feat(parser): Support Oracle/Postgres XMLNAMESPACES in XMLTABLE (#4643)

* feat(parser): Support Oracle/Postgres XMLNAMESPACES in XMLTABLE

* Update sqlglot/parser.py

* Update sqlglot/parser.py

---------

Co-authored-by: Jo <[email protected]>

Author: rbreejen

sqlglot/expressions.py

@@ -6782,7 +6782,17 @@ class XMLElement(Func):
 
 
 class XMLTable(Func):
-    arg_types = {"this": True, "passing": False, "columns": False, "by_ref": False}
+    arg_types = {
+        "this": True,
+        "namespaces": False,
+        "passing": False,
+        "columns": False,
+        "by_ref": False,
+    }
+
+
+class XMLNamespace(Expression):
+    pass
 
 
 class Year(Func):

sqlglot/generator.py

@@ -4738,9 +4738,15 @@ def analyze_sql(self, expression: exp.Analyze) -> str:
 
     def xmltable_sql(self, expression: exp.XMLTable) -> str:
         this = self.sql(expression, "this")
+        namespaces = self.expressions(expression, key="namespaces")
+        namespaces = f"XMLNAMESPACES({namespaces}), " if namespaces else ""
         passing = self.expressions(expression, key="passing")
         passing = f"{self.sep()}PASSING{self.seg(passing)}" if passing else ""
         columns = self.expressions(expression, key="columns")
         columns = f"{self.sep()}COLUMNS{self.seg(columns)}" if columns else ""
         by_ref = f"{self.sep()}RETURNING SEQUENCE BY REF" if expression.args.get("by_ref") else ""
-        return f"XMLTABLE({self.sep('')}{self.indent(this + passing + by_ref + columns)}{self.seg(')', sep='')}"
+        return f"XMLTABLE({self.sep('')}{self.indent(namespaces + this + passing + by_ref + columns)}{self.seg(')', sep='')}"
+
+    def xmlnamespace_sql(self, expression: exp.XMLNamespace) -> str:
+        this = self.sql(expression, "this")
+        return this if isinstance(expression.this, exp.Alias) else f"DEFAULT {this}"

sqlglot/parser.py

@@ -6212,11 +6212,16 @@ def _parse_convert(
         return self.expression(exp.Cast if strict else exp.TryCast, this=this, to=to, safe=safe)
 
     def _parse_xml_table(self) -> exp.XMLTable:
-        this = self._parse_string()
-
+        namespaces = None
         passing = None
         columns = None
 
+        if self._match_text_seq("XMLNAMESPACES", "("):
+            namespaces = self._parse_xml_namespace()
+            self._match_text_seq(")", ",")
+
+        this = self._parse_string()
+
         if self._match_text_seq("PASSING"):
             # The BY VALUE keywords are optional and are provided for semantic clarity
             self._match_text_seq("BY", "VALUE")
@@ -6228,9 +6233,28 @@ def _parse_xml_table(self) -> exp.XMLTable:
             columns = self._parse_csv(self._parse_field_def)
 
         return self.expression(
-            exp.XMLTable, this=this, passing=passing, columns=columns, by_ref=by_ref
+            exp.XMLTable,
+            this=this,
+            namespaces=namespaces,
+            passing=passing,
+            columns=columns,
+            by_ref=by_ref,
         )
 
+    def _parse_xml_namespace(self) -> t.List[exp.XMLNamespace]:
+        namespaces = []
+
+        while True:
+            if self._match_text_seq("DEFAULT"):
+                uri = self._parse_string()
+            else:
+                uri = self._parse_alias(self._parse_string())
+            namespaces.append(self.expression(exp.XMLNamespace, this=uri))
+            if not self._match(TokenType.COMMA):
+                break
+
+        return namespaces
+
     def _parse_decode(self) -> t.Optional[exp.Decode | exp.Case]:
         """
         There are generally two variants of the DECODE function:

tests/dialects/test_oracle.py

@@ -394,6 +394,9 @@ def test_xml_table(self):
         self.validate_identity(
             "XMLTABLE('x' RETURNING SEQUENCE BY REF COLUMNS a VARCHAR2, b FLOAT)"
         )
+        self.validate_identity(
+            "SELECT x.* FROM example t, XMLTABLE(XMLNAMESPACES(DEFAULT 'http://example.com/default', 'http://example.com/ns1' AS \"ns1\"), '/root/data' PASSING t.xml COLUMNS id NUMBER PATH '@id', value VARCHAR2(100) PATH 'ns1:value/text()') x"
+        )
 
         self.validate_all(
             """SELECT warehouse_name warehouse,

tests/dialects/test_postgres.py

@@ -76,7 +76,10 @@ def test_postgres(self):
         self.validate_identity("SELECT CURRENT_USER")
         self.validate_identity("SELECT * FROM ONLY t1")
         self.validate_identity(
-            "SELECT id, name FROM XMLTABLE('/root/user' PASSING xml_data COLUMNS id INT PATH '@id', name TEXT PATH 'name/text()') AS t"
+            "SELECT id, name FROM xml_data AS t, XMLTABLE('/root/user' PASSING t.xml COLUMNS id INT PATH '@id', name TEXT PATH 'name/text()') AS x"
+        )
+        self.validate_identity(
+            "SELECT id, value FROM xml_content AS t, XMLTABLE(XMLNAMESPACES('http://example.com/ns1' AS ns1, 'http://example.com/ns2' AS ns2), '/root/data' PASSING t.xml COLUMNS id INT PATH '@ns1:id', value TEXT PATH 'ns2:value/text()') AS x"
         )
         self.validate_identity(
             "SELECT * FROM t WHERE some_column >= CURRENT_DATE + INTERVAL '1 day 1 hour' AND some_another_column IS TRUE"