RegexLibC.cpp 45 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152
  1. /*
  2. * Copyright (c) 2020, Emanuel Sprung <emanuel.sprung@gmail.com>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include <LibTest/TestCase.h>
  7. #include <AK/StringBuilder.h>
  8. #include <LibC/regex.h>
  9. #include <stdio.h>
  10. TEST_CASE(catch_all)
  11. {
  12. String pattern = "^.*$";
  13. regex_t regex;
  14. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  15. EXPECT_EQ(regexec(&regex, "Hello World", 0, NULL, 0), REG_NOERR);
  16. regfree(&regex);
  17. }
  18. TEST_CASE(simple_start)
  19. {
  20. String pattern = "^hello friends";
  21. regex_t regex;
  22. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  23. EXPECT_EQ(regexec(&regex, "Hello!", 0, NULL, 0), REG_NOMATCH);
  24. EXPECT_EQ(regexec(&regex, "hello friends", 0, NULL, 0), REG_NOERR);
  25. EXPECT_EQ(regexec(&regex, "Well, hello friends", 0, NULL, 0), REG_NOMATCH);
  26. regfree(&regex);
  27. }
  28. TEST_CASE(simple_end)
  29. {
  30. String pattern = ".*hello\\.\\.\\. there$";
  31. regex_t regex;
  32. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  33. EXPECT_EQ(regexec(&regex, "Hallo", 0, NULL, 0), REG_NOMATCH);
  34. EXPECT_EQ(regexec(&regex, "I said fyhello... there", 0, NULL, 0), REG_NOERR);
  35. EXPECT_EQ(regexec(&regex, "ahello... therea", 0, NULL, 0), REG_NOMATCH);
  36. EXPECT_EQ(regexec(&regex, "hello.. there", 0, NULL, 0), REG_NOMATCH);
  37. regfree(&regex);
  38. }
  39. TEST_CASE(simple_period)
  40. {
  41. String pattern = "hello.";
  42. regex_t regex;
  43. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  44. EXPECT_EQ(regexec(&regex, "Hello1", 0, NULL, 0), REG_NOMATCH);
  45. EXPECT_EQ(regexec(&regex, "hello1", 0, NULL, 0), REG_NOERR);
  46. EXPECT_EQ(regexec(&regex, "hello2", 0, NULL, 0), REG_NOERR);
  47. EXPECT_EQ(regexec(&regex, "hello?", 0, NULL, 0), REG_NOERR);
  48. regfree(&regex);
  49. }
  50. TEST_CASE(simple_period_end)
  51. {
  52. String pattern = "hello.$";
  53. regex_t regex;
  54. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED | REG_NOSUB), REG_NOERR);
  55. EXPECT_EQ(regexec(&regex, "Hello1", 0, NULL, REG_NOSUB), REG_NOMATCH);
  56. EXPECT_EQ(regexec(&regex, "hello1hello1", 0, NULL, REG_GLOBAL), REG_NOERR);
  57. EXPECT_EQ(regexec(&regex, "hello2hell", 0, NULL, REG_GLOBAL), REG_NOMATCH);
  58. EXPECT_EQ(regexec(&regex, "hello?", 0, NULL, REG_NOSUB), REG_NOERR);
  59. regfree(&regex);
  60. }
  61. TEST_CASE(simple_escaped)
  62. {
  63. String pattern = "hello\\.";
  64. regex_t regex;
  65. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  66. EXPECT_EQ(regexec(&regex, "hello", 0, NULL, 0), REG_NOMATCH);
  67. EXPECT_EQ(regexec(&regex, "hello.", 0, NULL, 0), REG_NOERR);
  68. regfree(&regex);
  69. }
  70. TEST_CASE(simple_period2_end)
  71. {
  72. String pattern = ".*hi... there$";
  73. regex_t regex;
  74. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  75. EXPECT_EQ(regexec(&regex, "Hello there", 0, NULL, REG_GLOBAL), REG_NOMATCH);
  76. EXPECT_EQ(regexec(&regex, "I said fyhi... there", 0, NULL, REG_GLOBAL), REG_NOERR);
  77. EXPECT_EQ(regexec(&regex, "....hi... ", 0, NULL, REG_GLOBAL), REG_NOMATCH);
  78. EXPECT_EQ(regexec(&regex, "I said fyhihii there", 0, NULL, REG_GLOBAL), REG_NOERR);
  79. EXPECT_EQ(regexec(&regex, "I said fyhihi there", 0, NULL, REG_GLOBAL), REG_NOMATCH);
  80. regfree(&regex);
  81. }
  82. TEST_CASE(simple_plus)
  83. {
  84. String pattern = "a+";
  85. regex_t regex;
  86. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED | REG_NOSUB), REG_NOERR);
  87. EXPECT_EQ(regexec(&regex, "b", 0, NULL, REG_NOSUB), REG_NOMATCH);
  88. EXPECT_EQ(regexec(&regex, "a", 0, NULL, REG_NOSUB), REG_NOERR);
  89. EXPECT_EQ(regexec(&regex, "aaaaaabbbbb", 0, NULL, REG_GLOBAL), REG_NOERR);
  90. EXPECT_EQ(regexec(&regex, "aaaaaaaaaaa", 0, NULL, REG_GLOBAL), REG_NOERR);
  91. regfree(&regex);
  92. }
  93. TEST_CASE(simple_questionmark)
  94. {
  95. String pattern = "da?d";
  96. regex_t regex;
  97. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  98. EXPECT_EQ(regexec(&regex, "a", 0, NULL, REG_GLOBAL), REG_NOMATCH);
  99. EXPECT_EQ(regexec(&regex, "daa", 0, NULL, REG_GLOBAL), REG_NOMATCH);
  100. EXPECT_EQ(regexec(&regex, "ddddd", 0, NULL, REG_GLOBAL), REG_NOERR);
  101. EXPECT_EQ(regexec(&regex, "dd", 0, NULL, REG_GLOBAL), REG_NOERR);
  102. EXPECT_EQ(regexec(&regex, "dad", 0, NULL, REG_GLOBAL), REG_NOERR);
  103. EXPECT_EQ(regexec(&regex, "dada", 0, NULL, REG_GLOBAL), REG_NOERR);
  104. EXPECT_EQ(regexec(&regex, "adadaa", 0, NULL, REG_GLOBAL), REG_NOERR);
  105. regfree(&regex);
  106. }
  107. TEST_CASE(simple_questionmark_matchall)
  108. {
  109. String pattern = "da?d";
  110. regex_t regex;
  111. static constexpr int num_matches { 5 };
  112. regmatch_t matches[num_matches];
  113. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  114. EXPECT_EQ(regexec(&regex, "a", num_matches, matches, REG_GLOBAL), REG_NOMATCH);
  115. EXPECT_EQ(matches[0].rm_cnt, 0);
  116. EXPECT_EQ(regexec(&regex, "daa", num_matches, matches, REG_GLOBAL), REG_NOMATCH);
  117. EXPECT_EQ(matches[0].rm_cnt, 0);
  118. EXPECT_EQ(regexec(&regex, "ddddd", num_matches, matches, REG_GLOBAL), REG_NOERR);
  119. EXPECT_EQ(matches[0].rm_cnt, 2);
  120. EXPECT_EQ(matches[0].rm_so, 0);
  121. EXPECT_EQ(matches[0].rm_eo, 2);
  122. EXPECT_EQ(matches[1].rm_so, 2);
  123. EXPECT_EQ(matches[1].rm_eo, 4);
  124. EXPECT_EQ(regexec(&regex, "dd", num_matches, matches, REG_GLOBAL), REG_NOERR);
