Merge branch 'BABEL_6_X_DEV' into BABEL-6433

Author: anju15bharti

contrib/babelfishpg_tds/src/backend/tds/tdslogin.c

@@ -1162,6 +1162,46 @@ TdsResetLoginFlags()
  * Returns STATUS_OK or STATUS_ERROR, or might call ereport(FATAL) and
  * not return at all.
  */
+
+/*
+ * tds_truncate_identifier_md5
+ *
+ * Truncate an identifier exceeding NAMEDATALEN using clip + MD5 hash,
+ * matching the behavior of pltsql_truncate_identifier for CI_AS collations.
+ * Used for pre-auth user name truncation (where the tsql hook is unavailable)
+ * and for database name truncation (for consistency with the stored form).
+ */
+#define MD5_HASH_LEN 32
+
+static void
+tds_truncate_identifier_md5(char *ident, int len)
+{
+	char		md5[MD5_HASH_LEN + 1];
+	char		buf[NAMEDATALEN];
+	const char *errstr = NULL;
+	char	   *downcased;
+
+	Assert(len >= NAMEDATALEN);
+
+	downcased = downcase_identifier(ident, len, false, false);
+
+	if (!pg_md5_hash(downcased, strlen(downcased), md5, &errstr))
+		ereport(FATAL,
+				(errcode(ERRCODE_INTERNAL_ERROR),
+				 errmsg("could not compute %s hash: %s", "MD5", errstr)));
+
+	len = pg_mbcliplen(ident, len, NAMEDATALEN - MD5_HASH_LEN - 1);
+	Assert(len + MD5_HASH_LEN < NAMEDATALEN);
+	memcpy(buf, ident, len);
+	memcpy(buf + len, md5, MD5_HASH_LEN);
+	buf[len + MD5_HASH_LEN] = '\0';
+
+	pfree(downcased);
+	memcpy(ident, buf, len + MD5_HASH_LEN + 1);
+}
+
+#undef MD5_HASH_LEN
+
 static int
 ProcessLoginInternal(Port *port)
 {
@@ -1251,13 +1291,19 @@ ProcessLoginInternal(Port *port)
 	loginInfo = request;
 
 	/*
-	 * Truncate given database and user names to length of a Postgres name.
-	 * This avoids lookup failures when overlength names are given.
+	 * Truncate user name to fit Postgres NAMEDATALEN using TSQL-aware
+	 * truncation (clip + MD5 hash) so the result matches the role name
+	 * stored in the pg catalog. We cannot use truncate_identifier_hook here
+	 * because babelfishpg_tsql is not yet loaded at this stage.
+	 * Database name truncation is handled similarly in TdsSetDbContext().
 	 */
-	if (strlen(port->database_name) >= NAMEDATALEN)
-		port->database_name[NAMEDATALEN - 1] = '\0';
 	if (strlen(port->user_name) >= NAMEDATALEN)
-		port->user_name[NAMEDATALEN - 1] = '\0';
+	{
+		tds_truncate_identifier_md5(port->user_name, strlen(port->user_name));
+
+		pfree(loginInfo->username);
+		loginInfo->username = pstrdup(port->user_name);
+	}
 
 	/*
 	 * Done putting stuff in TopMemoryContext.
@@ -2043,6 +2089,9 @@ TdsSetDbContext()
 			 * SQL injection.
 			 */
 			StartTransactionCommand();
+			if (strlen(loginInfo->database) >= NAMEDATALEN)
+				tds_truncate_identifier_md5(loginInfo->database,
+											strlen(loginInfo->database));
 			db_id = pltsql_plugin_handler_ptr->pltsql_get_database_oid(loginInfo->database);
 			CommitTransactionCommand();
 			MemoryContextSwitchTo(oldContext);

contrib/babelfishpg_tsql/runtime/functions.c

@@ -6155,3 +6155,115 @@ openxml_simple(PG_FUNCTION_ARGS)
 #endif							/* USE_LIBXML */
 }
 
