GeneratorUtil.h 20 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631
  1. /*
  2. * Copyright (c) 2021, Tim Flynn <trflynn89@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #pragma once
  7. #include <AK/DeprecatedString.h>
  8. #include <AK/Function.h>
  9. #include <AK/HashFunctions.h>
  10. #include <AK/HashMap.h>
  11. #include <AK/JsonValue.h>
  12. #include <AK/LexicalPath.h>
  13. #include <AK/NumericLimits.h>
  14. #include <AK/Optional.h>
  15. #include <AK/QuickSort.h>
  16. #include <AK/SourceGenerator.h>
  17. #include <AK/StringBuilder.h>
  18. #include <AK/StringView.h>
  19. #include <AK/Traits.h>
  20. #include <AK/Vector.h>
  21. #include <LibCore/DirIterator.h>
  22. #include <LibCore/File.h>
  23. #include <LibLocale/Locale.h>
  24. template<class T>
  25. inline constexpr bool StorageTypeIsList = false;
  26. template<class T>
  27. inline constexpr bool StorageTypeIsList<Vector<T>> = true;
  28. template<typename T>
  29. concept IntegralOrEnum = Integral<T> || Enum<T>;
  30. template<IntegralOrEnum T>
  31. struct AK::Traits<Vector<T>> : public GenericTraits<Vector<T>> {
  32. static unsigned hash(Vector<T> const& list)
  33. {
  34. auto hash = int_hash(static_cast<u32>(list.size()));
  35. for (auto value : list) {
  36. if constexpr (Enum<T>)
  37. hash = pair_int_hash(hash, to_underlying(value));
  38. else
  39. hash = pair_int_hash(hash, value);
  40. }
  41. return hash;
  42. }
  43. };
  44. template<typename StorageType>
  45. class UniqueStorage {
  46. public:
  47. size_t ensure(StorageType value)
  48. {
  49. // We maintain a set of unique values in two structures: a vector which stores the values in
  50. // the order they are added, and a hash map which maps that value to its index in the vector.
  51. // The vector is to ensure the values are generated in an easily known order, and the map is
  52. // to allow quickly deciding if a value is actually unique (otherwise, we'd have to linearly
  53. // search the vector for each value).
  54. //
  55. // Also note that index 0 will be reserved for the default-initialized value, so the index
  56. // returned from this method is actually the real index in the vector + 1.
  57. if (auto index = m_storage_indices.get(value); index.has_value())
  58. return *index;
  59. m_storage.append(move(value));
  60. auto storage_index = m_storage.size();
  61. m_storage_indices.set(m_storage.last(), storage_index);
  62. return storage_index;
  63. }
  64. StorageType const& get(size_t index) const
  65. {
  66. if (index == 0) {
  67. static StorageType empty {};
  68. return empty;
  69. }
  70. VERIFY(index <= m_storage.size());
  71. return m_storage.at(index - 1);
  72. }
  73. StringView type_that_fits() const
  74. {
  75. if (m_storage.size() <= NumericLimits<u8>::max())
  76. return "u8"sv;
  77. if (m_storage.size() <= NumericLimits<u16>::max())
  78. return "u16"sv;
  79. if (m_storage.size() <= NumericLimits<u32>::max())
  80. return "u32"sv;
  81. return "u64"sv;
  82. }
  83. void generate(SourceGenerator& generator, StringView type, StringView name, size_t max_values_per_row)
  84. requires(!StorageTypeIsList<StorageType>)
  85. {
  86. generator.set("type"sv, type);
  87. generator.set("name"sv, name);
  88. generator.set("size"sv, DeprecatedString::number(m_storage.size()));
  89. generator.append(R"~~~(
  90. static constexpr Array<@type@, @size@ + 1> @name@ { {
  91. {})~~~");
  92. size_t values_in_current_row = 1;
  93. for (auto const& value : m_storage) {
  94. if (values_in_current_row++ > 0)
  95. generator.append(", ");
  96. if constexpr (IsSame<StorageType, DeprecatedString>)
  97. generator.append(DeprecatedString::formatted("\"{}\"sv", value));
  98. else
  99. generator.append(DeprecatedString::formatted("{}", value));
  100. if (values_in_current_row == max_values_per_row) {
