RegexLibC.cpp 45 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159
  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 <AK/Vector.h>
  9. #include <LibC/regex.h>
  10. #include <stdio.h>
  11. TEST_CASE(catch_all)
  12. {
  13. String pattern = "^.*$";
  14. regex_t regex;
  15. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  16. EXPECT_EQ(regexec(&regex, "Hello World", 0, NULL, 0), REG_NOERR);
  17. regfree(&regex);
  18. }
  19. TEST_CASE(simple_start)
  20. {
  21. String pattern = "^hello friends";
  22. regex_t regex;
  23. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  24. EXPECT_EQ(regexec(&regex, "Hello!", 0, NULL, 0), REG_NOMATCH);
  25. EXPECT_EQ(regexec(&regex, "hello friends", 0, NULL, 0), REG_NOERR);
  26. EXPECT_EQ(regexec(&regex, "Well, hello friends", 0, NULL, 0), REG_NOMATCH);
  27. regfree(&regex);
  28. }
  29. TEST_CASE(simple_end)
  30. {
  31. String pattern = ".*hello\\.\\.\\. there$";
  32. regex_t regex;
  33. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  34. EXPECT_EQ(regexec(&regex, "Hallo", 0, NULL, 0), REG_NOMATCH);
  35. EXPECT_EQ(regexec(&regex, "I said fyhello... there", 0, NULL, 0), REG_NOERR);
  36. EXPECT_EQ(regexec(&regex, "ahello... therea", 0, NULL, 0), REG_NOMATCH);
  37. EXPECT_EQ(regexec(&regex, "hello.. there", 0, NULL, 0), REG_NOMATCH);
  38. regfree(&regex);
  39. }
  40. TEST_CASE(simple_period)
  41. {
  42. String pattern = "hello.";
  43. regex_t regex;
  44. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  45. EXPECT_EQ(regexec(&regex, "Hello1", 0, NULL, 0), REG_NOMATCH);
  46. EXPECT_EQ(regexec(&regex, "hello1", 0, NULL, 0), REG_NOERR);
  47. EXPECT_EQ(regexec(&regex, "hello2", 0, NULL, 0), REG_NOERR);
  48. EXPECT_EQ(regexec(&regex, "hello?", 0, NULL, 0), REG_NOERR);
  49. regfree(&regex);
  50. }
  51. TEST_CASE(simple_period_end)
  52. {
  53. String pattern = "hello.$";
  54. regex_t regex;
  55. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED | REG_NOSUB), REG_NOERR);
  56. EXPECT_EQ(regexec(&regex, "Hello1", 0, NULL, REG_NOSUB), REG_NOMATCH);
  57. EXPECT_EQ(regexec(&regex, "hello1hello1", 0, NULL, REG_GLOBAL), REG_NOERR);
  58. EXPECT_EQ(regexec(&regex, "hello2hell", 0, NULL, REG_GLOBAL), REG_NOMATCH);
  59. EXPECT_EQ(regexec(&regex, "hello?", 0, NULL, REG_NOSUB), REG_NOERR);
  60. regfree(&regex);
  61. }
  62. TEST_CASE(simple_escaped)
  63. {
  64. String pattern = "hello\\.";
  65. regex_t regex;
  66. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  67. EXPECT_EQ(regexec(&regex, "hello", 0, NULL, 0), REG_NOMATCH);
  68. EXPECT_EQ(regexec(&regex, "hello.", 0, NULL, 0), REG_NOERR);
  69. regfree(&regex);
  70. }
  71. TEST_CASE(simple_period2_end)
  72. {
  73. String pattern = ".*hi... there$";
  74. regex_t regex;
  75. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  76. EXPECT_EQ(regexec(&regex, "Hello there", 0, NULL, REG_GLOBAL), REG_NOMATCH);
  77. EXPECT_EQ(regexec(&regex, "I said fyhi... there", 0, NULL, REG_GLOBAL), REG_NOERR);
  78. EXPECT_EQ(regexec(&regex, "....hi... ", 0, NULL, REG_GLOBAL), REG_NOMATCH);
  79. EXPECT_EQ(regexec(&regex, "I said fyhihii there", 0, NULL, REG_GLOBAL), REG_NOERR);
  80. EXPECT_EQ(regexec(&regex, "I said fyhihi there", 0, NULL, REG_GLOBAL), REG_NOMATCH);
  81. regfree(&regex);
  82. }
  83. TEST_CASE(simple_plus)
  84. {
  85. String pattern = "a+";
  86. regex_t regex;
  87. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED | REG_NOSUB), REG_NOERR);
  88. EXPECT_EQ(regexec(&regex, "b", 0, NULL, REG_NOSUB), REG_NOMATCH);
  89. EXPECT_EQ(regexec(&regex, "a", 0, NULL, REG_NOSUB), REG_NOERR);
  90. EXPECT_EQ(regexec(&regex, "aaaaaabbbbb", 0, NULL, REG_GLOBAL), REG_NOERR);
  91. EXPECT_EQ(regexec(&regex, "aaaaaaaaaaa", 0, NULL, REG_GLOBAL), REG_NOERR);
  92. regfree(&regex);
  93. }
  94. TEST_CASE(simple_questionmark)
  95. {
  96. String pattern = "da?d";
  97. regex_t regex;
  98. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  99. EXPECT_EQ(regexec(&regex, "a", 0, NULL, REG_GLOBAL), REG_NOMATCH);
  100. EXPECT_EQ(regexec(&regex, "daa", 0, NULL, REG_GLOBAL), REG_NOMATCH);
  101. EXPECT_EQ(regexec(&regex, "ddddd", 0, NULL, REG_GLOBAL), REG_NOERR);
  102. EXPECT_EQ(regexec(&regex, "dd", 0, NULL, REG_GLOBAL), REG_NOERR);
  103. EXPECT_EQ(regexec(&regex, "dad", 0, NULL, REG_GLOBAL), REG_NOERR);
  104. EXPECT_EQ(regexec(&regex, "dada", 0, NULL, REG_GLOBAL), REG_NOERR);
  105. EXPECT_EQ(regexec(&regex, "adadaa", 0, NULL, REG_GLOBAL), REG_NOERR);
  106. regfree(&regex);
  107. }
  108. TEST_CASE(simple_questionmark_matchall)
  109. {
  110. String pattern = "da?d";
  111. regex_t regex;
  112. static constexpr int num_matches { 5 };
  113. regmatch_t matches[num_matches];
  114. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  115. EXPECT_EQ(regexec(&regex, "a", num_matches, matches, REG_GLOBAL), REG_NOMATCH);
  116. EXPECT_EQ(matches[0].rm_cnt, 0);
  117. EXPECT_EQ(regexec(&regex, "daa", num_matches, matches, REG_GLOBAL), REG_NOMATCH);
