sqlglot 30.20.0__tar.gz → 30.21.0__tar.gz

{sqlglot-30.20.0 → sqlglot-30.21.0}/CHANGELOG.md

@@ -1,6 +1,74 @@
 Changelog
 =========
 
+## [v30.20.0] - 2026-09-27
+### :boom: BREAKING CHANGES
+- due to [`1d0e3f3`](https://github.com/tobymao/sqlglot/commit/1d0e3f34864037c54cf6bdf45af594db62ad7319) - Fix mysql roundtrip for DATE_ADD and DATE_SUB *(PR [#8417](https://github.com/tobymao/sqlglot/pull/8417) by [@fivetran-amrutabhimsenayachit](https://github.com/fivetran-amrutabhimsenayachit))*:
+
+  Fix mysql roundtrip for DATE_ADD and DATE_SUB (#8417)
+
+- due to [`284ea43`](https://github.com/tobymao/sqlglot/commit/284ea43fe773762a1aa1556171bbb4c03a197788) - support IGNORE NULLS and RESPECT NULLS in window functions [CLAUDE] *(PR [#8412](https://github.com/tobymao/sqlglot/pull/8412) by [@UCHPUCHMAK](https://github.com/UCHPUCHMAK))*:
+
+  support IGNORE NULLS and RESPECT NULLS in window functions [CLAUDE] (#8412)
+
+- due to [`abacf31`](https://github.com/tobymao/sqlglot/commit/abacf31d7b12b3bd70cd4a70b25cfcd2ad6c1f0a) - use row-preserving default projection for windowed aggregates *(PR [#8381](https://github.com/tobymao/sqlglot/pull/8381) by [@treysp](https://github.com/treysp))*:
+
+  use row-preserving default projection for windowed aggregates (#8381)
+
+- due to [`a388416`](https://github.com/tobymao/sqlglot/commit/a3884168f639633fa8f64334bf63e4b343bedfe2) - retain json <string> literals closes [#8413](https://github.com/tobymao/sqlglot/pull/8413) *(commit by [@georgesittas](https://github.com/georgesittas))*:
+
+  retain json <string> literals closes #8413
+
+- due to [`20fa8ae`](https://github.com/tobymao/sqlglot/commit/20fa8aedc48fccb780b973987908db63d0f76895) - keep `TRY_TO_NUMBER` safe when it falls back to a cast [CLAUDE] *(PR [#8424](https://github.com/tobymao/sqlglot/pull/8424) by [@anishmehta24](https://github.com/anishmehta24))*:
+
+  keep `TRY_TO_NUMBER` safe when it falls back to a cast [CLAUDE] (#8424)
+
+- due to [`b85a8c8`](https://github.com/tobymao/sqlglot/commit/b85a8c8b0d107410f6689c2ddad0921dbff9ddad) - exponential AST growth prevention in merge_subqueries *(PR [#8427](https://github.com/tobymao/sqlglot/pull/8427) by [@georgesittas](https://github.com/georgesittas))*:
+
+  exponential AST growth prevention in merge_subqueries (#8427)
+
+- due to [`4c11d07`](https://github.com/tobymao/sqlglot/commit/4c11d07af4b533274cabbcae9271234e9a0b039c) - avoid AUTOINCREMENT without a primary key *(PR [#8431](https://github.com/tobymao/sqlglot/pull/8431) by [@georgesittas](https://github.com/georgesittas))*:
+
+  avoid AUTOINCREMENT without a primary key (#8431)
+
+- due to [`d2fbb21`](https://github.com/tobymao/sqlglot/commit/d2fbb215ff6bd548547a151e8838c277aad076c1) - preserve TIMESTAMP_FROM_PARTS time zones ([#8422](https://github.com/tobymao/sqlglot/pull/8422)) *(PR [#8434](https://github.com/tobymao/sqlglot/pull/8434) by [@georgesittas](https://github.com/georgesittas))*:
+
+  preserve TIMESTAMP_FROM_PARTS time zones (#8422) (#8434)
+
+- due to [`afd6b69`](https://github.com/tobymao/sqlglot/commit/afd6b6946267b0b1100f48b299f3d44ff9a13767) - annotate `bit_or` for duckdb *(PR [#8438](https://github.com/tobymao/sqlglot/pull/8438) by [@deepika-kakde](https://github.com/deepika-kakde))*:
+
+  annotate `bit_or` for duckdb (#8438)
+
+- due to [`8bd7686`](https://github.com/tobymao/sqlglot/commit/8bd768664197fbd0ed86158c0eca955c14f43328) - annotate `bit_and` for duckdb *(PR [#8439](https://github.com/tobymao/sqlglot/pull/8439) by [@Anuja006](https://github.com/Anuja006))*:
+
+  annotate `bit_and` for duckdb (#8439)
+
+
+### :sparkles: New Features
+- [`a313c8b`](https://github.com/tobymao/sqlglot/commit/a313c8bee143b4b92c9a068fef72a5fd512bba61) - **parser**: support boolean expressions in update assignments closes [#8428](https://github.com/tobymao/sqlglot/pull/8428) *(commit by [@georgesittas](https://github.com/georgesittas))*
+- [`afd6b69`](https://github.com/tobymao/sqlglot/commit/afd6b6946267b0b1100f48b299f3d44ff9a13767) - **optimizer**: annotate `bit_or` for duckdb *(PR [#8438](https://github.com/tobymao/sqlglot/pull/8438) by [@deepika-kakde](https://github.com/deepika-kakde))*
+- [`8bd7686`](https://github.com/tobymao/sqlglot/commit/8bd768664197fbd0ed86158c0eca955c14f43328) - **optimizer**: annotate `bit_and` for duckdb *(PR [#8439](https://github.com/tobymao/sqlglot/pull/8439) by [@Anuja006](https://github.com/Anuja006))*
+
+### :bug: Bug Fixes
+- [`1d0e3f3`](https://github.com/tobymao/sqlglot/commit/1d0e3f34864037c54cf6bdf45af594db62ad7319) - **mysql**: Fix mysql roundtrip for DATE_ADD and DATE_SUB *(PR [#8417](https://github.com/tobymao/sqlglot/pull/8417) by [@fivetran-amrutabhimsenayachit](https://github.com/fivetran-amrutabhimsenayachit))*
+  - :arrow_lower_right: *fixes issue [#8392](https://github.com/tobymao/sqlglot/issues/8392) opened by [@linchangyang-sg](https://github.com/linchangyang-sg)*
+- [`284ea43`](https://github.com/tobymao/sqlglot/commit/284ea43fe773762a1aa1556171bbb4c03a197788) - **starrocks**: support IGNORE NULLS and RESPECT NULLS in window functions [CLAUDE] *(PR [#8412](https://github.com/tobymao/sqlglot/pull/8412) by [@UCHPUCHMAK](https://github.com/UCHPUCHMAK))*
+- [`abacf31`](https://github.com/tobymao/sqlglot/commit/abacf31d7b12b3bd70cd4a70b25cfcd2ad6c1f0a) - **optimizer**: use row-preserving default projection for windowed aggregates *(PR [#8381](https://github.com/tobymao/sqlglot/pull/8381) by [@treysp](https://github.com/treysp))*
+- [`a388416`](https://github.com/tobymao/sqlglot/commit/a3884168f639633fa8f64334bf63e4b343bedfe2) - retain json <string> literals closes [#8413](https://github.com/tobymao/sqlglot/pull/8413) *(commit by [@georgesittas](https://github.com/georgesittas))*
+- [`32a9297`](https://github.com/tobymao/sqlglot/commit/32a9297a24f06035293a2a3e935a7b65ece25755) - **optimizer**: avoid accumulating parentheses when merging repeated references *(PR [#8425](https://github.com/tobymao/sqlglot/pull/8425) by [@georgesittas](https://github.com/georgesittas))*
+- [`20fa8ae`](https://github.com/tobymao/sqlglot/commit/20fa8aedc48fccb780b973987908db63d0f76895) - keep `TRY_TO_NUMBER` safe when it falls back to a cast [CLAUDE] *(PR [#8424](https://github.com/tobymao/sqlglot/pull/8424) by [@anishmehta24](https://github.com/anishmehta24))*
