GenerateEncodingIndexes.cpp 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265
  1. /*
  2. * Copyright (c) 2024, Simon Wanner <simon@skyrising.xyz>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include <AK/Array.h>
  7. #include <AK/JsonObject.h>
  8. #include <AK/NumericLimits.h>
  9. #include <AK/SourceGenerator.h>
  10. #include <AK/StringBuilder.h>
  11. #include <AK/StringView.h>
  12. #include <AK/Vector.h>
  13. #include <LibCore/ArgsParser.h>
  14. #include <LibCore/File.h>
  15. #include <LibMain/Main.h>
  16. namespace {
  17. struct LookupTable {
  18. u32 first_pointer;
  19. u32 max_code_point;
  20. Vector<u32> code_points;
  21. bool generate_accessor;
  22. };
  23. struct LookupTables {
  24. JsonArray const& gb18030_ranges;
  25. OrderedHashMap<StringView, LookupTable> indexes;
  26. };
  27. enum class GenerateAccessor {
  28. No,
  29. Yes,
  30. };
  31. LookupTable prepare_table(JsonArray const& data, GenerateAccessor generate_accessor = GenerateAccessor::No)
  32. {
  33. Vector<u32> code_points;
  34. code_points.ensure_capacity(data.size());
  35. u32 max = 0;
  36. u32 first_pointer = 0;
  37. for (auto const& entry : data.values()) {
  38. if (entry.is_null()) {
  39. if (code_points.is_empty()) {
  40. first_pointer++;
  41. } else {
  42. code_points.append(0xfffd);
  43. max = AK::max(max, code_points.last());
  44. }
  45. } else {
  46. code_points.append(entry.as_integer<u32>());
  47. max = AK::max(max, code_points.last());
  48. }
  49. }
  50. if (generate_accessor == GenerateAccessor::Yes) {
  51. while (code_points.last() == 0xfffd)
  52. code_points.take_last();
  53. } else {
  54. VERIFY(first_pointer == 0);
  55. }
  56. return { first_pointer, max, move(code_points), generate_accessor == GenerateAccessor::Yes };
  57. }
  58. void generate_table(SourceGenerator generator, StringView name, LookupTable& table)
  59. {
  60. generator.set("name", name);
  61. generator.set("value_type", table.max_code_point > NumericLimits<u16>::max() ? "u32" : "u16");
  62. generator.set("first_pointer", MUST(String::number(table.first_pointer)));
  63. generator.set("size", MUST(String::number(table.code_points.size())));
  64. if (table.first_pointer > 0) {
  65. generator.appendln("static constexpr u32 s_@name@_index_first_pointer = @first_pointer@;");
  66. }
  67. generator.append("static constexpr Array<@value_type@, @size@> s_@name@_index {\n ");
  68. for (size_t i = 0; i < table.code_points.size(); i++) {
  69. generator.append(MUST(String::formatted("{:#04x}", table.code_points[i])));
  70. if (i != table.code_points.size() - 1)
  71. generator.append(i % 16 == 15 ? ",\n "sv : ", "sv);
  72. }
  73. generator.appendln("\n};");
  74. if (table.generate_accessor)
  75. generator.appendln("Optional<u32> index_@name@_code_point(u32 pointer);");
  76. }
  77. ErrorOr<void> generate_header_file(LookupTables& tables, Core::File& file)
  78. {
  79. StringBuilder builder;
  80. SourceGenerator generator { builder };
  81. generator.set("gb18030_ranges_size", MUST(String::number(tables.gb18030_ranges.size())));
  82. generator.append(R"~~~(
  83. #pragma once
  84. #include <AK/Array.h>
  85. #include <AK/Types.h>
  86. namespace TextCodec {
  87. struct Gb18030RangeEntry {
  88. u32 pointer;
  89. u32 code_point;
  90. };
  91. static constexpr Array<Gb18030RangeEntry, @gb18030_ranges_size@> s_gb18030_ranges { {
  92. )~~~");
  93. for (auto const& range : tables.gb18030_ranges.values()) {
  94. generator.appendln(MUST(String::formatted(" {{ {}, {:#04x} }},", range.as_array()[0].as_integer<u32>(), range.as_array()[1].as_integer<u32>())));
