Add WITH XMLNAMESPACES support

Author: Japleen Kaur

contrib/babelfishpg_tsql/runtime/functions.c

@@ -6096,28 +6096,23 @@ openxml_simple(PG_FUNCTION_ARGS)
 
 
 PG_FUNCTION_INFO_V1(bbf_xmlquery);
+PG_FUNCTION_INFO_V1(bbf_xmlquery_ns);
 
 /*
- * bbf_xmlquery - C implementation of XML .query() method
+ * bbf_xmlquery_internal - shared implementation for the .query() XML method.
  *
- * Signature:
- *   sys.bbf_xmlquery(xpath_pattern TEXT, xml_element ANYELEMENT)
- *
- * Returns XML result of evaluating the XPath expression against the input.
- * Returns empty XML if no nodes match.
- *
- * Validates:
- *   - Input must be XML type (or UDT based on XML)
- *   - QUOTED_IDENTIFIER must be ON
+ * Validates the input type and QUOTED_IDENTIFIER setting, then evaluates the
+ * XPath expression against the XML datum using the supplied namespace array
+ * (which may be empty).  Returns the concatenated XML fragments, or the
+ * empty XML string if no nodes match.
  */
-Datum
-bbf_xmlquery(PG_FUNCTION_ARGS)
+static Datum
+bbf_xmlquery_internal(FunctionCallInfo fcinfo, ArrayType *namespaces)
 {
 	text	   *xpath_expr;
 	Datum		xml_datum;
 	Oid			arg_type;
 	Oid			immediate_base_type;
-	ArrayType  *namespaces;
 	Datum		xpath_result;
 	ArrayType  *result_arr;
 	Datum	   *elems;
@@ -6161,13 +6156,11 @@ bbf_xmlquery(PG_FUNCTION_ARGS)
 						"SET options are correct for XML data type methods.")));
 
 	/*
-	 * Call the built-in xpath(text, xml, text[][]) directly with an empty
-	 * namespace array. Returns xml[] (array of XML fragments).
-	 *
-	 * TODO: when WITH XMLNAMESPACES is supported, populate this array with
-	 * the declared (prefix, uri) pairs from the active namespace context.
+	 * Call the built-in xpath(text, xml, text[][]).  Returns xml[] (array of
+	 * XML fragments).  When invoked without WITH XMLNAMESPACES the caller
+	 * passes an empty array so libxml2 only resolves prefixes already
+	 * declared in the document.
 	 */
-	namespaces = construct_empty_array(TEXTOID);
 	xpath_result = DirectFunctionCall3(xpath,
 									   PointerGetDatum(xpath_expr),
 									   xml_datum,
@@ -6206,3 +6199,35 @@ bbf_xmlquery(PG_FUNCTION_ARGS)
 
 	PG_RETURN_XML_P((xmltype *) cstring_to_text_with_len(buf.data, buf.len));
 }
+
+/*
+ * bbf_xmlquery - C implementation of XML .query() method (no namespaces).
+ *
+ * Signature:
+ *   sys.bbf_xmlquery(xpath_pattern TEXT, xml_element ANYELEMENT)
+ */
+Datum
+bbf_xmlquery(PG_FUNCTION_ARGS)
+{
+	ArrayType  *namespaces = construct_empty_array(TEXTOID);
+
+	return bbf_xmlquery_internal(fcinfo, namespaces);
+}
+
+/*
+ * bbf_xmlquery_ns - C implementation of XML .query() with namespace support.
+ *
+ * Signature:
+ *   sys.bbf_xmlquery(xpath_pattern TEXT, xml_element ANYELEMENT, nsarray TEXT[][])
+ *
+ * Used by the WITH XMLNAMESPACES rewrite path.  The ANTLR rewrite layer
+ * appends the namespace array as a 3rd argument when a WITH XMLNAMESPACES
+ * clause is in scope.
+ */
+Datum
+bbf_xmlquery_ns(PG_FUNCTION_ARGS)
+{
+	ArrayType  *namespaces = PG_GETARG_ARRAYTYPE_P(2);
+
+	return bbf_xmlquery_internal(fcinfo, namespaces);
+}