  118. EXPECT_EQ(matches[0].rm_cnt, 0);
  119. EXPECT_EQ(regexec(&regex, "ddddd", num_matches, matches, REG_GLOBAL), REG_NOERR);
  120. EXPECT_EQ(matches[0].rm_cnt, 2);
  121. EXPECT_EQ(matches[0].rm_so, 0);
  122. EXPECT_EQ(matches[0].rm_eo, 2);
  123. EXPECT_EQ(matches[1].rm_so, 2);
  124. EXPECT_EQ(matches[1].rm_eo, 4);
  125. EXPECT_EQ(regexec(&regex, "dd", num_matches, matches, REG_GLOBAL), REG_NOERR);
  126. EXPECT_EQ(matches[0].rm_cnt, 1);
  127. EXPECT_EQ(regexec(&regex, "dad", num_matches, matches, REG_GLOBAL), REG_NOERR);
  128. EXPECT_EQ(matches[0].rm_cnt, 1);
  129. EXPECT_EQ(regexec(&regex, "dada", num_matches, matches, REG_GLOBAL), REG_NOERR);
  130. EXPECT_EQ(matches[0].rm_cnt, 1);
  131. EXPECT_EQ(regexec(&regex, "adadaa", num_matches, matches, REG_GLOBAL), REG_NOERR);
  132. EXPECT_EQ(matches[0].rm_cnt, 1);
  133. regfree(&regex);
  134. }
  135. TEST_CASE(character_class)
  136. {
  137. String pattern = "[[:alpha:]]";
  138. regex_t regex;
  139. static constexpr int num_matches { 5 };
  140. regmatch_t matches[num_matches];
  141. String haystack = "[Window]\nOpacity=255\nAudibleBeep=0\n";
  142. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  143. EXPECT_EQ(regexec(&regex, haystack.characters(), num_matches, matches, 0), REG_NOMATCH);
  144. EXPECT_EQ(matches[0].rm_cnt, 0);
  145. EXPECT_EQ(regexec(&regex, haystack.characters(), num_matches, matches, REG_GLOBAL), REG_NOERR);
  146. EXPECT_EQ(matches[0].rm_cnt, 24);
  147. EXPECT_EQ(haystack.substring_view(matches[0].rm_so, matches[0].rm_eo - matches[0].rm_so), "W");
  148. EXPECT_EQ(haystack.substring_view(matches[1].rm_so, matches[1].rm_eo - matches[1].rm_so), "i");
  149. regfree(&regex);
  150. }
  151. TEST_CASE(character_class2)
  152. {
  153. String pattern = "[[:alpha:]]*=([[:digit:]]*)|\\[(.*)\\]";
  154. regex_t regex;
  155. static constexpr int num_matches { 9 };
  156. regmatch_t matches[num_matches];
  157. String haystack = "[Window]\nOpacity=255\nAudibleBeep=0\n";
  158. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED | REG_NEWLINE), REG_NOERR);
  159. EXPECT_EQ(regexec(&regex, haystack.characters(), num_matches, matches, 0), REG_NOERR);
  160. EXPECT_EQ(matches[0].rm_cnt, 3);
  161. #if 0
  162. for (int i = 0; i < num_matches; ++i) {
  163. 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);
  164. fprintf(stderr, "haystack length: %lu\n", haystack.length());
  165. if (matches[i].rm_so != -1)
  166. fprintf(stderr, "%s\n", haystack.substring_view(matches[i].rm_so, matches[i].rm_eo - matches[i].rm_so).to_string().characters());
  167. }
  168. #endif
  169. EXPECT_EQ(haystack.substring_view(matches[0].rm_so, matches[0].rm_eo - matches[0].rm_so), "[Window]");
  170. EXPECT_EQ(matches[1].rm_so, -1);
  171. EXPECT_EQ(matches[1].rm_eo, -1);
  172. EXPECT_EQ(matches[1].rm_cnt, 0);
  173. EXPECT_EQ(haystack.substring_view(matches[2].rm_so, matches[2].rm_eo - matches[2].rm_so), "Window");
  174. EXPECT_EQ(haystack.substring_view(matches[3].rm_so, matches[3].rm_eo - matches[3].rm_so), "Opacity=255");
  175. EXPECT_EQ(haystack.substring_view(matches[4].rm_so, matches[4].rm_eo - matches[4].rm_so), "255");
  176. EXPECT_EQ(matches[5].rm_so, -1);
  177. EXPECT_EQ(matches[5].rm_eo, -1);
  178. EXPECT_EQ(matches[5].rm_cnt, 0);
  179. EXPECT_EQ(haystack.substring_view(matches[6].rm_so, matches[6].rm_eo - matches[6].rm_so), "AudibleBeep=0");
  180. EXPECT_EQ(haystack.substring_view(matches[7].rm_so, matches[7].rm_eo - matches[7].rm_so), "0");
  181. EXPECT_EQ(matches[8].rm_so, -1);
  182. EXPECT_EQ(matches[8].rm_eo, -1);
  183. EXPECT_EQ(matches[8].rm_cnt, 0);
  184. regfree(&regex);
  185. }
  186. TEST_CASE(escaped_char_questionmark)
  187. {
  188. String pattern = "This\\.?And\\.?That";
  189. regex_t regex;
  190. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  191. EXPECT_EQ(regexec(&regex, "ThisAndThat", 0, NULL, 0), REG_NOERR);
  192. EXPECT_EQ(regexec(&regex, "This.And.That", 0, NULL, 0), REG_NOERR);
  193. EXPECT_EQ(regexec(&regex, "This And That", 0, NULL, 0), REG_NOMATCH);
  194. EXPECT_EQ(regexec(&regex, "This..And..That", 0, NULL, 0), REG_NOMATCH);
  195. regfree(&regex);
  196. }
  197. TEST_CASE(char_qualifier_asterisk)
  198. {
  199. String pattern = "regex*";
  200. regex_t regex;
  201. static constexpr int num_matches { 5 };
  202. regmatch_t matches[num_matches];
  203. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  204. EXPECT_EQ(regexec(&regex, "#include <regex.h>", num_matches, matches, REG_GLOBAL), REG_NOERR);
  205. EXPECT_EQ(matches[0].rm_cnt, 1);
  206. regfree(&regex);
  207. }
  208. TEST_CASE(char_utf8)
  209. {
  210. String pattern = "😀";
  211. regex_t regex;
  212. static constexpr int num_matches { 5 };
  213. regmatch_t matches[num_matches];
  214. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  215. EXPECT_EQ(regexec(&regex, "Привет, мир! 😀 γειά σου κόσμος 😀 こんにちは世界", num_matches, matches, REG_GLOBAL), REG_NOERR);
