or_expression, Expression):
6370        if copy:
6371            return sql_or_expression.copy()
6372        return sql_or_expression
6373
6374    if sql_or_expression is None:
6375        raise ParseError("SQL cannot be None")
6376
6377    import sqlglot
6378
6379    sql = str(sql_or_expression)
6380    if prefix:
6381        sql = f"{prefix} {sql}"
6382
6383    return sqlglot.parse_one(sql, read=dialect, into=into, **opts)
6384
6385
6386@t.overload
6387def maybe_copy(instance: None, copy: bool = True) -> None: ...
6388
6389
6390@t.overload
6391def maybe_copy(instance: E, copy: bool = True) -> E: ...
6392
6393
6394def maybe_copy(instance, copy=True):
6395    return instance.copy() if copy and instance else instance
6396
6397
6398def _to_s(node: t.Any, verbose: bool = False, level: int = 0) -> str:
6399    """Generate a textual representation of an Expression tree"""
6400    indent = "\n" + ("  " * (level + 1))
6401    delim = f",{indent}"
6402
6403    if isinstance(node, Expression):
6404        args = {k: v for k, v in node.args.items() if (v is not None and v != []) or verbose}
6405
6406        if (node.type or verbose) and not isinstance(node, DataType):
6407            args["_type"] = node.type
6408        if node.comments or verbose:
6409            args["_comments"] = node.comments
6410
6411        if verbose:
6412            args["_id"] = id(node)
6413
6414        # Inline leaves for a more compact representation
6415        if node.is_leaf():
6416            indent = ""
6417            delim = ", "
6418
6419        items = delim.join([f"{k}={_to_s(v, verbose, level + 1)}" for k, v in args.items()])
6420        return f"{node.__class__.__name__}({indent}{items})"
6421
6422    if isinstance(node, list):
6423        items = delim.join(_to_s(i, verbose, level + 1) for i in node)
6424        items = f"{indent}{items}" if items else ""
6425        return f"[{items}]"
6426
6427    # Indent multiline strings to match the current level
6428    return indent.join(textwrap.dedent(str(node).strip("\n")).splitlines())
6429
6430
6431def _is_wrong_expression(expression, into):
6432    return isinstance(expression, Expression) and not isinstance(expression, into)
6433
6434
6435def _apply_builder(
6436    expression,
6437    instance,
6438    arg,
6439    copy=True,
6440    prefix=None,
6441    into=None,
6442    dialect=None,
6443    into_arg="this",
6444    **opts,
6445):
6446    if _is_wrong_expression(expression, into):
6447        expression = into(**{into_arg: expression})
6448    instance = maybe_copy(instance, copy)
6449    expression = maybe_parse(
6450        sql_or_expression=expression,
6451        prefix=prefix,
6452        into=into,
6453        dialect=dialect,
6454        **opts,
6455    )
6456    instance.set(arg, expression)
6457    return instance
6458
6459
6460def _apply_child_list_builder(
6461    *expressions,
6462    instance,
6463    arg,
6464    append=True,
6465    copy=True,
6466    prefix=None,
6467    into=None,
6468    dialect=None,
6469    properties=None,
6470    **opts,
6471):
6472    instance = maybe_copy(instance, copy)
6473    parsed = []
6474    properties = {} if properties is None else properties
6475
6476    for expression in expressions:
6477        if expression is not None:
6478            if _is_wrong_expression(expression, into):
6479                expression = into(expressions=[expression])
6480
6481            expression = maybe_parse(
6482                expression,
6483                into=into,
6484                dialect=dialect,
6485                prefix=prefix,
6486                **opts,
6487            )
6488            for k, v in expression.args.items():
6489                if k == "expressions":
6490                    parsed.extend(v)
6491                else:
6492                    properties[k] = v
6493
6494    existing = instance.args.get(arg)
6495    if append and existing:
6496        parsed = existing.expressions + parsed
6497
6498    child = into(expressions=parsed)
6499    for k, v in properties.items():
6500        child.set(k, v)
6501    instance.set(arg, child)
6502
6503    return instance
6504
6505
6506def _apply_list_builder(
6507    *expressions,
6508    instance,
6509    arg,
6510    append=True,
6511    copy=True,
6512    prefix=None,
6513    into=None,
6514    dialect=None,
6515    **opts,
6516):
6517    inst = maybe_copy(instance, copy)
6518
6519    expressions = [
6520        maybe_parse(
6521            sql_or_expression=expression,
6522            into=into,
6523            prefix=prefix,
6524            dialect=dialect,
6525            **opts,
6526        )
6527        for expression in expressions
6528        if expression is not None
6529    ]
6530
6531    existing_expressions = inst.args.get(arg)
6532    if append and existing_expressions:
6533        expressions = existing_expressions + expressions
6534
6535    inst.set(arg, expressions)
6536    return inst
6537
6538
6539def _apply_conjunction_builder(
6540    *expressions,
6541    instance,
6542    arg,
6543    into=None,
6544    append=True,
6545    copy=True,
6546    dialect=None,
6547    **opts,
6548):
6549    expressions = [exp for exp in expressions if exp is not None and exp != ""]
6550    if not expressions:
6551        return instance
6552
6553    inst = maybe_copy(instance, copy)
6554
6555    existing = inst.args.get(arg)
6556    if append and existing is not None:
6557        expressions = [existing.this if into else existing] + list(expressions)
6558
6559    node = and_(*expressions, dialect=dialect, copy=copy, **opts)
6560
6561    inst.set(arg, into(this=node) if into else node)
6562    return inst
6563
6564
6565def _apply_cte_builder(
6566    instance: E,
6567    alias: ExpOrStr,
6568    as_: ExpOrStr,
6569    recursive: t.Optional[bool] = None,
6570    materialized: t.Optional[bool] = None,
6571    append: bool = True,
6572    dialect: DialectType = None,
6573    copy: bool = True,
6574    **opts,
6575) -> E:
6576    alias_expression = maybe_parse(alias, dialect=dialect, into=TableAlias, **opts)
6577    as_expression = maybe_parse(as_, dialect=dialect, **opts)
6578    cte = CTE(this=as_expression, alias=alias_expression, materialized=materialized)
6579    return _apply_child_list_builder(
6580        cte,
6581        instance=instance,
6582        arg="with",
6583        append=append,
6584        copy=copy,
6585        into=With,
6586        properties={"recursive": recursive or False},
6587    )
6588
6589
6590def _combine(
6591    expressions: t.Sequence[t.Optional[ExpOrStr]],
6592    operator: t.Type[Connector],
6593    dialect: DialectType = None,
6594    copy: bool = True,
6595    **opts,
6596) -> Expression:
6597    conditions = [
6598        condition(expression, dialect=dialect, copy=copy, **opts)
6599        for expression in expressions
6600        if expression is not None
6601    ]
6602
6603    this, *rest = conditions
6604    if rest:
6605        this = _wrap(this, Connector)
6606    for expression in rest:
6607        this = operator(this=this, expression=_wrap(expression, Connector))
6608
6609    return this
6610
6611
6612def _wrap(expression: E, kind: t.Type[Expression]) -> E | Paren:
6613    return Paren(this=expression) if isinstance(expression, kind) else expression
6614
6615
6616def union(
6617    left: ExpOrStr,
6618    right: ExpOrStr,
6619    distinct: bool = True,
6620    dialect: DialectType = None,
6621    copy: bool = True,
6622    **opts,
6623) -> Union:
6624    """
6625    Initializes a syntax tree from one UNION expression.
6626
6627    Example:
6628        >>> union("SELECT * FROM foo", "SELECT * FROM bla").sql()
6629        'SELECT * FROM foo UNION SELECT * FROM bla'
6630
6631    Args:
6632        left: the SQL code string corresponding to the left-hand side.
