sqlglot 27.20.0__tar.gz → 27.21.0__tar.gz

{sqlglot-27.20.0 → sqlglot-27.21.0}/CHANGELOG.md

@@ -1,6 +1,62 @@
 Changelog
 =========
 
+## [v27.20.0] - 2025-09-30
+### :boom: BREAKING CHANGES
+- due to [`13a30df`](https://github.com/tobymao/sqlglot/commit/13a30dfa37096df5bfc2c31538325c40a49f7917) - Annotate type for snowflake TRY_BASE64_DECODE_BINARY function *(PR [#5972](https://github.com/tobymao/sqlglot/pull/5972) by [@fivetran-amrutabhimsenayachit](https://github.com/fivetran-amrutabhimsenayachit))*:
+
+  Annotate type for snowflake TRY_BASE64_DECODE_BINARY function (#5972)
+
+- due to [`1f5fdd7`](https://github.com/tobymao/sqlglot/commit/1f5fdd799c047de167a4572f7ac26b7ad92167f2) - Annotate type for snowflake TRY_BASE64_DECODE_STRING function *(PR [#5974](https://github.com/tobymao/sqlglot/pull/5974) by [@fivetran-amrutabhimsenayachit](https://github.com/fivetran-amrutabhimsenayachit))*:
+
+  Annotate type for snowflake TRY_BASE64_DECODE_STRING function (#5974)
+
+- due to [`324e82f`](https://github.com/tobymao/sqlglot/commit/324e82fe1fb11722f91341010602a743b151e055) - Annotate type for snowflake TRY_HEX_DECODE_BINARY function *(PR [#5975](https://github.com/tobymao/sqlglot/pull/5975) by [@fivetran-amrutabhimsenayachit](https://github.com/fivetran-amrutabhimsenayachit))*:
+
+  Annotate type for snowflake TRY_HEX_DECODE_BINARY function (#5975)
+
+- due to [`6caf99d`](https://github.com/tobymao/sqlglot/commit/6caf99d556a3357ffaa6c294a9babcd30dd5fac5) - Annotate type for snowflake TRY_HEX_DECODE_STRING function *(PR [#5976](https://github.com/tobymao/sqlglot/pull/5976) by [@fivetran-amrutabhimsenayachit](https://github.com/fivetran-amrutabhimsenayachit))*:
+
+  Annotate type for snowflake TRY_HEX_DECODE_STRING function (#5976)
+
+- due to [`73186a8`](https://github.com/tobymao/sqlglot/commit/73186a812ce422c108ee81b3de11da6ee9a9e902) - annotate type for Snowflake REGEXP_COUNT function *(PR [#5963](https://github.com/tobymao/sqlglot/pull/5963) by [@fivetran-BradfordPaskewitz](https://github.com/fivetran-BradfordPaskewitz))*:
+
+  annotate type for Snowflake REGEXP_COUNT function (#5963)
+
+- due to [`c3bdb3c`](https://github.com/tobymao/sqlglot/commit/c3bdb3cd1af1809ed82be0ae40744d9fffc8ce18) - array start index is 1, support array_flatten, fixes [#5983](https://github.com/tobymao/sqlglot/pull/5983) *(commit by [@georgesittas](https://github.com/georgesittas))*:
+
+  array start index is 1, support array_flatten, fixes #5983
+
+- due to [`244fb48`](https://github.com/tobymao/sqlglot/commit/244fb48fc9c4776f427c08b825d139b1c172fd26) - annotate type for Snowflake SPLIT_PART function *(PR [#5988](https://github.com/tobymao/sqlglot/pull/5988) by [@fivetran-BradfordPaskewitz](https://github.com/fivetran-BradfordPaskewitz))*:
+
+  annotate type for Snowflake SPLIT_PART function (#5988)
+
+- due to [`0d772e0`](https://github.com/tobymao/sqlglot/commit/0d772e0b9d687b24d49203c05d7a90cc1dce02d5) - add ast node for `DIRECTORY` source *(PR [#5990](https://github.com/tobymao/sqlglot/pull/5990) by [@georgesittas](https://github.com/georgesittas))*:
+
+  add ast node for `DIRECTORY` source (#5990)
+
+
+### :sparkles: New Features
+- [`13a30df`](https://github.com/tobymao/sqlglot/commit/13a30dfa37096df5bfc2c31538325c40a49f7917) - **optimizer**: Annotate type for snowflake TRY_BASE64_DECODE_BINARY function *(PR [#5972](https://github.com/tobymao/sqlglot/pull/5972) by [@fivetran-amrutabhimsenayachit](https://github.com/fivetran-amrutabhimsenayachit))*
+- [`1f5fdd7`](https://github.com/tobymao/sqlglot/commit/1f5fdd799c047de167a4572f7ac26b7ad92167f2) - **optimizer**: Annotate type for snowflake TRY_BASE64_DECODE_STRING function *(PR [#5974](https://github.com/tobymao/sqlglot/pull/5974) by [@fivetran-amrutabhimsenayachit](https://github.com/fivetran-amrutabhimsenayachit))*