  216. EXPECT_EQ(matches[0].rm_cnt, 2);
  217. regfree(&regex);
  218. }
  219. TEST_CASE(parens)
  220. {
  221. String pattern = "test(hello)test";
  222. regex_t regex;
  223. static constexpr int num_matches { 5 };
  224. regmatch_t matches[num_matches];
  225. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  226. EXPECT_EQ(regexec(&regex, "testhellotest", num_matches, matches, 0), REG_NOERR);
  227. EXPECT_EQ(matches[0].rm_cnt, 1);
  228. EXPECT_EQ(matches[0].rm_so, 0);
  229. EXPECT_EQ(matches[0].rm_eo, 13);
  230. EXPECT_EQ(matches[1].rm_so, 4);
  231. EXPECT_EQ(matches[1].rm_eo, 9);
  232. regfree(&regex);
  233. }
  234. TEST_CASE(parser_error_parens)
  235. {
  236. String pattern = "test()test";
  237. regex_t regex;
  238. static constexpr int num_matches { 5 };
  239. regmatch_t matches[num_matches];
  240. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_EMPTY_EXPR);
  241. EXPECT_EQ(regexec(&regex, "testhellotest", num_matches, matches, 0), REG_EMPTY_EXPR);
  242. regfree(&regex);
  243. }
  244. TEST_CASE(parser_error_special_characters_used_at_wrong_place)
  245. {
  246. String pattern;
  247. regex_t regex;
  248. static constexpr int num_matches { 5 };
  249. regmatch_t matches[num_matches];
  250. Vector<char, 4> chars = { '*', '+', '?', '}' };
  251. StringBuilder b;
  252. for (auto& ch : chars) {
  253. auto error_code_to_check = ch == '}' ? REG_EBRACE : REG_BADRPT;
  254. // First in ere
  255. b.clear();
  256. b.append(ch);
  257. pattern = b.build();
  258. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), error_code_to_check);
  259. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), error_code_to_check);
  260. regfree(&regex);
  261. // After vertical line
  262. b.clear();
  263. b.append("a|"sv);
  264. b.append(ch);
  265. pattern = b.build();
  266. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), error_code_to_check);
  267. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), error_code_to_check);
  268. regfree(&regex);
  269. // After circumflex
  270. b.clear();
  271. b.append('^');
  272. b.append(ch);
  273. pattern = b.build();
  274. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), error_code_to_check);
  275. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), error_code_to_check);
  276. regfree(&regex);
  277. // After dollar
  278. b.clear();
  279. b.append('$');
  280. b.append(ch);
  281. pattern = b.build();
  282. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), error_code_to_check);
  283. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), error_code_to_check);
  284. regfree(&regex);
  285. // After left parens
  286. b.clear();
  287. b.append('(');
  288. b.append(ch);
  289. b.append(')');
  290. pattern = b.build();
  291. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), error_code_to_check);
  292. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), error_code_to_check);
  293. regfree(&regex);
  294. }
  295. }
  296. TEST_CASE(parser_error_vertical_line_used_at_wrong_place)
  297. {
  298. String pattern;
  299. regex_t regex;
  300. static constexpr int num_matches { 5 };
  301. regmatch_t matches[num_matches];
  302. // First in ere
  303. pattern = "|asdf";
  304. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_EMPTY_EXPR);
  305. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), REG_EMPTY_EXPR);
  306. regfree(&regex);
  307. // Last in ere
  308. pattern = "asdf|";
  309. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_EMPTY_EXPR);
  310. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), REG_EMPTY_EXPR);
  311. regfree(&regex);
  312. // After left parens
  313. pattern = "(|asdf)";
  314. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_EMPTY_EXPR);
  315. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), REG_EMPTY_EXPR);
  316. regfree(&regex);
  317. // Proceed right parens
  318. pattern = "(asdf)|";
  319. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_EMPTY_EXPR);
  320. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), REG_EMPTY_EXPR);
  321. regfree(&regex);
  322. }
  323. TEST_CASE(parens_qualifier_questionmark)
  324. {
  325. String pattern = "test(hello)?test";
  326. regex_t regex;
  327. static constexpr int num_matches { 5 };
  328. regmatch_t matches[num_matches];
  329. char const* match_str;
  330. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  331. match_str = "testtest";
  332. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  333. EXPECT_EQ(matches[0].rm_cnt, 1);
  334. EXPECT_EQ(matches[0].rm_so, 0);
  335. EXPECT_EQ(matches[0].rm_eo, 8);
  336. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testtest");
  337. match_str = "testhellotest";
  338. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  339. EXPECT_EQ(matches[0].rm_cnt, 1);
  340. EXPECT_EQ(matches[0].rm_so, 0);
  341. EXPECT_EQ(matches[0].rm_eo, 13);
  342. EXPECT_EQ(matches[1].rm_so, 4);