  95. }
  96. generator.appendln("} };\n");
  97. for (auto e : tables.indexes) {
  98. generate_table(generator.fork(), e.key, e.value);
  99. }
  100. generator.append("\n");
  101. generator.appendln("}");
  102. TRY(file.write_until_depleted(generator.as_string_view().bytes()));
  103. return {};
  104. }
  105. void generate_table_accessor(SourceGenerator generator, StringView name, LookupTable& table)
  106. {
  107. generator.set("name", name);
  108. generator.set("first_pointer", MUST(String::number(table.first_pointer)));
  109. generator.set("size", MUST(String::number(table.code_points.size())));
  110. if (table.first_pointer > 0) {
  111. generator.append(R"~~~(
  112. Optional<u32> index_@name@_code_point(u32 pointer)
  113. {
  114. if (pointer < s_@name@_index_first_pointer || pointer - s_@name@_index_first_pointer >= s_@name@_index.size())
  115. return {};
  116. auto value = s_@name@_index[pointer - s_@name@_index_first_pointer];
  117. if (value == 0xfffd)
  118. return {};
  119. return value;
  120. }
  121. )~~~");
  122. } else {
  123. generator.append(R"~~~(
  124. Optional<u32> index_@name@_code_point(u32 pointer)
  125. {
  126. if (pointer >= s_@name@_index.size())
  127. return {};
  128. auto value = s_@name@_index[pointer];
  129. if (value == 0xfffd)
  130. return {};
  131. return value;
  132. }
  133. )~~~");
  134. }
  135. }
  136. ErrorOr<void> generate_implementation_file(LookupTables& tables, Core::File& file)
  137. {
  138. StringBuilder builder;
  139. SourceGenerator generator { builder };
  140. generator.append(R"~~~(
  141. #include <LibTextCodec/LookupTables.h>
  142. namespace TextCodec {
  143. )~~~");
  144. for (auto& [key, table] : tables.indexes) {
  145. if (table.generate_accessor)
  146. generate_table_accessor(generator.fork(), key, table);
  147. }
  148. generator.appendln("\n}");
  149. TRY(file.write_until_depleted(generator.as_string_view().bytes()));
  150. return {};
  151. }
  152. } // end anonymous namespace
  153. ErrorOr<int> serenity_main(Main::Arguments arguments)
  154. {
  155. StringView generated_header_path;
  156. StringView generated_implementation_path;
  157. StringView json_path;
  158. Core::ArgsParser args_parser;
  159. args_parser.add_option(generated_header_path, "Path to the lookup table header file to generate", "generated-header-path", 'h', "generated-header-path");
  160. args_parser.add_option(generated_implementation_path, "Path to the lookup table implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  161. args_parser.add_option(json_path, "Path to the JSON file to read from", "json-path", 'j', "json-path");
  162. args_parser.parse(arguments);
  163. auto json_file = TRY(Core::File::open(json_path, Core::File::OpenMode::Read));
  164. auto json_data = TRY(json_file->read_until_eof());
  165. auto data = TRY(JsonValue::from_string(json_data)).as_object();
  166. auto gb18030_table = prepare_table(data.get("gb18030"sv)->as_array(), GenerateAccessor::Yes);
  167. // FIXME: Encoding specification is not updated to GB-18030-2022 yet (https://github.com/whatwg/encoding/issues/312)
  168. // NOTE: See https://commits.webkit.org/264918@main
  169. gb18030_table.code_points[7182] = 0xfe10;
  170. gb18030_table.code_points[7183] = 0xfe12;
  171. gb18030_table.code_points[7184] = 0xfe11;
  172. gb18030_table.code_points[7185] = 0xfe13;
  173. gb18030_table.code_points[7186] = 0xfe14;
  174. gb18030_table.code_points[7187] = 0xfe15;
  175. gb18030_table.code_points[7188] = 0xfe16;
  176. gb18030_table.code_points[7201] = 0xfe17;
  177. gb18030_table.code_points[7202] = 0xfe18;