  101. values_in_current_row = 0;
  102. generator.append(",\n ");
  103. }
  104. }
  105. generator.append(R"~~~(
  106. } };
  107. )~~~");
  108. }
  109. void generate(SourceGenerator& generator, StringView type, StringView name)
  110. requires(StorageTypeIsList<StorageType>)
  111. {
  112. generator.set("type"sv, type);
  113. generator.set("name"sv, name);
  114. for (size_t i = 0; i < m_storage.size(); ++i) {
  115. auto const& list = m_storage[i];
  116. generator.set("index"sv, DeprecatedString::number(i));
  117. generator.set("size"sv, DeprecatedString::number(list.size()));
  118. generator.append(R"~~~(
  119. static constexpr Array<@type@, @size@> @name@@index@ { {)~~~");
  120. bool first = true;
  121. for (auto const& value : list) {
  122. generator.append(first ? " "sv : ", "sv);
  123. generator.append(DeprecatedString::formatted("{}", value));
  124. first = false;
  125. }
  126. generator.append(" } };");
  127. }
  128. generator.set("size"sv, DeprecatedString::number(m_storage.size()));
  129. generator.append(R"~~~(
  130. static constexpr Array<ReadonlySpan<@type@>, @size@ + 1> @name@ { {
  131. {})~~~");
  132. constexpr size_t max_values_per_row = 10;
  133. size_t values_in_current_row = 1;
  134. for (size_t i = 0; i < m_storage.size(); ++i) {
  135. if (values_in_current_row++ > 0)
  136. generator.append(", ");
  137. generator.set("index"sv, DeprecatedString::number(i));
  138. generator.append("@name@@index@.span()");
  139. if (values_in_current_row == max_values_per_row) {
  140. values_in_current_row = 0;
  141. generator.append(",\n ");
  142. }
  143. }
  144. generator.append(R"~~~(
  145. } };
  146. )~~~");
  147. }
  148. protected:
  149. Vector<StorageType> m_storage;
  150. HashMap<StorageType, size_t> m_storage_indices;
  151. };
  152. class UniqueStringStorage : public UniqueStorage<DeprecatedString> {
  153. using Base = UniqueStorage<DeprecatedString>;
  154. public:
  155. // The goal of the string table generator is to ensure the table is located within the read-only
  156. // section of the shared library. If StringViews are generated directly, the table will be located
  157. // in the initialized data section. So instead, we generate run-length encoded (RLE) arrays to
  158. // represent the strings.
  159. void generate(SourceGenerator& generator) const
  160. {
  161. constexpr size_t max_values_per_row = 300;
  162. size_t values_in_current_row = 0;
  163. auto append_hex_value = [&](auto value) {
  164. if (values_in_current_row++ > 0)
  165. generator.append(", ");
  166. generator.append(DeprecatedString::formatted("{:#x}", value));
  167. if (values_in_current_row == max_values_per_row) {
  168. values_in_current_row = 0;
  169. generator.append(",\n ");
  170. }
  171. };
  172. Vector<u32> string_indices;
  173. string_indices.ensure_capacity(Base::m_storage.size());
  174. u32 next_index { 0 };
  175. for (auto const& string : Base::m_storage) {
  176. // Ensure the string length may be encoded as two u8s.
  177. VERIFY(string.length() <= NumericLimits<u16>::max());
  178. string_indices.unchecked_append(next_index);
  179. next_index += string.length() + 2;
  180. }
  181. generator.set("size", DeprecatedString::number(next_index));
  182. generator.append(R"~~~(
  183. static constexpr Array<u8, @size@> s_encoded_strings { {
  184. )~~~");
  185. for (auto const& string : Base::m_storage) {
  186. auto length = string.length();
  187. append_hex_value((length & 0xff00) >> 8);
  188. append_hex_value(length & 0x00ff);
  189. for (auto ch : string)
  190. append_hex_value(static_cast<u8>(ch));
  191. }
  192. generator.append(R"~~~(
  193. } };
  194. )~~~");
  195. generator.set("size", DeprecatedString::number(string_indices.size()));
  196. generator.append(R"~~~(