+- [`b85a8c8`](https://github.com/tobymao/sqlglot/commit/b85a8c8b0d107410f6689c2ddad0921dbff9ddad) - **optimizer**: exponential AST growth prevention in merge_subqueries *(PR [#8427](https://github.com/tobymao/sqlglot/pull/8427) by [@georgesittas](https://github.com/georgesittas))*
+- [`8d9f5d0`](https://github.com/tobymao/sqlglot/commit/8d9f5d08cce4b72b4df1a8f6394c2aa043cc7de8) - **parser**: handle singular ROW in FETCH and OFFSET clauses closes [#8429](https://github.com/tobymao/sqlglot/pull/8429) *(commit by [@georgesittas](https://github.com/georgesittas))*
+- [`981e492`](https://github.com/tobymao/sqlglot/commit/981e492413740f7740d2b0288e3da5b4028c767e) - **clickhouse**: preserve GLOBAL JOIN after ON conditions fixes [#8426](https://github.com/tobymao/sqlglot/pull/8426) *(commit by [@georgesittas](https://github.com/georgesittas))*
+- [`4c11d07`](https://github.com/tobymao/sqlglot/commit/4c11d07af4b533274cabbcae9271234e9a0b039c) - **sqlite**: avoid AUTOINCREMENT without a primary key *(PR [#8431](https://github.com/tobymao/sqlglot/pull/8431) by [@georgesittas](https://github.com/georgesittas))*
+  - :arrow_lower_right: *fixes issue [#8423](https://github.com/tobymao/sqlglot/issues/8423) opened by [@Rodrigo-Palma](https://github.com/Rodrigo-Palma)*
+- [`87b1402`](https://github.com/tobymao/sqlglot/commit/87b14027372c3820cbf6c89bc8adc2cbf8963404) - **optimizer**: preserve grouping when complementing comparisons *(PR [#8433](https://github.com/tobymao/sqlglot/pull/8433) by [@georgesittas](https://github.com/georgesittas))*
+  - :arrow_lower_right: *fixes issue [#8421](https://github.com/tobymao/sqlglot/issues/8421) opened by [@Rodrigo-Palma](https://github.com/Rodrigo-Palma)*
+- [`d2fbb21`](https://github.com/tobymao/sqlglot/commit/d2fbb215ff6bd548547a151e8838c277aad076c1) - **duckdb**: preserve TIMESTAMP_FROM_PARTS time zones ([#8422](https://github.com/tobymao/sqlglot/pull/8422)) *(PR [#8434](https://github.com/tobymao/sqlglot/pull/8434) by [@georgesittas](https://github.com/georgesittas))*
+  - :arrow_lower_right: *fixes issue [#8422](https://github.com/tobymao/sqlglot/issues/8422) opened by [@Rodrigo-Palma](https://github.com/Rodrigo-Palma)*
+- [`d01e946`](https://github.com/tobymao/sqlglot/commit/d01e9461a7a3fcbe50a965c7a2ddf55d41aca97d) - **optimizer**: preserve ordinal ordering with duplicate aliases [CODEX] *(PR [#8436](https://github.com/tobymao/sqlglot/pull/8436) by [@tobymao](https://github.com/tobymao))*
+
+
 ## [v30.19.0] - 2026-09-22
 ### :boom: BREAKING CHANGES
 - due to [`bbb717e`](https://github.com/tobymao/sqlglot/commit/bbb717ebb24c8dcbee94d77c9154e74da1dc6c10) - rename union scope type to set operation for consistency *(PR [#8306](https://github.com/tobymao/sqlglot/pull/8306) by [@georgesittas](https://github.com/georgesittas))*:
@@ -15921,3 +15989,4 @@ pip install "sqlglot[c]"   # compiled — faster, but no subclassing
 [v30.17.0]: https://github.com/tobymao/sqlglot/compare/v30.16.0...v30.17.0
 [v30.18.0]: https://github.com/tobymao/sqlglot/compare/v30.17.0...v30.18.0
 [v30.19.0]: https://github.com/tobymao/sqlglot/compare/v30.18.0...v30.19.0
+[v30.20.0]: https://github.com/tobymao/sqlglot/compare/v30.19.0...v30.20.0

