GeneratorUtil.h 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327
  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/LexicalPath.h>
  10. #include <AK/Optional.h>
  11. #include <AK/QuickSort.h>
  12. #include <AK/SourceGenerator.h>
  13. #include <AK/String.h>
  14. #include <AK/StringView.h>
  15. #include <AK/Vector.h>
  16. #include <LibCore/DirIterator.h>
  17. #include <LibCore/File.h>
  18. #include <LibUnicode/Locale.h>
  19. template<typename StringIndexType>
  20. class UniqueStringStorage {
  21. public:
  22. StringIndexType ensure(String string)
  23. {
  24. // We maintain a set of unique strings in two structures: a vector which owns the unique string,
  25. // and a hash map which maps that string to its index in the vector. The vector is to ensure the
  26. // strings are generated in an easily known order, and the map is to allow quickly deciding if a
  27. // string is actually unique (otherwise, we'd have to linear-search the vector for each string).
  28. //
  29. // Also note that index 0 will be reserved for the empty string, so the index returned from this
  30. // method is actually the real index in the vector + 1.
  31. if (auto index = m_unique_string_indices.get(string); index.has_value())
  32. return *index;
  33. m_unique_strings.append(move(string));
  34. size_t index = m_unique_strings.size();
  35. VERIFY(index < NumericLimits<StringIndexType>::max());
  36. auto string_index = static_cast<StringIndexType>(index);
  37. m_unique_string_indices.set(m_unique_strings.last(), string_index);
  38. return string_index;
  39. }
  40. StringView get(StringIndexType index) const
  41. {
  42. if (index == 0)
  43. return {};
  44. VERIFY(index <= m_unique_strings.size());
  45. return m_unique_strings.at(index - 1);
  46. }
  47. void generate(SourceGenerator& generator)
  48. {
  49. generator.set("size"sv, String::number(m_unique_strings.size()));
  50. generator.append(R"~~~(
  51. static constexpr Array<StringView, @size@ + 1> s_string_list { {
  52. {})~~~");
  53. constexpr size_t max_strings_per_row = 40;
  54. size_t strings_in_current_row = 1;
  55. for (auto const& string : m_unique_strings) {
  56. if (strings_in_current_row++ > 0)
  57. generator.append(", ");
  58. generator.append(String::formatted("\"{}\"sv", string));
  59. if (strings_in_current_row == max_strings_per_row) {
  60. strings_in_current_row = 0;
  61. generator.append(",\n ");
  62. }
  63. }
  64. generator.append(R"~~~(
  65. } };
  66. )~~~");
  67. }
  68. private:
  69. Vector<String> m_unique_strings;
  70. HashMap<StringView, StringIndexType> m_unique_string_indices;
  71. };
  72. struct Alias {
  73. String name;
  74. String alias;
  75. };
  76. template<typename StringIndexType>
  77. struct CanonicalLanguageID {
  78. static ErrorOr<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 Error::from_string_literal("Expected language subtag"sv);
  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 Error::from_string_literal("Expected variant subtag"sv);
  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 ErrorOr<Core::DirIterator> path_to_dir_iterator(String path, StringView subpath = "main"sv)
  114. {
  115. LexicalPath lexical_path(move(path));
  116. if (!subpath.is_empty())
  117. lexical_path = lexical_path.append(subpath);
  118. Core::DirIterator iterator(lexical_path.string(), Core::DirIterator::SkipParentAndBaseDir);
  119. if (iterator.has_error())
  120. return Error::from_string_literal(iterator.error_string());
  121. return iterator;
  122. }
  123. inline ErrorOr<String> next_path_from_dir_iterator(Core::DirIterator& iterator)
  124. {
  125. auto next_path = iterator.next_full_path();
  126. if (iterator.has_error())
  127. return Error::from_string_literal(iterator.error_string());
  128. return next_path;
  129. }
  130. inline void ensure_from_string_types_are_generated(SourceGenerator& generator)
  131. {
  132. static bool generated_from_string_types = false;
  133. if (generated_from_string_types)
  134. return;
  135. generator.append(R"~~~(
  136. template <typename ValueType>
  137. struct HashValuePair {
  138. unsigned hash { 0 };
  139. ValueType value {};
  140. };
  141. template <typename ValueType>
  142. struct HashValueComparator
  143. {
  144. constexpr int operator()(unsigned hash, HashValuePair<ValueType> const& pair)
  145. {
  146. if (hash > pair.hash)
  147. return 1;
  148. if (hash < pair.hash)
  149. return -1;
  150. return 0;
  151. }
  152. };
  153. )~~~");
  154. generated_from_string_types = true;
  155. }
  156. template<typename ValueType>
  157. using HashValueMap = HashMap<unsigned, ValueType>;
