LibUnicode: Add a method to check if a code point could start an emoji

trflynn89 · linusg · commit 1484d3d9f51e · 2023-02-24T19:48:47.000+01:00
diff --git a/Tests/LibUnicode/CMakeLists.txt b/Tests/LibUnicode/CMakeLists.txt
@@ -1,4 +1,5 @@
 set(TEST_SOURCES
+    TestEmoji.cpp
     TestUnicodeCharacterTypes.cpp
     TestUnicodeNormalization.cpp
 )
diff --git a/Tests/LibUnicode/TestEmoji.cpp b/Tests/LibUnicode/TestEmoji.cpp
@@ -0,0 +1,54 @@
+/*
+ * Copyright (c) 2023, Tim Flynn <trflynn89@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#include <AK/Array.h>
+#include <AK/CharacterTypes.h>
+#include <AK/String.h>
+#include <AK/Utf8View.h>
+#include <LibTest/TestCase.h>
+#include <LibUnicode/Emoji.h>
+
+// These emojis are the first subgroup in each Unicode-defined group of emojis, plus some interesting
+// hand-picked test cases (such as keycap emoji, which begin with ASCII symbols, and country flags).
+static constexpr auto s_smileys_emotion = Array { "😀"sv, "😃"sv, "😄"sv, "😁"sv, "😆"sv, "😅"sv, "🤣"sv, "😂"sv, "🙂"sv, "🙃"sv, "🫠"sv, "😉"sv, "😊"sv, "😇"sv };
+static constexpr auto s_people_body = Array { "👋"sv, "🤚"sv, "🖐️"sv, "🖐"sv, "✋"sv, "🫱"sv, "🫲"sv, "🫳"sv, "🫴"sv, "🫷"sv, "🫸"sv };
+static constexpr auto s_animals_nature = Array { "🐶"sv, "🐕"sv, "🐕‍🦺"sv, "🐩"sv, "🦊"sv, "🦝"sv, "🐱"sv, "🐈"sv, "🐈‍⬛"sv, "🦁"sv, "🐯"sv, "🐴"sv, "🫎"sv, "🫏"sv, "🐎"sv, "🦄"sv, "🦓"sv, "🦌"sv, "🦬"sv, "🐮"sv, "🐷"sv, "🐖"sv, "🐗"sv, "🐽"sv, "🐑"sv, "🦙"sv, "🦒"sv, "🐘"sv, "🐭"sv, "🐁"sv, "🐀"sv, "🐰"sv, "🐇"sv, "🐿️"sv, "🐿"sv, "🦔"sv, "🦇"sv, "🐻"sv, "🐻‍❄️"sv, "🐻‍❄"sv, "🐨"sv, "🐼"sv, "🦥"sv, "🦘"sv, "🦡"sv, "🐾"sv };
+static constexpr auto s_food_drink = Array { "🍇"sv, "🍈"sv, "🍉"sv, "🍊"sv, "🍋"sv, "🍌"sv, "🍍"sv, "🥭"sv, "🍎"sv, "🍏"sv, "🍐"sv, "🍑"sv, "🍒"sv, "🍓"sv, "🫐"sv, "🥝"sv, "🍅"sv, "🫒"sv, "🥥"sv };
+static constexpr auto s_travel_places = Array { "🌍"sv, "🌎"sv, "🌏"sv, "🌐"sv, "🗺️"sv, "🗺"sv, "🗾"sv, "🧭"sv };
+static constexpr auto s_activities = Array { "🎃"sv, "🎄"sv, "🎆"sv, "🎇"sv, "🧨"sv, "✨"sv, "🎈"sv, "🎉"sv, "🎊"sv, "🎋"sv, "🎍"sv, "🎏"sv, "🎑"sv, "🎀"sv, "🎁"sv, "🎗️"sv, "🎗"sv, "🎟️"sv, "🎟"sv, "🎫"sv };
