Added ANTLR parse tree persistent caching for procedures

.gitignore

@@ -45,3 +45,9 @@ contrib/babelfishpg_tsql/src/pl_gram.c
 contrib/babelfishpg_tsql/src/pl_gram.h
 contrib/babelfishpg_tsql/src/pl_gram.output
 
+# PLtsql serialization - auto-generated by gen_pltsql_node_support.pl
+contrib/babelfishpg_tsql/src/pltsql_serialize/pltsql_nodetags.h
+contrib/babelfishpg_tsql/src/pltsql_serialize/pltsql_outfuncs_gen.c
+contrib/babelfishpg_tsql/src/pltsql_serialize/pltsql_outfuncs_switch.c
+contrib/babelfishpg_tsql/src/pltsql_serialize/pltsql_readfuncs_gen.c
+contrib/babelfishpg_tsql/src/pltsql_serialize/pltsql_readfuncs_switch.c

contrib/babelfishpg_tsql/Makefile

@@ -86,6 +86,29 @@ export ANTLR4_RUNTIME_INCLUDE_DIR=/usr/local/include/antlr4-runtime
 export ANTLR4_RUNTIME_LIB_DIR=/usr/local/lib
 
 OBJS += src/pltsql_bulkcopy.o
+OBJS += src/pltsql_serialize/pltsql_node_stubs.o
+
+# PLtsql serialization code generation (babelfishpg_tsql.enable_routine_parse_cache)
+PLTSQL_SER_DIR = src/pltsql_serialize
+PLTSQL_SER_HEADERS = $(PLTSQL_SER_DIR)/pltsql_serializable_1.h $(PLTSQL_SER_DIR)/pltsql_serializable_2.h
+PLTSQL_GEN_SCRIPT = $(PLTSQL_SER_DIR)/gen_pltsql_node_support.pl
+
+$(PLTSQL_SER_DIR)/pltsql_nodetags.h $(PLTSQL_SER_DIR)/pltsql_outfuncs_gen.c $(PLTSQL_SER_DIR)/pltsql_readfuncs_gen.c $(PLTSQL_SER_DIR)/pltsql_outfuncs_switch.c $(PLTSQL_SER_DIR)/pltsql_readfuncs_switch.c: $(PLTSQL_SER_HEADERS) $(PLTSQL_GEN_SCRIPT)
+	$(PERL) $(PLTSQL_GEN_SCRIPT) --outdir $(PLTSQL_SER_DIR) $(PLTSQL_SER_HEADERS)
+
+# Wrapper .c files #include the generated .c files (mirroring engine pattern)
+# so we compile the wrappers, not the gen files directly.
+OBJS += $(PLTSQL_SER_DIR)/pltsql_outfuncs.o
+OBJS += $(PLTSQL_SER_DIR)/pltsql_readfuncs.o
+
+# pltsql_nodetags.h must be generated before any .o that includes pltsql.h
+# The first OBJS entry (src/pl_gram.o) triggers this via implicit ordering,
+# but we make it explicit for safety.
+src/pl_gram.o: $(PLTSQL_SER_DIR)/pltsql_nodetags.h
+
+# Wrapper .o depends on generated .c files (included via #include)
+$(PLTSQL_SER_DIR)/pltsql_outfuncs.o: $(PLTSQL_SER_DIR)/pltsql_outfuncs_gen.c $(PLTSQL_SER_DIR)/pltsql_outfuncs_switch.c
+$(PLTSQL_SER_DIR)/pltsql_readfuncs.o: $(PLTSQL_SER_DIR)/pltsql_readfuncs_gen.c $(PLTSQL_SER_DIR)/pltsql_readfuncs_switch.c
 
 PG_CXXFLAGS += -g -Werror -Wfloat-conversion
 PG_CXXFLAGS += -Wno-deprecated -Wno-error=attributes -Wno-suggest-attribute=format # disable some warnings from ANTLR runtime header

contrib/babelfishpg_tsql/sql/ownership.sql

@@ -41,6 +41,11 @@ CREATE TABLE sys.babelfish_function_ext (
 	create_date SYS.DATETIME NOT NULL,
 	modify_date SYS.DATETIME NOT NULL,
 	definition sys.NTEXT DEFAULT NULL,
+	antlr_parse_tree_text TEXT DEFAULT NULL, -- Native PG nodeToString() serialized parse tree
+	antlr_parse_tree_datums TEXT DEFAULT NULL, -- Native PG nodeToString() serialized datums array
+  antlr_parse_tree_modify_date SYS.DATETIME DEFAULT NULL,
+	antlr_parse_tree_bbf_version TEXT DEFAULT NULL,
+	antlr_cache_enabled BOOL DEFAULT false,
 	PRIMARY KEY(funcname, nspname, funcsignature)
 );
 GRANT SELECT ON sys.babelfish_function_ext TO PUBLIC;

