sqlglot 27.27.0__py3-none-any.whl → 28.4.0__py3-none-any.whl

sqlglot/expressions.py

@@ -16,6 +16,7 @@ import datetime
 import math
 import numbers
 import re
+import sys
 import textwrap
 import typing as t
 from collections import deque
@@ -54,6 +55,7 @@ class _Expression(type):
         # When an Expression class is created, its key is automatically set
         # to be the lowercase version of the class' name.
         klass.key = clsname.lower()
+        klass.required_args = {k for k, v in klass.arg_types.items() if v}
 
         # This is so that docstrings are not inherited in pdoc
         klass.__doc__ = klass.__doc__ or ""
@@ -66,6 +68,7 @@ SQLGLOT_ANONYMOUS = "sqlglot.anonymous"
 TABLE_PARTS = ("this", "db", "catalog")
 COLUMN_PARTS = ("this", "table", "db", "catalog")
 POSITION_META_KEYS = ("line", "col", "start", "end")
+UNITTEST = "unittest" in sys.modules or "pytest" in sys.modules
 
 
 class Expression(metaclass=_Expression):
@@ -102,6 +105,7 @@ class Expression(metaclass=_Expression):
 
     key = "expression"
     arg_types = {"this": True}
+    required_args = {"this"}
     __slots__ = ("args", "parent", "arg_key", "index", "comments", "_type", "_meta", "_hash")
 
     def __init__(self, **args: t.Any):
@@ -118,7 +122,7 @@ class Expression(metaclass=_Expression):
             self._set_parent(arg_key, value)
 
     def __eq__(self, other) -> bool:
-        return type(self) is type(other) and hash(self) == hash(other)
+        return self is other or (type(self) is type(other) and hash(self) == hash(other))
 
     def __hash__(self) -> int:
         if self._hash is None:
@@ -454,9 +458,8 @@ class Expression(metaclass=_Expression):
                 for v in reversed(vs) if reverse else vs:  # type: ignore
                     if hasattr(v, "parent"):
                         yield v
-            else:
-                if hasattr(vs, "parent"):
-                    yield vs
+            elif hasattr(vs, "parent"):
+                yield vs
 
     def find(self, *expression_types: t.Type[E], bfs: bool = True) -> t.Optional[E]:
         """
@@ -769,12 +772,14 @@ class Expression(metaclass=_Expression):
         """
         errors: t.List[str] = []
 
-        for k in self.args:
-            if k not in self.arg_types:
-                errors.append(f"Unexpected keyword: '{k}' for {self.__class__}")
-        for k, mandatory in self.arg_types.items():
+        if UNITTEST:
+            for k in self.args:
+                if k not in self.arg_types:
+                    raise TypeError(f"Unexpected keyword: '{k}' for {self.__class__}")
+
+        for k in self.required_args:
             v = self.args.get(k)
-            if mandatory and (v is None or (isinstance(v, list) and not v)):
+            if v is None or (type(v) is list and not v):
                 errors.append(f"Required keyword: '{k}' missing for {self.__class__}")
 
         if (
@@ -884,29 +889,39 @@ class Expression(metaclass=_Expression):
         return not_(self, copy=copy)
 
     def update_positions(
-        self: E, other: t.Optional[Token | Expression] = None, **kwargs: t.Any
+        self: E,
+        other: t.Optional[Token | Expression] = None,
+        line: t.Optional[int] = None,
+        col: t.Optional[int] = None,
+        start: t.Optional[int] = None,
+        end: t.Optional[int] = None,
     ) -> E:
         """
         Update this expression with positions from a token or other expression.
 
         Args:
             other: a token or expression to update this expression with.
+            line: the line number to use if other is None
+            col: column number
+            start: start char index
+            end:  end char index
 
         Returns:
             The updated expression.
         """
-        if isinstance(other, Expression):
-            self.meta.update({k: v for k, v in other.meta.items() if k in POSITION_META_KEYS})
-        elif other is not None:
-            self.meta.update(
-                {
-                    "line": other.line,
-                    "col": other.col,
-                    "start": other.start,
-                    "end": other.end,
-                }
-            )
-        self.meta.update({k: v for k, v in kwargs.items() if k in POSITION_META_KEYS})
+        if other is None:
+            self.meta["line"] = line
+            self.meta["col"] = col
+            self.meta["start"] = start
+            self.meta["end"] = end
+        elif hasattr(other, "meta"):
+            for k in POSITION_META_KEYS:
+                self.meta[k] = other.meta[k]
+        else:
+            self.meta["line"] = other.line
+            self.meta["col"] = other.col
+            self.meta["start"] = other.start
+            self.meta["end"] = other.end
         return self
 
     def as_(
@@ -1248,7 +1263,7 @@ class Query(Expression):
     @property
     def ctes(self) -> t.List[CTE]:
         """Returns a list of all the CTEs attached to this query."""
-        with_ = self.args.get("with")
+        with_ = self.args.get("with_")
         return with_.expressions if with_ else []
 
     @property
@@ -1338,7 +1353,7 @@ class Query(Expression):
         append: bool = True,
         dialect: DialectType = None,
         copy: bool = True,
-        scalar: bool = False,
+        scalar: t.Optional[bool] = None,
         **opts,
     ) -> Q:
         """
@@ -1469,14 +1484,14 @@ class Uncache(Expression):
 
 
 class Refresh(Expression):
-    pass
+    arg_types = {"this": True, "kind": True}
 
 
 class DDL(Expression):
     @property
     def ctes(self) -> t.List[CTE]:
         """Returns a list of all the CTEs attached to this statement."""
-        with_ = self.args.get("with")
+        with_ = self.args.get("with_")
         return with_.expressions if with_ else []
 
     @property
@@ -1537,7 +1552,7 @@ class DML(Expression):
 
 class Create(DDL):
     arg_types = {
-        "with": False,
+        "with_": False,
         "this": True,
         "kind": True,
         "expression": False,
@@ -1616,7 +1631,7 @@ class Detach(Expression):
 
