LibWeb: Use code unit offsets in Document::find_matching_text()

We were passing in byte offsets instead of UTF-16 code unit offsets,
which could lead to crashes if the offsets found exceeded the number of
code units in text fragments on the page.

Fixes #4908.

Co-authored-by: Tim Ledbetter <tim.ledbetter@ladybird.org>

Libraries/LibWeb/DOM/Document.cpp

@@ -6160,7 +6160,7 @@ void Document::set_needs_to_refresh_scroll_state(bool b)
         paintable->set_needs_to_refresh_scroll_state(b);
 }
 
-Vector<GC::Root<DOM::Range>> Document::find_matching_text(String const& query, CaseSensitivity case_sensitivity)
+Vector<GC::Root<Range>> Document::find_matching_text(String const& query, CaseSensitivity case_sensitivity)
 {
     // Ensure the layout tree exists before searching for text matches.
     update_layout(UpdateLayoutReason::DocumentFindMatchingText);
@@ -6172,16 +6172,19 @@ Vector<GC::Root<DOM::Range>> Document::find_matching_text(String const& query, C
     if (text_blocks.is_empty())
         return {};
 
-    Vector<GC::Root<DOM::Range>> matches;
+    auto utf16_query = MUST(AK::utf8_to_utf16(query));
+    Utf16View query_view { utf16_query };
+
+    Vector<GC::Root<Range>> matches;
     for (auto const& text_block : text_blocks) {
         size_t offset = 0;
         size_t i = 0;
-        auto const& text = text_block.text;
-        auto* match_start_position = &text_block.positions[0];
+        Utf16View text_view { text_block.text };
+        auto* match_start_position = text_block.positions.data();
         while (true) {
             auto match_index = case_sensitivity == CaseSensitivity::CaseInsensitive
-                ? text.find_byte_offset_ignoring_case(query, offset)
-                : text.find_byte_offset(query, offset);
+                ? text_view.find_code_unit_offset_ignoring_case(query_view, offset)
+                : text_view.find_code_unit_offset(query_view, offset);
             if (!match_index.has_value())
                 break;
 
@@ -6192,16 +6195,16 @@ Vector<GC::Root<DOM::Range>> Document::find_matching_text(String const& query, C
             auto& start_dom_node = match_start_position->dom_node;
 
             auto* match_end_position = match_start_position;
-            for (; i < text_block.positions.size() - 1 && (match_index.value() + query.bytes_as_string_view().length() > text_block.positions[i + 1].start_offset); ++i)
+            for (; i < text_block.positions.size() - 1 && (match_index.value() + query_view.length_in_code_units() > text_block.positions[i + 1].start_offset); ++i)
                 match_end_position = &text_block.positions[i + 1];
 
             auto& end_dom_node = match_end_position->dom_node;
-            auto end_position = match_index.value() + query.bytes_as_string_view().length() - match_end_position->start_offset;
+            auto end_position = match_index.value() + query_view.length_in_code_units() - match_end_position->start_offset;
 
             matches.append(Range::create(start_dom_node, start_position, end_dom_node, end_position));
             match_start_position = match_end_position;
-            offset = match_index.value() + query.bytes_as_string_view().length() + 1;
-            if (offset >= text.bytes_as_string_view().length())
+            offset = match_index.value() + query_view.length_in_code_units() + 1;
+            if (offset >= text_view.length_in_code_units())
                 break;
         }
     }

Libraries/LibWeb/DOM/Document.h

@@ -814,7 +814,7 @@ public:
     // Does document represent an embedded svg img
     [[nodiscard]] bool is_decoded_svg() const;
 
-    Vector<GC::Root<DOM::Range>> find_matching_text(String const&, CaseSensitivity);
+    Vector<GC::Root<Range>> find_matching_text(String const&, CaseSensitivity);
 
     void parse_html_from_a_string(StringView);
     static GC::Ref<Document> parse_html_unsafe(JS::VM&, StringView);

Libraries/LibWeb/Layout/Viewport.cpp

@@ -60,7 +60,7 @@ void Viewport::update_text_blocks()
 
         if (layout_node.is_box() || layout_node.is_generated()) {
             if (!builder.is_empty()) {
-                text_blocks.append({ builder.to_string_without_validation(), text_positions });
+                text_blocks.append({ MUST(AK::utf8_to_utf16(builder.string_view())), text_positions });
                 current_start_position = 0;
                 text_positions.clear_with_capacity();
                 builder.clear();
@@ -80,7 +80,8 @@ void Viewport::update_text_blocks()
                 }
 
                 auto const& current_node_text = text_node->text_for_rendering();
-                current_start_position += current_node_text.bytes_as_string_view().length();
+                auto const current_node_text_utf16 = MUST(AK::utf8_to_utf16(current_node_text));
+                current_start_position += current_node_text_utf16.data.size();
                 builder.append(move(current_node_text));
             }
         }
@@ -89,7 +90,7 @@ void Viewport::update_text_blocks()
     });
 
     if (!builder.is_empty())
-        text_blocks.append({ builder.to_string_without_validation(), text_positions });
+        text_blocks.append({ MUST(AK::utf8_to_utf16(builder.string_view())), text_positions });
 
     m_text_blocks = move(text_blocks);
 }

Libraries/LibWeb/Layout/Viewport.h

@@ -24,7 +24,7 @@ public:
         size_t start_offset { 0 };
     };
     struct TextBlock {
-        String text;
+        AK::Utf16ConversionResult text;
         Vector<TextPosition> positions;
     };
     Vector<TextBlock> const& text_blocks();

Tests/LibWeb/Crash/DOM/Window-find-code-units.html

@@ -0,0 +1,5 @@
+<!DOCTYPE html>
+😭a
+<script>
+window.find('a');
+</script>