GenerateEmojiData.cpp 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379
  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/DeprecatedString.h>
  9. #include <AK/SourceGenerator.h>
  10. #include <AK/StringUtils.h>
  11. #include <AK/Types.h>
  12. #include <LibCore/ArgsParser.h>
  13. #include <LibCore/DeprecatedFile.h>
  14. #include <LibCore/Directory.h>
  15. #include <LibCore/Stream.h>
  16. #include <LibUnicode/Emoji.h>
  17. struct Emoji {
  18. size_t name { 0 };
  19. Optional<DeprecatedString> image_path;
  20. Unicode::EmojiGroup group;
  21. DeprecatedString subgroup;
  22. u32 display_order { 0 };
  23. Vector<u32> code_points;
  24. DeprecatedString encoded_code_points;
  25. DeprecatedString status;
  26. size_t code_point_array_index { 0 };
  27. };
  28. struct EmojiData {
  29. UniqueStringStorage unique_strings;
  30. Vector<Emoji> emojis;
  31. };
  32. static void set_image_path_for_emoji(StringView emoji_resource_path, Emoji& emoji)
  33. {
  34. StringBuilder builder;
  35. for (auto code_point : emoji.code_points) {
  36. if (code_point == 0xfe0f)
  37. continue;
  38. if (!builder.is_empty())
  39. builder.append('_');
  40. builder.appendff("U+{:X}", code_point);
  41. }
  42. auto path = DeprecatedString::formatted("{}/{}.png", emoji_resource_path, builder.to_deprecated_string());
  43. if (Core::DeprecatedFile::exists(path))
  44. emoji.image_path = move(path);
  45. }
  46. static ErrorOr<void> parse_emoji_test_data(Core::BufferedFile& file, EmojiData& emoji_data)
  47. {
  48. static constexpr auto group_header = "# group: "sv;
  49. static constexpr auto subgroup_header = "# subgroup: "sv;
  50. Array<u8, 1024> buffer;
  51. Unicode::EmojiGroup group;
  52. DeprecatedString subgroup;
  53. u32 display_order { 0 };
  54. while (TRY(file.can_read_line())) {
  55. auto line = TRY(file.read_line(buffer));
  56. if (line.is_empty())
  57. continue;
  58. if (line.starts_with('#')) {
  59. if (line.starts_with(group_header)) {
  60. auto name = line.substring_view(group_header.length());
  61. group = Unicode::emoji_group_from_string(name);
  62. } else if (line.starts_with(subgroup_header)) {
  63. subgroup = line.substring_view(subgroup_header.length());
  64. }
  65. continue;
  66. }
  67. auto status_index = line.find(';');
  68. VERIFY(status_index.has_value());
  69. auto emoji_and_name_index = line.find('#', *status_index);
  70. VERIFY(emoji_and_name_index.has_value());
  71. Emoji emoji {};
  72. emoji.group = group;
  73. emoji.subgroup = subgroup;
  74. emoji.display_order = display_order++;
  75. auto code_points = line.substring_view(0, *status_index).split_view(' ');
  76. TRY(emoji.code_points.try_ensure_capacity(code_points.size()));
  77. for (auto code_point : code_points) {
  78. auto value = AK::StringUtils::convert_to_uint_from_hex<u32>(code_point);
  79. VERIFY(value.has_value());
  80. emoji.code_points.unchecked_append(*value);
  81. }
  82. auto emoji_and_name = line.substring_view(*emoji_and_name_index + 1);
  83. auto emoji_and_name_spaces = emoji_and_name.find_all(" "sv);
  84. VERIFY(emoji_and_name_spaces.size() > 2);
  85. auto name = emoji_and_name.substring_view(emoji_and_name_spaces[2]).trim_whitespace();
  86. emoji.name = emoji_data.unique_strings.ensure(name.to_titlecase_string());
  87. emoji.encoded_code_points = emoji_and_name.substring_view(0, emoji_and_name_spaces[1]).trim_whitespace();
  88. emoji.status = line.substring_view(*status_index + 1, *emoji_and_name_index - *status_index - 1).trim_whitespace();
  89. TRY(emoji_data.emojis.try_append(move(emoji)));
  90. }
  91. return {};
  92. }
  93. static ErrorOr<void> parse_emoji_serenity_data(Core::BufferedFile& file, EmojiData& emoji_data)
  94. {
  95. static constexpr auto code_point_header = "U+"sv;
  96. Array<u8, 1024> buffer;
  97. auto display_order = static_cast<u32>(emoji_data.emojis.size()) + 1u;
  98. while (TRY(file.can_read_line())) {
  99. auto line = TRY(file.read_line(buffer));
  100. if (line.is_empty())
  101. continue;
  102. auto index = line.find(code_point_header);
  103. if (!index.has_value())
  104. continue;
  105. line = line.substring_view(*index);
  106. StringBuilder builder;
  107. Emoji emoji {};
  108. emoji.group = Unicode::EmojiGroup::SerenityOS;
  109. emoji.display_order = display_order++;
  110. line.for_each_split_view(' ', SplitBehavior::Nothing, [&](auto segment) {