6633            If an `Expression` instance is passed, it will be used as-is.
6634        right: the SQL code string corresponding to the right-hand side.
6635            If an `Expression` instance is passed, it will be used as-is.
6636        distinct: set the DISTINCT flag if and only if this is true.
6637        dialect: the dialect used to parse the input expression.
6638        copy: whether to copy the expression.
6639        opts: other options to use to parse the input expressions.
6640
6641    Returns:
6642        The new Union instance.
6643    """
6644    left = maybe_parse(sql_or_expression=left, dialect=dialect, copy=copy, **opts)
6645    right = maybe_parse(sql_or_expression=right, dialect=dialect, copy=copy, **opts)
6646
6647    return Union(this=left, expression=right, distinct=distinct)
6648
6649
6650def intersect(
6651    left: ExpOrStr,
6652    right: ExpOrStr,
6653    distinct: bool = True,
6654    dialect: DialectType = None,
6655    copy: bool = True,
6656    **opts,
6657) -> Intersect:
6658    """
6659    Initializes a syntax tree from one INTERSECT expression.
6660
6661    Example:
6662        >>> intersect("SELECT * FROM foo", "SELECT * FROM bla").sql()
6663        'SELECT * FROM foo INTERSECT SELECT * FROM bla'
6664
6665    Args:
6666        left: the SQL code string corresponding to the left-hand side.
6667            If an `Expression` instance is passed, it will be used as-is.
6668        right: the SQL code string corresponding to the right-hand side.
6669            If an `Expression` instance is passed, it will be used as-is.
6670        distinct: set the DISTINCT flag if and only if this is true.
6671        dialect: the dialect used to parse the input expression.
6672        copy: whether to copy the expression.
6673        opts: other options to use to parse the input expressions.
6674
6675    Returns:
6676        The new Intersect instance.
6677    """
6678    left = maybe_parse(sql_or_expression=left, dialect=dialect, copy=copy, **opts)
6679    right = maybe_parse(sql_or_expression=right, dialect=dialect, copy=copy, **opts)
6680
6681    return Intersect(this=left, expression=right, distinct=distinct)
6682
6683
6684def except_(
6685    left: ExpOrStr,
6686    right: ExpOrStr,
6687    distinct: bool = True,
6688    dialect: DialectType = None,
6689    copy: bool = True,
6690    **opts,
6691) -> Except:
6692    """
6693    Initializes a syntax tree from one EXCEPT expression.
6694
6695    Example:
6696        >>> except_("SELECT * FROM foo", "SELECT * FROM bla").sql()
6697        'SELECT * FROM foo EXCEPT SELECT * FROM bla'
6698
6699    Args:
6700        left: the SQL code string corresponding to the left-hand side.
6701            If an `Expression` instance is passed, it will be used as-is.
6702        right: the SQL code string corresponding to the right-hand side.
6703            If an `Expression` instance is passed, it will be used as-is.
6704        distinct: set the DISTINCT flag if and only if this is true.
6705        dialect: the dialect used to parse the input expression.
6706        copy: whether to copy the expression.
6707        opts: other options to use to parse the input expressions.
6708
6709    Returns:
6710        The new Except instance.
6711    """
6712    left = maybe_parse(sql_or_expression=left, dialect=dialect, copy=copy, **opts)
6713    right = maybe_parse(sql_or_expression=right, dialect=dialect, copy=copy, **opts)
6714
6715    return Except(this=left, expression=right, distinct=distinct)
6716
6717
6718def select(*expressions: ExpOrStr, dialect: DialectType = None, **opts) -> Select:
6719    """
6720    Initializes a syntax tree from one or multiple SELECT expressions.
6721
6722    Example:
6723        >>> select("col1", "col2").from_("tbl").sql()
6724        'SELECT col1, col2 FROM tbl'
6725
6726    Args:
6727        *expressions: the SQL code string to parse as the expressions of a
6728            SELECT statement. If an Expression instance is passed, this is used as-is.
6729        dialect: the dialect used to parse the input expressions (in the case that an
6730            input expression is a SQL string).
6731        **opts: other options to use to parse the input expressions (again, in the case
6732            that an input expression is a SQL string).
6733
6734    Returns:
6735        Select: the syntax tree for the SELECT statement.
6736    """
6737    return Select().select(*expressions, dialect=dialect, **opts)
6738
6739
6740def from_(expression: ExpOrStr, dialect: DialectType = None, **opts) -> Select:
6741    """
6742    Initializes a syntax tree from a FROM expression.
6743
6744    Example:
6745        >>> from_("tbl").select("col1", "col2").sql()
6746        'SELECT col1, col2 FROM tbl'
6747
6748    Args:
6749        *expression: the SQL code string to parse as the FROM expressions of a
6750            SELECT statement. If an Expression instance is passed, this is used as-is.
6751        dialect: the dialect used to parse the input expression (in the case that the
6752            input expression is a SQL string).
6753        **opts: other options to use to parse the input expressions (again, in the case
6754            that the input expression is a SQL string).
6755
6756    Returns:
6757        Select: the syntax tree for the SELECT statement.
6758    """
6759    return Select().from_(expression, dialect=dialect, **opts)
6760
6761
6762def update(
6763    table: str | Table,
6764    properties: dict,
6765    where: t.Optional[ExpOrStr] = None,
6766    from_: t.Optional[ExpOrStr] = None,
6767    dialect: DialectType = None,
6768    **opts,
6769) -> Update:
6770    """
6771    Creates an update statement.
6772
6773    Example:
6774        >>> update("my_table", {"x": 1, "y": "2", "z": None}, from_="baz", where="id > 1").sql()
6775        "UPDATE my_table SET x = 1, y = '2', z = NULL FROM baz WHERE id > 1"
6776
6777    Args:
6778        *properties: dictionary of properties to set which are
6779            auto converted to sql objects eg None -> NULL
6780        where: sql conditional parsed into a WHERE statement
6781        from_: sql statement parsed into a FROM statement
6782        dialect: the dialect used to parse the input expressions.
6783        **opts: other options to use to parse the input expressions.