 # https://duckdb.org/docs/sql/statements/load_and_install.html
 class Install(Expression):
-    arg_types = {"this": True, "from": False, "force": False}
+    arg_types = {"this": True, "from_": False, "force": False}
 
 
 # https://duckdb.org/docs/guides/meta/summarize.html
@@ -1654,7 +1669,7 @@ class SetItem(Expression):
         "expressions": False,
         "kind": False,
         "collate": False,  # MySQL SET NAMES statement
-        "global": False,
+        "global_": False,
     }
 
 
@@ -1671,7 +1686,7 @@ class Show(Expression):
         "offset": False,
         "starts_with": False,
         "limit": False,
-        "from": False,
+        "from_": False,
         "like": False,
         "where": False,
         "db": False,
@@ -1681,7 +1696,7 @@ class Show(Expression):
         "mutex": False,
         "query": False,
         "channel": False,
-        "global": False,
+        "global_": False,
         "log": False,
         "position": False,
         "types": False,
@@ -1752,7 +1767,7 @@ class HexString(Condition):
 
 
 class ByteString(Condition):
-    pass
+    arg_types = {"this": True, "is_bytes": False}
 
 
 class RawString(Condition):
@@ -1804,6 +1819,10 @@ class Column(Condition):
         return Dot.build(deepcopy(parts)) if len(parts) > 1 else parts[0]
 
 
+class Pseudocolumn(Column):
+    pass
+
+
 class ColumnPosition(Expression):
     arg_types = {"this": False, "position": True}
 
@@ -1894,7 +1913,13 @@ class Comment(Expression):
 
 
 class Comprehension(Expression):
-    arg_types = {"this": True, "expression": True, "iterator": True, "condition": False}
+    arg_types = {
+        "this": True,
+        "expression": True,
+        "position": False,
+        "iterator": True,
+        "condition": False,
+    }
 
 
 # https://clickhouse.com/docs/en/engines/table-engines/mergetree-family/mergetree#mergetree-table-ttl
@@ -1947,6 +1972,10 @@ class AutoIncrementColumnConstraint(ColumnConstraintKind):
     pass
 
 
+class ZeroFillColumnConstraint(ColumnConstraint):
+    arg_types = {}
+
+
 class PeriodForSystemTimeConstraint(ColumnConstraintKind):
     arg_types = {"this": True, "expression": True}
 
@@ -2111,11 +2140,12 @@ class Constraint(Expression):
 
 class Delete(DML):
     arg_types = {
-        "with": False,
+        "with_": False,
         "this": False,
         "using": False,
         "where": False,
         "returning": False,
+        "order": False,
         "limit": False,
         "tables": False,  # Multiple-Table Syntax (MySQL)
         "cluster": False,  # Clickhouse
@@ -2290,7 +2320,7 @@ class ColumnPrefix(Expression):
 
 
 class PrimaryKey(Expression):
-    arg_types = {"expressions": True, "options": False, "include": False}
+    arg_types = {"this": False, "expressions": True, "options": False, "include": False}
 
 
 # https://www.postgresql.org/docs/9.1/sql-selectinto.html
@@ -2328,7 +2358,7 @@ class JoinHint(Expression):
 
 
 class Identifier(Expression):
-    arg_types = {"this": True, "quoted": False, "global": False, "temporary": False}
+    arg_types = {"this": True, "quoted": False, "global_": False, "temporary": False}
 
     @property
     def quoted(self) -> bool:
@@ -2371,7 +2401,7 @@ class IndexParameters(Expression):
 class Insert(DDL, DML):
     arg_types = {
         "hint": False,
-        "with": False,
+        "with_": False,
         "is_function": False,
         "this": False,
         "expression": False,
@@ -2598,7 +2628,7 @@ class Join(Expression):
         "kind": False,
         "using": False,
         "method": False,
-        "global": False,
+        "global_": False,
         "hint": False,
         "match_condition": False,  # Snowflake
         "expressions": False,
@@ -2777,7 +2807,7 @@ class Order(Expression):
 # https://clickhouse.com/docs/en/sql-reference/statements/select/order-by#order-by-expr-with-fill-modifier
 class WithFill(Expression):
     arg_types = {
-        "from": False,
+        "from_": False,
         "to": False,
         "step": False,
         "interpolate": False,
@@ -2881,7 +2911,7 @@ class DataBlocksizeProperty(Property):
 
 
 class DataDeletionProperty(Property):
-    arg_types = {"on": True, "filter_col": False, "retention_period": False}
+    arg_types = {"on": True, "filter_column": False, "retention_period": False}
 
 
 class DefinerProperty(Property):
@@ -3191,11 +3221,17 @@ class SchemaCommentProperty(Property):
 
 
 class SemanticView(Expression):
-    arg_types = {"this": True, "metrics": False, "dimensions": False, "where": False}
+    arg_types = {
+        "this": True,
+        "metrics": False,
+        "dimensions": False,
+        "facts": False,
+        "where": False,
+    }
 
 
 class SerdeProperties(Property):
-    arg_types = {"expressions": True, "with": False}
+    arg_types = {"expressions": True, "with_": False}
 
 
 class SetProperty(Property):
@@ -3223,7 +3259,7 @@ class SqlReadWriteProperty(Property):
 
 
 class SqlSecurityProperty(Property):
-    arg_types = {"definer": True}
+    arg_types = {"this": True}
 
 
 class StabilityProperty(Property):
@@ -3291,7 +3327,7 @@ class WithSystemVersioningProperty(Property):
         "this": False,
         "data_consistency": False,
         "retention_period": False,
-        "with": True,
+        "with_": True,
     }
 
 
@@ -3501,6 +3537,7 @@ class Table(Expression):
         "changes": False,
         "rows_from": False,
         "sample": False,
+        "indexed": False,
     }
 
