v8/src/api.cc

3397 lines
105 KiB
C++
Raw Normal View History

// Copyright 2009 the V8 project authors. All rights reserved.
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
// * Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
// * Redistributions in binary form must reproduce the above
// copyright notice, this list of conditions and the following
// disclaimer in the documentation and/or other materials provided
// with the distribution.
// * Neither the name of Google Inc. nor the names of its
// contributors may be used to endorse or promote products derived
// from this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#include "v8.h"
#include "api.h"
#include "bootstrapper.h"
#include "compiler.h"
#include "debug.h"
#include "execution.h"
#include "global-handles.h"
#include "platform.h"
#include "serialize.h"
#include "snapshot.h"
#include "v8threads.h"
#define LOG_API(expr) LOG(ApiEntryCall(expr))
#ifdef ENABLE_HEAP_PROTECTION
#define ENTER_V8 i::VMState __state__(i::OTHER)
#define LEAVE_V8 i::VMState __state__(i::EXTERNAL)
#else
#define ENTER_V8 ((void) 0)
#define LEAVE_V8 ((void) 0)
#endif
namespace v8 {
#define ON_BAILOUT(location, code) \
if (IsDeadCheck(location)) { \
code; \
UNREACHABLE(); \
}
#define EXCEPTION_PREAMBLE() \
thread_local.IncrementCallDepth(); \
ASSERT(!i::Top::external_caught_exception()); \
bool has_pending_exception = false
#define EXCEPTION_BAILOUT_CHECK(value) \
do { \
thread_local.DecrementCallDepth(); \
if (has_pending_exception) { \
if (thread_local.CallDepthIsZero() && i::Top::is_out_of_memory()) { \
if (!thread_local.IgnoreOutOfMemory()) \
i::V8::FatalProcessOutOfMemory(NULL); \
} \
bool call_depth_is_zero = thread_local.CallDepthIsZero(); \
i::Top::optional_reschedule_exception(call_depth_is_zero); \
return value; \
} \
} while (false)
#define API_ENTRY_CHECK(msg) \
do { \
if (v8::Locker::IsActive()) { \
ApiCheck(i::ThreadManager::IsLockedByCurrentThread(), \
msg, \
"Entering the V8 API without proper locking in place"); \
} \
} while (false)
// --- D a t a t h a t i s s p e c i f i c t o a t h r e a d ---
static i::HandleScopeImplementer thread_local;
// --- E x c e p t i o n B e h a v i o r ---
static bool has_shut_down = false;
static FatalErrorCallback exception_behavior = NULL;
static void DefaultFatalErrorHandler(const char* location,
const char* message) {
ENTER_V8;
API_Fatal(location, message);
}
static FatalErrorCallback& GetFatalErrorHandler() {
if (exception_behavior == NULL) {
exception_behavior = DefaultFatalErrorHandler;
}
return exception_behavior;
}
// When V8 cannot allocated memory FatalProcessOutOfMemory is called.
// The default fatal error handler is called and execution is stopped.
void i::V8::FatalProcessOutOfMemory(const char* location) {
has_shut_down = true;
FatalErrorCallback callback = GetFatalErrorHandler();
{
LEAVE_V8;
callback(location, "Allocation failed - process out of memory");
}
// If the callback returns, we stop execution.
UNREACHABLE();
}
void V8::SetFatalErrorHandler(FatalErrorCallback that) {
exception_behavior = that;
}
bool Utils::ReportApiFailure(const char* location, const char* message) {
FatalErrorCallback callback = GetFatalErrorHandler();
callback(location, message);
has_shut_down = true;
return false;
}
bool V8::IsDead() {
return has_shut_down;
}
static inline bool ApiCheck(bool condition,
const char* location,
const char* message) {
return condition ? true : Utils::ReportApiFailure(location, message);
}
static bool ReportV8Dead(const char* location) {
FatalErrorCallback callback = GetFatalErrorHandler();
callback(location, "V8 is no longer usable");
return true;
}
static bool ReportEmptyHandle(const char* location) {
FatalErrorCallback callback = GetFatalErrorHandler();
callback(location, "Reading from empty handle");
return true;
}
/**
* IsDeadCheck checks that the vm is usable. If, for instance, the vm has been
* out of memory at some point this check will fail. It should be called on
* entry to all methods that touch anything in the heap, except destructors
* which you sometimes can't avoid calling after the vm has crashed. Functions
* that call EnsureInitialized or ON_BAILOUT don't have to also call
* IsDeadCheck. ON_BAILOUT has the advantage over EnsureInitialized that you
* can arrange to return if the VM is dead. This is needed to ensure that no VM
* heap allocations are attempted on a dead VM. EnsureInitialized has the
* advantage over ON_BAILOUT that it actually initializes the VM if this has not
* yet been done.
*/
static inline bool IsDeadCheck(const char* location) {
return has_shut_down ? ReportV8Dead(location) : false;
}
static inline bool EmptyCheck(const char* location, v8::Handle<v8::Data> obj) {
return obj.IsEmpty() ? ReportEmptyHandle(location) : false;
}
static inline bool EmptyCheck(const char* location, const v8::Data* obj) {
return (obj == 0) ? ReportEmptyHandle(location) : false;
}
// --- S t a t i c s ---
static i::StringInputBuffer write_input_buffer;
static void EnsureInitialized(const char* location) {
if (IsDeadCheck(location)) return;
ApiCheck(v8::V8::Initialize(), location, "Error initializing V8");
}
ImplementationUtilities::HandleScopeData*
ImplementationUtilities::CurrentHandleScope() {
return &i::HandleScope::current_;
}
#ifdef DEBUG
void ImplementationUtilities::ZapHandleRange(void** begin, void** end) {
i::HandleScope::ZapRange(begin, end);
}
#endif
v8::Handle<v8::Primitive> ImplementationUtilities::Undefined() {
if (IsDeadCheck("v8::Undefined()")) return v8::Handle<v8::Primitive>();
EnsureInitialized("v8::Undefined()");
return v8::Handle<Primitive>(ToApi<Primitive>(i::Factory::undefined_value()));
}
v8::Handle<v8::Primitive> ImplementationUtilities::Null() {
if (IsDeadCheck("v8::Null()")) return v8::Handle<v8::Primitive>();
EnsureInitialized("v8::Null()");
return v8::Handle<Primitive>(ToApi<Primitive>(i::Factory::null_value()));
}
v8::Handle<v8::Boolean> ImplementationUtilities::True() {
if (IsDeadCheck("v8::True()")) return v8::Handle<v8::Boolean>();
EnsureInitialized("v8::True()");
return v8::Handle<v8::Boolean>(ToApi<Boolean>(i::Factory::true_value()));
}
v8::Handle<v8::Boolean> ImplementationUtilities::False() {
if (IsDeadCheck("v8::False()")) return v8::Handle<v8::Boolean>();
EnsureInitialized("v8::False()");
return v8::Handle<v8::Boolean>(ToApi<Boolean>(i::Factory::false_value()));
}
void V8::SetFlagsFromString(const char* str, int length) {
i::FlagList::SetFlagsFromString(str, length);
}
void V8::SetFlagsFromCommandLine(int* argc, char** argv, bool remove_flags) {
i::FlagList::SetFlagsFromCommandLine(argc, argv, remove_flags);
}
v8::Handle<Value> ThrowException(v8::Handle<v8::Value> value) {
if (IsDeadCheck("v8::ThrowException()")) return v8::Handle<Value>();
ENTER_V8;
// If we're passed an empty handle, we throw an undefined exception
// to deal more gracefully with out of memory situations.
if (value.IsEmpty()) {
i::Top::ScheduleThrow(i::Heap::undefined_value());
} else {
i::Top::ScheduleThrow(*Utils::OpenHandle(*value));
}
return v8::Undefined();
}
RegisteredExtension* RegisteredExtension::first_extension_ = NULL;
RegisteredExtension::RegisteredExtension(Extension* extension)
: extension_(extension), state_(UNVISITED) { }
void RegisteredExtension::Register(RegisteredExtension* that) {
that->next_ = RegisteredExtension::first_extension_;
RegisteredExtension::first_extension_ = that;
}
void RegisterExtension(Extension* that) {
RegisteredExtension* extension = new RegisteredExtension(that);
RegisteredExtension::Register(extension);
}
Extension::Extension(const char* name,
const char* source,
int dep_count,
const char** deps)
: name_(name),
source_(source),
dep_count_(dep_count),
deps_(deps),
auto_enable_(false) { }
v8::Handle<Primitive> Undefined() {
LOG_API("Undefined");
return ImplementationUtilities::Undefined();
}
v8::Handle<Primitive> Null() {
LOG_API("Null");
return ImplementationUtilities::Null();
}
v8::Handle<Boolean> True() {
LOG_API("True");
return ImplementationUtilities::True();
}
v8::Handle<Boolean> False() {
LOG_API("False");
return ImplementationUtilities::False();
}
ResourceConstraints::ResourceConstraints()
: max_young_space_size_(0),
max_old_space_size_(0),
stack_limit_(NULL) { }
bool SetResourceConstraints(ResourceConstraints* constraints) {
bool result = i::Heap::ConfigureHeap(constraints->max_young_space_size(),
constraints->max_old_space_size());
if (!result) return false;
if (constraints->stack_limit() != NULL) {
uintptr_t limit = reinterpret_cast<uintptr_t>(constraints->stack_limit());
i::StackGuard::SetStackLimit(limit);
}
return true;
}
void** V8::GlobalizeReference(void** obj) {
if (IsDeadCheck("V8::Persistent::New")) return NULL;
LOG_API("Persistent::New");
i::Handle<i::Object> result =
i::GlobalHandles::Create(*reinterpret_cast<i::Object**>(obj));
return reinterpret_cast<void**>(result.location());
}
void V8::MakeWeak(void** object, void* parameters,
WeakReferenceCallback callback) {
LOG_API("MakeWeak");
i::GlobalHandles::MakeWeak(reinterpret_cast<i::Object**>(object), parameters,
callback);
}
void V8::ClearWeak(void** obj) {
LOG_API("ClearWeak");
i::GlobalHandles::ClearWeakness(reinterpret_cast<i::Object**>(obj));
}
bool V8::IsGlobalNearDeath(void** obj) {
LOG_API("IsGlobalNearDeath");
if (has_shut_down) return false;
return i::GlobalHandles::IsNearDeath(reinterpret_cast<i::Object**>(obj));
}
bool V8::IsGlobalWeak(void** obj) {
LOG_API("IsGlobalWeak");
if (has_shut_down) return false;
return i::GlobalHandles::IsWeak(reinterpret_cast<i::Object**>(obj));
}
void V8::DisposeGlobal(void** obj) {
LOG_API("DisposeGlobal");
if (has_shut_down) return;
i::Object** ptr = reinterpret_cast<i::Object**>(obj);
if ((*ptr)->IsGlobalContext()) i::Heap::NotifyContextDisposed();
i::GlobalHandles::Destroy(ptr);
}
// --- H a n d l e s ---
HandleScope::HandleScope() : is_closed_(false) {
API_ENTRY_CHECK("HandleScope::HandleScope");
i::HandleScope::Enter(&previous_);
}
HandleScope::~HandleScope() {
if (!is_closed_) {
i::HandleScope::Leave(&previous_);
}
}
int HandleScope::NumberOfHandles() {
return i::HandleScope::NumberOfHandles();
}
void** v8::HandleScope::CreateHandle(void* value) {
return i::HandleScope::CreateHandle(value);
}
void Context::Enter() {
if (IsDeadCheck("v8::Context::Enter()")) return;
ENTER_V8;
i::Handle<i::Context> env = Utils::OpenHandle(this);
thread_local.EnterContext(env);
thread_local.SaveContext(i::GlobalHandles::Create(i::Top::context()));
i::Top::set_context(*env);
}
void Context::Exit() {
if (has_shut_down) return;
if (!ApiCheck(thread_local.LeaveLastContext(),
"v8::Context::Exit()",
"Cannot exit non-entered context")) {
return;
}
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
// Content of 'last_context' could be NULL.
i::Handle<i::Object> last_context = thread_local.RestoreContext();
i::Top::set_context(static_cast<i::Context*>(*last_context));
i::GlobalHandles::Destroy(last_context.location());
}
void** v8::HandleScope::RawClose(void** value) {
if (!ApiCheck(!is_closed_,
"v8::HandleScope::Close()",
"Local scope has already been closed")) {
return 0;
}
LOG_API("CloseHandleScope");
// Read the result before popping the handle block.
i::Object* result = reinterpret_cast<i::Object*>(*value);
is_closed_ = true;
i::HandleScope::Leave(&previous_);
// Allocate a new handle on the previous handle block.
i::Handle<i::Object> handle(result);
return reinterpret_cast<void**>(handle.location());
}
// --- N e a n d e r ---
// A constructor cannot easily return an error value, therefore it is necessary
// to check for a dead VM with ON_BAILOUT before constructing any Neander
// objects. To remind you about this there is no HandleScope in the
// NeanderObject constructor. When you add one to the site calling the
// constructor you should check that you ensured the VM was not dead first.