contrib/babelfishpg_tsql/sql/sys_functions.sql

@@ -8,7 +8,8 @@ CREATE OR REPLACE FUNCTION sys.tsql_query_to_xml_sfunc(
     root_name text,
     elements boolean,
     xsinil boolean,
-    auto_metadata text
+    auto_metadata text,
+    ns_decls text
 ) RETURNS INTERNAL
 AS 'babelfishpg_tsql', 'tsql_query_to_xml_sfunc'
 LANGUAGE C STABLE;
@@ -35,7 +36,8 @@ CREATE OR REPLACE AGGREGATE sys.tsql_select_for_xml_agg(
     root_name text,
     elements boolean,
     xsinil boolean,
-    auto_metadata text)
+    auto_metadata text,
+    ns_decls text)
 (
     STYPE = INTERNAL,
     SFUNC = tsql_query_to_xml_sfunc,
@@ -50,7 +52,8 @@ CREATE OR REPLACE AGGREGATE sys.tsql_select_for_xml_text_agg(
     root_name text,
     elements boolean,
     xsinil boolean,
-    auto_metadata text)
+    auto_metadata text,
+    ns_decls text)
 (
     STYPE = INTERNAL,
     SFUNC = tsql_query_to_xml_sfunc,
@@ -170,6 +173,91 @@ RETURNS XML
 AS 'babelfishpg_tsql', 'bbf_xmlquery'
 LANGUAGE C STABLE STRICT PARALLEL SAFE;
 