  197. static constexpr Array<u32, @size@> s_encoded_string_indices { {
  198. )~~~");
  199. values_in_current_row = 0;
  200. for (auto index : string_indices)
  201. append_hex_value(index);
  202. generator.append(R"~~~(
  203. } };
  204. static constexpr StringView decode_string(size_t index)
  205. {
  206. if (index == 0)
  207. return {};
  208. index = s_encoded_string_indices[index - 1];
  209. auto length_high = s_encoded_strings[index];
  210. auto length_low = s_encoded_strings[index + 1];
  211. size_t length = (length_high << 8) | length_low;
  212. if (length == 0)
  213. return {};
  214. auto const* start = &s_encoded_strings[index + 2];
  215. return { reinterpret_cast<char const*>(start), length };
  216. }
  217. )~~~");
  218. }
  219. };
  220. struct Alias {
  221. DeprecatedString name;
  222. DeprecatedString alias;
  223. };
  224. struct CanonicalLanguageID {
  225. static ErrorOr<CanonicalLanguageID> parse(UniqueStringStorage& unique_strings, StringView language)
  226. {
  227. CanonicalLanguageID language_id {};
  228. auto segments = language.split_view('-');
  229. VERIFY(!segments.is_empty());
  230. size_t index = 0;
  231. if (Locale::is_unicode_language_subtag(segments[index])) {
  232. language_id.language = unique_strings.ensure(segments[index]);
  233. if (segments.size() == ++index)
  234. return language_id;
  235. } else {
  236. return Error::from_string_literal("Expected language subtag");
  237. }
  238. if (Locale::is_unicode_script_subtag(segments[index])) {
  239. language_id.script = unique_strings.ensure(segments[index]);
  240. if (segments.size() == ++index)
  241. return language_id;
  242. }
  243. if (Locale::is_unicode_region_subtag(segments[index])) {
  244. language_id.region = unique_strings.ensure(segments[index]);
  245. if (segments.size() == ++index)
  246. return language_id;
  247. }
  248. while (index < segments.size()) {
  249. if (!Locale::is_unicode_variant_subtag(segments[index]))
  250. return Error::from_string_literal("Expected variant subtag");
  251. language_id.variants.append(unique_strings.ensure(segments[index++]));
  252. }
  253. return language_id;
  254. }
  255. size_t language { 0 };
  256. size_t script { 0 };
  257. size_t region { 0 };
  258. Vector<size_t> variants {};
  259. };
  260. inline ErrorOr<NonnullOwnPtr<Core::BufferedFile>> open_file(StringView path, Core::File::OpenMode mode)
  261. {
  262. if (path.is_empty())
  263. return Error::from_string_literal("Provided path is empty, please provide all command line options");
  264. auto file = TRY(Core::File::open(path, mode));
  265. return Core::BufferedFile::create(move(file));
  266. }
  267. inline ErrorOr<JsonValue> read_json_file(StringView path)
  268. {
  269. auto file = TRY(open_file(path, Core::File::OpenMode::Read));
  270. auto buffer = TRY(file->read_until_eof());
  271. return JsonValue::from_string(buffer);
  272. }
  273. inline ErrorOr<Core::DirIterator> path_to_dir_iterator(DeprecatedString path, StringView subpath = "main"sv)
  274. {
  275. LexicalPath lexical_path(move(path));
  276. if (!subpath.is_empty())
  277. lexical_path = lexical_path.append(subpath);
  278. Core::DirIterator iterator(lexical_path.string(), Core::DirIterator::SkipParentAndBaseDir);
  279. if (iterator.has_error()) {
  280. // FIXME: Make Core::DirIterator return a StringView for its error
  281. // string.
  282. auto const* error_string_ptr = iterator.error_string();
  283. return Error::from_string_view({ error_string_ptr, strlen(error_string_ptr) });
  284. }
  285. return iterator;
  286. }
  287. inline ErrorOr<DeprecatedString> next_path_from_dir_iterator(Core::DirIterator& iterator)
  288. {
  289. auto next_path = iterator.next_full_path();
  290. if (iterator.has_error()) {
  291. // FIXME: Make Core::DirIterator return a StringView for its error
  292. // string.
