LibJS: Add initial support for creating PrimitiveStrings with AK::String

This will temporarily bloat the size of PrimitiveString as LibJS is
transitioned to use String throughout, but will make doing so piecemeal
much easier.

4 files changed, 74 insertions, 4 deletions

diff --git a/Userland/Libraries/LibJS/Heap/Heap.cpp b/Userland/Libraries/LibJS/Heap/Heap.cpp
index 254cecf19a..d58758bfdd 100644
--- a/Userland/Libraries/LibJS/Heap/Heap.cpp
+++ b/Userland/Libraries/LibJS/Heap/Heap.cpp
@@ -57,6 +57,7 @@ Heap::Heap(VM& vm)
 
 Heap::~Heap()
 {
+    vm().string_cache().clear();
     vm().deprecated_string_cache().clear();
     collect_garbage(CollectionType::CollectEverything);
 }
diff --git a/Userland/Libraries/LibJS/Runtime/PrimitiveString.cpp b/Userland/Libraries/LibJS/Runtime/PrimitiveString.cpp
index 9b57954020..abfaaceb8e 100644
--- a/Userland/Libraries/LibJS/Runtime/PrimitiveString.cpp
+++ b/Userland/Libraries/LibJS/Runtime/PrimitiveString.cpp
@@ -25,6 +25,11 @@ PrimitiveString::PrimitiveString(PrimitiveString& lhs, PrimitiveString& rhs)
 {
 }
 
+PrimitiveString::PrimitiveString(String string)
+    : m_utf8_string(move(string))
+{
+}
+
 PrimitiveString::PrimitiveString(DeprecatedString string)
     : m_deprecated_string(move(string))
 {
@@ -37,6 +42,8 @@ PrimitiveString::PrimitiveString(Utf16String string)
 
 PrimitiveString::~PrimitiveString()
 {
+    if (has_utf8_string())
+        vm().string_cache().remove(*m_utf8_string);
     if (has_deprecated_string())
         vm().deprecated_string_cache().remove(*m_deprecated_string);
 }
@@ -59,18 +66,41 @@ bool PrimitiveString::is_empty() const
 
     if (has_utf16_string())
         return m_utf16_string->is_empty();
+    if (has_utf8_string())
+        return m_utf8_string->is_empty();
     if (has_deprecated_string())
         return m_deprecated_string->is_empty();
     VERIFY_NOT_REACHED();
 }
 
+ThrowCompletionOr<String> PrimitiveString::utf8_string() const
+{
+    auto& vm = this->vm();
+    TRY(resolve_rope_if_needed());
+
+    if (!has_utf8_string()) {
+        if (has_deprecated_string())
+            m_utf8_string = TRY_OR_THROW_OOM(vm, String::from_utf8(*m_deprecated_string));
+        else if (has_utf16_string())
+            m_utf8_string = TRY(m_utf16_string->to_utf8(vm));
+        else
+            VERIFY_NOT_REACHED();
+    }
+
+    return *m_utf8_string;
+}
+
 ThrowCompletionOr<DeprecatedString> PrimitiveString::deprecated_string() const
 {
     TRY(resolve_rope_if_needed());
 
     if (!has_deprecated_string()) {
-        VERIFY(has_utf16_string());
-        m_deprecated_string = TRY(m_utf16_string->to_deprecated_string(vm()));
+        if (has_utf8_string())
+            m_deprecated_string = m_utf8_string->to_deprecated_string();
+        else if (has_utf16_string())
+            m_deprecated_string = TRY(m_utf16_string->to_deprecated_string(vm()));
+        else
+            VERIFY_NOT_REACHED();
     }
 
     return *m_deprecated_string;
@@ -81,8 +111,12 @@ ThrowCompletionOr<Utf16String> PrimitiveString::utf16_string() const
     TRY(resolve_rope_if_needed());
 
     if (!has_utf16_string()) {
-        VERIFY(has_deprecated_string());
-        m_utf16_string = TRY(Utf16String::create(vm(), *m_deprecated_string));
+        if (has_utf8_string()) {
+            m_utf16_string = TRY(Utf16String::create(vm(), m_utf8_string->bytes_as_string_view()));
+        } else {
+            VERIFY(has_deprecated_string());
+            m_utf16_string = TRY(Utf16String::create(vm(), *m_deprecated_string));
+        }
     }
 
     return *m_utf16_string;
@@ -128,6 +162,26 @@ NonnullGCPtr<PrimitiveString> PrimitiveString::create(VM& vm, Utf16String string
     return vm.heap().allocate_without_realm<PrimitiveString>(move(string));
 }
 
