From: Michael Wallner <mike@php.net>
Date: Fri, 2 Nov 2018 06:36:31 +0000 (+0100)
Subject: interned strings
X-Git-Url: https://git.m6w6.name/?a=commitdiff_plain;h=2559f07a178b681dc240e20153ff5abdff54cb34;p=m6w6%2Fext-psi

interned strings
---

diff --git a/Makefile.frag b/Makefile.frag
index 1cd17bd..7ad6301 100644
--- a/Makefile.frag
+++ b/Makefile.frag
@@ -36,7 +36,8 @@ psi-generated: $(PHP_PSI_GENERATED)
 PHP_PSI_DEPEND = $(patsubst $(PHP_PSI_SRCDIR)/%,$(PHP_PSI_BUILDDIR)/%,$(PHP_PSI_SOURCES:.c=.dep))
 
 $(PHP_PSI_BUILDDIR)/%.dep: $(PHP_PSI_SRCDIR)/%.c | $(PHP_PSI_GENERATED)
-	$(CC) -MM -MG -MF $@ -MT $(patsubst $(PHP_PSI_SRCDIR)/%,$(PHP_PSI_BUILDDIR)/%,$(@:.dep=.lo)) \
+	@echo Generating $@ ...
+	@$(CC) -MM -MG -MF $@ -MT $(patsubst $(PHP_PSI_SRCDIR)/%,$(PHP_PSI_BUILDDIR)/%,$(@:.dep=.lo)) \
 		$(CPPFLAGS) $(DEFS) $(INCLUDES) $< \
 			|| touch $@
 
diff --git a/src/call.c b/src/call.c
index c5aa4b7..22516c7 100644
--- a/src/call.c
+++ b/src/call.c
@@ -199,7 +199,7 @@ zval *psi_call_frame_sub_argument(struct psi_call_frame *frame,
 	if (!iarg) {
 		struct psi_call_frame_argument *frame_arg;
 		impl_val empty_val = {0};
-		zend_string *type_str = zend_string_init(ZEND_STRL("mixed"), 1);
+		zend_string *type_str = zend_string_init_interned(ZEND_STRL("mixed"), 1);
 		struct psi_impl_arg *carg_spec = psi_impl_arg_init(
 				psi_impl_type_init(PSI_T_MIXED, type_str),
 				psi_impl_var_copy(inner_var), NULL);
diff --git a/src/context.c b/src/context.c
index da36c8e..8fdbf22 100644
--- a/src/context.c
+++ b/src/context.c
@@ -196,6 +196,9 @@ zend_function_entry *psi_context_compile(struct psi_context *C)
 			case PSI_T_STRING:
 			case PSI_T_QUOTED_STRING:
 				ZVAL_NEW_STR(&zc.value, zend_string_copy(c->val->ival.zend.str));
+				if (ZSTR_IS_INTERNED(Z_STR(zc.value))) {
+					Z_TYPE_FLAGS(zc.value) = 0;
+				}
 				break;
 			default:
 				assert(0);
diff --git a/src/cpp.c b/src/cpp.c
index 64a1b55..1300d57 100644
--- a/src/cpp.c
+++ b/src/cpp.c
@@ -173,7 +173,7 @@ static bool psi_cpp_stage1(struct psi_cpp *cpp)
 
 					no_ws->type = PSI_T_NO_WHITESPACE;
 					zend_string_release(no_ws->text);
-					no_ws->text = zend_string_init("\xA0", 1, 1);
+					no_ws->text = zend_string_init_interned("\xA0", 1, 1);
 					psi_cpp_tokiter_add(cpp, no_ws);
 					continue;
 				}
diff --git a/src/parser.c b/src/parser.c
index dfeb970..47b89cb 100644
--- a/src/parser.c
+++ b/src/parser.c
@@ -1,4 +1,4 @@
-/* Generated by re2c 1.1.1 on Thu Nov  1 08:10:31 2018 */
+/* Generated by re2c 1.1.1 on Fri Nov  2 07:35:40 2018 */
 #line 1 "src/parser.re"
 /*******************************************************************************
  Copyright (c) 2016, Michael Wallner <mike@php.net>.
@@ -103,7 +103,7 @@ struct psi_parser_input *psi_parser_open_file(struct psi_parser *P, const char *
 	}
 
 	fb->length = sb.st_size;
-	fb->file = zend_string_init(filename, strlen(filename), 1);
+	fb->file = zend_string_init_interned(filename, strlen(filename), 1);
 
 	return fb;
 }
@@ -123,7 +123,7 @@ struct psi_parser_input *psi_parser_open_string(struct psi_parser *P, const char
 	memset(sb->buffer + length, 0, YYMAXFILL);
 
 	sb->length = length;
-	sb->file = zend_string_init("<stdin>", strlen("<stdin>"), 1);
+	sb->file = zend_string_init_interned("<stdin>", strlen("<stdin>"), 1);
 