NeanderObject::NeanderObject(int size) {
EnsureInitialized("v8::Nowhere");
ENTER_V8;
value_ = i::Factory::NewNeanderObject();
i::Handle<i::FixedArray> elements = i::Factory::NewFixedArray(size);
value_->set_elements(*elements);
}
int NeanderObject::size() {
return i::FixedArray::cast(value_->elements())->length();
}
NeanderArray::NeanderArray() : obj_(2) {
obj_.set(0, i::Smi::FromInt(0));
}
int NeanderArray::length() {
return i::Smi::cast(obj_.get(0))->value();
}
i::Object* NeanderArray::get(int offset) {
ASSERT(0 <= offset);
ASSERT(offset < length());
return obj_.get(offset + 1);
}
// This method cannot easily return an error value, therefore it is necessary
// to check for a dead VM with ON_BAILOUT before calling it. To remind you
// about this there is no HandleScope in this method. When you add one to the
// site calling this method you should check that you ensured the VM was not
// dead first.
void NeanderArray::add(i::Handle<i::Object> value) {
int length = this->length();
int size = obj_.size();
if (length == size - 1) {
i::Handle<i::FixedArray> new_elms = i::Factory::NewFixedArray(2 * size);
for (int i = 0; i < length; i++)
new_elms->set(i + 1, get(i));
obj_.value()->set_elements(*new_elms);
}
obj_.set(length + 1, *value);
obj_.set(0, i::Smi::FromInt(length + 1));
}
void NeanderArray::set(int index, i::Object* value) {
if (index < 0 || index >= this->length()) return;
obj_.set(index + 1, value);
}
// --- T e m p l a t e ---
static void InitializeTemplate(i::Handle<i::TemplateInfo> that, int type) {
that->set_tag(i::Smi::FromInt(type));
}
void Template::Set(v8::Handle<String> name, v8::Handle<Data> value,
v8::PropertyAttribute attribute) {
if (IsDeadCheck("v8::Template::SetProperty()")) return;
ENTER_V8;
HandleScope scope;
i::Handle<i::Object> list(Utils::OpenHandle(this)->property_list());
if (list->IsUndefined()) {
list = NeanderArray().value();
Utils::OpenHandle(this)->set_property_list(*list);
}
NeanderArray array(list);
array.add(Utils::OpenHandle(*name));
array.add(Utils::OpenHandle(*value));
array.add(Utils::OpenHandle(*v8::Integer::New(attribute)));
}
// --- F u n c t i o n T e m p l a t e ---
static void InitializeFunctionTemplate(
i::Handle<i::FunctionTemplateInfo> info) {
info->set_tag(i::Smi::FromInt(Consts::FUNCTION_TEMPLATE));
info->set_flag(0);
}
Local<ObjectTemplate> FunctionTemplate::PrototypeTemplate() {
if (IsDeadCheck("v8::FunctionTemplate::PrototypeTemplate()")) {
return Local<ObjectTemplate>();
}
ENTER_V8;
i::Handle<i::Object> result(Utils::OpenHandle(this)->prototype_template());
if (result->IsUndefined()) {
result = Utils::OpenHandle(*ObjectTemplate::New());
Utils::OpenHandle(this)->set_prototype_template(*result);
}
return Local<ObjectTemplate>(ToApi<ObjectTemplate>(result));
}
void FunctionTemplate::Inherit(v8::Handle<FunctionTemplate> value) {
if (IsDeadCheck("v8::FunctionTemplate::Inherit()")) return;
ENTER_V8;
Utils::OpenHandle(this)->set_parent_template(*Utils::OpenHandle(*value));
}
// To distinguish the function templates, so that we can find them in the
// function cache of the global context.
static int next_serial_number = 0;
Local<FunctionTemplate> FunctionTemplate::New(InvocationCallback callback,
v8::Handle<Value> data, v8::Handle<Signature> signature) {
EnsureInitialized("v8::FunctionTemplate::New()");
LOG_API("FunctionTemplate::New");
ENTER_V8;
i::Handle<i::Struct> struct_obj =
i::Factory::NewStruct(i::FUNCTION_TEMPLATE_INFO_TYPE);
i::Handle<i::FunctionTemplateInfo> obj =
i::Handle<i::FunctionTemplateInfo>::cast(struct_obj);
InitializeFunctionTemplate(obj);
obj->set_serial_number(i::Smi::FromInt(next_serial_number++));
if (callback != 0) {
if (data.IsEmpty()) data = v8::Undefined();
Utils::ToLocal(obj)->SetCallHandler(callback, data);
}
obj->set_undetectable(false);
obj->set_needs_access_check(false);
if (!signature.IsEmpty())
obj->set_signature(*Utils::OpenHandle(*signature));
return Utils::ToLocal(obj);
}
Local<Signature> Signature::New(Handle<FunctionTemplate> receiver,
int argc, Handle<FunctionTemplate> argv[]) {
EnsureInitialized("v8::Signature::New()");
LOG_API("Signature::New");
ENTER_V8;
i::Handle<i::Struct> struct_obj =
i::Factory::NewStruct(i::SIGNATURE_INFO_TYPE);
i::Handle<i::SignatureInfo> obj =
i::Handle<i::SignatureInfo>::cast(struct_obj);
if (!receiver.IsEmpty()) obj->set_receiver(*Utils::OpenHandle(*receiver));
if (argc > 0) {
i::Handle<i::FixedArray> args = i::Factory::NewFixedArray(argc);
for (int i = 0; i < argc; i++) {
if (!argv[i].IsEmpty())
args->set(i, *Utils::OpenHandle(*argv[i]));
}
obj->set_args(*args);
}
return Utils::ToLocal(obj);
}
Local<TypeSwitch> TypeSwitch::New(Handle<FunctionTemplate> type) {
Handle<FunctionTemplate> types[1] = { type };
return TypeSwitch::New(1, types);
}
Local<TypeSwitch> TypeSwitch::New(int argc, Handle<FunctionTemplate> types[]) {
EnsureInitialized("v8::TypeSwitch::New()");
LOG_API("TypeSwitch::New");
ENTER_V8;
i::Handle<i::FixedArray> vector = i::Factory::NewFixedArray(argc);
for (int i = 0; i < argc; i++)
vector->set(i, *Utils::OpenHandle(*types[i]));
i::Handle<i::Struct> struct_obj =
i::Factory::NewStruct(i::TYPE_SWITCH_INFO_TYPE);
i::Handle<i::TypeSwitchInfo> obj =
i::Handle<i::TypeSwitchInfo>::cast(struct_obj);
obj->set_types(*vector);
return Utils::ToLocal(obj);
}
int TypeSwitch::match(v8::Handle<Value> value) {
LOG_API("TypeSwitch::match");
i::Handle<i::Object> obj = Utils::OpenHandle(*value);
i::Handle<i::TypeSwitchInfo> info = Utils::OpenHandle(this);
i::FixedArray* types = i::FixedArray::cast(info->types());
for (int i = 0; i < types->length(); i++) {
if (obj->IsInstanceOf(i::FunctionTemplateInfo::cast(types->get(i))))
return i + 1;
}
return 0;
}
void FunctionTemplate::SetCallHandler(InvocationCallback callback,
v8::Handle<Value> data) {
if (IsDeadCheck("v8::FunctionTemplate::SetCallHandler()")) return;
ENTER_V8;
HandleScope scope;
i::Handle<i::Struct> struct_obj =
i::Factory::NewStruct(i::CALL_HANDLER_INFO_TYPE);
i::Handle<i::CallHandlerInfo> obj =
i::Handle<i::CallHandlerInfo>::cast(struct_obj);
obj->set_callback(*FromCData(callback));
if (data.IsEmpty()) data = v8::Undefined();
obj->set_data(*Utils::OpenHandle(*data));
Utils::OpenHandle(this)->set_call_code(*obj);
}
void FunctionTemplate::AddInstancePropertyAccessor(
v8::Handle<String> name,
AccessorGetter getter,
AccessorSetter setter,
v8::Handle<Value> data,
v8::AccessControl settings,
v8::PropertyAttribute attributes) {
if (IsDeadCheck("v8::FunctionTemplate::AddInstancePropertyAccessor()")) {
return;
}
ENTER_V8;
HandleScope scope;
i::Handle<i::AccessorInfo> obj = i::Factory::NewAccessorInfo();
ASSERT(getter != NULL);
obj->set_getter(*FromCData(getter));
obj->set_setter(*FromCData(setter));
if (data.IsEmpty()) data = v8::Undefined();
obj->set_data(*Utils::OpenHandle(*data));
obj->set_name(*Utils::OpenHandle(*name));
if (settings & ALL_CAN_READ) obj->set_all_can_read(true);
if (settings & ALL_CAN_WRITE) obj->set_all_can_write(true);
if (settings & PROHIBITS_OVERWRITING) obj->set_prohibits_overwriting(true);
obj->set_property_attributes(static_cast<PropertyAttributes>(attributes));
i::Handle<i::Object> list(Utils::OpenHandle(this)->property_accessors());
if (list->IsUndefined()) {
list = NeanderArray().value();
Utils::OpenHandle(this)->set_property_accessors(*list);
}
NeanderArray array(list);
array.add(obj);
}
Local<ObjectTemplate> FunctionTemplate::InstanceTemplate() {
if (IsDeadCheck("v8::FunctionTemplate::InstanceTemplate()")
|| EmptyCheck("v8::FunctionTemplate::InstanceTemplate()", this))
return Local<ObjectTemplate>();
ENTER_V8;
if (Utils::OpenHandle(this)->instance_template()->IsUndefined()) {
Local<ObjectTemplate> templ =
ObjectTemplate::New(v8::Handle<FunctionTemplate>(this));
Utils::OpenHandle(this)->set_instance_template(*Utils::OpenHandle(*templ));
}
i::Handle<i::ObjectTemplateInfo> result(i::ObjectTemplateInfo::cast(
Utils::OpenHandle(this)->instance_template()));
return Utils::ToLocal(result);
}
void FunctionTemplate::SetClassName(Handle<String> name) {
if (IsDeadCheck("v8::FunctionTemplate::SetClassName()")) return;
ENTER_V8;
Utils::OpenHandle(this)->set_class_name(*Utils::OpenHandle(*name));
}
void FunctionTemplate::SetHiddenPrototype(bool value) {
if (IsDeadCheck("v8::FunctionTemplate::SetHiddenPrototype()")) return;
ENTER_V8;
Utils::OpenHandle(this)->set_hidden_prototype(value);
}
void FunctionTemplate::SetNamedInstancePropertyHandler(
NamedPropertyGetter getter,
NamedPropertySetter setter,
NamedPropertyQuery query,
NamedPropertyDeleter remover,
NamedPropertyEnumerator enumerator,
Handle<Value> data) {
if (IsDeadCheck("v8::FunctionTemplate::SetNamedInstancePropertyHandler()")) {
return;
}
ENTER_V8;
HandleScope scope;
i::Handle<i::Struct> struct_obj =
i::Factory::NewStruct(i::INTERCEPTOR_INFO_TYPE);
i::Handle<i::InterceptorInfo> obj =
i::Handle<i::InterceptorInfo>::cast(struct_obj);
if (getter != 0) obj->set_getter(*FromCData(getter));
if (setter != 0) obj->set_setter(*FromCData(setter));
if (query != 0) obj->set_query(*FromCData(query));
if (remover != 0) obj->set_deleter(*FromCData(remover));
if (enumerator != 0) obj->set_enumerator(*FromCData(enumerator));
if (data.IsEmpty()) data = v8::Undefined();
obj->set_data(*Utils::OpenHandle(*data));
Utils::OpenHandle(this)->set_named_property_handler(*obj);
}
void FunctionTemplate::SetIndexedInstancePropertyHandler(
IndexedPropertyGetter getter,
IndexedPropertySetter setter,
IndexedPropertyQuery query,
IndexedPropertyDeleter remover,
IndexedPropertyEnumerator enumerator,
Handle<Value> data) {
if (IsDeadCheck(
"v8::FunctionTemplate::SetIndexedInstancePropertyHandler()")) {
return;
}
ENTER_V8;
HandleScope scope;
i::Handle<i::Struct> struct_obj =
i::Factory::NewStruct(i::INTERCEPTOR_INFO_TYPE);
i::Handle<i::InterceptorInfo> obj =
i::Handle<i::InterceptorInfo>::cast(struct_obj);
if (getter != 0) obj->set_getter(*FromCData(getter));
if (setter != 0) obj->set_setter(*FromCData(setter));
if (query != 0) obj->set_query(*FromCData(query));
if (remover != 0) obj->set_deleter(*FromCData(remover));
if (enumerator != 0) obj->set_enumerator(*FromCData(enumerator));
if (data.IsEmpty()) data = v8::Undefined();
obj->set_data(*Utils::OpenHandle(*data));
Utils::OpenHandle(this)->set_indexed_property_handler(*obj);
}
void FunctionTemplate::SetInstanceCallAsFunctionHandler(
InvocationCallback callback,
Handle<Value> data) {
if (IsDeadCheck("v8::FunctionTemplate::SetInstanceCallAsFunctionHandler()")) {
return;
}
ENTER_V8;
HandleScope scope;
i::Handle<i::Struct> struct_obj =
i::Factory::NewStruct(i::CALL_HANDLER_INFO_TYPE);
i::Handle<i::CallHandlerInfo> obj =
i::Handle<i::CallHandlerInfo>::cast(struct_obj);
obj->set_callback(*FromCData(callback));
if (data.IsEmpty()) data = v8::Undefined();
obj->set_data(*Utils::OpenHandle(*data));
Utils::OpenHandle(this)->set_instance_call_handler(*obj);
}
// --- O b j e c t T e m p l a t e ---
Local<ObjectTemplate> ObjectTemplate::New() {
return New(Local<FunctionTemplate>());
}
Local<ObjectTemplate> ObjectTemplate::New(
v8::Handle<FunctionTemplate> constructor) {
if (IsDeadCheck("v8::ObjectTemplate::New()")) return Local<ObjectTemplate>();
EnsureInitialized("v8::ObjectTemplate::New()");
LOG_API("ObjectTemplate::New");
ENTER_V8;
i::Handle<i::Struct> struct_obj =
i::Factory::NewStruct(i::OBJECT_TEMPLATE_INFO_TYPE);
i::Handle<i::ObjectTemplateInfo> obj =
i::Handle<i::ObjectTemplateInfo>::cast(struct_obj);
InitializeTemplate(obj, Consts::OBJECT_TEMPLATE);
if (!constructor.IsEmpty())
obj->set_constructor(*Utils::OpenHandle(*constructor));
obj->set_internal_field_count(i::Smi::FromInt(0));
return Utils::ToLocal(obj);
}
// Ensure that the object template has a constructor. If no
// constructor is available we create one.