  343. EXPECT_EQ(matches[1].rm_eo, 9);
  344. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testhellotest");
  345. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  346. regfree(&regex);
  347. }
  348. TEST_CASE(parens_qualifier_asterisk)
  349. {
  350. String pattern = "test(hello)*test";
  351. regex_t regex;
  352. static constexpr int num_matches { 6 };
  353. regmatch_t matches[num_matches];
  354. char const* match_str;
  355. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  356. match_str = "testtest";
  357. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  358. EXPECT_EQ(matches[0].rm_cnt, 1);
  359. EXPECT_EQ(matches[0].rm_so, 0);
  360. EXPECT_EQ(matches[0].rm_eo, 8);
  361. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testtest");
  362. match_str = "testhellohellotest";
  363. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  364. EXPECT_EQ(matches[0].rm_cnt, 1);
  365. EXPECT_EQ(matches[0].rm_so, 0);
  366. EXPECT_EQ(matches[0].rm_eo, 18);
  367. EXPECT_EQ(matches[1].rm_so, 9);
  368. EXPECT_EQ(matches[1].rm_eo, 14);
  369. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testhellohellotest");
  370. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  371. match_str = "testhellohellotest, testhellotest";
  372. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  373. EXPECT_EQ(matches[0].rm_cnt, 2);
  374. EXPECT_EQ(matches[0].rm_so, 0);
  375. EXPECT_EQ(matches[0].rm_eo, 18);
  376. EXPECT_EQ(matches[1].rm_so, 9);
  377. EXPECT_EQ(matches[1].rm_eo, 14);
  378. EXPECT_EQ(matches[2].rm_so, 20);
  379. EXPECT_EQ(matches[2].rm_eo, 33);
  380. EXPECT_EQ(matches[3].rm_so, 24);
  381. EXPECT_EQ(matches[3].rm_eo, 29);
  382. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testhellohellotest");
  383. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  384. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "testhellotest");
  385. EXPECT_EQ(StringView(&match_str[matches[3].rm_so], matches[3].rm_eo - matches[3].rm_so), "hello");
  386. regfree(&regex);
  387. }
  388. TEST_CASE(parens_qualifier_asterisk_2)
  389. {
  390. String pattern = "test(.*)test";
  391. regex_t regex;
  392. static constexpr int num_matches { 6 };
  393. regmatch_t matches[num_matches];
  394. char const* match_str;
  395. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  396. match_str = "testasdftest";
  397. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  398. EXPECT_EQ(matches[0].rm_cnt, 1);
  399. EXPECT_EQ(matches[0].rm_so, 0);
  400. EXPECT_EQ(matches[0].rm_eo, 12);
  401. EXPECT_EQ(matches[1].rm_so, 4);
  402. EXPECT_EQ(matches[1].rm_eo, 8);
  403. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testasdftest");
  404. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "asdf");
  405. match_str = "testasdfasdftest";
  406. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  407. EXPECT_EQ(matches[0].rm_cnt, 1);
  408. EXPECT_EQ(matches[0].rm_so, 0);
  409. EXPECT_EQ(matches[0].rm_eo, 16);
  410. EXPECT_EQ(matches[1].rm_so, 4);
  411. EXPECT_EQ(matches[1].rm_eo, 12);
  412. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testasdfasdftest");
  413. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "asdfasdf");
  414. match_str = "testaaaatest, testbbbtest, testtest";
  415. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  416. EXPECT_EQ(matches[0].rm_cnt, 1);
  417. EXPECT_EQ(matches[0].rm_so, 0);
  418. EXPECT_EQ(matches[0].rm_eo, 35);
  419. EXPECT_EQ(matches[1].rm_so, 4);
  420. EXPECT_EQ(matches[1].rm_eo, 31);
  421. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testaaaatest, testbbbtest, testtest");
  422. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "aaaatest, testbbbtest, test");
  423. regfree(&regex);
  424. }
  425. TEST_CASE(mulit_parens_qualifier_too_less_result_values)
  426. {
  427. String pattern = "test(a)?(b)?(c)?test";
  428. regex_t regex;
  429. static constexpr int num_matches { 4 };
  430. regmatch_t matches[num_matches];
  431. char const* match_str;
  432. matches[3] = { -2, -2, 100 };
  433. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  434. match_str = "testabtest";
  435. EXPECT_EQ(regexec(&regex, match_str, num_matches - 1, matches, 0), REG_NOERR);
  436. EXPECT_EQ(matches[0].rm_cnt, 1);
  437. EXPECT_EQ(matches[0].rm_so, 0);
  438. EXPECT_EQ(matches[0].rm_eo, 10);
  439. EXPECT_EQ(matches[1].rm_so, 4);
  440. EXPECT_EQ(matches[1].rm_eo, 5);
  441. EXPECT_EQ(matches[2].rm_so, 5);
  442. EXPECT_EQ(matches[2].rm_eo, 6);
  443. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testabtest");
  444. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "a");
  445. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "b");
  446. EXPECT_EQ(matches[3].rm_so, -2);
  447. EXPECT_EQ(matches[3].rm_eo, -2);
  448. EXPECT_EQ(matches[3].rm_cnt, 100);
  449. match_str = "testabctest";
  450. EXPECT_EQ(regexec(&regex, match_str, num_matches - 1, matches, 0), REG_NOERR);