6784
6785    Returns:
6786        Update: the syntax tree for the UPDATE statement.
6787    """
6788    update_expr = Update(this=maybe_parse(table, into=Table, dialect=dialect))
6789    update_expr.set(
6790        "expressions",
6791        [
6792            EQ(this=maybe_parse(k, dialect=dialect, **opts), expression=convert(v))
6793            for k, v in properties.items()
6794        ],
6795    )
6796    if from_:
6797        update_expr.set(
6798            "from",
6799            maybe_parse(from_, into=From, dialect=dialect, prefix="FROM", **opts),
6800        )
6801    if isinstance(where, Condition):
6802        where = Where(this=where)
6803    if where:
6804        update_expr.set(
6805            "where",
6806            maybe_parse(where, into=Where, dialect=dialect, prefix="WHERE", **opts),
6807        )
6808    return update_expr
6809
6810
6811def delete(
6812    table: ExpOrStr,
6813    where: t.Optional[ExpOrStr] = None,
6814    returning: t.Optional[ExpOrStr] = None,
6815    dialect: DialectType = None,
6816    **opts,
6817) -> Delete:
6818    """
6819    Builds a delete statement.
6820
6821    Example:
6822        >>> delete("my_table", where="id > 1").sql()
6823        'DELETE FROM my_table WHERE id > 1'
6824
6825    Args:
6826        where: sql conditional parsed into a WHERE statement
6827        returning: sql conditional parsed into a RETURNING statement
6828        dialect: the dialect used to parse the input expressions.
6829        **opts: other options to use to parse the input expressions.
6830
6831    Returns:
6832        Delete: the syntax tree for the DELETE statement.
6833    """
6834    delete_expr = Delete().delete(table, dialect=dialect, copy=False, **opts)
6835    if where:
6836        delete_expr = delete_expr.where(where, dialect=dialect, copy=False, **opts)
6837    if returning:
6838        delete_expr = t.cast(
6839            Delete, delete_expr.returning(returning, dialect=dialect, copy=False, **opts)
6840        )
6841    return delete_expr
6842
6843
6844def insert(
6845    expression: ExpOrStr,
6846    into: ExpOrStr,
6847    columns: t.Optional[t.Sequence[str | Identifier]] = None,
6848    overwrite: t.Optional[bool] = None,
6849    returning: t.Optional[ExpOrStr] = None,
6850    dialect: DialectType = None,
6851    copy: bool = True,
6852    **opts,
6853) -> Insert:
6854    """
6855    Builds an INSERT statement.
6856
6857    Example:
6858        >>> insert("VALUES (1, 2, 3)", "tbl").sql()
6859        'INSERT INTO tbl VALUES (1, 2, 3)'
6860
6861    Args:
6862        expression: the sql string or expression of the INSERT statement
6863        into: the tbl to insert data to.
6864        columns: optionally the table's column names.
6865        overwrite: whether to INSERT OVERWRITE or not.
6866        returning: sql conditional parsed into a RETURNING statement
6867        dialect: the dialect used to parse the input expressions.
6868        copy: whether to copy the expression.
6869        **opts: other options to use to parse the input expressions.
6870
6871    Returns:
6872        Insert: the syntax tree for the INSERT statement.
6873    """
6874    expr = maybe_parse(expression, dialect=dialect, copy=copy, **opts)
6875    this: Table | Schema = maybe_parse(into, into=Table, dialect=dialect, copy=copy, **opts)
6876
6877    if columns:
6878        this = Schema(this=this, expressions=[to_identifier(c, copy=copy) for c in columns])
6879
6880    insert = Insert(this=this, expression=expr, overwrite=overwrite)
6881
6882    if returning:
6883        insert = t.cast(Insert, insert.returning(returning, dialect=dialect, copy=False, **opts))
6884
6885    return insert
6886
6887
6888def condition(
6889    expression: ExpOrStr, dialect: DialectType = None, copy: bool = True, **opts
6890) -> Condition:
6891    """
6892    Initialize a logical condition expression.
6893
6894    Example:
6895        >>> condition("x=1").sql()
6896        'x = 1'
6897
6898        This is helpful for composing larger logical syntax trees:
6899        >>> where = condition("x=1")
6900        >>> where = where.and_("y=1")
6901        >>> Select().from_("tbl").select("*").where(where).sql()
6902        'SELECT * FROM tbl WHERE x = 1 AND y = 1'
6903
6904    Args:
6905        *expression: the SQL code string to parse.
6906            If an Expression instance is passed, this is used as-is.
6907        dialect: the dialect used to parse the input expression (in the case that the
6908            input expression is a SQL string).
6909        copy: Whether to copy `expression` (only applies to expressions).
6910        **opts: other options to use to parse the input expressions (again, in the case
6911            that the input expression is a SQL string).
6912
6913    Returns:
6914        The new Condition instance
6915    """
6916    return maybe_parse(
6917        expression,
6918        into=Condition,
6919        dialect=dialect,
6920        copy=copy,
6921        **opts,
6922    )
6923
6924
6925def and_(
6926    *expressions: t.Optional[ExpOrStr], dialect: DialectType = None, copy: bool = True, **opts
6927) -> Condition:
6928    """
6929    Combine multiple conditions with an AND logical operator.
6930
6931    Example:
6932        >>> and_("x=1", and_("y=1", "z=1")).sql()
6933        'x = 1 AND (y = 1 AND z = 1)'
6934
6935    Args:
6936        *expressions: the SQL code strings to parse.
6937            If an Expression instance is passed, this is used as-is.
6938        dialect: the dialect used to parse the input expression.
6939        copy: whether to copy `expressions` (only applies to Expressions).
6940        **opts: other options to use to parse the input expressions.
6941
6942    Returns:
6943        The new condition
6944    """
6945    return t.cast(Condition, _combine(expressions, And, dialect, copy=copy, **opts))
6946
6947
6948def or_(
6949    *expressions: t.Optional[ExpOrStr], dialect: DialectType = None, copy: bool = True, **opts
6950) -> Condition:
6951    """
6952    Combine multiple conditions with an OR logical operator.
6953
6954    Example:
6955        >>> or_("x=1", or_("y=1", "z=1")).sql()
6956        'x = 1 OR (y = 1 OR z = 1)'
6957
6958    Args:
6959        *expressions: the SQL code strings to parse.
6960            If an Expression instance is passed, this is used as-is.
6961        dialect: the dialect used to parse the input expression.
6962        copy: whether to copy `expressions` (only applies to Expressions).
6963        **opts: other options to use to parse the input expressions.
6964
6965    Returns:
6966        The new condition
6967    """
6968    return t.cast(Condition, _combine(expressions, Or, dialect, copy=copy, **opts))
6969
6970
6971def xor(
6972    *expressions: t.Optional[ExpOrStr], dialect: DialectType = None, copy: bool = True, **opts
6973) -> Condition:
6974    """
6975    Combine multiple conditions with an XOR logical operator.
6976
6977    Example:
6978        >>> xor("x=1", xor("y=1", "z=1")).sql()
6979        'x = 1 XOR (y = 1 XOR z = 1)'
6980
6981    Args:
6982        *expressions: the SQL code strings to parse.