 	return sb;
 }
@@ -172,7 +172,7 @@ void psi_parser_postprocess(struct psi_parser *P)
 				smart_str_appendl_ex(&ns_name, ZEND_STRL("psi\\"), 1);
 				smart_str_append_ex(&ns_name, name, 1);
 				name_str = smart_str_extract(&ns_name);
-				type_str = zend_string_init(ZEND_STRL("<eval number>"), 1);
+				type_str = zend_string_init_interned(ZEND_STRL("<eval number>"), 1);
 
 				num = psi_num_exp_copy(scope.macro->exp);
 				def = psi_impl_def_val_init(PSI_T_NUMBER, num);
@@ -197,7 +197,7 @@ void psi_parser_postprocess(struct psi_parser *P)
 						smart_str_appendl_ex(&ns_name, ZEND_STRL("psi\\"), 1);
 						smart_str_append_ex(&ns_name, name, 1);
 						name_str = smart_str_extract(&ns_name);
-						type_str = zend_string_init(ZEND_STRL("string"), 1);
+						type_str = zend_string_init_interned(ZEND_STRL("string"), 1);
 
 						type = psi_impl_type_init(PSI_T_STRING, type_str);
 						def = psi_impl_def_val_init(PSI_T_QUOTED_STRING, t->text);
@@ -264,17 +264,17 @@ void psi_parser_free(struct psi_parser **P)
 	++I->lines
 
 #define NEWTOKEN(t) \
-	token = psi_token_init(t, tok, cur - tok, tok - eol + 1, I->lines, I->file); \
+	if (t == PSI_T_COMMENT || t == PSI_T_WHITESPACE) { \
+		token = psi_token_init(t, "", 0, tok - eol + 1, I->lines, I->file); \
+	} else { \
+		token = psi_token_init(t, tok, cur - tok, tok - eol + 1, I->lines, I->file); \
+	} \
 	tokens = psi_plist_add(tokens, &token); \
 	if (P->flags & PSI_DEBUG) { \
 		fprintf(stderr, "PSI< "); \
 		psi_token_dump(2, token); \
 	}
 
-union int_suffix {
-	char s[4];
-	uint32_t i;
-};
 
 struct psi_plist *psi_parser_scan(struct psi_parser *P, struct psi_parser_input *I)
 {
diff --git a/src/parser.re b/src/parser.re
index 926cc7c..72587f0 100644
--- a/src/parser.re
+++ b/src/parser.re
@@ -100,7 +100,7 @@ struct psi_parser_input *psi_parser_open_file(struct psi_parser *P, const char *
 	}
 
 	fb->length = sb.st_size;
-	fb->file = zend_string_init(filename, strlen(filename), 1);
+	fb->file = zend_string_init_interned(filename, strlen(filename), 1);
 
 	return fb;
 }
@@ -120,7 +120,7 @@ struct psi_parser_input *psi_parser_open_string(struct psi_parser *P, const char
 	memset(sb->buffer + length, 0, YYMAXFILL);
 
 	sb->length = length;
-	sb->file = zend_string_init("<stdin>", strlen("<stdin>"), 1);
+	sb->file = zend_string_init_interned("<stdin>", strlen("<stdin>"), 1);
 
 	return sb;
 }
@@ -169,7 +169,7 @@ void psi_parser_postprocess(struct psi_parser *P)
 				smart_str_appendl_ex(&ns_name, ZEND_STRL("psi\\"), 1);
 				smart_str_append_ex(&ns_name, name, 1);
 				name_str = smart_str_extract(&ns_name);
-				type_str = zend_string_init(ZEND_STRL("<eval number>"), 1);
+				type_str = zend_string_init_interned(ZEND_STRL("<eval number>"), 1);
 
 				num = psi_num_exp_copy(scope.macro->exp);
 				def = psi_impl_def_val_init(PSI_T_NUMBER, num);
@@ -194,7 +194,7 @@ void psi_parser_postprocess(struct psi_parser *P)
 						smart_str_appendl_ex(&ns_name, ZEND_STRL("psi\\"), 1);
 						smart_str_append_ex(&ns_name, name, 1);
 						name_str = smart_str_extract(&ns_name);
-						type_str = zend_string_init(ZEND_STRL("string"), 1);
+						type_str = zend_string_init_interned(ZEND_STRL("string"), 1);
 
 						type = psi_impl_type_init(PSI_T_STRING, type_str);
 						def = psi_impl_def_val_init(PSI_T_QUOTED_STRING, t->text);
@@ -261,17 +261,17 @@ void psi_parser_free(struct psi_parser **P)
 	++I->lines
 
 #define NEWTOKEN(t) \
-	token = psi_token_init(t, tok, cur - tok, tok - eol + 1, I->lines, I->file); \
+	if (t == PSI_T_COMMENT || t == PSI_T_WHITESPACE) { \
+		token = psi_token_init(t, "", 0, tok - eol + 1, I->lines, I->file); \
+	} else { \
+		token = psi_token_init(t, tok, cur - tok, tok - eol + 1, I->lines, I->file); \
+	} \
 	tokens = psi_plist_add(tokens, &token); \
 	if (P->flags & PSI_DEBUG) { \
 		fprintf(stderr, "PSI< "); \
 		psi_token_dump(2, token); \
 	}
 