+- [`324e82f`](https://github.com/tobymao/sqlglot/commit/324e82fe1fb11722f91341010602a743b151e055) - **optimizer**: Annotate type for snowflake TRY_HEX_DECODE_BINARY function *(PR [#5975](https://github.com/tobymao/sqlglot/pull/5975) by [@fivetran-amrutabhimsenayachit](https://github.com/fivetran-amrutabhimsenayachit))*
+- [`6caf99d`](https://github.com/tobymao/sqlglot/commit/6caf99d556a3357ffaa6c294a9babcd30dd5fac5) - **optimizer**: Annotate type for snowflake TRY_HEX_DECODE_STRING function *(PR [#5976](https://github.com/tobymao/sqlglot/pull/5976) by [@fivetran-amrutabhimsenayachit](https://github.com/fivetran-amrutabhimsenayachit))*
+- [`73186a8`](https://github.com/tobymao/sqlglot/commit/73186a812ce422c108ee81b3de11da6ee9a9e902) - **optimizer**: annotate type for Snowflake REGEXP_COUNT function *(PR [#5963](https://github.com/tobymao/sqlglot/pull/5963) by [@fivetran-BradfordPaskewitz](https://github.com/fivetran-BradfordPaskewitz))*
+- [`6124de7`](https://github.com/tobymao/sqlglot/commit/6124de76fa6d6725e844cd37e09ebfe99469b0ec) - **optimizer**: Annotate type for snowflake SOUNDEX function *(PR [#5986](https://github.com/tobymao/sqlglot/pull/5986) by [@fivetran-amrutabhimsenayachit](https://github.com/fivetran-amrutabhimsenayachit))*
+- [`244fb48`](https://github.com/tobymao/sqlglot/commit/244fb48fc9c4776f427c08b825d139b1c172fd26) - **optimizer**: annotate type for Snowflake SPLIT_PART function *(PR [#5988](https://github.com/tobymao/sqlglot/pull/5988) by [@fivetran-BradfordPaskewitz](https://github.com/fivetran-BradfordPaskewitz))*
+- [`0d772e0`](https://github.com/tobymao/sqlglot/commit/0d772e0b9d687b24d49203c05d7a90cc1dce02d5) - **snowflake**: add ast node for `DIRECTORY` source *(PR [#5990](https://github.com/tobymao/sqlglot/pull/5990) by [@georgesittas](https://github.com/georgesittas))*
+
+### :bug: Bug Fixes
+- [`7a3744f`](https://github.com/tobymao/sqlglot/commit/7a3744f203b93211e5dd97e6730b6bf59d6d96e0) - **sqlite**: support `RANGE CURRENT ROW` in window spec *(commit by [@georgesittas](https://github.com/georgesittas))*
+- [`c3bdb3c`](https://github.com/tobymao/sqlglot/commit/c3bdb3cd1af1809ed82be0ae40744d9fffc8ce18) - **starrocks**: array start index is 1, support array_flatten, fixes [#5983](https://github.com/tobymao/sqlglot/pull/5983) *(commit by [@georgesittas](https://github.com/georgesittas))*
+
+### :recycle: Refactors
+- [`d425ba2`](https://github.com/tobymao/sqlglot/commit/d425ba26b96b368801f8f486fa375cd75105993d) - make hash and eq non recursive *(PR [#5966](https://github.com/tobymao/sqlglot/pull/5966) by [@tobymao](https://github.com/tobymao))*
+
+### :wrench: Chores
+- [`345c6a1`](https://github.com/tobymao/sqlglot/commit/345c6a153481a22d6df1b12ef1863e2133688fdf) - add uv support to Makefile *(PR [#5973](https://github.com/tobymao/sqlglot/pull/5973) by [@eakmanrq](https://github.com/eakmanrq))*
+
+
 ## [v27.19.0] - 2025-09-26
 ### :boom: BREAKING CHANGES
 - due to [`68473ac`](https://github.com/tobymao/sqlglot/commit/68473ac3ec8dc76512dc76819892a1b0324c7ddc) - Annotate type for snowflake PARSE_URL function *(PR [#5962](https://github.com/tobymao/sqlglot/pull/5962) by [@fivetran-amrutabhimsenayachit](https://github.com/fivetran-amrutabhimsenayachit))*:
@@ -7561,3 +7617,4 @@ Changelog
 [v27.17.0]: https://github.com/tobymao/sqlglot/compare/v27.16.3...v27.17.0
 [v27.18.0]: https://github.com/tobymao/sqlglot/compare/v27.17.0...v27.18.0
 [v27.19.0]: https://github.com/tobymao/sqlglot/compare/v27.18.0...v27.19.0
+[v27.20.0]: https://github.com/tobymao/sqlglot/compare/v27.19.0...v27.20.0