6983            If an Expression instance is passed, this is used as-is.
6984        dialect: the dialect used to parse the input expression.
6985        copy: whether to copy `expressions` (only applies to Expressions).
6986        **opts: other options to use to parse the input expressions.
6987
6988    Returns:
6989        The new condition
6990    """
6991    return t.cast(Condition, _combine(expressions, Xor, dialect, copy=copy, **opts))
6992
6993
6994def not_(expression: ExpOrStr, dialect: DialectType = None, copy: bool = True, **opts) -> Not:
6995    """
6996    Wrap a condition with a NOT operator.
6997
6998    Example:
6999        >>> not_("this_suit='black'").sql()
7000        "NOT this_suit = 'black'"
7001
7002    Args:
7003        expression: the SQL code string to parse.
7004            If an Expression instance is passed, this is used as-is.
7005        dialect: the dialect used to parse the input expression.
7006        copy: whether to copy the expression or not.
7007        **opts: other options to use to parse the input expressions.
7008
7009    Returns:
7010        The new condition.
7011    """
7012    this = condition(
7013        expression,
7014        dialect=dialect,
7015        copy=copy,
7016        **opts,
7017    )
7018    return Not(this=_wrap(this, Connector))
7019
7020
7021def paren(expression: ExpOrStr, copy: bool = True) -> Paren:
7022    """
7023    Wrap an expression in parentheses.
7024
7025    Example:
7026        >>> paren("5 + 3").sql()
7027        '(5 + 3)'
7028
7029    Args:
7030        expression: the SQL code string to parse.
7031            If an Expression instance is passed, this is used as-is.
7032        copy: whether to copy the expression or not.
7033
7034    Returns:
7035        The wrapped expression.
7036    """
7037    return Paren(this=maybe_parse(expression, copy=copy))
7038
7039
7040SAFE_IDENTIFIER_RE: t.Pattern[str] = re.compile(r"^[_a-zA-Z][\w]*$")
7041
7042
7043@t.overload
7044def to_identifier(name: None, quoted: t.Optional[bool] = None, copy: bool = True) -> None: ...
7045
7046
7047@t.overload
7048def to_identifier(
7049    name: str | Identifier, quoted: t.Optional[bool] = None, copy: bool = True
7050) -> Identifier: ...
7051
7052
7053def to_identifier(name, quoted=None, copy=True):
7054    """Builds an identifier.
7055
7056    Args:
7057        name: The name to turn into an identifier.
7058        quoted: Whether to force quote the identifier.
7059        copy: Whether to copy name if it's an Identifier.
7060
7061    Returns:
7062        The identifier ast node.
7063    """
7064
7065    if name is None:
7066        return None
7067
7068    if isinstance(name, Identifier):
7069        identifier = maybe_copy(name, copy)
7070    elif isinstance(name, str):
7071        identifier = Identifier(
7072            this=name,
7073            quoted=not SAFE_IDENTIFIER_RE.match(name) if quoted is None else quoted,
7074        )
7075    else:
7076        raise ValueError(f"Name needs to be a string or an Identifier, got: {name.__class__}")
7077    return identifier
7078
7079
7080def parse_identifier(name: str | Identifier, dialect: DialectType = None) -> Identifier:
7081    """
7082    Parses a given string into an identifier.
7083
7084    Args:
7085        name: The name to parse into an identifier.
7086        dialect: The dialect to parse against.
7087
7088    Returns:
7089        The identifier ast node.
7090    """
7091    try:
7092        expression = maybe_parse(name, dialect=dialect, into=Identifier)
7093    except (ParseError, TokenError):
7094        expression = to_identifier(name)
7095
7096    return expression
7097
7098
7099INTERVAL_STRING_RE = re.compile(r"\s*([0-9]+)\s*([a-zA-Z]+)\s*")
7100
7101
7102def to_interval(interval: str | Literal) -> Interval:
7103    """Builds an interval expression from a string like '1 day' or '5 months'."""
7104    if isinstance(interval, Literal):
7105        if not interval.is_string:
7106            raise ValueError("Invalid interval string.")
7107
7108        interval = interval.this
7109
7110    interval_parts = INTERVAL_STRING_RE.match(interval)  # type: ignore
7111
7112    if not interval_parts:
7113        raise ValueError("Invalid interval string.")
7114
7115    return Interval(
7116        this=Literal.string(interval_parts.group(1)),
7117        unit=Var(this=interval_parts.group(2).upper()),
7118    )
7119
7120
7121def to_table(
7122    sql_path: str | Table, dialect: DialectType = None, copy: bool = True, **kwargs
7123) -> Table:
7124    """
7125    Create a table expression from a `[catalog].[schema].[table]` sql path. Catalog and schema are optional.
7126    If a table is passed in then that table is returned.
7127
7128    Args:
7129        sql_path: a `[catalog].[schema].[table]` string.
7130        dialect: the source dialect according to which the table name will be parsed.
7131        copy: Whether to copy a table if it is passed in.
7132        kwargs: the kwargs to instantiate the resulting `Table` expression with.
7133
7134    Returns:
7135        A table expression.
7136    """
7137    if isinstance(sql_path, Table):
7138        return maybe_copy(sql_path, copy=copy)
7139
7140    table = maybe_parse(sql_path, into=Table, dialect=dialect)
7141
7142    for k, v in kwargs.items():
7143        table.set(k, v)
7144
7145    return table
7146
7147
7148def to_column(
7149    sql_path: str | Column,
7150    quoted: t.Optional[bool] = None,
7151    dialect: DialectType = None,
7152    copy: bool = True,
7153    **kwargs,
7154) -> Column:
7155    """
7156    Create a column from a `[table].[column]` sql path. Table is optional.
7157    If a column is passed in then that column is returned.
7158
7159    Args:
7160        sql_path: a `[table].[column]` string.
7161        quoted: Whether or not to force quote identifiers.
7162        dialect: the source dialect according to which the column name will be parsed.
7163        copy: Whether to copy a column if it is passed in.
7164        kwargs: the kwargs to instantiate the resulting `Column` expression with.
7165
7166    Returns:
7167        A column expression.
7168    """
7169    if isinstance(sql_path, Column):
7170        return maybe_copy(sql_path, copy=copy)
7171
7172    try:
7173        col = maybe_parse(sql_path, into=Column, dialect=dialect)
7174    except ParseError:
7175        return column(*reversed(sql_path.split(".")), quoted=quoted, **kwargs)
7176
7177    for k, v in kwargs.items():
7178        col.set(k, v)
7179
7180    if quoted:
7181        for i in col.find_all(Identifier):
7182            i.set("quoted", True)
7183
7184    return col
7185
7186
7187def alias_(
7188    expression: ExpOrStr,
7189    alias: t.Optional[str | Identifier],
7190    table: bool | t.Sequence[str | Identifier] = False,
7191    quoted: t.Optional[bool] = None,
7192    dialect: DialectType = None,
7193    copy: bool = True,
7194    **opts,
7195):
7196    """Create an Alias expression.
