GeneratorUtil.h 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344
  1. /*
  2. * Copyright (c) 2021, Tim Flynn <trflynn89@pm.me>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #pragma once
  7. #include <AK/Format.h>
  8. #include <AK/HashMap.h>
  9. #include <AK/JsonArray.h>
  10. #include <AK/JsonObject.h>
  11. #include <AK/JsonParser.h>
  12. #include <AK/JsonValue.h>
  13. #include <AK/LexicalPath.h>
  14. #include <AK/Optional.h>
  15. #include <AK/QuickSort.h>
  16. #include <AK/SourceGenerator.h>
  17. #include <AK/String.h>
  18. #include <AK/StringView.h>
  19. #include <AK/Vector.h>
  20. #include <LibCore/DirIterator.h>
  21. #include <LibCore/File.h>
  22. #include <LibUnicode/Locale.h>
  23. template<typename StringIndexType>
  24. class UniqueStringStorage {
  25. public:
  26. StringIndexType ensure(String string)
  27. {
  28. // We maintain a set of unique strings in two structures: a vector which owns the unique string,
  29. // and a hash map which maps that string to its index in the vector. The vector is to ensure the
  30. // strings are generated in an easily known order, and the map is to allow quickly deciding if a
  31. // string is actually unique (otherwise, we'd have to linear-search the vector for each string).
  32. //
  33. // Also note that index 0 will be reserved for the empty string, so the index returned from this
  34. // method is actually the real index in the vector + 1.
  35. if (auto index = m_unique_string_indices.get(string); index.has_value())
  36. return *index;
  37. m_unique_strings.append(move(string));
  38. size_t index = m_unique_strings.size();
  39. VERIFY(index < NumericLimits<StringIndexType>::max());
  40. auto string_index = static_cast<StringIndexType>(index);
  41. m_unique_string_indices.set(m_unique_strings.last(), string_index);
  42. return string_index;
  43. }
  44. StringView get(StringIndexType index) const
  45. {
  46. if (index == 0)
  47. return {};
  48. VERIFY(index <= m_unique_strings.size());
  49. return m_unique_strings.at(index - 1);
  50. }
  51. void generate(SourceGenerator& generator)
  52. {
  53. generator.set("size"sv, String::number(m_unique_strings.size()));
  54. generator.append(R"~~~(
  55. static constexpr Array<StringView, @size@ + 1> s_string_list { {
  56. {})~~~");
  57. constexpr size_t max_strings_per_row = 40;
  58. size_t strings_in_current_row = 1;
  59. for (auto const& string : m_unique_strings) {
  60. if (strings_in_current_row++ > 0)
  61. generator.append(", ");
  62. generator.append(String::formatted("\"{}\"sv", string));
  63. if (strings_in_current_row == max_strings_per_row) {
  64. strings_in_current_row = 0;
  65. generator.append(",\n ");
  66. }
  67. }
  68. generator.append(R"~~~(
  69. } };
  70. )~~~");
  71. }
  72. private:
  73. Vector<String> m_unique_strings;
  74. HashMap<StringView, StringIndexType> m_unique_string_indices;
  75. };
  76. template<typename StringIndexType>
  77. struct CanonicalLanguageID {
  78. static Optional<CanonicalLanguageID> parse(UniqueStringStorage<StringIndexType>& unique_strings, StringView language)
  79. {
  80. CanonicalLanguageID language_id {};
  81. auto segments = language.split_view('-');
  82. VERIFY(!segments.is_empty());
  83. size_t index = 0;
  84. if (Unicode::is_unicode_language_subtag(segments[index])) {
  85. language_id.language = unique_strings.ensure(segments[index]);
  86. if (segments.size() == ++index)
  87. return language_id;
  88. } else {
  89. return {};
  90. }
  91. if (Unicode::is_unicode_script_subtag(segments[index])) {
  92. language_id.script = unique_strings.ensure(segments[index]);
  93. if (segments.size() == ++index)
  94. return language_id;
  95. }
  96. if (Unicode::is_unicode_region_subtag(segments[index])) {
