ladybird/Userland/Libraries/LibJS/Runtime/ECMAScriptFunctionObject.cpp

870 lines
41 KiB
C++
Raw Normal View History

/*
* Copyright (c) 2020, Stephan Unverwerth <s.unverwerth@serenityos.org>
* Copyright (c) 2020-2022, Linus Groh <linusg@serenityos.org>
*
* SPDX-License-Identifier: BSD-2-Clause
*/
#include <AK/Debug.h>
#include <AK/Function.h>
#include <LibJS/AST.h>
#include <LibJS/Bytecode/BasicBlock.h>
#include <LibJS/Bytecode/Generator.h>
#include <LibJS/Bytecode/Interpreter.h>
#include <LibJS/Interpreter.h>
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
#include <LibJS/Runtime/AbstractOperations.h>
2020-05-04 16:05:13 +01:00
#include <LibJS/Runtime/Array.h>
#include <LibJS/Runtime/AsyncFunctionDriverWrapper.h>
#include <LibJS/Runtime/ECMAScriptFunctionObject.h>
#include <LibJS/Runtime/Error.h>
#include <LibJS/Runtime/ExecutionContext.h>
#include <LibJS/Runtime/FunctionEnvironment.h>
#include <LibJS/Runtime/GeneratorObject.h>
#include <LibJS/Runtime/GeneratorPrototype.h>
#include <LibJS/Runtime/GlobalObject.h>
#include <LibJS/Runtime/NativeFunction.h>
#include <LibJS/Runtime/PromiseConstructor.h>
#include <LibJS/Runtime/PromiseReaction.h>
#include <LibJS/Runtime/Value.h>
namespace JS {
2021-10-11 20:29:31 +02:00
ECMAScriptFunctionObject* ECMAScriptFunctionObject::create(GlobalObject& global_object, FlyString name, Statement const& ecmascript_code, Vector<FunctionNode::Parameter> parameters, i32 m_function_length, Environment* parent_scope, PrivateEnvironment* private_scope, FunctionKind kind, bool is_strict, bool might_need_arguments_object, bool contains_direct_call_to_eval, bool is_arrow_function)
{
Object* prototype = nullptr;
switch (kind) {
case FunctionKind::Normal:
prototype = global_object.function_prototype();
break;
case FunctionKind::Generator:
prototype = global_object.generator_function_prototype();
break;
case FunctionKind::Async:
prototype = global_object.async_function_prototype();
break;
2021-11-15 01:53:24 +01:00
case FunctionKind::AsyncGenerator:
prototype = global_object.async_generator_function_prototype();
break;
}
2021-10-11 20:29:31 +02:00
return global_object.heap().allocate<ECMAScriptFunctionObject>(global_object, move(name), ecmascript_code, move(parameters), m_function_length, parent_scope, private_scope, *prototype, kind, is_strict, might_need_arguments_object, contains_direct_call_to_eval, is_arrow_function);
}
ECMAScriptFunctionObject* ECMAScriptFunctionObject::create(GlobalObject& global_object, FlyString name, Object& prototype, Statement const& ecmascript_code, Vector<FunctionNode::Parameter> parameters, i32 m_function_length, Environment* parent_scope, PrivateEnvironment* private_scope, FunctionKind kind, bool is_strict, bool might_need_arguments_object, bool contains_direct_call_to_eval, bool is_arrow_function)
{
return global_object.heap().allocate<ECMAScriptFunctionObject>(global_object, move(name), ecmascript_code, move(parameters), m_function_length, parent_scope, private_scope, prototype, kind, is_strict, might_need_arguments_object, contains_direct_call_to_eval, is_arrow_function);
}
2021-10-11 20:29:31 +02:00
ECMAScriptFunctionObject::ECMAScriptFunctionObject(FlyString name, Statement const& ecmascript_code, Vector<FunctionNode::Parameter> formal_parameters, i32 function_length, Environment* parent_scope, PrivateEnvironment* private_scope, Object& prototype, FunctionKind kind, bool strict, bool might_need_arguments_object, bool contains_direct_call_to_eval, bool is_arrow_function)
: FunctionObject(prototype)
, m_environment(parent_scope)
2021-10-11 20:29:31 +02:00
, m_private_environment(private_scope)
, m_formal_parameters(move(formal_parameters))
, m_ecmascript_code(ecmascript_code)
, m_realm(global_object().associated_realm())
, m_strict(strict)
, m_name(move(name))
, m_function_length(function_length)
, m_kind(kind)
, m_might_need_arguments_object(might_need_arguments_object)
, m_contains_direct_call_to_eval(contains_direct_call_to_eval)
, m_is_arrow_function(is_arrow_function)
{
// NOTE: This logic is from OrdinaryFunctionCreate, https://tc39.es/ecma262/#sec-ordinaryfunctioncreate
if (m_is_arrow_function)
m_this_mode = ThisMode::Lexical;
else if (m_strict)
m_this_mode = ThisMode::Strict;
else
m_this_mode = ThisMode::Global;
// 15.1.3 Static Semantics: IsSimpleParameterList, https://tc39.es/ecma262/#sec-static-semantics-issimpleparameterlist
m_has_simple_parameter_list = all_of(m_formal_parameters, [&](auto& parameter) {
if (parameter.is_rest)
return false;
if (parameter.default_value)
return false;
if (!parameter.binding.template has<FlyString>())
return false;
return true;
});
}
void ECMAScriptFunctionObject::initialize(GlobalObject& global_object)
{
auto& vm = this->vm();
Base::initialize(global_object);
// Note: The ordering of these properties must be: length, name, prototype which is the order
// they are defined in the spec: https://tc39.es/ecma262/#sec-function-instances .