7197
7198    Example:
7199        >>> alias_('foo', 'bar').sql()
7200        'foo AS bar'
7201
7202        >>> alias_('(select 1, 2)', 'bar', table=['a', 'b']).sql()
7203        '(SELECT 1, 2) AS bar(a, b)'
7204
7205    Args:
7206        expression: the SQL code strings to parse.
7207            If an Expression instance is passed, this is used as-is.
7208        alias: the alias name to use. If the name has
7209            special characters it is quoted.
7210        table: Whether to create a table alias, can also be a list of columns.
7211        quoted: whether to quote the alias
7212        dialect: the dialect used to parse the input expression.
7213        copy: Whether to copy the expression.
7214        **opts: other options to use to parse the input expressions.
7215
7216    Returns:
7217        Alias: the aliased expression
7218    """
7219    exp = maybe_parse(expression, dialect=dialect, copy=copy, **opts)
7220    alias = to_identifier(alias, quoted=quoted)
7221
7222    if table:
7223        table_alias = TableAlias(this=alias)
7224        exp.set("alias", table_alias)
7225
7226        if not isinstance(table, bool):
7227            for column in table:
7228                table_alias.append("columns", to_identifier(column, quoted=quoted))
7229
7230        return exp
7231
7232    # We don't set the "alias" arg for Window expressions, because that would add an IDENTIFIER node in
7233    # the AST, representing a "named_window" [1] construct (eg. bigquery). What we want is an ALIAS node
7234    # for the complete Window expression.
7235    #
7236    # [1]: https://cloud.google.com/bigquery/docs/reference/standard-sql/window-function-calls
7237
7238    if "alias" in exp.arg_types and not isinstance(exp, Window):
7239        exp.set("alias", alias)
7240        return exp
7241    return Alias(this=exp, alias=alias)
7242
7243
7244def subquery(
7245    expression: ExpOrStr,
7246    alias: t.Optional[Identifier | str] = None,
7247    dialect: DialectType = None,
7248    **opts,
7249) -> Select:
7250    """
7251    Build a subquery expression that's selected from.
7252
7253    Example:
7254        >>> subquery('select x from tbl', 'bar').select('x').sql()
7255        'SELECT x FROM (SELECT x FROM tbl) AS bar'
7256
7257    Args:
7258        expression: the SQL code strings to parse.
7259            If an Expression instance is passed, this is used as-is.
7260        alias: the alias name to use.
7261        dialect: the dialect used to parse the input expression.
7262        **opts: other options to use to parse the input expressions.
7263
7264    Returns:
7265        A new Select instance with the subquery expression included.
7266    """
7267
7268    expression = maybe_parse(expression, dialect=dialect, **opts).subquery(alias, **opts)
7269    return Select().from_(expression, dialect=dialect, **opts)
7270
7271
7272@t.overload
7273def column(
7274    col: str | Identifier,
7275    table: t.Optional[str | Identifier] = None,
7276    db: t.Optional[str | Identifier] = None,
7277    catalog: t.Optional[str | Identifier] = None,
7278    *,
7279    fields: t.Collection[t.Union[str, Identifier]],
7280    quoted: t.Optional[bool] = None,
7281    copy: bool = True,
7282) -> Dot:
7283    pass
7284
7285
7286@t.overload
7287def column(
7288    col: str | Identifier,
7289    table: t.Optional[str | Identifier] = None,
7290    db: t.Optional[str | Identifier] = None,
7291    catalog: t.Optional[str | Identifier] = None,
7292    *,
7293    fields: Lit[None] = None,
7294    quoted: t.Optional[bool] = None,
7295    copy: bool = True,
7296) -> Column:
7297    pass
7298
7299
7300def column(
7301    col,
7302    table=None,
7303    db=None,
7304    catalog=None,
7305    *,
7306    fields=None,
7307    quoted=None,
7308    copy=True,
7309):
7310    """
7311    Build a Column.
7312
7313    Args:
7314        col: Column name.
7315        table: Table name.
7316        db: Database name.
7317        catalog: Catalog name.
7318        fields: Additional fields using dots.
7319        quoted: Whether to force quotes on the column's identifiers.
7320        copy: Whether to copy identifiers if passed in.
7321
7322    Returns:
7323        The new Column instance.
7324    """
7325    this = Column(
7326        this=to_identifier(col, quoted=quoted, copy=copy),
7327        table=to_identifier(table, quoted=quoted, copy=copy),
7328        db=to_identifier(db, quoted=quoted, copy=copy),
7329        catalog=to_identifier(catalog, quoted=quoted, copy=copy),
7330    )
7331
7332    if fields:
7333        this = Dot.build(
7334            (this, *(to_identifier(field, quoted=quoted, copy=copy) for field in fields))
7335        )
7336    return this
7337
7338
7339def cast(
7340    expression: ExpOrStr, to: DATA_TYPE, copy: bool = True, dialect: DialectType = None, **opts
7341) -> Cast:
7342    """Cast an expression to a data type.
7343
7344    Example:
7345        >>> cast('x + 1', 'int').sql()
7346        'CAST(x + 1 AS INT)'
7347
7348    Args:
7349        expression: The expression to cast.
7350        to: The datatype to cast to.
7351        copy: Whether to copy the supplied expressions.
7352        dialect: The target dialect. This is used to prevent a re-cast in the following scenario:
7353            - The expression to be cast is already a exp.Cast expression
7354            - The existing cast is to a type that is logically equivalent to new type
7355
7356            For example, if :expression='CAST(x as DATETIME)' and :to=Type.TIMESTAMP,
7357            but in the target dialect DATETIME is mapped to TIMESTAMP, then we will NOT return `CAST(x (as DATETIME) as TIMESTAMP)`
7358            and instead just return the original expression `CAST(x as DATETIME)`.
7359
7360            This is to prevent it being output as a double cast `CAST(x (as TIMESTAMP) as TIMESTAMP)` once the DATETIME -> TIMESTAMP
7361            mapping is applied in the target dialect generator.
7362
7363    Returns:
7364        The new Cast instance.
7365    """
7366    expr = maybe_parse(expression, copy=copy, dialect=dialect, **opts)
7367    data_type = DataType.build(to, copy=copy, dialect=dialect, **opts)
7368
7369    # dont re-cast if the expression is already a cast to the correct type
7370    if isinstance(expr, Cast):
7371        from sqlglot.dialects.dialect import Dialect
7372
7373        target_dialect = Dialect.get_or_raise(dialect)
7374        type_mapping = target_dialect.generator_class.TYPE_MAPPING
7375
7376        existing_cast_type: DataType.Type = expr.to.this
7377        new_cast_type: DataType.Type = data_type.this
7378        types_are_equivalent = type_mapping.get(
7379            existing_cast_type, existing_cast_type
7380        ) == type_mapping.get(new_cast_type, new_cast_type)
7381        if expr.is_type(data_type) or types_are_equivalent:
7382            return expr
7383
7384    expr = Cast(this=expr, to=data_type)
7385    expr.type = data_type
7386
7387    return expr
7388
7389
7390def table_(
7391    table: Identifier | str,
7392    db: t.Optional[Identifier | str] = None,
7393    catalog: t.Optional[Identifier | str] = None,
7394    quoted: t.Optional[bool] = None,
7395    alias: t.Optional[Identifier | str] = None,
7396) -> Table:
7397    """Build a Table.