  97. language_id.region = unique_strings.ensure(segments[index]);
  98. if (segments.size() == ++index)
  99. return language_id;
  100. }
  101. while (index < segments.size()) {
  102. if (!Unicode::is_unicode_variant_subtag(segments[index]))
  103. return {};
  104. language_id.variants.append(unique_strings.ensure(segments[index++]));
  105. }
  106. return language_id;
  107. }
  108. StringIndexType language { 0 };
  109. StringIndexType script { 0 };
  110. StringIndexType region { 0 };
  111. Vector<StringIndexType> variants {};
  112. };
  113. inline Core::DirIterator path_to_dir_iterator(String path)
  114. {
  115. LexicalPath lexical_path(move(path));
  116. lexical_path = lexical_path.append("main"sv);
  117. VERIFY(Core::File::is_directory(lexical_path.string()));
  118. Core::DirIterator iterator(lexical_path.string(), Core::DirIterator::SkipParentAndBaseDir);
  119. if (iterator.has_error()) {
  120. warnln("{}: {}", lexical_path.string(), iterator.error_string());
  121. VERIFY_NOT_REACHED();
  122. }
  123. return iterator;
  124. }
  125. template<typename LocaleDataType>
  126. void parse_default_content_locales(String core_path, LocaleDataType& locale_data)
  127. {
  128. LexicalPath default_content_path(move(core_path));
  129. default_content_path = default_content_path.append("defaultContent.json"sv);
  130. VERIFY(Core::File::exists(default_content_path.string()));
  131. auto default_content_file_or_error = Core::File::open(default_content_path.string(), Core::OpenMode::ReadOnly);
  132. VERIFY(!default_content_file_or_error.is_error());
  133. auto default_content = JsonValue::from_string(default_content_file_or_error.value()->read_all()).release_value_but_fixme_should_propagate_errors();
  134. auto const& default_content_array = default_content.as_object().get("defaultContent"sv);
  135. default_content_array.as_array().for_each([&](JsonValue const& value) {
  136. auto locale = value.as_string();
  137. StringView default_locale = locale;
  138. while (true) {
  139. if (locale_data.locales.contains(default_locale))
  140. break;
  141. auto pos = default_locale.find_last('-');
  142. if (!pos.has_value())
  143. return;
  144. default_locale = default_locale.substring_view(0, *pos);
  145. }
  146. locale_data.locales.set(locale, locale_data.locales.get(default_locale).value());
  147. });
  148. }
  149. inline void ensure_from_string_types_are_generated(SourceGenerator& generator)
  150. {
  151. static bool generated_from_string_types = false;
  152. if (generated_from_string_types)
  153. return;
  154. generator.append(R"~~~(
  155. template <typename ValueType>
  156. struct HashValuePair {
  157. unsigned hash { 0 };
  158. ValueType value {};
  159. };
  160. template <typename ValueType>
  161. struct HashValueComparator
  162. {
  163. constexpr int operator()(unsigned hash, HashValuePair<ValueType> const& pair)
  164. {
  165. if (hash > pair.hash)
  166. return 1;
  167. if (hash < pair.hash)
  168. return -1;
  169. return 0;
  170. }
  171. };
  172. )~~~");
  173. generated_from_string_types = true;
  174. }
  175. template<typename ValueType>
  176. using HashValueMap = HashMap<unsigned, ValueType>;
  177. template<typename ValueType>
  178. void generate_value_from_string(SourceGenerator& generator, StringView method_name_format, StringView value_type, StringView value_name, HashValueMap<ValueType> hashes, Optional<StringView> return_type = {}, StringView return_format = "{}"sv)
  179. {
  180. ensure_from_string_types_are_generated(generator);
  181. generator.set("method_name", String::formatted(method_name_format, value_name));
  182. generator.set("value_type", value_type);
  183. generator.set("value_name", value_name);