// This is observable through something like: https://tc39.es/ecma262/#sec-ordinaryownpropertykeys
// which must give the properties in chronological order which in this case is the order they
// are defined in the spec.
MUST(define_property_or_throw(vm.names.length, { .value = Value(m_function_length), .writable = false, .enumerable = false, .configurable = true }));
MUST(define_property_or_throw(vm.names.name, { .value = js_string(vm, m_name.is_null() ? "" : m_name), .writable = false, .enumerable = false, .configurable = true }));
if (!m_is_arrow_function) {
Object* prototype = nullptr;
switch (m_kind) {
case FunctionKind::Normal:
prototype = vm.heap().allocate<Object>(global_object, *global_object.new_ordinary_function_prototype_object_shape());
MUST(prototype->define_property_or_throw(vm.names.constructor, { .value = this, .writable = true, .enumerable = false, .configurable = true }));
break;
case FunctionKind::Generator:
// prototype is "g1.prototype" in figure-2 (https://tc39.es/ecma262/img/figure-2.png)
prototype = global_object.generator_prototype();
break;
case FunctionKind::Async:
break;
2021-11-15 01:53:24 +01:00
case FunctionKind::AsyncGenerator:
// FIXME: Add the AsyncGeneratorObject and set it as prototype.
break;
}
define_direct_property(vm.names.prototype, prototype, Attribute::Writable);
}
}
ECMAScriptFunctionObject::~ECMAScriptFunctionObject()
{
}
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
// 10.2.1 [[Call]] ( thisArgument, argumentsList ), https://tc39.es/ecma262/#sec-ecmascript-function-objects-call-thisargument-argumentslist
ThrowCompletionOr<Value> ECMAScriptFunctionObject::internal_call(Value this_argument, MarkedValueList arguments_list)
{
auto& vm = this->vm();
// 1. Let callerContext be the running execution context.
// NOTE: No-op, kept by the VM in its execution context stack.
ExecutionContext callee_context(heap());
// Non-standard
callee_context.arguments.extend(move(arguments_list));
if (auto* interpreter = vm.interpreter_if_exists())
callee_context.current_node = interpreter->current_node();
// 2. Let calleeContext be PrepareForOrdinaryCall(F, undefined).
// NOTE: We throw if the end of the native stack is reached, so unlike in the spec this _does_ need an exception check.
TRY(prepare_for_ordinary_call(callee_context, nullptr));
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
// 3. Assert: calleeContext is now the running execution context.
VERIFY(&vm.running_execution_context() == &callee_context);
// 4. If F.[[IsClassConstructor]] is true, then
if (m_is_class_constructor) {
// a. Let error be a newly created TypeError object.
// b. NOTE: error is created in calleeContext with F's associated Realm Record.
auto throw_completion = vm.throw_completion<TypeError>(global_object(), ErrorType::ClassConstructorWithoutNew, m_name);
// c. Remove calleeContext from the execution context stack and restore callerContext as the running execution context.
vm.pop_execution_context();
// d. Return ThrowCompletion(error).
return throw_completion;
}
// 5. Perform OrdinaryCallBindThis(F, calleeContext, thisArgument).
ordinary_call_bind_this(callee_context, this_argument);
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
// 6. Let result be OrdinaryCallEvaluateBody(F, argumentsList).
auto result = ordinary_call_evaluate_body();
// 7. Remove calleeContext from the execution context stack and restore callerContext as the running execution context.
vm.pop_execution_context();
// 8. If result.[[Type]] is return, return NormalCompletion(result.[[Value]]).
if (result.type() == Completion::Type::Return)
return result.value();
// 9. ReturnIfAbrupt(result).
if (result.is_abrupt()) {
VERIFY(result.is_error());
return result;
}
// 10. Return NormalCompletion(undefined).
return js_undefined();
}
// 10.2.2 [[Construct]] ( argumentsList, newTarget ), https://tc39.es/ecma262/#sec-ecmascript-function-objects-construct-argumentslist-newtarget
ThrowCompletionOr<Object*> ECMAScriptFunctionObject::internal_construct(MarkedValueList arguments_list, FunctionObject& new_target)
{
auto& vm = this->vm();
auto& global_object = this->global_object();
// 1. Let callerContext be the running execution context.
// NOTE: No-op, kept by the VM in its execution context stack.
