GenerateEmojiData.cpp 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378
  1. /*
  2. * Copyright (c) 2022, Tim Flynn <trflynn89@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include "GeneratorUtil.h"
  7. #include <AK/AnyOf.h>
  8. #include <AK/SourceGenerator.h>
  9. #include <AK/String.h>
  10. #include <AK/StringUtils.h>
  11. #include <AK/Types.h>
  12. #include <LibCore/ArgsParser.h>
  13. #include <LibCore/Directory.h>
  14. #include <LibCore/Stream.h>
  15. #include <LibUnicode/Emoji.h>
  16. struct Emoji {
  17. size_t name { 0 };
  18. Optional<String> image_path;
  19. Unicode::EmojiGroup group;
  20. String subgroup;
  21. u32 display_order { 0 };
  22. Vector<u32> code_points;
  23. String encoded_code_points;
  24. String status;
  25. size_t code_point_array_index { 0 };
  26. };
  27. struct EmojiData {
  28. UniqueStringStorage unique_strings;
  29. Vector<Emoji> emojis;
  30. };
  31. static void set_image_path_for_emoji(StringView emoji_resource_path, Emoji& emoji)
  32. {
  33. StringBuilder builder;
  34. for (auto code_point : emoji.code_points) {
  35. if (code_point == 0xfe0f)
  36. continue;
  37. if (!builder.is_empty())
  38. builder.append('_');
  39. builder.appendff("U+{:X}", code_point);
  40. }
  41. auto path = String::formatted("{}/{}.png", emoji_resource_path, builder.build());
  42. if (Core::Stream::File::exists(path))
  43. emoji.image_path = move(path);
  44. }
  45. static ErrorOr<void> parse_emoji_test_data(Core::Stream::BufferedFile& file, EmojiData& emoji_data)
  46. {
  47. static constexpr auto group_header = "# group: "sv;
  48. static constexpr auto subgroup_header = "# subgroup: "sv;
  49. Array<u8, 1024> buffer;
  50. Unicode::EmojiGroup group;
  51. String subgroup;
  52. u32 display_order { 0 };
  53. while (TRY(file.can_read_line())) {
  54. auto line = TRY(file.read_line(buffer));
  55. if (line.is_empty())
  56. continue;
  57. if (line.starts_with('#')) {
  58. if (line.starts_with(group_header)) {
  59. auto name = line.substring_view(group_header.length());
  60. group = Unicode::emoji_group_from_string(name);
  61. } else if (line.starts_with(subgroup_header)) {
  62. subgroup = line.substring_view(subgroup_header.length());
  63. }
  64. continue;
  65. }
  66. auto status_index = line.find(';');
  67. VERIFY(status_index.has_value());
  68. auto emoji_and_name_index = line.find('#', *status_index);
  69. VERIFY(emoji_and_name_index.has_value());
  70. Emoji emoji {};
  71. emoji.group = group;
  72. emoji.subgroup = subgroup;
  73. emoji.display_order = display_order++;
  74. auto code_points = line.substring_view(0, *status_index).split_view(' ');
  75. TRY(emoji.code_points.try_ensure_capacity(code_points.size()));
  76. for (auto code_point : code_points) {
  77. auto value = AK::StringUtils::convert_to_uint_from_hex<u32>(code_point);
  78. VERIFY(value.has_value());
  79. emoji.code_points.unchecked_append(*value);
  80. }
  81. auto emoji_and_name = line.substring_view(*emoji_and_name_index + 1);
  82. auto emoji_and_name_spaces = emoji_and_name.find_all(" "sv);
  83. VERIFY(emoji_and_name_spaces.size() > 2);
  84. auto name = emoji_and_name.substring_view(emoji_and_name_spaces[2]).trim_whitespace();
  85. emoji.name = emoji_data.unique_strings.ensure(name.to_titlecase_string());
  86. emoji.encoded_code_points = emoji_and_name.substring_view(0, emoji_and_name_spaces[1]).trim_whitespace();
  87. emoji.status = line.substring_view(*status_index + 1, *emoji_and_name_index - *status_index - 1).trim_whitespace();
  88. TRY(emoji_data.emojis.try_append(move(emoji)));
  89. }
  90. return {};
  91. }
  92. static ErrorOr<void> parse_emoji_serenity_data(Core::Stream::BufferedFile& file, EmojiData& emoji_data)
  93. {
  94. static constexpr auto code_point_header = "U+"sv;
  95. Array<u8, 1024> buffer;
  96. auto display_order = static_cast<u32>(emoji_data.emojis.size()) + 1u;
  97. while (TRY(file.can_read_line())) {
  98. auto line = TRY(file.read_line(buffer));
  99. if (line.is_empty())
  100. continue;
  101. auto index = line.find(code_point_header);
  102. if (!index.has_value())
  103. continue;
  104. line = line.substring_view(*index);
  105. StringBuilder builder;
  106. Emoji emoji {};
  107. emoji.group = Unicode::EmojiGroup::SerenityOS;
  108. emoji.display_order = display_order++;
