Merge branch 'master' into dependabot/github_actions/actions/checkout-7

Author: macbre

.github/workflows/python-ci.yml

@@ -32,7 +32,7 @@ jobs:
         python-version: ${{ matrix.python-version }}
 
     - name: Load ~/.cache directory and Poetry .venv
-      uses: actions/cache@v5
+      uses: actions/cache@v6
       with:
         path: |
           ~/.cache

sql_metadata/nested_resolver.py

@@ -207,7 +207,7 @@ def __init__(
     def extract_cte_names(
         self,
         cte_name_map: dict[str, str],
-    ) -> list[str]:
+    ) -> UniqueList:
         """Extract CTE names from the AST.
 
         Called by :attr:`Parser.with_names`.
@@ -247,7 +247,7 @@ def extract_cte_bodies(
     @staticmethod
     def extract_subqueries(
         ast: exp.Expression,
-    ) -> tuple[list[str], dict[str, str]]:
+    ) -> tuple[UniqueList, dict[str, str]]:
         """Extract subquery names and bodies in a single post-order walk.
 
         Aliased subqueries keep their alias as the name.  Unaliased
@@ -262,7 +262,7 @@ def extract_subqueries(
         :returns: ``(names, bodies)`` where *names* is ordered innermost-first,
             e.g. ``(["subquery_1", "sub"], {...})``.
         """
-        names: list[str] = UniqueList()
+        names = UniqueList()
         bodies: dict[str, str] = {}
         NestedResolver._walk_subqueries(ast, names, bodies, 0)
         return names, bodies

sql_metadata/parser.py

@@ -63,13 +63,13 @@ def __init__(self, sql: str = "", disable_logging: bool = False) -> None:
         self._columns_aliases_dict: dict[str, UniqueList] = {}
         self._output_columns: list[str] = []
 
-        self._tables: list[str] | None = None
+        self._tables: UniqueList | None = None
         self._table_aliases: dict[str, str] | None = None
 
-        self._with_names: list[str] | None = None
+        self._with_names: UniqueList | None = None
         self._with_queries: dict[str, str] | None = None
         self._subqueries: dict[str, str] | None = None
-        self._subqueries_names: list[str] | None = None
+        self._subqueries_names: UniqueList | None = None
 
         self._limit_and_offset: tuple[int, int] | None = None
 
@@ -168,7 +168,7 @@ def tokens(self) -> list[str]:
         return self._tokens
 
     @property
-    def columns(self) -> list[str]:
+    def columns(self) -> UniqueList:
         """Return the list of column names referenced in the query.
 
         Walks the sqlglot AST via :class:`ColumnExtractor` in a single DFS
@@ -177,7 +177,7 @@ def columns(self) -> list[str]:
         SQL), falls back to a regex extraction of ``INTO … (col1, col2)``
         column lists.
 
-        :rtype: list[str]
+        :rtype: UniqueList
         """
         if self._columns_extracted:
             return self._columns
@@ -276,10 +276,10 @@ def columns_aliases_dict(self) -> dict[str, UniqueList]:
         return self._columns_aliases_dict
 
     @property
-    def columns_aliases_names(self) -> list[str]:
+    def columns_aliases_names(self) -> UniqueList:
         """Return the names of all column aliases used in the query.
 
-        :rtype: list[str]
+        :rtype: UniqueList
         """
         if not self._columns_extracted:
             _ = self.columns
@@ -299,14 +299,14 @@ def output_columns(self) -> list[str]:
         return self._output_columns
 
     @property
-    def tables(self) -> list[str]:
+    def tables(self) -> UniqueList:
         """Return the list of table names referenced in the query.
 
         Tables are extracted from the AST by :class:`TableExtractor`,
         sorted by their position in the SQL text, and filtered to exclude
         CTE names (which appear in :attr:`with_names` instead).
 
-        :rtype: list[str]
+        :rtype: UniqueList
         """
         if self._tables is not None:
             return self._tables
@@ -339,10 +339,10 @@ def tables_aliases(self) -> dict[str, str]:
         return self._table_aliases
 
     @property
-    def with_names(self) -> list[str]:
+    def with_names(self) -> UniqueList:
         """Return the CTE (Common Table Expression) names from the query.
 
-        :rtype: list[str]
+        :rtype: UniqueList
         """
         if self._with_names is not None:
             return self._with_names
@@ -387,13 +387,13 @@ def subqueries(self) -> dict[str, str]:
         return self._subqueries
 
     @property
-    def subqueries_names(self) -> list[str]:
+    def subqueries_names(self) -> UniqueList:
         """Return the names of all subqueries (innermost first).
 
         Aliased subqueries use their alias; unaliased ones get
         auto-generated names (``subquery_1``, ``subquery_2``, …).
 
-        :rtype: list[str]
+        :rtype: UniqueList
         """
         if self._subqueries_names is not None:
             return self._subqueries_names
@@ -482,7 +482,9 @@ def values_dict(self) -> dict[str, Any] | None:
         is_multi = values and isinstance(values[0], list)
         first_row = values[0] if is_multi else values
         if not columns:
-            columns = [f"column_{ind + 1}" for ind in range(len(first_row))]
+            columns = UniqueList(
+                f"column_{ind + 1}" for ind in range(len(first_row))
+            )
 
         if is_multi:
             self._values_dict = {

sql_metadata/table_extractor.py

@@ -141,7 +141,7 @@ def __init__(
     # Public API
     # -------------------------------------------------------------------
 
-    def extract(self) -> list[str]:
+    def extract(self) -> UniqueList:
         """Extract table names, excluding CTE definitions.
 
         For ``CREATE TABLE`` statements, the target table is always placed

test/test_getting_tables.py

@@ -948,6 +948,28 @@ def test_unmatched_parentheses_graceful():
     assert isinstance(tables, list)
 
 
+def test_mysql_view_definition_with_bracketed_join():
+    # solved: https://github.com/macbre/sql-metadata/issues/253
+    # MySQL view definitions wrap the FROM source in parentheses and
+    # double-bracket the ON clause; used to crash with AttributeError.
+    query = (
+        "select `t`.`symbol` AS `symbol` "
+        "from (`stock`.`top_momentum_sector` `s` "
+        "join `stock`.`daily_companies` `t` "
+        "on((`s`.`symbol` = `t`.`symbol`)))"
+    )
+    parser = Parser(query)
+    assert parser.tables == ["stock.top_momentum_sector", "stock.daily_companies"]
+    assert parser.columns == [
+        "stock.daily_companies.symbol",
+        "stock.top_momentum_sector.symbol",
+    ]
+    assert parser.tables_aliases == {
+        "s": "stock.top_momentum_sector",
+        "t": "stock.daily_companies",
+    }
+
+
 def test_degraded_parse_falls_through_to_last_dialect():
     """SELECT UNIQUE triggers multi-dialect retry."""
     p = Parser("SELECT UNIQUE col FROM t")