Added support for XML VALUE() datatype method

contrib/babelfishpg_tsql/sql/sys_functions.sql

@@ -108,6 +108,64 @@ END
 $BODY$
 LANGUAGE plpgsql STABLE STRICT PARALLEL SAFE;
 
+-- helper functions for XML VALUE(xpath)
+CREATE OR REPLACE FUNCTION sys.bbf_xmlvalue(xpath_pattern TEXT, xml_element ANYELEMENT, datatype 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
+        -- for User Defined Datatype, use immediate base type to check for xml_element datatype validation
+        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;
+
+    temp_datatype := sys.translate_pg_type_to_tsql(datatype::regtype::oid);
+    IF temp_datatype IS NULL THEN
+        -- for User Defined Datatype, use immediate base type to check for datatype validation
+        temp_basetype := sys.bbf_get_immediate_base_type_of_UDT(datatype::regtype::oid);
+        temp_datatype := sys.translate_pg_type_to_tsql(temp_basetype);
+    END IF;
+
+    IF (temp_datatype IN ('xml', 'image', 'text', 'ntext', 'sql_variant', 'geometry', 'geography', 'vector', 'sparsevec', 'halfvec')) THEN
+        RAISE EXCEPTION 'The data type ''%'' used in the VALUE method is invalid.', datatype;
+    END IF;
+
+    result_set := xpath(xpath_pattern, xml_element);
+    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))[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--5.2.0--5.3.0.sql

@@ -208,6 +208,64 @@ RETURNS INTEGER AS
 'babelfishpg_tsql', 'isnumeric'
 LANGUAGE C IMMUTABLE PARALLEL SAFE;
 
+CREATE OR REPLACE FUNCTION sys.bbf_xmlvalue(xpath_pattern TEXT, xml_element ANYELEMENT, datatype 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
+        -- for User Defined Datatype, use immediate base type to check for xml_element datatype validation
+        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;
+
+    temp_datatype := sys.translate_pg_type_to_tsql(datatype::regtype::oid);
+    IF temp_datatype IS NULL THEN
+        -- for User Defined Datatype, use immediate base type to check for datatype validation
+        temp_basetype := sys.bbf_get_immediate_base_type_of_UDT(datatype::regtype::oid);
+        temp_datatype := sys.translate_pg_type_to_tsql(temp_basetype);
+    END IF;
+
+    IF (temp_datatype IN ('xml', 'image', 'text', 'ntext', 'sql_variant', 'geometry', 'geography', 'vector', 'sparsevec', 'halfvec')) THEN
+        RAISE EXCEPTION 'The data type ''%'' used in the VALUE method is invalid.', datatype;
+    END IF;
+
+    result_set := xpath(xpath_pattern, xml_element);
+    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))[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;
+
+
 CREATE OR REPLACE FUNCTION sys.tsql_type_precision_helper(IN type TEXT, IN typemod INT) RETURNS sys.TINYINT
 AS $$
 DECLARE

contrib/babelfishpg_tsql/src/backend_parser/gram-tsql-decl.y

@@ -128,6 +128,7 @@
 	TSQL_CONTAINS TSQL_FREETEXT
 	TSQL_RECOMPILE
 	TSQL_UNPIVOT
+	TSQL_XML_VALUE
 
 /*
  * WITH_paren is added to support table hints syntax WITH (<table_hint> [[,]...n]),

contrib/babelfishpg_tsql/src/backend_parser/gram-tsql-rule.y

@@ -2119,6 +2119,19 @@ func_expr_common_subexpr:
 											   COERCE_EXPLICIT_CALL,
 											   @1);
 				}
+			| TSQL_XML_VALUE '(' c_expr ',' Typename ',' c_expr ')'
+				{
+					Node       *helperFuncCall;
+					List       *args;
+
+					args = list_make3($3, $7, makeStringConst(TypeNameToString($5), @5));
+					helperFuncCall = (Node *) makeFuncCall(TsqlSystemFuncName("bbf_xmlvalue"), 
+														    args, 
+														    COERCE_EXPLICIT_CALL,
+														    @1);
+
+					$$ = TsqlFunctionConvert($5, helperFuncCall, NULL, false, @1);
+				}
 		;
 
 opt_var_name_list:
@@ -4855,6 +4868,7 @@ reserved_keyword:
 			| TSQL_TRY_CONVERT
 			| TSQL_TRY_PARSE
 			| TSQL_EXEC
+			| TSQL_XML_VALUE
 		;
 
 bare_label_keyword:

contrib/babelfishpg_tsql/src/backend_parser/kwlist.h

@@ -38,6 +38,7 @@ PG_KEYWORD("authorization", AUTHORIZATION, TYPE_FUNC_NAME_KEYWORD)
 PG_KEYWORD("auto", TSQL_AUTO, UNRESERVED_KEYWORD)
 PG_KEYWORD("backward", BACKWARD, UNRESERVED_KEYWORD)
 PG_KEYWORD("base64", TSQL_BASE64, UNRESERVED_KEYWORD)
+PG_KEYWORD("bbf_xmlvalue", TSQL_XML_VALUE, RESERVED_KEYWORD)
 PG_KEYWORD("before", BEFORE, UNRESERVED_KEYWORD)
 PG_KEYWORD("begin", BEGIN_P, UNRESERVED_KEYWORD)
 PG_KEYWORD("between", BETWEEN, COL_NAME_KEYWORD)