+static constexpr auto s_objects = Array { "👓"sv, "🕶️"sv, "🕶"sv, "🦺"sv, "👔"sv, "👖"sv, "🧦"sv, "👗"sv, "🥻"sv, "🩱"sv, "🩲"sv, "🩳"sv, "👙"sv, "🪭"sv, "👛"sv, "👜"sv, "🛍️"sv, "🛍"sv, "🩴"sv, "👡"sv, "👢"sv, "🪮"sv, "👑"sv, "🎩"sv, "🎓"sv, "🪖"sv, "⛑️"sv, "⛑"sv, "💄"sv, "💍"sv, "💎"sv };
+static constexpr auto s_symbols = Array { "🚮"sv, "🚰"sv, "♿"sv, "🚹"sv, "🚺"sv, "🚾"sv, "🛂"sv, "🛃"sv, "🛄"sv, "🛅"sv, "#️⃣"sv, "#⃣"sv, "*️⃣"sv, "*⃣"sv, "0️⃣"sv, "0⃣"sv, "1️⃣"sv, "1⃣"sv, "2️⃣"sv, "2⃣"sv, "3️⃣"sv, "3⃣"sv, "4️⃣"sv, "4⃣"sv, "5️⃣"sv, "5⃣"sv, "6️⃣"sv, "6⃣"sv, "7️⃣"sv, "7⃣"sv, "8️⃣"sv, "8⃣"sv, "9️⃣"sv, "9⃣"sv, "🔟"sv };
+static constexpr auto s_flags = Array { "🏁"sv, "🚩"sv, "🎌"sv, "🏴"sv, "🏳️"sv, "🏳"sv, "🏳️‍🌈"sv, "🏳‍🌈"sv, "🏳️‍⚧️"sv, "🏳‍⚧️"sv, "🏳️‍⚧"sv, "🏳‍⚧"sv, "🏴‍☠️"sv, "🏴‍☠"sv, "🇦🇨"sv, "🇦🇩"sv, "🇦🇪"sv, "🇦🇫"sv, "🇦🇬"sv, "🇦🇮"sv, "🇦🇱"sv, "🇦🇲"sv, "🇦🇴"sv, "🇦🇶"sv, "🇦🇷"sv, "🇦🇸"sv, "🇦🇹"sv, "🇦🇺"sv, "🇦🇼"sv, "🇦🇽"sv, "🇦🇿"sv, "🇧🇦"sv, "🇧🇧"sv, "🇧🇩"sv, "🇧🇪"sv, "🇧🇫"sv, "🇧🇬"sv, "🇧🇭"sv, "🇧🇮"sv, "🇧🇯"sv, "🇧🇱"sv, "🇧🇲"sv, "🇧🇳"sv, "🇧🇴"sv, "🇧🇶"sv, "🇧🇷"sv, "🇧🇸"sv };
+
+TEST_CASE(emoji)
+{
+    auto test_emojis = [](auto const& emojis) {
+        for (auto emoji : emojis) {
+            Utf8View view { emoji };
+            EXPECT(Unicode::could_be_start_of_emoji_sequence(view.begin()));
+        }
+    };
+
+    test_emojis(s_smileys_emotion);
+    test_emojis(s_people_body);
+    test_emojis(s_animals_nature);
+    test_emojis(s_food_drink);
+    test_emojis(s_travel_places);
+    test_emojis(s_activities);
+    test_emojis(s_objects);
+    test_emojis(s_symbols);
+    test_emojis(s_flags);
+}
+
+TEST_CASE(ascii_is_not_emoji)
+{
+    for (u32 code_point = 0u; is_ascii(code_point); ++code_point) {
+        auto string = String::from_code_point(code_point);
+        Utf8View view { string };
+
+        EXPECT(!Unicode::could_be_start_of_emoji_sequence(view.begin()));
+    }
+}
diff --git a/Userland/Libraries/LibUnicode/Emoji.cpp b/Userland/Libraries/LibUnicode/Emoji.cpp
@@ -1,13 +1,115 @@
 /*
- * Copyright (c) 2022, Tim Flynn <trflynn89@serenityos.org>
+ * Copyright (c) 2022-2023, Tim Flynn <trflynn89@serenityos.org>
  *
  * SPDX-License-Identifier: BSD-2-Clause
  */
 