// 2. Let kind be F.[[ConstructorKind]].
auto kind = m_constructor_kind;
Object* this_argument = nullptr;
// 3. If kind is base, then
if (kind == ConstructorKind::Base) {
// a. Let thisArgument be ? OrdinaryCreateFromConstructor(newTarget, "%Object.prototype%").
this_argument = TRY(ordinary_create_from_constructor<Object>(global_object, new_target, &GlobalObject::object_prototype));
}
ExecutionContext callee_context(heap());
// Non-standard
callee_context.arguments.extend(move(arguments_list));
if (auto* interpreter = vm.interpreter_if_exists())
callee_context.current_node = interpreter->current_node();
// 4. Let calleeContext be PrepareForOrdinaryCall(F, newTarget).
// NOTE: We throw if the end of the native stack is reached, so unlike in the spec this _does_ need an exception check.
TRY(prepare_for_ordinary_call(callee_context, &new_target));
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
// 5. Assert: calleeContext is now the running execution context.
VERIFY(&vm.running_execution_context() == &callee_context);
// 6. If kind is base, then
if (kind == ConstructorKind::Base) {
// a. Perform OrdinaryCallBindThis(F, calleeContext, thisArgument).
ordinary_call_bind_this(callee_context, this_argument);
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
// b. Let initializeResult be InitializeInstanceElements(thisArgument, F).
auto initialize_result = vm.initialize_instance_elements(*this_argument, *this);
// c. If initializeResult is an abrupt completion, then
if (initialize_result.is_throw_completion()) {
// i. Remove calleeContext from the execution context stack and restore callerContext as the running execution context.
vm.pop_execution_context();
// ii. Return Completion(initializeResult).
return initialize_result.throw_completion();
}
}
// 7. Let constructorEnv be the LexicalEnvironment of calleeContext.
auto* constructor_env = callee_context.lexical_environment;
// 8. Let result be OrdinaryCallEvaluateBody(F, argumentsList).
auto result = ordinary_call_evaluate_body();
// 9. Remove calleeContext from the execution context stack and restore callerContext as the running execution context.
vm.pop_execution_context();
// 10. If result.[[Type]] is return, then
if (result.type() == Completion::Type::Return) {
// FIXME: This is leftover from untangling the call/construct mess - doesn't belong here in any way, but removing it breaks derived classes.
// Likely fixed by making ClassDefinitionEvaluation fully spec compliant.
if (kind == ConstructorKind::Derived && result.value()->is_object()) {
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
auto prototype = TRY(new_target.get(vm.names.prototype));
if (prototype.is_object())
TRY(result.value()->as_object().internal_set_prototype_of(&prototype.as_object()));
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
}
// EOF (End of FIXME)
// a. If Type(result.[[Value]]) is Object, return NormalCompletion(result.[[Value]]).
if (result.value()->is_object())
return &result.value()->as_object();
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
// b. If kind is base, return NormalCompletion(thisArgument).
if (kind == ConstructorKind::Base)
return this_argument;
// c. If result.[[Value]] is not undefined, throw a TypeError exception.
if (!result.value()->is_undefined())
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
return vm.throw_completion<TypeError>(global_object, ErrorType::DerivedConstructorReturningInvalidValue);
}
// 11. Else, ReturnIfAbrupt(result).
else if (result.is_abrupt()) {
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
VERIFY(result.is_error());
return result;
}
// 12. Return ? constructorEnv.GetThisBinding().
auto this_binding = TRY(constructor_env->get_this_binding(global_object));
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
return &this_binding.as_object();
}
void ECMAScriptFunctionObject::visit_edges(Visitor& visitor)
{
Base::visit_edges(visitor);
visitor.visit(m_environment);
visitor.visit(m_private_environment);
visitor.visit(m_realm);
visitor.visit(m_home_object);
for (auto& field : m_fields) {
if (auto* property_name_ptr = field.name.get_pointer<PropertyKey>(); property_name_ptr && property_name_ptr->is_symbol())
visitor.visit(property_name_ptr->as_symbol());
visitor.visit(field.initializer);
}
}
// 10.2.7 MakeMethod ( F, homeObject ), https://tc39.es/ecma262/#sec-makemethod
void ECMAScriptFunctionObject::make_method(Object& home_object)
{
// 1. Set F.[[HomeObject]] to homeObject.
m_home_object = &home_object;
// 2. Return NormalCompletion(undefined).
