Merging upstream version 21.1.2.

Signed-off-by: Daniel Baumann <daniel@debian.org>

sqlglot/optimizer/normalize.py

@@ -76,7 +76,7 @@ def normalized(expression: exp.Expression, dnf: bool = False) -> bool:
 
     Args:
         expression: The expression to check if it's normalized.
-        dnf: Whether or not to check if the expression is in Disjunctive Normal Form (DNF).
+        dnf: Whether to check if the expression is in Disjunctive Normal Form (DNF).
             Default: False, i.e. we check if it's in Conjunctive Normal Form (CNF).
     """
     ancestor, root = (exp.And, exp.Or) if dnf else (exp.Or, exp.And)
@@ -99,7 +99,7 @@ def normalization_distance(expression: exp.Expression, dnf: bool = False) -> int
 
     Args:
         expression: The expression to compute the normalization distance for.
-        dnf: Whether or not to check if the expression is in Disjunctive Normal Form (DNF).
+        dnf: Whether to check if the expression is in Disjunctive Normal Form (DNF).
             Default: False, i.e. we check if it's in Conjunctive Normal Form (CNF).
 
     Returns:

sqlglot/optimizer/qualify.py

@@ -48,15 +48,15 @@ def qualify(
         db: Default database name for tables.
         catalog: Default catalog name for tables.
         schema: Schema to infer column names and types.
-        expand_alias_refs: Whether or not to expand references to aliases.
-        expand_stars: Whether or not to expand star queries. This is a necessary step
+        expand_alias_refs: Whether to expand references to aliases.
+        expand_stars: Whether to expand star queries. This is a necessary step
             for most of the optimizer's rules to work; do not set to False unless you
             know what you're doing!
-        infer_schema: Whether or not to infer the schema if missing.
-        isolate_tables: Whether or not to isolate table selects.
-        qualify_columns: Whether or not to qualify columns.
-        validate_qualify_columns: Whether or not to validate columns.
-        quote_identifiers: Whether or not to run the quote_identifiers step.
+        infer_schema: Whether to infer the schema if missing.
+        isolate_tables: Whether to isolate table selects.
+        qualify_columns: Whether to qualify columns.
+        validate_qualify_columns: Whether to validate columns.
+        quote_identifiers: Whether to run the quote_identifiers step.
             This step is necessary to ensure correctness for case sensitive queries.
             But this flag is provided in case this step is performed at a later time.
         identify: If True, quote all identifiers, else only necessary ones.

sqlglot/optimizer/qualify_columns.py

@@ -35,11 +35,11 @@ def qualify_columns(
     Args:
         expression: Expression to qualify.
         schema: Database schema.
-        expand_alias_refs: Whether or not to expand references to aliases.
-        expand_stars: Whether or not to expand star queries. This is a necessary step
+        expand_alias_refs: Whether to expand references to aliases.
+        expand_stars: Whether to expand star queries. This is a necessary step
             for most of the optimizer's rules to work; do not set to False unless you
             know what you're doing!
-        infer_schema: Whether or not to infer the schema if missing.
+        infer_schema: Whether to infer the schema if missing.
 
     Returns:
         The qualified expression.
@@ -164,12 +164,7 @@ def _expand_using(scope: Scope, resolver: Resolver) -> t.Dict[str, t.Any]:
 
             table = table or source_table
             conditions.append(
-                exp.condition(
-                    exp.EQ(
-                        this=exp.column(identifier, table=table),
-                        expression=exp.column(identifier, table=join_table),
-                    )
-                )
+                exp.column(identifier, table=table).eq(exp.column(identifier, table=join_table))
             )
 
             # Set all values in the dict to None, because we only care about the key ordering
@@ -449,10 +444,9 @@ def _expand_stars(
                     continue
 
             for name in columns:
+                if name in columns_to_exclude or name in coalesced_columns:
+                    continue
                 if name in using_column_tables and table in using_column_tables[name]:
-                    if name in coalesced_columns:
-                        continue
-
                     coalesced_columns.add(name)
                     tables = using_column_tables[name]
                     coalesce = [exp.column(name, table=table) for table in tables]
@@ -464,7 +458,7 @@ def _expand_stars(
                             copy=False,
                         )
                     )
-                elif name not in columns_to_exclude:
+                else:
                     alias_ = replace_columns.get(table_id, {}).get(name, name)
                     column = exp.column(name, table=table)
                     new_selections.append(

sqlglot/optimizer/scope.py

@@ -254,7 +254,7 @@ class Scope:
             self._columns = []
             for column in columns + external_columns:
                 ancestor = column.find_ancestor(
-                    exp.Select, exp.Qualify, exp.Order, exp.Having, exp.Hint, exp.Table
+                    exp.Select, exp.Qualify, exp.Order, exp.Having, exp.Hint, exp.Table, exp.Star
                 )
                 if (
                     not ancestor

sqlglot/optimizer/simplify.py

@@ -41,7 +41,7 @@ def simplify(
 
     Args:
         expression (sqlglot.Expression): expression to simplify
-        constant_propagation: whether or not the constant propagation rule should be used
+        constant_propagation: whether the constant propagation rule should be used
 
     Returns:
         sqlglot.Expression: simplified expression

sqlglot/optimizer/unnest_subqueries.py

@@ -248,7 +248,7 @@ def decorrelate(select, parent_select, external_columns, next_alias_name):
             key.replace(exp.to_identifier("_x"))
             parent_predicate = _replace(
                 parent_predicate,
-                f'({parent_predicate} AND ARRAY_ANY({nested}, "_x" -> {predicate}))',
+                f"({parent_predicate} AND ARRAY_ANY({nested}, _x -> {predicate}))",
             )
 
     parent_select.join(