{sqlglot-30.20.0/sqlglot.egg-info → sqlglot-30.21.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sqlglot
-Version: 30.20.0
+Version: 30.21.0
 Summary: An easily customizable SQL parser and transpiler
 Author-email: Toby Mao <toby.mao@gmail.com>
 License-Expression: MIT
@@ -34,10 +34,10 @@ Requires-Dist: types-pytz; extra == "dev"
 Requires-Dist: typing_extensions; extra == "dev"
 Requires-Dist: pyperf; extra == "dev"
 Provides-Extra: c
-Requires-Dist: sqlglotc==30.20.0; python_version >= "3.10" and extra == "c"
+Requires-Dist: sqlglotc==30.21.0; python_version >= "3.10" and extra == "c"
 Provides-Extra: rs
 Requires-Dist: sqlglotrs==0.13.0; extra == "rs"
-Requires-Dist: sqlglotc==30.20.0; python_version >= "3.10" and extra == "rs"
+Requires-Dist: sqlglotc==30.21.0; python_version >= "3.10" and extra == "rs"
 Dynamic: license-file
 Dynamic: provides-extra
 

{sqlglot-30.20.0 → sqlglot-30.21.0}/sqlglot/_version.py

@@ -18,7 +18,7 @@ version_tuple: tuple[int | str, ...]
 commit_id: str | None
 __commit_id__: str | None
 
-__version__ = version = '30.20.0'
-__version_tuple__ = version_tuple = (30, 20, 0)
+__version__ = version = '30.21.0'
+__version_tuple__ = version_tuple = (30, 21, 0)
 
-__commit_id__ = commit_id = 'g8bd768664'
+__commit_id__ = commit_id = 'g5d16fa83b'

{sqlglot-30.20.0 → sqlglot-30.21.0}/sqlglot/dialects/athena.py

@@ -2,7 +2,7 @@ from __future__ import annotations
 
 
 from sqlglot import tokens
-from sqlglot.dialects.dialect import Dialect, DialectType
+from sqlglot.dialects.dialect import DialectType
 from sqlglot.generators.athena import AthenaGenerator
 from sqlglot.parsers.athena import AthenaParser
 from sqlglot.tokens import TokenType, Token
@@ -10,7 +10,7 @@ from sqlglot.dialects.trino import Trino
 from sqlglot.dialects.hive import Hive
 
 
-class Athena(Dialect):
+class Athena(Trino):
     """
     Over the years, it looks like AWS has taken various execution engines, bolted on AWS-specific
     modifications and then built the Athena service around them.

{sqlglot-30.20.0 → sqlglot-30.21.0}/sqlglot/dialects/dialect.py

@@ -346,6 +346,24 @@ class _Dialect(type):
             if not v.isprintable() or v == "\\"
         }
 
+        # Prefer hex, then octal, then Unicode for numeric control-character escapes.
+        numeric_escapes = klass.tokenizer_class.NUMERIC_ESCAPES
+        if "x" in numeric_escapes:
+            fmt = "\\x{:02x}"
+        elif "0" in numeric_escapes:
+            fmt = "\\{:03o}"
+        elif "u" in numeric_escapes:
+            fmt = "\\u{:04x}"
+        else:
+            fmt = ""
+
+        if fmt:
+            # ASCII controls are 0-31 and DEL (127). Preserve existing escape mappings and
+            # leave tab (9), newline (10), and carriage return (13) to their existing handling.
+            for code in (*range(32), 127):
+                if code not in (9, 10, 13):
+                    klass.ESCAPED_SEQUENCES.setdefault(chr(code), fmt.format(code))
+
         klass.SUPPORTS_COLUMN_JOIN_MARKS = "(+)" in klass.tokenizer_class.KEYWORDS
 
         if enum not in ("", "bigquery", "snowflake"):

{sqlglot-30.20.0 → sqlglot-30.21.0}/sqlglot/dialects/snowflake.py

@@ -136,6 +136,8 @@ class Snowflake(Dialect):
 
     class Tokenizer(tokens.Tokenizer):
         STRING_ESCAPES = ["\\", "'"]
+        NUMERIC_ESCAPES = {"x": (16, 2, 2, 0xFF), "u": (16, 4, 4, 0xFFFF), "0": (8, 1, 3, 0xFF)}
+        DROP_UNKNOWN_ESCAPES = True
         HEX_STRINGS = [("x'", "'"), ("X'", "'")]
         RAW_STRINGS = ["$$"]
         COMMENTS = ["--", "//", ("/*", "*/")]