  184. generator.set("return_type", return_type.has_value() ? *return_type : value_type);
  185. generator.set("size", String::number(hashes.size()));
  186. generator.append(R"~~~(
  187. Optional<@return_type@> @method_name@(StringView key)
  188. {
  189. constexpr Array<HashValuePair<@value_type@>, @size@> hash_pairs { {
  190. )~~~");
  191. auto hash_keys = hashes.keys();
  192. quick_sort(hash_keys);
  193. constexpr size_t max_values_per_row = 10;
  194. size_t values_in_current_row = 0;
  195. for (auto hash_key : hash_keys) {
  196. if (values_in_current_row++ > 0)
  197. generator.append(" ");
  198. if constexpr (IsIntegral<ValueType>)
  199. generator.set("value"sv, String::number(hashes.get(hash_key).value()));
  200. else
  201. generator.set("value"sv, String::formatted("{}::{}", value_type, hashes.get(hash_key).value()));
  202. generator.set("hash"sv, String::number(hash_key));
  203. generator.append("{ @hash@U, @value@ },"sv);
  204. if (values_in_current_row == max_values_per_row) {
  205. generator.append("\n ");
  206. values_in_current_row = 0;
  207. }
  208. }
  209. generator.set("return_statement", String::formatted(return_format, "value->value"sv));
  210. generator.append(R"~~~(
  211. } };
  212. if (auto const* value = binary_search(hash_pairs, key.hash(), nullptr, HashValueComparator<@value_type@> {}))
  213. return @return_statement@;
  214. return {};
  215. }
  216. )~~~");
  217. }
  218. template<typename IdentifierFormatter>
  219. void generate_enum(SourceGenerator& generator, IdentifierFormatter&& format_identifier, StringView name, StringView default_, Vector<String>& values)
  220. {
  221. quick_sort(values);
  222. generator.set("name", name);
  223. generator.set("underlying", ((values.size() + !default_.is_empty()) < 256) ? "u8"sv : "u16"sv);
  224. generator.append(R"~~~(
  225. enum class @name@ : @underlying@ {)~~~");
  226. if (!default_.is_empty()) {
  227. generator.set("default", default_);
  228. generator.append(R"~~~(
  229. @default@,)~~~");
  230. }
  231. for (auto const& value : values) {
  232. generator.set("value", format_identifier(name, value));
  233. generator.append(R"~~~(
  234. @value@,)~~~");
  235. }
  236. generator.append(R"~~~(
  237. };
  238. )~~~");
  239. }
  240. template<typename LocalesType, typename ListFormatter>
  241. void generate_mapping(SourceGenerator& generator, LocalesType const& locales, StringView type, StringView name, StringView format, ListFormatter&& format_list)
  242. {
  243. auto format_mapping_name = [](StringView format, StringView name) {
  244. auto mapping_name = name.to_lowercase_string().replace("-"sv, "_"sv, true);
  245. return String::formatted(format, mapping_name);
  246. };
  247. Vector<String> mapping_names;
  248. for (auto const& locale : locales) {
  249. auto mapping_name = format_mapping_name(format, locale.key);
  250. format_list(mapping_name, locale.value);
  251. mapping_names.append(move(mapping_name));
  252. }
  253. quick_sort(mapping_names);
  254. generator.set("type", type);
  255. generator.set("name", name);
  256. generator.set("size", String::number(locales.size()));
  257. generator.append(R"~~~(
  258. static constexpr Array<Span<@type@ const>, @size@> @name@ { {
  259. )~~~");
  260. constexpr size_t max_values_per_row = 10;
  261. size_t values_in_current_row = 0;
  262. for (auto& mapping_name : mapping_names) {
  263. if (values_in_current_row++ > 0)
  264. generator.append(" ");
  265. generator.set("name", move(mapping_name));
  266. generator.append("@name@.span(),");
  267. if (values_in_current_row == max_values_per_row) {
  268. values_in_current_row = 0;
  269. generator.append("\n ");
  270. }
  271. }
  272. generator.append(R"~~~(
  273. } };
  274. )~~~");
  275. }