contrib/babelfishpg_tsql/sql/sys_functions.sql

@@ -5312,3 +5312,10 @@ RETURNS table (
 AS 'babelfishpg_tsql', 'openxml_simple'
 LANGUAGE C IMMUTABLE;
 
+
+-- Function-specific ANTLR parse tree cache GUC control
+CREATE OR REPLACE FUNCTION sys.enable_routine_parse_cache(
+    IN func_identifier TEXT,
+    IN enable_flag BOOLEAN
+) RETURNS BOOLEAN
+AS 'babelfishpg_tsql', 'enable_routine_parse_cache' LANGUAGE C;

contrib/babelfishpg_tsql/sql/upgrades/babelfishpg_tsql--5.5.0--5.6.0.sql

@@ -231,6 +231,21 @@ $$
     end;
 $$;
 
+-- BABELFISH_FUNCTION_EXT
+SET allow_system_table_mods = on;
+ALTER TABLE sys.babelfish_function_ext ADD COLUMN IF NOT EXISTS antlr_parse_tree_text TEXT DEFAULT NULL;
+ALTER TABLE sys.babelfish_function_ext ADD COLUMN IF NOT EXISTS antlr_parse_tree_datums TEXT DEFAULT NULL;
+ALTER TABLE sys.babelfish_function_ext ADD COLUMN IF NOT EXISTS antlr_parse_tree_modify_date SYS.DATETIME DEFAULT NULL;
+ALTER TABLE sys.babelfish_function_ext ADD COLUMN IF NOT EXISTS antlr_parse_tree_bbf_version TEXT DEFAULT NULL;
+ALTER TABLE sys.babelfish_function_ext ADD COLUMN IF NOT EXISTS antlr_cache_enabled BOOL DEFAULT false;
+RESET allow_system_table_mods;
+
+CREATE OR REPLACE FUNCTION sys.enable_routine_parse_cache(
+    IN func_identifier TEXT,
+    IN enable_flag BOOLEAN
+) RETURNS BOOLEAN
+AS 'babelfishpg_tsql', 'enable_routine_parse_cache' LANGUAGE C;
+
 -- Please add your SQLs here
 
 -- Deprecate and drop old aggregates first (they depend on the function)

contrib/babelfishpg_tsql/src/catalog.c

@@ -3655,6 +3655,18 @@ rename_procfunc_update_bbf_catalog(RenameStmt *stmt)
 		new_record_repl_func_ext[Anum_bbf_function_ext_orig_name - 1] = true;
 	}
 
+	/* Invalidate cached parse tree — procedure name is embedded in the parse tree namespace stack */
+	/* Alternatively, if guc enabled, explore if worth looking up session cache PLtsql_HashTable for
+	 * updated parse tree result and serialize and re-populate babelfish_function_ext's antlr_parse_tree column */
+	new_record_nulls_func_ext[Anum_bbf_function_ext_antlr_parse_tree_text - 1] = true;
+	new_record_repl_func_ext[Anum_bbf_function_ext_antlr_parse_tree_text - 1] = true;
+	new_record_nulls_func_ext[Anum_bbf_function_ext_antlr_parse_tree_datums - 1] = true;
+	new_record_repl_func_ext[Anum_bbf_function_ext_antlr_parse_tree_datums - 1] = true;
+	new_record_nulls_func_ext[Anum_bbf_function_ext_antlr_parse_tree_modify_date - 1] = true;
+	new_record_repl_func_ext[Anum_bbf_function_ext_antlr_parse_tree_modify_date - 1] = true;
+	new_record_nulls_func_ext[Anum_bbf_function_ext_antlr_parse_tree_bbf_version - 1] = true;
+	new_record_repl_func_ext[Anum_bbf_function_ext_antlr_parse_tree_bbf_version - 1] = true;
+
 	new_tuple = heap_modify_tuple(usertuple,
 								  bbf_func_ext_dsc,
 								  new_record_func_ext,
@@ -6490,4 +6502,179 @@ bbf_check_member_has_direct_priv_to_grant_role(Oid member, Oid role)
 