+-- helper function for XML QUERY(xpath) with namespace support (used by WITH XMLNAMESPACES)
+CREATE OR REPLACE FUNCTION sys.bbf_xmlquery(xpath_pattern TEXT, xml_element ANYELEMENT, nsarray TEXT[][])
+RETURNS XML
+AS 'babelfishpg_tsql', 'bbf_xmlquery_ns'
+LANGUAGE C STABLE STRICT PARALLEL SAFE;
+
+-- helper function for XML EXIST(xpath) with namespace support (used by WITH XMLNAMESPACES)
+CREATE OR REPLACE FUNCTION sys.bbf_xmlexist(xpath_pattern TEXT, xml_element ANYELEMENT, nsarray TEXT[][])
+RETURNS sys.BIT
+AS
+$BODY$
+DECLARE
+    arg_datatype text;
+    arg_datatype_oid oid;
+    basetype oid;
+    pltsql_quoted_identifier text;
+BEGIN
+    arg_datatype_oid := pg_typeof(xml_element)::oid;
+    arg_datatype := sys.translate_pg_type_to_tsql(arg_datatype_oid);
+    IF arg_datatype IS NULL THEN
+        basetype := sys.bbf_get_immediate_base_type_of_UDT(arg_datatype_oid);
+        arg_datatype := sys.translate_pg_type_to_tsql(basetype);
+    END IF;
+
+    IF (arg_datatype != 'xml') THEN
+        RAISE EXCEPTION 'Cannot call methods on %.', arg_datatype;
+    END IF;
+
+    pltsql_quoted_identifier := current_setting('babelfishpg_tsql.quoted_identifier');
+
+    IF (pltsql_quoted_identifier = 'off') THEN
+        RAISE EXCEPTION 'SELECT failed because the following SET options have incorrect settings: ''QUOTED_IDENTIFIER''. Verify that SET options are correct for XML data type methods.';
+    END IF;
+
+    RETURN (cardinality(xpath(xpath_pattern, xml_element, nsarray)) > 0)::int::sys.BIT;
+END
+$BODY$
+LANGUAGE plpgsql STABLE STRICT PARALLEL SAFE;
+
+-- helper function for XML VALUE(xpath) with namespace support (used by WITH XMLNAMESPACES)
+CREATE OR REPLACE FUNCTION sys.bbf_xmlvalue(xpath_pattern TEXT, datatype TEXT, xml_element ANYELEMENT, nsarray TEXT[][])
+RETURNS sys.NVARCHAR
+AS
+$BODY$
+DECLARE
+    temp_datatype text;
+    temp_basetype oid;
+    result_set xml[];
+    result sys.NVARCHAR;
+    pltsql_quoted_identifier text;
+BEGIN
+    temp_datatype := sys.translate_pg_type_to_tsql(pg_typeof(xml_element)::oid);
+    IF temp_datatype IS NULL THEN
+        temp_basetype := sys.bbf_get_immediate_base_type_of_UDT(pg_typeof(xml_element)::oid);
+        temp_datatype := sys.translate_pg_type_to_tsql(temp_basetype);
+    END IF;
+
+    IF (temp_datatype != 'xml') THEN
+        RAISE EXCEPTION 'Cannot call methods on %.', temp_datatype;
+    END IF;
+
+    pltsql_quoted_identifier := current_setting('babelfishpg_tsql.quoted_identifier');
+
+    IF (pltsql_quoted_identifier = 'off') THEN
+        RAISE EXCEPTION 'SELECT failed because the following SET options have incorrect settings: ''QUOTED_IDENTIFIER''. Verify that SET options are correct for XML data type methods.';
+    END IF;
+
+    result_set := xpath(xpath_pattern, xml_element, nsarray);
+    IF (cardinality(result_set) > 1) THEN
+        RAISE EXCEPTION 'XML Value result is not a single value.';
+    ELSIF (cardinality(result_set) = 0) THEN
+        RETURN NULL;
+    ELSE
+        result := (xpath('string(' || xpath_pattern || ')', xml_element, nsarray))[1];
+        result := pg_catalog.replace(result, '&lt;', '<');
+        result := pg_catalog.replace(result, '&gt;', '>');
+        result := pg_catalog.replace(result, '&apos;', '''');
+        result := pg_catalog.replace(result, '&quot;', '"');
+        result := pg_catalog.replace(result, '&amp;', '&');
+        return result;
+    END IF;
+END
+$BODY$
+LANGUAGE plpgsql STABLE STRICT PARALLEL SAFE;
+
 -- SELECT FOR JSON
 CREATE OR REPLACE FUNCTION sys.tsql_query_to_json_sfunc(
     state INTERNAL,

contrib/babelfishpg_tsql/sql/upgrades/babelfishpg_tsql--5.6.0--5.7.0.sql

@@ -201,6 +201,149 @@ CREATE OR REPLACE AGGREGATE sys.tsql_select_for_xml_text_agg(
     FINALFUNC = tsql_query_to_xml_text_ffunc
 );
 
