LibGfx: Use LibUnicode to filter code points that cannot start an emoji

Userland/Libraries/LibGfx/Font/Emoji.cpp

@@ -51,35 +51,10 @@ Bitmap const* Emoji::emoji_for_code_points(ReadonlySpan<u32> const& code_points)
     return bitmap.ptr();
 }
 
-template<typename CodePointIterator>
-static bool could_be_emoji(CodePointIterator const& it)
-{
-    if (it.done())
-        return false;
-
-    static constexpr u32 supplementary_private_use_area_b_first_code_point = 0x100000;
-    if (*it >= supplementary_private_use_area_b_first_code_point) {
-        // We use Supplementary Private Use Area-B for custom Serenity emoji.
-        return true;
-    }
-
-    static auto const emoji_property = Unicode::property_from_string("Emoji"sv);
-    if (!emoji_property.has_value()) {
-        // This means Unicode data generation is disabled. Always check the disk in that case.
-        return true;
-    }
-
-    return Unicode::code_point_has_property(*it, *emoji_property);
-}
-
 template<typename CodePointIterator>
 static Bitmap const* emoji_for_code_point_iterator_impl(CodePointIterator& it)
 {
-    // NOTE: I'm sure this could be more efficient, e.g. by checking if each code point falls
+    if (!Unicode::could_be_start_of_emoji_sequence(it))
-    // into a certain range in the loop below (emojis, modifiers, variation selectors, ZWJ),
-    // and bailing out early if not. Current worst case is 10 file lookups for any sequence of
-    // code points (if the first glyph isn't part of the font in regular text rendering).
-    if (!could_be_emoji(it))
         return nullptr;
 
     constexpr size_t max_emoji_code_point_sequence_length = 10;

Userland/Libraries/LibGfx/Painter.cpp

@@ -34,6 +34,7 @@
 #include <LibGfx/TextDirection.h>
 #include <LibGfx/TextLayout.h>
 #include <LibUnicode/CharacterTypes.h>
+#include <LibUnicode/Emoji.h>
 #include <stdio.h>
 
 #if defined(AK_COMPILER_GCC)
@@ -1396,13 +1397,14 @@ void Painter::draw_glyph_or_emoji(FloatPoint point, u32 code_point, Font const&
 
 void Painter::draw_glyph_or_emoji(FloatPoint point, Utf8CodePointIterator& it, Font const& font, Color color)
 {
-    static auto const emoji_component_property = Unicode::property_from_string("Emoji_Component"sv);
-    static auto const variation_selector = Unicode::property_from_string("Variation_Selector"sv);
-
     u32 code_point = *it;
     auto next_code_point = it.peek(1);
 
     ScopeGuard consume_variation_selector = [&, initial_it = it] {
+        static auto const variation_selector = Unicode::property_from_string("Variation_Selector"sv);
+        if (!variation_selector.has_value())
+            return;
+
         // If we advanced the iterator to consume an emoji sequence, don't look for another variation selector.
         if (initial_it != it)
             return;
@@ -1413,14 +1415,7 @@ void Painter::draw_glyph_or_emoji(FloatPoint point, Utf8CodePointIterator& it, F
     };
 
     auto font_contains_glyph = font.contains_glyph(code_point);
-    auto check_for_emoji = false;
+    auto check_for_emoji = Unicode::could_be_start_of_emoji_sequence(it);
-
-    if (emoji_component_property.has_value()) {
-        auto code_point_is_emoji_component = Unicode::code_point_has_property(code_point, *emoji_component_property);
-        auto next_code_point_is_emoji_component = next_code_point.has_value() && Unicode::code_point_has_property(*next_code_point, *emoji_component_property);
-
-        check_for_emoji = code_point_is_emoji_component || next_code_point_is_emoji_component;
-    }
 
     // If the font contains the glyph, and we know it's not the start of an emoji, draw a text glyph.
     if (font_contains_glyph && !check_for_emoji) {