 	ReleaseSysCacheList(memlist);
 	return false;
-}
+}
+
+/*
+ * Helper: update antlr_cache_enabled flag and optionally clear cache columns
+ * in babelfish_function_ext for a given bbf function tuple.
+ */
+static void
+update_bbf_function_cache_enabled(HeapTuple bbffunctuple, bool enable_flag)
+{
+	Relation	rel;
+	TupleDesc	dsc;
+	HeapTuple	newtup;
+	Datum		new_record[BBF_FUNCTION_EXT_NUM_COLS];
+	bool		new_record_nulls[BBF_FUNCTION_EXT_NUM_COLS];
+	bool		new_record_replaces[BBF_FUNCTION_EXT_NUM_COLS];
+
+	rel = table_open(get_bbf_function_ext_oid(), RowExclusiveLock);
+	dsc = RelationGetDescr(rel);
+
+	MemSet(new_record, 0, sizeof(new_record));
+	MemSet(new_record_nulls, false, sizeof(new_record_nulls));
+	MemSet(new_record_replaces, false, sizeof(new_record_replaces));
+
+	/* Set antlr_cache_enabled */
+	new_record[Anum_bbf_function_ext_antlr_cache_enabled - 1] = BoolGetDatum(enable_flag);
+	new_record_replaces[Anum_bbf_function_ext_antlr_cache_enabled - 1] = true;
+
+	/* If disabling, NULL out the 4 cache columns for immediate invalidation */
+	if (!enable_flag)
+	{
+		new_record_nulls[Anum_bbf_function_ext_antlr_parse_tree_text - 1] = true;
+		new_record_replaces[Anum_bbf_function_ext_antlr_parse_tree_text - 1] = true;
+		new_record_nulls[Anum_bbf_function_ext_antlr_parse_tree_datums - 1] = true;
+		new_record_replaces[Anum_bbf_function_ext_antlr_parse_tree_datums - 1] = true;
+		new_record_nulls[Anum_bbf_function_ext_antlr_parse_tree_modify_date - 1] = true;
+		new_record_replaces[Anum_bbf_function_ext_antlr_parse_tree_modify_date - 1] = true;
+		new_record_nulls[Anum_bbf_function_ext_antlr_parse_tree_bbf_version - 1] = true;
+		new_record_replaces[Anum_bbf_function_ext_antlr_parse_tree_bbf_version - 1] = true;
+	}
+
+	newtup = heap_modify_tuple(bbffunctuple, dsc,
+							   new_record, new_record_nulls, new_record_replaces);
+	CatalogTupleUpdate(rel, &newtup->t_self, newtup);
+
+	heap_freetuple(newtup);
+	table_close(rel, RowExclusiveLock);
+}
+
+/*
+ * enable_routine_parse_cache
+ *
+ * SQL-callable: sys.enable_routine_parse_cache(func_identifier TEXT, enable_flag BOOLEAN)
+ *
+ * Enables or disables ANTLR parse result caching for a specific function.
+ *
+ * func_identifier formats:
+ *   'schema.funcname'              — unique function (no overloads)
+ *   'funcname'                     — defaults to dbo schema
+ *   'schema.funcname(argtypes)'    — disambiguate overloads, e.g. 'dbo.myProc("sys"."varchar")'
+ *
+ * Uses PROCNAMENSPSIGNATURE syscache (funcname, nspname, funcsignature).
+ * When no arg types given, uses 2-key list lookup; errors if ambiguous.
+ */
+PG_FUNCTION_INFO_V1(enable_routine_parse_cache);
+Datum
+enable_routine_parse_cache(PG_FUNCTION_ARGS)
+{
+	text	   *func_id_text = PG_GETARG_TEXT_PP(0);
+	bool		enable_flag = PG_GETARG_BOOL(1);
+	char	   *func_id = text_to_cstring(func_id_text);
+	char	   *dot;
+	char	   *schema_name;
+	char	   *func_part;
+	char	   *funcname_str;
+	char	   *paren;
+	NameData	nsp_name;
+	NameData	funcname_data;
+
+	/* Split schema from func_part on first dot */
+	dot = strchr(func_id, '.');
+	if (dot != NULL)
+	{
+		*dot = '\0';
+		schema_name = func_id;
+		func_part = dot + 1;
+	}
+	else
+	{
+		schema_name = "dbo";
+		func_part = func_id;
+	}
+
+	/* Extract bare funcname (everything before '(' if present) */
+	funcname_str = pstrdup(func_part);
+	paren = strchr(funcname_str, '(');
+	if (paren != NULL)
+		*paren = '\0';
+
+	/* Convert logical schema name (e.g. 'dbo') to physical (e.g. 'master_dbo') */
+	{
+		char	   *cur_db = get_cur_db_name();
+		char	   *physical_schema = get_physical_schema_name(cur_db, schema_name);
+
+		namestrcpy(&nsp_name, physical_schema);
+		pfree(cur_db);
+		pfree(physical_schema);
+	}
+	namestrcpy(&funcname_data, funcname_str);
+
+	if (strchr(func_part, '(') != NULL)
+	{
+		/* Full signature provided — exact 3-key lookup */
+		HeapTuple	bbffunctuple;
+
+		bbffunctuple = SearchSysCache3(PROCNAMENSPSIGNATURE,
+									   NameGetDatum(&funcname_data),
+									   NameGetDatum(&nsp_name),
+									   CStringGetTextDatum(func_part));
+
+		if (!HeapTupleIsValid(bbffunctuple))
+			ereport(ERROR,
+					(errcode(ERRCODE_UNDEFINED_OBJECT),
+					 errmsg("function \"%s\" not found in schema \"%s\"",
+							func_part, schema_name),
+					 errhint("Use sys.babelfish_get_pltsql_function_signature(oid) to find the exact signature.")));
+
+		{
+			HeapTuple	copytup = heap_copytuple(bbffunctuple);
+
+			ReleaseSysCache(bbffunctuple);
+			update_bbf_function_cache_enabled(copytup, enable_flag);
+			heap_freetuple(copytup);
+		}
+	}
+	else
+	{
+		/* No arg types — 2-key list lookup */
+		CatCList   *catlist;
+
+		catlist = SearchSysCacheList2(PROCNAMENSPSIGNATURE,
+									  NameGetDatum(&funcname_data),
+									  NameGetDatum(&nsp_name));
+
+		if (catlist->n_members == 0)
+		{
+			ReleaseSysCacheList(catlist);
+			ereport(ERROR,
+					(errcode(ERRCODE_UNDEFINED_OBJECT),
+					 errmsg("function \"%s\" not found in schema \"%s\"",
+							funcname_str, schema_name)));
+		}
+		else if (catlist->n_members > 1)
+		{
+			ReleaseSysCacheList(catlist);
+			ereport(ERROR,
+					(errcode(ERRCODE_AMBIGUOUS_FUNCTION),
+					 errmsg("multiple overloads of \"%s\" exist in schema \"%s\"",
+							funcname_str, schema_name),
+					 errhint("Specify the full signature, e.g. 'dbo.%s(integer)'. "
+							 "Use sys.babelfish_get_pltsql_function_signature(oid) to find the exact signature.",
+							 funcname_str)));
+		}
+		else
+		{
+			HeapTuple	copytup = heap_copytuple(&catlist->members[0]->tuple);
+
+			ReleaseSysCacheList(catlist);
+			update_bbf_function_cache_enabled(copytup, enable_flag);
+			heap_freetuple(copytup);
+		}
+	}
+
+	pfree(funcname_str);
+	pfree(func_id);
+	PG_RETURN_BOOL(enable_flag);
+}