{sqlglot-27.20.0 → sqlglot-27.21.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sqlglot
-Version: 27.20.0
+Version: 27.21.0
 Summary: An easily customizable SQL parser and transpiler
 Author-email: Toby Mao <toby.mao@gmail.com>
 License-Expression: MIT
@@ -33,7 +33,7 @@ Requires-Dist: typing_extensions; extra == "dev"
 Requires-Dist: maturin<2.0,>=1.4; extra == "dev"
 Requires-Dist: pyperf; extra == "dev"
 Provides-Extra: rs
-Requires-Dist: sqlglotrs==0.6.2; extra == "rs"
+Requires-Dist: sqlglotrs==0.7.0; extra == "rs"
 Dynamic: license-file
 Dynamic: provides-extra
 

{sqlglot-27.20.0 → sqlglot-27.21.0}/sqlglot/_version.py

@@ -28,7 +28,7 @@ version_tuple: VERSION_TUPLE
 commit_id: COMMIT_ID
 __commit_id__: COMMIT_ID
 
-__version__ = version = '27.20.0'
-__version_tuple__ = version_tuple = (27, 20, 0)
+__version__ = version = '27.21.0'
+__version_tuple__ = version_tuple = (27, 21, 0)
 
-__commit_id__ = commit_id = 'g0d772e0b9'
+__commit_id__ = commit_id = 'g5dd2ed3c6'

{sqlglot-27.20.0 → sqlglot-27.21.0}/sqlglot/dialects/bigquery.py

@@ -867,6 +867,8 @@ class BigQuery(Dialect):
             "FROM_HEX": exp.Unhex.from_arg_list,
             "WEEK": lambda args: exp.WeekStart(this=exp.var(seq_get(args, 0))),
         }
+        # Remove SEARCH to avoid parameter routing issues - let it fall back to Anonymous function
+        FUNCTIONS.pop("SEARCH")
 
         FUNCTION_PARSERS = {
             **parser.Parser.FUNCTION_PARSERS,

{sqlglot-27.20.0 → sqlglot-27.21.0}/sqlglot/dialects/dialect.py

@@ -1715,7 +1715,7 @@ def unit_to_str(expression: exp.Expression, default: str = "DAY") -> t.Optional[
 def unit_to_var(expression: exp.Expression, default: str = "DAY") -> t.Optional[exp.Expression]:
     unit = expression.args.get("unit")
 
-    if isinstance(unit, (exp.Var, exp.Placeholder, exp.WeekStart)):
+    if isinstance(unit, (exp.Var, exp.Placeholder, exp.WeekStart, exp.Column)):
         return unit
 
     value = unit.name if unit else default
@@ -1736,7 +1736,9 @@ def map_date_part(
 
 def map_date_part(part, dialect: DialectType = Dialect):
     mapped = (
-        Dialect.get_or_raise(dialect).DATE_PART_MAPPING.get(part.name.upper()) if part else None
+        Dialect.get_or_raise(dialect).DATE_PART_MAPPING.get(part.name.upper())
+        if part and not (isinstance(part, exp.Column) and len(part.parts) != 1)
+        else None
     )
     if mapped:
         return exp.Literal.string(mapped) if part.is_string else exp.var(mapped)

{sqlglot-27.20.0 → sqlglot-27.21.0}/sqlglot/dialects/duckdb.py

@@ -311,6 +311,7 @@ class DuckDB(Dialect):
             "PIVOT_WIDER": TokenType.PIVOT,
             "POSITIONAL": TokenType.POSITIONAL,
             "RESET": TokenType.COMMAND,
+            "ROW": TokenType.STRUCT,
             "SIGNED": TokenType.INT,
             "STRING": TokenType.TEXT,
             "SUMMARIZE": TokenType.SUMMARIZE,
@@ -337,16 +338,14 @@ class DuckDB(Dialect):
     class Parser(parser.Parser):
         MAP_KEYS_ARE_ARBITRARY_EXPRESSIONS = True
 
-        BITWISE = {
-            **parser.Parser.BITWISE,
-            TokenType.TILDA: exp.RegexpLike,
-        }
+        BITWISE = parser.Parser.BITWISE.copy()
         BITWISE.pop(TokenType.CARET)
 
         RANGE_PARSERS = {
             **parser.Parser.RANGE_PARSERS,
             TokenType.DAMP: binary_range_parser(exp.ArrayOverlaps),
             TokenType.CARET_AT: binary_range_parser(exp.StartsWith),
+            TokenType.TILDA: binary_range_parser(exp.RegexpFullMatch),
         }
 