     @property
@@ -3559,7 +3596,7 @@ class Table(Expression):
 
 class SetOperation(Query):
     arg_types = {
-        "with": False,
+        "with_": False,
         "this": True,
         "expression": True,
         "distinct": False,
@@ -3587,7 +3624,10 @@ class SetOperation(Query):
 
     @property
     def named_selects(self) -> t.List[str]:
-        return self.this.unnest().named_selects
+        expression = self
+        while isinstance(expression, SetOperation):
+            expression = expression.this.unnest()
+        return expression.named_selects
 
     @property
     def is_star(self) -> bool:
@@ -3595,7 +3635,10 @@ class SetOperation(Query):
 
     @property
     def selects(self) -> t.List[Expression]:
-        return self.this.unnest().selects
+        expression = self
+        while isinstance(expression, SetOperation):
+            expression = expression.this.unnest()
+        return expression.selects
 
     @property
     def left(self) -> Query:
@@ -3628,10 +3671,10 @@ class Intersect(SetOperation):
 
 class Update(DML):
     arg_types = {
-        "with": False,
+        "with_": False,
         "this": False,
-        "expressions": True,
-        "from": False,
+        "expressions": False,
+        "from_": False,
         "where": False,
         "returning": False,
         "order": False,
@@ -3779,7 +3822,7 @@ class Update(DML):
         return _apply_builder(
             expression=expression,
             instance=self,
-            arg="from",
+            arg="from_",
             into=From,
             prefix="FROM",
             dialect=dialect,
@@ -3834,8 +3877,15 @@ class Update(DML):
         )
 
 
+# DuckDB supports VALUES followed by https://duckdb.org/docs/stable/sql/query_syntax/limit
 class Values(UDTF):
-    arg_types = {"expressions": True, "alias": False}
+    arg_types = {
+        "expressions": True,
+        "alias": False,
+        "order": False,
+        "limit": False,
+        "offset": False,
+    }
 
 
 class Var(Expression):
@@ -3868,13 +3918,13 @@ class Lock(Expression):
 
 class Select(Query):
     arg_types = {
-        "with": False,
+        "with_": False,
         "kind": False,
         "expressions": False,
         "hint": False,
         "distinct": False,
         "into": False,
-        "from": False,
+        "from_": False,
         "operation_modifiers": False,
         **QUERY_MODIFIERS,
     }
@@ -3903,7 +3953,7 @@ class Select(Query):
         return _apply_builder(
             expression=expression,
             instance=self,
-            arg="from",
+            arg="from_",
             into=From,
             prefix="FROM",
             dialect=dialect,
@@ -4405,7 +4455,7 @@ class Subquery(DerivedTable, Query):
     arg_types = {
         "this": True,
         "alias": False,
-        "with": False,
+        "with_": False,
         **QUERY_MODIFIERS,
     }
 
@@ -4493,6 +4543,7 @@ class Pivot(Expression):
         "include_nulls": False,
         "default_on_null": False,
         "into": False,
+        "with_": False,
     }
 
     @property
@@ -4542,7 +4593,7 @@ class Where(Expression):
 
 
 class Star(Expression):
-    arg_types = {"except": False, "replace": False, "rename": False}
+    arg_types = {"except_": False, "replace": False, "rename": False}
 
     @property
     def name(self) -> str:
@@ -4614,6 +4665,7 @@ class DataType(Expression):
         SIMPLEAGGREGATEFUNCTION = auto()
         BIGDECIMAL = auto()
         BIGINT = auto()
+        BIGNUM = auto()
         BIGSERIAL = auto()
         BINARY = auto()
         BIT = auto()
@@ -4633,11 +4685,13 @@ class DataType(Expression):
         DECIMAL64 = auto()
         DECIMAL128 = auto()
         DECIMAL256 = auto()
+        DECFLOAT = auto()
         DOUBLE = auto()
         DYNAMIC = auto()
         ENUM = auto()
         ENUM8 = auto()
         ENUM16 = auto()
+        FILE = auto()
         FIXEDSTRING = auto()
         FLOAT = auto()
         GEOGRAPHY = auto()
@@ -4700,6 +4754,7 @@ class DataType(Expression):
         TINYTEXT = auto()
         TIME = auto()
         TIMETZ = auto()
+        TIME_NS = auto()
         TIMESTAMP = auto()
         TIMESTAMPNTZ = auto()
         TIMESTAMPLTZ = auto()
@@ -4734,6 +4789,7 @@ class DataType(Expression):
         TDIGEST = auto()
 
     STRUCT_TYPES = {
+        Type.FILE,
         Type.NESTED,
         Type.OBJECT,
         Type.STRUCT,
@@ -4799,6 +4855,7 @@ class DataType(Expression):
         Type.DECIMAL64,
         Type.DECIMAL128,
         Type.DECIMAL256,
+        Type.DECFLOAT,
         Type.MONEY,
         Type.SMALLMONEY,
         Type.UDECIMAL,
@@ -5082,7 +5139,7 @@ class Connector(Binary):
 
 
 class BitwiseAnd(Binary):
-    pass
+    arg_types = {"this": True, "expression": True, "padside": False}
 
 
 class BitwiseLeftShift(Binary):
@@ -5090,7 +5147,7 @@ class BitwiseLeftShift(Binary):
 
 
 class BitwiseOr(Binary):
-    pass
+    arg_types = {"this": True, "expression": True, "padside": False}
 
 
 class BitwiseRightShift(Binary):
@@ -5098,7 +5155,7 @@ class BitwiseRightShift(Binary):
 
 
 class BitwiseXor(Binary):
-    pass
+    arg_types = {"this": True, "expression": True, "padside": False}
 
 
 class Div(Binary):
@@ -5109,6 +5166,14 @@ class Overlaps(Binary):
     pass
 
 
+class ExtendsLeft(Binary):
+    pass
+
+
+class ExtendsRight(Binary):
+    pass
+
+
 class Dot(Binary):
     @property
     def is_star(self) -> bool:
@@ -5211,6 +5276,10 @@ class Like(Binary, Predicate):
     pass
 
