LibJS: Implement the RegExpStringIterator object

This implementation closely follows the StringIterator object in that
the abstract closure meant to be created in CreateRegExpStringIterator
is instead unrolled into RegExpStringIterator.prototype.next.

5 files changed, 181 insertions, 0 deletions

diff --git a/Userland/Libraries/LibJS/Runtime/GlobalObject.cpp b/Userland/Libraries/LibJS/Runtime/GlobalObject.cpp
index 57011398d6..3a1fad31d9 100644
--- a/Userland/Libraries/LibJS/Runtime/GlobalObject.cpp
+++ b/Userland/Libraries/LibJS/Runtime/GlobalObject.cpp
@@ -59,6 +59,7 @@
 #include <LibJS/Runtime/ReflectObject.h>
 #include <LibJS/Runtime/RegExpConstructor.h>
 #include <LibJS/Runtime/RegExpPrototype.h>
+#include <LibJS/Runtime/RegExpStringIteratorPrototype.h>
 #include <LibJS/Runtime/SetConstructor.h>
 #include <LibJS/Runtime/SetIteratorPrototype.h>
 #include <LibJS/Runtime/SetPrototype.h>
diff --git a/Userland/Libraries/LibJS/Runtime/RegExpStringIterator.cpp b/Userland/Libraries/LibJS/Runtime/RegExpStringIterator.cpp
new file mode 100644
index 0000000000..573e7bb32e
--- /dev/null
+++ b/Userland/Libraries/LibJS/Runtime/RegExpStringIterator.cpp
@@ -0,0 +1,27 @@
+/*
+ * Copyright (c) 2021, Tim Flynn <trflynn89@pm.me>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#include <LibJS/Runtime/GlobalObject.h>
+#include <LibJS/Runtime/RegExpStringIterator.h>
+
+namespace JS {
+
+// 22.2.7.1 CreateRegExpStringIterator ( R, S, global, fullUnicode ), https://tc39.es/ecma262/#sec-createregexpstringiterator
+RegExpStringIterator* RegExpStringIterator::create(GlobalObject& global_object, Object& regexp_object, String string, bool global, bool unicode)
+{
+    return global_object.heap().allocate<RegExpStringIterator>(global_object, *global_object.regexp_string_iterator_prototype(), regexp_object, move(string), global, unicode);
+}
+
+RegExpStringIterator::RegExpStringIterator(Object& prototype, Object& regexp_object, String string, bool global, bool unicode)
+    : Object(prototype)
+    , m_regexp_object(regexp_object)
+    , m_string(move(string))
+    , m_global(global)
+    , m_unicode(unicode)
+{
+}
+
+}
diff --git a/Userland/Libraries/LibJS/Runtime/RegExpStringIterator.h b/Userland/Libraries/LibJS/Runtime/RegExpStringIterator.h
new file mode 100644
index 0000000000..876dd796a9
--- /dev/null
+++ b/Userland/Libraries/LibJS/Runtime/RegExpStringIterator.h
@@ -0,0 +1,38 @@
+/*
+ * Copyright (c) 2021, Tim Flynn <trflynn89@pm.me>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+#include <LibJS/Runtime/Object.h>
+
+namespace JS {
+
+class RegExpStringIterator final : public Object {
+    JS_OBJECT(RegExpStringIterator, Object);
+
+public:
+    static RegExpStringIterator* create(GlobalObject&, Object& regexp_object, String string, bool global, bool unicode);
+
+    explicit RegExpStringIterator(Object& prototype, Object& regexp_object, String string, bool global, bool unicode);
+    virtual ~RegExpStringIterator() override = default;
+
+    Object& regexp_object() { return m_regexp_object; }
+    String const& string() const { return m_string; }
+    bool global() const { return m_global; }
+    bool unicode() const { return m_unicode; }
+
+    bool done() const { return m_done; }
+    void set_done() { m_done = true; }
+
+private:
+    Object& m_regexp_object;
+    String m_string;
+    bool m_global { false };
+    bool m_unicode { false };
+    bool m_done { false };
+};
+
+}
diff --git a/Userland/Libraries/LibJS/Runtime/RegExpStringIteratorPrototype.cpp b/Userland/Libraries/LibJS/Runtime/RegExpStringIteratorPrototype.cpp
new file mode 100644
index 0000000000..9b61656d20
--- /dev/null
+++ b/Userland/Libraries/LibJS/Runtime/RegExpStringIteratorPrototype.cpp
@@ -0,0 +1,89 @@
+/*
+ * Copyright (c) 2021, Tim Flynn <trflynn89@pm.me>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#include <LibJS/Runtime/GlobalObject.h>
+#include <LibJS/Runtime/IteratorOperations.h>