         EXPONENT = {

{sqlglot-27.20.0 → sqlglot-27.21.0}/sqlglot/dialects/hive.py

@@ -531,7 +531,6 @@ class Hive(Dialect):
 
         TRANSFORMS = {
             **generator.Generator.TRANSFORMS,
-            exp.Group: transforms.preprocess([transforms.unalias_group]),
             exp.Property: property_sql,
             exp.AnyValue: rename_func("FIRST"),
             exp.ApproxDistinct: approx_count_distinct_sql,

{sqlglot-27.20.0 → sqlglot-27.21.0}/sqlglot/dialects/oracle.py

@@ -307,7 +307,6 @@ class Oracle(Dialect):
             ),
             exp.DateTrunc: lambda self, e: self.func("TRUNC", e.this, e.unit),
             exp.EuclideanDistance: rename_func("L2_DISTANCE"),
-            exp.Group: transforms.preprocess([transforms.unalias_group]),
             exp.ILike: no_ilike_sql,
             exp.LogicalOr: rename_func("MAX"),
             exp.LogicalAnd: rename_func("MIN"),

{sqlglot-27.20.0 → sqlglot-27.21.0}/sqlglot/dialects/presto.py

@@ -475,7 +475,6 @@ class Presto(Dialect):
             e: f"WITH_TIMEZONE({self.sql(e, 'this')}, {self.sql(e, 'zone')}) AT TIME ZONE 'UTC'",
             exp.GenerateSeries: sequence_sql,
             exp.GenerateDateArray: sequence_sql,
-            exp.Group: transforms.preprocess([transforms.unalias_group]),
             exp.If: if_sql(),
             exp.ILike: no_ilike_sql,
             exp.Initcap: _initcap_sql,

{sqlglot-27.20.0 → sqlglot-27.21.0}/sqlglot/dialects/risingwave.py

@@ -25,6 +25,20 @@ class RisingWave(Postgres):
             "KEY": lambda self: self._parse_encode_property(key=True),
         }
 
+        CONSTRAINT_PARSERS = {
+            **Postgres.Parser.CONSTRAINT_PARSERS,
+            "WATERMARK": lambda self: self.expression(
+                exp.WatermarkColumnConstraint,
+                this=self._match(TokenType.FOR) and self._parse_column(),
+                expression=self._match(TokenType.ALIAS) and self._parse_disjunction(),
+            ),
+        }
+
+        SCHEMA_UNNAMED_CONSTRAINTS = {
+            *Postgres.Parser.SCHEMA_UNNAMED_CONSTRAINTS,
+            "WATERMARK",
+        }
+
         def _parse_table_hints(self) -> t.Optional[t.List[exp.Expression]]:
             # There is no hint in risingwave.
             # Do nothing here to avoid WITH keywords conflict in CREATE SINK statement.

{sqlglot-27.20.0 → sqlglot-27.21.0}/sqlglot/dialects/snowflake.py

@@ -41,7 +41,18 @@ if t.TYPE_CHECKING:
     from sqlglot._typing import E, B
 
 
-# from https://docs.snowflake.com/en/sql-reference/functions/to_timestamp.html
+def _build_strtok(args: t.List) -> exp.SplitPart:
+    # Add default delimiter (space) if missing - per Snowflake docs
+    if len(args) == 1:
+        args.append(exp.Literal.string(" "))
+
+    # Add default part_index (1) if missing
+    if len(args) == 2:
+        args.append(exp.Literal.number(1))
+
+    return exp.SplitPart.from_arg_list(args)
+
+
 def _build_datetime(
     name: str, kind: exp.DataType.Type, safe: bool = False
 ) -> t.Callable[[t.List], exp.Func]:
@@ -137,12 +148,35 @@ def _build_if_from_div0(args: t.List) -> exp.If:
     return exp.If(this=cond, true=true, false=false)
 
 
+# https://docs.snowflake.com/en/sql-reference/functions/div0null
+def _build_if_from_div0null(args: t.List) -> exp.If:
+    lhs = exp._wrap(seq_get(args, 0), exp.Binary)
+    rhs = exp._wrap(seq_get(args, 1), exp.Binary)
+
+    # Returns 0 when divisor is 0 OR NULL
+    cond = exp.EQ(this=rhs, expression=exp.Literal.number(0)).or_(
+        exp.Is(this=rhs, expression=exp.null())
+    )
+    true = exp.Literal.number(0)
+    false = exp.Div(this=lhs, expression=rhs)
+    return exp.If(this=cond, true=true, false=false)
+
+
 # https://docs.snowflake.com/en/sql-reference/functions/zeroifnull
 def _build_if_from_zeroifnull(args: t.List) -> exp.If:
     cond = exp.Is(this=seq_get(args, 0), expression=exp.Null())
     return exp.If(this=cond, true=exp.Literal.number(0), false=seq_get(args, 0))
 