  451. EXPECT_EQ(matches[0].rm_cnt, 1);
  452. EXPECT_EQ(matches[0].rm_so, 0);
  453. EXPECT_EQ(matches[0].rm_eo, 11);
  454. EXPECT_EQ(matches[1].rm_so, 4);
  455. EXPECT_EQ(matches[1].rm_eo, 5);
  456. EXPECT_EQ(matches[2].rm_so, 5);
  457. EXPECT_EQ(matches[2].rm_eo, 6);
  458. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testabctest");
  459. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "a");
  460. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "b");
  461. EXPECT_EQ(matches[3].rm_so, -2);
  462. EXPECT_EQ(matches[3].rm_eo, -2);
  463. EXPECT_EQ(matches[3].rm_cnt, 100);
  464. match_str = "testabctest, testabctest";
  465. EXPECT_EQ(regexec(&regex, match_str, num_matches - 1, matches, REG_GLOBAL), REG_NOERR);
  466. EXPECT_EQ(matches[0].rm_cnt, 2);
  467. EXPECT_EQ(matches[0].rm_so, 0);
  468. EXPECT_EQ(matches[0].rm_eo, 11);
  469. EXPECT_EQ(matches[1].rm_so, 4);
  470. EXPECT_EQ(matches[1].rm_eo, 5);
  471. EXPECT_EQ(matches[2].rm_so, 5);
  472. EXPECT_EQ(matches[2].rm_eo, 6);
  473. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testabctest");
  474. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "a");
  475. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "b");
  476. EXPECT_EQ(matches[3].rm_so, -2);
  477. EXPECT_EQ(matches[3].rm_eo, -2);
  478. EXPECT_EQ(matches[3].rm_cnt, 100);
  479. regfree(&regex);
  480. }
  481. TEST_CASE(multi_parens_qualifier_questionmark)
  482. {
  483. String pattern = "test(a)?(b)?(c)?test";
  484. regex_t regex;
  485. static constexpr int num_matches { 8 };
  486. regmatch_t matches[num_matches];
  487. char const* match_str;
  488. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  489. match_str = "testtest";
  490. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  491. EXPECT_EQ(matches[0].rm_cnt, 1);
  492. EXPECT_EQ(matches[0].rm_so, 0);
  493. EXPECT_EQ(matches[0].rm_eo, 8);
  494. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testtest");
  495. match_str = "testabctest";
  496. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  497. EXPECT_EQ(matches[0].rm_cnt, 1);
  498. EXPECT_EQ(matches[0].rm_so, 0);
  499. EXPECT_EQ(matches[0].rm_eo, 11);
  500. EXPECT_EQ(matches[1].rm_so, 4);
  501. EXPECT_EQ(matches[1].rm_eo, 5);
  502. EXPECT_EQ(matches[2].rm_so, 5);
  503. EXPECT_EQ(matches[2].rm_eo, 6);
  504. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testabctest");
  505. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "a");
  506. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "b");
  507. match_str = "testabctest, testactest";
  508. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  509. EXPECT_EQ(matches[0].rm_cnt, 2);
  510. EXPECT_EQ(matches[0].rm_so, 0);
  511. EXPECT_EQ(matches[0].rm_eo, 11);
  512. EXPECT_EQ(matches[1].rm_so, 4);
  513. EXPECT_EQ(matches[1].rm_eo, 5);
  514. EXPECT_EQ(matches[2].rm_so, 5);
  515. EXPECT_EQ(matches[2].rm_eo, 6);
  516. EXPECT_EQ(matches[3].rm_so, 6);
  517. EXPECT_EQ(matches[3].rm_eo, 7);
  518. EXPECT_EQ(matches[4].rm_so, 13);
  519. EXPECT_EQ(matches[4].rm_eo, 23);
  520. EXPECT_EQ(matches[5].rm_so, 17);
  521. EXPECT_EQ(matches[5].rm_eo, 18);
  522. EXPECT_EQ(matches[6].rm_so, -1);
  523. EXPECT_EQ(matches[6].rm_eo, -1);
  524. EXPECT_EQ(matches[7].rm_so, 18);
  525. EXPECT_EQ(matches[7].rm_eo, 19);
  526. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testabctest");
  527. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "a");
  528. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "b");
  529. EXPECT_EQ(StringView(&match_str[matches[3].rm_so], matches[3].rm_eo - matches[3].rm_so), "c");
  530. EXPECT_EQ(StringView(&match_str[matches[4].rm_so], matches[4].rm_eo - matches[4].rm_so), "testactest");
  531. EXPECT_EQ(StringView(&match_str[matches[5].rm_so], matches[5].rm_eo - matches[5].rm_so), "a");
  532. EXPECT_EQ(StringView(&match_str[matches[6].rm_so], matches[6].rm_eo - matches[6].rm_so), "");
  533. EXPECT_EQ(StringView(&match_str[matches[7].rm_so], matches[7].rm_eo - matches[7].rm_so), "c");
  534. regfree(&regex);
  535. }
  536. TEST_CASE(simple_alternative)
  537. {
  538. String pattern = "test|hello|friends";
  539. regex_t regex;
  540. static constexpr int num_matches { 1 };
  541. regmatch_t matches[num_matches];
  542. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  543. EXPECT_EQ(regexec(&regex, "test", num_matches, matches, 0), REG_NOERR);
  544. EXPECT_EQ(matches[0].rm_cnt, 1);
  545. EXPECT_EQ(matches[0].rm_so, 0);
  546. EXPECT_EQ(matches[0].rm_eo, 4);
  547. EXPECT_EQ(regexec(&regex, "hello", num_matches, matches, 0), REG_NOERR);
  548. EXPECT_EQ(matches[0].rm_cnt, 1);
  549. EXPECT_EQ(matches[0].rm_so, 0);
  550. EXPECT_EQ(matches[0].rm_eo, 5);
  551. EXPECT_EQ(regexec(&regex, "friends", num_matches, matches, 0), REG_NOERR);
  552. EXPECT_EQ(matches[0].rm_cnt, 1);
  553. EXPECT_EQ(matches[0].rm_so, 0);
  554. EXPECT_EQ(matches[0].rm_eo, 7);
  555. regfree(&regex);
  556. }