7398
7399    Args:
7400        table: Table name.
7401        db: Database name.
7402        catalog: Catalog name.
7403        quote: Whether to force quotes on the table's identifiers.
7404        alias: Table's alias.
7405
7406    Returns:
7407        The new Table instance.
7408    """
7409    return Table(
7410        this=to_identifier(table, quoted=quoted) if table else None,
7411        db=to_identifier(db, quoted=quoted) if db else None,
7412        catalog=to_identifier(catalog, quoted=quoted) if catalog else None,
7413        alias=TableAlias(this=to_identifier(alias)) if alias else None,
7414    )
7415
7416
7417def values(
7418    values: t.Iterable[t.Tuple[t.Any, ...]],
7419    alias: t.Optional[str] = None,
7420    columns: t.Optional[t.Iterable[str] | t.Dict[str, DataType]] = None,
7421) -> Values:
7422    """Build VALUES statement.
7423
7424    Example:
7425        >>> values([(1, '2')]).sql()
7426        "VALUES (1, '2')"
7427
7428    Args:
7429        values: values statements that will be converted to SQL
7430        alias: optional alias
7431        columns: Optional list of ordered column names or ordered dictionary of column names to types.
7432         If either are provided then an alias is also required.
7433
7434    Returns:
7435        Values: the Values expression object
7436    """
7437    if columns and not alias:
7438        raise ValueError("Alias is required when providing columns")
7439
7440    return Values(
7441        expressions=[convert(tup) for tup in values],
7442        alias=(
7443            TableAlias(this=to_identifier(alias), columns=[to_identifier(x) for x in columns])
7444            if columns
7445            else (TableAlias(this=to_identifier(alias)) if alias else None)
7446        ),
7447    )
7448
7449
7450def var(name: t.Optional[ExpOrStr]) -> Var:
7451    """Build a SQL variable.
7452
7453    Example:
7454        >>> repr(var('x'))
7455        'Var(this=x)'
7456
7457        >>> repr(var(column('x', table='y')))
7458        'Var(this=x)'
7459
7460    Args:
7461        name: The name of the var or an expression who's name will become the var.
7462
7463    Returns:
7464        The new variable node.
7465    """
7466    if not name:
7467        raise ValueError("Cannot convert empty name into var.")
7468
7469    if isinstance(name, Expression):
7470        name = name.name
7471    return Var(this=name)
7472
7473
7474def rename_table(
7475    old_name: str | Table,
7476    new_name: str | Table,
7477    dialect: DialectType = None,
7478) -> Alter:
7479    """Build ALTER TABLE... RENAME... expression
7480
7481    Args:
7482        old_name: The old name of the table
7483        new_name: The new name of the table
7484        dialect: The dialect to parse the table.
7485
7486    Returns:
7487        Alter table expression
7488    """
7489    old_table = to_table(old_name, dialect=dialect)
7490    new_table = to_table(new_name, dialect=dialect)
7491    return Alter(
7492        this=old_table,
7493        kind="TABLE",
7494        actions=[
7495            RenameTable(this=new_table),
7496        ],
7497    )
7498
7499
7500def rename_column(
7501    table_name: str | Table,
7502    old_column_name: str | Column,
7503    new_column_name: str | Column,
7504    exists: t.Optional[bool] = None,
7505    dialect: DialectType = None,
7506) -> Alter:
7507    """Build ALTER TABLE... RENAME COLUMN... expression
7508
7509    Args:
7510        table_name: Name of the table
7511        old_column: The old name of the column
7512        new_column: The new name of the column
7513        exists: Whether to add the `IF EXISTS` clause
7514        dialect: The dialect to parse the table/column.
7515
7516    Returns:
7517        Alter table expression
7518    """
7519    table = to_table(table_name, dialect=dialect)
7520    old_column = to_column(old_column_name, dialect=dialect)
7521    new_column = to_column(new_column_name, dialect=dialect)
7522    return Alter(
7523        this=table,
7524        kind="TABLE",
7525        actions=[
7526            RenameColumn(this=old_column, to=new_column, exists=exists),
7527        ],
7528    )
7529
7530
7531def convert(value: t.Any, copy: bool = False) -> Expression:
7532    """Convert a python value into an expression object.
7533
7534    Raises an error if a conversion is not possible.
7535
7536    Args:
7537        value: A python object.
7538        copy: Whether to copy `value` (only applies to Expressions and collections).
7539
7540    Returns:
7541        The equivalent expression object.
7542    """
7543    if isinstance(value, Expression):
7544        return maybe_copy(value, copy)
7545    if isinstance(value, str):
7546        return Literal.string(value)
7547    if isinstance(value, bool):
7548        return Boolean(this=value)
7549    if value is None or (isinstance(value, float) and math.isnan(value)):
7550        return null()
7551    if isinstance(value, numbers.Number):
7552        return Literal.number(value)
7553    if isinstance(value, bytes):
7554        return HexString(this=value.hex())
7555    if isinstance(value, datetime.datetime):
7556        datetime_literal = Literal.string(value.isoformat(sep=" "))
7557
7558        tz = None
7559        if value.tzinfo:
7560            # this works for zoneinfo.ZoneInfo, pytz.timezone and datetime.datetime.utc to return IANA timezone names like "America/Los_Angeles"
7561            # instead of abbreviations like "PDT". This is for consistency with other timezone handling functions in SQLGlot
7562            tz = Literal.string(str(value.tzinfo))
7563
7564        return TimeStrToTime(this=datetime_literal, zone=tz)
7565    if isinstance(value, datetime.date):
7566        date_literal = Literal.string(value.strftime("%Y-%m-%d"))
7567        return DateStrToDate(this=date_literal)
7568    if isinstance(value, tuple):
7569        if hasattr(value, "_fields"):
7570            return Struct(
7571                expressions=[
7572                    PropertyEQ(
7573                        this=to_identifier(k), expression=convert(getattr(value, k), copy=copy)
7574                    )
7575                    for k in value._fields
7576                ]
7577            )
7578        return Tuple(expressions=[convert(v, copy=copy) for v in value])
7579    if isinstance(value, list):
7580        return Array(expressions=[convert(v, copy=copy) for v in value])
7581    if isinstance(value, dict):
7582        return Map(
7583            keys=Array(expressions=[convert(k, copy=copy) for k in value]),
7584            values=Array(expressions=[convert(v, copy=copy) for v in value.values()]),
7585        )
7586    if hasattr(value, "__dict__"):
7587        return Struct(
7588            expressions=[
7589                PropertyEQ(this=to_identifier(k), expression=convert(v, copy=copy))
7590                for k, v in value.__dict__.items()
7591            ]
7592        )
7593    raise ValueError(f"Cannot convert {value}")
7594
7595
7596def replace_children(expression: Expression, fun: t.Callable, *args, **kwargs) -> None:
7597    """