static void EnsureConstructor(ObjectTemplate* object_template) {
if (Utils::OpenHandle(object_template)->constructor()->IsUndefined()) {
Local<FunctionTemplate> templ = FunctionTemplate::New();
i::Handle<i::FunctionTemplateInfo> constructor = Utils::OpenHandle(*templ);
constructor->set_instance_template(*Utils::OpenHandle(object_template));
Utils::OpenHandle(object_template)->set_constructor(*constructor);
}
}
void ObjectTemplate::SetAccessor(v8::Handle<String> name,
AccessorGetter getter,
AccessorSetter setter,
v8::Handle<Value> data,
AccessControl settings,
PropertyAttribute attribute) {
if (IsDeadCheck("v8::ObjectTemplate::SetAccessor()")) return;
ENTER_V8;
HandleScope scope;
EnsureConstructor(this);
i::FunctionTemplateInfo* constructor =
i::FunctionTemplateInfo::cast(Utils::OpenHandle(this)->constructor());
i::Handle<i::FunctionTemplateInfo> cons(constructor);
Utils::ToLocal(cons)->AddInstancePropertyAccessor(name,
getter,
setter,
data,
settings,
attribute);
}
void ObjectTemplate::SetNamedPropertyHandler(NamedPropertyGetter getter,
NamedPropertySetter setter,
NamedPropertyQuery query,
NamedPropertyDeleter remover,
NamedPropertyEnumerator enumerator,
Handle<Value> data) {
if (IsDeadCheck("v8::ObjectTemplate::SetNamedPropertyHandler()")) return;
ENTER_V8;
HandleScope scope;
EnsureConstructor(this);
i::FunctionTemplateInfo* constructor =
i::FunctionTemplateInfo::cast(Utils::OpenHandle(this)->constructor());
i::Handle<i::FunctionTemplateInfo> cons(constructor);
Utils::ToLocal(cons)->SetNamedInstancePropertyHandler(getter,
setter,
query,
remover,
enumerator,
data);
}
void ObjectTemplate::MarkAsUndetectable() {
if (IsDeadCheck("v8::ObjectTemplate::MarkAsUndetectable()")) return;
ENTER_V8;
HandleScope scope;
EnsureConstructor(this);
i::FunctionTemplateInfo* constructor =
i::FunctionTemplateInfo::cast(Utils::OpenHandle(this)->constructor());
i::Handle<i::FunctionTemplateInfo> cons(constructor);
cons->set_undetectable(true);
}
void ObjectTemplate::SetAccessCheckCallbacks(
NamedSecurityCallback named_callback,
IndexedSecurityCallback indexed_callback,
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
Handle<Value> data,
bool turned_on_by_default) {
if (IsDeadCheck("v8::ObjectTemplate::SetAccessCheckCallbacks()")) return;
ENTER_V8;
HandleScope scope;
EnsureConstructor(this);
i::Handle<i::Struct> struct_info =
i::Factory::NewStruct(i::ACCESS_CHECK_INFO_TYPE);
i::Handle<i::AccessCheckInfo> info =
i::Handle<i::AccessCheckInfo>::cast(struct_info);
info->set_named_callback(*FromCData(named_callback));
info->set_indexed_callback(*FromCData(indexed_callback));
if (data.IsEmpty()) data = v8::Undefined();
info->set_data(*Utils::OpenHandle(*data));
i::FunctionTemplateInfo* constructor =
i::FunctionTemplateInfo::cast(Utils::OpenHandle(this)->constructor());
i::Handle<i::FunctionTemplateInfo> cons(constructor);
cons->set_access_check_info(*info);
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
cons->set_needs_access_check(turned_on_by_default);
}
void ObjectTemplate::SetIndexedPropertyHandler(
IndexedPropertyGetter getter,
IndexedPropertySetter setter,
IndexedPropertyQuery query,
IndexedPropertyDeleter remover,
IndexedPropertyEnumerator enumerator,
Handle<Value> data) {
if (IsDeadCheck("v8::ObjectTemplate::SetIndexedPropertyHandler()")) return;
ENTER_V8;
HandleScope scope;
EnsureConstructor(this);
i::FunctionTemplateInfo* constructor =
i::FunctionTemplateInfo::cast(Utils::OpenHandle(this)->constructor());
i::Handle<i::FunctionTemplateInfo> cons(constructor);
Utils::ToLocal(cons)->SetIndexedInstancePropertyHandler(getter,
setter,
query,
remover,
enumerator,
data);
}
void ObjectTemplate::SetCallAsFunctionHandler(InvocationCallback callback,
Handle<Value> data) {
if (IsDeadCheck("v8::ObjectTemplate::SetCallAsFunctionHandler()")) return;
ENTER_V8;
HandleScope scope;
EnsureConstructor(this);
i::FunctionTemplateInfo* constructor =
i::FunctionTemplateInfo::cast(Utils::OpenHandle(this)->constructor());
i::Handle<i::FunctionTemplateInfo> cons(constructor);
Utils::ToLocal(cons)->SetInstanceCallAsFunctionHandler(callback, data);
}
int ObjectTemplate::InternalFieldCount() {
if (IsDeadCheck("v8::ObjectTemplate::InternalFieldCount()")) {
return 0;
}
return i::Smi::cast(Utils::OpenHandle(this)->internal_field_count())->value();
}
void ObjectTemplate::SetInternalFieldCount(int value) {
if (IsDeadCheck("v8::ObjectTemplate::SetInternalFieldCount()")) return;
if (!ApiCheck(i::Smi::IsValid(value),
"v8::ObjectTemplate::SetInternalFieldCount()",
"Invalid internal field count")) {
return;
}
ENTER_V8;
if (value > 0) {
// The internal field count is set by the constructor function's
// construct code, so we ensure that there is a constructor
// function to do the setting.
EnsureConstructor(this);
}
Utils::OpenHandle(this)->set_internal_field_count(i::Smi::FromInt(value));
}
// --- S c r i p t D a t a ---
ScriptData* ScriptData::PreCompile(const char* input, int length) {
unibrow::Utf8InputBuffer<> buf(input, length);
return i::PreParse(&buf, NULL);
}
ScriptData* ScriptData::New(unsigned* data, int length) {
return new i::ScriptDataImpl(i::Vector<unsigned>(data, length));
}
// --- S c r i p t ---
Local<Script> Script::Compile(v8::Handle<String> source,
v8::ScriptOrigin* origin,
v8::ScriptData* script_data) {
ON_BAILOUT("v8::Script::Compile()", return Local<Script>());
LOG_API("Script::Compile");
ENTER_V8;
i::Handle<i::String> str = Utils::OpenHandle(*source);
i::Handle<i::Object> name_obj;
int line_offset = 0;
int column_offset = 0;
if (origin != NULL) {
if (!origin->ResourceName().IsEmpty()) {
name_obj = Utils::OpenHandle(*origin->ResourceName());
}
if (!origin->ResourceLineOffset().IsEmpty()) {
line_offset = static_cast<int>(origin->ResourceLineOffset()->Value());
}
if (!origin->ResourceColumnOffset().IsEmpty()) {
column_offset = static_cast<int>(origin->ResourceColumnOffset()->Value());
}
}
EXCEPTION_PREAMBLE();
i::ScriptDataImpl* pre_data = static_cast<i::ScriptDataImpl*>(script_data);
// We assert that the pre-data is sane, even though we can actually
// handle it if it turns out not to be in release mode.
ASSERT(pre_data == NULL || pre_data->SanityCheck());
// If the pre-data isn't sane we simply ignore it
if (pre_data != NULL && !pre_data->SanityCheck())
pre_data = NULL;
i::Handle<i::JSFunction> boilerplate = i::Compiler::Compile(str,
name_obj,
line_offset,
column_offset,
NULL,
pre_data);
has_pending_exception = boilerplate.is_null();
EXCEPTION_BAILOUT_CHECK(Local<Script>());
i::Handle<i::JSFunction> result =
i::Factory::NewFunctionFromBoilerplate(boilerplate,
i::Top::global_context());
return Local<Script>(ToApi<Script>(result));
}
Local<Script> Script::Compile(v8::Handle<String> source,
v8::Handle<Value> file_name) {
ScriptOrigin origin(file_name);
return Compile(source, &origin);
}
Local<Value> Script::Run() {
ON_BAILOUT("v8::Script::Run()", return Local<Value>());
LOG_API("Script::Run");
ENTER_V8;
i::Object* raw_result = NULL;
{
HandleScope scope;
i::Handle<i::JSFunction> fun = Utils::OpenHandle(this);
EXCEPTION_PREAMBLE();
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
i::Handle<i::Object> receiver(i::Top::context()->global_proxy());
i::Handle<i::Object> result =
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
i::Execution::Call(fun, receiver, 0, NULL, &has_pending_exception);
EXCEPTION_BAILOUT_CHECK(Local<Value>());
raw_result = *result;
}
i::Handle<i::Object> result(raw_result);
return Utils::ToLocal(result);
}
Local<Value> Script::Id() {
ON_BAILOUT("v8::Script::Id()", return Local<Value>());
LOG_API("Script::Id");
i::Object* raw_id = NULL;
{
HandleScope scope;
i::Handle<i::JSFunction> fun = Utils::OpenHandle(this);
i::Handle<i::Script> script(i::Script::cast(fun->shared()->script()));
i::Handle<i::Object> id(script->id());
raw_id = *id;
}
i::Handle<i::Object> id(raw_id);
return Utils::ToLocal(id);
}
void Script::SetData(v8::Handle<Value> data) {
ON_BAILOUT("v8::Script::SetData()", return);
LOG_API("Script::SetData");
{
HandleScope scope;
i::Handle<i::JSFunction> fun = Utils::OpenHandle(this);
i::Handle<i::Object> raw_data = Utils::OpenHandle(*data);
i::Handle<i::Script> script(i::Script::cast(fun->shared()->script()));
script->set_data(*raw_data);
}
}
// --- E x c e p t i o n s ---
v8::TryCatch::TryCatch()
: next_(i::Top::try_catch_handler()),
exception_(i::Heap::the_hole_value()),
message_(i::Smi::FromInt(0)),
is_verbose_(false),
capture_message_(true),
js_handler_(NULL) {
i::Top::RegisterTryCatchHandler(this);
}
v8::TryCatch::~TryCatch() {
i::Top::UnregisterTryCatchHandler(this);
}
bool v8::TryCatch::HasCaught() const {
return !reinterpret_cast<i::Object*>(exception_)->IsTheHole();
}
v8::Local<Value> v8::TryCatch::Exception() const {
if (HasCaught()) {
// Check for out of memory exception.
i::Object* exception = reinterpret_cast<i::Object*>(exception_);
return v8::Utils::ToLocal(i::Handle<i::Object>(exception));
} else {
return v8::Local<Value>();
}
}
v8::Local<v8::Message> v8::TryCatch::Message() const {
if (HasCaught() && message_ != i::Smi::FromInt(0)) {
i::Object* message = reinterpret_cast<i::Object*>(message_);
return v8::Utils::MessageToLocal(i::Handle<i::Object>(message));
} else {
return v8::Local<v8::Message>();
}
}
void v8::TryCatch::Reset() {
exception_ = i::Heap::the_hole_value();
message_ = i::Smi::FromInt(0);
}
void v8::TryCatch::SetVerbose(bool value) {
is_verbose_ = value;
}
void v8::TryCatch::SetCaptureMessage(bool value) {
capture_message_ = value;
}
// --- M e s s a g e ---
Local<String> Message::Get() const {
ON_BAILOUT("v8::Message::Get()", return Local<String>());
ENTER_V8;
HandleScope scope;
i::Handle<i::Object> obj = Utils::OpenHandle(this);
i::Handle<i::String> raw_result = i::MessageHandler::GetMessage(obj);
Local<String> result = Utils::ToLocal(raw_result);
return scope.Close(result);
}
v8::Handle<Value> Message::GetScriptResourceName() const {
if (IsDeadCheck("v8::Message::GetScriptResourceName()")) {
return Local<String>();
}
ENTER_V8;
HandleScope scope;
i::Handle<i::JSObject> obj =
i::Handle<i::JSObject>::cast(Utils::OpenHandle(this));
// Return this.script.name.
i::Handle<i::JSValue> script =
i::Handle<i::JSValue>::cast(GetProperty(obj, "script"));
i::Handle<i::Object> resource_name(i::Script::cast(script->value())->name());
return scope.Close(Utils::ToLocal(resource_name));
}
v8::Handle<Value> Message::GetScriptData() const {
if (IsDeadCheck("v8::Message::GetScriptResourceData()")) {
return Local<Value>();
}
ENTER_V8;
HandleScope scope;
i::Handle<i::JSObject> obj =
i::Handle<i::JSObject>::cast(Utils::OpenHandle(this));
// Return this.script.data.
