GenerateEmojiData.cpp 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428
  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/DirIterator.h>
  15. #include <LibCore/Directory.h>
  16. #include <LibUnicode/Emoji.h>
  17. struct Emoji {
  18. size_t name { 0 };
  19. Optional<size_t> 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, EmojiData& emoji_data, 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 file = DeprecatedString::formatted("{}.png", builder.to_deprecated_string());
  43. auto path = DeprecatedString::formatted("{}/{}", emoji_resource_path, file);
  44. if (!Core::DeprecatedFile::exists(path))
  45. return;
  46. emoji.image_path = emoji_data.unique_strings.ensure(move(file));
  47. }
  48. static ErrorOr<void> parse_emoji_test_data(Core::BufferedFile& file, EmojiData& emoji_data)
  49. {
  50. static constexpr auto group_header = "# group: "sv;
  51. static constexpr auto subgroup_header = "# subgroup: "sv;
  52. Array<u8, 1024> buffer;
  53. Unicode::EmojiGroup group;
  54. DeprecatedString subgroup;
  55. u32 display_order { 0 };
  56. while (TRY(file.can_read_line())) {
  57. auto line = TRY(file.read_line(buffer));
  58. if (line.is_empty())
  59. continue;
  60. if (line.starts_with('#')) {
  61. if (line.starts_with(group_header)) {
  62. auto name = line.substring_view(group_header.length());
  63. group = Unicode::emoji_group_from_string(name);
  64. } else if (line.starts_with(subgroup_header)) {
  65. subgroup = line.substring_view(subgroup_header.length());
  66. }
  67. continue;
  68. }
  69. auto status_index = line.find(';');
  70. VERIFY(status_index.has_value());
  71. auto emoji_and_name_index = line.find('#', *status_index);
  72. VERIFY(emoji_and_name_index.has_value());
  73. Emoji emoji {};
  74. emoji.group = group;
  75. emoji.subgroup = subgroup;
  76. emoji.display_order = display_order++;
  77. auto code_points = line.substring_view(0, *status_index).split_view(' ');
  78. TRY(emoji.code_points.try_ensure_capacity(code_points.size()));
  79. for (auto code_point : code_points) {
  80. auto value = AK::StringUtils::convert_to_uint_from_hex<u32>(code_point);
  81. VERIFY(value.has_value());
  82. emoji.code_points.unchecked_append(*value);
  83. }
  84. auto emoji_and_name = line.substring_view(*emoji_and_name_index + 1);
  85. auto emoji_and_name_spaces = emoji_and_name.find_all(" "sv);
  86. VERIFY(emoji_and_name_spaces.size() > 2);
  87. auto name = emoji_and_name.substring_view(emoji_and_name_spaces[2]).trim_whitespace();
  88. emoji.name = emoji_data.unique_strings.ensure(name.to_titlecase_string());
  89. emoji.encoded_code_points = emoji_and_name.substring_view(0, emoji_and_name_spaces[1]).trim_whitespace();
  90. emoji.status = line.substring_view(*status_index + 1, *emoji_and_name_index - *status_index - 1).trim_whitespace();
  91. TRY(emoji_data.emojis.try_append(move(emoji)));
  92. }
  93. return {};
  94. }
  95. static ErrorOr<void> parse_emoji_serenity_data(Core::BufferedFile& file, EmojiData& emoji_data)
  96. {
  97. static constexpr auto code_point_header = "U+"sv;
  98. Array<u8, 1024> buffer;
  99. auto display_order = static_cast<u32>(emoji_data.emojis.size()) + 1u;
  100. while (TRY(file.can_read_line())) {
  101. auto line = TRY(file.read_line(buffer));
  102. if (line.is_empty())
  103. continue;
  104. auto index = line.find(code_point_header);
  105. if (!index.has_value())
  106. continue;
  107. line = line.substring_view(*index);
  108. StringBuilder builder;
  109. Emoji emoji {};