  109. line.for_each_split_view(' ', SplitBehavior::Nothing, [&](auto segment) {
  110. if (segment.starts_with(code_point_header)) {
  111. segment = segment.substring_view(code_point_header.length());
  112. auto code_point = AK::StringUtils::convert_to_uint_from_hex<u32>(segment);
  113. VERIFY(code_point.has_value());
  114. emoji.code_points.append(*code_point);
  115. } else {
  116. if (!builder.is_empty())
  117. builder.append(' ');
  118. builder.append(segment);
  119. }
  120. });
  121. auto name = builder.build();
  122. if (!any_of(name, is_ascii_lower_alpha))
  123. name = name.to_titlecase();
  124. emoji.name = emoji_data.unique_strings.ensure(move(name));
  125. TRY(emoji_data.emojis.try_append(move(emoji)));
  126. }
  127. return {};
  128. }
  129. static ErrorOr<void> generate_emoji_data_header(Core::Stream::BufferedFile& file, EmojiData const&)
  130. {
  131. StringBuilder builder;
  132. SourceGenerator generator { builder };
  133. TRY(file.write(generator.as_string_view().bytes()));
  134. return {};
  135. }
  136. static ErrorOr<void> generate_emoji_data_implementation(Core::Stream::BufferedFile& file, EmojiData const& emoji_data)
  137. {
  138. StringBuilder builder;
  139. SourceGenerator generator { builder };
  140. generator.set("string_index_type"sv, emoji_data.unique_strings.type_that_fits());
  141. generator.set("emojis_size"sv, String::number(emoji_data.emojis.size()));
  142. generator.append(R"~~~(
  143. #include <AK/Array.h>
  144. #include <AK/BinarySearch.h>
  145. #include <AK/Span.h>
  146. #include <AK/StringView.h>
  147. #include <AK/Types.h>
  148. #include <LibUnicode/Emoji.h>
  149. #include <LibUnicode/EmojiData.h>
  150. namespace Unicode {
  151. )~~~");
  152. emoji_data.unique_strings.generate(generator);
  153. size_t total_code_point_count { 0 };
  154. for (auto const& emoji : emoji_data.emojis) {
  155. total_code_point_count += emoji.code_points.size();
  156. }
  157. generator.set("total_code_point_count", String::number(total_code_point_count));
  158. generator.append(R"~~~(
  159. static constexpr Array<u32, @total_code_point_count@> s_emoji_code_points { {)~~~");
  160. bool first = true;
  161. for (auto const& emoji : emoji_data.emojis) {
  162. for (auto code_point : emoji.code_points) {
  163. generator.append(first ? " "sv : ", "sv);
  164. generator.append(String::formatted("{:#x}", code_point));
  165. first = false;
  166. }
  167. }
  168. generator.append(" } };"sv);
  169. generator.append(R"~~~(
  170. struct EmojiData {
  171. constexpr Emoji to_unicode_emoji() const
  172. {
  173. Emoji emoji {};
  174. emoji.name = decode_string(name);
  175. emoji.group = static_cast<EmojiGroup>(group);
  176. emoji.display_order = display_order;
  177. emoji.code_points = code_points();
  178. return emoji;
  179. }
  180. constexpr Span<u32 const> code_points() const
  181. {
  182. return Span<u32 const>(s_emoji_code_points.data() + code_point_start, code_point_count);
  183. }
  184. @string_index_type@ name { 0 };
  185. u8 group { 0 };
  186. u32 display_order { 0 };
  187. size_t code_point_start { 0 };
  188. size_t code_point_count { 0 };
  189. };
  190. )~~~");
  191. generator.append(R"~~~(
  192. static constexpr Array<EmojiData, @emojis_size@> s_emojis { {)~~~");
  193. for (auto const& emoji : emoji_data.emojis) {
  194. generator.set("name"sv, String::number(emoji.name));
  195. generator.set("group"sv, String::number(to_underlying(emoji.group)));
  196. generator.set("display_order"sv, String::number(emoji.display_order));
  197. generator.set("code_point_start"sv, String::number(emoji.code_point_array_index));
  198. generator.set("code_point_count"sv, String::number(emoji.code_points.size()));
  199. generator.append(R"~~~(
  200. { @name@, @group@, @display_order@, @code_point_start@, @code_point_count@ },)~~~");
  201. }
  202. generator.append(R"~~~(
  203. } };
  204. Optional<Emoji> find_emoji_for_code_points(Span<u32 const> code_points)
  205. {
  206. for (auto& emoji : s_emojis) {
  207. if (emoji.code_points() == code_points)
  208. return emoji.to_unicode_emoji();
  209. }
  210. return {};
  211. }
  212. }
  213. )~~~");
  214. TRY(file.write(generator.as_string_view().bytes()));
  215. return {};
  216. }
  217. static ErrorOr<void> generate_emoji_installation(Core::Stream::BufferedFile& file, EmojiData const& emoji_data)
  218. {