  557. TEST_CASE(alternative_match_groups)
  558. {
  559. String pattern = "test(a)?(b)?|hello ?(dear|my)? friends";
  560. regex_t regex;
  561. static constexpr int num_matches { 8 };
  562. regmatch_t matches[num_matches];
  563. char const* match_str;
  564. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  565. match_str = "test";
  566. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  567. EXPECT_EQ(matches[0].rm_cnt, 1);
  568. EXPECT_EQ(matches[0].rm_so, 0);
  569. EXPECT_EQ(matches[0].rm_eo, 4);
  570. EXPECT_EQ(matches[1].rm_so, -1);
  571. EXPECT_EQ(matches[1].rm_eo, -1);
  572. EXPECT_EQ(matches[2].rm_so, -1);
  573. EXPECT_EQ(matches[2].rm_eo, -1);
  574. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "test");
  575. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "");
  576. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "");
  577. match_str = "testa";
  578. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  579. EXPECT_EQ(matches[0].rm_cnt, 1);
  580. EXPECT_EQ(matches[0].rm_so, 0);
  581. EXPECT_EQ(matches[0].rm_eo, 5);
  582. EXPECT_EQ(matches[1].rm_so, 4);
  583. EXPECT_EQ(matches[1].rm_eo, 5);
  584. EXPECT_EQ(matches[2].rm_so, -1);
  585. EXPECT_EQ(matches[2].rm_eo, -1);
  586. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testa");
  587. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "a");
  588. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "");
  589. match_str = "testb";
  590. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  591. EXPECT_EQ(matches[0].rm_cnt, 1);
  592. EXPECT_EQ(matches[0].rm_so, 0);
  593. EXPECT_EQ(matches[0].rm_eo, 5);
  594. EXPECT_EQ(matches[1].rm_so, -1);
  595. EXPECT_EQ(matches[1].rm_eo, -1);
  596. EXPECT_EQ(matches[2].rm_so, 4);
  597. EXPECT_EQ(matches[2].rm_eo, 5);
  598. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "testb");
  599. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "");
  600. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "b");
  601. match_str = "hello friends";
  602. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  603. EXPECT_EQ(matches[0].rm_cnt, 1);
  604. EXPECT_EQ(matches[0].rm_so, 0);
  605. EXPECT_EQ(matches[0].rm_eo, 13);
  606. EXPECT_EQ(matches[1].rm_so, -1);
  607. EXPECT_EQ(matches[1].rm_eo, -1);
  608. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hello friends");
  609. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "");
  610. match_str = "hello dear friends";
  611. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  612. EXPECT_EQ(matches[0].rm_cnt, 1);
  613. EXPECT_EQ(matches[0].rm_so, 0);
  614. EXPECT_EQ(matches[0].rm_eo, 18);
  615. EXPECT_EQ(matches[1].rm_so, -1);
  616. EXPECT_EQ(matches[1].rm_eo, -1);
  617. EXPECT_EQ(matches[2].rm_so, -1);
  618. EXPECT_EQ(matches[2].rm_eo, -1);
  619. EXPECT_EQ(matches[3].rm_so, 6);
  620. EXPECT_EQ(matches[3].rm_eo, 10);
  621. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hello dear friends");
  622. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "");
  623. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "");
  624. EXPECT_EQ(StringView(&match_str[matches[3].rm_so], matches[3].rm_eo - matches[3].rm_so), "dear");
  625. match_str = "hello my friends";
  626. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  627. EXPECT_EQ(matches[0].rm_cnt, 1);
  628. EXPECT_EQ(matches[0].rm_so, 0);
  629. EXPECT_EQ(matches[0].rm_eo, 16);
  630. EXPECT_EQ(matches[1].rm_so, -1);
  631. EXPECT_EQ(matches[1].rm_eo, -1);
  632. EXPECT_EQ(matches[2].rm_so, -1);
  633. EXPECT_EQ(matches[2].rm_eo, -1);
  634. EXPECT_EQ(matches[3].rm_so, 6);
  635. EXPECT_EQ(matches[3].rm_eo, 8);
  636. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hello my friends");
  637. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "");
  638. EXPECT_EQ(StringView(&match_str[matches[2].rm_so], matches[2].rm_eo - matches[2].rm_so), "");
  639. EXPECT_EQ(StringView(&match_str[matches[3].rm_so], matches[3].rm_eo - matches[3].rm_so), "my");
  640. match_str = "testabc";
  641. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOMATCH);
  642. EXPECT_EQ(matches[0].rm_cnt, 0);
  643. EXPECT_EQ(matches[0].rm_so, -1);
  644. EXPECT_EQ(matches[0].rm_eo, -1);
  645. match_str = "hello test friends";
  646. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOMATCH);
  647. EXPECT_EQ(matches[0].rm_cnt, 0);
  648. EXPECT_EQ(matches[0].rm_so, -1);
  649. EXPECT_EQ(matches[0].rm_eo, -1);
  650. regfree(&regex);
  651. }
  652. TEST_CASE(parens_qualifier_exact)
  653. {
  654. String pattern = "(hello){3}";
  655. regex_t regex;
  656. static constexpr int num_matches { 5 };
  657. regmatch_t matches[num_matches];
  658. char const* match_str;
  659. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  660. match_str = "hello";
  661. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOMATCH);
  662. EXPECT_EQ(matches[0].rm_cnt, 0);
  663. match_str = "hellohellohello";