 
+def _build_search(args: t.List) -> exp.Search:
+    kwargs = {
+        "this": seq_get(args, 0),
+        "expression": seq_get(args, 1),
+        **{arg.name.lower(): arg for arg in args[2:] if isinstance(arg, exp.Kwarg)},
+    }
+    return exp.Search(**kwargs)
+
+
 # https://docs.snowflake.com/en/sql-reference/functions/zeroifnull
 def _build_if_from_nullifzero(args: t.List) -> exp.If:
     cond = exp.EQ(this=seq_get(args, 0), expression=exp.Literal.number(0))
@@ -529,6 +563,16 @@ class Snowflake(Dialect):
 
     TYPE_TO_EXPRESSIONS = {
         **Dialect.TYPE_TO_EXPRESSIONS,
+        exp.DataType.Type.DOUBLE: {
+            *Dialect.TYPE_TO_EXPRESSIONS[exp.DataType.Type.DOUBLE],
+            exp.Cos,
+            exp.Cosh,
+            exp.Cot,
+            exp.Degrees,
+            exp.Exp,
+            exp.Sin,
+            exp.Tan,
+        },
         exp.DataType.Type.INT: {
             *Dialect.TYPE_TO_EXPRESSIONS[exp.DataType.Type.INT],
             exp.Ascii,
@@ -539,6 +583,7 @@ class Snowflake(Dialect):
             exp.Levenshtein,
             exp.JarowinklerSimilarity,
             exp.StrPosition,
+            exp.Unicode,
         },
         exp.DataType.Type.VARCHAR: {
             *Dialect.TYPE_TO_EXPRESSIONS[exp.DataType.Type.VARCHAR],
@@ -564,8 +609,10 @@ class Snowflake(Dialect):
             exp.SHA,
             exp.SHA2,
             exp.Soundex,
+            exp.SoundexP123,
             exp.Space,
             exp.SplitPart,
+            exp.Translate,
             exp.Uuid,
         },
         exp.DataType.Type.BINARY: {
@@ -587,6 +634,8 @@ class Snowflake(Dialect):
         },
         exp.DataType.Type.ARRAY: {
             exp.Split,
+            exp.RegexpExtractAll,
+            exp.StringToArray,
         },
         exp.DataType.Type.OBJECT: {
             exp.ParseUrl,
@@ -595,6 +644,10 @@ class Snowflake(Dialect):
         exp.DataType.Type.DECIMAL: {
             exp.RegexpCount,
         },
+        exp.DataType.Type.BOOLEAN: {
+            *Dialect.TYPE_TO_EXPRESSIONS[exp.DataType.Type.BOOLEAN],
+            exp.Search,
+        },
     }
 
     ANNOTATORS = {
@@ -614,11 +667,17 @@ class Snowflake(Dialect):
                 exp.Substring,
             )
         },
+        **{
+            expr_type: lambda self, e: self._annotate_with_type(
+                e, exp.DataType.build("NUMBER", dialect="snowflake")
+            )
+            for expr_type in (
+                exp.RegexpCount,
+                exp.RegexpInstr,
+            )
+        },
         exp.ConcatWs: lambda self, e: self._annotate_by_args(e, "expressions"),
         exp.Reverse: _annotate_reverse,
-        exp.RegexpCount: lambda self, e: self._annotate_with_type(
-            e, exp.DataType.build("NUMBER", dialect="snowflake")
-        ),
     }
 
     TIME_MAPPING = {
@@ -691,7 +750,7 @@ class Snowflake(Dialect):
             "APPROX_PERCENTILE": exp.ApproxQuantile.from_arg_list,
             "ARRAY_CONSTRUCT": lambda args: exp.Array(expressions=args),
             "ARRAY_CONTAINS": lambda args: exp.ArrayContains(
-                this=seq_get(args, 1), expression=seq_get(args, 0)
+                this=seq_get(args, 1), expression=seq_get(args, 0), ensure_variant=False
             ),
             "ARRAY_GENERATE_RANGE": lambda args: exp.GenerateSeries(
                 # ARRAY_GENERATE_RANGE has an exlusive end; we normalize it to be inclusive
@@ -727,6 +786,7 @@ class Snowflake(Dialect):
             "DATEDIFF": _build_datediff,
             "DAYOFWEEKISO": exp.DayOfWeekIso.from_arg_list,
             "DIV0": _build_if_from_div0,
+            "DIV0NULL": _build_if_from_div0null,
             "EDITDISTANCE": lambda args: exp.Levenshtein(
                 this=seq_get(args, 0), expression=seq_get(args, 1), max_dist=seq_get(args, 2)
             ),
@@ -765,6 +825,7 @@ class Snowflake(Dialect):
             "SHA2_BINARY": exp.SHA2Digest.from_arg_list,
             "SHA2_HEX": exp.SHA2.from_arg_list,
             "SQUARE": lambda args: exp.Pow(this=seq_get(args, 0), expression=exp.Literal.number(2)),
+            "STRTOK": _build_strtok,
             "TABLE": lambda args: exp.TableFromRows(this=seq_get(args, 0)),
             "TIMEADD": _build_date_time_add(exp.TimeAdd),
             "TIMEDIFF": _build_datediff,
@@ -799,6 +860,7 @@ class Snowflake(Dialect):
             "ZEROIFNULL": _build_if_from_zeroifnull,
             "LIKE": _build_like(exp.Like),
             "ILIKE": _build_like(exp.ILike),
+            "SEARCH": _build_search,
         }
         FUNCTIONS.pop("PREDICT")
 