  125. EXPECT_EQ(matches[0].rm_cnt, 1);
  126. EXPECT_EQ(regexec(&regex, "dad", num_matches, matches, REG_GLOBAL), REG_NOERR);
  127. EXPECT_EQ(matches[0].rm_cnt, 1);
  128. EXPECT_EQ(regexec(&regex, "dada", num_matches, matches, REG_GLOBAL), REG_NOERR);
  129. EXPECT_EQ(matches[0].rm_cnt, 1);
  130. EXPECT_EQ(regexec(&regex, "adadaa", num_matches, matches, REG_GLOBAL), REG_NOERR);
  131. EXPECT_EQ(matches[0].rm_cnt, 1);
  132. regfree(&regex);
  133. }
  134. TEST_CASE(character_class)
  135. {
  136. String pattern = "[[:alpha:]]";
  137. regex_t regex;
  138. static constexpr int num_matches { 5 };
  139. regmatch_t matches[num_matches];
  140. String haystack = "[Window]\nOpacity=255\nAudibleBeep=0\n";
  141. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  142. EXPECT_EQ(regexec(&regex, haystack.characters(), num_matches, matches, 0), REG_NOMATCH);
  143. EXPECT_EQ(matches[0].rm_cnt, 0);
  144. EXPECT_EQ(regexec(&regex, haystack.characters(), num_matches, matches, REG_GLOBAL), REG_NOERR);
  145. EXPECT_EQ(matches[0].rm_cnt, 24);
  146. EXPECT_EQ(haystack.substring_view(matches[0].rm_so, matches[0].rm_eo - matches[0].rm_so), "W");
  147. EXPECT_EQ(haystack.substring_view(matches[1].rm_so, matches[1].rm_eo - matches[1].rm_so), "i");
  148. regfree(&regex);
  149. }
  150. TEST_CASE(character_class2)
  151. {
  152. String pattern = "[[:alpha:]]*=([[:digit:]]*)|\\[(.*)\\]";
  153. regex_t regex;
  154. static constexpr int num_matches { 9 };
  155. regmatch_t matches[num_matches];
  156. String haystack = "[Window]\nOpacity=255\nAudibleBeep=0\n";
  157. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED | REG_NEWLINE), REG_NOERR);
  158. EXPECT_EQ(regexec(&regex, haystack.characters(), num_matches, matches, 0), REG_NOERR);
  159. EXPECT_EQ(matches[0].rm_cnt, 3);
  160. #if 0
  161. for (int i = 0; i < num_matches; ++i) {
  162. fprintf(stderr, "Matches[%i].rm_so: %li, .rm_eo: %li .rm_cnt: %li: ", i, matches[i].rm_so, matches[i].rm_eo, matches[i].rm_cnt);
  163. fprintf(stderr, "haystack length: %lu\n", haystack.length());
  164. if (matches[i].rm_so != -1)
  165. fprintf(stderr, "%s\n", haystack.substring_view(matches[i].rm_so, matches[i].rm_eo - matches[i].rm_so).to_string().characters());
  166. }
  167. #endif
  168. EXPECT_EQ(haystack.substring_view(matches[0].rm_so, matches[0].rm_eo - matches[0].rm_so), "[Window]");
  169. EXPECT_EQ(matches[1].rm_so, -1);
  170. EXPECT_EQ(matches[1].rm_eo, -1);
  171. EXPECT_EQ(matches[1].rm_cnt, 0);
  172. EXPECT_EQ(haystack.substring_view(matches[2].rm_so, matches[2].rm_eo - matches[2].rm_so), "Window");
  173. EXPECT_EQ(haystack.substring_view(matches[3].rm_so, matches[3].rm_eo - matches[3].rm_so), "Opacity=255");
  174. EXPECT_EQ(haystack.substring_view(matches[4].rm_so, matches[4].rm_eo - matches[4].rm_so), "255");
  175. EXPECT_EQ(matches[5].rm_so, -1);
  176. EXPECT_EQ(matches[5].rm_eo, -1);
  177. EXPECT_EQ(matches[5].rm_cnt, 0);
  178. EXPECT_EQ(haystack.substring_view(matches[6].rm_so, matches[6].rm_eo - matches[6].rm_so), "AudibleBeep=0");
  179. EXPECT_EQ(haystack.substring_view(matches[7].rm_so, matches[7].rm_eo - matches[7].rm_so), "0");
  180. EXPECT_EQ(matches[8].rm_so, -1);
  181. EXPECT_EQ(matches[8].rm_eo, -1);
  182. EXPECT_EQ(matches[8].rm_cnt, 0);
  183. regfree(&regex);
  184. }
  185. TEST_CASE(escaped_char_questionmark)
  186. {
  187. String pattern = "This\\.?And\\.?That";
  188. regex_t regex;
  189. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  190. EXPECT_EQ(regexec(&regex, "ThisAndThat", 0, NULL, 0), REG_NOERR);
  191. EXPECT_EQ(regexec(&regex, "This.And.That", 0, NULL, 0), REG_NOERR);
  192. EXPECT_EQ(regexec(&regex, "This And That", 0, NULL, 0), REG_NOMATCH);
  193. EXPECT_EQ(regexec(&regex, "This..And..That", 0, NULL, 0), REG_NOMATCH);
  194. regfree(&regex);
  195. }
  196. TEST_CASE(char_qualifier_asterisk)
  197. {
  198. String pattern = "regex*";
  199. regex_t regex;
  200. static constexpr int num_matches { 5 };
  201. regmatch_t matches[num_matches];
  202. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  203. EXPECT_EQ(regexec(&regex, "#include <regex.h>", num_matches, matches, REG_GLOBAL), REG_NOERR);
  204. EXPECT_EQ(matches[0].rm_cnt, 1);
  205. regfree(&regex);
  206. }
  207. TEST_CASE(char_utf8)
  208. {
  209. String pattern = "😀";
  210. regex_t regex;
  211. static constexpr int num_matches { 5 };
  212. regmatch_t matches[num_matches];
  213. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  214. EXPECT_EQ(regexec(&regex, "Привет, мир! 😀 γειά σου κόσμος 😀 こんにちは世界", num_matches, matches, REG_GLOBAL), REG_NOERR);
  215. EXPECT_EQ(matches[0].rm_cnt, 2);
  216. regfree(&regex);
  217. }
  218. TEST_CASE(parens)
  219. {
  220. String pattern = "test(hello)test";
  221. regex_t regex;
  222. static constexpr int num_matches { 5 };
  223. regmatch_t matches[num_matches];
  224. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  225. EXPECT_EQ(regexec(&regex, "testhellotest", num_matches, matches, 0), REG_NOERR);
  226. EXPECT_EQ(matches[0].rm_cnt, 1);
  227. EXPECT_EQ(matches[0].rm_so, 0);
  228. EXPECT_EQ(matches[0].rm_eo, 13);
  229. EXPECT_EQ(matches[1].rm_so, 4);
  230. EXPECT_EQ(matches[1].rm_eo, 9);
  231. regfree(&regex);
  232. }
  233. TEST_CASE(parser_error_parens)
  234. {
  235. String pattern = "test()test";
  236. regex_t regex;
  237. static constexpr int num_matches { 5 };
  238. regmatch_t matches[num_matches];
  239. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_EMPTY_EXPR);