  219. StringBuilder builder;
  220. SourceGenerator generator { builder };
  221. auto current_group = Unicode::EmojiGroup::Unknown;
  222. StringView current_subgroup;
  223. for (auto const& emoji : emoji_data.emojis) {
  224. if (!emoji.image_path.has_value())
  225. continue;
  226. if (emoji.group == Unicode::EmojiGroup::SerenityOS)
  227. continue; // SerenityOS emojis are in emoji-serenity.txt
  228. if (current_group != emoji.group) {
  229. if (!builder.is_empty())
  230. generator.append("\n"sv);
  231. generator.set("group"sv, Unicode::emoji_group_to_string(emoji.group));
  232. generator.append("# group: @group@\n");
  233. current_group = emoji.group;
  234. }
  235. if (current_subgroup != emoji.subgroup) {
  236. generator.set("subgroup"sv, emoji.subgroup);
  237. generator.append("\n# subgroup: @subgroup@\n");
  238. current_subgroup = emoji.subgroup;
  239. }
  240. generator.set("emoji"sv, emoji.encoded_code_points);
  241. generator.set("name"sv, emoji_data.unique_strings.get(emoji.name));
  242. generator.set("status"sv, emoji.status);
  243. generator.append("@emoji@"sv);
  244. generator.append(" - "sv);
  245. generator.append(String::join(" "sv, emoji.code_points, "U+{:X}"sv));
  246. generator.append(" @name@ (@status@)\n"sv);
  247. }
  248. TRY(file.write(generator.as_string_view().bytes()));
  249. return {};
  250. }
  251. ErrorOr<int> serenity_main(Main::Arguments arguments)
  252. {
  253. StringView generated_header_path;
  254. StringView generated_implementation_path;
  255. StringView generated_installation_path;
  256. StringView emoji_test_path;
  257. StringView emoji_serenity_path;
  258. StringView emoji_resource_path;
  259. Core::ArgsParser args_parser;
  260. args_parser.add_option(generated_header_path, "Path to the Unicode Data header file to generate", "generated-header-path", 'h', "generated-header-path");
  261. args_parser.add_option(generated_implementation_path, "Path to the Unicode Data implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  262. args_parser.add_option(generated_installation_path, "Path to the emoji.txt file to generate", "generated-installation-path", 'i', "generated-installation-path");
  263. args_parser.add_option(emoji_test_path, "Path to emoji-test.txt file", "emoji-test-path", 'e', "emoji-test-path");
  264. args_parser.add_option(emoji_serenity_path, "Path to emoji-serenity.txt file", "emoji-serenity-path", 's', "emoji-serenity-path");
  265. args_parser.add_option(emoji_resource_path, "Path to the /res/emoji directory", "emoji-resource-path", 'r', "emoji-resource-path");
  266. args_parser.parse(arguments);
  267. auto emoji_test_file = TRY(open_file(emoji_test_path, Core::Stream::OpenMode::Read));
  268. VERIFY(!emoji_resource_path.is_empty() && Core::Stream::File::exists(emoji_resource_path));
  269. EmojiData emoji_data {};
  270. TRY(parse_emoji_test_data(*emoji_test_file, emoji_data));
  271. if (!emoji_serenity_path.is_empty()) {
  272. auto emoji_serenity_file = TRY(open_file(emoji_serenity_path, Core::Stream::OpenMode::Read));
  273. TRY(parse_emoji_serenity_data(*emoji_serenity_file, emoji_data));
  274. }
  275. size_t code_point_array_index { 0 };
  276. for (auto& emoji : emoji_data.emojis) {
  277. emoji.code_point_array_index = code_point_array_index;
  278. code_point_array_index += emoji.code_points.size();
  279. }
  280. if (!generated_header_path.is_empty()) {
  281. auto generated_header_file = TRY(open_file(generated_header_path, Core::Stream::OpenMode::Write));
  282. TRY(generate_emoji_data_header(*generated_header_file, emoji_data));
  283. }
  284. if (!generated_implementation_path.is_empty()) {
  285. auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::Stream::OpenMode::Write));
  286. TRY(generate_emoji_data_implementation(*generated_implementation_file, emoji_data));
  287. }
  288. if (!generated_installation_path.is_empty()) {
  289. TRY(Core::Directory::create(LexicalPath { generated_installation_path }.parent(), Core::Directory::CreateDirectories::Yes));
  290. for (auto& emoji : emoji_data.emojis)
  291. set_image_path_for_emoji(emoji_resource_path, emoji);
  292. auto generated_installation_file = TRY(open_file(generated_installation_path, Core::Stream::OpenMode::Write));
  293. TRY(generate_emoji_installation(*generated_installation_file, emoji_data));
  294. }
  295. return 0;
  296. }