Zend type tree

Zend/Optimizer/dfa_pass.c

@@ -282,7 +282,7 @@ static inline bool can_elide_list_type(
 			zend_string *lcname = zend_string_tolower(ZEND_TYPE_NAME(*single_type));
 			zend_class_entry *ce = zend_optimizer_get_class_entry(script, op_array, lcname);
 			zend_string_release(lcname);
-			bool result = ce && safe_instanceof(use_info->ce, ce);
+			bool result = ce && !ce->required_scope && safe_instanceof(use_info->ce, ce);
 			if (result == !is_intersection) {
 				return result;
 			}

Zend/tests/errmsg/errmsg_027.phpt

@@ -13,4 +13,4 @@ class test {
 echo "Done\n";
 ?>
 --EXPECTF--
-Fatal error: Class declarations may not be nested in %s on line %d
+Fatal error: Class declarations may not be declared inside functions in %s on line %d

Zend/zend.h

@@ -164,6 +164,10 @@ struct _zend_class_entry {
 	HashTable properties_info;
 	HashTable constants_table;
 
+	zend_class_entry *required_scope;
+	zend_class_entry *lexical_scope;
+	bool required_scope_absolute;
+
 	ZEND_MAP_PTR_DEF(zend_class_mutable_data*, mutable_data);
 	zend_inheritance_cache_entry *inheritance_cache;
 

Zend/zend_API.c

@@ -1816,6 +1816,27 @@ static zend_always_inline zend_result _object_and_properties_init(zval *arg, zen
 		return FAILURE;
 	}
 
+	if (class_type->required_scope) {
+		const zend_class_entry *scope = zend_get_executed_scope();
+		if (UNEXPECTED(scope == NULL)) {
+			zend_type_error("Cannot instantiate class %s from the global scope", ZSTR_VAL(class_type->name));
+			ZVAL_NULL(arg);
+			Z_OBJ_P(arg) = NULL;
+			return FAILURE;
+		}
+
+		if (class_type->required_scope_absolute) {
+			if (scope != class_type->required_scope && scope->lexical_scope != class_type->required_scope) {
+				zend_type_error("Cannot instantiate private class %s from scope %s", ZSTR_VAL(class_type->name), ZSTR_VAL(scope->name));
+				ZVAL_NULL(arg);
+				Z_OBJ_P(arg) = NULL;
+				return FAILURE;
+			}
+		} else if (!instanceof_function(scope, class_type->required_scope) && !instanceof_function(scope->lexical_scope, class_type->required_scope)) {
+			zend_type_error("Cannot instantiate protected class %s from scope %s", ZSTR_VAL(class_type->name), ZSTR_VAL(scope->name));
+		}
+	}
+
 	if (UNEXPECTED(!(class_type->ce_flags & ZEND_ACC_CONSTANTS_UPDATED))) {
 		if (UNEXPECTED(zend_update_class_constants(class_type) != SUCCESS)) {
 			ZVAL_NULL(arg);
@@ -2898,7 +2919,223 @@ 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 +3158,8 @@ 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 +3429,7 @@ 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 +4904,9 @@ 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);