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