+NonnullGCPtr<PrimitiveString> PrimitiveString::create(VM& vm, String string)
+{
+    if (string.is_empty())
+        return vm.empty_string();
+
+    if (auto bytes = string.bytes_as_string_view(); bytes.length() == 1) {
+        auto ch = static_cast<u8>(bytes[0]);
+        if (is_ascii(ch))
+            return vm.single_ascii_character_string(ch);
+    }
+
+    auto& string_cache = vm.string_cache();
+    if (auto it = string_cache.find(string); it != string_cache.end())
+        return *it->value;
+
+    auto new_string = vm.heap().allocate_without_realm<PrimitiveString>(string);
+    string_cache.set(move(string), new_string);
+    return *new_string;
+}
+
 NonnullGCPtr<PrimitiveString> PrimitiveString::create(VM& vm, DeprecatedString string)
 {
     if (string.is_empty())
diff --git a/Userland/Libraries/LibJS/Runtime/PrimitiveString.h b/Userland/Libraries/LibJS/Runtime/PrimitiveString.h
index 9684fb64b1..96e5accb99 100644
--- a/Userland/Libraries/LibJS/Runtime/PrimitiveString.h
+++ b/Userland/Libraries/LibJS/Runtime/PrimitiveString.h
@@ -9,6 +9,7 @@
 
 #include <AK/DeprecatedString.h>
 #include <AK/Optional.h>
+#include <AK/String.h>
 #include <AK/StringView.h>
 #include <LibJS/Forward.h>
 #include <LibJS/Heap/Cell.h>
@@ -23,6 +24,7 @@ class PrimitiveString final : public Cell {
 
 public:
     [[nodiscard]] static NonnullGCPtr<PrimitiveString> create(VM&, Utf16String);
+    [[nodiscard]] static NonnullGCPtr<PrimitiveString> create(VM&, String);
     [[nodiscard]] static NonnullGCPtr<PrimitiveString> create(VM&, DeprecatedString);
     [[nodiscard]] static NonnullGCPtr<PrimitiveString> create(VM&, PrimitiveString&, PrimitiveString&);
 
@@ -34,6 +36,9 @@ public:
     bool is_empty() const;
     u32 hash() const;
 
+    ThrowCompletionOr<String> utf8_string() const;
+    bool has_utf8_string() const { return m_utf8_string.has_value(); }
+
     ThrowCompletionOr<DeprecatedString> deprecated_string() const;
     bool has_deprecated_string() const { return m_deprecated_string.has_value(); }
 
@@ -45,6 +50,7 @@ public:
 
 private:
     explicit PrimitiveString(PrimitiveString&, PrimitiveString&);
+    explicit PrimitiveString(String);
     explicit PrimitiveString(DeprecatedString);
     explicit PrimitiveString(Utf16String);
 
@@ -57,6 +63,7 @@ private:
     mutable PrimitiveString* m_lhs { nullptr };
     mutable PrimitiveString* m_rhs { nullptr };
 
+    mutable Optional<String> m_utf8_string;
     mutable Optional<DeprecatedString> m_deprecated_string;
     mutable Optional<Utf16String> m_utf16_string;
 };
diff --git a/Userland/Libraries/LibJS/Runtime/VM.h b/Userland/Libraries/LibJS/Runtime/VM.h
index 8ac1aa219f..4cc9c23894 100644
--- a/Userland/Libraries/LibJS/Runtime/VM.h
+++ b/Userland/Libraries/LibJS/Runtime/VM.h
@@ -73,11 +73,18 @@ public:
     JS_ENUMERATE_WELL_KNOWN_SYMBOLS
 #undef __JS_ENUMERATE
 
+    HashMap<String, PrimitiveString*>& string_cache()
+    {
+        return m_string_cache;
+    }
+
     HashMap<DeprecatedString, PrimitiveString*>& deprecated_string_cache()
     {
         return m_deprecated_string_cache;
     }
+
     PrimitiveString& empty_string() { return *m_empty_string; }
+
     PrimitiveString& single_ascii_character_string(u8 character)
     {
         VERIFY(character < 0x80);
@@ -257,6 +264,7 @@ private:
     void import_module_dynamically(ScriptOrModule referencing_script_or_module, ModuleRequest module_request, PromiseCapability const& promise_capability);
     void finish_dynamic_import(ScriptOrModule referencing_script_or_module, ModuleRequest module_request, PromiseCapability const& promise_capability, Promise* inner_promise);
 
+    HashMap<String, PrimitiveString*> m_string_cache;
     HashMap<DeprecatedString, PrimitiveString*> m_deprecated_string_cache;
 
     Heap m_heap;