i::Handle<i::JSValue> script =
i::Handle<i::JSValue>::cast(GetProperty(obj, "script"));
i::Handle<i::Object> data(i::Script::cast(script->value())->data());
return scope.Close(Utils::ToLocal(data));
}
static i::Handle<i::Object> CallV8HeapFunction(const char* name,
i::Handle<i::Object> recv,
int argc,
i::Object** argv[],
bool* has_pending_exception) {
i::Handle<i::String> fmt_str = i::Factory::LookupAsciiSymbol(name);
i::Object* object_fun = i::Top::builtins()->GetProperty(*fmt_str);
i::Handle<i::JSFunction> fun =
i::Handle<i::JSFunction>(i::JSFunction::cast(object_fun));
i::Handle<i::Object> value =
i::Execution::Call(fun, recv, argc, argv, has_pending_exception);
return value;
}
static i::Handle<i::Object> CallV8HeapFunction(const char* name,
i::Handle<i::Object> data,
bool* has_pending_exception) {
i::Object** argv[1] = { data.location() };
return CallV8HeapFunction(name,
i::Top::builtins(),
1,
argv,
has_pending_exception);
}
int Message::GetLineNumber() const {
ON_BAILOUT("v8::Message::GetLineNumber()", return -1);
ENTER_V8;
HandleScope scope;
EXCEPTION_PREAMBLE();
i::Handle<i::Object> result = CallV8HeapFunction("GetLineNumber",
Utils::OpenHandle(this),
&has_pending_exception);
EXCEPTION_BAILOUT_CHECK(0);
return static_cast<int>(result->Number());
}
int Message::GetStartPosition() const {
if (IsDeadCheck("v8::Message::GetStartPosition()")) return 0;
ENTER_V8;
HandleScope scope;
i::Handle<i::JSObject> data_obj = Utils::OpenHandle(this);
return static_cast<int>(GetProperty(data_obj, "startPos")->Number());
}
int Message::GetEndPosition() const {
if (IsDeadCheck("v8::Message::GetEndPosition()")) return 0;
ENTER_V8;
HandleScope scope;
i::Handle<i::JSObject> data_obj = Utils::OpenHandle(this);
return static_cast<int>(GetProperty(data_obj, "endPos")->Number());
}
int Message::GetStartColumn() const {
if (IsDeadCheck("v8::Message::GetStartColumn()")) return 0;
ENTER_V8;
HandleScope scope;
i::Handle<i::JSObject> data_obj = Utils::OpenHandle(this);
EXCEPTION_PREAMBLE();
i::Handle<i::Object> start_col_obj = CallV8HeapFunction(
"GetPositionInLine",
data_obj,
&has_pending_exception);
EXCEPTION_BAILOUT_CHECK(0);
return static_cast<int>(start_col_obj->Number());
}
int Message::GetEndColumn() const {
if (IsDeadCheck("v8::Message::GetEndColumn()")) return 0;
ENTER_V8;
HandleScope scope;
i::Handle<i::JSObject> data_obj = Utils::OpenHandle(this);
EXCEPTION_PREAMBLE();
i::Handle<i::Object> start_col_obj = CallV8HeapFunction(
"GetPositionInLine",
data_obj,
&has_pending_exception);
EXCEPTION_BAILOUT_CHECK(0);
int start = static_cast<int>(GetProperty(data_obj, "startPos")->Number());
int end = static_cast<int>(GetProperty(data_obj, "endPos")->Number());
return static_cast<int>(start_col_obj->Number()) + (end - start);
}
Local<String> Message::GetSourceLine() const {
ON_BAILOUT("v8::Message::GetSourceLine()", return Local<String>());
ENTER_V8;
HandleScope scope;
EXCEPTION_PREAMBLE();
i::Handle<i::Object> result = CallV8HeapFunction("GetSourceLine",
Utils::OpenHandle(this),
&has_pending_exception);
EXCEPTION_BAILOUT_CHECK(Local<v8::String>());
if (result->IsString()) {
return scope.Close(Utils::ToLocal(i::Handle<i::String>::cast(result)));
} else {
return Local<String>();
}
}
void Message::PrintCurrentStackTrace(FILE* out) {
if (IsDeadCheck("v8::Message::PrintCurrentStackTrace()")) return;
ENTER_V8;
i::Top::PrintCurrentStackTrace(out);
}
// --- D a t a ---
bool Value::IsUndefined() const {
if (IsDeadCheck("v8::Value::IsUndefined()")) return false;
return Utils::OpenHandle(this)->IsUndefined();
}
bool Value::IsNull() const {
if (IsDeadCheck("v8::Value::IsNull()")) return false;
return Utils::OpenHandle(this)->IsNull();
}
bool Value::IsTrue() const {
if (IsDeadCheck("v8::Value::IsTrue()")) return false;
return Utils::OpenHandle(this)->IsTrue();
}
bool Value::IsFalse() const {
if (IsDeadCheck("v8::Value::IsFalse()")) return false;
return Utils::OpenHandle(this)->IsFalse();
}
bool Value::IsFunction() const {
if (IsDeadCheck("v8::Value::IsFunction()")) return false;
return Utils::OpenHandle(this)->IsJSFunction();
}
bool Value::IsString() const {
if (IsDeadCheck("v8::Value::IsString()")) return false;
return Utils::OpenHandle(this)->IsString();
}
bool Value::IsArray() const {
if (IsDeadCheck("v8::Value::IsArray()")) return false;
return Utils::OpenHandle(this)->IsJSArray();
}
bool Value::IsObject() const {
if (IsDeadCheck("v8::Value::IsObject()")) return false;
return Utils::OpenHandle(this)->IsJSObject();
}
bool Value::IsNumber() const {
if (IsDeadCheck("v8::Value::IsNumber()")) return false;
return Utils::OpenHandle(this)->IsNumber();
}
bool Value::IsBoolean() const {
if (IsDeadCheck("v8::Value::IsBoolean()")) return false;
return Utils::OpenHandle(this)->IsBoolean();
}
bool Value::IsExternal() const {
if (IsDeadCheck("v8::Value::IsExternal()")) return false;
return Utils::OpenHandle(this)->IsProxy();
}
bool Value::IsInt32() const {
if (IsDeadCheck("v8::Value::IsInt32()")) return false;
i::Handle<i::Object> obj = Utils::OpenHandle(this);
if (obj->IsSmi()) return true;
if (obj->IsNumber()) {
double value = obj->Number();
return i::FastI2D(i::FastD2I(value)) == value;
}
return false;
}
bool Value::IsDate() const {
if (IsDeadCheck("v8::Value::IsDate()")) return false;
i::Handle<i::Object> obj = Utils::OpenHandle(this);
return obj->HasSpecificClassOf(i::Heap::Date_symbol());
}
Local<String> Value::ToString() const {
if (IsDeadCheck("v8::Value::ToString()")) return Local<String>();
LOG_API("ToString");
i::Handle<i::Object> obj = Utils::OpenHandle(this);
i::Handle<i::Object> str;
if (obj->IsString()) {
str = obj;
} else {
ENTER_V8;
EXCEPTION_PREAMBLE();
str = i::Execution::ToString(obj, &has_pending_exception);
EXCEPTION_BAILOUT_CHECK(Local<String>());
}
return Local<String>(ToApi<String>(str));
}
Local<String> Value::ToDetailString() const {
if (IsDeadCheck("v8::Value::ToDetailString()")) return Local<String>();
LOG_API("ToDetailString");
i::Handle<i::Object> obj = Utils::OpenHandle(this);
i::Handle<i::Object> str;
if (obj->IsString()) {
str = obj;
} else {
ENTER_V8;
EXCEPTION_PREAMBLE();
str = i::Execution::ToDetailString(obj, &has_pending_exception);
EXCEPTION_BAILOUT_CHECK(Local<String>());
}
return Local<String>(ToApi<String>(str));
}
Local<v8::Object> Value::ToObject() const {
if (IsDeadCheck("v8::Value::ToObject()")) return Local<v8::Object>();
LOG_API("ToObject");
i::Handle<i::Object> obj = Utils::OpenHandle(this);
i::Handle<i::Object> val;
if (obj->IsJSObject()) {
val = obj;
} else {
ENTER_V8;
EXCEPTION_PREAMBLE();
val = i::Execution::ToObject(obj, &has_pending_exception);
EXCEPTION_BAILOUT_CHECK(Local<v8::Object>());
}
return Local<v8::Object>(ToApi<Object>(val));
}
Local<Boolean> Value::ToBoolean() const {
if (IsDeadCheck("v8::Value::ToBoolean()")) return Local<Boolean>();
LOG_API("ToBoolean");
i::Handle<i::Object> obj = Utils::OpenHandle(this);
if (obj->IsBoolean()) {
return Local<Boolean>(ToApi<Boolean>(obj));
} else {
ENTER_V8;
i::Handle<i::Object> val = i::Execution::ToBoolean(obj);
return Local<Boolean>(ToApi<Boolean>(val));
}
}
Local<Number> Value::ToNumber() const {
if (IsDeadCheck("v8::Value::ToNumber()")) return Local<Number>();
LOG_API("ToNumber");
i::Handle<i::Object> obj = Utils::OpenHandle(this);
i::Handle<i::Object> num;
if (obj->IsNumber()) {
num = obj;
} else {
ENTER_V8;
EXCEPTION_PREAMBLE();
num = i::Execution::ToNumber(obj, &has_pending_exception);
EXCEPTION_BAILOUT_CHECK(Local<Number>());
}
return Local<Number>(ToApi<Number>(num));
}
Local<Integer> Value::ToInteger() const {
if (IsDeadCheck("v8::Value::ToInteger()")) return Local<Integer>();
LOG_API("ToInteger");
i::Handle<i::Object> obj = Utils::OpenHandle(this);
i::Handle<i::Object> num;
if (obj->IsSmi()) {
num = obj;
} else {
ENTER_V8;
EXCEPTION_PREAMBLE();
num = i::Execution::ToInteger(obj, &has_pending_exception);
EXCEPTION_BAILOUT_CHECK(Local<Integer>());
}
return Local<Integer>(ToApi<Integer>(num));
}
External* External::Cast(v8::Value* that) {
if (IsDeadCheck("v8::External::Cast()")) return 0;
i::Handle<i::Object> obj = Utils::OpenHandle(that);
ApiCheck(obj->IsProxy(),
"v8::External::Cast()",
"Could not convert to external");
return static_cast<External*>(that);
}
v8::Object* v8::Object::Cast(Value* that) {
if (IsDeadCheck("v8::Object::Cast()")) return 0;
i::Handle<i::Object> obj = Utils::OpenHandle(that);
ApiCheck(obj->IsJSObject(),
"v8::Object::Cast()",
"Could not convert to object");
return static_cast<v8::Object*>(that);
}
v8::Function* v8::Function::Cast(Value* that) {
if (IsDeadCheck("v8::Function::Cast()")) return 0;
i::Handle<i::Object> obj = Utils::OpenHandle(that);
ApiCheck(obj->IsJSFunction(),
"v8::Function::Cast()",
"Could not convert to function");
return static_cast<v8::Function*>(that);
}
v8::String* v8::String::Cast(v8::Value* that) {
if (IsDeadCheck("v8::String::Cast()")) return 0;
i::Handle<i::Object> obj = Utils::OpenHandle(that);
ApiCheck(obj->IsString(),
"v8::String::Cast()",
"Could not convert to string");
return static_cast<v8::String*>(that);
}
v8::Number* v8::Number::Cast(v8::Value* that) {
if (IsDeadCheck("v8::Number::Cast()")) return 0;
i::Handle<i::Object> obj = Utils::OpenHandle(that);
ApiCheck(obj->IsNumber(),
"v8::Number::Cast()",
"Could not convert to number");
return static_cast<v8::Number*>(that);
}
v8::Integer* v8::Integer::Cast(v8::Value* that) {
if (IsDeadCheck("v8::Integer::Cast()")) return 0;
i::Handle<i::Object> obj = Utils::OpenHandle(that);
ApiCheck(obj->IsNumber(),
"v8::Integer::Cast()",
"Could not convert to number");
return static_cast<v8::Integer*>(that);
}
v8::Array* v8::Array::Cast(Value* that) {
if (IsDeadCheck("v8::Array::Cast()")) return 0;
i::Handle<i::Object> obj = Utils::OpenHandle(that);
ApiCheck(obj->IsJSArray(),
"v8::Array::Cast()",
"Could not convert to array");
return static_cast<v8::Array*>(that);
}
v8::Date* v8::Date::Cast(v8::Value* that) {
if (IsDeadCheck("v8::Date::Cast()")) return 0;
i::Handle<i::Object> obj = Utils::OpenHandle(that);
ApiCheck(obj->HasSpecificClassOf(i::Heap::Date_symbol()),
"v8::Date::Cast()",
"Could not convert to date");
return static_cast<v8::Date*>(that);
}
bool Value::BooleanValue() const {
if (IsDeadCheck("v8::Value::BooleanValue()")) return false;
LOG_API("BooleanValue");
i::Handle<i::Object> obj = Utils::OpenHandle(this);
if (obj->IsBoolean()) {
return obj->IsTrue();
} else {
ENTER_V8;
i::Handle<i::Object> value = i::Execution::ToBoolean(obj);
return value->IsTrue();
}
}
double Value::NumberValue() const {
if (IsDeadCheck("v8::Value::NumberValue()")) return i::OS::nan_value();
LOG_API("NumberValue");
i::Handle<i::Object> obj = Utils::OpenHandle(this);
i::Handle<i::Object> num;
if (obj->IsNumber()) {
num = obj;
} else {
ENTER_V8;
EXCEPTION_PREAMBLE();
num = i::Execution::ToNumber(obj, &has_pending_exception);
EXCEPTION_BAILOUT_CHECK(i::OS::nan_value());
}
return num->Number();
}
int64_t Value::IntegerValue() const {
if (IsDeadCheck("v8::Value::IntegerValue()")) return 0;
LOG_API("IntegerValue");
i::Handle<i::Object> obj = Utils::OpenHandle(this);
i::Handle<i::Object> num;
if (obj->IsNumber()) {
num = obj;
} else {
ENTER_V8;
EXCEPTION_PREAMBLE();
num = i::Execution::ToInteger(obj, &has_pending_exception);
EXCEPTION_BAILOUT_CHECK(0);
}
if (num->IsSmi()) {
return i::Smi::cast(*num)->value();
} else {
return static_cast<int64_t>(num->Number());
}
}
Local<Int32> Value::ToInt32() const {
if (IsDeadCheck("v8::Value::ToInt32()")) return Local<Int32>();
LOG_API("ToInt32");
i::Handle<i::Object> obj = Utils::OpenHandle(this);
i::Handle<i::Object> num;
if (obj->IsSmi()) {
num = obj;
} else {
ENTER_V8;
EXCEPTION_PREAMBLE();
num = i::Execution::ToInt32(obj, &has_pending_exception);
EXCEPTION_BAILOUT_CHECK(Local<Int32>());
}
return Local<Int32>(ToApi<Int32>(num));
}
Local<Uint32> Value::ToUint32() const {
if (IsDeadCheck("v8::Value::ToUint32()")) return Local<Uint32>();
LOG_API("ToUInt32");
i::Handle<i::Object> obj = Utils::OpenHandle(this);
i::Handle<i::Object> num;
if (obj->IsSmi()) {
num = obj;
} else {
ENTER_V8;
EXCEPTION_PREAMBLE();
num = i::Execution::ToUint32(obj, &has_pending_exception);
EXCEPTION_BAILOUT_CHECK(Local<Uint32>());
}
return Local<Uint32>(ToApi<Uint32>(num));
}
Local<Uint32> Value::ToArrayIndex() const {
if (IsDeadCheck("v8::Value::ToArrayIndex()")) return Local<Uint32>();
LOG_API("ToArrayIndex");
i::Handle<i::Object> obj = Utils::OpenHandle(this);
if (obj->IsSmi()) {
if (i::Smi::cast(*obj)->value() >= 0) return Utils::Uint32ToLocal(obj);
return Local<Uint32>();
}
ENTER_V8;
EXCEPTION_PREAMBLE();
i::Handle<i::Object> string_obj =
i::Execution::ToString(obj, &has_pending_exception);
EXCEPTION_BAILOUT_CHECK(Local<Uint32>());
i::Handle<i::String> str = i::Handle<i::String>::cast(string_obj);
uint32_t index;
if (str->AsArrayIndex(&index)) {
i::Handle<i::Object> value;
if (index <= static_cast<uint32_t>(i::Smi::kMaxValue)) {
value = i::Handle<i::Object>(i::Smi::FromInt(index));
} else {
value = i::Factory::NewNumber(index);
}
return Utils::Uint32ToLocal(value);
}
return Local<Uint32>();
}
int32_t Value::Int32Value() const {
if (IsDeadCheck("v8::Value::Int32Value()")) return 0;
LOG_API("Int32Value");
i::Handle<i::Object> obj = Utils::OpenHandle(this);
if (obj->IsSmi()) {
return i::Smi::cast(*obj)->value();
} else {
LOG_API("Int32Value (slow)");
ENTER_V8;
EXCEPTION_PREAMBLE();
i::Handle<i::Object> num =
i::Execution::ToInt32(obj, &has_pending_exception);
EXCEPTION_BAILOUT_CHECK(0);
if (num->IsSmi()) {
return i::Smi::cast(*num)->value();
} else {
return static_cast<int32_t>(num->Number());
}
}
}
bool Value::Equals(Handle<Value> that) const {
if (IsDeadCheck("v8::Value::Equals()")
|| EmptyCheck("v8::Value::Equals()", this)
|| EmptyCheck("v8::Value::Equals()", that)) {
return false;
}
LOG_API("Equals");
ENTER_V8;
i::Handle<i::Object> obj = Utils::OpenHandle(this);
i::Handle<i::Object> other = Utils::OpenHandle(*that);
i::Object** args[1] = { other.location() };
EXCEPTION_PREAMBLE();
i::Handle<i::Object> result =
CallV8HeapFunction("EQUALS", obj, 1, args, &has_pending_exception);
EXCEPTION_BAILOUT_CHECK(false);
return *result == i::Smi::FromInt(i::EQUAL);
}
bool Value::StrictEquals(Handle<Value> that) const {
if (IsDeadCheck("v8::Value::StrictEquals()")
|| EmptyCheck("v8::Value::StrictEquals()", this)
|| EmptyCheck("v8::Value::StrictEquals()", that)) {
return false;
}
LOG_API("StrictEquals");
i::Handle<i::Object> obj = Utils::OpenHandle(this);
i::Handle<i::Object> other = Utils::OpenHandle(*that);
// Must check HeapNumber first, since NaN !== NaN.