+#include <AK/CharacterTypes.h>
+#include <AK/Utf32View.h>
+#include <AK/Utf8View.h>
+#include <LibUnicode/CharacterTypes.h>
 #include <LibUnicode/Emoji.h>
 
+#if ENABLE_UNICODE_DATA
+#    include <LibUnicode/UnicodeData.h>
+#endif
+
 namespace Unicode {
 
 Optional<Emoji> __attribute__((weak)) find_emoji_for_code_points(ReadonlySpan<u32>) { return {}; }
 
+#if ENABLE_UNICODE_DATA
+
+// https://unicode.org/reports/tr51/#def_emoji_core_sequence
+static bool could_be_start_of_emoji_core_sequence(u32 code_point, Optional<u32> const& next_code_point)
+{
+    // emoji_core_sequence := emoji_character | emoji_presentation_sequence | emoji_keycap_sequence | emoji_modifier_sequence | emoji_flag_sequence
+
+    static constexpr auto emoji_presentation_selector = 0xFE0Fu;
+    static constexpr auto combining_enclosing_keycap = 0x20E3u;
+
+    // https://unicode.org/reports/tr51/#def_emoji_keycap_sequence
+    // emoji_keycap_sequence := [0-9#*] \x{FE0F 20E3}
+    if (is_ascii_digit(code_point) || code_point == '#' || code_point == '*')
+        return next_code_point == emoji_presentation_selector || next_code_point == combining_enclosing_keycap;
+
+    // A little non-standard, but all other ASCII code points are not the beginning of any emoji sequence.
+    if (is_ascii(code_point))
+        return false;
+
+    // https://unicode.org/reports/tr51/#def_emoji_character
+    if (code_point_has_property(code_point, Property::Emoji))
+        return true;
+
+    // https://unicode.org/reports/tr51/#def_emoji_presentation_sequence
+    // emoji_presentation_sequence := emoji_character emoji_presentation_selector
+    if (next_code_point == emoji_presentation_selector)
+        return true;
+
+    // https://unicode.org/reports/tr51/#def_emoji_modifier_sequence
+    // emoji_modifier_sequence := emoji_modifier_base emoji_modifier
+    if (code_point_has_property(code_point, Property::Emoji_Modifier_Base))
+        return true;
+
+    // https://unicode.org/reports/tr51/#def_emoji_flag_sequence
+    // emoji_flag_sequence := regional_indicator regional_indicator
+    if (code_point_has_property(code_point, Property::Regional_Indicator))
+        return true;
+
+    return false;
+}
+
+static bool could_be_start_of_serenity_emoji(u32 code_point)
+{
+    // We use Supplementary Private Use Area-B for custom Serenity emoji, starting at U+10CD00.
+    static constexpr auto first_custom_serenity_emoji_code_point = 0x10CD00u;
+
+    return code_point >= first_custom_serenity_emoji_code_point;
+}
+
+#endif
+
+// https://unicode.org/reports/tr51/#def_emoji_sequence
+template<typename CodePointIterator>
+static bool could_be_start_of_emoji_sequence_impl(CodePointIterator const& it)
+{
+    // emoji_sequence := emoji_core_sequence | emoji_zwj_sequence | emoji_tag_sequence
+
+    if (it.done())
+        return false;
+
+#if ENABLE_UNICODE_DATA
+    // The purpose of this method is to quickly filter out code points that cannot be the start of
+    // an emoji. The emoji_core_sequence definition alone captures the start of all possible
+    // emoji_zwj_sequence and emoji_tag_sequence emojis, because:
+    //
+    //     * emoji_zwj_sequence must begin with emoji_zwj_element, which is:
+    //       emoji_zwj_sequence := emoji_core_sequence | emoji_tag_sequence
+    //
+    //     * emoji_tag_sequence must begin with tag_base, which is:
+    //       tag_base := emoji_character | emoji_modifier_sequence | emoji_presentation_sequence
+    //       Note that this is a subset of emoji_core_sequence.
+    auto code_point = *it;
+    auto next_code_point = it.peek(1);
+
+    if (could_be_start_of_emoji_core_sequence(code_point, next_code_point))
+        return true;
+    if (could_be_start_of_serenity_emoji(code_point))
+        return true;
+    return false;
+#else
+    return true;
+#endif
+}
+
+bool could_be_start_of_emoji_sequence(Utf8CodePointIterator const& it)
+{
+    return could_be_start_of_emoji_sequence_impl(it);
+}
+
+bool could_be_start_of_emoji_sequence(Utf32CodePointIterator const& it)
+{
+    return could_be_start_of_emoji_sequence_impl(it);
+}
+
 }
diff --git a/Userland/Libraries/LibUnicode/Emoji.h b/Userland/Libraries/LibUnicode/Emoji.h
@@ -1,11 +1,12 @@
 /*
- * Copyright (c) 2022, Tim Flynn <trflynn89@serenityos.org>
+ * Copyright (c) 2022-2023, Tim Flynn <trflynn89@serenityos.org>
  *
  * SPDX-License-Identifier: BSD-2-Clause
  */
 
 #pragma once
 
+#include <AK/Forward.h>
 #include <AK/Optional.h>
 #include <AK/StringView.h>
 #include <AK/Types.h>
@@ -46,6 +47,9 @@ Optional<Emoji> find_emoji_for_code_points(u32 const (&code_points)[Size])
     return find_emoji_for_code_points(ReadonlySpan<u32> { code_points });
 }
 
+bool could_be_start_of_emoji_sequence(Utf8CodePointIterator const&);
+bool could_be_start_of_emoji_sequence(Utf32CodePointIterator const&);
+
 constexpr StringView emoji_group_to_string(EmojiGroup group)
 {
     switch (group) {

Original file line number	Diff line number	Diff line change
`@@ -1,4 +1,5 @@`
`1`	`1`	`set(TEST_SOURCES`
	`2`	`+ TestEmoji.cpp`
`2`	`3`	`TestUnicodeCharacterTypes.cpp`
`3`	`4`	`TestUnicodeNormalization.cpp`
`4`	`5`	`)`