contrib/babelfishpg_tsql/src/catalog.h

@@ -277,7 +277,12 @@ typedef FormData_bbf_servers_def *Form_bbf_servers_def;
 #define Anum_bbf_function_ext_create_date 8
 #define Anum_bbf_function_ext_modify_date 9
 #define Anum_bbf_function_ext_definition 10
-#define BBF_FUNCTION_EXT_NUM_COLS 10
+#define Anum_bbf_function_ext_antlr_parse_tree_text 11
+#define Anum_bbf_function_ext_antlr_parse_tree_datums 12
+#define Anum_bbf_function_ext_antlr_parse_tree_modify_date 13
+#define Anum_bbf_function_ext_antlr_parse_tree_bbf_version 14
+#define Anum_bbf_function_ext_antlr_cache_enabled 15
+#define BBF_FUNCTION_EXT_NUM_COLS 15
 #define FLAG_IS_ANSI_NULLS_ON (1<<0)
 #define FLAG_USES_QUOTED_IDENTIFIER (1<<1)
 #define FLAG_CREATED_WITH_RECOMPILE (1<<2)

contrib/babelfishpg_tsql/src/codegen.c

@@ -147,7 +147,7 @@ create_goto(int lineno)
 {
 	PLtsql_stmt_goto *stmt_goto;
 