@@ -1364,7 +1426,13 @@ class Snowflake(Dialect):
             exp.ArgMax: rename_func("MAX_BY"),
             exp.ArgMin: rename_func("MIN_BY"),
             exp.ArrayConcat: lambda self, e: self.arrayconcat_sql(e, name="ARRAY_CAT"),
-            exp.ArrayContains: lambda self, e: self.func("ARRAY_CONTAINS", e.expression, e.this),
+            exp.ArrayContains: lambda self, e: self.func(
+                "ARRAY_CONTAINS",
+                e.expression
+                if e.args.get("ensure_variant") is False
+                else exp.cast(e.expression, exp.DataType.Type.VARIANT, copy=False),
+                e.this,
+            ),
             exp.ArrayIntersect: rename_func("ARRAY_INTERSECTION"),
             exp.AtTimeZone: lambda self, e: self.func(
                 "CONVERT_TIMEZONE", e.args.get("zone"), e.this
@@ -1894,3 +1962,13 @@ class Snowflake(Dialect):
                 return self.func("TO_CHAR", expression.expressions[0])
 
             return self.function_fallback_sql(expression)
+
+        def splitpart_sql(self, expression: exp.SplitPart) -> str:
+            # Set part_index to 1 if missing
+            if not expression.args.get("delimiter"):
+                expression.set("delimiter", exp.Literal.string(" "))
+
+            if not expression.args.get("part_index"):
+                expression.set("part_index", exp.Literal.number(1))
+
+            return rename_func("SPLIT_PART")(self, expression)

{sqlglot-27.20.0 → sqlglot-27.21.0}/sqlglot/dialects/spark.py

@@ -230,7 +230,6 @@ class Spark(Spark2):
         }
         TRANSFORMS.pop(exp.AnyValue)
         TRANSFORMS.pop(exp.DateDiff)
-        TRANSFORMS.pop(exp.Group)
 
         def bracket_sql(self, expression: exp.Bracket) -> str:
             if expression.args.get("safe"):

{sqlglot-27.20.0 → sqlglot-27.21.0}/sqlglot/expressions.py

@@ -5385,7 +5385,7 @@ class TimeUnit(Expression):
 
     def __init__(self, **args):
         unit = args.get("unit")
-        if type(unit) in self.VAR_LIKE:
+        if type(unit) in self.VAR_LIKE and not (isinstance(unit, Column) and len(unit.parts) != 1):
             args["unit"] = Var(
                 this=(self.UNABBREVIATED_UNIT_NAME.get(unit.name) or unit.name).upper()
             )
@@ -5525,6 +5525,10 @@ class Coth(Func):
     pass
 
 
+class Cos(Func):
+    pass
+
+
 class Csc(Func):
     pass
 
@@ -5549,6 +5553,18 @@ class Sinh(Func):
     pass
 
 
+class Tan(Func):
+    pass
+
+
+class Degrees(Func):
+    pass
+
+
+class Cosh(Func):
+    pass
+
+
 class CosineDistance(Func):
     arg_types = {"this": True, "expression": True}
 
@@ -5840,6 +5856,7 @@ class ArrayConstructCompact(Func):
 
 
 class ArrayContains(Binary, Func):
+    arg_types = {"this": True, "expression": True, "ensure_variant": False}
     _sql_names = ["ARRAY_CONTAINS", "ARRAY_HAS"]
 
 
@@ -6172,7 +6189,9 @@ class DateTrunc(Func):
         unabbreviate = args.pop("unabbreviate", True)
 
         unit = args.get("unit")
-        if isinstance(unit, TimeUnit.VAR_LIKE):
+        if isinstance(unit, TimeUnit.VAR_LIKE) and not (
+            isinstance(unit, Column) and len(unit.parts) != 1
+        ):
             unit_name = unit.name.upper()
             if unabbreviate and unit_name in TimeUnit.UNABBREVIATED_UNIT_NAME:
                 unit_name = TimeUnit.UNABBREVIATED_UNIT_NAME[unit_name]
@@ -7279,6 +7298,10 @@ class RegexpILike(Binary, Func):
     arg_types = {"this": True, "expression": True, "flag": False}
 