  111. if (segment.starts_with(code_point_header)) {
  112. segment = segment.substring_view(code_point_header.length());
  113. auto code_point = AK::StringUtils::convert_to_uint_from_hex<u32>(segment);
  114. VERIFY(code_point.has_value());
  115. emoji.code_points.append(*code_point);
  116. } else {
  117. if (!builder.is_empty())
  118. builder.append(' ');
  119. builder.append(segment);
  120. }
  121. });
  122. auto name = builder.to_deprecated_string();
  123. if (!any_of(name, is_ascii_lower_alpha))
  124. name = name.to_titlecase();
  125. emoji.name = emoji_data.unique_strings.ensure(move(name));
  126. TRY(emoji_data.emojis.try_append(move(emoji)));
  127. }
  128. return {};
  129. }
  130. static ErrorOr<void> generate_emoji_data_header(Core::BufferedFile& file, EmojiData const&)
  131. {
  132. StringBuilder builder;
  133. SourceGenerator generator { builder };
  134. TRY(file.write(generator.as_string_view().bytes()));
  135. return {};
  136. }
  137. static ErrorOr<void> generate_emoji_data_implementation(Core::BufferedFile& file, EmojiData const& emoji_data)
  138. {
  139. StringBuilder builder;
  140. SourceGenerator generator { builder };
  141. generator.set("string_index_type"sv, emoji_data.unique_strings.type_that_fits());
  142. generator.set("emojis_size"sv, DeprecatedString::number(emoji_data.emojis.size()));
  143. generator.append(R"~~~(
  144. #include <AK/Array.h>
  145. #include <AK/BinarySearch.h>
  146. #include <AK/Span.h>
  147. #include <AK/StringView.h>
  148. #include <AK/Types.h>
  149. #include <LibUnicode/Emoji.h>
  150. #include <LibUnicode/EmojiData.h>
  151. namespace Unicode {
  152. )~~~");
  153. emoji_data.unique_strings.generate(generator);
  154. size_t total_code_point_count { 0 };
  155. for (auto const& emoji : emoji_data.emojis) {
  156. total_code_point_count += emoji.code_points.size();
  157. }
  158. generator.set("total_code_point_count", DeprecatedString::number(total_code_point_count));
  159. generator.append(R"~~~(
  160. static constexpr Array<u32, @total_code_point_count@> s_emoji_code_points { {)~~~");
  161. bool first = true;
  162. for (auto const& emoji : emoji_data.emojis) {
  163. for (auto code_point : emoji.code_points) {
  164. generator.append(first ? " "sv : ", "sv);
  165. generator.append(DeprecatedString::formatted("{:#x}", code_point));
  166. first = false;
  167. }
  168. }
  169. generator.append(" } };"sv);
  170. generator.append(R"~~~(
  171. struct EmojiData {
  172. constexpr Emoji to_unicode_emoji() const
  173. {
  174. Emoji emoji {};
  175. emoji.name = decode_string(name);
  176. emoji.group = static_cast<EmojiGroup>(group);
  177. emoji.display_order = display_order;
  178. emoji.code_points = code_points();
  179. return emoji;
  180. }
  181. constexpr ReadonlySpan<u32> code_points() const
  182. {
  183. return ReadonlySpan<u32>(s_emoji_code_points.data() + code_point_start, code_point_count);
  184. }
  185. @string_index_type@ name { 0 };
  186. u8 group { 0 };
  187. u32 display_order { 0 };
  188. size_t code_point_start { 0 };
  189. size_t code_point_count { 0 };
  190. };
  191. )~~~");
  192. generator.append(R"~~~(
  193. static constexpr Array<EmojiData, @emojis_size@> s_emojis { {)~~~");
  194. for (auto const& emoji : emoji_data.emojis) {
  195. generator.set("name"sv, DeprecatedString::number(emoji.name));
  196. generator.set("group"sv, DeprecatedString::number(to_underlying(emoji.group)));
  197. generator.set("display_order"sv, DeprecatedString::number(emoji.display_order));
  198. generator.set("code_point_start"sv, DeprecatedString::number(emoji.code_point_array_index));
  199. generator.set("code_point_count"sv, DeprecatedString::number(emoji.code_points.size()));
  200. generator.append(R"~~~(
  201. { @name@, @group@, @display_order@, @code_point_start@, @code_point_count@ },)~~~");
  202. }
  203. generator.append(R"~~~(
  204. } };
  205. Optional<Emoji> find_emoji_for_code_points(ReadonlySpan<u32> code_points)
  206. {
  207. for (auto& emoji : s_emojis) {
  208. if (emoji.code_points() == code_points)
  209. return emoji.to_unicode_emoji();
  210. }
  211. return {};
  212. }
  213. }
  214. )~~~");
  215. TRY(file.write(generator.as_string_view().bytes()));
  216. return {};
  217. }
  218. static ErrorOr<void> generate_emoji_installation(Core::BufferedFile& file, EmojiData const& emoji_data)