{sqlglot-30.20.0 → sqlglot-30.21.0}/sqlglot/expressions/query.py

@@ -804,6 +804,19 @@ class Lateral(Expression, UDTF):
         "ordinality": False,
     }
 
+    @property
+    def selects(self) -> list[Expr]:
+        from sqlglot.expressions.array import Unnest
+
+        columns = super().selects
+
+        # UNNEST ... WITH ORDINALITY stores the ordinality column's name in Unnest.offset
+        offset = self.this.args.get("offset") if isinstance(self.this, Unnest) else None
+        if isinstance(offset, Identifier):
+            columns = columns + [offset]
+
+        return columns
+
 
 class TableFromRows(Expression, UDTF):
     arg_types = {

{sqlglot-30.20.0 → sqlglot-30.21.0}/sqlglot/generator.py

@@ -2959,7 +2959,20 @@ class Generator:
             op_sql = self.seg(f"LATERAL VIEW{' OUTER' if expression.args.get('outer') else ''}")
             return f"{op_sql}{self.sep()}{this}{table}{columns}"
 
-        alias = self.sql(expression, "alias")
+        table_alias = expression.args.get("alias")
+        offset = expression.this.args.get("offset")
+
+        if (
+            self.UNNEST_WITH_ORDINALITY
+            and table_alias
+            and isinstance(expression.this, exp.Unnest)
+            and isinstance(offset, exp.Identifier)
+        ):
+            # UNNEST ... WITH ORDINALITY stores the ordinality column's name in Unnest.offset
+            table_alias = table_alias.copy()
+            table_alias.append("columns", offset.copy())
+
+        alias = self.sql(table_alias)
         alias = f" AS {alias}" if alias else ""
 
         ordinality = expression.args.get("ordinality") or ""
@@ -5309,6 +5322,28 @@ class Generator:
             sqlglot.dialects.dialect.unit_to_str(expression),
         )
 
+    def arrayinsert_sql(self, expression: exp.ArrayInsert, index_offset: int = 0) -> str:
+        this = expression.this
+        position = expression.args["position"]
+        offset = index_offset - (expression.args.get("offset") or 0)
+
+        if offset:
+            if position.is_int:
+                value = position.to_py()
+                if value >= 0:
+                    position = exp.Literal.number(value + offset)
+                elif offset < 0 and value == -1:
+                    # 1-based -1 appends, which a 0-based position can only express as the size
+                    position = exp.ArraySize(this=this.copy())
+                else:
+                    # Negative positions count from the end, so they shift in the opposite
+                    # direction, e.g. 0-based -1 (before the last element) is 1-based -2
+                    position = exp.Literal.number(value - offset)
+            else:
+                self.unsupported("ARRAY_INSERT position can only be converted if it's a literal")
+
+        return self.func("ARRAY_INSERT", this, position, expression.expression)
+
     def arrayany_sql(self, expression: exp.ArrayAny) -> str:
         if self.CAN_IMPLEMENT_ARRAY_ANY:
             filtered = exp.ArrayFilter(this=expression.this, expression=expression.expression)

{sqlglot-30.20.0 → sqlglot-30.21.0}/sqlglot/generators/athena.py

@@ -122,7 +122,7 @@ class AthenaTrinoGenerator(TrinoGenerator):
     }
 
 
-class AthenaGenerator(generator.Generator):
+class AthenaGenerator(AthenaTrinoGenerator):
     SELECT_KINDS: tuple[str, ...] = ()
     SUPPORTS_DECODE_CASE = False
 

{sqlglot-30.20.0 → sqlglot-30.21.0}/sqlglot/generators/clickhouse.py

@@ -370,6 +370,7 @@ class ClickHouseGenerator(generator.Generator):
         exp.UnixToTime: _unix_to_time_sql,
         exp.Trim: lambda self, e: trim_sql(self, e, default_trim_type="BOTH"),
         exp.Variance: rename_func("varSamp"),
+        exp.VariancePop: rename_func("varPop"),
         exp.SchemaCommentProperty: lambda self, e: self.naked_property(e),
         exp.Stddev: rename_func("stddevSamp"),
         exp.Chr: rename_func("CHAR"),