-union int_suffix {
-	char s[4];
-	uint32_t i;
-};
 
 struct psi_plist *psi_parser_scan(struct psi_parser *P, struct psi_parser_input *I)
 {
diff --git a/src/parser_proc_grammar.y b/src/parser_proc_grammar.y
index 32e673b..1b40555 100644
--- a/src/parser_proc_grammar.y
+++ b/src/parser_proc_grammar.y
@@ -947,7 +947,7 @@ decl_type_simple[type]:
 }
 ;
 
-decl_real_type[type]:
+decl_real_type[type]: /* allocated, so free, if set */
 	FLOAT[type_] {
 	$type = psi_token_copy($type_);
 }
@@ -964,7 +964,7 @@ int_signed[i]:
 |	UNSIGNED
 ;
 
-int_width[i]:
+int_width[i]: /* allocated, so free, if set */
 	SHORT {
 	$i = psi_token_copy($SHORT);
 }
@@ -976,7 +976,7 @@ int_width[i]:
 }
 ;
 
-decl_int_type[type]:
+decl_int_type[type]: /* allocated, so free, if set */
 	CHAR {
 	$type = psi_token_copy($CHAR);
 }
@@ -1002,7 +1002,7 @@ decl_int_type[type]:
 }
 ;
 
-int_signed_types[type]:
+int_signed_types[type]: /* allocated, so free, if set */
 	%empty {
 	$type = NULL;
 }
@@ -1046,7 +1046,7 @@ signed_long_types[type]:
 |	LONG INT
 ;
 
-int_width_types[type]:
+int_width_types[type]: /* allocated, so free, if set */
 	%empty {
 	$type = NULL;
 }
diff --git a/src/token.c b/src/token.c
index 5dbb2ec..316c734 100644
--- a/src/token.c
+++ b/src/token.c
@@ -46,7 +46,7 @@ struct psi_token *psi_token_init(token_t token_typ, const char *token_txt,
 	T->col = col;
 	T->line = line;
 	T->file = zend_string_copy(file);
-	T->text = zend_string_init(token_txt, token_len, 1);
+	T->text = zend_string_init_interned(token_txt, token_len, 1);
 #if PSI_DEBUG_TOKEN_ALLOC
 	fprintf(stderr, "PSI: token_init %p\n", T);
 #endif
diff --git a/src/types/decl_abi.c b/src/types/decl_abi.c
index bf05239..794005b 100644
--- a/src/types/decl_abi.c
+++ b/src/types/decl_abi.c
@@ -31,7 +31,7 @@ struct psi_decl_abi *psi_decl_abi_init(zend_string *convention)
 	struct psi_decl_abi *abi = calloc(1, sizeof(*abi));
 	abi->convention = convention
 			? zend_string_copy(convention)
-			: zend_string_init(ZEND_STRL("default"), 1);
+			: zend_string_init_interned(ZEND_STRL("default"), 1);
 	return abi;
 }
 
diff --git a/src/types/decl_extvar.c b/src/types/decl_extvar.c
index fc4a5ef..c6a77aa 100644
--- a/src/types/decl_extvar.c
+++ b/src/types/decl_extvar.c
@@ -115,7 +115,7 @@ void psi_decl_extvar_dump(int fd, struct psi_decl_extvar *evar)
 
 struct psi_decl *psi_decl_extvar_setter(struct psi_decl_extvar *evar)
 {
-	zend_string *type_str = zend_string_init(ZEND_STRS("void"), 1);
+	zend_string *type_str = zend_string_init_interned(ZEND_STRS("void"), 1);
 	struct psi_decl_type *func_type = psi_decl_type_init(PSI_T_VOID, type_str);
 	struct psi_decl_var *func_var = psi_decl_var_copy(evar->arg->var);
 	struct psi_decl_arg *func = psi_decl_arg_init(func_type, func_var);
diff --git a/src/types/impl_def_val.c b/src/types/impl_def_val.c
index 311b39d..f78a6b9 100644
--- a/src/types/impl_def_val.c
+++ b/src/types/impl_def_val.c
@@ -113,13 +113,13 @@ bool psi_impl_def_val_validate(struct psi_data *data,
 				val->type = PSI_T_FLOAT;
 				type->type = PSI_T_FLOAT;
 				zend_string_release(type->name);
-				type->name = zend_string_init(ZEND_STRL("float"), 1);
+				type->name = zend_string_init_interned(ZEND_STRL("float"), 1);
 				break;
 			default:
 				val->type = PSI_T_INT;
 				type->type = PSI_T_INT;
 				zend_string_release(type->name);
-				type->name = zend_string_init(ZEND_STRL("int"), 1);
+				type->name = zend_string_init_interned(ZEND_STRL("int"), 1);
 				break;
 			}
 			psi_num_exp_free(&val->data.num);