initial type tree implementation

Zend/zend_API.c

@@ -2898,7 +2898,223 @@ ZEND_API void zend_add_magic_method(zend_class_entry *ce, zend_function *fptr, z
 ZEND_BEGIN_ARG_WITH_RETURN_TYPE_INFO_EX(arg_info_toString, 0, 0, IS_STRING, 0)
 ZEND_END_ARG_INFO()
 
-static zend_always_inline void zend_normalize_internal_type(zend_type *type) {
+static HashTable *interned_type_tree = NULL;
+
+// todo: move to zend_types.h
+#define ADD_TO_TREE(list, count, value) \
+	do { \
+		list = erealloc(list, sizeof(zend_type) * (count + 1)); \
+		list[count++] = value; \
+	} while (0)
+
+static int compare_simple_types(zend_type a, zend_type b) {
+	uint32_t a_mask = ZEND_TYPE_FULL_MASK(a);
+	uint32_t b_mask = ZEND_TYPE_FULL_MASK(b);
+
+	if (a_mask != b_mask) {
+		return a_mask < b_mask ? -1 : 1;
+	}
+
+	bool a_has_name = ZEND_TYPE_HAS_NAME(a);
+	bool b_has_name = ZEND_TYPE_HAS_NAME(b);
+
+	if (a_has_name && b_has_name) {
+		zend_string *a_name = ZEND_TYPE_NAME(a);
+		zend_string *b_name = ZEND_TYPE_NAME(b);
+		int cmp = ZSTR_VAL(a_name) == ZSTR_VAL(b_name);
+		if (cmp != 0) {
+			return cmp;
+		}
+	}
+
+	bool a_nullable = ZEND_TYPE_ALLOW_NULL(a);
+	bool b_nullable = ZEND_TYPE_ALLOW_NULL(b);
+
+	if (a_nullable != b_nullable) {
+		return a_nullable ? 1 : -1;
+	}
+
+	// Types are equal
+	return 0;
+}
+
+static int compare_type_nodes(const void *a_, const void *b_) {
+	zend_type_node *a = *(zend_type_node **)a_;
+	zend_type_node *b = *(zend_type_node **)b_;
+
+	if (a->kind != b->kind) {
+		return a->kind - b->kind;
+	}
+
+	if (a->kind == ZEND_TYPE_SIMPLE) {
+		return compare_simple_types(a->simple_type, b->simple_type);
+	}
+
+	if (a->compound.num_types != b->compound.num_types) {
+		return (int)a->compound.num_types - (int)b->compound.num_types;
+	}
+
+	for (uint32_t i = 0; i < a->compound.num_types; i++) {
+		const int cmp = compare_type_nodes(&a->compound.types[i], &b->compound.types[i]);
+		if (cmp != 0) {
+			return cmp;
+		}
+	}
+
+	return 0;
+}
+
+zend_ulong zend_type_node_hash(zend_type_node *node) {
+	zend_ulong hash = 2166136261u; // FNV-1a offset basis
+
+	hash ^= (zend_ulong)node->kind;
+	hash *= 16777619;
+
+	switch (node->kind) {
+		case ZEND_TYPE_SIMPLE: {
+			zend_type type = node->simple_type;
+			hash ^= (zend_ulong)ZEND_TYPE_FULL_MASK(type);
+			hash *= 16777619;
+
+			if (ZEND_TYPE_HAS_NAME(type)) {
+				zend_string *name = ZEND_TYPE_NAME(type);
+				hash ^= zend_string_hash_val(name);
+				hash *= 16777619;
+			}
+
+			break;
+		}
+
+		case ZEND_TYPE_UNION:
+		case ZEND_TYPE_INTERSECTION: {
+			for (uint32_t i = 0; i < node->compound.num_types; ++i) {
+				zend_ulong child_hash = zend_type_node_hash(node->compound.types[i]);
+				hash ^= child_hash;
+				hash *= 16777619;
+			}
+			break;
+		}
+	}
+
+	return hash;
+}
+
+bool zend_type_node_equals(zend_type_node *a, zend_type_node *b) {
+	if (a == b) return true;
+	if (a->kind != b->kind) return false;
+
+	if (a->kind == ZEND_TYPE_SIMPLE) {
+		zend_type at = a->simple_type;
+		zend_type bt = b->simple_type;
+
+		if (ZEND_TYPE_FULL_MASK(at) != ZEND_TYPE_FULL_MASK(bt)) {
+			return false;
+		}
+
+		bool a_has_name = ZEND_TYPE_HAS_NAME(at);
+		bool b_has_name = ZEND_TYPE_HAS_NAME(bt);
+		if (a_has_name != b_has_name) {
+			return false;
+		}
+
+		if (a_has_name) {
+			zend_string *a_name = ZEND_TYPE_NAME(at);