  219. {
  220. StringBuilder builder;
  221. SourceGenerator generator { builder };
  222. auto current_group = Unicode::EmojiGroup::Unknown;
  223. StringView current_subgroup;
  224. for (auto const& emoji : emoji_data.emojis) {
  225. if (!emoji.image_path.has_value())
  226. continue;
  227. if (emoji.group == Unicode::EmojiGroup::SerenityOS)
  228. continue; // SerenityOS emojis are in emoji-serenity.txt
  229. if (current_group != emoji.group) {
  230. if (!builder.is_empty())
  231. generator.append("\n"sv);
  232. generator.set("group"sv, Unicode::emoji_group_to_string(emoji.group));
  233. generator.append("# group: @group@\n");
  234. current_group = emoji.group;
  235. }
  236. if (current_subgroup != emoji.subgroup) {
  237. generator.set("subgroup"sv, emoji.subgroup);
  238. generator.append("\n# subgroup: @subgroup@\n");
  239. current_subgroup = emoji.subgroup;
  240. }
  241. generator.set("emoji"sv, emoji.encoded_code_points);
  242. generator.set("name"sv, emoji_data.unique_strings.get(emoji.name));
  243. generator.set("status"sv, emoji.status);
  244. generator.append("@emoji@"sv);
  245. generator.append(" - "sv);
  246. generator.append(DeprecatedString::join(" "sv, emoji.code_points, "U+{:X}"sv));
  247. generator.append(" @name@ (@status@)\n"sv);
  248. }
  249. TRY(file.write(generator.as_string_view().bytes()));
  250. return {};
  251. }
  252. ErrorOr<int> serenity_main(Main::Arguments arguments)
  253. {
  254. StringView generated_header_path;
  255. StringView generated_implementation_path;
  256. StringView generated_installation_path;
  257. StringView emoji_test_path;
  258. StringView emoji_serenity_path;
  259. StringView emoji_resource_path;
  260. Core::ArgsParser args_parser;
  261. args_parser.add_option(generated_header_path, "Path to the Unicode Data header file to generate", "generated-header-path", 'h', "generated-header-path");
  262. args_parser.add_option(generated_implementation_path, "Path to the Unicode Data implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  263. args_parser.add_option(generated_installation_path, "Path to the emoji.txt file to generate", "generated-installation-path", 'i', "generated-installation-path");
  264. args_parser.add_option(emoji_test_path, "Path to emoji-test.txt file", "emoji-test-path", 'e', "emoji-test-path");
  265. args_parser.add_option(emoji_serenity_path, "Path to emoji-serenity.txt file", "emoji-serenity-path", 's', "emoji-serenity-path");
  266. args_parser.add_option(emoji_resource_path, "Path to the /res/emoji directory", "emoji-resource-path", 'r', "emoji-resource-path");
  267. args_parser.parse(arguments);
  268. auto emoji_test_file = TRY(open_file(emoji_test_path, Core::File::OpenMode::Read));
  269. VERIFY(!emoji_resource_path.is_empty() && Core::DeprecatedFile::exists(emoji_resource_path));
  270. EmojiData emoji_data {};
  271. TRY(parse_emoji_test_data(*emoji_test_file, emoji_data));
  272. if (!emoji_serenity_path.is_empty()) {
  273. auto emoji_serenity_file = TRY(open_file(emoji_serenity_path, Core::File::OpenMode::Read));
  274. TRY(parse_emoji_serenity_data(*emoji_serenity_file, emoji_data));
  275. }
  276. size_t code_point_array_index { 0 };
  277. for (auto& emoji : emoji_data.emojis) {
  278. emoji.code_point_array_index = code_point_array_index;
  279. code_point_array_index += emoji.code_points.size();
  280. }
  281. if (!generated_header_path.is_empty()) {
  282. auto generated_header_file = TRY(open_file(generated_header_path, Core::File::OpenMode::Write));
  283. TRY(generate_emoji_data_header(*generated_header_file, emoji_data));
  284. }
  285. if (!generated_implementation_path.is_empty()) {
  286. auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::File::OpenMode::Write));
  287. TRY(generate_emoji_data_implementation(*generated_implementation_file, emoji_data));
  288. }
  289. if (!generated_installation_path.is_empty()) {
  290. TRY(Core::Directory::create(LexicalPath { generated_installation_path }.parent(), Core::Directory::CreateDirectories::Yes));
  291. for (auto& emoji : emoji_data.emojis)
  292. set_image_path_for_emoji(emoji_resource_path, emoji);
  293. auto generated_installation_file = TRY(open_file(generated_installation_path, Core::File::OpenMode::Write));
  294. TRY(generate_emoji_installation(*generated_installation_file, emoji_data));
  295. }
  296. return 0;
  297. }