+
+PG_FUNCTION_INFO_V1(bbf_xmlquery);
+
+/*
+ * bbf_xmlquery - C implementation of XML .query() 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
+ */
+Datum
+bbf_xmlquery(PG_FUNCTION_ARGS)
+{
+	text	   *xpath_expr;
+	Datum		xml_datum;
+	Oid			arg_type;
+	Oid			immediate_base_type;
+	ArrayType  *namespaces;
+	Datum		xpath_result;
+	ArrayType  *result_arr;
+	Datum	   *elems;
+	bool	   *nulls;
+	int			nitems;
+	StringInfoData buf;
+	int			i;
+
+	xpath_expr = PG_GETARG_TEXT_PP(0);
+	xml_datum = PG_GETARG_DATUM(1);
+
+	/* Lookup the datatype of the supplied argument */
+	arg_type = get_fn_expr_argtype(fcinfo->flinfo, 1);
+
+	/* UDT handling: resolve to immediate base type if it's a UDT */
+	immediate_base_type = get_immediate_base_type_of_UDT_internal(arg_type);
+	if (OidIsValid(immediate_base_type))
+		arg_type = immediate_base_type;
+
+	if (arg_type != XMLOID)
+	{
+		const char *typname = NULL;
+
+		/* Get T-SQL type name for error message */
+		if (common_utility_plugin_ptr)
+			typname = (*common_utility_plugin_ptr->resolve_pg_type_to_tsql)(arg_type);
+		if (typname == NULL)
+			typname = format_type_be(arg_type);
+
+		ereport(ERROR,
+				(errcode(ERRCODE_DATATYPE_MISMATCH),
+				 errmsg("Cannot call methods on %s.", typname)));
+	}
+
+	/* Check QUOTED_IDENTIFIER setting (required for XML methods in T-SQL) */
+	if (!pltsql_quoted_identifier)
+		ereport(ERROR,
+				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
+				 errmsg("SELECT failed because the following SET options have "
+						"incorrect settings: 'QUOTED_IDENTIFIER'. Verify that "
+						"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.
+	 */
+	namespaces = construct_empty_array(TEXTOID);
+	xpath_result = DirectFunctionCall3(xpath,
+									   PointerGetDatum(xpath_expr),
+									   xml_datum,
+									   PointerGetDatum(namespaces));
+
+	result_arr = DatumGetArrayTypeP(xpath_result);
+
+	/* Deconstruct the result array */
+	deconstruct_array(result_arr, XMLOID, -1, false, TYPALIGN_INT,
+					  &elems, &nulls, &nitems);
+
+	/* Empty result → return empty string as XML (matches T-SQL behavior) */
+	if (nitems == 0)
+		PG_RETURN_XML_P((xmltype *) cstring_to_text(""));
+
+	/* Single result → return directly (common fast path) */
+	if (nitems == 1 && !nulls[0])
+		PG_RETURN_DATUM(elems[0]);
+
+	/*
+	 * Multiple results - concatenate all XML fragments.
+	 * Equivalent to: SELECT xmlagg(x) FROM unnest(result_set) AS x
+	 */
+	initStringInfo(&buf);
+	for (i = 0; i < nitems; i++)
+	{
+		if (!nulls[i])
+		{
+			text *fragment = DatumGetTextPP(elems[i]);
+
+			appendBinaryStringInfo(&buf,
+								   VARDATA_ANY(fragment),
+								   VARSIZE_ANY_EXHDR(fragment));
+		}
+	}
+
+	PG_RETURN_XML_P((xmltype *) cstring_to_text_with_len(buf.data, buf.len));
+}

contrib/babelfishpg_tsql/sql/sys_functions.sql

@@ -159,11 +159,17 @@ BEGIN
         result := pg_catalog.replace(result, '"', '"');
         result := pg_catalog.replace(result, '&', '&');
         return result;
-    END IF; 
+    END IF;
 END
 $BODY$
 LANGUAGE plpgsql STABLE STRICT PARALLEL SAFE;
 
+-- helper function for XML QUERY(xpath)
+CREATE OR REPLACE FUNCTION sys.bbf_xmlquery(xpath_pattern TEXT, xml_element ANYELEMENT)
+RETURNS XML
+AS 'babelfishpg_tsql', 'bbf_xmlquery'
+LANGUAGE C 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--6.1.0--6.2.0.sql

@@ -73,6 +73,12 @@ $$
     end;
 $$;
 
+-- helper function for XML QUERY(xpath)
+CREATE OR REPLACE FUNCTION sys.bbf_xmlquery(xpath_pattern TEXT, xml_element ANYELEMENT)
+RETURNS XML
+AS 'babelfishpg_tsql', 'bbf_xmlquery'
+LANGUAGE C STABLE STRICT PARALLEL SAFE;
+
 -- BABELFISH_FUNCTION_EXT (antlr_parse_cache)
 SET allow_system_table_mods = on;
 ALTER TABLE sys.babelfish_function_ext ADD COLUMN IF NOT EXISTS antlr_parse_cache_tree TEXT DEFAULT NULL;

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

@@ -60,6 +60,17 @@ construct_unique_index_name(char *index_name, char *relation_name)
 	index_len = strlen(index_name);
 	relation_len = strlen(relation_name);
 