}
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
// 10.2.11 FunctionDeclarationInstantiation ( func, argumentsList ), https://tc39.es/ecma262/#sec-functiondeclarationinstantiation
ThrowCompletionOr<void> ECMAScriptFunctionObject::function_declaration_instantiation(Interpreter* interpreter)
{
auto& vm = this->vm();
auto& callee_context = vm.running_execution_context();
// Needed to extract declarations and functions
ScopeNode const* scope_body = nullptr;
if (is<ScopeNode>(*m_ecmascript_code))
scope_body = static_cast<ScopeNode const*>(m_ecmascript_code.ptr());
bool has_parameter_expressions = false;
// FIXME: Maybe compute has duplicates at parse time? (We need to anyway since it's an error in some cases)
bool has_duplicates = false;
HashTable<FlyString> parameter_names;
for (auto& parameter : m_formal_parameters) {
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
if (parameter.default_value)
has_parameter_expressions = true;
parameter.binding.visit(
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
[&](FlyString const& name) {
if (parameter_names.set(name) != AK::HashSetResult::InsertedNewEntry)
has_duplicates = true;
},
[&](NonnullRefPtr<BindingPattern> const& pattern) {
if (pattern->contains_expression())
has_parameter_expressions = true;
pattern->for_each_bound_name([&](auto& name) {
if (parameter_names.set(name) != AK::HashSetResult::InsertedNewEntry)
has_duplicates = true;
});
});
}
auto arguments_object_needed = m_might_need_arguments_object;
if (this_mode() == ThisMode::Lexical)
arguments_object_needed = false;
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
if (parameter_names.contains(vm.names.arguments.as_string()))
arguments_object_needed = false;
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
HashTable<FlyString> function_names;
Vector<FunctionDeclaration const&> functions_to_initialize;
if (scope_body) {
scope_body->for_each_var_function_declaration_in_reverse_order([&](FunctionDeclaration const& function) {
if (function_names.set(function.name()) == AK::HashSetResult::InsertedNewEntry)
functions_to_initialize.append(function);
});
auto const& arguments_name = vm.names.arguments.as_string();
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
if (!has_parameter_expressions && function_names.contains(arguments_name))
arguments_object_needed = false;
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
if (!has_parameter_expressions && arguments_object_needed) {
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
scope_body->for_each_lexically_declared_name([&](auto const& name) {
if (name == arguments_name) {
arguments_object_needed = false;
return IterationDecision::Break;
}
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
return IterationDecision::Continue;
});
}
} else {
arguments_object_needed = false;
}
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
Environment* environment;
if (is_strict_mode() || !has_parameter_expressions) {
environment = callee_context.lexical_environment;
} else {
environment = new_declarative_environment(*callee_context.lexical_environment);
VERIFY(callee_context.variable_environment == callee_context.lexical_environment);
callee_context.lexical_environment = environment;
}
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
for (auto const& parameter_name : parameter_names) {
if (MUST(environment->has_binding(parameter_name)))
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
continue;
MUST(environment->create_mutable_binding(global_object(), parameter_name, false));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
if (has_duplicates)
MUST(environment->initialize_binding(global_object(), parameter_name, js_undefined()));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
}
if (arguments_object_needed) {
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
Object* arguments_object;
if (is_strict_mode() || !has_simple_parameter_list())
arguments_object = create_unmapped_arguments_object(global_object(), vm.running_execution_context().arguments);
else
arguments_object = create_mapped_arguments_object(global_object(), *this, formal_parameters(), vm.running_execution_context().arguments, *environment);
if (is_strict_mode())
MUST(environment->create_immutable_binding(global_object(), vm.names.arguments.as_string(), false));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
else
MUST(environment->create_mutable_binding(global_object(), vm.names.arguments.as_string(), false));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
MUST(environment->initialize_binding(global_object(), vm.names.arguments.as_string(), arguments_object));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
parameter_names.set(vm.names.arguments.as_string());
}
// We now treat parameterBindings as parameterNames.
// The spec makes an iterator here to do IteratorBindingInitialization but we just do it manually
auto& execution_context_arguments = vm.running_execution_context().arguments;
for (size_t i = 0; i < m_formal_parameters.size(); ++i) {
auto& parameter = m_formal_parameters[i];
TRY(parameter.binding.visit(
[&](auto const& param) -> ThrowCompletionOr<void> {
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
Value argument_value;
if (parameter.is_rest) {
auto* array = MUST(Array::create(global_object(), 0));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
for (size_t rest_index = i; rest_index < execution_context_arguments.size(); ++rest_index)
array->indexed_properties().append(execution_context_arguments[rest_index]);
argument_value = array;
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
} else if (i < execution_context_arguments.size() && !execution_context_arguments[i].is_undefined()) {
argument_value = execution_context_arguments[i];
} else if (parameter.default_value) {
// FIXME: Support default arguments in the bytecode world!