 
+class Match(Binary, Predicate):
+    pass
+
+
 class LT(Binary, Predicate):
     pass
 
@@ -5240,10 +5309,6 @@ class SimilarTo(Binary, Predicate):
     pass
 
 
-class Slice(Binary):
-    arg_types = {"this": False, "expression": False}
-
-
 class Sub(Binary):
     pass
 
@@ -5578,10 +5643,18 @@ class CosineDistance(Func):
     arg_types = {"this": True, "expression": True}
 
 
+class DotProduct(Func):
+    arg_types = {"this": True, "expression": True}
+
+
 class EuclideanDistance(Func):
     arg_types = {"this": True, "expression": True}
 
 
+class ManhattanDistance(Func):
+    arg_types = {"this": True, "expression": True}
+
+
 class JarowinklerSimilarity(Func):
     arg_types = {"this": True, "expression": True}
 
@@ -5602,7 +5675,31 @@ class BitwiseXorAgg(AggFunc):
     pass
 
 
-class BitwiseCountAgg(AggFunc):
+class BoolxorAgg(AggFunc):
+    pass
+
+
+class BitwiseCount(Func):
+    pass
+
+
+class BitmapBucketNumber(Func):
+    pass
+
+
+class BitmapCount(Func):
+    pass
+
+
+class BitmapBitPosition(Func):
+    pass
+
+
+class BitmapConstructAgg(AggFunc):
+    pass
+
+
+class BitmapOrAgg(AggFunc):
     pass
 
 
@@ -5610,6 +5707,18 @@ class ByteLength(Func):
     pass
 
 
+class Boolnot(Func):
+    pass
+
+
+class Booland(Func):
+    arg_types = {"this": True, "expression": True}
+
+
+class Boolor(Func):
+    arg_types = {"this": True, "expression": True}
+
+
 # https://cloud.google.com/bigquery/docs/reference/standard-sql/json_functions#bool_for_json
 class JSONBool(Func):
     pass
@@ -5641,6 +5750,21 @@ class ApproxTopK(AggFunc):
     arg_types = {"this": True, "expression": False, "counters": False}
 
 
+# https://docs.snowflake.com/en/sql-reference/functions/approx_top_k_accumulate
+# https://spark.apache.org/docs/preview/api/sql/index.html#approx_top_k_accumulate
+class ApproxTopKAccumulate(AggFunc):
+    arg_types = {"this": True, "expression": False}
+
+
+# https://docs.snowflake.com/en/sql-reference/functions/approx_top_k_combine
+class ApproxTopKCombine(AggFunc):
+    arg_types = {"this": True, "expression": False}
+
+
+class ApproxTopKEstimate(Func):
+    arg_types = {"this": True, "expression": False}
+
+
 class ApproxTopSum(AggFunc):
     arg_types = {"this": True, "expression": True, "count": True}
 
@@ -5649,6 +5773,27 @@ class ApproxQuantiles(AggFunc):
     arg_types = {"this": True, "expression": False}
 
 
+# https://docs.snowflake.com/en/sql-reference/functions/approx_percentile_combine
+class ApproxPercentileCombine(AggFunc):
+    pass
+
+
+# https://docs.snowflake.com/en/sql-reference/functions/minhash
+class Minhash(AggFunc):
+    arg_types = {"this": True, "expressions": True}
+    is_var_len_args = True
+
+
+# https://docs.snowflake.com/en/sql-reference/functions/minhash_combine
+class MinhashCombine(AggFunc):
+    pass
+
+
+# https://docs.snowflake.com/en/sql-reference/functions/approximate_similarity
+class ApproximateSimilarity(AggFunc):
+    _sql_names = ["APPROXIMATE_SIMILARITY", "APPROXIMATE_JACCARD_INDEX"]
+
+
 class FarmFingerprint(Func):
     arg_types = {"expressions": True}
     is_var_len_args = True
@@ -5669,7 +5814,7 @@ class Transform(Func):
 
 
 class Translate(Func):
-    arg_types = {"this": True, "from": True, "to": True}
+    arg_types = {"this": True, "from_": True, "to": True}
 
 
 class Grouping(AggFunc):
@@ -5677,6 +5822,11 @@ class Grouping(AggFunc):
     is_var_len_args = True
 
 
+class GroupingId(AggFunc):
+    arg_types = {"expressions": True}
+    is_var_len_args = True
+
+
 class Anonymous(Func):
     arg_types = {"this": True, "expressions": False}
     is_var_len_args = True
@@ -5700,6 +5850,12 @@ class CombinedParameterizedAgg(ParameterizedAgg):
     arg_types = {"this": True, "expressions": True, "params": True}
 
 
+# https://docs.snowflake.com/en/sql-reference/functions/hash_agg
+class HashAgg(AggFunc):
+    arg_types = {"this": True, "expressions": False}
+    is_var_len_args = True
+
+
 # https://docs.snowflake.com/en/sql-reference/functions/hll
 # https://docs.aws.amazon.com/redshift/latest/dg/r_HLL_function.html
 class Hll(AggFunc):
@@ -5717,7 +5873,11 @@ class Apply(Func):
 
 
 class Array(Func):
-    arg_types = {"expressions": False, "bracket_notation": False}
+    arg_types = {
+        "expressions": False,
+        "bracket_notation": False,
+        "struct_name_inheritance": False,
+    }
     is_var_len_args = True
 
 
@@ -5730,6 +5890,10 @@ class ToArray(Func):
     pass
 
 
+class ToBoolean(Func):
+    arg_types = {"this": True, "safe": False}
+
+
 # https://materialize.com/docs/sql/types/list/
 class List(Func):
     arg_types = {"expressions": False}
@@ -5765,6 +5929,8 @@ class ToNumber(Func):
         "nlsparam": False,
         "precision": False,
         "scale": False,
+        "safe": False,
+        "safe_name": False,
     }
 