+-- WITH XMLNAMESPACES support: extend FOR XML aggregate to carry namespace declarations
+-- and add namespace-aware overloads for XML data type methods.
+--
+-- The previous block above already migrated the aggregate from 7 args (legacy) to
+-- 8 args (adding auto_metadata for FOR XML AUTO). We now drop those 8-arg
+-- definitions and recreate at 9 args (adding ns_decls for WITH XMLNAMESPACES),
+-- so the final shape matches sys_functions.sql.
+CALL sys.babelfish_drop_deprecated_object('aggregate', 'sys', 'tsql_select_for_xml_agg', 'ANYELEMENT, int, text, boolean, text, boolean, boolean, text');
+CALL sys.babelfish_drop_deprecated_object('aggregate', 'sys', 'tsql_select_for_xml_text_agg', 'ANYELEMENT, int, text, boolean, text, boolean, boolean, text');
+CALL sys.babelfish_drop_deprecated_object('function', 'sys', 'tsql_query_to_xml_sfunc', 'INTERNAL, ANYELEMENT, int, text, boolean, text, boolean, boolean, text');
+
+CREATE OR REPLACE FUNCTION sys.tsql_query_to_xml_sfunc(
+    state INTERNAL,
+    rec ANYELEMENT,
+    mode int,
+    element_name text,
+    binary_base64 boolean,
+    root_name text,
+    elements boolean,
+    xsinil boolean,
+    auto_metadata text,
+    ns_decls text
+) RETURNS INTERNAL
+AS 'babelfishpg_tsql', 'tsql_query_to_xml_sfunc'
+LANGUAGE C STABLE;
+
+CREATE OR REPLACE AGGREGATE sys.tsql_select_for_xml_agg(
+    rec ANYELEMENT,
+    mode int,
+    element_name text,
+    binary_base64 boolean,
+    root_name text,
+    elements boolean,
+    xsinil boolean,
+    auto_metadata text,
+    ns_decls text)
+(
+    STYPE = INTERNAL,
+    SFUNC = tsql_query_to_xml_sfunc,
+    FINALFUNC = tsql_query_to_xml_ffunc
+);
+
+CREATE OR REPLACE AGGREGATE sys.tsql_select_for_xml_text_agg(
+    rec ANYELEMENT,
+    mode int,
+    element_name text,
+    binary_base64 boolean,
+    root_name text,
+    elements boolean,
+    xsinil boolean,
+    auto_metadata text,
+    ns_decls text)
+(
+    STYPE = INTERNAL,
+    SFUNC = tsql_query_to_xml_sfunc,
+    FINALFUNC = tsql_query_to_xml_text_ffunc
+);
+
+-- helper function for XML QUERY(xpath) with namespace support
+CREATE OR REPLACE FUNCTION sys.bbf_xmlquery(xpath_pattern TEXT, xml_element ANYELEMENT, nsarray TEXT[][])
+RETURNS XML
+AS 'babelfishpg_tsql', 'bbf_xmlquery_ns'
+LANGUAGE C STABLE STRICT PARALLEL SAFE;
+
+-- helper function for XML EXIST(xpath) with namespace support
+CREATE OR REPLACE FUNCTION sys.bbf_xmlexist(xpath_pattern TEXT, xml_element ANYELEMENT, nsarray TEXT[][])
+RETURNS sys.BIT
+AS
+$BODY$
+DECLARE
+    arg_datatype text;
+    arg_datatype_oid oid;
+    basetype oid;
+    pltsql_quoted_identifier text;
+BEGIN
+    arg_datatype_oid := pg_typeof(xml_element)::oid;
+    arg_datatype := sys.translate_pg_type_to_tsql(arg_datatype_oid);
+    IF arg_datatype IS NULL THEN
+        basetype := sys.bbf_get_immediate_base_type_of_UDT(arg_datatype_oid);
+        arg_datatype := sys.translate_pg_type_to_tsql(basetype);
+    END IF;
+
+    IF (arg_datatype != 'xml') THEN
+        RAISE EXCEPTION 'Cannot call methods on %.', arg_datatype;
+    END IF;
+
+    pltsql_quoted_identifier := current_setting('babelfishpg_tsql.quoted_identifier');
+
+    IF (pltsql_quoted_identifier = 'off') THEN
+        RAISE EXCEPTION 'SELECT failed because the following SET options have incorrect settings: ''QUOTED_IDENTIFIER''. Verify that SET options are correct for XML data type methods.';
+    END IF;
+
+    RETURN (cardinality(xpath(xpath_pattern, xml_element, nsarray)) > 0)::int::sys.BIT;
+END
+$BODY$
+LANGUAGE plpgsql STABLE STRICT PARALLEL SAFE;
+
+-- helper function for XML VALUE(xpath) with namespace support
+CREATE OR REPLACE FUNCTION sys.bbf_xmlvalue(xpath_pattern TEXT, datatype TEXT, xml_element ANYELEMENT, nsarray TEXT[][])
+RETURNS sys.NVARCHAR
+AS
+$BODY$
+DECLARE
+    temp_datatype text;
+    temp_basetype oid;
+    result_set xml[];
+    result sys.NVARCHAR;
+    pltsql_quoted_identifier text;
+BEGIN
+    temp_datatype := sys.translate_pg_type_to_tsql(pg_typeof(xml_element)::oid);
+    IF temp_datatype IS NULL THEN
+        temp_basetype := sys.bbf_get_immediate_base_type_of_UDT(pg_typeof(xml_element)::oid);
+        temp_datatype := sys.translate_pg_type_to_tsql(temp_basetype);
+    END IF;
+
+    IF (temp_datatype != 'xml') THEN
+        RAISE EXCEPTION 'Cannot call methods on %.', temp_datatype;
+    END IF;
+
+    pltsql_quoted_identifier := current_setting('babelfishpg_tsql.quoted_identifier');
+
+    IF (pltsql_quoted_identifier = 'off') THEN
+        RAISE EXCEPTION 'SELECT failed because the following SET options have incorrect settings: ''QUOTED_IDENTIFIER''. Verify that SET options are correct for XML data type methods.';
+    END IF;
+
+    result_set := xpath(xpath_pattern, xml_element, nsarray);
+    IF (cardinality(result_set) > 1) THEN
+        RAISE EXCEPTION 'XML Value result is not a single value.';
+    ELSIF (cardinality(result_set) = 0) THEN
+        RETURN NULL;
+    ELSE
+        result := (xpath('string(' || xpath_pattern || ')', xml_element, nsarray))[1];
+        result := pg_catalog.replace(result, '&lt;', '<');
+        result := pg_catalog.replace(result, '&gt;', '>');
+        result := pg_catalog.replace(result, '&apos;', '''');
+        result := pg_catalog.replace(result, '&quot;', '"');
+        result := pg_catalog.replace(result, '&amp;', '&');
+        return result;
+    END IF;
+END
+$BODY$
+LANGUAGE plpgsql STABLE STRICT PARALLEL SAFE;
+
 -- Drops the temporary procedure used by the upgrade script.
 -- Please have this be one of the last statements executed in this upgrade script.
 DROP PROCEDURE sys.babelfish_drop_deprecated_object(varchar, varchar, varchar, varchar);

