From 733a92820bfe1e4e30531c65226a06539d5dea11 Mon Sep 17 00:00:00 2001
From: Timothy Flynn <trflynn89@pm.me>
Date: Mon, 19 Jul 2021 17:02:44 -0400
Subject: [PATCH] LibJS: Implement String.prototype.replaceAll with UTF-16 code
 units

---
 .../LibJS/Runtime/StringPrototype.cpp         | 33 ++++++++++++-------
 .../String/String.prototype.replaceAll.js     | 12 +++++++
 2 files changed, 33 insertions(+), 12 deletions(-)

diff --git a/Userland/Libraries/LibJS/Runtime/StringPrototype.cpp b/Userland/Libraries/LibJS/Runtime/StringPrototype.cpp
index f98e93ec20..acd43b7063 100644
--- a/Userland/Libraries/LibJS/Runtime/StringPrototype.cpp
+++ b/Userland/Libraries/LibJS/Runtime/StringPrototype.cpp
@@ -997,33 +997,40 @@ JS_DEFINE_NATIVE_FUNCTION(StringPrototype::replace_all)
         return {};
 
     if (!replace_value.is_function()) {
-        auto replace_string = replace_value.to_string(global_object);
+        auto replace_string = replace_value.to_utf16_string(global_object);
         if (vm.exception())
             return {};
-
-        replace_value = js_string(vm, move(replace_string));
+        replace_value = js_string(vm, Utf16View { replace_string });
         if (vm.exception())
             return {};
     }
 
+    auto utf16_string = AK::utf8_to_utf16(string);
+    Utf16View utf16_string_view { utf16_string };
+    auto string_length = utf16_string_view.length_in_code_units();
+
+    auto utf16_search_string = AK::utf8_to_utf16(search_string);
+    Utf16View utf16_search_view { utf16_search_string };
+    auto search_length = utf16_search_view.length_in_code_units();
+
     Vector<size_t> match_positions;
-    size_t advance_by = max(1u, search_string.length());
-    auto position = string.find(search_string);
+    size_t advance_by = max(1u, search_length);
+    auto position = string_index_of(utf16_string_view, utf16_search_view, 0);
 
     while (position.has_value()) {
         match_positions.append(*position);
-        position = string.find(search_string, *position + advance_by);
+        position = string_index_of(utf16_string_view, utf16_search_view, *position + advance_by);
     }
 
     size_t end_of_last_match = 0;
     StringBuilder result;
 
     for (auto position : match_positions) {
-        auto preserved = string.substring_view(end_of_last_match, position - end_of_last_match);
+        auto preserved = utf16_string_view.substring_view(end_of_last_match, position - end_of_last_match);
         String replacement;
 
         if (replace_value.is_function()) {
-            auto result = vm.call(replace_value.as_function(), js_undefined(), js_string(vm, search_string), Value(position), js_string(vm, string));
+            auto result = vm.call(replace_value.as_function(), js_undefined(), js_string(vm, utf16_search_view), Value(position), js_string(vm, utf16_string_view));
             if (vm.exception())
                 return {};
 
@@ -1036,14 +1043,16 @@ JS_DEFINE_NATIVE_FUNCTION(StringPrototype::replace_all)
                 return {};
         }
 
-        result.append(preserved);
+        result.append(preserved.to_utf8(Utf16View::AllowInvalidCodeUnits::Yes));
         result.append(replacement);
 
-        end_of_last_match = position + search_string.length();
+        end_of_last_match = position + search_length;
     }
 
-    if (end_of_last_match < string.length())
-        result.append(string.substring_view(end_of_last_match));
+    if (end_of_last_match < string_length) {
+        utf16_string_view = utf16_string_view.substring_view(end_of_last_match);
+        result.append(utf16_string_view.to_utf8(Utf16View::AllowInvalidCodeUnits::Yes));
+    }
 
     return js_string(vm, result.build());
 }
diff --git a/Userland/Libraries/LibJS/Tests/builtins/String/String.prototype.replaceAll.js b/Userland/Libraries/LibJS/Tests/builtins/String/String.prototype.replaceAll.js
index d2a6d3be9f..271bcc2bc9 100644
--- a/Userland/Libraries/LibJS/Tests/builtins/String/String.prototype.replaceAll.js
+++ b/Userland/Libraries/LibJS/Tests/builtins/String/String.prototype.replaceAll.js
@@ -143,3 +143,15 @@ test("search value is coerced to a string", () => {
     expect(newString).toBe("abc");
     expect(coerced).toBe("x");
 });
+
+test("UTF-16", () => {
+    expect("😀".replaceAll("😀", "")).toBe("");
+    expect("😀".replaceAll("\ud83d", "")).toBe("\ude00");
+    expect("😀".replaceAll("\ude00", "")).toBe("\ud83d");
+    expect("😀😀😀".replaceAll("\ud83d", "")).toBe("\ude00\ude00\ude00");
+    expect("😀😀😀".replaceAll("\ude00", "")).toBe("\ud83d\ud83d\ud83d");
+
+    // FIXME: RegExp.prototype [ @@replace ] also needs to support UTF-16.
+    // expect("😀".replaceAll(/\ud83d/g, "")).toBe("\ude00");
+    // expect("😀".replaceAll(/\ude00/g, "")).toBe("\ud83d");
+});