if (interpreter)
argument_value = TRY(parameter.default_value->execute(*interpreter, global_object())).release_value();
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
} else {
argument_value = js_undefined();
}
Environment* used_environment = has_duplicates ? nullptr : environment;
if constexpr (IsSame<FlyString const&, decltype(param)>) {
Reference reference = TRY(vm.resolve_binding(param, used_environment));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
// Here the difference from hasDuplicates is important
if (has_duplicates)
return reference.put_value(global_object(), argument_value);
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
else
return reference.initialize_referenced_binding(global_object(), argument_value);
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
} else if (IsSame<NonnullRefPtr<BindingPattern> const&, decltype(param)>) {
// Here the difference from hasDuplicates is important
return vm.binding_initialization(param, argument_value, used_environment, global_object());
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
}
}));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
}
Environment* var_environment;
HashTable<FlyString> instantiated_var_names;
if (scope_body)
instantiated_var_names.ensure_capacity(scope_body->var_declaration_count());
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
if (!has_parameter_expressions) {
if (scope_body) {
scope_body->for_each_var_declared_name([&](auto const& name) {
if (!parameter_names.contains(name) && instantiated_var_names.set(name) == AK::HashSetResult::InsertedNewEntry) {
MUST(environment->create_mutable_binding(global_object(), name, false));
MUST(environment->initialize_binding(global_object(), name, js_undefined()));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
}
});
}
var_environment = environment;
} else {
var_environment = new_declarative_environment(*environment);
callee_context.variable_environment = var_environment;
if (scope_body) {
scope_body->for_each_var_declared_name([&](auto const& name) {
if (instantiated_var_names.set(name) != AK::HashSetResult::InsertedNewEntry)
return IterationDecision::Continue;
MUST(var_environment->create_mutable_binding(global_object(), name, false));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
Value initial_value;
if (!parameter_names.contains(name) || function_names.contains(name))
initial_value = js_undefined();
else
initial_value = MUST(environment->get_binding_value(global_object(), name, false));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
MUST(var_environment->initialize_binding(global_object(), name, initial_value));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
return IterationDecision::Continue;
});
}
}
// B.3.2.1 Changes to FunctionDeclarationInstantiation, https://tc39.es/ecma262/#sec-web-compat-functiondeclarationinstantiation
if (!m_strict && scope_body) {
scope_body->for_each_function_hoistable_with_annexB_extension([&](FunctionDeclaration& function_declaration) {
auto& function_name = function_declaration.name();
if (parameter_names.contains(function_name))
return IterationDecision::Continue;
// The spec says 'initializedBindings' here but that does not exist and it then adds it to 'instantiatedVarNames' so it probably means 'instantiatedVarNames'.
if (!instantiated_var_names.contains(function_name) && function_name != vm.names.arguments.as_string()) {
MUST(var_environment->create_mutable_binding(global_object(), function_name, false));
MUST(var_environment->initialize_binding(global_object(), function_name, js_undefined()));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
instantiated_var_names.set(function_name);
}
function_declaration.set_should_do_additional_annexB_steps();
return IterationDecision::Continue;
});
}
Environment* lex_environment;
// 30. If strict is false, then
if (!is_strict_mode()) {
// Optimization: We avoid creating empty top-level declarative environments in non-strict mode, if both of these conditions are true:
// 1. there is no direct call to eval() within this function
// 2. there are no lexical declarations that would go into the environment
bool can_elide_declarative_environment = !m_contains_direct_call_to_eval && (!scope_body || !scope_body->has_lexical_declarations());
if (can_elide_declarative_environment) {
lex_environment = var_environment;
} else {
// a. Let lexEnv be NewDeclarativeEnvironment(varEnv).
// b. NOTE: Non-strict functions use a separate Environment Record for top-level lexical declarations so that a direct eval
// can determine whether any var scoped declarations introduced by the eval code conflict with pre-existing top-level
// lexically scoped declarations. This is not needed for strict functions because a strict direct eval always places
// all declarations into a new Environment Record.
lex_environment = new_declarative_environment(*var_environment);
}
} else {
// 31. Else, let lexEnv be varEnv.
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
lex_environment = var_environment;
}
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
// 32. Set the LexicalEnvironment of calleeContext to lexEnv.
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
callee_context.lexical_environment = lex_environment;
if (!scope_body)
return {};
scope_body->for_each_lexically_scoped_declaration([&](Declaration const& declaration) {
declaration.for_each_bound_name([&](auto const& name) {
if (declaration.is_constant_declaration())
MUST(lex_environment->create_immutable_binding(global_object(), name, true));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
else
MUST(lex_environment->create_mutable_binding(global_object(), name, false));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
return IterationDecision::Continue;
});
});
VERIFY(!vm.exception());
2021-10-11 20:29:31 +02:00
auto* private_environment = callee_context.private_environment;
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
for (auto& declaration : functions_to_initialize) {
2021-10-11 20:29:31 +02:00
auto* function = ECMAScriptFunctionObject::create(global_object(), declaration.name(), declaration.body(), declaration.parameters(), declaration.function_length(), lex_environment, private_environment, declaration.kind(), declaration.is_strict_mode(), declaration.might_need_arguments_object(), declaration.contains_direct_call_to_eval());
MUST(var_environment->set_mutable_binding(global_object(), declaration.name(), function, false));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
}
return {};
}
// 10.2.1.1 PrepareForOrdinaryCall ( F, newTarget ), https://tc39.es/ecma262/#sec-prepareforordinarycall
ThrowCompletionOr<void> ECMAScriptFunctionObject::prepare_for_ordinary_call(ExecutionContext& callee_context, Object* new_target)
{
auto& vm = this->vm();
// Non-standard
callee_context.is_strict_mode = m_strict;
// 1. Let callerContext be the running execution context.
// 2. Let calleeContext be a new ECMAScript code execution context.
// NOTE: In the specification, PrepareForOrdinaryCall "returns" a new callee execution context.