  110. emoji.group = Unicode::EmojiGroup::SerenityOS;
  111. emoji.display_order = display_order++;
  112. TRY(line.for_each_split_view(' ', SplitBehavior::Nothing, [&](auto segment) -> ErrorOr<void> {
  113. if (segment.starts_with(code_point_header)) {
  114. segment = segment.substring_view(code_point_header.length());
  115. auto code_point = AK::StringUtils::convert_to_uint_from_hex<u32>(segment);
  116. VERIFY(code_point.has_value());
  117. TRY(emoji.code_points.try_append(*code_point));
  118. } else {
  119. if (!builder.is_empty())
  120. TRY(builder.try_append(' '));
  121. TRY(builder.try_append(segment));
  122. }
  123. return {};
  124. }));
  125. auto name = builder.to_deprecated_string();
  126. if (!any_of(name, is_ascii_lower_alpha))
  127. name = name.to_titlecase();
  128. emoji.name = emoji_data.unique_strings.ensure(move(name));
  129. TRY(emoji_data.emojis.try_append(move(emoji)));
  130. }
  131. return {};
  132. }
  133. static ErrorOr<void> validate_emoji(StringView emoji_resource_path, EmojiData& emoji_data)
  134. {
  135. Core::DirIterator iterator(emoji_resource_path, Core::DirIterator::SkipDots);
  136. while (iterator.has_next()) {
  137. auto filename = iterator.next_path();
  138. auto lexical_path = LexicalPath(filename);
  139. if (lexical_path.extension() != "png")
  140. continue;
  141. auto basename = lexical_path.basename();
  142. if (!basename.starts_with("U+"sv))
  143. continue;
  144. basename = basename.substring_view(0, basename.length() - lexical_path.extension().length() - 1);
  145. Vector<u32> code_points;
  146. TRY(basename.for_each_split_view('_', SplitBehavior::Nothing, [&](auto segment) -> ErrorOr<void> {
  147. auto code_point = AK::StringUtils::convert_to_uint_from_hex<u32>(segment.substring_view(2));
  148. VERIFY(code_point.has_value());
  149. TRY(code_points.try_append(*code_point));
  150. return {};
  151. }));
  152. auto it = emoji_data.emojis.find_if([&](auto const& emoji) {
  153. return emoji.code_points == code_points;
  154. });
  155. if (it == emoji_data.emojis.end()) {
  156. warnln("\x1b[1;31mError!\x1b[0m Emoji data for \x1b[35m{}\x1b[0m not found. Please check emoji-test.txt and emoji-serenity.txt.", filename);
  157. return Error::from_errno(ENOENT);
  158. }
  159. }
  160. return {};
  161. }
  162. static ErrorOr<void> generate_emoji_data_header(Core::BufferedFile& file, EmojiData const&)
  163. {
  164. StringBuilder builder;
  165. SourceGenerator generator { builder };
  166. TRY(file.write(generator.as_string_view().bytes()));
  167. return {};
  168. }
  169. static ErrorOr<void> generate_emoji_data_implementation(Core::BufferedFile& file, EmojiData const& emoji_data)
  170. {
  171. StringBuilder builder;
  172. SourceGenerator generator { builder };
  173. generator.set("string_index_type"sv, emoji_data.unique_strings.type_that_fits());
  174. generator.set("emojis_size"sv, DeprecatedString::number(emoji_data.emojis.size()));
  175. generator.append(R"~~~(
  176. #include <AK/Array.h>
  177. #include <AK/BinarySearch.h>
  178. #include <AK/Span.h>
  179. #include <AK/StringView.h>
  180. #include <AK/Types.h>
  181. #include <LibUnicode/Emoji.h>
  182. #include <LibUnicode/EmojiData.h>
  183. namespace Unicode {
  184. )~~~");
  185. emoji_data.unique_strings.generate(generator);
  186. size_t total_code_point_count { 0 };
  187. for (auto const& emoji : emoji_data.emojis) {
  188. total_code_point_count += emoji.code_points.size();
  189. }
  190. generator.set("total_code_point_count", DeprecatedString::number(total_code_point_count));