7598    Replace children of an expression with the result of a lambda fun(child) -> exp.
7599    """
7600    for k, v in tuple(expression.args.items()):
7601        is_list_arg = type(v) is list
7602
7603        child_nodes = v if is_list_arg else [v]
7604        new_child_nodes = []
7605
7606        for cn in child_nodes:
7607            if isinstance(cn, Expression):
7608                for child_node in ensure_collection(fun(cn, *args, **kwargs)):
7609                    new_child_nodes.append(child_node)
7610            else:
7611                new_child_nodes.append(cn)
7612
7613        expression.set(k, new_child_nodes if is_list_arg else seq_get(new_child_nodes, 0))
7614
7615
7616def replace_tree(
7617    expression: Expression,
7618    fun: t.Callable,
7619    prune: t.Optional[t.Callable[[Expression], bool]] = None,
7620) -> Expression:
7621    """
7622    Replace an entire tree with the result of function calls on each node.
7623
7624    This will be traversed in reverse dfs, so leaves first.
7625    If new nodes are created as a result of function calls, they will also be traversed.
7626    """
7627    stack = list(expression.dfs(prune=prune))
7628
7629    while stack:
7630        node = stack.pop()
7631        new_node = fun(node)
7632
7633        if new_node is not node:
7634            node.replace(new_node)
7635
7636            if isinstance(new_node, Expression):
7637                stack.append(new_node)
7638
7639    return new_node
7640
7641
7642def column_table_names(expression: Expression, exclude: str = "") -> t.Set[str]:
7643    """
7644    Return all table names referenced through columns in an expression.
7645
7646    Example:
7647        >>> import sqlglot
7648        >>> sorted(column_table_names(sqlglot.parse_one("a.b AND c.d AND c.e")))
7649        ['a', 'c']
7650
7651    Args:
7652        expression: expression to find table names.
7653        exclude: a table name to exclude
7654
7655    Returns:
7656        A list of unique names.
7657    """
7658    return {
7659        table
7660        for table in (column.table for column in expression.find_all(Column))
7661        if table and table != exclude
7662    }
7663
7664
7665def table_name(table: Table | str, dialect: DialectType = None, identify: bool = False) -> str:
7666    """Get the full name of a table as a string.
7667
7668    Args:
7669        table: Table expression node or string.
7670        dialect: The dialect to generate the table name for.
7671        identify: Determines when an identifier should be quoted. Possible values are:
7672            False (default): Never quote, except in cases where it's mandatory by the dialect.
7673            True: Always quote.
7674
7675    Examples:
7676        >>> from sqlglot import exp, parse_one
7677        >>> table_name(parse_one("select * from a.b.c").find(exp.Table))
7678        'a.b.c'
7679
7680    Returns:
7681        The table name.
7682    """
7683
7684    table = maybe_parse(table, into=Table, dialect=dialect)
7685
7686    if not table:
7687        raise ValueError(f"Cannot parse {table}")
7688
7689    return ".".join(
7690        (
7691            part.sql(dialect=dialect, identify=True, copy=False)
7692            if identify or not SAFE_IDENTIFIER_RE.match(part.name)
7693            else part.name
7694        )
7695        for part in table.parts
7696    )
7697
7698
7699def normalize_table_name(table: str | Table, dialect: DialectType = None, copy: bool = True) -> str:
7700    """Returns a case normalized table name without quotes.
7701
7702    Args:
7703        table: the table to normalize
7704        dialect: the dialect to use for normalization rules
7705        copy: whether to copy the expression.
7706
7707    Examples:
7708        >>> normalize_table_name("`A-B`.c", dialect="bigquery")
7709        'A-B.c'
7710    """
7711    from sqlglot.optimizer.normalize_identifiers import normalize_identifiers
7712
7713    return ".".join(
7714        p.name
7715        for p in normalize_identifiers(
7716            to_table(table, dialect=dialect, copy=copy), dialect=dialect
7717        ).parts
7718    )
7719
7720
7721def replace_tables(
7722    expression: E, mapping: t.Dict[str, str], dialect: DialectType = None, copy: bool = True
7723) -> E:
7724    """Replace all tables in expression according to the mapping.
7725
7726    Args:
7727        expression: expression node to be transformed and replaced.
7728        mapping: mapping of table names.
7729        dialect: the dialect of the mapping table
7730        copy: whether to copy the expression.
7731
7732    Examples:
7733        >>> from sqlglot import exp, parse_one
7734        >>> replace_tables(parse_one("select * from a.b"), {"a.b": "c"}).sql()
7735        'SELECT * FROM c /* a.b */'
7736
7737    Returns:
7738        The mapped expression.
7739    """
7740
7741    mapping = {normalize_table_name(k, dialect=dialect): v for k, v in mapping.items()}
7742
7743    def _replace_tables(node: Expression) -> Expression:
7744        if isinstance(node, Table):
7745            original = normalize_table_name(node, dialect=dialect)
7746            new_name = mapping.get(original)
7747
7748            if new_name:
7749                table = to_table(
7750                    new_name,
7751                    **{k: v for k, v in node.args.items() if k not in TABLE_PARTS},
7752                    dialect=dialect,
7753                )
7754                table.add_comments([original])
7755                return table
7756        return node
7757
7758    return expression.transform(_replace_tables, copy=copy)  # type: ignore
7759
7760
7761def replace_placeholders(expression: Expression, *args, **kwargs) -> Expression:
7762    """Replace placeholders in an expression.
7763
7764    Args:
7765        expression: expression node to be transformed and replaced.
7766        args: positional names that will substitute unnamed placeholders in the given order.
7767        kwargs: keyword arguments that will substitute named placeholders.
7768
7769    Examples:
7770        >>> from sqlglot import exp, parse_one
7771        >>> replace_placeholders(
7772        ...     parse_one("select * from :tbl where ? = ?"),
7773        ...     exp.to_identifier("str_col"), "b", tbl=exp.to_identifier("foo")
7774        ... ).sql()
7775        "SELECT * FROM foo WHERE str_col = 'b'"
7776
7777    Returns:
7778        The mapped expression.
7779    """
7780
7781    def _replace_placeholders(node: Expression, args, **kwargs) -> Expression:
7782        if isinstance(node, Placeholder):
7783            if node.this:
7784                new_name = kwargs.get(node.this)
7785                if new_name is not None:
7786                    return convert(new_name)
7787            else:
7788                try:
7789                    return convert(next(args))
7790                except StopIteration:
7791                    pass
7792        return node
7793
7794    return expression.transform(_replace_placeholders, iter(args), **kwargs)
7795
7796
7797def expand(
7798    expression: Expression,
7799    sources: t.Dict[str, Query],
7800    dialect: DialectType = None,
7801    copy: bool = True,
7802) -> Expression:
7803    """Transforms an expression by expanding all referenced sources into subqueries.