// To avoid heap allocations, we put our ExecutionContext objects on the C++ stack instead.
// Whoever calls us should put an ExecutionContext on their stack and pass that as the `callee_context`.
// 3. Set the Function of calleeContext to F.
callee_context.function = this;
callee_context.function_name = m_name;
// 4. Let calleeRealm be F.[[Realm]].
auto* callee_realm = m_realm;
// NOTE: This non-standard fallback is needed until we can guarantee that literally
// every function has a realm - especially in LibWeb that's sometimes not the case
// when a function is created while no JS is running, as we currently need to rely on
// that (:acid2:, I know - see set_event_handler_attribute() for an example).
// If there's no 'current realm' either, we can't continue and crash.
if (!callee_realm)
callee_realm = vm.current_realm();
VERIFY(callee_realm);
// 5. Set the Realm of calleeContext to calleeRealm.
callee_context.realm = callee_realm;
// 6. Set the ScriptOrModule of calleeContext to F.[[ScriptOrModule]].
// FIXME: Our execution context struct currently does not track this item.
// 7. Let localEnv be NewFunctionEnvironment(F, newTarget).
auto* local_environment = new_function_environment(*this, new_target);
// 8. Set the LexicalEnvironment of calleeContext to localEnv.
callee_context.lexical_environment = local_environment;
// 9. Set the VariableEnvironment of calleeContext to localEnv.
callee_context.variable_environment = local_environment;
// 10. Set the PrivateEnvironment of calleeContext to F.[[PrivateEnvironment]].
2021-10-11 20:29:31 +02:00
callee_context.private_environment = m_private_environment;
// 11. If callerContext is not already suspended, suspend callerContext.
// FIXME: We don't have this concept yet.
// 12. Push calleeContext onto the execution context stack; calleeContext is now the running execution context.
TRY(vm.push_execution_context(callee_context, global_object()));
// 13. NOTE: Any exception objects produced after this point are associated with calleeRealm.
// 14. Return calleeContext. (See NOTE above about how contexts are allocated on the C++ stack.)
return {};
}
// 10.2.1.2 OrdinaryCallBindThis ( F, calleeContext, thisArgument ), https://tc39.es/ecma262/#sec-ordinarycallbindthis
void ECMAScriptFunctionObject::ordinary_call_bind_this(ExecutionContext& callee_context, Value this_argument)
{
auto& vm = this->vm();
// 1. Let thisMode be F.[[ThisMode]].
auto this_mode = m_this_mode;
// If thisMode is lexical, return NormalCompletion(undefined).
if (this_mode == ThisMode::Lexical)
return;
// 3. Let calleeRealm be F.[[Realm]].
auto* callee_realm = m_realm;
// NOTE: This non-standard fallback is needed until we can guarantee that literally
// every function has a realm - especially in LibWeb that's sometimes not the case
// when a function is created while no JS is running, as we currently need to rely on
// that (:acid2:, I know - see set_event_handler_attribute() for an example).
// If there's no 'current realm' either, we can't continue and crash.
if (!callee_realm)
callee_realm = vm.current_realm();
VERIFY(callee_realm);
// 4. Let localEnv be the LexicalEnvironment of calleeContext.
auto* local_env = callee_context.lexical_environment;
Value this_value;
// 5. If thisMode is strict, let thisValue be thisArgument.
if (this_mode == ThisMode::Strict) {
this_value = this_argument;
}
// 6. Else,
else {
// a. If thisArgument is undefined or null, then
if (this_argument.is_nullish()) {
// i. Let globalEnv be calleeRealm.[[GlobalEnv]].
// ii. Assert: globalEnv is a global Environment Record.
auto& global_env = callee_realm->global_environment();
// iii. Let thisValue be globalEnv.[[GlobalThisValue]].
this_value = &global_env.global_this_value();
}
// b. Else,
else {
// i. Let thisValue be ! ToObject(thisArgument).
this_value = MUST(this_argument.to_object(global_object()));
// ii. NOTE: ToObject produces wrapper objects using calleeRealm.
// FIXME: It currently doesn't, as we pass the function's global object.
}
}
// 7. Assert: localEnv is a function Environment Record.
// 8. Assert: The next step never returns an abrupt completion because localEnv.[[ThisBindingStatus]] is not initialized.
// 9. Return localEnv.BindThisValue(thisValue).
MUST(verify_cast<FunctionEnvironment>(local_env)->bind_this_value(global_object(), this_value));
}
// 27.7.5.1 AsyncFunctionStart ( promiseCapability, asyncFunctionBody ), https://tc39.es/ecma262/#sec-async-functions-abstract-operations-async-function-start
void ECMAScriptFunctionObject::async_function_start(PromiseCapability const& promise_capability)
{
auto& vm = this->vm();
// 1. Let runningContext be the running execution context.
auto& running_context = vm.running_execution_context();
// 2. Let asyncContext be a copy of runningContext.
auto async_context = running_context.copy();
// 3. NOTE: Copying the execution state is required for AsyncBlockStart to resume its execution. It is ill-defined to resume a currently executing context.
