Restrict identifier lengths to 128 characters and local temp tables to 116 characters (#4832)

Restrict identifier lengths to 128 characters and local temp tables to 116 characters

Issues Resolved
BABEL-6434

Authored-by: Anju Bharti <abanju@amazon.com>

Author: anju15bharti

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);
 					return IDENT;
 				}

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);

contrib/babelfishpg_tsql/src/pl_exec-2.c

@@ -4388,11 +4388,14 @@ exec_stmt_partition_function(PLtsql_execstate *estate, PLtsql_stmt_partition_fun
 	 */
 
 	/* check if given name is exceeding the allowed limit */
-	if (strlen(partition_function_name) > 128)
+	if (pg_mbstrlen(partition_function_name) > 128)
 	{
-		ereport(ERROR, 
-			(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
-				errmsg("The identifier that starts with '%.128s' is too long. Maximum length is 128.", partition_function_name)));
+		int		cliplen = pg_mbcliplen(partition_function_name, strlen(partition_function_name), 128);
+
+		ereport(ERROR,
+				(errcode(ERRCODE_NAME_TOO_LONG),
+				 errmsg("The identifier that starts with '%.*s' is too long. Maximum length is 128.",
+						cliplen, partition_function_name)));
 	}
 
 	/*
@@ -4645,12 +4648,14 @@ exec_stmt_partition_scheme(PLtsql_execstate *estate, PLtsql_stmt_partition_schem
 	 */
 
 	/* check if given name is exceeding the allowed limit */
-	if (strlen(partition_scheme_name) > 128)
+	if (pg_mbstrlen(partition_scheme_name) > 128)
 	{
-		ereport(ERROR, 
-			(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
-				errmsg("The identifier that starts with '%.128s' is too long. Maximum length is 128.",
-						partition_scheme_name)));
+		int		cliplen = pg_mbcliplen(partition_scheme_name, strlen(partition_scheme_name), 128);
+
+		ereport(ERROR,
+				(errcode(ERRCODE_NAME_TOO_LONG),
+				 errmsg("The identifier that starts with '%.*s' is too long. Maximum length is 128.",
+						cliplen, partition_scheme_name)));
 	}
 
 	/* raise error if provided partition function doesn't exists in the current database */

contrib/babelfishpg_tsql/src/pl_handler.c

@@ -6170,6 +6170,22 @@ pltsql_truncate_identifier(char *ident, int len, bool warn)
 
 	Assert(len >= NAMEDATALEN); /* should be already checked */
 
+	/*
+	 * Restrict TSQL identifier length to 128 characters.
+	 * Only enforce for user-supplied identifiers (warn=true from scanner).
+	 * Internally constructed identifiers (warn=false) such as combined
+	 * index+table names may exceed 128 chars and should just be truncated.
+	 */
+	if (warn && pg_mbstrlen_with_len(ident, len) > 128)
+	{
+		int		cliplen = pg_mbcliplen(ident, len, 128);
+
+		ereport(ERROR,
+				(errcode(ERRCODE_NAME_TOO_LONG),
+				 errmsg("The identifier that starts with '%.*s' is too long. Maximum length is 128.",
+						cliplen, ident)));
+	}
+
 	if (tsql_is_database_or_server_collation_CI())
 	{
 		/* md5 should be generated by case-insensitive way */
@@ -6216,6 +6232,9 @@ pltsql_cstr_to_name(char *s, int len)
 			bool		success;
 			const char *errstr = NULL;
 
+			/* Identifier should already be validated at the entry point */
+			Assert(pg_mbstrlen_with_len(s, len) <= 128);
+
 			if (tsql_is_database_or_server_collation_CI())
 			{
 				/* md5 should be generated in a case-insensitive way */

contrib/babelfishpg_tsql/src/pltsql_partition.c

@@ -379,6 +379,9 @@ construct_unique_hash(char *relation_name)
 	bool		success;
 	const char	*errstr = NULL;
 