 
@@ -5773,6 +5939,32 @@ class ToDouble(Func):
     arg_types = {
         "this": True,
         "format": False,
+        "safe": False,
+    }
+
+
+# https://docs.snowflake.com/en/sql-reference/functions/to_decfloat
+class ToDecfloat(Func):
+    arg_types = {
+        "this": True,
+        "format": False,
+    }
+
+
+# https://docs.snowflake.com/en/sql-reference/functions/try_to_decfloat
+class TryToDecfloat(Func):
+    arg_types = {
+        "this": True,
+        "format": False,
+    }
+
+
+# https://docs.snowflake.com/en/sql-reference/functions/to_file
+class ToFile(Func):
+    arg_types = {
+        "this": True,
+        "path": False,
+        "safe": False,
     }
 
 
@@ -5786,7 +5978,7 @@ class Columns(Func):
 
 # https://learn.microsoft.com/en-us/sql/t-sql/functions/cast-and-convert-transact-sql?view=sql-server-ver16#syntax
 class Convert(Func):
-    arg_types = {"this": True, "expression": True, "style": False}
+    arg_types = {"this": True, "expression": True, "style": False, "safe": False}
 
 
 # https://docs.oracle.com/en/database/oracle/oracle-database/19/sqlrf/CONVERT.html
@@ -5860,7 +6052,7 @@ class ArrayConcatAgg(AggFunc):
 
 
 class ArrayConstructCompact(Func):
-    arg_types = {"expressions": True}
+    arg_types = {"expressions": False}
     is_var_len_args = True
 
 
@@ -5964,11 +6156,11 @@ class Lead(AggFunc):
 # some dialects have a distinction between first and first_value, usually first is an aggregate func
 # and first_value is a window func
 class First(AggFunc):
-    pass
+    arg_types = {"this": True, "expression": False}
 
 
 class Last(AggFunc):
-    pass
+    arg_types = {"this": True, "expression": False}
 
 
 class FirstValue(AggFunc):
@@ -5983,6 +6175,10 @@ class NthValue(AggFunc):
     arg_types = {"this": True, "offset": True}
 
 
+class ObjectAgg(AggFunc):
+    arg_types = {"this": True, "expression": True}
+
+
 class Case(Func):
     arg_types = {"this": False, "ifs": True, "default": False}
 
@@ -6069,6 +6265,14 @@ class CastToStrType(Func):
     arg_types = {"this": True, "to": True}
 
 
+class CheckJson(Func):
+    arg_types = {"this": True}
+
+
+class CheckXml(Func):
+    arg_types = {"this": True, "disable_auto_convert": False}
+
+
 # https://docs.teradata.com/r/Enterprise_IntelliFlex_VMware/SQL-Functions-Expressions-and-Predicates/String-Operators-and-Functions/TRANSLATE/TRANSLATE-Function-Syntax
 class TranslateCharacters(Expression):
     arg_types = {"this": True, "expression": True, "with_error": False}
@@ -6132,6 +6336,58 @@ class Cbrt(Func):
     pass
 
 
+class CurrentAccount(Func):
+    arg_types = {}
+
+
+class CurrentAccountName(Func):
+    arg_types = {}
+
+
+class CurrentAvailableRoles(Func):
+    arg_types = {}
+
+
+class CurrentClient(Func):
+    arg_types = {}
+
+
+class CurrentIpAddress(Func):
+    arg_types = {}
+
+
+class CurrentDatabase(Func):
+    arg_types = {}
+
+
+class CurrentSchemas(Func):
+    arg_types = {"this": False}
+
+
+class CurrentSecondaryRoles(Func):
+    arg_types = {}
+
+
+class CurrentSession(Func):
+    arg_types = {}
+
+
+class CurrentStatement(Func):
+    arg_types = {}
+
+
+class CurrentVersion(Func):
+    arg_types = {}
+
+
+class CurrentTransaction(Func):
+    arg_types = {}
+
+
+class CurrentWarehouse(Func):
+    arg_types = {}
+
+
 class CurrentDate(Func):
     arg_types = {"this": False}
 
@@ -6144,6 +6400,16 @@ class CurrentTime(Func):
     arg_types = {"this": False}
 
 
+# https://www.postgresql.org/docs/current/functions-datetime.html#FUNCTIONS-DATETIME-CURRENT
+# In Postgres, the difference between CURRENT_TIME vs LOCALTIME etc is that the latter does not have tz
+class Localtime(Func):
+    arg_types = {"this": False}
+
+
+class Localtimestamp(Func):
+    arg_types = {"this": False}
+
+
 class CurrentTimestamp(Func):
     arg_types = {"this": False, "sysdate": False}
 
@@ -6152,6 +6418,18 @@ class CurrentTimestampLTZ(Func):
     arg_types = {}
 
 
+class CurrentTimezone(Func):
+    arg_types = {}
+
+
+class Sysdate(Func):
+    arg_types = {}
+
+
+class CurrentOrganizationName(Func):
+    arg_types = {}
+
+
 class CurrentSchema(Func):
     arg_types = {"this": False}
 
@@ -6160,6 +6438,30 @@ class CurrentUser(Func):
     arg_types = {"this": False}
 
 
+class CurrentCatalog(Func):
+    arg_types = {}
+
+
+class CurrentRegion(Func):
+    arg_types = {}
+
+
+class CurrentRole(Func):
+    arg_types = {}
+
+
+class CurrentRoleType(Func):
+    arg_types = {}
+
+
+class CurrentOrganizationUser(Func):
+    arg_types = {}
+
+
+class SessionUser(Func):
+    arg_types = {}
+
+
 class UtcDate(Func):
     arg_types = {}
 
