ISO8601.h 8.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225
  1. /*
  2. * Copyright (c) 2021-2022, Linus Groh <linusg@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #pragma once
  7. #include <AK/GenericLexer.h>
  8. #include <AK/Optional.h>
  9. #include <AK/StringView.h>
  10. #include <AK/Vector.h>
  11. namespace JS::Temporal {
  12. struct Annotation {
  13. bool critical { false };
  14. StringView key;
  15. StringView value;
  16. };
  17. struct ParseResult {
  18. Optional<StringView> sign;
  19. Optional<StringView> date_year;
  20. Optional<StringView> date_month;
  21. Optional<StringView> date_day;
  22. Optional<StringView> time_hour;
  23. Optional<StringView> time_minute;
  24. Optional<StringView> time_second;
  25. Optional<StringView> time_fraction;
  26. Optional<StringView> utc_designator;
  27. Optional<StringView> time_zone_annotation;
  28. Optional<StringView> time_zone_numeric_utc_offset;
  29. Optional<StringView> time_zone_utc_offset_sign;
  30. Optional<StringView> time_zone_utc_offset_hour;
  31. Optional<StringView> time_zone_utc_offset_minute;
  32. Optional<StringView> time_zone_utc_offset_second;
  33. Optional<StringView> time_zone_utc_offset_fraction;
  34. Optional<StringView> time_zone_identifier;
  35. Optional<StringView> duration_years;
  36. Optional<StringView> duration_months;
  37. Optional<StringView> duration_weeks;
  38. Optional<StringView> duration_days;
  39. Optional<StringView> duration_whole_hours;
  40. Optional<StringView> duration_hours_fraction;
  41. Optional<StringView> duration_whole_minutes;
  42. Optional<StringView> duration_minutes_fraction;
  43. Optional<StringView> duration_whole_seconds;
  44. Optional<StringView> duration_seconds_fraction;
  45. Optional<StringView> annotation_key;
  46. Optional<StringView> annotation_value;
  47. Vector<Annotation> annotations;
  48. };
  49. enum class Production {
  50. TemporalInstantString,
  51. TemporalDateTimeString,
  52. TemporalDurationString,
  53. TemporalMonthDayString,
  54. TemporalTimeString,
  55. TemporalYearMonthString,
  56. TemporalZonedDateTimeString,
  57. TimeZoneIdentifier,
  58. TimeZoneNumericUTCOffset,
  59. AnnotationValue,
  60. DateMonth,
  61. };
  62. Optional<ParseResult> parse_iso8601(Production, StringView);
  63. namespace Detail {
  64. // 13.33 ISO 8601 grammar, https://tc39.es/proposal-temporal/#sec-temporal-iso8601grammar
  65. class ISO8601Parser {
  66. public:
  67. explicit ISO8601Parser(StringView input)
  68. : m_input(input)
  69. , m_state({
  70. .lexer = GenericLexer { input },
  71. .parse_result = {},
  72. })
  73. {
  74. }
  75. [[nodiscard]] GenericLexer const& lexer() const { return m_state.lexer; }
  76. [[nodiscard]] ParseResult const& parse_result() const { return m_state.parse_result; }
  77. [[nodiscard]] bool parse_decimal_digits();
  78. [[nodiscard]] bool parse_decimal_digit();
  79. [[nodiscard]] bool parse_non_zero_digit();
  80. [[nodiscard]] bool parse_ascii_sign();
  81. [[nodiscard]] bool parse_sign();
  82. [[nodiscard]] bool parse_unpadded_hour();
  83. [[nodiscard]] bool parse_hour();
  84. [[nodiscard]] bool parse_minute_second();
  85. [[nodiscard]] bool parse_decimal_separator();
  86. [[nodiscard]] bool parse_days_designator();
  87. [[nodiscard]] bool parse_hours_designator();
  88. [[nodiscard]] bool parse_minutes_designator();
  89. [[nodiscard]] bool parse_months_designator();
  90. [[nodiscard]] bool parse_duration_designator();
  91. [[nodiscard]] bool parse_seconds_designator();
  92. [[nodiscard]] bool parse_date_time_separator();
  93. [[nodiscard]] bool parse_time_designator();
  94. [[nodiscard]] bool parse_weeks_designator();
  95. [[nodiscard]] bool parse_years_designator();
  96. [[nodiscard]] bool parse_utc_designator();
  97. [[nodiscard]] bool parse_annotation_critical_flag();
  98. [[nodiscard]] bool parse_date_year();
  99. [[nodiscard]] bool parse_date_month();
  100. [[nodiscard]] bool parse_date_month_with_thirty_days();
  101. [[nodiscard]] bool parse_date_day();
  102. [[nodiscard]] bool parse_date_spec_year_month();
  103. [[nodiscard]] bool parse_date_spec_month_day();
  104. [[nodiscard]] bool parse_valid_month_day();
  105. [[nodiscard]] bool parse_date();
  106. [[nodiscard]] bool parse_time_hour();
  107. [[nodiscard]] bool parse_time_minute();
  108. [[nodiscard]] bool parse_time_second();
  109. [[nodiscard]] bool parse_fractional_part();