// 4. Perform ! AsyncBlockStart(promiseCapability, asyncFunctionBody, asyncContext).
async_block_start(promise_capability, async_context);
}
// 27.7.5.2 AsyncBlockStart ( promiseCapability, asyncBody, asyncContext ), https://tc39.es/ecma262/#sec-asyncblockstart
void ECMAScriptFunctionObject::async_block_start(PromiseCapability const& promise_capability, ExecutionContext& async_context)
{
auto& vm = this->vm();
// 1. Assert: promiseCapability is a PromiseCapability Record.
// 2. Let runningContext be the running execution context.
auto& running_context = vm.running_execution_context();
// 3. Set the code evaluation state of asyncContext such that when evaluation is resumed for that execution context the following steps will be performed:
auto* execution_steps = NativeFunction::create(global_object(), "", [async_body = m_ecmascript_code, &promise_capability](auto& vm, auto& global_object) -> ThrowCompletionOr<Value> {
// a. Let result be the result of evaluating asyncBody.
auto result = async_body->execute(vm.interpreter(), global_object);
// b. Assert: If we return here, the async function either threw an exception or performed an implicit or explicit return; all awaiting is done.
// c. Remove asyncContext from the execution context stack and restore the execution context that is at the top of the execution context stack as the running execution context.
vm.pop_execution_context();
// d. If result.[[Type]] is normal, then
if (result.type() == Completion::Type::Normal) {
// i. Perform ! Call(promiseCapability.[[Resolve]], undefined, « undefined »).
MUST(call(global_object, promise_capability.resolve, js_undefined(), js_undefined()));
}
// e. Else if result.[[Type]] is return, then
else if (result.type() == Completion::Type::Return) {
// i. Perform ! Call(promiseCapability.[[Resolve]], undefined, « result.[[Value]] »).
MUST(call(global_object, promise_capability.resolve, js_undefined(), *result.value()));
}
// f. Else,
else {
// i. Assert: result.[[Type]] is throw.
VERIFY(result.type() == Completion::Type::Throw);
// ii. Perform ! Call(promiseCapability.[[Reject]], undefined, « result.[[Value]] »).
vm.clear_exception();
MUST(call(global_object, promise_capability.reject, js_undefined(), *result.value()));
}
// g. Return.
return js_undefined();
});
// 4. Push asyncContext onto the execution context stack; asyncContext is now the running execution context.
auto push_result = vm.push_execution_context(async_context, global_object());
if (push_result.is_error())
return;
// 5. Resume the suspended evaluation of asyncContext. Let result be the value returned by the resumed computation.
auto result = vm.call(*execution_steps, async_context.this_value.is_empty() ? js_undefined() : async_context.this_value);
// 6. Assert: When we return here, asyncContext has already been removed from the execution context stack and runningContext is the currently running execution context.
VERIFY(&vm.running_execution_context() == &running_context);
// 7. Assert: result is a normal completion with a value of undefined. The possible sources of completion values are Await or, if the async function doesn't await anything, step 3.g above.
VERIFY(result.has_value() && result.value().is_undefined());
// 8. Return.
}
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
// 10.2.1.4 OrdinaryCallEvaluateBody ( F, argumentsList ), https://tc39.es/ecma262/#sec-ordinarycallevaluatebody
// 15.8.4 Runtime Semantics: EvaluateAsyncFunctionBody, https://tc39.es/ecma262/#sec-runtime-semantics-evaluatefunctionbody
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
Completion ECMAScriptFunctionObject::ordinary_call_evaluate_body()
{
auto& vm = this->vm();
auto* bytecode_interpreter = Bytecode::Interpreter::current();
2021-11-15 01:53:24 +01:00
if (m_kind == FunctionKind::AsyncGenerator)
return vm.throw_completion<InternalError>(global_object(), ErrorType::NotImplemented, "Async Generator function execution");
if (bytecode_interpreter) {
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
// FIXME: pass something to evaluate default arguments with
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
TRY(function_declaration_instantiation(nullptr));
if (!m_bytecode_executable.has_value()) {
m_bytecode_executable = Bytecode::Generator::generate(m_ecmascript_code, m_kind);
m_bytecode_executable->name = m_name;
auto& passes = JS::Bytecode::Interpreter::optimization_pipeline();
passes.perform(*m_bytecode_executable);
if constexpr (JS_BYTECODE_DEBUG) {
dbgln("Optimisation passes took {}us", passes.elapsed());
dbgln("Compiled Bytecode::Block for function '{}':", m_name);
}
if (JS::Bytecode::g_dump_bytecode)
m_bytecode_executable->dump();
}
auto result_and_frame = bytecode_interpreter->run_and_return_frame(*m_bytecode_executable, nullptr);
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
if (auto* exception = vm.exception())
return throw_completion(exception->value());
VERIFY(result_and_frame.frame != nullptr);
if (result_and_frame.value.is_error()) {
vm.throw_exception(bytecode_interpreter->global_object(), *result_and_frame.value.release_error().value());
return throw_completion(vm.exception()->value());
}
auto result = result_and_frame.value.release_value();
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
// NOTE: Running the bytecode should eventually return a completion.