if (obj->IsHeapNumber()) {
if (!other->IsNumber()) return false;
double x = obj->Number();
double y = other->Number();
// Must check explicitly for NaN:s on Windows, but -0 works fine.
return x == y && !isnan(x) && !isnan(y);
} else if (*obj == *other) { // Also covers Booleans.
return true;
} else if (obj->IsSmi()) {
return other->IsNumber() && obj->Number() == other->Number();
} else if (obj->IsString()) {
return other->IsString() &&
i::String::cast(*obj)->Equals(i::String::cast(*other));
} else if (obj->IsUndefined() || obj->IsUndetectableObject()) {
return other->IsUndefined() || other->IsUndetectableObject();
} else {
return false;
}
}
uint32_t Value::Uint32Value() const {
if (IsDeadCheck("v8::Value::Uint32Value()")) return 0;
LOG_API("Uint32Value");
i::Handle<i::Object> obj = Utils::OpenHandle(this);
if (obj->IsSmi()) {
return i::Smi::cast(*obj)->value();
} else {
ENTER_V8;
EXCEPTION_PREAMBLE();
i::Handle<i::Object> num =
i::Execution::ToUint32(obj, &has_pending_exception);
EXCEPTION_BAILOUT_CHECK(0);
if (num->IsSmi()) {
return i::Smi::cast(*num)->value();
} else {
return static_cast<uint32_t>(num->Number());
}
}
}
bool v8::Object::Set(v8::Handle<Value> key, v8::Handle<Value> value,
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
v8::PropertyAttribute attribs) {
ON_BAILOUT("v8::Object::Set()", return false);
ENTER_V8;
i::Handle<i::Object> self = Utils::OpenHandle(this);
i::Handle<i::Object> key_obj = Utils::OpenHandle(*key);
i::Handle<i::Object> value_obj = Utils::OpenHandle(*value);
EXCEPTION_PREAMBLE();
i::Handle<i::Object> obj = i::SetProperty(
self,
key_obj,
value_obj,
static_cast<PropertyAttributes>(attribs));
has_pending_exception = obj.is_null();
EXCEPTION_BAILOUT_CHECK(false);
return true;
}
bool v8::Object::ForceSet(v8::Handle<Value> key,
v8::Handle<Value> value,
v8::PropertyAttribute attribs) {
ON_BAILOUT("v8::Object::ForceSet()", return false);
ENTER_V8;
i::Handle<i::JSObject> self = Utils::OpenHandle(this);
i::Handle<i::Object> key_obj = Utils::OpenHandle(*key);
i::Handle<i::Object> value_obj = Utils::OpenHandle(*value);
EXCEPTION_PREAMBLE();
i::Handle<i::Object> obj = i::ForceSetProperty(
self,
key_obj,
value_obj,
static_cast<PropertyAttributes>(attribs));
has_pending_exception = obj.is_null();
EXCEPTION_BAILOUT_CHECK(false);
return true;
}
Local<Value> v8::Object::Get(v8::Handle<Value> key) {
ON_BAILOUT("v8::Object::Get()", return Local<v8::Value>());
ENTER_V8;
i::Handle<i::Object> self = Utils::OpenHandle(this);
i::Handle<i::Object> key_obj = Utils::OpenHandle(*key);
EXCEPTION_PREAMBLE();
i::Handle<i::Object> result = i::GetProperty(self, key_obj);
has_pending_exception = result.is_null();
EXCEPTION_BAILOUT_CHECK(Local<Value>());
return Utils::ToLocal(result);
}
Local<Value> v8::Object::GetPrototype() {
ON_BAILOUT("v8::Object::GetPrototype()", return Local<v8::Value>());
ENTER_V8;
i::Handle<i::Object> self = Utils::OpenHandle(this);
i::Handle<i::Object> result = i::GetPrototype(self);
return Utils::ToLocal(result);
}
Local<Array> v8::Object::GetPropertyNames() {
ON_BAILOUT("v8::Object::GetPropertyNames()", return Local<v8::Array>());
ENTER_V8;
v8::HandleScope scope;
i::Handle<i::JSObject> self = Utils::OpenHandle(this);
i::Handle<i::FixedArray> value = i::GetKeysInFixedArrayFor(self);
// Because we use caching to speed up enumeration it is important
// to never change the result of the basic enumeration function so
// we clone the result.
i::Handle<i::FixedArray> elms = i::Factory::CopyFixedArray(value);
i::Handle<i::JSArray> result = i::Factory::NewJSArrayWithElements(elms);
return scope.Close(Utils::ToLocal(result));
}
Local<String> v8::Object::ObjectProtoToString() {
ON_BAILOUT("v8::Object::ObjectProtoToString()", return Local<v8::String>());
ENTER_V8;
i::Handle<i::JSObject> self = Utils::OpenHandle(this);
i::Handle<i::Object> name(self->class_name());
// Native implementation of Object.prototype.toString (v8natives.js):
// var c = %ClassOf(this);
// if (c === 'Arguments') c = 'Object';
// return "[object " + c + "]";
if (!name->IsString()) {
return v8::String::New("[object ]");
} else {
i::Handle<i::String> class_name = i::Handle<i::String>::cast(name);
if (class_name->IsEqualTo(i::CStrVector("Arguments"))) {
return v8::String::New("[object Object]");
} else {
const char* prefix = "[object ";
Local<String> str = Utils::ToLocal(class_name);
const char* postfix = "]";
size_t prefix_len = strlen(prefix);
size_t str_len = str->Length();
size_t postfix_len = strlen(postfix);
size_t buf_len = prefix_len + str_len + postfix_len;
char* buf = i::NewArray<char>(buf_len);
// Write prefix.
char* ptr = buf;
memcpy(ptr, prefix, prefix_len * v8::internal::kCharSize);
ptr += prefix_len;
// Write real content.
str->WriteAscii(ptr, 0, str_len);
ptr += str_len;
// Write postfix.
memcpy(ptr, postfix, postfix_len * v8::internal::kCharSize);
// Copy the buffer into a heap-allocated string and return it.
Local<String> result = v8::String::New(buf, buf_len);
i::DeleteArray(buf);
return result;
}
}
}
bool v8::Object::Delete(v8::Handle<String> key) {
ON_BAILOUT("v8::Object::Delete()", return false);
ENTER_V8;
HandleScope scope;
i::Handle<i::JSObject> self = Utils::OpenHandle(this);
i::Handle<i::String> key_obj = Utils::OpenHandle(*key);
return i::DeleteProperty(self, key_obj)->IsTrue();
}
bool v8::Object::Has(v8::Handle<String> key) {
ON_BAILOUT("v8::Object::Has()", return false);
ENTER_V8;
i::Handle<i::JSObject> self = Utils::OpenHandle(this);
i::Handle<i::String> key_obj = Utils::OpenHandle(*key);
return self->HasProperty(*key_obj);
}
bool v8::Object::Delete(uint32_t index) {
ON_BAILOUT("v8::Object::DeleteProperty()", return false);
ENTER_V8;
HandleScope scope;
i::Handle<i::JSObject> self = Utils::OpenHandle(this);
return i::DeleteElement(self, index)->IsTrue();
}
bool v8::Object::Has(uint32_t index) {
ON_BAILOUT("v8::Object::HasProperty()", return false);
i::Handle<i::JSObject> self = Utils::OpenHandle(this);
return self->HasElement(index);
}
bool v8::Object::HasRealNamedProperty(Handle<String> key) {
ON_BAILOUT("v8::Object::HasRealNamedProperty()", return false);
return Utils::OpenHandle(this)->HasRealNamedProperty(
*Utils::OpenHandle(*key));
}
bool v8::Object::HasRealIndexedProperty(uint32_t index) {
ON_BAILOUT("v8::Object::HasRealIndexedProperty()", return false);
return Utils::OpenHandle(this)->HasRealElementProperty(index);
}
bool v8::Object::HasRealNamedCallbackProperty(Handle<String> key) {
ON_BAILOUT("v8::Object::HasRealNamedCallbackProperty()", return false);
ENTER_V8;
return Utils::OpenHandle(this)->HasRealNamedCallbackProperty(
*Utils::OpenHandle(*key));
}
bool v8::Object::HasNamedLookupInterceptor() {
ON_BAILOUT("v8::Object::HasNamedLookupInterceptor()", return false);
return Utils::OpenHandle(this)->HasNamedInterceptor();
}
bool v8::Object::HasIndexedLookupInterceptor() {
ON_BAILOUT("v8::Object::HasIndexedLookupInterceptor()", return false);
return Utils::OpenHandle(this)->HasIndexedInterceptor();
}
Handle<Value> v8::Object::GetRealNamedPropertyInPrototypeChain(
Handle<String> key) {
ON_BAILOUT("v8::Object::GetRealNamedPropertyInPrototypeChain()",
return Local<Value>());
ENTER_V8;
i::Handle<i::JSObject> self_obj = Utils::OpenHandle(this);
i::Handle<i::String> key_obj = Utils::OpenHandle(*key);
i::LookupResult lookup;
self_obj->LookupRealNamedPropertyInPrototypes(*key_obj, &lookup);
if (lookup.IsValid()) {
PropertyAttributes attributes;
i::Handle<i::Object> result(self_obj->GetProperty(*self_obj,
&lookup,
*key_obj,
&attributes));
return Utils::ToLocal(result);
}
return Local<Value>(); // No real property was found in prototype chain.
}
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
// Turns on access checks by copying the map and setting the check flag.
// Because the object gets a new map, existing inline cache caching
// the old map of this object will fail.
void v8::Object::TurnOnAccessCheck() {
ON_BAILOUT("v8::Object::TurnOnAccessCheck()", return);
ENTER_V8;
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
i::Handle<i::JSObject> obj = Utils::OpenHandle(this);
i::Handle<i::Map> new_map =
i::Factory::CopyMapDropTransitions(i::Handle<i::Map>(obj->map()));
new_map->set_is_access_check_needed(true);
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
obj->set_map(*new_map);
}
Local<v8::Object> v8::Object::Clone() {
ON_BAILOUT("v8::Object::Clone()", return Local<Object>());
ENTER_V8;
i::Handle<i::JSObject> self = Utils::OpenHandle(this);
EXCEPTION_PREAMBLE();
i::Handle<i::JSObject> result = i::Copy(self);
has_pending_exception = result.is_null();
EXCEPTION_BAILOUT_CHECK(Local<Object>());
return Utils::ToLocal(result);
}
int v8::Object::GetIdentityHash() {
ON_BAILOUT("v8::Object::GetIdentityHash()", return 0);
ENTER_V8;
i::Handle<i::JSObject> self = Utils::OpenHandle(this);
i::Handle<i::Object> hidden_props(i::GetHiddenProperties(self, true));
i::Handle<i::Object> hash_symbol = i::Factory::identity_hash_symbol();
i::Handle<i::Object> hash = i::GetProperty(hidden_props, hash_symbol);
int hash_value;
if (hash->IsSmi()) {
hash_value = i::Smi::cast(*hash)->value();
} else {
hash_value = random() & i::Smi::kMaxValue; // Limit range to fit a smi.
i::SetProperty(hidden_props,
hash_symbol,
i::Handle<i::Object>(i::Smi::FromInt(hash_value)),
static_cast<PropertyAttributes>(None));
}
return hash_value;
}
bool v8::Object::SetHiddenValue(v8::Handle<v8::String> key,
v8::Handle<v8::Value> value) {
ON_BAILOUT("v8::Object::SetHiddenValue()", return false);
ENTER_V8;
i::Handle<i::JSObject> self = Utils::OpenHandle(this);
i::Handle<i::Object> hidden_props(i::GetHiddenProperties(self, true));
i::Handle<i::Object> key_obj = Utils::OpenHandle(*key);
i::Handle<i::Object> value_obj = Utils::OpenHandle(*value);
EXCEPTION_PREAMBLE();
i::Handle<i::Object> obj = i::SetProperty(
hidden_props,
key_obj,
value_obj,
static_cast<PropertyAttributes>(None));
has_pending_exception = obj.is_null();
EXCEPTION_BAILOUT_CHECK(false);
return true;
}
v8::Local<v8::Value> v8::Object::GetHiddenValue(v8::Handle<v8::String> key) {
ON_BAILOUT("v8::Object::GetHiddenValue()", return Local<v8::Value>());
ENTER_V8;
i::Handle<i::JSObject> self = Utils::OpenHandle(this);
i::Handle<i::Object> hidden_props(i::GetHiddenProperties(self, false));
if (hidden_props->IsUndefined()) {
return v8::Local<v8::Value>();
}
i::Handle<i::String> key_obj = Utils::OpenHandle(*key);
EXCEPTION_PREAMBLE();
i::Handle<i::Object> result = i::GetProperty(hidden_props, key_obj);
has_pending_exception = result.is_null();
EXCEPTION_BAILOUT_CHECK(v8::Local<v8::Value>());
if (result->IsUndefined()) {
return v8::Local<v8::Value>();
}
return Utils::ToLocal(result);
}
bool v8::Object::DeleteHiddenValue(v8::Handle<v8::String> key) {
ON_BAILOUT("v8::DeleteHiddenValue()", return false);
ENTER_V8;
i::Handle<i::JSObject> self = Utils::OpenHandle(this);
i::Handle<i::Object> hidden_props(i::GetHiddenProperties(self, false));
if (hidden_props->IsUndefined()) {
return true;
}
i::Handle<i::JSObject> js_obj(i::JSObject::cast(*hidden_props));
i::Handle<i::String> key_obj = Utils::OpenHandle(*key);
return i::DeleteProperty(js_obj, key_obj)->IsTrue();
}
Local<v8::Object> Function::NewInstance() const {
return NewInstance(0, NULL);
}
Local<v8::Object> Function::NewInstance(int argc,
v8::Handle<v8::Value> argv[]) const {
ON_BAILOUT("v8::Function::NewInstance()", return Local<v8::Object>());
LOG_API("Function::NewInstance");
ENTER_V8;
HandleScope scope;
i::Handle<i::JSFunction> function = Utils::OpenHandle(this);
STATIC_ASSERT(sizeof(v8::Handle<v8::Value>) == sizeof(i::Object**));
i::Object*** args = reinterpret_cast<i::Object***>(argv);
EXCEPTION_PREAMBLE();
i::Handle<i::Object> returned =
i::Execution::New(function, argc, args, &has_pending_exception);
EXCEPTION_BAILOUT_CHECK(Local<v8::Object>());
return scope.Close(Utils::ToLocal(i::Handle<i::JSObject>::cast(returned)));
}
Local<v8::Value> Function::Call(v8::Handle<v8::Object> recv, int argc,
v8::Handle<v8::Value> argv[]) {
ON_BAILOUT("v8::Function::Call()", return Local<v8::Value>());
LOG_API("Function::Call");
ENTER_V8;
i::Object* raw_result = NULL;
{
HandleScope scope;
i::Handle<i::JSFunction> fun = Utils::OpenHandle(this);
i::Handle<i::Object> recv_obj = Utils::OpenHandle(*recv);
STATIC_ASSERT(sizeof(v8::Handle<v8::Value>) == sizeof(i::Object**));
i::Object*** args = reinterpret_cast<i::Object***>(argv);
EXCEPTION_PREAMBLE();
i::Handle<i::Object> returned =
i::Execution::Call(fun, recv_obj, argc, args, &has_pending_exception);
EXCEPTION_BAILOUT_CHECK(Local<Object>());
raw_result = *returned;
}
i::Handle<i::Object> result(raw_result);
return Utils::ToLocal(result);
}
void Function::SetName(v8::Handle<v8::String> name) {
ENTER_V8;
i::Handle<i::JSFunction> func = Utils::OpenHandle(this);
func->shared()->set_name(*Utils::OpenHandle(*name));
}
Handle<Value> Function::GetName() const {
i::Handle<i::JSFunction> func = Utils::OpenHandle(this);
return Utils::ToLocal(i::Handle<i::Object>(func->shared()->name()));
}
int String::Length() const {
if (IsDeadCheck("v8::String::Length()")) return 0;
return Utils::OpenHandle(this)->length();
}
int String::Utf8Length() const {
if (IsDeadCheck("v8::String::Utf8Length()")) return 0;
return Utils::OpenHandle(this)->Utf8Length();
}
int String::WriteUtf8(char* buffer, int capacity) const {
if (IsDeadCheck("v8::String::WriteUtf8()")) return 0;
LOG_API("String::WriteUtf8");
ENTER_V8;
i::Handle<i::String> str = Utils::OpenHandle(this);
write_input_buffer.Reset(0, *str);
int len = str->length();
// Encode the first K - 3 bytes directly into the buffer since we
// know there's room for them. If no capacity is given we copy all
// of them here.