  240. EXPECT_EQ(regexec(&regex, "testhellotest", num_matches, matches, 0), REG_EMPTY_EXPR);
  241. regfree(&regex);
  242. }
  243. TEST_CASE(parser_error_special_characters_used_at_wrong_place)
  244. {
  245. String pattern;
  246. regex_t regex;
  247. static constexpr int num_matches { 5 };
  248. regmatch_t matches[num_matches];
  249. Vector<char, 4> chars = { '*', '+', '?', '}' };
  250. StringBuilder b;
  251. for (auto& ch : chars) {
  252. auto error_code_to_check = ch == '}' ? REG_EBRACE : REG_BADRPT;
  253. // First in ere
  254. b.clear();
  255. b.append(ch);
  256. pattern = b.build();
  257. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), error_code_to_check);
  258. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), error_code_to_check);
  259. regfree(&regex);
  260. // After vertical line
  261. b.clear();
  262. b.append("a|");
  263. b.append(ch);
  264. pattern = b.build();
  265. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), error_code_to_check);
  266. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), error_code_to_check);
  267. regfree(&regex);
  268. // After circumflex
  269. b.clear();
  270. b.append("^");
  271. b.append(ch);
  272. pattern = b.build();
  273. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), error_code_to_check);
  274. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), error_code_to_check);
  275. regfree(&regex);
  276. // After dollar
  277. b.clear();
  278. b.append("$");
  279. b.append(ch);
  280. pattern = b.build();
  281. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), error_code_to_check);
  282. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), error_code_to_check);
  283. regfree(&regex);
  284. // After left parens
  285. b.clear();
  286. b.append("(");
  287. b.append(ch);
  288. b.append(")");
  289. pattern = b.build();
  290. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), error_code_to_check);
  291. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), error_code_to_check);
  292. regfree(&regex);
  293. }
  294. }
  295. TEST_CASE(parser_error_vertical_line_used_at_wrong_place)
  296. {
  297. String pattern;
  298. regex_t regex;
  299. static constexpr int num_matches { 5 };
  300. regmatch_t matches[num_matches];
  301. // First in ere
  302. pattern = "|asdf";
  303. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_EMPTY_EXPR);
  304. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), REG_EMPTY_EXPR);
  305. regfree(&regex);
  306. // Last in ere
  307. pattern = "asdf|";
  308. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_EMPTY_EXPR);
  309. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), REG_EMPTY_EXPR);
  310. regfree(&regex);
  311. // After left parens
  312. pattern = "(|asdf)";
  313. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_EMPTY_EXPR);
  314. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), REG_EMPTY_EXPR);
  315. regfree(&regex);
  316. // Proceed right parens
  317. pattern = "(asdf)|";
  318. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_EMPTY_EXPR);
  319. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), REG_EMPTY_EXPR);
  320. regfree(&regex);
  321. }
  322. TEST_CASE(parens_qualifier_questionmark)
  323. {
  324. String pattern = "test(hello)?test";
  325. regex_t regex;
  326. static constexpr int num_matches { 5 };
  327. regmatch_t matches[num_matches];
  328. const char* match_str;
  329. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  330. match_str = "testtest";
  331. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  332. EXPECT_EQ(matches[0].rm_cnt, 1);
  333. EXPECT_EQ(matches[0].rm_so, 0);
  334. EXPECT_EQ(matches[0].rm_eo, 8);
  335. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testtest");
  336. match_str = "testhellotest";
  337. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  338. EXPECT_EQ(matches[0].rm_cnt, 1);
  339. EXPECT_EQ(matches[0].rm_so, 0);
  340. EXPECT_EQ(matches[0].rm_eo, 13);
  341. EXPECT_EQ(matches[1].rm_so, 4);
  342. EXPECT_EQ(matches[1].rm_eo, 9);
  343. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testhellotest");
  344. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  345. regfree(&regex);
  346. }
  347. TEST_CASE(parens_qualifier_asterisk)
  348. {
  349. String pattern = "test(hello)*test";
  350. regex_t regex;
  351. static constexpr int num_matches { 6 };
  352. regmatch_t matches[num_matches];
  353. const char* match_str;
  354. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  355. match_str = "testtest";
  356. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  357. EXPECT_EQ(matches[0].rm_cnt, 1);
  358. EXPECT_EQ(matches[0].rm_so, 0);
  359. EXPECT_EQ(matches[0].rm_eo, 8);
  360. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testtest");
  361. match_str = "testhellohellotest";
  362. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  363. EXPECT_EQ(matches[0].rm_cnt, 1);
  364. EXPECT_EQ(matches[0].rm_so, 0);
  365. EXPECT_EQ(matches[0].rm_eo, 18);
  366. EXPECT_EQ(matches[1].rm_so, 9);
  367. EXPECT_EQ(matches[1].rm_eo, 14);
  368. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testhellohellotest");
  369. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  370. match_str = "testhellohellotest, testhellotest";
  371. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  372. EXPECT_EQ(matches[0].rm_cnt, 2);
  373. EXPECT_EQ(matches[0].rm_so, 0);
  374. EXPECT_EQ(matches[0].rm_eo, 18);
  375. EXPECT_EQ(matches[1].rm_so, 9);
  376. EXPECT_EQ(matches[1].rm_eo, 14);