// Until it does, we assume "return" and include the undefined fallback from the call site.
if (m_kind == FunctionKind::Normal)
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
return { Completion::Type::Return, result.value_or(js_undefined()), {} };
auto generator_object = TRY(GeneratorObject::create(global_object(), result, this, vm.running_execution_context().copy(), move(*result_and_frame.frame)));
// NOTE: Async functions are entirely transformed to generator functions, and wrapped in a custom driver that returns a promise
// See AwaitExpression::generate_bytecode() for the transformation.
if (m_kind == FunctionKind::Async)
return { Completion::Type::Return, TRY(AsyncFunctionDriverWrapper::create(global_object(), generator_object)), {} };
VERIFY(m_kind == FunctionKind::Generator);
return { Completion::Type::Return, generator_object, {} };
} else {
if (m_kind == FunctionKind::Generator)
return vm.throw_completion<InternalError>(global_object(), ErrorType::NotImplemented, "Generator function execution in AST interpreter");
OwnPtr<Interpreter> local_interpreter;
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
Interpreter* ast_interpreter = vm.interpreter_if_exists();
if (!ast_interpreter) {
local_interpreter = Interpreter::create_with_existing_realm(*realm());
ast_interpreter = local_interpreter.ptr();
}
VM::InterpreterExecutionScope scope(*ast_interpreter);
// FunctionBody : FunctionStatementList
if (m_kind == FunctionKind::Normal) {
// 1. Perform ? FunctionDeclarationInstantiation(functionObject, argumentsList).
TRY(function_declaration_instantiation(ast_interpreter));
// 2. Return the result of evaluating FunctionStatementList.
return m_ecmascript_code->execute(*ast_interpreter, global_object());
}
// AsyncFunctionBody : FunctionBody
else if (m_kind == FunctionKind::Async) {
// 1. Let promiseCapability be ! NewPromiseCapability(%Promise%).
auto promise_capability = MUST(new_promise_capability(global_object(), global_object().promise_constructor()));
// 2. Let declResult be FunctionDeclarationInstantiation(functionObject, argumentsList).
auto declaration_result = function_declaration_instantiation(ast_interpreter);
// 3. If declResult is not an abrupt completion, then
if (!declaration_result.is_throw_completion()) {
// a. Perform ! AsyncFunctionStart(promiseCapability, FunctionBody).
async_function_start(promise_capability);
}
// 4. Else,
else {
// a. Perform ! Call(promiseCapability.[[Reject]], undefined, « declResult.[[Value]] »).
MUST(call(global_object(), promise_capability.reject, js_undefined(), *declaration_result.throw_completion().value()));
}
// 5. Return Completion { [[Type]]: return, [[Value]]: promiseCapability.[[Promise]], [[Target]]: empty }.
return Completion { Completion::Type::Return, promise_capability.promise, {} };
}
}
LibJS: Implement [[Call]] and [[Construct]] internal slots properly This patch implements: - Spec compliant [[Call]] and [[Construct]] internal slots, as virtual FunctionObject::internal_{call,construct}(). These effectively replace the old virtual FunctionObject::{call,construct}(), but with several advantages: - Clear and consistent naming, following the object internal methods - Use of completions - internal_construct() returns an Object, and not Value! This has been a source of confusion for a long time, since in the spec there's always an Object returned but the Value return type in LibJS meant that this could not be fully trusted and something could screw you over. - Arguments are passed explicitly in form of a MarkedValueList, allowing manipulation (BoundFunction). We still put them on the execution context as a lot of code depends on it (VM::arguments()), but not from the Call() / Construct() AOs anymore, which now allows for bypassing them and invoking [[Call]] / [[Construct]] directly. Nothing but Call() / Construct() themselves do that at the moment, but future additions to ECMA262 or already existing web specs might. - Spec compliant, standalone Call() and Construct() AOs: currently the closest we have is VM::{call,construct}(), but those try to cater to all the different function object subclasses at once, resulting in a horrible mess and calling AOs with functions they should never be called with; most prominently PrepareForOrdinaryCall and OrdinaryCallBindThis, which are only for ECMAScriptFunctionObject. As a result this also contains an implicit optimization: we no longer need to create a new function environment for NativeFunctions - which, worth mentioning, is what started this whole crusade in the first place :^)
2021-10-08 20:37:21 +01:00
VERIFY_NOT_REACHED();
}
void ECMAScriptFunctionObject::set_name(const FlyString& name)
{
VERIFY(!name.is_null());
auto& vm = this->vm();
m_name = name;
auto success = MUST(define_property_or_throw(vm.names.name, { .value = js_string(vm, m_name), .writable = false, .enumerable = false, .configurable = true }));
VERIFY(success);
}
void ECMAScriptFunctionObject::add_field(ClassElement::ClassElementName property_key, ECMAScriptFunctionObject* initializer)
{
m_fields.empend(property_key, initializer);
}
}