+#include <LibJS/Runtime/RegExpPrototype.h>
+#include <LibJS/Runtime/RegExpStringIterator.h>
+#include <LibJS/Runtime/RegExpStringIteratorPrototype.h>
+
+namespace JS {
+
+RegExpStringIteratorPrototype::RegExpStringIteratorPrototype(GlobalObject& global_object)
+    : Object(*global_object.iterator_prototype())
+{
+}
+
+void RegExpStringIteratorPrototype::initialize(GlobalObject& global_object)
+{
+    Object::initialize(global_object);
+    auto& vm = this->vm();
+
+    u8 attr = Attribute::Writable | Attribute::Configurable;
+    define_native_function(vm.names.next, next, 0, attr);
+
+    // 22.2.7.2.2 %RegExpStringIteratorPrototype% [ @@toStringTag ], https://tc39.es/ecma262/#sec-%regexpstringiteratorprototype%-@@tostringtag
+    define_direct_property(*vm.well_known_symbol_to_string_tag(), js_string(global_object.heap(), "RegExp String Iterator"), Attribute::Configurable);
+}
+
+// 22.2.7.2.1 %RegExpStringIteratorPrototype%.next ( ), https://tc39.es/ecma262/#sec-%regexpstringiteratorprototype%.next
+JS_DEFINE_NATIVE_FUNCTION(RegExpStringIteratorPrototype::next)
+{
+    // For details, see the 'closure' of: https://tc39.es/ecma262/#sec-createregexpstringiterator
+    auto this_value = vm.this_value(global_object);
+    if (!this_value.is_object() || !is<RegExpStringIterator>(this_value.as_object())) {
+        vm.throw_exception<TypeError>(global_object, ErrorType::NotA, "RegExp String Iterator");
+        return {};
+    }
+
+    auto& iterator = static_cast<RegExpStringIterator&>(this_value.as_object());
+    if (iterator.done())
+        return create_iterator_result_object(global_object, js_undefined(), true);
+
+    auto match = regexp_exec(global_object, iterator.regexp_object(), iterator.string());
+    if (vm.exception())
+        return {};
+
+    if (match.is_null()) {
+        iterator.set_done();
+        return create_iterator_result_object(global_object, js_undefined(), true);
+    }
+
+    if (!iterator.global()) {
+        iterator.set_done();
+        return create_iterator_result_object(global_object, match, false);
+    }
+
+    auto* match_object = match.to_object(global_object);
+    if (!match_object)
+        return {};
+    auto match_string_value = match_object->get(0);
+    if (vm.exception())
+        return {};
+    auto match_string = match_string_value.to_string(global_object);
+    if (vm.exception())
+        return {};
+
+    if (match_string.is_empty()) {
+        auto last_index_value = iterator.regexp_object().get(vm.names.lastIndex);
+        if (vm.exception())
+            return {};
+        auto last_index = last_index_value.to_length(global_object);
+        if (vm.exception())
+            return {};
+
+        // FIXME: Implement AdvanceStringIndex to take Unicode code points into account - https://tc39.es/ecma262/#sec-advancestringindex
+        ++last_index;
+
+        iterator.regexp_object().set(vm.names.lastIndex, Value(last_index), true);
+        if (vm.exception())
+            return {};
+    }
+
+    return create_iterator_result_object(global_object, match, false);
+}
+
+}
diff --git a/Userland/Libraries/LibJS/Runtime/RegExpStringIteratorPrototype.h b/Userland/Libraries/LibJS/Runtime/RegExpStringIteratorPrototype.h
new file mode 100644
index 0000000000..075cffec74
--- /dev/null
+++ b/Userland/Libraries/LibJS/Runtime/RegExpStringIteratorPrototype.h
@@ -0,0 +1,26 @@
+/*
+ * Copyright (c) 2021, Tim Flynn <trflynn89@pm.me>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+#include <LibJS/Runtime/Object.h>
+
+namespace JS {
+
+class RegExpStringIteratorPrototype final : public Object {
+    JS_OBJECT(RegExpStringIteratorPrototype, Object)
+
+public:
+    explicit RegExpStringIteratorPrototype(GlobalObject&);
+    virtual ~RegExpStringIteratorPrototype() override = default;
+
+    virtual void initialize(GlobalObject&) override;
+
+private:
+    JS_DECLARE_NATIVE_FUNCTION(next);
+};
+
+}