  377. EXPECT_EQ(matches[2].rm_so, 20);
  378. EXPECT_EQ(matches[2].rm_eo, 33);
  379. EXPECT_EQ(matches[3].rm_so, 24);
  380. EXPECT_EQ(matches[3].rm_eo, 29);
  381. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testhellohellotest");
  382. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  383. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "testhellotest");
  384. EXPECT_EQ(StringView(&match_str[matches[3].rm_so], matches[3].rm_eo - matches[3].rm_so), "hello");
  385. regfree(&regex);
  386. }
  387. TEST_CASE(parens_qualifier_asterisk_2)
  388. {
  389. String pattern = "test(.*)test";
  390. regex_t regex;
  391. static constexpr int num_matches { 6 };
  392. regmatch_t matches[num_matches];
  393. const char* match_str;
  394. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  395. match_str = "testasdftest";
  396. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  397. EXPECT_EQ(matches[0].rm_cnt, 1);
  398. EXPECT_EQ(matches[0].rm_so, 0);
  399. EXPECT_EQ(matches[0].rm_eo, 12);
  400. EXPECT_EQ(matches[1].rm_so, 4);
  401. EXPECT_EQ(matches[1].rm_eo, 8);
  402. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testasdftest");
  403. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "asdf");
  404. match_str = "testasdfasdftest";
  405. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  406. EXPECT_EQ(matches[0].rm_cnt, 1);
  407. EXPECT_EQ(matches[0].rm_so, 0);
  408. EXPECT_EQ(matches[0].rm_eo, 16);
  409. EXPECT_EQ(matches[1].rm_so, 4);
  410. EXPECT_EQ(matches[1].rm_eo, 12);
  411. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testasdfasdftest");
  412. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "asdfasdf");
  413. match_str = "testaaaatest, testbbbtest, testtest";
  414. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  415. EXPECT_EQ(matches[0].rm_cnt, 1);
  416. EXPECT_EQ(matches[0].rm_so, 0);
  417. EXPECT_EQ(matches[0].rm_eo, 35);
  418. EXPECT_EQ(matches[1].rm_so, 4);
  419. EXPECT_EQ(matches[1].rm_eo, 31);
  420. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testaaaatest, testbbbtest, testtest");
  421. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "aaaatest, testbbbtest, test");
  422. regfree(&regex);
  423. }
  424. TEST_CASE(mulit_parens_qualifier_too_less_result_values)
  425. {
  426. String pattern = "test(a)?(b)?(c)?test";
  427. regex_t regex;
  428. static constexpr int num_matches { 4 };
  429. regmatch_t matches[num_matches];
  430. const char* match_str;
  431. matches[3] = { -2, -2, 100 };
  432. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  433. match_str = "testabtest";
  434. EXPECT_EQ(regexec(&regex, match_str, num_matches - 1, matches, 0), REG_NOERR);
  435. EXPECT_EQ(matches[0].rm_cnt, 1);
  436. EXPECT_EQ(matches[0].rm_so, 0);
  437. EXPECT_EQ(matches[0].rm_eo, 10);
  438. EXPECT_EQ(matches[1].rm_so, 4);
  439. EXPECT_EQ(matches[1].rm_eo, 5);
  440. EXPECT_EQ(matches[2].rm_so, 5);
  441. EXPECT_EQ(matches[2].rm_eo, 6);
  442. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testabtest");
  443. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "a");
  444. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "b");
  445. EXPECT_EQ(matches[3].rm_so, -2);
  446. EXPECT_EQ(matches[3].rm_eo, -2);
  447. EXPECT_EQ(matches[3].rm_cnt, 100);
  448. match_str = "testabctest";
  449. EXPECT_EQ(regexec(&regex, match_str, num_matches - 1, matches, 0), REG_NOERR);
  450. EXPECT_EQ(matches[0].rm_cnt, 1);
  451. EXPECT_EQ(matches[0].rm_so, 0);
  452. EXPECT_EQ(matches[0].rm_eo, 11);
  453. EXPECT_EQ(matches[1].rm_so, 4);
  454. EXPECT_EQ(matches[1].rm_eo, 5);
  455. EXPECT_EQ(matches[2].rm_so, 5);
  456. EXPECT_EQ(matches[2].rm_eo, 6);
  457. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testabctest");
  458. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "a");
  459. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "b");
  460. EXPECT_EQ(matches[3].rm_so, -2);
  461. EXPECT_EQ(matches[3].rm_eo, -2);
  462. EXPECT_EQ(matches[3].rm_cnt, 100);
  463. match_str = "testabctest, testabctest";
  464. EXPECT_EQ(regexec(&regex, match_str, num_matches - 1, matches, REG_GLOBAL), REG_NOERR);
  465. EXPECT_EQ(matches[0].rm_cnt, 2);
  466. EXPECT_EQ(matches[0].rm_so, 0);
  467. EXPECT_EQ(matches[0].rm_eo, 11);
  468. EXPECT_EQ(matches[1].rm_so, 4);
  469. EXPECT_EQ(matches[1].rm_eo, 5);
  470. EXPECT_EQ(matches[2].rm_so, 5);
  471. EXPECT_EQ(matches[2].rm_eo, 6);
  472. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testabctest");
  473. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "a");
  474. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "b");
  475. EXPECT_EQ(matches[3].rm_so, -2);
  476. EXPECT_EQ(matches[3].rm_eo, -2);
  477. EXPECT_EQ(matches[3].rm_cnt, 100);
  478. regfree(&regex);
  479. }
  480. TEST_CASE(multi_parens_qualifier_questionmark)
  481. {
  482. String pattern = "test(a)?(b)?(c)?test";
  483. regex_t regex;
  484. static constexpr int num_matches { 8 };
  485. regmatch_t matches[num_matches];
  486. const char* match_str;
  487. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  488. match_str = "testtest";
  489. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  490. EXPECT_EQ(matches[0].rm_cnt, 1);
  491. EXPECT_EQ(matches[0].rm_so, 0);
  492. EXPECT_EQ(matches[0].rm_eo, 8);
  493. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testtest");
  494. match_str = "testabctest";
  495. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  496. EXPECT_EQ(matches[0].rm_cnt, 1);