+			zend_string *b_name = ZEND_TYPE_NAME(bt);
+			if (!zend_string_equals(a_name, b_name)) {
+				return false;
+			}
+		}
+
+		return true;
+	}
+
+	// Compound type: union or intersection
+	if (a->compound.num_types != b->compound.num_types) {
+		return false;
+	}
+
+	for (uint32_t i = 0; i < a->compound.num_types; ++i) {
+		if (!zend_type_node_equals(a->compound.types[i], b->compound.types[i])) {
+			return false;
+		}
+	}
+
+	return true;
+}
+
+
+static zend_type_node *intern_type_node(zend_type_node *node) {
+	zend_ulong hash = zend_type_node_hash(node);
+	zend_type_node *existing;
+
+	if (interned_type_tree == NULL) {
+		interned_type_tree = pemalloc(sizeof(HashTable), 1);
+		zend_hash_init(interned_type_tree, 64, NULL, NULL, 1);
+	}
+
+	if ((existing = zend_hash_index_find_ptr(interned_type_tree, hash))) {
+		if (zend_type_node_equals(existing, node)) {
+			return existing; // reuse interned node
+		}
+	}
+
+	zend_hash_index_add_new_ptr(interned_type_tree, hash, node);
+	return node;
+}
+
+
+ZEND_API zend_type_node *zend_type_to_interned_tree(zend_type type) {
+	if (type.type_mask == 0) {
+		return NULL;
+	}
+
+	if (!ZEND_TYPE_HAS_LIST(type)) {
+		zend_type_node *node = pemalloc(sizeof(zend_type_node), 1);
+		node->kind = ZEND_TYPE_SIMPLE;
+		node->simple_type = type;
+		return intern_type_node(node);
+	}
+
+	zend_type_list *list = ZEND_TYPE_LIST(type);
+	zend_type_node_kind kind = ZEND_TYPE_IS_INTERSECTION(type) ?
+		ZEND_TYPE_INTERSECTION : ZEND_TYPE_UNION;
+
+	zend_type_node **children = NULL;
+	uint32_t num_children = 0;
+
+	zend_type *subtype;
+
+	ZEND_TYPE_LIST_FOREACH(list, subtype) {
+		zend_type_node *child = zend_type_to_interned_tree(*subtype);
+
+		if (child->kind == kind) {
+			for (uint32_t i = 0; child->compound.num_types; i++) {
+				ADD_TO_TREE(children, num_children, child->compound.types[i]);
+			}
+		} else {
+			ADD_TO_TREE(children, num_children, child);
+		}
+	} ZEND_TYPE_LIST_FOREACH_END();
+
+	qsort(children, num_children, sizeof(zend_type_node*), compare_type_nodes);
+
+	size_t deduped_count = 0;
+	for (size_t i = 0; i < num_children; i++) {
+		if (i == 0 || compare_type_nodes(&children[i], &children[i - 1]) != 0) {
+			children[deduped_count++] = children[i];
+		}
+	}
+
+	zend_type_node *node = pemalloc(sizeof(zend_type_node), 1);
+	node->kind = kind;
+	node->compound.num_types = deduped_count;
+	node->compound.types = pemalloc(sizeof(zend_type_node *) * deduped_count, 1);
+	memcpy(node->compound.types, children, sizeof(zend_type_node *) * deduped_count);
+
+	return intern_type_node(node);
+}
+
+static zend_always_inline zend_type_node *zend_normalize_internal_type(zend_type *type) {
 	ZEND_ASSERT(!ZEND_TYPE_HAS_LITERAL_NAME(*type));
 	if (ZEND_TYPE_PURE_MASK(*type) != MAY_BE_ANY) {
 		ZEND_ASSERT(!ZEND_TYPE_CONTAINS_CODE(*type, IS_RESOURCE) && "resource is not allowed in a zend_type");
@@ -2921,6 +3137,8 @@ static zend_always_inline void zend_normalize_internal_type(zend_type *type) {
 			} ZEND_TYPE_FOREACH_END();
 		}
 	} ZEND_TYPE_FOREACH_END();
+
+	return zend_type_to_interned_tree(*type);
 }
 
 /* registers all functions in *library_functions in the function hash */
@@ -3190,7 +3408,7 @@ ZEND_API zend_result zend_register_functions(zend_class_entry *scope, const zend
 					new_arg_info[i].type = legacy_iterable;
 				}
 