contrib/babelfishpg_tsql/src/pltsql_instr.h

@@ -45,9 +45,7 @@ typedef enum PgTsqlInstrMetricType
 	INSTR_UNSUPPORTED_TSQL_FREETEXT,
 	INSTR_UNSUPPORTED_TSQL_NEXT_VALUE_FOR,
 	INSTR_UNSUPPORTED_TSQL_XML_NODES,
-	INSTR_UNSUPPORTED_TSQL_XML_VALUE,
 	INSTR_UNSUPPORTED_TSQL_XML_QUERY,
-	INSTR_UNSUPPORTED_TSQL_XML_EXIST,
 	INSTR_UNSUPPORTED_TSQL_XML_MODIFY,
 	INSTR_UNSUPPORTED_TSQL_ALTER_FUNCTION,
 	INSTR_UNSUPPORTED_TSQL_ALTER_FUNCTION_ENCRYPTION_OPTION,

contrib/babelfishpg_tsql/src/tsqlIface.cpp

@@ -1048,7 +1048,7 @@ class tsqlCommonMutator : public TSqlParserBaseListener
 
 	void exitXml_func_arg(TSqlParser::Xml_func_argContext *ctx) override
 	{
-		if (ctx->EXIST())
+		if (ctx->EXIST() || ctx->VALUE())
 		{
 			size_t startPosition = ctx->start->getStartIndex();
 			rewritten_query_fragment.emplace(std::make_pair(startPosition, std::make_pair("", "bbf_xml")));
@@ -9282,12 +9282,16 @@ handleGeospatialFunctionsInFunctionCall(TSqlParser::Function_callContext *ctx)
 }
 
 static void
-validateXMLFunctionArgs(TSqlParser::Xml_func_argContext *xml_func, TSqlParser::Expression_listContext *expr_list)
+validateAndRewriteXMLFunctionArgs(TSqlParser::Xml_func_argContext *xml_func, TSqlParser::Expression_listContext *expr_list)
 {
 	/* XML EXIST function requires only 1 argument */
 	if (xml_func->EXIST() && (expr_list == NULL || expr_list->expression().size() != 1))
 		throw PGErrorWrapperException(ERROR, ERRCODE_UNDEFINED_FUNCTION, "The exist function requires 1 argument(s).", getLineAndPos(xml_func));
 
+	/* XML VALUE function requires only 2 argument */
+	if (xml_func->VALUE() && (expr_list == NULL || expr_list->expression().size() != 2))
+		throw PGErrorWrapperException(ERROR, ERRCODE_UNDEFINED_FUNCTION, "The value function requires 2 argument(s).", getLineAndPos(xml_func));
+
 	/* Only String Literal is allowed as agument for XML Functions */
 	if (expr_list)
 	{
@@ -9303,6 +9307,21 @@ validateXMLFunctionArgs(TSqlParser::Xml_func_argContext *xml_func, TSqlParser::E
 						getLineAndPos(expr));
 		}
 	}
+
+	/*
+	 * Rewrite second argument of VALUE method by removing the quotes around it as follows
+	 * 'typename(typmod)'  -> typename(typmod)
+	 */
+	if (xml_func->VALUE())
+	{
+		TSqlParser::ExpressionContext *expr = expr_list->expression()[1];
+		std::string arg_str = ::getFullText(expr);
+		size_t startPosition = expr->start->getStartIndex();
+		size_t stopPosition = expr->stop->getStopIndex();
+
+		std::string rewritten_arg_str = arg_str.substr(1, stopPosition - startPosition - 1);
+		rewritten_query_fragment.emplace(std::make_pair(startPosition, std::make_pair(arg_str, rewritten_arg_str)));
+	}
 }
 
 static void
@@ -9312,7 +9331,7 @@ handleXMLFunctionsInFunctionCall(TSqlParser::Function_callContext *ctx)
 	if (ctx->xml_proc_name_table_column())
 	{
 		/* validate the xml method arguments before rewriting */
-		validateXMLFunctionArgs(ctx->xml_proc_name_table_column()->xml_func_arg(), ctx->expression_list());
+		validateAndRewriteXMLFunctionArgs(ctx->xml_proc_name_table_column()->xml_func_arg(), ctx->expression_list());
 
 		rewrite_function_call_dot_func_ref_args(ctx);
 	}
@@ -9348,7 +9367,7 @@ handleClrUdtFuncCall(TSqlParser::Clr_udt_func_callContext *ctx)
 			if (method->xml_methods())
 			{
 				/* validate the xml method arguments before rewriting */
-				validateXMLFunctionArgs(method->xml_methods()->xml_func_arg(), method->xml_methods()->expression_list());
+				validateAndRewriteXMLFunctionArgs(method->xml_methods()->xml_func_arg(), method->xml_methods()->expression_list());
 
 				size_t expr_list_start_index = method->xml_methods()->expression_list()->start->getStartIndex();
 				size_t expr_list_stop_index = method->xml_methods()->expression_list()->stop->getStopIndex();

contrib/babelfishpg_tsql/src/tsqlUnsupportedFeatureHandler.cpp

@@ -1527,9 +1527,7 @@ antlrcpp::Any TsqlUnsupportedFeatureHandlerImpl::visitId(TSqlParser::IdContext *
 
 antlrcpp::Any TsqlUnsupportedFeatureHandlerImpl::visitXml_func_arg(TSqlParser::Xml_func_argContext *ctx)
 {
-	if (ctx->VALUE())
-		handle(INSTR_UNSUPPORTED_TSQL_XML_VALUE, "XML VALUE", getLineAndPos(ctx));
-	else if (ctx->QUERY())
+	if (ctx->QUERY())
 		handle(INSTR_UNSUPPORTED_TSQL_XML_QUERY, "XML QUERY", getLineAndPos(ctx));
 	else if (ctx->MODIFY())
 		handle(INSTR_UNSUPPORTED_TSQL_XML_QUERY, "XML MODIFY", getLineAndPos(ctx));