  497. EXPECT_EQ(matches[0].rm_so, 0);
  498. EXPECT_EQ(matches[0].rm_eo, 11);
  499. EXPECT_EQ(matches[1].rm_so, 4);
  500. EXPECT_EQ(matches[1].rm_eo, 5);
  501. EXPECT_EQ(matches[2].rm_so, 5);
  502. EXPECT_EQ(matches[2].rm_eo, 6);
  503. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testabctest");
  504. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "a");
  505. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "b");
  506. match_str = "testabctest, testactest";
  507. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  508. EXPECT_EQ(matches[0].rm_cnt, 2);
  509. EXPECT_EQ(matches[0].rm_so, 0);
  510. EXPECT_EQ(matches[0].rm_eo, 11);
  511. EXPECT_EQ(matches[1].rm_so, 4);
  512. EXPECT_EQ(matches[1].rm_eo, 5);
  513. EXPECT_EQ(matches[2].rm_so, 5);
  514. EXPECT_EQ(matches[2].rm_eo, 6);
  515. EXPECT_EQ(matches[3].rm_so, 6);
  516. EXPECT_EQ(matches[3].rm_eo, 7);
  517. EXPECT_EQ(matches[4].rm_so, 13);
  518. EXPECT_EQ(matches[4].rm_eo, 23);
  519. EXPECT_EQ(matches[5].rm_so, 17);
  520. EXPECT_EQ(matches[5].rm_eo, 18);
  521. EXPECT_EQ(matches[6].rm_so, -1);
  522. EXPECT_EQ(matches[6].rm_eo, -1);
  523. EXPECT_EQ(matches[7].rm_so, 18);
  524. EXPECT_EQ(matches[7].rm_eo, 19);
  525. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testabctest");
  526. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "a");
  527. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "b");
  528. EXPECT_EQ(StringView(&match_str[matches[3].rm_so], matches[3].rm_eo - matches[3].rm_so), "c");
  529. EXPECT_EQ(StringView(&match_str[matches[4].rm_so], matches[4].rm_eo - matches[4].rm_so), "testactest");
  530. EXPECT_EQ(StringView(&match_str[matches[5].rm_so], matches[5].rm_eo - matches[5].rm_so), "a");
  531. EXPECT_EQ(StringView(&match_str[matches[6].rm_so], matches[6].rm_eo - matches[6].rm_so), "");
  532. EXPECT_EQ(StringView(&match_str[matches[7].rm_so], matches[7].rm_eo - matches[7].rm_so), "c");
  533. regfree(&regex);
  534. }
  535. TEST_CASE(simple_alternative)
  536. {
  537. String pattern = "test|hello|friends";
  538. regex_t regex;
  539. static constexpr int num_matches { 1 };
  540. regmatch_t matches[num_matches];
  541. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  542. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), REG_NOERR);
  543. EXPECT_EQ(matches[0].rm_cnt, 1);
  544. EXPECT_EQ(matches[0].rm_so, 0);
  545. EXPECT_EQ(matches[0].rm_eo, 4);
  546. EXPECT_EQ(regexec(&regex, "hello", num_matches, matches, 0), REG_NOERR);
  547. EXPECT_EQ(matches[0].rm_cnt, 1);
  548. EXPECT_EQ(matches[0].rm_so, 0);
  549. EXPECT_EQ(matches[0].rm_eo, 5);
  550. EXPECT_EQ(regexec(&regex, "friends", num_matches, matches, 0), REG_NOERR);
  551. EXPECT_EQ(matches[0].rm_cnt, 1);
  552. EXPECT_EQ(matches[0].rm_so, 0);
  553. EXPECT_EQ(matches[0].rm_eo, 7);
  554. regfree(&regex);
  555. }
  556. TEST_CASE(alternative_match_groups)
  557. {
  558. String pattern = "test(a)?(b)?|hello ?(dear|my)? friends";
  559. regex_t regex;
  560. static constexpr int num_matches { 8 };
  561. regmatch_t matches[num_matches];
  562. const char* match_str;
  563. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  564. match_str = "test";
  565. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  566. EXPECT_EQ(matches[0].rm_cnt, 1);
  567. EXPECT_EQ(matches[0].rm_so, 0);
  568. EXPECT_EQ(matches[0].rm_eo, 4);
  569. EXPECT_EQ(matches[1].rm_so, -1);
  570. EXPECT_EQ(matches[1].rm_eo, -1);
  571. EXPECT_EQ(matches[2].rm_so, -1);
  572. EXPECT_EQ(matches[2].rm_eo, -1);
  573. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "test");
  574. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "");
  575. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "");
  576. match_str = "testa";
  577. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  578. EXPECT_EQ(matches[0].rm_cnt, 1);
  579. EXPECT_EQ(matches[0].rm_so, 0);
  580. EXPECT_EQ(matches[0].rm_eo, 5);
  581. EXPECT_EQ(matches[1].rm_so, 4);
  582. EXPECT_EQ(matches[1].rm_eo, 5);
  583. EXPECT_EQ(matches[2].rm_so, -1);
  584. EXPECT_EQ(matches[2].rm_eo, -1);
  585. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testa");
  586. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "a");
  587. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "");
  588. match_str = "testb";
  589. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  590. EXPECT_EQ(matches[0].rm_cnt, 1);
  591. EXPECT_EQ(matches[0].rm_so, 0);
  592. EXPECT_EQ(matches[0].rm_eo, 5);
  593. EXPECT_EQ(matches[1].rm_so, -1);
  594. EXPECT_EQ(matches[1].rm_eo, -1);
  595. EXPECT_EQ(matches[2].rm_so, 4);
  596. EXPECT_EQ(matches[2].rm_eo, 5);
  597. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testb");
  598. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "");
  599. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "b");
  600. match_str = "hello friends";
  601. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  602. EXPECT_EQ(matches[0].rm_cnt, 1);
  603. EXPECT_EQ(matches[0].rm_so, 0);
  604. EXPECT_EQ(matches[0].rm_eo, 13);
  605. EXPECT_EQ(matches[1].rm_so, -1);
  606. EXPECT_EQ(matches[1].rm_eo, -1);
  607. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hello friends");
  608. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "");
  609. match_str = "hello dear friends";
  610. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  611. EXPECT_EQ(matches[0].rm_cnt, 1);
  612. EXPECT_EQ(matches[0].rm_so, 0);
  613. EXPECT_EQ(matches[0].rm_eo, 18);
  614. EXPECT_EQ(matches[1].rm_so, -1);