int fast_end = capacity - (unibrow::Utf8::kMaxEncodedSize - 1);
int i;
int pos = 0;
for (i = 0; i < len && (capacity == -1 || pos < fast_end); i++) {
i::uc32 c = write_input_buffer.GetNext();
int written = unibrow::Utf8::Encode(buffer + pos, c);
pos += written;
}
if (i < len) {
// For the last characters we need to check the length for each one
// because they may be longer than the remaining space in the
// buffer.
char intermediate[unibrow::Utf8::kMaxEncodedSize];
for (; i < len && pos < capacity; i++) {
i::uc32 c = write_input_buffer.GetNext();
int written = unibrow::Utf8::Encode(intermediate, c);
if (pos + written <= capacity) {
for (int j = 0; j < written; j++)
buffer[pos + j] = intermediate[j];
pos += written;
} else {
// We've reached the end of the buffer
break;
}
}
}
if (i == len && (capacity == -1 || pos < capacity))
buffer[pos++] = '\0';
return pos;
}
int String::WriteAscii(char* buffer, int start, int length) const {
if (IsDeadCheck("v8::String::WriteAscii()")) return 0;
LOG_API("String::WriteAscii");
ENTER_V8;
ASSERT(start >= 0 && length >= -1);
i::Handle<i::String> str = Utils::OpenHandle(this);
// Flatten the string for efficiency. This applies whether we are
// using StringInputBuffer or Get(i) to access the characters.
str->TryFlattenIfNotFlat();
int end = length;
if ( (length == -1) || (length > str->length() - start) )
end = str->length() - start;
if (end < 0) return 0;
write_input_buffer.Reset(start, *str);
int i;
for (i = 0; i < end; i++) {
char c = static_cast<char>(write_input_buffer.GetNext());
if (c == '\0') c = ' ';
buffer[i] = c;
}
if (length == -1 || i < length)
buffer[i] = '\0';
return i;
}
int String::Write(uint16_t* buffer, int start, int length) const {
if (IsDeadCheck("v8::String::Write()")) return 0;
LOG_API("String::Write");
ENTER_V8;
ASSERT(start >= 0 && length >= -1);
i::Handle<i::String> str = Utils::OpenHandle(this);
// Flatten the string for efficiency. This applies whether we are
// using StringInputBuffer or Get(i) to access the characters.
str->TryFlattenIfNotFlat();
int end = length;
if ( (length == -1) || (length > str->length() - start) )
end = str->length() - start;
if (end < 0) return 0;
write_input_buffer.Reset(start, *str);
int i;
for (i = 0; i < end; i++)
buffer[i] = write_input_buffer.GetNext();
if (length == -1 || i < length)
buffer[i] = '\0';
return i;
}
bool v8::String::IsExternal() const {
EnsureInitialized("v8::String::IsExternal()");
i::Handle<i::String> str = Utils::OpenHandle(this);
return i::StringShape(*str).IsExternalTwoByte();
}
bool v8::String::IsExternalAscii() const {
EnsureInitialized("v8::String::IsExternalAscii()");
i::Handle<i::String> str = Utils::OpenHandle(this);
return i::StringShape(*str).IsExternalAscii();
}
v8::String::ExternalStringResource*
v8::String::GetExternalStringResource() const {
EnsureInitialized("v8::String::GetExternalStringResource()");
i::Handle<i::String> str = Utils::OpenHandle(this);
ASSERT(str->IsExternalTwoByteString());
void* resource = i::Handle<i::ExternalTwoByteString>::cast(str)->resource();
return reinterpret_cast<ExternalStringResource*>(resource);
}
v8::String::ExternalAsciiStringResource*
v8::String::GetExternalAsciiStringResource() const {
EnsureInitialized("v8::String::GetExternalAsciiStringResource()");
i::Handle<i::String> str = Utils::OpenHandle(this);
ASSERT(str->IsExternalAsciiString());
void* resource = i::Handle<i::ExternalAsciiString>::cast(str)->resource();
return reinterpret_cast<ExternalAsciiStringResource*>(resource);
}
double Number::Value() const {
if (IsDeadCheck("v8::Number::Value()")) return 0;
i::Handle<i::Object> obj = Utils::OpenHandle(this);
return obj->Number();
}
bool Boolean::Value() const {
if (IsDeadCheck("v8::Boolean::Value()")) return false;
i::Handle<i::Object> obj = Utils::OpenHandle(this);
return obj->IsTrue();
}
int64_t Integer::Value() const {
if (IsDeadCheck("v8::Integer::Value()")) return 0;
i::Handle<i::Object> obj = Utils::OpenHandle(this);
if (obj->IsSmi()) {
return i::Smi::cast(*obj)->value();
} else {
return static_cast<int64_t>(obj->Number());
}
}
int32_t Int32::Value() const {
if (IsDeadCheck("v8::Int32::Value()")) return 0;
i::Handle<i::Object> obj = Utils::OpenHandle(this);
if (obj->IsSmi()) {
return i::Smi::cast(*obj)->value();
} else {
return static_cast<int32_t>(obj->Number());
}
}
int v8::Object::InternalFieldCount() {
if (IsDeadCheck("v8::Object::InternalFieldCount()")) return 0;
i::Handle<i::JSObject> obj = Utils::OpenHandle(this);
return obj->GetInternalFieldCount();
}
Local<Value> v8::Object::GetInternalField(int index) {
if (IsDeadCheck("v8::Object::GetInternalField()")) return Local<Value>();
i::Handle<i::JSObject> obj = Utils::OpenHandle(this);
if (!ApiCheck(index < obj->GetInternalFieldCount(),
"v8::Object::GetInternalField()",
"Reading internal field out of bounds")) {
return Local<Value>();
}
i::Handle<i::Object> value(obj->GetInternalField(index));
return Utils::ToLocal(value);
}
void v8::Object::SetInternalField(int index, v8::Handle<Value> value) {
if (IsDeadCheck("v8::Object::SetInternalField()")) return;
i::Handle<i::JSObject> obj = Utils::OpenHandle(this);
if (!ApiCheck(index < obj->GetInternalFieldCount(),
"v8::Object::SetInternalField()",
"Writing internal field out of bounds")) {
return;
}
ENTER_V8;
i::Handle<i::Object> val = Utils::OpenHandle(*value);
obj->SetInternalField(index, *val);
}
// --- E n v i r o n m e n t ---
bool v8::V8::Initialize() {
if (i::V8::HasBeenSetup()) return true;
ENTER_V8;
HandleScope scope;
if (i::Snapshot::Initialize()) {
return true;
} else {
return i::V8::Initialize(NULL);
}
}
bool v8::V8::Dispose() {
i::V8::TearDown();
return true;
}
const char* v8::V8::GetVersion() {
return "1.2.1 (candidate)";
}
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
static i::Handle<i::FunctionTemplateInfo>
EnsureConstructor(i::Handle<i::ObjectTemplateInfo> templ) {
if (templ->constructor()->IsUndefined()) {
Local<FunctionTemplate> constructor = FunctionTemplate::New();
Utils::OpenHandle(*constructor)->set_instance_template(*templ);
templ->set_constructor(*Utils::OpenHandle(*constructor));
}
return i::Handle<i::FunctionTemplateInfo>(
i::FunctionTemplateInfo::cast(templ->constructor()));
}
Persistent<Context> v8::Context::New(
v8::ExtensionConfiguration* extensions,
v8::Handle<ObjectTemplate> global_template,
v8::Handle<Value> global_object) {
EnsureInitialized("v8::Context::New()");
LOG_API("Context::New");
ON_BAILOUT("v8::Context::New()", return Persistent<Context>());
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
// Enter V8 via an ENTER_V8 scope.
i::Handle<i::Context> env;
{
ENTER_V8;
// Give the heap a chance to cleanup if we've disposed contexts.
i::Heap::CollectAllGarbageIfContextDisposed();
v8::Handle<ObjectTemplate> proxy_template = global_template;
i::Handle<i::FunctionTemplateInfo> proxy_constructor;
i::Handle<i::FunctionTemplateInfo> global_constructor;
if (!global_template.IsEmpty()) {
// Make sure that the global_template has a constructor.
global_constructor =
EnsureConstructor(Utils::OpenHandle(*global_template));
// Create a fresh template for the global proxy object.
proxy_template = ObjectTemplate::New();
proxy_constructor =
EnsureConstructor(Utils::OpenHandle(*proxy_template));
// Set the global template to be the prototype template of
// global proxy template.
proxy_constructor->set_prototype_template(
*Utils::OpenHandle(*global_template));
// Migrate security handlers from global_template to
// proxy_template. Temporarily removing access check
// information from the global template.
if (!global_constructor->access_check_info()->IsUndefined()) {
proxy_constructor->set_access_check_info(
global_constructor->access_check_info());
proxy_constructor->set_needs_access_check(
global_constructor->needs_access_check());
global_constructor->set_needs_access_check(false);
global_constructor->set_access_check_info(i::Heap::undefined_value());
}
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
}
// Create the environment.
env = i::Bootstrapper::CreateEnvironment(
Utils::OpenHandle(*global_object),
proxy_template,
extensions);
// Restore the access check info on the global template.
if (!global_template.IsEmpty()) {
ASSERT(!global_constructor.is_null());
ASSERT(!proxy_constructor.is_null());
global_constructor->set_access_check_info(
proxy_constructor->access_check_info());
global_constructor->set_needs_access_check(
proxy_constructor->needs_access_check());
}
}
// Leave V8.