@@ -6186,7 +6488,14 @@ class DateSub(Func, IntervalOp):
 
 class DateDiff(Func, TimeUnit):
     _sql_names = ["DATEDIFF", "DATE_DIFF"]
-    arg_types = {"this": True, "expression": True, "unit": False, "zone": False}
+    arg_types = {
+        "this": True,
+        "expression": True,
+        "unit": False,
+        "zone": False,
+        "big_int": False,
+        "date_part_boundary": False,
+    }
 
 
 class DateTrunc(Func):
@@ -6266,6 +6575,14 @@ class WeekOfYear(Func):
     _sql_names = ["WEEK_OF_YEAR", "WEEKOFYEAR"]
 
 
+class YearOfWeek(Func):
+    _sql_names = ["YEAR_OF_WEEK", "YEAROFWEEK"]
+
+
+class YearOfWeekIso(Func):
+    _sql_names = ["YEAR_OF_WEEK_ISO", "YEAROFWEEKISO"]
+
+
 class MonthsBetween(Func):
     arg_types = {"this": True, "expression": True, "roundoff": False}
 
@@ -6286,6 +6603,10 @@ class LastDay(Func, TimeUnit):
     arg_types = {"this": True, "unit": False}
 
 
+class PreviousDay(Func):
+    arg_types = {"this": True, "expression": True}
+
+
 class LaxBool(Func):
     pass
 
@@ -6310,8 +6631,13 @@ class Exists(Func, SubqueryPredicate):
     arg_types = {"this": True, "expression": False}
 
 
+class Elt(Func):
+    arg_types = {"this": True, "expressions": True}
+    is_var_len_args = True
+
+
 class Timestamp(Func):
-    arg_types = {"this": False, "zone": False, "with_tz": False}
+    arg_types = {"this": False, "zone": False, "with_tz": False, "safe": False}
 
 
 class TimestampAdd(Func, TimeUnit):
@@ -6331,6 +6657,10 @@ class TimestampTrunc(Func, TimeUnit):
     arg_types = {"this": True, "unit": True, "zone": False}
 
 
+class TimeSlice(Func, TimeUnit):
+    arg_types = {"this": True, "expression": True, "unit": True, "kind": False}
+
+
 class TimeAdd(Func, TimeUnit):
     arg_types = {"this": True, "expression": True, "unit": False}
 
@@ -6349,7 +6679,7 @@ class TimeTrunc(Func, TimeUnit):
 
 class DateFromParts(Func):
     _sql_names = ["DATE_FROM_PARTS", "DATEFROMPARTS"]
-    arg_types = {"year": True, "month": True, "day": True}
+    arg_types = {"year": True, "month": False, "day": False}
 
 
 class TimeFromParts(Func):
@@ -6408,6 +6738,10 @@ class Encode(Func):
     arg_types = {"this": True, "charset": True}
 
 
+class EqualNull(Func):
+    arg_types = {"this": True, "expression": True}
+
+
 class Exp(Func):
     pass
 
@@ -6480,6 +6814,10 @@ class ToBase64(Func):
     pass
 
 
+class ToBinary(Func):
+    arg_types = {"this": True, "format": False, "safe": False}
+
+
 # https://docs.snowflake.com/en/sql-reference/functions/base64_decode_binary
 class Base64DecodeBinary(Func):
     arg_types = {"this": True, "alphabet": False}
@@ -6547,8 +6885,22 @@ class GetExtract(Func):
     arg_types = {"this": True, "expression": True}
 
 
+class Getbit(Func):
+    arg_types = {"this": True, "expression": True}
+
+
 class Greatest(Func):
-    arg_types = {"this": True, "expressions": False}
+    arg_types = {"this": True, "expressions": False, "null_if_any_null": False}
+    is_var_len_args = True
+
+
+class GreatestIgnoreNulls(Func):
+    arg_types = {"expressions": True}
+    is_var_len_args = True
+
+
+class LeastIgnoreNulls(Func):
+    arg_types = {"expressions": True}
     is_var_len_args = True
 
 
@@ -6576,6 +6928,18 @@ class HexEncode(Func):
     arg_types = {"this": True, "case": False}
 
 
+class Hour(Func):
+    pass
+
+
+class Minute(Func):
+    pass
+
+
+class Second(Func):
+    pass
+
+
 # T-SQL: https://learn.microsoft.com/en-us/sql/t-sql/functions/compress-transact-sql?view=sql-server-ver17
 # Snowflake: https://docs.snowflake.com/en/sql-reference/functions/compress
 class Compress(Func):
@@ -6638,9 +7002,13 @@ class IsInf(Func):
     _sql_names = ["IS_INF", "ISINF"]
 
 
+class IsNullValue(Func):
+    pass
+
+
 # https://www.postgresql.org/docs/current/functions-json.html
 class JSON(Expression):
-    arg_types = {"this": False, "with": False, "unique": False}
+    arg_types = {"this": False, "with_": False, "unique": False}
 
 
 class JSONPath(Expression):
@@ -6701,7 +7069,7 @@ class FormatJson(Expression):
 
 
 class Format(Func):
-    arg_types = {"this": True, "expressions": True}
+    arg_types = {"this": True, "expressions": False}
     is_var_len_args = True
 
 
@@ -6761,7 +7129,13 @@ class JSONArrayAgg(Func):
 
 
 class JSONExists(Func):
-    arg_types = {"this": True, "path": True, "passing": False, "on_condition": False}
+    arg_types = {
+        "this": True,
+        "path": True,
+        "passing": False,
+        "on_condition": False,
+        "from_dcolonqmark": False,
+    }
 
 
 # https://docs.oracle.com/en/database/oracle/oracle-database/19/sqlrf/JSON_TABLE.html
@@ -6918,6 +7292,7 @@ class JSONExtractScalar(Binary, Func):
         "only_json_types": False,
         "expressions": False,
         "json_type": False,
+        "scalar_only": False,
     }
     _sql_names = ["JSON_EXTRACT_SCALAR"]
     is_var_len_args = True
@@ -6993,7 +7368,7 @@ class ParseDatetime(Func):
 
 
 class Least(Func):
-    arg_types = {"this": True, "expressions": False}
+    arg_types = {"this": True, "expressions": False, "null_if_any_null": False}
     is_var_len_args = True
 