  158. template<typename ValueType>
  159. 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)
  160. {
  161. ensure_from_string_types_are_generated(generator);
  162. generator.set("method_name", String::formatted(method_name_format, value_name));
  163. generator.set("value_type", value_type);
  164. generator.set("value_name", value_name);
  165. generator.set("return_type", return_type.has_value() ? *return_type : value_type);
  166. generator.set("size", String::number(hashes.size()));
  167. generator.append(R"~~~(
  168. Optional<@return_type@> @method_name@(StringView key)
  169. {
  170. constexpr Array<HashValuePair<@value_type@>, @size@> hash_pairs { {
  171. )~~~");
  172. auto hash_keys = hashes.keys();
  173. quick_sort(hash_keys);
  174. constexpr size_t max_values_per_row = 10;
  175. size_t values_in_current_row = 0;
  176. for (auto hash_key : hash_keys) {
  177. if (values_in_current_row++ > 0)
  178. generator.append(" ");
  179. if constexpr (IsIntegral<ValueType>)
  180. generator.set("value"sv, String::number(hashes.get(hash_key).value()));
  181. else
  182. generator.set("value"sv, String::formatted("{}::{}", value_type, hashes.get(hash_key).value()));
  183. generator.set("hash"sv, String::number(hash_key));
  184. generator.append("{ @hash@U, @value@ },"sv);
  185. if (values_in_current_row == max_values_per_row) {
  186. generator.append("\n ");
  187. values_in_current_row = 0;
  188. }
  189. }
  190. generator.set("return_statement", String::formatted(return_format, "value->value"sv));
  191. generator.append(R"~~~(
  192. } };
  193. if (auto const* value = binary_search(hash_pairs, key.hash(), nullptr, HashValueComparator<@value_type@> {}))
  194. return @return_statement@;
  195. return {};
  196. }
  197. )~~~");
  198. }
  199. template<typename IdentifierFormatter>
  200. void generate_enum(SourceGenerator& generator, IdentifierFormatter&& format_identifier, StringView name, StringView default_, Vector<String>& values, Vector<Alias> aliases = {})
  201. {
  202. quick_sort(values, [](auto const& value1, auto const& value2) { return value1.to_lowercase() < value2.to_lowercase(); });
  203. quick_sort(aliases, [](auto const& alias1, auto const& alias2) { return alias1.alias.to_lowercase() < alias2.alias.to_lowercase(); });
  204. generator.set("name", name);
  205. generator.set("underlying", ((values.size() + !default_.is_empty()) < 256) ? "u8"sv : "u16"sv);
  206. generator.append(R"~~~(
  207. enum class @name@ : @underlying@ {)~~~");
  208. if (!default_.is_empty()) {
  209. generator.set("default", default_);
  210. generator.append(R"~~~(
  211. @default@,)~~~");
  212. }
  213. for (auto const& value : values) {
  214. generator.set("value", format_identifier(name, value));
  215. generator.append(R"~~~(
  216. @value@,)~~~");
  217. }
  218. for (auto const& alias : aliases) {
  219. generator.set("alias", format_identifier(name, alias.alias));
  220. generator.set("value", format_identifier(name, alias.name));
  221. generator.append(R"~~~(
  222. @alias@ = @value@,)~~~");
  223. }
  224. generator.append(R"~~~(
  225. };
  226. )~~~");
  227. }
  228. template<typename LocalesType, typename ListFormatter>
  229. void generate_mapping(SourceGenerator& generator, LocalesType const& locales, StringView type, StringView name, StringView format, ListFormatter&& format_list)
  230. {
  231. auto format_mapping_name = [](StringView format, StringView name) {
  232. auto mapping_name = name.to_lowercase_string().replace("-"sv, "_"sv, true);
  233. return String::formatted(format, mapping_name);
  234. };
  235. Vector<String> mapping_names;
  236. for (auto const& locale : locales) {
  237. auto mapping_name = format_mapping_name(format, locale.key);
  238. format_list(mapping_name, locale.value);
  239. mapping_names.append(move(mapping_name));
  240. }
  241. quick_sort(mapping_names);
  242. generator.set("type", type);
  243. generator.set("name", name);
  244. generator.set("size", String::number(locales.size()));
  245. generator.append(R"~~~(
  246. static constexpr Array<Span<@type@ const>, @size@> @name@ { {
  247. )~~~");
  248. constexpr size_t max_values_per_row = 10;
  249. size_t values_in_current_row = 0;
  250. for (auto& mapping_name : mapping_names) {
  251. if (values_in_current_row++ > 0)
  252. generator.append(" ");
  253. generator.set("name", move(mapping_name));
  254. generator.append("@name@.span(),");
  255. if (values_in_current_row == max_values_per_row) {
  256. values_in_current_row = 0;
  257. generator.append("\n ");
  258. }
  259. }
  260. generator.append(R"~~~(
  261. } };
  262. )~~~");
  263. }