-				zend_normalize_internal_type(&new_arg_info[i].type);
+				new_arg_info[i].type_tree = zend_normalize_internal_type(&new_arg_info[i].type);
 			}
 		}
 
@@ -4665,7 +4883,9 @@ ZEND_API zend_property_info *zend_declare_typed_property(zend_class_entry *ce, z
 	property_info->type = type;
 
 	if (is_persistent_class(ce)) {
-		zend_normalize_internal_type(&property_info->type);
+		property_info->type_tree = zend_normalize_internal_type(&property_info->type);
+	} else {
+		property_info->type_tree = zend_type_to_interned_tree(property_info->type);
 	}
 
 	zend_hash_update_ptr(&ce->properties_info, name, property_info);

Zend/zend_API.h

@@ -128,46 +128,46 @@ typedef struct _zend_fcall_info_cache {
 
 /* Arginfo structures without type information */
 #define ZEND_ARG_INFO(pass_by_ref, name) \
-	{ #name, ZEND_TYPE_INIT_NONE(_ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), NULL },
+	{ #name, ZEND_TYPE_INIT_NONE(_ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), NULL, NULL },
 #define ZEND_ARG_INFO_WITH_DEFAULT_VALUE(pass_by_ref, name, default_value) \
-	{ #name, ZEND_TYPE_INIT_NONE(_ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), default_value },
+	{ #name, ZEND_TYPE_INIT_NONE(_ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), NULL, default_value },
 #define ZEND_ARG_VARIADIC_INFO(pass_by_ref, name) \
-	{ #name, ZEND_TYPE_INIT_NONE(_ZEND_ARG_INFO_FLAGS(pass_by_ref, 1, 0)), NULL },
+	{ #name, ZEND_TYPE_INIT_NONE(_ZEND_ARG_INFO_FLAGS(pass_by_ref, 1, 0)), NULL, NULL },
 
 /* Arginfo structures with simple type information */
 #define ZEND_ARG_TYPE_INFO(pass_by_ref, name, type_hint, allow_null) \
-	{ #name, ZEND_TYPE_INIT_CODE(type_hint, allow_null, _ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), NULL },
+	{ #name, ZEND_TYPE_INIT_CODE(type_hint, allow_null, _ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), NULL, NULL },
 #define ZEND_ARG_TYPE_INFO_WITH_DEFAULT_VALUE(pass_by_ref, name, type_hint, allow_null, default_value) \
-	{ #name, ZEND_TYPE_INIT_CODE(type_hint, allow_null, _ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), default_value },
+	{ #name, ZEND_TYPE_INIT_CODE(type_hint, allow_null, _ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), NULL, default_value },
 #define ZEND_ARG_VARIADIC_TYPE_INFO(pass_by_ref, name, type_hint, allow_null) \
-	{ #name, ZEND_TYPE_INIT_CODE(type_hint, allow_null, _ZEND_ARG_INFO_FLAGS(pass_by_ref, 1, 0)), NULL },
+	{ #name, ZEND_TYPE_INIT_CODE(type_hint, allow_null, _ZEND_ARG_INFO_FLAGS(pass_by_ref, 1, 0)), NULL, NULL },
 
 /* Arginfo structures with complex type information */
 #define ZEND_ARG_TYPE_MASK(pass_by_ref, name, type_mask, default_value) \
-	{ #name, ZEND_TYPE_INIT_MASK(type_mask | _ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), default_value },
+	{ #name, ZEND_TYPE_INIT_MASK(type_mask | _ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), NULL, default_value },
 #define ZEND_ARG_OBJ_TYPE_MASK(pass_by_ref, name, class_name, type_mask, default_value) \
-	{ #name, ZEND_TYPE_INIT_CLASS_CONST_MASK(#class_name, type_mask | _ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), default_value },
+	{ #name, ZEND_TYPE_INIT_CLASS_CONST_MASK(#class_name, type_mask | _ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), NULL, default_value },
 #define ZEND_ARG_VARIADIC_OBJ_TYPE_MASK(pass_by_ref, name, class_name, type_mask) \
-	{ #name, ZEND_TYPE_INIT_CLASS_CONST_MASK(#class_name, type_mask | _ZEND_ARG_INFO_FLAGS(pass_by_ref, 1, 0)), NULL },
+	{ #name, ZEND_TYPE_INIT_CLASS_CONST_MASK(#class_name, type_mask | _ZEND_ARG_INFO_FLAGS(pass_by_ref, 1, 0)), NULL, NULL },
 