  191. generator.append(R"~~~(
  192. static constexpr Array<u32, @total_code_point_count@> s_emoji_code_points { {)~~~");
  193. bool first = true;
  194. for (auto const& emoji : emoji_data.emojis) {
  195. for (auto code_point : emoji.code_points) {
  196. generator.append(first ? " "sv : ", "sv);
  197. generator.append(DeprecatedString::formatted("{:#x}", code_point));
  198. first = false;
  199. }
  200. }
  201. generator.append(" } };"sv);
  202. generator.append(R"~~~(
  203. struct EmojiData {
  204. Emoji to_unicode_emoji() const
  205. {
  206. Emoji emoji {};
  207. emoji.name = decode_string(name);
  208. if (image_path != 0)
  209. emoji.image_path = decode_string(image_path);
  210. emoji.group = static_cast<EmojiGroup>(group);
  211. emoji.display_order = display_order;
  212. emoji.code_points = code_points();
  213. return emoji;
  214. }
  215. constexpr ReadonlySpan<u32> code_points() const
  216. {
  217. return ReadonlySpan<u32>(s_emoji_code_points.data() + code_point_start, code_point_count);
  218. }
  219. @string_index_type@ name { 0 };
  220. @string_index_type@ image_path { 0 };
  221. u8 group { 0 };
  222. u32 display_order { 0 };
  223. size_t code_point_start { 0 };
  224. size_t code_point_count { 0 };
  225. };
  226. )~~~");
  227. generator.append(R"~~~(
  228. static constexpr Array<EmojiData, @emojis_size@> s_emojis { {)~~~");
  229. for (auto const& emoji : emoji_data.emojis) {
  230. generator.set("name"sv, DeprecatedString::number(emoji.name));
  231. generator.set("image_path"sv, DeprecatedString::number(emoji.image_path.value_or(0)));
  232. generator.set("group"sv, DeprecatedString::number(to_underlying(emoji.group)));
  233. generator.set("display_order"sv, DeprecatedString::number(emoji.display_order));
  234. generator.set("code_point_start"sv, DeprecatedString::number(emoji.code_point_array_index));
  235. generator.set("code_point_count"sv, DeprecatedString::number(emoji.code_points.size()));
  236. generator.append(R"~~~(
  237. { @name@, @image_path@, @group@, @display_order@, @code_point_start@, @code_point_count@ },)~~~");
  238. }
  239. generator.append(R"~~~(
  240. } };
  241. Optional<Emoji> find_emoji_for_code_points(ReadonlySpan<u32> code_points)
  242. {
  243. for (auto& emoji : s_emojis) {
  244. if (emoji.code_points() == code_points)
  245. return emoji.to_unicode_emoji();
  246. }
  247. return {};
  248. }
  249. }
  250. )~~~");
  251. TRY(file.write(generator.as_string_view().bytes()));
  252. return {};
  253. }
  254. static ErrorOr<void> generate_emoji_installation(Core::BufferedFile& file, EmojiData const& emoji_data)
  255. {
  256. StringBuilder builder;
  257. SourceGenerator generator { builder };
  258. auto current_group = Unicode::EmojiGroup::Unknown;
  259. StringView current_subgroup;
  260. for (auto const& emoji : emoji_data.emojis) {
  261. if (!emoji.image_path.has_value())
  262. continue;
  263. if (emoji.group == Unicode::EmojiGroup::SerenityOS)
  264. continue; // SerenityOS emojis are in emoji-serenity.txt
  265. if (current_group != emoji.group) {
  266. if (!builder.is_empty())
  267. generator.append("\n"sv);
  268. generator.set("group"sv, Unicode::emoji_group_to_string(emoji.group));
  269. generator.append("# group: @group@\n");
  270. current_group = emoji.group;
  271. }
  272. if (current_subgroup != emoji.subgroup) {
  273. generator.set("subgroup"sv, emoji.subgroup);
  274. generator.append("\n# subgroup: @subgroup@\n");
  275. current_subgroup = emoji.subgroup;