  664. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  665. EXPECT_EQ(matches[0].rm_cnt, 1);
  666. EXPECT_EQ(matches[0].rm_so, 0);
  667. EXPECT_EQ(matches[0].rm_eo, 15);
  668. EXPECT_EQ(matches[1].rm_so, 10);
  669. EXPECT_EQ(matches[1].rm_eo, 15);
  670. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  671. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  672. match_str = "hellohellohellohello";
  673. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  674. EXPECT_EQ(matches[0].rm_cnt, 1);
  675. EXPECT_EQ(matches[0].rm_so, 0);
  676. EXPECT_EQ(matches[0].rm_eo, 15);
  677. EXPECT_EQ(matches[1].rm_so, 10);
  678. EXPECT_EQ(matches[1].rm_eo, 15);
  679. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  680. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  681. match_str = "test hellohellohello";
  682. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  683. EXPECT_EQ(matches[0].rm_cnt, 1);
  684. EXPECT_EQ(matches[0].rm_so, 5);
  685. EXPECT_EQ(matches[0].rm_eo, 20);
  686. EXPECT_EQ(matches[1].rm_so, 15);
  687. EXPECT_EQ(matches[1].rm_eo, 20);
  688. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  689. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  690. regfree(&regex);
  691. }
  692. TEST_CASE(parens_qualifier_minimum)
  693. {
  694. String pattern = "(hello){3,}";
  695. regex_t regex;
  696. static constexpr int num_matches { 5 };
  697. regmatch_t matches[num_matches];
  698. char const* match_str;
  699. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  700. match_str = "hello";
  701. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOMATCH);
  702. EXPECT_EQ(matches[0].rm_cnt, 0);
  703. match_str = "hellohellohello";
  704. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  705. EXPECT_EQ(matches[0].rm_cnt, 1);
  706. EXPECT_EQ(matches[0].rm_so, 0);
  707. EXPECT_EQ(matches[0].rm_eo, 15);
  708. EXPECT_EQ(matches[1].rm_so, 10);
  709. EXPECT_EQ(matches[1].rm_eo, 15);
  710. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  711. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  712. match_str = "hellohellohellohello";
  713. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_SEARCH), REG_NOERR);
  714. EXPECT_EQ(matches[0].rm_cnt, 1);
  715. EXPECT_EQ(matches[0].rm_so, 0);
  716. EXPECT_EQ(matches[0].rm_eo, 20);
  717. EXPECT_EQ(matches[1].rm_so, 15);
  718. EXPECT_EQ(matches[1].rm_eo, 20);
  719. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohellohello");
  720. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  721. match_str = "test hellohellohello";
  722. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  723. EXPECT_EQ(matches[0].rm_cnt, 1);
  724. EXPECT_EQ(matches[0].rm_so, 5);
  725. EXPECT_EQ(matches[0].rm_eo, 20);
  726. EXPECT_EQ(matches[1].rm_so, 15);
  727. EXPECT_EQ(matches[1].rm_eo, 20);
  728. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  729. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  730. match_str = "test hellohellohellohello";
  731. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  732. EXPECT_EQ(matches[0].rm_cnt, 1);
  733. EXPECT_EQ(matches[0].rm_so, 5);
  734. EXPECT_EQ(matches[0].rm_eo, 25);
  735. EXPECT_EQ(matches[1].rm_so, 20);
  736. EXPECT_EQ(matches[1].rm_eo, 25);
  737. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohellohello");
  738. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  739. regfree(&regex);
  740. }
  741. TEST_CASE(parens_qualifier_maximum)
  742. {
  743. String pattern = "(hello){2,3}";
  744. regex_t regex;
  745. static constexpr int num_matches { 5 };
  746. regmatch_t matches[num_matches];
  747. char const* match_str;
  748. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  749. match_str = "hello";
  750. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOMATCH);
  751. EXPECT_EQ(matches[0].rm_cnt, 0);
  752. match_str = "hellohellohello";
  753. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, 0), REG_NOERR);
  754. EXPECT_EQ(matches[0].rm_cnt, 1);
  755. EXPECT_EQ(matches[0].rm_so, 0);
  756. EXPECT_EQ(matches[0].rm_eo, 15);
  757. EXPECT_EQ(matches[1].rm_so, 10);
  758. EXPECT_EQ(matches[1].rm_eo, 15);
  759. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  760. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  761. match_str = "hellohellohellohello";
  762. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  763. EXPECT_EQ(matches[0].rm_cnt, 1);
  764. EXPECT_EQ(matches[0].rm_so, 0);
  765. EXPECT_EQ(matches[0].rm_eo, 15);
  766. EXPECT_EQ(matches[1].rm_so, 10);
  767. EXPECT_EQ(matches[1].rm_eo, 15);
  768. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  769. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  770. match_str = "test hellohellohello";
  771. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  772. EXPECT_EQ(matches[0].rm_cnt, 1);
  773. EXPECT_EQ(matches[0].rm_so, 5);
  774. EXPECT_EQ(matches[0].rm_eo, 20);
  775. EXPECT_EQ(matches[1].rm_so, 15);
  776. EXPECT_EQ(matches[1].rm_eo, 20);