{sqlglot-30.20.0 → sqlglot-30.21.0}/sqlglot/generators/singlestore.py

@@ -1529,7 +1529,12 @@ class SingleStoreGenerator(MySQLGenerator):
 
     def collate_sql(self, expression: exp.Collate) -> str:
         # SingleStore does not support setting a collation for column in the SELECT query,
-        # so we cast column to a LONGTEXT type with specific collation
+        # so we cast column to a LONGTEXT type with specific collation. If the cast is
+        # already there (e.g. when regenerating previously generated SQL), don't add
+        # another one, so that generation stays idempotent.
+        this = expression.this
+        if isinstance(this, exp.Cast) and this.to.this == exp.DType.LONGTEXT:
+            return self.binary(expression, "COLLATE")
         return self.binary(expression, ":> LONGTEXT COLLATE")
 
     def currentdate_sql(self, expression: exp.CurrentDate) -> str:

{sqlglot-30.20.0 → sqlglot-30.21.0}/sqlglot/generators/spark.py

@@ -95,9 +95,6 @@ class SparkGenerator(Spark2Generator):
             exp.ArrayConstructCompact: lambda self, e: self.func(
                 "ARRAY_COMPACT", self.func("ARRAY", *e.expressions)
             ),
-            exp.ArrayInsert: lambda self, e: self.func(
-                "ARRAY_INSERT", e.this, e.args.get("position"), e.expression
-            ),
             exp.ArrayAppend: array_append_sql("ARRAY_APPEND"),
             exp.ArrayPrepend: array_append_sql("ARRAY_PREPEND"),
             exp.BitwiseAndAgg: rename_func("BIT_AND"),
@@ -152,6 +149,9 @@ class SparkGenerator(Spark2Generator):
     def ignorenulls_sql(self, expression: exp.IgnoreNulls) -> str:
         return generator.Generator.ignorenulls_sql(self, expression)
 
+    def arrayinsert_sql(self, expression: exp.ArrayInsert, index_offset: int = 0) -> str:
+        return super().arrayinsert_sql(expression, index_offset=1)
+
     def bracket_sql(self, expression: exp.Bracket) -> str:
         if expression.args.get("safe"):
             key = seq_get(self.bracket_offset_expressions(expression, index_offset=1), 0)

{sqlglot-30.20.0 → sqlglot-30.21.0}/sqlglot/generators/spark2.py

@@ -220,6 +220,16 @@ class Spark2Generator(HiveGenerator):
 
     def cast_sql(self, expression: exp.Cast, safe_prefix: str | None = None) -> str:
         arg = expression.this
+        if (
+            isinstance(arg, exp.Literal)
+            and arg.is_number
+            and expression.to.is_type(exp.DType.DECIMAL)
+            and len(expression.to.expressions) == 2
+            and isinstance(expression.to.expressions[1].this, exp.Neg)
+        ):
+            # Spark 2 supports negative scales in literals, but not in DECIMAL type syntax.
+            return f"{self.sql(arg)}BD"
+
         is_json_extract = isinstance(
             arg, (exp.JSONExtract, exp.JSONExtractScalar)
         ) and not arg.args.get("variant_extract")

{sqlglot-30.20.0 → sqlglot-30.21.0}/sqlglot/optimizer/canonicalize_internal_names.py

@@ -205,10 +205,12 @@ def canonicalize_internal_names(expression: E) -> E:
                             ],
                         )
 
-                # BigQuery UNNEST ... WITH OFFSET AS <id> declares a pseudo-column via
-                # the offset arg (not the alias).
-                if isinstance(alias_holder, exp.Unnest):
-                    offset_id = alias_holder.args.get("offset")
+                # UNNEST stores its offset column outside the alias, including under LATERAL.
+                unnest = (
+                    alias_holder.this if isinstance(alias_holder, exp.Lateral) else alias_holder
+                )
+                if isinstance(unnest, exp.Unnest):
+                    offset_id = unnest.args.get("offset")
                     if isinstance(offset_id, exp.Identifier) and offset_id.name in name_map:
                         _canon(offset_id, name_map[offset_id.name])