  110. [[nodiscard]] bool parse_fraction();
  111. [[nodiscard]] bool parse_time_fraction();
  112. [[nodiscard]] bool parse_time_zone_utc_offset_sign();
  113. [[nodiscard]] bool parse_time_zone_utc_offset_hour();
  114. [[nodiscard]] bool parse_time_zone_utc_offset_minute();
  115. [[nodiscard]] bool parse_time_zone_utc_offset_second();
  116. [[nodiscard]] bool parse_time_zone_utc_offset_fractional_part();
  117. [[nodiscard]] bool parse_time_zone_utc_offset_fraction();
  118. [[nodiscard]] bool parse_time_zone_numeric_utc_offset();
  119. [[nodiscard]] bool parse_time_zone_utc_offset();
  120. [[nodiscard]] bool parse_time_zone_utc_offset_name();
  121. [[nodiscard]] bool parse_tz_leading_char();
  122. [[nodiscard]] bool parse_tz_char();
  123. [[nodiscard]] bool parse_time_zone_iana_component();
  124. [[nodiscard]] bool parse_time_zone_iana_name_tail();
  125. [[nodiscard]] bool parse_time_zone_iana_legacy_name();
  126. [[nodiscard]] bool parse_time_zone_iana_name();
  127. [[nodiscard]] bool parse_time_zone_identifier();
  128. [[nodiscard]] bool parse_time_zone_annotation();
  129. [[nodiscard]] bool parse_a_key_leading_char();
  130. [[nodiscard]] bool parse_a_key_char();
  131. [[nodiscard]] bool parse_a_val_char();
  132. [[nodiscard]] bool parse_annotation_key_tail();
  133. [[nodiscard]] bool parse_annotation_key();
  134. [[nodiscard]] bool parse_annotation_value_component();
  135. [[nodiscard]] bool parse_annotation_value_tail();
  136. [[nodiscard]] bool parse_annotation_value();
  137. [[nodiscard]] bool parse_annotation();
  138. [[nodiscard]] bool parse_annotations();
  139. [[nodiscard]] bool parse_time_spec();
  140. [[nodiscard]] bool parse_time_spec_with_optional_offset_not_ambiguous();
  141. [[nodiscard]] bool parse_date_time();
  142. [[nodiscard]] bool parse_annotated_time();
  143. [[nodiscard]] bool parse_annotated_date_time();
  144. [[nodiscard]] bool parse_annotated_date_time_time_required();
  145. [[nodiscard]] bool parse_annotated_year_month();
  146. [[nodiscard]] bool parse_annotated_month_day();
  147. [[nodiscard]] bool parse_duration_whole_seconds();
  148. [[nodiscard]] bool parse_duration_seconds_fraction();
  149. [[nodiscard]] bool parse_duration_seconds_part();
  150. [[nodiscard]] bool parse_duration_whole_minutes();
  151. [[nodiscard]] bool parse_duration_minutes_fraction();
  152. [[nodiscard]] bool parse_duration_minutes_part();
  153. [[nodiscard]] bool parse_duration_whole_hours();
  154. [[nodiscard]] bool parse_duration_hours_fraction();
  155. [[nodiscard]] bool parse_duration_hours_part();
  156. [[nodiscard]] bool parse_duration_time();
  157. [[nodiscard]] bool parse_duration_days();
  158. [[nodiscard]] bool parse_duration_days_part();
  159. [[nodiscard]] bool parse_duration_weeks();
  160. [[nodiscard]] bool parse_duration_weeks_part();
  161. [[nodiscard]] bool parse_duration_months();
  162. [[nodiscard]] bool parse_duration_months_part();
  163. [[nodiscard]] bool parse_duration_years();
  164. [[nodiscard]] bool parse_duration_years_part();
  165. [[nodiscard]] bool parse_duration_date();
  166. [[nodiscard]] bool parse_duration();
  167. [[nodiscard]] bool parse_temporal_instant_string();
  168. [[nodiscard]] bool parse_temporal_date_time_string();
  169. [[nodiscard]] bool parse_temporal_duration_string();
  170. [[nodiscard]] bool parse_temporal_month_day_string();
  171. [[nodiscard]] bool parse_temporal_time_string();
  172. [[nodiscard]] bool parse_temporal_year_month_string();
  173. [[nodiscard]] bool parse_temporal_zoned_date_time_string();
  174. private:
  175. struct State {
  176. GenericLexer lexer;
  177. ParseResult parse_result;
  178. };
  179. struct StateTransaction {
  180. explicit StateTransaction(ISO8601Parser& parser)
  181. : m_parser(parser)
  182. , m_saved_state(parser.m_state)
  183. , m_start_index(parser.m_state.lexer.tell())
  184. {
  185. }
  186. ~StateTransaction()
  187. {
  188. if (!m_commit)
  189. m_parser.m_state = move(m_saved_state);
  190. }
  191. void commit() { m_commit = true; }
  192. StringView parsed_string_view() const
  193. {
  194. return m_parser.m_input.substring_view(m_start_index, m_parser.m_state.lexer.tell() - m_start_index);
  195. }
  196. private:
  197. ISO8601Parser& m_parser;
  198. State m_saved_state;
  199. size_t m_start_index { 0 };
  200. bool m_commit { false };
  201. };
  202. StringView m_input;
  203. State m_state;
  204. };
  205. }
  206. }