 
@@ -7081,6 +7456,10 @@ class ScopeResolution(Expression):
     arg_types = {"this": False, "expression": True}
 
 
+class Slice(Expression):
+    arg_types = {"this": False, "expression": False, "step": False}
+
+
 class Stream(Expression):
     pass
 
@@ -7135,6 +7514,10 @@ class Median(AggFunc):
     pass
 
 
+class Mode(AggFunc):
+    arg_types = {"this": False, "deterministic": False}
+
+
 class Min(AggFunc):
     arg_types = {"this": True, "expressions": False}
     is_var_len_args = True
@@ -7144,6 +7527,10 @@ class Month(Func):
     pass
 
 
+class Monthname(Func):
+    pass
+
+
 class AddMonths(Func):
     arg_types = {"this": True, "expression": True}
 
@@ -7160,8 +7547,17 @@ class Normalize(Func):
     arg_types = {"this": True, "form": False, "is_casefold": False}
 
 
+class Normal(Func):
+    arg_types = {"this": True, "stddev": True, "gen": True}
+
+
+# https://cloud.google.com/bigquery/docs/reference/standard-sql/net_functions#nethost
+class NetHost(Func):
+    _sql_names = ["NET.HOST"]
+
+
 class Overlay(Func):
-    arg_types = {"this": True, "expression": True, "from": True, "for": False}
+    arg_types = {"this": True, "expression": True, "from_": True, "for_": False}
 
 
 # https://cloud.google.com/bigquery/docs/reference/standard-sql/bigqueryml-syntax-predict#mlpredict_function
@@ -7242,6 +7638,16 @@ class ApproxQuantile(Quantile):
     }
 
 
+# https://docs.snowflake.com/en/sql-reference/functions/approx_percentile_accumulate
+class ApproxPercentileAccumulate(AggFunc):
+    pass
+
+
+# https://docs.snowflake.com/en/sql-reference/functions/approx_percentile_estimate
+class ApproxPercentileEstimate(Func):
+    arg_types = {"this": True, "percentile": True}
+
+
 class Quarter(Func):
     pass
 
@@ -7257,6 +7663,10 @@ class Randn(Func):
     arg_types = {"this": False}
 
 
+class Randstr(Func):
+    arg_types = {"this": True, "generator": False}
+
+
 class RangeN(Func):
     arg_types = {"this": True, "expressions": True, "each": False}
 
@@ -7276,6 +7686,11 @@ class ReadCSV(Func):
     arg_types = {"this": True, "expressions": False}
 
 
+class ReadParquet(Func):
+    is_var_len_args = True
+    arg_types = {"expressions": True}
+
+
 class Reduce(Func):
     arg_types = {"this": True, "initial": True, "merge": True, "finish": False}
 
@@ -7288,6 +7703,7 @@ class RegexpExtract(Func):
         "occurrence": False,
         "parameters": False,
         "group": False,
+        "null_if_pos_overflow": False,  # for transpilation target behavior
     }
 
 
@@ -7295,10 +7711,10 @@ class RegexpExtractAll(Func):
     arg_types = {
         "this": True,
         "expression": True,
+        "group": False,
+        "parameters": False,
         "position": False,
         "occurrence": False,
-        "parameters": False,
-        "group": False,
     }
 
 
@@ -7310,6 +7726,7 @@ class RegexpReplace(Func):
         "position": False,
         "occurrence": False,
         "modifiers": False,
+        "single_replace": False,
     }
 
 
@@ -7352,6 +7769,50 @@ class RegexpCount(Func):
     }
 
 
+class RegrValx(AggFunc):
+    arg_types = {"this": True, "expression": True}
+
+
+class RegrValy(AggFunc):
+    arg_types = {"this": True, "expression": True}
+
+
+class RegrAvgy(AggFunc):
+    arg_types = {"this": True, "expression": True}
+
+
+class RegrAvgx(AggFunc):
+    arg_types = {"this": True, "expression": True}
+
+
+class RegrCount(AggFunc):
+    arg_types = {"this": True, "expression": True}
+
+
+class RegrIntercept(AggFunc):
+    arg_types = {"this": True, "expression": True}
+
+
+class RegrR2(AggFunc):
+    arg_types = {"this": True, "expression": True}
+
+
+class RegrSxx(AggFunc):
+    arg_types = {"this": True, "expression": True}
+
+
+class RegrSxy(AggFunc):
+    arg_types = {"this": True, "expression": True}
+
+
+class RegrSyy(AggFunc):
+    arg_types = {"this": True, "expression": True}
+
+
+class RegrSlope(AggFunc):
+    arg_types = {"this": True, "expression": True}
+
+
 class Repeat(Func):
     arg_types = {"this": True, "times": True}
 
@@ -7368,7 +7829,12 @@ class Radians(Func):
 # https://learn.microsoft.com/en-us/sql/t-sql/functions/round-transact-sql?view=sql-server-ver16
 # tsql third argument function == trunctaion if not 0
 class Round(Func):
-    arg_types = {"this": True, "decimals": False, "truncate": False}
+    arg_types = {
+        "this": True,
+        "decimals": False,
+        "truncate": False,
+        "casts_non_integer_decimals": False,
+    }
 
 
 class RowNumber(Func):
@@ -7500,6 +7966,11 @@ class Search(Func):
     }
 
 
+# Snowflake: https://docs.snowflake.com/en/sql-reference/functions/search_ip
+class SearchIp(Func):
+    arg_types = {"this": True, "expression": True}
+
+
 class StrToDate(Func):
     arg_types = {"this": True, "format": False, "safe": False}
 