  293. auto const* error_string_ptr = iterator.error_string();
  294. return Error::from_string_view({ error_string_ptr, strlen(error_string_ptr) });
  295. }
  296. return next_path;
  297. }
  298. inline void ensure_from_string_types_are_generated(SourceGenerator& generator)
  299. {
  300. static bool generated_from_string_types = false;
  301. if (generated_from_string_types)
  302. return;
  303. generator.append(R"~~~(
  304. template <typename ValueType>
  305. struct HashValuePair {
  306. unsigned hash { 0 };
  307. ValueType value {};
  308. };
  309. template <typename ValueType>
  310. struct HashValueComparator
  311. {
  312. constexpr int operator()(unsigned hash, HashValuePair<ValueType> const& pair)
  313. {
  314. if (hash > pair.hash)
  315. return 1;
  316. if (hash < pair.hash)
  317. return -1;
  318. return 0;
  319. }
  320. };
  321. )~~~");
  322. generated_from_string_types = true;
  323. }
  324. template<typename ValueType>
  325. using HashValueMap = HashMap<unsigned, ValueType>;
  326. struct ValueFromStringOptions {
  327. Optional<StringView> return_type {};
  328. StringView return_format { "{}"sv };
  329. CaseSensitivity sensitivity { CaseSensitivity::CaseSensitive };
  330. };
  331. template<typename ValueType>
  332. void generate_value_from_string(SourceGenerator& generator, StringView method_name_format, StringView value_type, StringView value_name, HashValueMap<ValueType> hashes, ValueFromStringOptions options = {})
  333. {
  334. ensure_from_string_types_are_generated(generator);
  335. generator.set("method_name", DeprecatedString::formatted(method_name_format, value_name));
  336. generator.set("value_type", value_type);
  337. generator.set("value_name", value_name);
  338. generator.set("return_type", options.return_type.has_value() ? *options.return_type : value_type);
  339. generator.set("size", DeprecatedString::number(hashes.size()));
  340. generator.append(R"~~~(
  341. Optional<@return_type@> @method_name@(StringView key)
  342. {
  343. constexpr Array<HashValuePair<@value_type@>, @size@> hash_pairs { {
  344. )~~~");
  345. auto hash_keys = hashes.keys();
  346. quick_sort(hash_keys);
  347. constexpr size_t max_values_per_row = 10;
  348. size_t values_in_current_row = 0;
  349. for (auto hash_key : hash_keys) {
  350. if (values_in_current_row++ > 0)
  351. generator.append(" ");
  352. if constexpr (IsIntegral<ValueType>)
  353. generator.set("value"sv, DeprecatedString::number(hashes.get(hash_key).value()));
  354. else
  355. generator.set("value"sv, DeprecatedString::formatted("{}::{}", value_type, hashes.get(hash_key).value()));
  356. generator.set("hash"sv, DeprecatedString::number(hash_key));
  357. generator.append("{ @hash@U, @value@ },"sv);
  358. if (values_in_current_row == max_values_per_row) {
  359. generator.append("\n ");
  360. values_in_current_row = 0;
  361. }
  362. }
  363. generator.set("return_statement", DeprecatedString::formatted(options.return_format, "value->value"sv));
  364. generator.append(R"~~~(
  365. } };
  366. )~~~");
  367. if (options.sensitivity == CaseSensitivity::CaseSensitive) {
  368. generator.append(R"~~~(
  369. auto hash = key.hash();
  370. )~~~");
  371. } else {
  372. generator.append(R"~~~(
  373. auto hash = CaseInsensitiveStringViewTraits::hash(key);
  374. )~~~");
  375. }
  376. generator.append(R"~~~(
  377. if (auto const* value = binary_search(hash_pairs, hash, nullptr, HashValueComparator<@value_type@> {}))
  378. return @return_statement@;
  379. return {};
  380. }
  381. )~~~");
  382. }
  383. template<typename IdentifierFormatter>
  384. void generate_value_to_string(SourceGenerator& generator, StringView method_name_format, StringView value_type, StringView value_name, IdentifierFormatter&& format_identifier, ReadonlySpan<DeprecatedString> values)
  385. {
  386. generator.set("method_name", DeprecatedString::formatted(method_name_format, value_name));
  387. generator.set("value_type", value_type);
  388. generator.set("value_name", value_name);
  389. generator.append(R"~~~(
  390. StringView @method_name@(@value_type@ @value_name@)
  391. {
  392. using enum @value_type@;
  393. switch (@value_name@) {)~~~");
  394. for (auto const& value : values) {
  395. generator.set("enum_value", format_identifier(value_type, value));
  396. generator.set("string_value", value);
  397. generator.append(R"~~~(
  398. case @enum_value@:
  399. return "@string_value@"sv;)~~~");