+	/* Identifier should already be validated at the entry point */
+	Assert(pg_mbstrlen_with_len(relation_name, strlen(relation_name)) <= 128);
+
 	md5 = (char *) palloc(MD5_HASH_LEN + 1);
 
 	success = pg_md5_hash(relation_name, strlen(relation_name), md5, &errstr);

test/JDBC/expected/BABEL-3201-vu-verify.out

@@ -655,10 +655,10 @@ DBCC CHECKIDENT('[babel_DBCC_CHECK_schema .with .DOT_and_spaces]."babel_dbcc_CHE
 GO
 
 -- db name longer then 63 and doing cross db call
-CREATE DATABASE babel_dbcc_checkident_database_longer_than_63_0abcdefgij1abcdefgij2abcdefgij3abcdefgij4abcdefgij5abcdefgij6abcdefgij7abcdefgij8abcdefghij9abcdefghij
+CREATE DATABASE babel_dbcc_checkident_database_longer_than_63_0abcdefgij1abcdefgij2abcdefgij3abcdefgij4abcdefgij5abcdefgij6abcdefgij7abcdefgij8a
 GO
 
-USE babel_dbcc_checkident_database_longer_than_63_0abcdefgij1abcdefgij2abcdefgij3abcdefgij4abcdefgij5abcdefgij6abcdefgij7abcdefgij8abcdefghij9abcdefghij
+USE babel_dbcc_checkident_database_longer_than_63_0abcdefgij1abcdefgij2abcdefgij3abcdefgij4abcdefgij5abcdefgij6abcdefgij7abcdefgij8a
 GO
 
 CREATE TABLE babel_3201_longer__name_db_table (a int identity, b int);
@@ -667,15 +667,15 @@ GO
 USE master;
 go
 
-DBCC CHECKIDENT('babel_dbcc_checkident_database_longer_than_63_0abcdefgij1abcdefgij2abcdefgij3abcdefgij4abcdefgij5abcdefgij6abcdefgij7abcdefgij8abcdefghij9abcdefghij.dbo.babel_3201_longer__name_db_table', noreseed);
+DBCC CHECKIDENT('babel_dbcc_checkident_database_longer_than_63_0abcdefgij1abcdefgij2abcdefgij3abcdefgij4abcdefgij5abcdefgij6abcdefgij7abcdefgij8a.dbo.babel_3201_longer__name_db_table', noreseed);
 GO
 
 -- db name longer and mixed case as well. 
-DBCC CHECKIDENT('bAbEl_dBcC_ChEcKiDeNt_dAtAbAsE_LoNgEr_tHaN_63_0AbCdEfGiJ1AbCdEfGiJ2AbCdEfGiJ3AbCdEfGiJ4AbCdEfGiJ5AbCdEfGiJ6AbCdEfGiJ7AbCdEfGiJ8AbCdEfGhIj9aBcDeFgHiJ.dbo.babel_3201_longer__name_db_table', noreseed);
+DBCC CHECKIDENT('babel_dbcc_checkident_database_longer_than_63_0abcdefgij1abcdefgij2abcdefgij3abcdefgij4abcdefgij5abcdefgij6abcdefgij7abcdefgij8a.dbo.babel_3201_longer__name_db_table', noreseed);
 GO
 
 -- drop this db because of single_db mode
-DROP DATABASE babel_dbcc_checkident_database_longer_than_63_0abcdefgij1abcdefgij2abcdefgij3abcdefgij4abcdefgij5abcdefgij6abcdefgij7abcdefgij8abcdefghij9abcdefghij
+DROP DATABASE babel_dbcc_checkident_database_longer_than_63_0abcdefgij1abcdefgij2abcdefgij3abcdefgij4abcdefgij5abcdefgij6abcdefgij7abcdefgij8a
 GO
 
 -- create database to test cross db behavior.