if (!ApiCheck(!env.is_null(),
"v8::Context::New()",
"Could not initialize environment"))
return Persistent<Context>();
return Persistent<Context>(Utils::ToLocal(env));
}
void v8::Context::SetSecurityToken(Handle<Value> token) {
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
if (IsDeadCheck("v8::Context::SetSecurityToken()")) return;
ENTER_V8;
i::Handle<i::Context> env = Utils::OpenHandle(this);
i::Handle<i::Object> token_handle = Utils::OpenHandle(*token);
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
env->set_security_token(*token_handle);
}
void v8::Context::UseDefaultSecurityToken() {
if (IsDeadCheck("v8::Context::UseDefaultSecurityToken()")) return;
ENTER_V8;
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
i::Handle<i::Context> env = Utils::OpenHandle(this);
env->set_security_token(env->global());
}
Handle<Value> v8::Context::GetSecurityToken() {
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
if (IsDeadCheck("v8::Context::GetSecurityToken()")) return Handle<Value>();
i::Handle<i::Context> env = Utils::OpenHandle(this);
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
i::Object* security_token = env->security_token();
i::Handle<i::Object> token_handle(security_token);
return Utils::ToLocal(token_handle);
}
bool Context::HasOutOfMemoryException() {
i::Handle<i::Context> env = Utils::OpenHandle(this);
return env->has_out_of_memory();
}
bool Context::InContext() {
return i::Top::context() != NULL;
}
v8::Local<v8::Context> Context::GetEntered() {
if (IsDeadCheck("v8::Context::GetEntered()")) return Local<Context>();
i::Handle<i::Object> last = thread_local.LastEnteredContext();
if (last.is_null()) return Local<Context>();
i::Handle<i::Context> context = i::Handle<i::Context>::cast(last);
return Utils::ToLocal(context);
}
v8::Local<v8::Context> Context::GetCurrent() {
if (IsDeadCheck("v8::Context::GetCurrent()")) return Local<Context>();
i::Handle<i::Context> context(i::Top::global_context());
return Utils::ToLocal(context);
}
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
v8::Local<v8::Object> Context::Global() {
if (IsDeadCheck("v8::Context::Global()")) return Local<v8::Object>();
i::Object** ctx = reinterpret_cast<i::Object**>(this);
i::Handle<i::Context> context =
i::Handle<i::Context>::cast(i::Handle<i::Object>(ctx));
i::Handle<i::Object> global(context->global_proxy());
return Utils::ToLocal(i::Handle<i::JSObject>::cast(global));
}
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
void Context::DetachGlobal() {
if (IsDeadCheck("v8::Context::DetachGlobal()")) return;
ENTER_V8;
i::Object** ctx = reinterpret_cast<i::Object**>(this);
i::Handle<i::Context> context =
i::Handle<i::Context>::cast(i::Handle<i::Object>(ctx));
Split window support from V8. Here is a description of the background and design of split window in Chrome and V8: https://docs.google.com/a/google.com/Doc?id=chhjkpg_47fwddxbfr This change list splits the window object into two parts: 1) an inner window object used as the global object of contexts; 2) an outer window object exposed to JavaScript and accessible by the name 'window'. Firefox did it awhile ago, here are some discussions: https://wiki.mozilla.org/Gecko:SplitWindow. One additional benefit of splitting window in Chrome is that accessing global variables don't need security checks anymore, it can improve applications that use many global variables. V8 support of split window: There are a small number of changes on V8 api to support split window: Security context is removed from V8, so does related API functions; A global object can be detached from its context and reused by a new context; Access checks on an object template can be turned on/off by default; An object can turn on its access checks later; V8 has a new object type, ApiGlobalObject, which is the outer window object type. The existing JSGlobalObject becomes the inner window object type. Security checks are moved from JSGlobalObject to ApiGlobalObject. ApiGlobalObject is the one exposed to JavaScript, it is accessible through Context::Global(). ApiGlobalObject's prototype is set to JSGlobalObject so that property lookups are forwarded to JSGlobalObject. ApiGlobalObject forwards all other property access requests to JSGlobalObject, such as SetProperty, DeleteProperty, etc. Security token is moved to a global context, and ApiGlobalObject has a reference to its global context. JSGlobalObject has a reference to its global context as well. When accessing properties on a global object in JavaScript, the domain security check is performed by comparing the security token of the lexical context (Top::global_context()) to the token of global object's context. The check is only needed when the receiver is a window object, such as 'window.document'. Accessing global variables, such as 'var foo = 3; foo' does not need checks because the receiver is the inner window object. When an outer window is detached from its global context (when a frame navigates away from a page), it is completely detached from the inner window. A new context is created for the new page, and the outer global object is reused. At this point, the access check on the DOMWindow wrapper of the old context is turned on. The code in old context is still able to access DOMWindow properties, but it has to go through domain security checks. It is debatable on how to implement the outer window object. Currently each property access function has to check if the receiver is ApiGlobalObject type. This approach might be error-prone that one may forget to check the receiver when adding new functions. It is unlikely a performance issue because accessing global variables are more common than 'window.foo' style coding. I am still working on the ARM port, and I'd like to hear comments and suggestions on the best way to support it in V8. Review URL: http://codereview.chromium.org/7366 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@540 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2008-10-21 19:07:58 +00:00
i::Bootstrapper::DetachGlobal(context);
}
Local<v8::Object> ObjectTemplate::NewInstance() {
ON_BAILOUT("v8::ObjectTemplate::NewInstance()", return Local<v8::Object>());
LOG_API("ObjectTemplate::NewInstance");
ENTER_V8;
EXCEPTION_PREAMBLE();
i::Handle<i::Object> obj =
i::Execution::InstantiateObject(Utils::OpenHandle(this),
&has_pending_exception);
EXCEPTION_BAILOUT_CHECK(Local<v8::Object>());
return Utils::ToLocal(i::Handle<i::JSObject>::cast(obj));
}
Local<v8::Function> FunctionTemplate::GetFunction() {
ON_BAILOUT("v8::FunctionTemplate::GetFunction()",
return Local<v8::Function>());
LOG_API("FunctionTemplate::GetFunction");
ENTER_V8;
EXCEPTION_PREAMBLE();
i::Handle<i::Object> obj =
i::Execution::InstantiateFunction(Utils::OpenHandle(this),
&has_pending_exception);
EXCEPTION_BAILOUT_CHECK(Local<v8::Function>());
return Utils::ToLocal(i::Handle<i::JSFunction>::cast(obj));
}
bool FunctionTemplate::HasInstance(v8::Handle<v8::Value> value) {
ON_BAILOUT("v8::FunctionTemplate::HasInstanceOf()", return false);
i::Object* obj = *Utils::OpenHandle(*value);
return obj->IsInstanceOf(*Utils::OpenHandle(this));
}
static Local<External> ExternalNewImpl(void* data) {
return Utils::ToLocal(i::Factory::NewProxy(static_cast<i::Address>(data)));
}
static void* ExternalValueImpl(i::Handle<i::Object> obj) {
return reinterpret_cast<void*>(i::Proxy::cast(*obj)->proxy());
}
static const intptr_t kAlignedPointerMask = 3;
static const int kAlignedPointerShift = 2;
Local<Value> v8::External::Wrap(void* data) {
STATIC_ASSERT(sizeof(data) == sizeof(i::Address));
LOG_API("External::Wrap");
EnsureInitialized("v8::External::Wrap()");
ENTER_V8;
if ((reinterpret_cast<intptr_t>(data) & kAlignedPointerMask) == 0) {
uintptr_t data_ptr = reinterpret_cast<uintptr_t>(data);
int data_value = static_cast<int>(data_ptr >> kAlignedPointerShift);
STATIC_ASSERT(sizeof(data_ptr) == sizeof(data_value));
i::Handle<i::Object> obj(i::Smi::FromInt(data_value));
return Utils::ToLocal(obj);
}
return ExternalNewImpl(data);
}
void* v8::External::Unwrap(v8::Handle<v8::Value> value) {
if (IsDeadCheck("v8::External::Unwrap()")) return 0;
i::Handle<i::Object> obj = Utils::OpenHandle(*value);
if (obj->IsSmi()) {
// The external value was an aligned pointer.
uintptr_t result = i::Smi::cast(*obj)->value() << kAlignedPointerShift;
return reinterpret_cast<void*>(result);
}
return ExternalValueImpl(obj);
}
Local<External> v8::External::New(void* data) {
STATIC_ASSERT(sizeof(data) == sizeof(i::Address));
LOG_API("External::New");
EnsureInitialized("v8::External::New()");
ENTER_V8;
return ExternalNewImpl(data);
}
void* External::Value() const {
if (IsDeadCheck("v8::External::Value()")) return 0;
i::Handle<i::Object> obj = Utils::OpenHandle(this);
return ExternalValueImpl(obj);
}
Local<String> v8::String::Empty() {
EnsureInitialized("v8::String::Empty()");
LOG_API("String::Empty()");
return Utils::ToLocal(i::Factory::empty_symbol());
}
Local<String> v8::String::New(const char* data, int length) {
EnsureInitialized("v8::String::New()");
LOG_API("String::New(char)");
if (length == 0) return Empty();
ENTER_V8;
if (length == -1) length = strlen(data);
i::Handle<i::String> result =
i::Factory::NewStringFromUtf8(i::Vector<const char>(data, length));
return Utils::ToLocal(result);
}
Local<String> v8::String::NewUndetectable(const char* data, int length) {
EnsureInitialized("v8::String::NewUndetectable()");
LOG_API("String::NewUndetectable(char)");
ENTER_V8;
if (length == -1) length = strlen(data);
i::Handle<i::String> result =
i::Factory::NewStringFromUtf8(i::Vector<const char>(data, length));
result->MarkAsUndetectable();
return Utils::ToLocal(result);
}
static int TwoByteStringLength(const uint16_t* data) {
int length = 0;
while (data[length] != '\0') length++;
return length;
}
Local<String> v8::String::New(const uint16_t* data, int length) {
EnsureInitialized("v8::String::New()");
LOG_API("String::New(uint16_)");
if (length == 0) return Empty();
ENTER_V8;
if (length == -1) length = TwoByteStringLength(data);
i::Handle<i::String> result =
i::Factory::NewStringFromTwoByte(i::Vector<const uint16_t>(data, length));
return Utils::ToLocal(result);
}
Local<String> v8::String::NewUndetectable(const uint16_t* data, int length) {
EnsureInitialized("v8::String::NewUndetectable()");
LOG_API("String::NewUndetectable(uint16_)");
ENTER_V8;
if (length == -1) length = TwoByteStringLength(data);
i::Handle<i::String> result =
i::Factory::NewStringFromTwoByte(i::Vector<const uint16_t>(data, length));
result->MarkAsUndetectable();
return Utils::ToLocal(result);
}
i::Handle<i::String> NewExternalStringHandle(
v8::String::ExternalStringResource* resource) {
i::Handle<i::String> result =
i::Factory::NewExternalStringFromTwoByte(resource);
return result;
}
i::Handle<i::String> NewExternalAsciiStringHandle(
v8::String::ExternalAsciiStringResource* resource) {
i::Handle<i::String> result =
i::Factory::NewExternalStringFromAscii(resource);
return result;
}
static void DisposeExternalString(v8::Persistent<v8::Value> obj,
void* parameter) {
ENTER_V8;
i::ExternalTwoByteString* str =
i::ExternalTwoByteString::cast(*Utils::OpenHandle(*obj));
// External symbols are deleted when they are pruned out of the symbol
// table. Generally external symbols are not registered with the weak handle
// callbacks unless they are upgraded to a symbol after being externalized.
if (!str->IsSymbol()) {
v8::String::ExternalStringResource* resource =
reinterpret_cast<v8::String::ExternalStringResource*>(parameter);
if (resource != NULL) {
const size_t total_size = resource->length() * sizeof(*resource->data());
i::Counters::total_external_string_memory.Decrement(total_size);
// The object will continue to live in the JavaScript heap until the
// handle is entirely cleaned out by the next GC. For example the
// destructor for the resource below could bring it back to life again.
// Which is why we make sure to not have a dangling pointer here.
str->set_resource(NULL);
delete resource;
}
}
// In any case we do not need this handle any longer.
obj.Dispose();
}
static void DisposeExternalAsciiString(v8::Persistent<v8::Value> obj,
void* parameter) {
ENTER_V8;
i::ExternalAsciiString* str =
i::ExternalAsciiString::cast(*Utils::OpenHandle(*obj));
// External symbols are deleted when they are pruned out of the symbol
// table. Generally external symbols are not registered with the weak handle
// callbacks unless they are upgraded to a symbol after being externalized.
if (!str->IsSymbol()) {
v8::String::ExternalAsciiStringResource* resource =
reinterpret_cast<v8::String::ExternalAsciiStringResource*>(parameter);
if (resource != NULL) {
const size_t total_size = resource->length() * sizeof(*resource->data());
i::Counters::total_external_string_memory.Decrement(total_size);
// The object will continue to live in the JavaScript heap until the
// handle is entirely cleaned out by the next GC. For example the
// destructor for the resource below could bring it back to life again.
// Which is why we make sure to not have a dangling pointer here.
str->set_resource(NULL);
delete resource;
}
}
// In any case we do not need this handle any longer.
obj.Dispose();
}
Local<String> v8::String::NewExternal(
v8::String::ExternalStringResource* resource) {
EnsureInitialized("v8::String::NewExternal()");
LOG_API("String::NewExternal");
ENTER_V8;
const size_t total_size = resource->length() * sizeof(*resource->data());
i::Counters::total_external_string_memory.Increment(total_size);
i::Handle<i::String> result = NewExternalStringHandle(resource);
i::Handle<i::Object> handle = i::GlobalHandles::Create(*result);
i::GlobalHandles::MakeWeak(handle.location(),
resource,
&DisposeExternalString);
return Utils::ToLocal(result);
}
bool v8::String::MakeExternal(v8::String::ExternalStringResource* resource) {
if (IsDeadCheck("v8::String::MakeExternal()")) return false;
if (this->IsExternal()) return false; // Already an external string.
ENTER_V8;
i::Handle <i::String> obj = Utils::OpenHandle(this);
bool result = obj->MakeExternal(resource);
if (result && !obj->IsSymbol()) {
// Operation was successful and the string is not a symbol. In this case
// we need to make sure that the we call the destructor for the external
// resource when no strong references to the string remain.
i::Handle<i::Object> handle = i::GlobalHandles::Create(*obj);
i::GlobalHandles::MakeWeak(handle.location(),
resource,
&DisposeExternalString);
}
return result;
}
Local<String> v8::String::NewExternal(
v8::String::ExternalAsciiStringResource* resource) {
EnsureInitialized("v8::String::NewExternal()");
LOG_API("String::NewExternal");
ENTER_V8;
const size_t total_size = resource->length() * sizeof(*resource->data());
i::Counters::total_external_string_memory.Increment(total_size);
i::Handle<i::String> result = NewExternalAsciiStringHandle(resource);
i::Handle<i::Object> handle = i::GlobalHandles::Create(*result);
i::GlobalHandles::MakeWeak(handle.location(),
resource,
&DisposeExternalAsciiString);
return Utils::ToLocal(result);
}
bool v8::String::MakeExternal(
v8::String::ExternalAsciiStringResource* resource) {
if (IsDeadCheck("v8::String::MakeExternal()")) return false;
if (this->IsExternal()) return false; // Already an external string.
ENTER_V8;
i::Handle <i::String> obj = Utils::OpenHandle(this);
bool result = obj->MakeExternal(resource);
if (result && !obj->IsSymbol()) {
// Operation was successful and the string is not a symbol. In this case
// we need to make sure that the we call the destructor for the external
// resource when no strong references to the string remain.
i::Handle<i::Object> handle = i::GlobalHandles::Create(*obj);
i::GlobalHandles::MakeWeak(handle.location(),
resource,
&DisposeExternalAsciiString);
}
return result;
}
Local<v8::Object> v8::Object::New() {
EnsureInitialized("v8::Object::New()");
LOG_API("Object::New");
ENTER_V8;
i::Handle<i::JSObject> obj =
i::Factory::NewJSObject(i::Top::object_function());
return Utils::ToLocal(obj);
}
Local<v8::Value> v8::Date::New(double time) {
EnsureInitialized("v8::Date::New()");
LOG_API("Date::New");
ENTER_V8;
EXCEPTION_PREAMBLE();
i::Handle<i::Object> obj =
i::Execution::NewDate(time, &has_pending_exception);
EXCEPTION_BAILOUT_CHECK(Local<v8::Value>());
return Utils::ToLocal(obj);
}
double v8::Date::NumberValue() const {
if (IsDeadCheck("v8::Date::NumberValue()")) return 0;
LOG_API("Date::NumberValue");
i::Handle<i::Object> obj = Utils::OpenHandle(this);
i::Handle<i::JSValue> jsvalue = i::Handle<i::JSValue>::cast(obj);
return jsvalue->value()->Number();
}
Local<v8::Array> v8::Array::New(int length) {
EnsureInitialized("v8::Array::New()");
LOG_API("Array::New");
ENTER_V8;
i::Handle<i::JSArray> obj = i::Factory::NewJSArray(length);
return Utils::ToLocal(obj);
}
uint32_t v8::Array::Length() const {
if (IsDeadCheck("v8::Array::Length()")) return 0;
i::Handle<i::JSArray> obj = Utils::OpenHandle(this);
i::Object* length = obj->length();
if (length->IsSmi()) {
return i::Smi::cast(length)->value();
} else {
return static_cast<uint32_t>(length->Number());
}
}
Local<String> v8::String::NewSymbol(const char* data, int length) {
EnsureInitialized("v8::String::NewSymbol()");
LOG_API("String::NewSymbol(char)");
ENTER_V8;
if (length == -1) length = strlen(data);
i::Handle<i::String> result =
i::Factory::LookupSymbol(i::Vector<const char>(data, length));
return Utils::ToLocal(result);
}
Local<Number> v8::Number::New(double value) {
EnsureInitialized("v8::Number::New()");
ENTER_V8;
i::Handle<i::Object> result = i::Factory::NewNumber(value);
return Utils::NumberToLocal(result);
}
Local<Integer> v8::Integer::New(int32_t value) {
EnsureInitialized("v8::Integer::New()");
if (i::Smi::IsValid(value)) {
return Utils::IntegerToLocal(i::Handle<i::Object>(i::Smi::FromInt(value)));
}
ENTER_V8;
i::Handle<i::Object> result = i::Factory::NewNumber(value);
return Utils::IntegerToLocal(result);
}
void V8::IgnoreOutOfMemoryException() {
thread_local.SetIgnoreOutOfMemory(true);
}
bool V8::AddMessageListener(MessageCallback that, Handle<Value> data) {
EnsureInitialized("v8::V8::AddMessageListener()");
ON_BAILOUT("v8::V8::AddMessageListener()", return false);
ENTER_V8;
HandleScope scope;
NeanderArray listeners(i::Factory::message_listeners());
NeanderObject obj(2);
obj.set(0, *i::Factory::NewProxy(FUNCTION_ADDR(that)));
obj.set(1, data.IsEmpty() ?