  276. }
  277. generator.set("emoji"sv, emoji.encoded_code_points);
  278. generator.set("name"sv, emoji_data.unique_strings.get(emoji.name));
  279. generator.set("status"sv, emoji.status);
  280. generator.append("@emoji@"sv);
  281. generator.append(" - "sv);
  282. generator.append(DeprecatedString::join(" "sv, emoji.code_points, "U+{:X}"sv));
  283. generator.append(" @name@ (@status@)\n"sv);
  284. }
  285. TRY(file.write(generator.as_string_view().bytes()));
  286. return {};
  287. }
  288. ErrorOr<int> serenity_main(Main::Arguments arguments)
  289. {
  290. StringView generated_header_path;
  291. StringView generated_implementation_path;
  292. StringView generated_installation_path;
  293. StringView emoji_test_path;
  294. StringView emoji_serenity_path;
  295. StringView emoji_resource_path;
  296. Core::ArgsParser args_parser;
  297. args_parser.add_option(generated_header_path, "Path to the Unicode Data header file to generate", "generated-header-path", 'h', "generated-header-path");
  298. args_parser.add_option(generated_implementation_path, "Path to the Unicode Data implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  299. args_parser.add_option(generated_installation_path, "Path to the emoji.txt file to generate", "generated-installation-path", 'i', "generated-installation-path");
  300. args_parser.add_option(emoji_test_path, "Path to emoji-test.txt file", "emoji-test-path", 'e', "emoji-test-path");
  301. args_parser.add_option(emoji_serenity_path, "Path to emoji-serenity.txt file", "emoji-serenity-path", 's', "emoji-serenity-path");
  302. args_parser.add_option(emoji_resource_path, "Path to the /res/emoji directory", "emoji-resource-path", 'r', "emoji-resource-path");
  303. args_parser.parse(arguments);
  304. VERIFY(!emoji_resource_path.is_empty() && Core::DeprecatedFile::exists(emoji_resource_path));
  305. auto emoji_test_file = TRY(open_file(emoji_test_path, Core::File::OpenMode::Read));
  306. EmojiData emoji_data {};
  307. TRY(parse_emoji_test_data(*emoji_test_file, emoji_data));
  308. if (!emoji_serenity_path.is_empty()) {
  309. auto emoji_serenity_file = TRY(open_file(emoji_serenity_path, Core::File::OpenMode::Read));
  310. TRY(parse_emoji_serenity_data(*emoji_serenity_file, emoji_data));
  311. TRY(validate_emoji(emoji_resource_path, emoji_data));
  312. }
  313. size_t code_point_array_index { 0 };
  314. for (auto& emoji : emoji_data.emojis) {
  315. emoji.code_point_array_index = code_point_array_index;
  316. code_point_array_index += emoji.code_points.size();
  317. set_image_path_for_emoji(emoji_resource_path, emoji_data, emoji);
  318. }
  319. if (!generated_header_path.is_empty()) {
  320. auto generated_header_file = TRY(open_file(generated_header_path, Core::File::OpenMode::Write));
  321. TRY(generate_emoji_data_header(*generated_header_file, emoji_data));
  322. }
  323. if (!generated_implementation_path.is_empty()) {
  324. auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::File::OpenMode::Write));
  325. TRY(generate_emoji_data_implementation(*generated_implementation_file, emoji_data));
  326. }
  327. if (!generated_installation_path.is_empty()) {
  328. TRY(Core::Directory::create(LexicalPath { generated_installation_path }.parent(), Core::Directory::CreateDirectories::Yes));
  329. auto generated_installation_file = TRY(open_file(generated_installation_path, Core::File::OpenMode::Write));
  330. TRY(generate_emoji_installation(*generated_installation_file, emoji_data));
  331. }
  332. return 0;
  333. }