  400. }
  401. generator.append(R"~~~(
  402. }
  403. VERIFY_NOT_REACHED();
  404. }
  405. )~~~");
  406. }
  407. template<typename IdentifierFormatter>
  408. void generate_enum(SourceGenerator& generator, IdentifierFormatter&& format_identifier, StringView name, StringView default_, Vector<DeprecatedString>& values, Vector<Alias> aliases = {})
  409. {
  410. quick_sort(values, [](auto const& value1, auto const& value2) { return value1.to_lowercase() < value2.to_lowercase(); });
  411. quick_sort(aliases, [](auto const& alias1, auto const& alias2) { return alias1.alias.to_lowercase() < alias2.alias.to_lowercase(); });
  412. generator.set("name", name);
  413. generator.set("underlying", ((values.size() + !default_.is_empty()) < 256) ? "u8"sv : "u16"sv);
  414. generator.append(R"~~~(
  415. enum class @name@ : @underlying@ {)~~~");
  416. if (!default_.is_empty()) {
  417. generator.set("default", default_);
  418. generator.append(R"~~~(
  419. @default@,)~~~");
  420. }
  421. for (auto const& value : values) {
  422. generator.set("value", format_identifier(name, value));
  423. generator.append(R"~~~(
  424. @value@,)~~~");
  425. }
  426. for (auto const& alias : aliases) {
  427. generator.set("alias", format_identifier(name, alias.alias));
  428. generator.set("value", format_identifier(name, alias.name));
  429. generator.append(R"~~~(
  430. @alias@ = @value@,)~~~");
  431. }
  432. generator.append(R"~~~(
  433. };
  434. )~~~");
  435. }
  436. template<typename LocalesType, typename IdentifierFormatter, typename ListFormatter>
  437. void generate_mapping(SourceGenerator& generator, LocalesType const& locales, StringView type, StringView name, StringView format, IdentifierFormatter&& format_identifier, ListFormatter&& format_list)
  438. {
  439. auto format_mapping_name = [&](StringView format, StringView name) {
  440. DeprecatedString mapping_name;
  441. if constexpr (IsNullPointer<IdentifierFormatter>)
  442. mapping_name = name.replace("-"sv, "_"sv, ReplaceMode::All);
  443. else
  444. mapping_name = format_identifier(type, name);
  445. return DeprecatedString::formatted(format, mapping_name.to_lowercase());
  446. };
  447. Vector<DeprecatedString> mapping_names;
  448. for (auto const& locale : locales) {
  449. DeprecatedString mapping_name;
  450. if constexpr (requires { locale.key; }) {
  451. mapping_name = format_mapping_name(format, locale.key);
  452. format_list(mapping_name, locale.value);
  453. } else {
  454. mapping_name = format_mapping_name(format, locale);
  455. format_list(mapping_name, locale);
  456. }
  457. mapping_names.append(move(mapping_name));
  458. }
  459. quick_sort(mapping_names);
  460. generator.set("type", type);
  461. generator.set("name", name);
  462. generator.set("size", DeprecatedString::number(locales.size()));
  463. generator.append(R"~~~(
  464. static constexpr Array<ReadonlySpan<@type@>, @size@> @name@ { {
  465. )~~~");
  466. constexpr size_t max_values_per_row = 10;
  467. size_t values_in_current_row = 0;
  468. for (auto& mapping_name : mapping_names) {
  469. if (values_in_current_row++ > 0)
  470. generator.append(" ");
  471. generator.set("name", move(mapping_name));
  472. generator.append("@name@.span(),");
  473. if (values_in_current_row == max_values_per_row) {
  474. values_in_current_row = 0;
  475. generator.append("\n ");
  476. }
  477. }
  478. generator.append(R"~~~(
  479. } };
  480. )~~~");
  481. }
  482. template<typename T>
  483. void generate_available_values(SourceGenerator& generator, StringView name, Vector<T> const& values, Vector<Alias> const& aliases = {}, Function<bool(StringView)> value_filter = {})
  484. {
  485. generator.set("name", name);
  486. generator.append(R"~~~(
  487. ReadonlySpan<StringView> @name@()
  488. {
  489. static constexpr auto values = Array {)~~~");
  490. bool first = true;
  491. for (auto const& value : values) {
  492. if (value_filter && !value_filter(value))
  493. continue;
  494. generator.append(first ? " "sv : ", "sv);
  495. first = false;
  496. if (auto it = aliases.find_if([&](auto const& alias) { return alias.alias == value; }); it != aliases.end())
  497. generator.append(DeprecatedString::formatted("\"{}\"sv", it->name));
  498. else
  499. generator.append(DeprecatedString::formatted("\"{}\"sv", value));
  500. }
  501. generator.append(R"~~~( };
  502. return values.span();
  503. }
  504. )~~~");
  505. }