  615. EXPECT_EQ(matches[1].rm_eo, -1);
  616. EXPECT_EQ(matches[2].rm_so, -1);
  617. EXPECT_EQ(matches[2].rm_eo, -1);
  618. EXPECT_EQ(matches[3].rm_so, 6);
  619. EXPECT_EQ(matches[3].rm_eo, 10);
  620. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hello dear friends");
  621. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "");
  622. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "");
  623. EXPECT_EQ(StringView(&match_str[matches[3].rm_so], matches[3].rm_eo - matches[3].rm_so), "dear");
  624. match_str = "hello my friends";
  625. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  626. EXPECT_EQ(matches[0].rm_cnt, 1);
  627. EXPECT_EQ(matches[0].rm_so, 0);
  628. EXPECT_EQ(matches[0].rm_eo, 16);
  629. EXPECT_EQ(matches[1].rm_so, -1);
  630. EXPECT_EQ(matches[1].rm_eo, -1);
  631. EXPECT_EQ(matches[2].rm_so, -1);
  632. EXPECT_EQ(matches[2].rm_eo, -1);
  633. EXPECT_EQ(matches[3].rm_so, 6);
  634. EXPECT_EQ(matches[3].rm_eo, 8);
  635. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hello my friends");
  636. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "");
  637. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "");
  638. EXPECT_EQ(StringView(&match_str[matches[3].rm_so], matches[3].rm_eo - matches[3].rm_so), "my");
  639. match_str = "testabc";
  640. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOMATCH);
  641. EXPECT_EQ(matches[0].rm_cnt, 0);
  642. EXPECT_EQ(matches[0].rm_so, -1);
  643. EXPECT_EQ(matches[0].rm_eo, -1);
  644. match_str = "hello test friends";
  645. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOMATCH);
  646. EXPECT_EQ(matches[0].rm_cnt, 0);
  647. EXPECT_EQ(matches[0].rm_so, -1);
  648. EXPECT_EQ(matches[0].rm_eo, -1);
  649. regfree(&regex);
  650. }
  651. TEST_CASE(parens_qualifier_exact)
  652. {
  653. String pattern = "(hello){3}";
  654. regex_t regex;
  655. static constexpr int num_matches { 5 };
  656. regmatch_t matches[num_matches];
  657. const char* match_str;
  658. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  659. match_str = "hello";
  660. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOMATCH);
  661. EXPECT_EQ(matches[0].rm_cnt, 0);
  662. match_str = "hellohellohello";
  663. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  664. EXPECT_EQ(matches[0].rm_cnt, 1);
  665. EXPECT_EQ(matches[0].rm_so, 0);
  666. EXPECT_EQ(matches[0].rm_eo, 15);
  667. EXPECT_EQ(matches[1].rm_so, 10);
  668. EXPECT_EQ(matches[1].rm_eo, 15);
  669. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  670. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  671. match_str = "hellohellohellohello";
  672. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  673. EXPECT_EQ(matches[0].rm_cnt, 1);
  674. EXPECT_EQ(matches[0].rm_so, 0);
  675. EXPECT_EQ(matches[0].rm_eo, 15);
  676. EXPECT_EQ(matches[1].rm_so, 10);
  677. EXPECT_EQ(matches[1].rm_eo, 15);
  678. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  679. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  680. match_str = "test hellohellohello";
  681. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  682. EXPECT_EQ(matches[0].rm_cnt, 1);
  683. EXPECT_EQ(matches[0].rm_so, 5);
  684. EXPECT_EQ(matches[0].rm_eo, 20);
  685. EXPECT_EQ(matches[1].rm_so, 15);
  686. EXPECT_EQ(matches[1].rm_eo, 20);
  687. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  688. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  689. regfree(&regex);
  690. }
  691. TEST_CASE(parens_qualifier_minimum)
  692. {
  693. String pattern = "(hello){3,}";
  694. regex_t regex;
  695. static constexpr int num_matches { 5 };
  696. regmatch_t matches[num_matches];
  697. const char* match_str;
  698. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  699. match_str = "hello";
  700. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOMATCH);
  701. EXPECT_EQ(matches[0].rm_cnt, 0);
  702. match_str = "hellohellohello";
  703. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  704. EXPECT_EQ(matches[0].rm_cnt, 1);
  705. EXPECT_EQ(matches[0].rm_so, 0);
  706. EXPECT_EQ(matches[0].rm_eo, 15);
  707. EXPECT_EQ(matches[1].rm_so, 10);
  708. EXPECT_EQ(matches[1].rm_eo, 15);
  709. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  710. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  711. match_str = "hellohellohellohello";
  712. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_SEARCH), REG_NOERR);
  713. EXPECT_EQ(matches[0].rm_cnt, 1);
  714. EXPECT_EQ(matches[0].rm_so, 0);
  715. EXPECT_EQ(matches[0].rm_eo, 20);
  716. EXPECT_EQ(matches[1].rm_so, 15);
  717. EXPECT_EQ(matches[1].rm_eo, 20);
  718. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohellohello");
  719. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  720. match_str = "test hellohellohello";
  721. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  722. EXPECT_EQ(matches[0].rm_cnt, 1);
  723. EXPECT_EQ(matches[0].rm_so, 5);
  724. EXPECT_EQ(matches[0].rm_eo, 20);
  725. EXPECT_EQ(matches[1].rm_so, 15);
  726. EXPECT_EQ(matches[1].rm_eo, 20);
  727. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  728. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  729. match_str = "test hellohellohellohello";
  730. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  731. EXPECT_EQ(matches[0].rm_cnt, 1);
  732. EXPECT_EQ(matches[0].rm_so, 5);
  733. EXPECT_EQ(matches[0].rm_eo, 25);
  734. EXPECT_EQ(matches[1].rm_so, 20);
  735. EXPECT_EQ(matches[1].rm_eo, 25);