7804
7805    Examples:
7806        >>> from sqlglot import parse_one
7807        >>> expand(parse_one("select * from x AS z"), {"x": parse_one("select * from y")}).sql()
7808        'SELECT * FROM (SELECT * FROM y) AS z /* source: x */'
7809
7810        >>> expand(parse_one("select * from x AS z"), {"x": parse_one("select * from y"), "y": parse_one("select * from z")}).sql()
7811        'SELECT * FROM (SELECT * FROM (SELECT * FROM z) AS y /* source: y */) AS z /* source: x */'
7812
7813    Args:
7814        expression: The expression to expand.
7815        sources: A dictionary of name to Queries.
7816        dialect: The dialect of the sources dict.
7817        copy: Whether to copy the expression during transformation. Defaults to True.
7818
7819    Returns:
7820        The transformed expression.
7821    """
7822    sources = {normalize_table_name(k, dialect=dialect): v for k, v in sources.items()}
7823
7824    def _expand(node: Expression):
7825        if isinstance(node, Table):
7826            name = normalize_table_name(node, dialect=dialect)
7827            source = sources.get(name)
7828            if source:
7829                subquery = source.subquery(node.alias or name)
7830                subquery.comments = [f"source: {name}"]
7831                return subquery.transform(_expand, copy=False)
7832        return node
7833
7834    return expression.transform(_expand, copy=copy)
7835
7836
7837def func(name: str, *args, copy: bool = True, dialect: DialectType = None, **kwargs) -> Func:
7838    """
7839    Returns a Func expression.
7840
7841    Examples:
7842        >>> func("abs", 5).sql()
7843        'ABS(5)'
7844
7845        >>> func("cast", this=5, to=DataType.build("DOUBLE")).sql()
7846        'CAST(5 AS DOUBLE)'
7847
7848    Args:
7849        name: the name of the function to build.
7850        args: the args used to instantiate the function of interest.
7851        copy: whether to copy the argument expressions.
7852        dialect: the source dialect.
7853        kwargs: the kwargs used to instantiate the function of interest.
7854
7855    Note:
7856        The arguments `args` and `kwargs` are mutually exclusive.
7857
7858    Returns:
7859        An instance of the function of interest, or an anonymous function, if `name` doesn't
7860        correspond to an existing `sqlglot.expressions.Func` class.
7861    """
7862    if args and kwargs:
7863        raise ValueError("Can't use both args and kwargs to instantiate a function.")
7864
7865    from sqlglot.dialects.dialect import Dialect
7866
7867    dialect = Dialect.get_or_raise(dialect)
7868
7869    converted: t.List[Expression] = [maybe_parse(arg, dialect=dialect, copy=copy) for arg in args]
7870    kwargs = {key: maybe_parse(value, dialect=dialect, copy=copy) for key, value in kwargs.items()}
7871
7872    constructor = dialect.parser_class.FUNCTIONS.get(name.upper())
7873    if constructor:
7874        if converted:
7875            if "dialect" in constructor.__code__.co_varnames:
7876                function = constructor(converted, dialect=dialect)
7877            else:
7878                function = constructor(converted)
7879        elif constructor.__name__ == "from_arg_list":
7880            function = constructor.__self__(**kwargs)  # type: ignore
7881        else:
7882            constructor = FUNCTION_BY_NAME.get(name.upper())
7883            if constructor:
7884                function = constructor(**kwargs)
7885            else:
7886                raise ValueError(
7887                    f"Unable to convert '{name}' into a Func. Either manually construct "
7888                    "the Func expression of interest or parse the function call."
7889                )
7890    else:
7891        kwargs = kwargs or {"expressions": converted}
7892        function = Anonymous(this=name, **kwargs)
7893
7894    for error_message in function.error_messages(converted):
7895        raise ValueError(error_message)
7896
7897    return function
7898
7899
7900def case(
7901    expression: t.Optional[ExpOrStr] = None,
7902    **opts,
7903) -> Case:
7904    """
7905    Initialize a CASE statement.
7906
7907    Example:
7908        case().when("a = 1", "foo").else_("bar")
7909
7910    Args:
7911        expression: Optionally, the input expression (not all dialects support this)
7912        **opts: Extra keyword arguments for parsing `expression`
7913    """
7914    if expression is not None:
7915        this = maybe_parse(expression, **opts)
7916    else:
7917        this = None
7918    return Case(this=this, ifs=[])
7919
7920
7921def array(
7922    *expressions: ExpOrStr, copy: bool = True, dialect: DialectType = None, **kwargs
7923) -> Array:
7924    """
7925    Returns an array.
7926
7927    Examples:
7928        >>> array(1, 'x').sql()
7929        'ARRAY(1, x)'
7930
7931    Args:
7932        expressions: the expressions to add to the array.
7933        copy: whether to copy the argument expressions.
7934        dialect: the source dialect.
7935        kwargs: the kwargs used to instantiate the function of interest.
7936
7937    Returns:
7938        An array expression.
7939    """
7940    return Array(
7941        expressions=[
7942            maybe_parse(expression, copy=copy, dialect=dialect, **kwargs)
7943            for expression in expressions
7944        ]
7945    )
7946
7947
7948def tuple_(
7949    *expressions: ExpOrStr, copy: bool = True, dialect: DialectType = None, **kwargs
7950) -> Tuple:
7951    """
7952    Returns an tuple.
7953
7954    Examples:
7955        >>> tuple_(1, 'x').sql()
7956        '(1, x)'
7957
7958    Args:
7959        expressions: the expressions to add to the tuple.
7960        copy: whether to copy the argument expressions.
7961        dialect: the source dialect.
7962        kwargs: the kwargs used to instantiate the function of interest.
7963
7964    Returns:
7965        A tuple expression.
7966    """
7967    return Tuple(
7968        expressions=[
7969            maybe_parse(expression, copy=copy, dialect=dialect, **kwargs)
7970            for expression in expressions
7971        ]
7972    )
7973
7974
7975def true() -> Boolean:
7976    """
7977    Returns a true Boolean expression.
7978    """
7979    return Boolean(this=True)
7980
7981
7982def false() -> Boolean:
7983    """
7984    Returns a false Boolean expression.
7985    """
7986    return Boolean(this=False)
7987
7988
7989def null() -> Null:
7990    """
7991    Returns a Null expression.
7992    """
7993    return Null()
7994
7995
7996NONNULL_CONSTANTS = (
7997    Literal,
7998    Boolean,
7999)
8000
8001CONSTANTS = (
8002    Literal,
8003    Boolean,
8004    Null,
8005)