 /* Arginfo structures with object type information */
 #define ZEND_ARG_OBJ_INFO(pass_by_ref, name, class_name, allow_null) \
-	{ #name, ZEND_TYPE_INIT_CLASS_CONST(#class_name, allow_null, _ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), NULL },
+	{ #name, ZEND_TYPE_INIT_CLASS_CONST(#class_name, allow_null, _ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), NULL, NULL },
 #define ZEND_ARG_OBJ_INFO_WITH_DEFAULT_VALUE(pass_by_ref, name, class_name, allow_null, default_value) \
-	{ #name, ZEND_TYPE_INIT_CLASS_CONST(#class_name, allow_null, _ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), default_value },
+	{ #name, ZEND_TYPE_INIT_CLASS_CONST(#class_name, allow_null, _ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), NULL, default_value },
 #define ZEND_ARG_VARIADIC_OBJ_INFO(pass_by_ref, name, class_name, allow_null) \
-	{ #name, ZEND_TYPE_INIT_CLASS_CONST(#class_name, allow_null, _ZEND_ARG_INFO_FLAGS(pass_by_ref, 1, 0)), NULL },
+	{ #name, ZEND_TYPE_INIT_CLASS_CONST(#class_name, allow_null, _ZEND_ARG_INFO_FLAGS(pass_by_ref, 1, 0)), NULL, NULL },
 
 /* Legacy arginfo structures */
 #define ZEND_ARG_ARRAY_INFO(pass_by_ref, name, allow_null) \
-	{ #name, ZEND_TYPE_INIT_CODE(IS_ARRAY, allow_null, _ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), NULL },
+	{ #name, ZEND_TYPE_INIT_CODE(IS_ARRAY, allow_null, _ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), NULL, NULL },
 #define ZEND_ARG_CALLABLE_INFO(pass_by_ref, name, allow_null) \
-	{ #name, ZEND_TYPE_INIT_CODE(IS_CALLABLE, allow_null, _ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), NULL },
+	{ #name, ZEND_TYPE_INIT_CODE(IS_CALLABLE, allow_null, _ZEND_ARG_INFO_FLAGS(pass_by_ref, 0, 0)), NULL, NULL },
 
 #define ZEND_BEGIN_ARG_WITH_RETURN_OBJ_INFO_EX2(name, return_reference, required_num_args, class_name, allow_null, is_tentative_return_type) \
 	static const zend_internal_arg_info name[] = { \
 		{ (const char*)(uintptr_t)(required_num_args), \
-			ZEND_TYPE_INIT_CLASS_CONST(#class_name, allow_null, _ZEND_ARG_INFO_FLAGS(return_reference, 0, is_tentative_return_type)), NULL },
+			ZEND_TYPE_INIT_CLASS_CONST(#class_name, allow_null, _ZEND_ARG_INFO_FLAGS(return_reference, 0, is_tentative_return_type)), NULL, NULL },
 
 #define ZEND_BEGIN_ARG_WITH_RETURN_OBJ_INFO_EX(name, return_reference, required_num_args, class_name, allow_null) \
 	ZEND_BEGIN_ARG_WITH_RETURN_OBJ_INFO_EX2(name, return_reference, required_num_args, class_name, allow_null, 0)
@@ -180,7 +180,7 @@ typedef struct _zend_fcall_info_cache {
 
 #define ZEND_BEGIN_ARG_WITH_RETURN_TYPE_MASK_EX2(name, return_reference, required_num_args, type, is_tentative_return_type) \
 	static const zend_internal_arg_info name[] = { \
-		{ (const char*)(uintptr_t)(required_num_args), ZEND_TYPE_INIT_MASK(type | _ZEND_ARG_INFO_FLAGS(return_reference, 0, is_tentative_return_type)), NULL },
+		{ (const char*)(uintptr_t)(required_num_args), ZEND_TYPE_INIT_MASK(type | _ZEND_ARG_INFO_FLAGS(return_reference, 0, is_tentative_return_type)), NULL, NULL },
 
 #define ZEND_BEGIN_ARG_WITH_RETURN_TYPE_MASK_EX(name, return_reference, required_num_args, type) \
 	ZEND_BEGIN_ARG_WITH_RETURN_TYPE_MASK_EX2(name, return_reference, required_num_args, type, 0)
@@ -190,7 +190,7 @@ typedef struct _zend_fcall_info_cache {
 
 #define ZEND_BEGIN_ARG_WITH_RETURN_OBJ_TYPE_MASK_EX2(name, return_reference, required_num_args, class_name, type, is_tentative_return_type) \
 	static const zend_internal_arg_info name[] = { \
-		{ (const char*)(uintptr_t)(required_num_args), ZEND_TYPE_INIT_CLASS_CONST_MASK(#class_name, type | _ZEND_ARG_INFO_FLAGS(return_reference, 0, is_tentative_return_type)), NULL },
+		{ (const char*)(uintptr_t)(required_num_args), ZEND_TYPE_INIT_CLASS_CONST_MASK(#class_name, type | _ZEND_ARG_INFO_FLAGS(return_reference, 0, is_tentative_return_type)), NULL, NULL },
 