@@ -7660,6 +8131,10 @@ class Unicode(Func):
     pass
 
 
+class Uniform(Func):
+    arg_types = {"this": True, "expression": True, "gen": False, "seed": False}
+
+
 # https://cloud.google.com/bigquery/docs/reference/standard-sql/date_functions#unix_date
 class UnixDate(Func):
     pass
@@ -7712,21 +8187,41 @@ class UnixMillis(Func):
 class Uuid(Func):
     _sql_names = ["UUID", "GEN_RANDOM_UUID", "GENERATE_UUID", "UUID_STRING"]
 
-    arg_types = {"this": False, "name": False}
+    arg_types = {"this": False, "name": False, "is_string": False}
+
+
+TIMESTAMP_PARTS = {
+    "year": False,
+    "month": False,
+    "day": False,
+    "hour": False,
+    "min": False,
+    "sec": False,
+    "nano": False,
+}
 
 
 class TimestampFromParts(Func):
     _sql_names = ["TIMESTAMP_FROM_PARTS", "TIMESTAMPFROMPARTS"]
     arg_types = {
-        "year": True,
-        "month": True,
-        "day": True,
-        "hour": True,
-        "min": True,
-        "sec": True,
-        "nano": False,
+        **TIMESTAMP_PARTS,
         "zone": False,
         "milli": False,
+        "this": False,
+        "expression": False,
+    }
+
+
+class TimestampLtzFromParts(Func):
+    _sql_names = ["TIMESTAMP_LTZ_FROM_PARTS", "TIMESTAMPLTZFROMPARTS"]
+    arg_types = TIMESTAMP_PARTS.copy()
+
+
+class TimestampTzFromParts(Func):
+    _sql_names = ["TIMESTAMP_TZ_FROM_PARTS", "TIMESTAMPTZFROMPARTS"]
+    arg_types = {
+        **TIMESTAMP_PARTS,
+        "zone": False,
     }
 
 
@@ -7752,6 +8247,14 @@ class VariancePop(AggFunc):
     _sql_names = ["VARIANCE_POP", "VAR_POP"]
 
 
+class Skewness(AggFunc):
+    pass
+
+
+class WidthBucket(Func):
+    arg_types = {"this": True, "min_value": True, "max_value": True, "num_buckets": True}
+
+
 class CovarSamp(Binary, AggFunc):
     pass
 
@@ -7768,11 +8271,20 @@ class WeekStart(Expression):
     pass
 
 
+class NextDay(Func):
+    arg_types = {"this": True, "expression": True}
+
+
 class XMLElement(Func):
     _sql_names = ["XMLELEMENT"]
     arg_types = {"this": True, "expressions": False}
 
 
+class XMLGet(Func):
+    _sql_names = ["XMLGET"]
+    arg_types = {"this": True, "expression": True, "instance": False}
+
+
 class XMLTable(Func):
     arg_types = {
         "this": True,
@@ -7796,6 +8308,10 @@ class Year(Func):
     pass
 
 
+class Zipf(Func):
+    arg_types = {"this": True, "elementcount": True, "gen": True}
+
+
 class Use(Expression):
     arg_types = {"this": False, "expressions": False, "kind": False}
 
@@ -7804,9 +8320,10 @@ class Merge(DML):
     arg_types = {
         "this": True,
         "using": True,
-        "on": True,
+        "on": False,
+        "using_cond": False,
         "whens": True,
-        "with": False,
+        "with_": False,
         "returning": False,
     }
 
@@ -7839,10 +8356,10 @@ class TableColumn(Expression):
     pass
 
 
-ALL_FUNCTIONS = subclasses(__name__, Func, (AggFunc, Anonymous, Func))
+ALL_FUNCTIONS = subclasses(__name__, Func, {AggFunc, Anonymous, Func})
 FUNCTION_BY_NAME = {name: func for func in ALL_FUNCTIONS for name in func.sql_names()}
 
-JSON_PATH_PARTS = subclasses(__name__, JSONPathPart, (JSONPathPart,))
+JSON_PATH_PARTS = subclasses(__name__, JSONPathPart, {JSONPathPart})
 
 PERCENTILES = (PercentileCont, PercentileDisc)
 
@@ -8115,7 +8632,7 @@ def _apply_cte_builder(
     append: bool = True,
     dialect: DialectType = None,
     copy: bool = True,
-    scalar: bool = False,
+    scalar: t.Optional[bool] = None,
     **opts,
 ) -> E:
     alias_expression = maybe_parse(alias, dialect=dialect, into=TableAlias, **opts)
@@ -8127,11 +8644,11 @@ def _apply_cte_builder(
     return _apply_child_list_builder(
         cte,
         instance=instance,
-        arg="with",
+        arg="with_",
         append=append,
         copy=copy,
         into=With,
-        properties={"recursive": recursive or False},
+        properties={"recursive": recursive} if recursive else {},
     )
 
 
@@ -8360,7 +8877,7 @@ def update(
         )
     if from_:
         update_expr.set(
-            "from",
+            "from_",
             maybe_parse(from_, into=From, dialect=dialect, prefix="FROM", **opts),
         )
     if isinstance(where, Condition):
@@ -8376,7 +8893,7 @@ def update(
             for alias, qry in with_.items()
         ]
         update_expr.set(
-            "with",
+            "with_",
             With(expressions=cte_list),
         )
     return update_expr
@@ -9308,6 +9825,26 @@ def replace_tree(
     return new_node
 
 
+def find_tables(expression: Expression) -> t.Set[Table]:
+    """
+    Find all tables referenced in a query.
+
+    Args:
+        expressions: The query to find the tables in.
+
+    Returns:
+        A set of all the tables.
+    """
+    from sqlglot.optimizer.scope import traverse_scope
+
+    return {
+        table
+        for scope in traverse_scope(expression)
+        for table in scope.tables
+        if table.name and table.name not in scope.cte_sources
+    }
+
+
 def column_table_names(expression: Expression, exclude: str = "") -> t.Set[str]:
     """
     Return all table names referenced through columns in an expression.