  178. gb18030_table.code_points[7208] = 0xfe19;
  179. gb18030_table.code_points[23775] = 0x9fb4;
  180. gb18030_table.code_points[23783] = 0x9fb5;
  181. gb18030_table.code_points[23788] = 0x9fb6;
  182. gb18030_table.code_points[23789] = 0x9fb7;
  183. gb18030_table.code_points[23795] = 0x9fb8;
  184. gb18030_table.code_points[23812] = 0x9fb9;
  185. gb18030_table.code_points[23829] = 0x9fba;
  186. gb18030_table.code_points[23845] = 0x9fbb;
  187. LookupTables tables {
  188. .gb18030_ranges = data.get("gb18030-ranges"sv)->as_array(),
  189. .indexes = {
  190. { "gb18030"sv, move(gb18030_table) },
  191. { "big5"sv, prepare_table(data.get("big5"sv)->as_array(), GenerateAccessor::Yes) },
  192. { "jis0208"sv, prepare_table(data.get("jis0208"sv)->as_array(), GenerateAccessor::Yes) },
  193. { "jis0212"sv, prepare_table(data.get("jis0212"sv)->as_array(), GenerateAccessor::Yes) },
  194. { "euc_kr"sv, prepare_table(data.get("euc-kr"sv)->as_array(), GenerateAccessor::Yes) },
  195. { "ibm866"sv, prepare_table(data.get("ibm866"sv)->as_array()) },
  196. { "iso_8859_2"sv, prepare_table(data.get("iso-8859-2"sv)->as_array()) },
  197. { "iso_8859_3"sv, prepare_table(data.get("iso-8859-3"sv)->as_array()) },
  198. { "iso_8859_4"sv, prepare_table(data.get("iso-8859-4"sv)->as_array()) },
  199. { "iso_8859_5"sv, prepare_table(data.get("iso-8859-5"sv)->as_array()) },
  200. { "iso_8859_6"sv, prepare_table(data.get("iso-8859-6"sv)->as_array()) },
  201. { "iso_8859_7"sv, prepare_table(data.get("iso-8859-7"sv)->as_array()) },
  202. { "iso_8859_8"sv, prepare_table(data.get("iso-8859-8"sv)->as_array()) },
  203. { "iso_8859_10"sv, prepare_table(data.get("iso-8859-10"sv)->as_array()) },
  204. { "iso_8859_13"sv, prepare_table(data.get("iso-8859-13"sv)->as_array()) },
  205. { "iso_8859_14"sv, prepare_table(data.get("iso-8859-14"sv)->as_array()) },
  206. { "iso_8859_15"sv, prepare_table(data.get("iso-8859-15"sv)->as_array()) },
  207. { "iso_8859_16"sv, prepare_table(data.get("iso-8859-16"sv)->as_array()) },
  208. { "koi8_r"sv, prepare_table(data.get("koi8-r"sv)->as_array()) },
  209. { "koi8_u"sv, prepare_table(data.get("koi8-u"sv)->as_array()) },
  210. { "macintosh"sv, prepare_table(data.get("macintosh"sv)->as_array()) },
  211. { "windows_874"sv, prepare_table(data.get("windows-874"sv)->as_array()) },
  212. { "windows_1250"sv, prepare_table(data.get("windows-1250"sv)->as_array()) },
  213. { "windows_1251"sv, prepare_table(data.get("windows-1251"sv)->as_array()) },
  214. { "windows_1252"sv, prepare_table(data.get("windows-1252"sv)->as_array()) },
  215. { "windows_1253"sv, prepare_table(data.get("windows-1253"sv)->as_array()) },
  216. { "windows_1254"sv, prepare_table(data.get("windows-1254"sv)->as_array()) },
  217. { "windows_1255"sv, prepare_table(data.get("windows-1255"sv)->as_array()) },
  218. { "windows_1256"sv, prepare_table(data.get("windows-1256"sv)->as_array()) },
  219. { "windows_1257"sv, prepare_table(data.get("windows-1257"sv)->as_array()) },
  220. { "windows_1258"sv, prepare_table(data.get("windows-1258"sv)->as_array()) },
  221. { "x_mac_cyrillic"sv, prepare_table(data.get("x-mac-cyrillic"sv)->as_array()) },
  222. },
  223. };
  224. auto generated_header_file = TRY(Core::File::open(generated_header_path, Core::File::OpenMode::Write));
  225. auto generated_implementation_file = TRY(Core::File::open(generated_implementation_path, Core::File::OpenMode::Write));
  226. TRY(generate_header_file(tables, *generated_header_file));
  227. TRY(generate_implementation_file(tables, *generated_implementation_file));
  228. return 0;
  229. }