  736. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohellohello");
  737. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  738. regfree(&regex);
  739. }
  740. TEST_CASE(parens_qualifier_maximum)
  741. {
  742. String pattern = "(hello){2,3}";
  743. regex_t regex;
  744. static constexpr int num_matches { 5 };
  745. regmatch_t matches[num_matches];
  746. const char* match_str;
  747. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  748. match_str = "hello";
  749. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOMATCH);
  750. EXPECT_EQ(matches[0].rm_cnt, 0);
  751. match_str = "hellohellohello";
  752. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  753. EXPECT_EQ(matches[0].rm_cnt, 1);
  754. EXPECT_EQ(matches[0].rm_so, 0);
  755. EXPECT_EQ(matches[0].rm_eo, 15);
  756. EXPECT_EQ(matches[1].rm_so, 10);
  757. EXPECT_EQ(matches[1].rm_eo, 15);
  758. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  759. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  760. match_str = "hellohellohellohello";
  761. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  762. EXPECT_EQ(matches[0].rm_cnt, 1);
  763. EXPECT_EQ(matches[0].rm_so, 0);
  764. EXPECT_EQ(matches[0].rm_eo, 15);
  765. EXPECT_EQ(matches[1].rm_so, 10);
  766. EXPECT_EQ(matches[1].rm_eo, 15);
  767. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  768. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  769. match_str = "test hellohellohello";
  770. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  771. EXPECT_EQ(matches[0].rm_cnt, 1);
  772. EXPECT_EQ(matches[0].rm_so, 5);
  773. EXPECT_EQ(matches[0].rm_eo, 20);
  774. EXPECT_EQ(matches[1].rm_so, 15);
  775. EXPECT_EQ(matches[1].rm_eo, 20);
  776. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  777. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  778. match_str = "test hellohellohellohello";
  779. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  780. EXPECT_EQ(matches[0].rm_cnt, 1);
  781. EXPECT_EQ(matches[0].rm_so, 5);
  782. EXPECT_EQ(matches[0].rm_eo, 20);
  783. EXPECT_EQ(matches[1].rm_so, 15);
  784. EXPECT_EQ(matches[1].rm_eo, 20);
  785. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  786. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  787. regfree(&regex);
  788. }
  789. TEST_CASE(char_qualifier_min_max)
  790. {
  791. String pattern = "c{3,30}";
  792. regex_t regex;
  793. static constexpr int num_matches { 5 };
  794. regmatch_t matches[num_matches];
  795. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  796. EXPECT_EQ(regexec(&regex, "cc", num_matches, matches, 0), REG_NOMATCH);
  797. EXPECT_EQ(regexec(&regex, "ccc", num_matches, matches, 0), REG_NOERR);
  798. EXPECT_EQ(regexec(&regex, "cccccccccccccccccccccccccccccc", num_matches, matches, 0), REG_NOERR);
  799. EXPECT_EQ(matches[0].rm_cnt, 1);
  800. EXPECT_EQ(regexec(&regex, "ccccccccccccccccccccccccccccccc", num_matches, matches, 0), REG_NOMATCH);
  801. EXPECT_EQ(regexec(&regex, "ccccccccccccccccccccccccccccccc", num_matches, matches, REG_GLOBAL), REG_NOERR);
  802. EXPECT_EQ(regexec(&regex, "cccccccccccccccccccccccccccccccc", num_matches, matches, 0), REG_NOMATCH);
  803. regfree(&regex);
  804. }
  805. TEST_CASE(simple_bracket_chars)
  806. {
  807. String pattern = "[abc]";
  808. regex_t regex;
  809. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  810. EXPECT_EQ(regexec(&regex, "a", 0, NULL, 0), REG_NOERR);
  811. EXPECT_EQ(regexec(&regex, "b", 0, NULL, 0), REG_NOERR);
  812. EXPECT_EQ(regexec(&regex, "c", 0, NULL, 0), REG_NOERR);
  813. EXPECT_EQ(regexec(&regex, "d", 0, NULL, 0), REG_NOMATCH);
  814. EXPECT_EQ(regexec(&regex, "e", 0, NULL, 0), REG_NOMATCH);
  815. regfree(&regex);
  816. }
  817. TEST_CASE(simple_bracket_chars_inverse)
  818. {
  819. String pattern = "[^abc]";
  820. regex_t regex;
  821. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  822. EXPECT_EQ(regexec(&regex, "a", 0, NULL, 0), REG_NOMATCH);
  823. EXPECT_EQ(regexec(&regex, "b", 0, NULL, 0), REG_NOMATCH);
  824. EXPECT_EQ(regexec(&regex, "c", 0, NULL, 0), REG_NOMATCH);
  825. EXPECT_EQ(regexec(&regex, "d", 0, NULL, 0), REG_NOERR);
  826. EXPECT_EQ(regexec(&regex, "e", 0, NULL, 0), REG_NOERR);
  827. regfree(&regex);
  828. }
  829. TEST_CASE(simple_bracket_chars_range)
  830. {
  831. String pattern = "[a-d]";
  832. regex_t regex;
  833. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  834. EXPECT_EQ(regexec(&regex, "a", 0, NULL, 0), REG_NOERR);
  835. EXPECT_EQ(regexec(&regex, "b", 0, NULL, 0), REG_NOERR);
  836. EXPECT_EQ(regexec(&regex, "c", 0, NULL, 0), REG_NOERR);
  837. EXPECT_EQ(regexec(&regex, "d", 0, NULL, 0), REG_NOERR);
  838. EXPECT_EQ(regexec(&regex, "e", 0, NULL, 0), REG_NOMATCH);
  839. regfree(&regex);
  840. }
  841. TEST_CASE(simple_bracket_chars_range_inverse)
  842. {
  843. String pattern = "[^a-df-z]";
  844. regex_t regex;
  845. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  846. EXPECT_EQ(regexec(&regex, "a", 0, NULL, 0), REG_NOMATCH);
  847. EXPECT_EQ(regexec(&regex, "b", 0, NULL, 0), REG_NOMATCH);
  848. EXPECT_EQ(regexec(&regex, "c", 0, NULL, 0), REG_NOMATCH);
  849. EXPECT_EQ(regexec(&regex, "d", 0, NULL, 0), REG_NOMATCH);
  850. EXPECT_EQ(regexec(&regex, "e", 0, NULL, 0), REG_NOERR);
  851. EXPECT_EQ(regexec(&regex, "k", 0, NULL, 0), REG_NOMATCH);
  852. EXPECT_EQ(regexec(&regex, "z", 0, NULL, 0), REG_NOMATCH);
  853. regfree(&regex);
  854. }
  855. TEST_CASE(bracket_character_class_uuid)
  856. {
  857. String pattern = "^([[:xdigit:]]{8})-([[:xdigit:]]{4})-([[:xdigit:]]{4})-([[:xdigit:]]{4})-([[:xdigit:]]{12})$";