i::Heap::undefined_value() :
*Utils::OpenHandle(*data));
listeners.add(obj.value());
return true;
}
void V8::RemoveMessageListeners(MessageCallback that) {
EnsureInitialized("v8::V8::RemoveMessageListener()");
ON_BAILOUT("v8::V8::RemoveMessageListeners()", return);
ENTER_V8;
HandleScope scope;
NeanderArray listeners(i::Factory::message_listeners());
for (int i = 0; i < listeners.length(); i++) {
if (listeners.get(i)->IsUndefined()) continue; // skip deleted ones
NeanderObject listener(i::JSObject::cast(listeners.get(i)));
i::Handle<i::Proxy> callback_obj(i::Proxy::cast(listener.get(0)));
if (callback_obj->proxy() == FUNCTION_ADDR(that)) {
listeners.set(i, i::Heap::undefined_value());
}
}
}
void V8::SetCounterFunction(CounterLookupCallback callback) {
if (IsDeadCheck("v8::V8::SetCounterFunction()")) return;
i::StatsTable::SetCounterFunction(callback);
}
void V8::SetCreateHistogramFunction(CreateHistogramCallback callback) {
if (IsDeadCheck("v8::V8::SetCreateHistogramFunction()")) return;
i::StatsTable::SetCreateHistogramFunction(callback);
}
void V8::SetAddHistogramSampleFunction(AddHistogramSampleCallback callback) {
if (IsDeadCheck("v8::V8::SetAddHistogramSampleFunction()")) return;
i::StatsTable::SetAddHistogramSampleFunction(callback);
}
void V8::EnableSlidingStateWindow() {
if (IsDeadCheck("v8::V8::EnableSlidingStateWindow()")) return;
i::Logger::EnableSlidingStateWindow();
}
void V8::SetFailedAccessCheckCallbackFunction(
FailedAccessCheckCallback callback) {
if (IsDeadCheck("v8::V8::SetFailedAccessCheckCallbackFunction()")) return;
i::Top::SetFailedAccessCheckCallback(callback);
}
void V8::AddObjectGroup(Persistent<Value>* objects, size_t length) {
if (IsDeadCheck("v8::V8::AddObjectGroup()")) return;
STATIC_ASSERT(sizeof(Persistent<Value>) == sizeof(i::Object**));
i::GlobalHandles::AddGroup(reinterpret_cast<i::Object***>(objects), length);
}
int V8::AdjustAmountOfExternalAllocatedMemory(int change_in_bytes) {
if (IsDeadCheck("v8::V8::AdjustAmountOfExternalAllocatedMemory()")) return 0;
return i::Heap::AdjustAmountOfExternalAllocatedMemory(change_in_bytes);
}
void V8::SetGlobalGCPrologueCallback(GCCallback callback) {
if (IsDeadCheck("v8::V8::SetGlobalGCPrologueCallback()")) return;
i::Heap::SetGlobalGCPrologueCallback(callback);
}
void V8::SetGlobalGCEpilogueCallback(GCCallback callback) {
if (IsDeadCheck("v8::V8::SetGlobalGCEpilogueCallback()")) return;
i::Heap::SetGlobalGCEpilogueCallback(callback);
}
void V8::PauseProfiler() {
#ifdef ENABLE_LOGGING_AND_PROFILING
i::Logger::PauseProfiler();
#endif
}
void V8::ResumeProfiler() {
#ifdef ENABLE_LOGGING_AND_PROFILING
i::Logger::ResumeProfiler();
#endif
}
String::Utf8Value::Utf8Value(v8::Handle<v8::Value> obj) {
EnsureInitialized("v8::String::Utf8Value::Utf8Value()");
if (obj.IsEmpty()) {
str_ = NULL;
length_ = 0;
return;
}
ENTER_V8;
HandleScope scope;
TryCatch try_catch;
Handle<String> str = obj->ToString();
if (str.IsEmpty()) {
str_ = NULL;
length_ = 0;
} else {
length_ = str->Utf8Length();
str_ = i::NewArray<char>(length_ + 1);
str->WriteUtf8(str_);
}
}
String::Utf8Value::~Utf8Value() {
i::DeleteArray(str_);
}
String::AsciiValue::AsciiValue(v8::Handle<v8::Value> obj) {
EnsureInitialized("v8::String::AsciiValue::AsciiValue()");
if (obj.IsEmpty()) {
str_ = NULL;
length_ = 0;
return;
}
ENTER_V8;
HandleScope scope;
TryCatch try_catch;
Handle<String> str = obj->ToString();
if (str.IsEmpty()) {
str_ = NULL;
length_ = 0;
} else {
length_ = str->Length();
str_ = i::NewArray<char>(length_ + 1);
str->WriteAscii(str_);
}
}
String::AsciiValue::~AsciiValue() {
i::DeleteArray(str_);
}
String::Value::Value(v8::Handle<v8::Value> obj) {
EnsureInitialized("v8::String::Value::Value()");
if (obj.IsEmpty()) {
str_ = NULL;
length_ = 0;
return;
}
ENTER_V8;
HandleScope scope;
TryCatch try_catch;
Handle<String> str = obj->ToString();
if (str.IsEmpty()) {
str_ = NULL;
length_ = 0;
} else {
length_ = str->Length();
str_ = i::NewArray<uint16_t>(length_ + 1);
str->Write(str_);
}
}
String::Value::~Value() {
i::DeleteArray(str_);
}
Local<Value> Exception::RangeError(v8::Handle<v8::String> raw_message) {
LOG_API("RangeError");
ON_BAILOUT("v8::Exception::RangeError()", return Local<Value>());
ENTER_V8;
i::Object* error;
{
HandleScope scope;
i::Handle<i::String> message = Utils::OpenHandle(*raw_message);
i::Handle<i::Object> result = i::Factory::NewRangeError(message);
error = *result;
}
i::Handle<i::Object> result(error);
return Utils::ToLocal(result);
}
Local<Value> Exception::ReferenceError(v8::Handle<v8::String> raw_message) {
LOG_API("ReferenceError");
ON_BAILOUT("v8::Exception::ReferenceError()", return Local<Value>());
ENTER_V8;
i::Object* error;
{
HandleScope scope;
i::Handle<i::String> message = Utils::OpenHandle(*raw_message);
i::Handle<i::Object> result = i::Factory::NewReferenceError(message);
error = *result;
}
i::Handle<i::Object> result(error);
return Utils::ToLocal(result);
}
Local<Value> Exception::SyntaxError(v8::Handle<v8::String> raw_message) {
LOG_API("SyntaxError");
ON_BAILOUT("v8::Exception::SyntaxError()", return Local<Value>());
ENTER_V8;
i::Object* error;
{
HandleScope scope;
i::Handle<i::String> message = Utils::OpenHandle(*raw_message);
i::Handle<i::Object> result = i::Factory::NewSyntaxError(message);
error = *result;
}
i::Handle<i::Object> result(error);
return Utils::ToLocal(result);
}
Local<Value> Exception::TypeError(v8::Handle<v8::String> raw_message) {
LOG_API("TypeError");
ON_BAILOUT("v8::Exception::TypeError()", return Local<Value>());
ENTER_V8;
i::Object* error;
{
HandleScope scope;
i::Handle<i::String> message = Utils::OpenHandle(*raw_message);
i::Handle<i::Object> result = i::Factory::NewTypeError(message);
error = *result;
}
i::Handle<i::Object> result(error);
return Utils::ToLocal(result);
}
Local<Value> Exception::Error(v8::Handle<v8::String> raw_message) {
LOG_API("Error");
ON_BAILOUT("v8::Exception::Error()", return Local<Value>());
ENTER_V8;
i::Object* error;
{
HandleScope scope;
i::Handle<i::String> message = Utils::OpenHandle(*raw_message);
i::Handle<i::Object> result = i::Factory::NewError(message);
error = *result;
}
i::Handle<i::Object> result(error);
return Utils::ToLocal(result);
}
// --- D e b u g S u p p o r t ---
#ifdef ENABLE_DEBUGGER_SUPPORT
bool Debug::SetDebugEventListener(EventCallback that, Handle<Value> data) {
EnsureInitialized("v8::Debug::SetDebugEventListener()");
ON_BAILOUT("v8::Debug::SetDebugEventListener()", return false);
ENTER_V8;
HandleScope scope;
i::Handle<i::Object> proxy = i::Factory::undefined_value();
if (that != NULL) {
proxy = i::Factory::NewProxy(FUNCTION_ADDR(that));
}
i::Debugger::SetEventListener(proxy, Utils::OpenHandle(*data));
return true;
}
bool Debug::SetDebugEventListener(v8::Handle<v8::Object> that,
Handle<Value> data) {
ON_BAILOUT("v8::Debug::SetDebugEventListener()", return false);
ENTER_V8;
i::Debugger::SetEventListener(Utils::OpenHandle(*that),
Utils::OpenHandle(*data));
return true;
}
void Debug::DebugBreak() {
if (!i::V8::HasBeenSetup()) return;
i::StackGuard::DebugBreak();
}
void Debug::SetMessageHandler(v8::Debug::MessageHandler handler,
bool message_handler_thread) {
EnsureInitialized("v8::Debug::SetMessageHandler");
ENTER_V8;
// Message handler thread not supported any more. Parameter temporally left in
// the API for client compatability reasons.
CHECK(!message_handler_thread);
i::Debugger::SetMessageHandler(handler);
}
void Debug::SendCommand(const uint16_t* command, int length,
ClientData* client_data) {
if (!i::V8::HasBeenSetup()) return;
i::Debugger::ProcessCommand(i::Vector<const uint16_t>(command, length),
client_data);
}
void Debug::SetHostDispatchHandler(HostDispatchHandler handler,
int period) {
EnsureInitialized("v8::Debug::SetHostDispatchHandler");
ENTER_V8;
i::Debugger::SetHostDispatchHandler(handler, period);
}
Handle<Value> Debug::Call(v8::Handle<v8::Function> fun,
v8::Handle<v8::Value> data) {
if (!i::V8::HasBeenSetup()) return Handle<Value>();
ON_BAILOUT("v8::Debug::Call()", return Handle<Value>());
ENTER_V8;
i::Handle<i::Object> result;
EXCEPTION_PREAMBLE();
if (data.IsEmpty()) {
result = i::Debugger::Call(Utils::OpenHandle(*fun),
i::Factory::undefined_value(),
&has_pending_exception);
} else {
result = i::Debugger::Call(Utils::OpenHandle(*fun),
Utils::OpenHandle(*data),
&has_pending_exception);
}
EXCEPTION_BAILOUT_CHECK(Local<Value>());
return Utils::ToLocal(result);
}
bool Debug::EnableAgent(const char* name, int port) {
return i::Debugger::StartAgent(name, port);
}
#endif // ENABLE_DEBUGGER_SUPPORT
namespace internal {
HandleScopeImplementer* HandleScopeImplementer::instance() {
return &thread_local;
}
char* HandleScopeImplementer::ArchiveThread(char* storage) {
return thread_local.ArchiveThreadHelper(storage);
}
char* HandleScopeImplementer::ArchiveThreadHelper(char* storage) {
v8::ImplementationUtilities::HandleScopeData* current =
v8::ImplementationUtilities::CurrentHandleScope();
handle_scope_data_ = *current;
memcpy(storage, this, sizeof(*this));
Initialize();
current->Initialize();
return storage + ArchiveSpacePerThread();
}
int HandleScopeImplementer::ArchiveSpacePerThread() {
return sizeof(thread_local);
}
char* HandleScopeImplementer::RestoreThread(char* storage) {
return thread_local.RestoreThreadHelper(storage);
}
char* HandleScopeImplementer::RestoreThreadHelper(char* storage) {
memcpy(this, storage, sizeof(*this));
*v8::ImplementationUtilities::CurrentHandleScope() = handle_scope_data_;
return storage + ArchiveSpacePerThread();
}
void HandleScopeImplementer::Iterate(
ObjectVisitor* v,
List<void**>* blocks,
v8::ImplementationUtilities::HandleScopeData* handle_data) {
// Iterate over all handles in the blocks except for the last.
for (int i = blocks->length() - 2; i >= 0; --i) {
Object** block =
reinterpret_cast<Object**>(blocks->at(i));
v->VisitPointers(block, &block[kHandleBlockSize]);
}
// Iterate over live handles in the last block (if any).
if (!blocks->is_empty()) {
v->VisitPointers(reinterpret_cast<Object**>(blocks->last()),
reinterpret_cast<Object**>(handle_data->next));
}
}
void HandleScopeImplementer::Iterate(ObjectVisitor* v) {
v8::ImplementationUtilities::HandleScopeData* current =
v8::ImplementationUtilities::CurrentHandleScope();
Iterate(v, thread_local.Blocks(), current);
}
char* HandleScopeImplementer::Iterate(ObjectVisitor* v, char* storage) {
HandleScopeImplementer* thread_local =
reinterpret_cast<HandleScopeImplementer*>(storage);
List<void**>* blocks_of_archived_thread = thread_local->Blocks();
v8::ImplementationUtilities::HandleScopeData* handle_data_of_archived_thread =
&thread_local->handle_scope_data_;
Iterate(v, blocks_of_archived_thread, handle_data_of_archived_thread);
return storage + ArchiveSpacePerThread();
}
} } // namespace v8::internal