contrib/babelfishpg_tsql/src/backend_parser/gram-tsql-epilogue.y.c

@@ -1930,6 +1930,25 @@ TsqlForXMLMakeFuncCall(TSQL_ForClause *forclause)
 	func_args = lappend(func_args, makeBoolAConst(xsinil, -1));
 	/* 8th arg: auto_metadata placeholder (empty string, filled in by handleForXmlAuto) */
 	func_args = lappend(func_args, makeStringConst("", -1));
+	/*
+	 * 9th arg: namespace declarations from WITH XMLNAMESPACES, if any.
+	 * The C++ ANTLR layer captures the WITH XMLNAMESPACES clause and stores
+	 * a formatted decls string ('xmlns:p1="u1" xmlns:p2="u2"') on the
+	 * enclosing PLtsql_stmt_execsql. We read it back from the currently
+	 * executing stmt via get_current_tsql_estate().
+	 */
+	{
+		PLtsql_execstate *estate = get_current_tsql_estate();
+		char	   *ns_decls = NULL;
+
+		if (estate && estate->err_stmt && estate->err_stmt->cmd_type == PLTSQL_STMT_EXECSQL)
+			ns_decls = ((PLtsql_stmt_execsql *) estate->err_stmt)->xml_namespace_decls;
+
+		if (ns_decls && ns_decls[0] != '\0')
+			func_args = lappend(func_args, makeStringConst(ns_decls, -1));
+		else
+			func_args = lappend(func_args, makeStringConst("", -1));
+	}
 	fc = makeFuncCall(func_name, func_args, COERCE_EXPLICIT_CALL, -1);
 
 	/*