  858. regex_t regex;
  859. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  860. EXPECT_EQ(regexec(&regex, "fb9b62a2-1579-4e3a-afba-76239ccb6583", 0, NULL, 0), REG_NOERR);
  861. EXPECT_EQ(regexec(&regex, "fb9b62a2", 0, NULL, 0), REG_NOMATCH);
  862. regfree(&regex);
  863. }
  864. TEST_CASE(simple_bracket_character_class_inverse)
  865. {
  866. String pattern = "[^[:digit:]]";
  867. regex_t regex;
  868. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  869. EXPECT_EQ(regexec(&regex, "1", 0, NULL, 0), REG_NOMATCH);
  870. EXPECT_EQ(regexec(&regex, "2", 0, NULL, 0), REG_NOMATCH);
  871. EXPECT_EQ(regexec(&regex, "3", 0, NULL, 0), REG_NOMATCH);
  872. EXPECT_EQ(regexec(&regex, "d", 0, NULL, 0), REG_NOERR);
  873. EXPECT_EQ(regexec(&regex, "e", 0, NULL, 0), REG_NOERR);
  874. regfree(&regex);
  875. }
  876. TEST_CASE(email_address)
  877. {
  878. String pattern = "^[A-Z0-9a-z._%+-]{1,64}@(?:[A-Za-z0-9-]{1,63}\\.){1,125}[A-Za-z]{2,63}$";
  879. regex_t regex;
  880. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  881. EXPECT_EQ(regexec(&regex, "emanuel.sprung@gmail.com", 0, NULL, 0), REG_NOERR);
  882. EXPECT_EQ(regexec(&regex, "kling@serenityos.org", 0, NULL, 0), REG_NOERR);
  883. regfree(&regex);
  884. }
  885. TEST_CASE(error_message)
  886. {
  887. String pattern = "^[A-Z0-9[a-z._%+-]{1,64}@[A-Za-z0-9-]{1,63}\\.{1,125}[A-Za-z]{2,63}$";
  888. regex_t regex;
  889. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_EBRACK);
  890. EXPECT_EQ(regexec(&regex, "asdf@asdf.com", 0, NULL, 0), REG_EBRACK);
  891. char buf[1024];
  892. size_t buflen = 1024;
  893. auto len = regerror(0, &regex, buf, buflen);
  894. String expected = "Error during parsing of regular expression:\n ^[A-Z0-9[a-z._%+-]{1,64}@[A-Za-z0-9-]{1,63}\\.{1,125}[A-Za-z]{2,63}$\n ^---- [ ] imbalance.";
  895. for (size_t i = 0; i < len; ++i) {
  896. EXPECT_EQ(buf[i], expected[i]);
  897. }
  898. regfree(&regex);
  899. }
  900. TEST_CASE(simple_ignorecase)
  901. {
  902. String pattern = "^hello friends";
  903. regex_t regex;
  904. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED | REG_NOSUB | REG_ICASE), REG_NOERR);
  905. EXPECT_EQ(regexec(&regex, "Hello Friends", 0, NULL, 0), REG_NOERR);
  906. EXPECT_EQ(regexec(&regex, "hello Friends", 0, NULL, 0), REG_NOERR);
  907. EXPECT_EQ(regexec(&regex, "hello Friends!", 0, NULL, 0), REG_NOMATCH);
  908. EXPECT_EQ(regexec(&regex, "hello Friends!", 0, NULL, REG_GLOBAL), REG_NOERR);
  909. EXPECT_EQ(regexec(&regex, "hell Friends", 0, NULL, 0), REG_NOMATCH);
  910. EXPECT_EQ(regexec(&regex, "hell Friends", 0, NULL, REG_GLOBAL), REG_NOMATCH);
  911. regfree(&regex);
  912. }
  913. TEST_CASE(simple_notbol_noteol)
  914. {
  915. String pattern = "^hello friends$";
  916. String pattern2 = "hello friends";
  917. regex_t regex, regex2;
  918. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED | REG_NOSUB | REG_ICASE), REG_NOERR);
  919. EXPECT_EQ(regcomp(&regex2, pattern2.characters(), REG_EXTENDED | REG_NOSUB | REG_ICASE), REG_NOERR);
  920. EXPECT_EQ(regexec(&regex, "hello friends", 0, NULL, REG_NOTBOL), REG_NOMATCH);
  921. EXPECT_EQ(regexec(&regex, "hello friends", 0, NULL, REG_NOTEOL), REG_NOMATCH);
  922. EXPECT_EQ(regexec(&regex, "hello friends", 0, NULL, REG_NOTBOL | REG_NOTEOL), REG_NOMATCH);
  923. EXPECT_EQ(regexec(&regex, "a hello friends b", 0, NULL, REG_NOTBOL), REG_NOMATCH);
  924. EXPECT_EQ(regexec(&regex, "a hello friends", 0, NULL, REG_NOTBOL), REG_NOMATCH);
  925. EXPECT_EQ(regexec(&regex, "a hello friends", 0, NULL, REG_NOTBOL | REG_SEARCH), REG_NOERR);
  926. EXPECT_EQ(regexec(&regex, "a hello friends b", 0, NULL, REG_NOTBOL | REG_SEARCH), REG_NOERR);
  927. EXPECT_EQ(regexec(&regex, "a hello friends b", 0, NULL, REG_NOTEOL), REG_NOMATCH);
  928. EXPECT_EQ(regexec(&regex, "hello friends b", 0, NULL, REG_NOTEOL), REG_NOMATCH);
  929. EXPECT_EQ(regexec(&regex, "hello friends b", 0, NULL, REG_NOTEOL | REG_SEARCH), REG_NOERR);
  930. EXPECT_EQ(regexec(&regex, "a hello friends b", 0, NULL, REG_NOTEOL | REG_SEARCH), REG_NOMATCH);
  931. EXPECT_EQ(regexec(&regex, "a hello friends b", 0, NULL, REG_NOTBOL | REG_NOTEOL), REG_NOMATCH);
  932. EXPECT_EQ(regexec(&regex, "a hello friends b", 0, NULL, REG_NOTBOL | REG_NOTEOL | REG_SEARCH), REG_NOMATCH);
  933. EXPECT_EQ(regexec(&regex2, "hello friends", 0, NULL, REG_NOTBOL), REG_NOMATCH);
  934. EXPECT_EQ(regexec(&regex2, "hello friends", 0, NULL, REG_NOTEOL), REG_NOMATCH);
  935. regfree(&regex);
  936. regfree(&regex2);
  937. }
  938. TEST_CASE(bre_basic)
  939. {
  940. regex_t regex;
  941. EXPECT_EQ(regcomp(&regex, "hello friends", REG_NOSUB | REG_ICASE), REG_NOERR);
  942. EXPECT_EQ(regexec(&regex, "hello friends", 0, NULL, 0), REG_NOERR);
  943. regfree(&regex);
  944. EXPECT_EQ(regcomp(&regex, "\\(15\\)\\1", REG_NOSUB | REG_ICASE), REG_NOERR);
  945. EXPECT_EQ(regexec(&regex, "1515", 0, NULL, 0), REG_NOERR);
  946. EXPECT_EQ(regexec(&regex, "55", 0, NULL, 0), REG_NOMATCH);
  947. regfree(&regex);
  948. EXPECT_EQ(regcomp(&regex, "15\\{1,2\\}", REG_NOSUB | REG_ICASE), REG_NOERR);
  949. EXPECT_EQ(regexec(&regex, "15", 0, NULL, 0), REG_NOERR);
  950. EXPECT_EQ(regexec(&regex, "1515", 0, NULL, 0), REG_NOMATCH);
  951. EXPECT_EQ(regexec(&regex, "55", 0, NULL, 0), REG_NOMATCH);
  952. regfree(&regex);
  953. EXPECT_EQ(regcomp(&regex, "15{1,2}", REG_NOSUB | REG_ICASE), REG_NOERR);
  954. EXPECT_EQ(regexec(&regex, "15{1,2}", 0, NULL, 0), REG_NOERR);
  955. regfree(&regex);
  956. }