 
+class RegexpFullMatch(Binary, Func):
+    arg_types = {"this": True, "expression": True, "options": False}
+
+
 class RegexpInstr(Func):
     arg_types = {
         "this": True,
@@ -7380,13 +7403,20 @@ class Soundex(Func):
     pass
 
 
+# https://docs.snowflake.com/en/sql-reference/functions/soundex_p123
+class SoundexP123(Func):
+    pass
+
+
 class Split(Func):
     arg_types = {"this": True, "expression": True, "limit": False}
 
 
 # https://spark.apache.org/docs/latest/api/python/reference/pyspark.sql/api/pyspark.sql.functions.split_part.html
+# https://docs.snowflake.com/en/sql-reference/functions/split_part
+# https://docs.snowflake.com/en/sql-reference/functions/strtok
 class SplitPart(Func):
-    arg_types = {"this": True, "delimiter": True, "part_index": True}
+    arg_types = {"this": True, "delimiter": False, "part_index": False}
 
 
 # Start may be omitted in the case of postgres
@@ -7430,6 +7460,19 @@ class StrPosition(Func):
     }
 
 
+# Snowflake: https://docs.snowflake.com/en/sql-reference/functions/search
+# BigQuery: https://cloud.google.com/bigquery/docs/reference/standard-sql/search_functions#search
+class Search(Func):
+    arg_types = {
+        "this": True,  # data_to_search / search_data
+        "expression": True,  # search_query / search_string
+        "json_scope": False,  # BigQuery: JSON_VALUES | JSON_KEYS | JSON_KEYS_AND_VALUES
+        "analyzer": False,  # Both: analyzer / ANALYZER
+        "analyzer_options": False,  # BigQuery: analyzer_options_values
+        "search_mode": False,  # Snowflake: OR | AND
+    }
+
+
 class StrToDate(Func):
     arg_types = {"this": True, "format": False, "safe": False}
 

{sqlglot-27.20.0 → sqlglot-27.21.0}/sqlglot/optimizer/canonicalize.py

@@ -77,7 +77,7 @@ def coerce_type(node: exp.Expression, promote_to_inferred_datetime_type: bool) -
         _coerce_date(node.left, node.right, promote_to_inferred_datetime_type)
     elif isinstance(node, exp.Between):
         _coerce_date(node.this, node.args["low"], promote_to_inferred_datetime_type)
-    elif isinstance(node, exp.Extract) and not node.expression.type.is_type(
+    elif isinstance(node, exp.Extract) and not node.expression.is_type(
         *exp.DataType.TEMPORAL_TYPES
     ):
         _replace_cast(node.expression, exp.DataType.Type.DATETIME)

{sqlglot-27.20.0 → sqlglot-27.21.0}/sqlglot/optimizer/merge_subqueries.py

@@ -201,6 +201,7 @@ def _mergeable(
         and not outer_scope.pivots
         and not any(e.find(exp.AggFunc, exp.Select, exp.Explode) for e in inner_select.expressions)
         and not (leave_tables_isolated and len(outer_scope.selected_sources) > 1)
+        and not (isinstance(from_or_join, exp.Join) and inner_select.args.get("joins"))
         and not (
             isinstance(from_or_join, exp.Join)
             and inner_select.args.get("where")
@@ -282,6 +283,7 @@ def _merge_joins(outer_scope: Scope, inner_scope: Scope, from_or_join: FromOrJoi
     new_joins = []
 
     joins = inner_scope.expression.args.get("joins") or []
+
     for join in joins:
         new_joins.append(join)
         outer_scope.add_source(join.alias_or_name, inner_scope.sources[join.alias_or_name])

{sqlglot-27.20.0 → sqlglot-27.21.0}/sqlglot/parser.py

@@ -1141,11 +1141,6 @@ class Parser(metaclass=_Parser):
         "TTL": lambda self: self.expression(exp.MergeTreeTTL, expressions=[self._parse_bitwise()]),
         "UNIQUE": lambda self: self._parse_unique(),
         "UPPERCASE": lambda self: self.expression(exp.UppercaseColumnConstraint),
-        "WATERMARK": lambda self: self.expression(
-            exp.WatermarkColumnConstraint,
-            this=self._match(TokenType.FOR) and self._parse_column(),
-            expression=self._match(TokenType.ALIAS) and self._parse_disjunction(),
-        ),
         "WITH": lambda self: self.expression(
             exp.Properties, expressions=self._parse_wrapped_properties()
         ),
@@ -1211,7 +1206,6 @@ class Parser(metaclass=_Parser):
         "PERIOD",
         "PRIMARY KEY",
         "UNIQUE",
-        "WATERMARK",
         "BUCKET",
         "TRUNCATE",
     }
@@ -4592,14 +4586,10 @@ class Parser(metaclass=_Parser):
             before_with_index = self._index
             with_prefix = self._match(TokenType.WITH)
 