+	/* Validate original index name against T-SQL 128-char limit */
+	if (pg_mbstrlen_with_len(index_name, index_len) > 128)
+	{
+		int			cliplen = pg_mbcliplen(index_name, index_len, 128);
+
+		ereport(ERROR,
+				(errcode(ERRCODE_NAME_TOO_LONG),
+				 errmsg("The identifier that starts with '%.*s' is too long. Maximum length is 128.",
+						cliplen, index_name)));
+	}
+
 	success = pg_md5_hash(index_name, index_len, md5, &errstr);
 	if (unlikely(!success))
 	{							/* OOM */

contrib/babelfishpg_tsql/src/backend_parser/gram-tsql-prologue.y.h

@@ -6,6 +6,7 @@
 
 #include "access/htup_details.h"
 #include "catalog/pg_type.h"
+#include "mb/pg_wchar.h"
 #include "parser/parse_type.h"
 #include "parser/scansup.h"
 #include "utils/builtins.h"

contrib/babelfishpg_tsql/src/backend_parser/scan-tsql-prologue-top.l.h

@@ -6,3 +6,16 @@
 #include "common/string.h"
 #include "nodes/parsenodes.h"
 #include "src/backend_parser/gramparse.h"
+
+#define CHECK_LOCAL_TEMP_TABLE_LENGTH(ident) \
+	do { \
+		if ((ident)[0] == '#' && (ident)[1] != '#' && \
+			pg_mbstrlen(ident) > 116) \
+		{ \
+			int cliplen = pg_mbcliplen((ident), strlen(ident), 116); \
+			ereport(ERROR, \
+					(errcode(ERRCODE_NAME_TOO_LONG), \
+					 errmsg("The identifier that starts with '%.*s' is too long. Maximum length for local temporary table is 116.", \
+							cliplen, (ident)))); \
+		} \
+	} while (0)

contrib/babelfishpg_tsql/src/backend_parser/scan-tsql-rule.l

@@ -76,6 +76,7 @@
 					if (yyextra->literallen == 0)
 						yyerror("zero-length delimited identifier");
 					ident = litbufdup(yyscanner);
+					CHECK_LOCAL_TEMP_TABLE_LENGTH(ident);
 					if (yyextra->literallen >= NAMEDATALEN)
 						truncate_identifier(ident, yyextra->literallen, true);
 					yylval->str = ident;
@@ -152,6 +153,7 @@
 
 <tsql>{tsql_ttname}	{
 					SET_YYLLOC();
+					CHECK_LOCAL_TEMP_TABLE_LENGTH(yytext);
 					yylval->str = pstrdup(yytext);
 					if (yyleng >= NAMEDATALEN)
 						truncate_identifier(yylval->str, yyleng, true);

contrib/babelfishpg_tsql/src/cursor.c

@@ -194,6 +194,18 @@ pltsql_declare_cursor(PLtsql_execstate *estate, PLtsql_var *var, PLtsql_expr *ex
 	Portal		portal;
 	char		mangled_name[NAMEDATALEN];
 
+	/* Enforce 128-character limit on cursor name */
+	Assert(var->refname != NULL);
+	if (pg_mbstrlen(var->refname) > 128)
+	{
+		int		cliplen = pg_mbcliplen(var->refname, strlen(var->refname), 128);
+
+		ereport(ERROR,
+				(errcode(ERRCODE_NAME_TOO_LONG),
+				 errmsg("The identifier that starts with '%.*s' is too long. Maximum length is 128.",
+						cliplen, var->refname)));
+	}
+
 	if (!var->isnull)
 	{
 		curname = TextDatumGetCString(var->value);
@@ -223,12 +235,6 @@ pltsql_declare_cursor(PLtsql_execstate *estate, PLtsql_var *var, PLtsql_expr *ex
 	 * cursor, its lifespan is longer so we have to use different memory
 	 * context.
 	 */
-	Assert(var->refname != NULL);
-	if (strlen(var->refname) + strlen(LOCAL_CURSOR_INFIX) + 19 > NAMEDATALEN)
-		ereport(ERROR,
-				(errcode(ERRCODE_INTERNAL_ERROR),
-				 errmsg("internal cursor name is too long: %s", var->refname)));
-
 	snprintf(mangled_name, NAMEDATALEN, "%s%s%p", var->refname, LOCAL_CURSOR_INFIX, var);
 
 	assign_text_var(estate, var, mangled_name);

contrib/babelfishpg_tsql/src/multidb.c

@@ -1,5 +1,6 @@
 #include "postgres.h"
 
+#include "mb/pg_wchar.h"
 #include "miscadmin.h"
 #include "nodes/parsenodes.h"
 #include "nodes/primnodes.h"
@@ -1195,6 +1196,17 @@ get_physical_schema_name_by_mode(char *db_name, const char *schema_name, Migrati
 	if (len == 0)
 		return NULL;
 
+	/* Validate original schema name against T-SQL 128-char limit */
+	if (pg_mbstrlen_with_len(schema_name, len) > 128)
+	{
+		int		cliplen = pg_mbcliplen(schema_name, len, 128);
+
+		ereport(ERROR,
+				(errcode(ERRCODE_NAME_TOO_LONG),
+				 errmsg("The identifier that starts with '%.*s' is too long. Maximum length is 128.",
+						cliplen, schema_name)));
+	}
+
 	/* always return a new copy */
 	len = len > MAX_BBF_NAMEDATALEND ? len : MAX_BBF_NAMEDATALEND;
 	name = palloc0(len + 1);