 #define ZEND_BEGIN_ARG_WITH_RETURN_OBJ_TYPE_MASK_EX(name, return_reference, required_num_args, class_name, type) \
 	ZEND_BEGIN_ARG_WITH_RETURN_OBJ_TYPE_MASK_EX2(name, return_reference, required_num_args, class_name, type, 0)
@@ -200,7 +200,7 @@ typedef struct _zend_fcall_info_cache {
 
 #define ZEND_BEGIN_ARG_WITH_RETURN_TYPE_INFO_EX2(name, return_reference, required_num_args, type, allow_null, is_tentative_return_type) \
 	static const zend_internal_arg_info name[] = { \
-		{ (const char*)(uintptr_t)(required_num_args), ZEND_TYPE_INIT_CODE(type, allow_null, _ZEND_ARG_INFO_FLAGS(return_reference, 0, is_tentative_return_type)), NULL },
+		{ (const char*)(uintptr_t)(required_num_args), ZEND_TYPE_INIT_CODE(type, allow_null, _ZEND_ARG_INFO_FLAGS(return_reference, 0, is_tentative_return_type)), NULL, NULL },
 
 #define ZEND_BEGIN_ARG_WITH_RETURN_TYPE_INFO_EX(name, return_reference, required_num_args, type, allow_null) \
 	ZEND_BEGIN_ARG_WITH_RETURN_TYPE_INFO_EX2(name, return_reference, required_num_args, type, allow_null, 0)
@@ -213,7 +213,7 @@ typedef struct _zend_fcall_info_cache {
 
 #define ZEND_BEGIN_ARG_INFO_EX(name, _unused, return_reference, required_num_args)	\
 	static const zend_internal_arg_info name[] = { \
-		{ (const char*)(uintptr_t)(required_num_args), ZEND_TYPE_INIT_NONE(_ZEND_ARG_INFO_FLAGS(return_reference, 0, 0)), NULL },
+		{ (const char*)(uintptr_t)(required_num_args), ZEND_TYPE_INIT_NONE(_ZEND_ARG_INFO_FLAGS(return_reference, 0, 0)), NULL, NULL },
 #define ZEND_BEGIN_ARG_INFO(name, _unused)	\
 	ZEND_BEGIN_ARG_INFO_EX(name, {}, ZEND_RETURN_VALUE, -1)
 #define ZEND_END_ARG_INFO()		};
@@ -448,6 +448,8 @@ ZEND_API zend_result zend_update_class_constant(zend_class_constant *c, const ze
 ZEND_API zend_result zend_update_class_constants(zend_class_entry *class_type);
 ZEND_API HashTable *zend_separate_class_constants_table(zend_class_entry *class_type);
 
+ZEND_API zend_type_node *zend_type_to_interned_tree(zend_type type);
+
 static zend_always_inline HashTable *zend_class_constants_table(zend_class_entry *ce) {
 	if ((ce->ce_flags & ZEND_ACC_HAS_AST_CONSTANTS) && ZEND_MAP_PTR(ce->mutable_data)) {
 		zend_class_mutable_data *mutable_data =

Zend/zend_compile.c

@@ -7348,6 +7348,7 @@ static zend_type zend_compile_typename_ex(
 	}
 
 	ast->attr = orig_ast_attr;
+
 	return type;
 }
 /* }}} */
@@ -7584,6 +7585,7 @@ static void zend_compile_params(zend_ast *ast, zend_ast *return_type_ast, uint32
 		} else {
 			arg_infos->type = (zend_type) ZEND_TYPE_INIT_CODE(fallback_return_type, 0, 0);
 		}
+		arg_infos->type_tree = zend_type_to_interned_tree(arg_infos->type);
 		arg_infos++;
 		op_array->fn_flags |= ZEND_ACC_HAS_RETURN_TYPE;
 
@@ -7830,6 +7832,8 @@ static void zend_compile_params(zend_ast *ast, zend_ast *return_type_ast, uint32
 					&prop->attributes, attributes_ast, 0, ZEND_ATTRIBUTE_TARGET_PROPERTY, ZEND_ATTRIBUTE_TARGET_PARAMETER);
 			}
 		}
+
+		arg_info->type_tree = zend_type_to_interned_tree(arg_info->type);
 	}
 
 	/* These are assigned at the end to avoid uninitialized memory in case of an error */