-            if self._match(TokenType.ROLLUP):
-                elements["rollup"].append(
-                    self._parse_cube_or_rollup(exp.Rollup, with_prefix=with_prefix)
-                )
-            elif self._match(TokenType.CUBE):
-                elements["cube"].append(
-                    self._parse_cube_or_rollup(exp.Cube, with_prefix=with_prefix)
-                )
+            cube_or_rollup = self._parse_cube_or_rollup(with_prefix=with_prefix)
+            if cube_or_rollup:
+                key = "rollup" if isinstance(cube_or_rollup, exp.Rollup) else "cube"
+                elements[key].append(cube_or_rollup)
             elif self._match(TokenType.GROUPING_SETS):
                 elements["grouping_sets"].append(
                     self.expression(
@@ -4619,18 +4609,20 @@ class Parser(metaclass=_Parser):
 
         return self.expression(exp.Group, comments=comments, **elements)  # type: ignore
 
-    def _parse_cube_or_rollup(self, kind: t.Type[E], with_prefix: bool = False) -> E:
+    def _parse_cube_or_rollup(self, with_prefix: bool = False) -> t.Optional[exp.Cube | exp.Rollup]:
+        if self._match(TokenType.CUBE):
+            kind: t.Type[exp.Cube | exp.Rollup] = exp.Cube
+        elif self._match(TokenType.ROLLUP):
+            kind = exp.Rollup
+        else:
+            return None
+
         return self.expression(
             kind, expressions=[] if with_prefix else self._parse_wrapped_csv(self._parse_column)
         )
 
     def _parse_grouping_set(self) -> t.Optional[exp.Expression]:
-        if self._match(TokenType.L_PAREN):
-            grouping_set = self._parse_csv(self._parse_bitwise)
-            self._match_r_paren()
-            return self.expression(exp.Tuple, expressions=grouping_set)
-
-        return self._parse_column()
+        return self._parse_cube_or_rollup() or self._parse_bitwise()
 
     def _parse_having(self, skip_having_token: bool = False) -> t.Optional[exp.Having]:
         if not skip_having_token and not self._match(TokenType.HAVING):
@@ -4749,11 +4741,15 @@ class Parser(metaclass=_Parser):
             exp.Ordered, this=this, desc=desc, nulls_first=nulls_first, with_fill=with_fill
         )
 
-    def _parse_limit_options(self) -> exp.LimitOptions:
-        percent = self._match(TokenType.PERCENT)
+    def _parse_limit_options(self) -> t.Optional[exp.LimitOptions]:
+        percent = self._match_set((TokenType.PERCENT, TokenType.MOD))
         rows = self._match_set((TokenType.ROW, TokenType.ROWS))
         self._match_text_seq("ONLY")
         with_ties = self._match_text_seq("WITH", "TIES")
+
+        if not (percent or rows or with_ties):
+            return None
+
         return self.expression(exp.LimitOptions, percent=percent, rows=rows, with_ties=with_ties)
 
     def _parse_limit(
@@ -4771,10 +4767,13 @@ class Parser(metaclass=_Parser):
                 if limit_paren:
                     self._match_r_paren()
 
-                limit_options = self._parse_limit_options()
             else:
-                limit_options = None
-                expression = self._parse_term()
+                # Parsing LIMIT x% (i.e x PERCENT) as a term leads to an error, since
+                # we try to build an exp.Mod expr. For that matter, we backtrack and instead
+                # consume the factor plus parse the percentage separately
+                expression = self._try_parse(self._parse_term) or self._parse_factor()
+
+            limit_options = self._parse_limit_options()
 
             if self._match(TokenType.COMMA):
                 offset = expression

{sqlglot-27.20.0 → sqlglot-27.21.0}/sqlglot/tokens.py

@@ -1421,7 +1421,11 @@ class Tokenizer(metaclass=_Tokenizer):
                         raise_unmatched=not self.HEREDOC_TAG_IS_IDENTIFIER,
                     )
 
-                if tag and self.HEREDOC_TAG_IS_IDENTIFIER and (self._end or not tag.isidentifier()):
+                if (
+                    tag
+                    and self.HEREDOC_TAG_IS_IDENTIFIER
+                    and (self._end or tag.isdigit() or any(c.isspace() for c in tag))
+                ):
                     if not self._end:
                         self._advance(-1)