-	stmt_goto = palloc(sizeof(PLtsql_stmt_goto));
+	stmt_goto = makeNode(PLtsql_stmt_goto);
 	stmt_goto->cmd_type = PLTSQL_STMT_GOTO;
 	stmt_goto->lineno = lineno;
 	stmt_goto->cond = NULL;		/* unconditional goto */
@@ -159,7 +159,7 @@ create_goto(int lineno)
 static PLtsql_stmt_save_ctx *
 create_save_ctx(int lineno)
 {
-	PLtsql_stmt_save_ctx *save_ctx = palloc(sizeof(PLtsql_stmt_save_ctx));
+	PLtsql_stmt_save_ctx *save_ctx = makeNode(PLtsql_stmt_save_ctx);
 
 	save_ctx->cmd_type = PLTSQL_STMT_SAVE_CTX;
 	save_ctx->lineno = lineno;
@@ -171,7 +171,7 @@ create_save_ctx(int lineno)
 static PLtsql_stmt_restore_ctx_full *
 create_restore_ctx_full(int lineno)
 {
-	PLtsql_stmt_restore_ctx_full *restore_ctx = palloc(sizeof(PLtsql_stmt_restore_ctx_full));
+	PLtsql_stmt_restore_ctx_full *restore_ctx = makeNode(PLtsql_stmt_restore_ctx_full);
 
 	restore_ctx->cmd_type = PLTSQL_STMT_RESTORE_CTX_FULL;
 	restore_ctx->lineno = lineno;
@@ -181,7 +181,7 @@ create_restore_ctx_full(int lineno)
 static PLtsql_stmt_restore_ctx_partial *
 create_restore_ctx_partial(int lineno)
 {
-	PLtsql_stmt_restore_ctx_partial *restore_ctx = palloc(sizeof(PLtsql_stmt_restore_ctx_partial));
+	PLtsql_stmt_restore_ctx_partial *restore_ctx = makeNode(PLtsql_stmt_restore_ctx_partial);
 
 	restore_ctx->cmd_type = PLTSQL_STMT_RESTORE_CTX_PARTIAL;
 	restore_ctx->lineno = lineno;

contrib/babelfishpg_tsql/src/guc.c

@@ -72,6 +72,7 @@ char	   *pltsql_host_service_pack_level = NULL;
 
 bool		pltsql_enable_create_alter_view_from_pg = false;
 bool		pltsql_enable_alter_owner_from_pg = false;
+bool		pltsql_enable_routine_parse_cache = false;
 
 static const struct config_enum_entry explain_format_options[] = {
 	{"text", EXPLAIN_FORMAT_TEXT, false},
@@ -1140,6 +1141,18 @@ define_custom_variables(void)
 							 GUC_NOT_IN_SAMPLE | GUC_DISALLOW_IN_FILE | GUC_DISALLOW_IN_AUTO_FILE,
 							 NULL, NULL, NULL);
 
+	/*
+	 * Enable/disable procedure ANTLR parse result caching for cross-session (sys.babelfish_function_ext->antlr_parse_tree) performance optimization.
+	 */
+	DefineCustomBoolVariable("babelfishpg_tsql.enable_routine_parse_cache",
+							 gettext_noop("Enables caching of ANTLR parse results for stored procedures across sessions"),
+							 NULL,
+							 &pltsql_enable_routine_parse_cache,
+							 false,
+							 PGC_USERSET,
+							 GUC_NOT_IN_SAMPLE | GUC_DISALLOW_IN_FILE | GUC_DISALLOW_IN_AUTO_FILE,
+							 NULL, NULL, NULL);
+
 	/* Dump and Restore */
 	DefineCustomBoolVariable("babelfishpg_tsql.dump_restore",
 							 gettext_noop("Enable special handlings during dump and restore"),

contrib/babelfishpg_tsql/src/guc.h

@@ -18,6 +18,7 @@ typedef enum IsolationOptions
 extern bool pltsql_fmtonly;
 extern bool pltsql_enable_create_alter_view_from_pg;
 extern bool	pltsql_enable_alter_owner_from_pg;
+extern bool pltsql_enable_routine_parse_cache;
 extern bool pltsql_enable_linked_servers;
 extern bool pltsql_enable_ownership_chaining;
 extern bool pltsql_allow_windows_login;