  777. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  778. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  779. match_str = "test hellohellohellohello";
  780. EXPECT_EQ(regexec(&regex, match_str, num_matches, matches, REG_GLOBAL), REG_NOERR);
  781. EXPECT_EQ(matches[0].rm_cnt, 1);
  782. EXPECT_EQ(matches[0].rm_so, 5);
  783. EXPECT_EQ(matches[0].rm_eo, 20);
  784. EXPECT_EQ(matches[1].rm_so, 15);
  785. EXPECT_EQ(matches[1].rm_eo, 20);
  786. EXPECT_EQ(StringView(&match_str[matches[0].rm_so], matches[0].rm_eo - matches[0].rm_so), "hellohellohello");
  787. EXPECT_EQ(StringView(&match_str[matches[1].rm_so], matches[1].rm_eo - matches[1].rm_so), "hello");
  788. regfree(&regex);
  789. }
  790. TEST_CASE(char_qualifier_min_max)
  791. {
  792. String pattern = "c{3,30}";
  793. regex_t regex;
  794. static constexpr int num_matches { 5 };
  795. regmatch_t matches[num_matches];
  796. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  797. EXPECT_EQ(regexec(&regex, "cc", num_matches, matches, 0), REG_NOMATCH);
  798. EXPECT_EQ(regexec(&regex, "ccc", num_matches, matches, 0), REG_NOERR);
  799. EXPECT_EQ(regexec(&regex, "cccccccccccccccccccccccccccccc", num_matches, matches, 0), REG_NOERR);
  800. EXPECT_EQ(matches[0].rm_cnt, 1);
  801. EXPECT_EQ(regexec(&regex, "ccccccccccccccccccccccccccccccc", num_matches, matches, 0), REG_NOMATCH);
  802. EXPECT_EQ(regexec(&regex, "ccccccccccccccccccccccccccccccc", num_matches, matches, REG_GLOBAL), REG_NOERR);
  803. EXPECT_EQ(regexec(&regex, "cccccccccccccccccccccccccccccccc", num_matches, matches, 0), REG_NOMATCH);
  804. regfree(&regex);
  805. }
  806. TEST_CASE(simple_bracket_chars)
  807. {
  808. String pattern = "[abc]";
  809. regex_t regex;
  810. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  811. EXPECT_EQ(regexec(&regex, "a", 0, NULL, 0), REG_NOERR);
  812. EXPECT_EQ(regexec(&regex, "b", 0, NULL, 0), REG_NOERR);
  813. EXPECT_EQ(regexec(&regex, "c", 0, NULL, 0), REG_NOERR);
  814. EXPECT_EQ(regexec(&regex, "d", 0, NULL, 0), REG_NOMATCH);
  815. EXPECT_EQ(regexec(&regex, "e", 0, NULL, 0), REG_NOMATCH);
  816. regfree(&regex);
  817. }
  818. TEST_CASE(simple_bracket_chars_inverse)
  819. {
  820. String pattern = "[^abc]";
  821. regex_t regex;
  822. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  823. EXPECT_EQ(regexec(&regex, "a", 0, NULL, 0), REG_NOMATCH);
  824. EXPECT_EQ(regexec(&regex, "b", 0, NULL, 0), REG_NOMATCH);
  825. EXPECT_EQ(regexec(&regex, "c", 0, NULL, 0), REG_NOMATCH);
  826. EXPECT_EQ(regexec(&regex, "d", 0, NULL, 0), REG_NOERR);
  827. EXPECT_EQ(regexec(&regex, "e", 0, NULL, 0), REG_NOERR);
  828. regfree(&regex);
  829. }
  830. TEST_CASE(simple_bracket_chars_range)
  831. {
  832. String pattern = "[a-d]";
  833. regex_t regex;
  834. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  835. EXPECT_EQ(regexec(&regex, "a", 0, NULL, 0), REG_NOERR);
  836. EXPECT_EQ(regexec(&regex, "b", 0, NULL, 0), REG_NOERR);
  837. EXPECT_EQ(regexec(&regex, "c", 0, NULL, 0), REG_NOERR);
  838. EXPECT_EQ(regexec(&regex, "d", 0, NULL, 0), REG_NOERR);
  839. EXPECT_EQ(regexec(&regex, "e", 0, NULL, 0), REG_NOMATCH);
  840. regfree(&regex);
  841. }
  842. TEST_CASE(simple_bracket_chars_range_inverse)
  843. {
  844. String pattern = "[^a-df-z]";
  845. regex_t regex;
  846. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  847. EXPECT_EQ(regexec(&regex, "a", 0, NULL, 0), REG_NOMATCH);
  848. EXPECT_EQ(regexec(&regex, "b", 0, NULL, 0), REG_NOMATCH);
  849. EXPECT_EQ(regexec(&regex, "c", 0, NULL, 0), REG_NOMATCH);
  850. EXPECT_EQ(regexec(&regex, "d", 0, NULL, 0), REG_NOMATCH);
  851. EXPECT_EQ(regexec(&regex, "e", 0, NULL, 0), REG_NOERR);
  852. EXPECT_EQ(regexec(&regex, "k", 0, NULL, 0), REG_NOMATCH);
  853. EXPECT_EQ(regexec(&regex, "z", 0, NULL, 0), REG_NOMATCH);
  854. regfree(&regex);
  855. }
  856. TEST_CASE(bracket_character_class_uuid)
  857. {
  858. String pattern = "^([[:xdigit:]]{8})-([[:xdigit:]]{4})-([[:xdigit:]]{4})-([[:xdigit:]]{4})-([[:xdigit:]]{12})$";
  859. regex_t regex;
  860. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  861. EXPECT_EQ(regexec(&regex, "fb9b62a2-1579-4e3a-afba-76239ccb6583", 0, NULL, 0), REG_NOERR);
  862. EXPECT_EQ(regexec(&regex, "fb9b62a2", 0, NULL, 0), REG_NOMATCH);
  863. regfree(&regex);
  864. }
  865. TEST_CASE(simple_bracket_character_class_inverse)
  866. {
  867. String pattern = "[^[:digit:]]";
  868. regex_t regex;
  869. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  870. EXPECT_EQ(regexec(&regex, "1", 0, NULL, 0), REG_NOMATCH);
  871. EXPECT_EQ(regexec(&regex, "2", 0, NULL, 0), REG_NOMATCH);
  872. EXPECT_EQ(regexec(&regex, "3", 0, NULL, 0), REG_NOMATCH);
  873. EXPECT_EQ(regexec(&regex, "d", 0, NULL, 0), REG_NOERR);
  874. EXPECT_EQ(regexec(&regex, "e", 0, NULL, 0), REG_NOERR);
  875. regfree(&regex);
  876. }
  877. TEST_CASE(email_address)
  878. {
  879. String pattern = "^[A-Z0-9a-z._%+-]{1,64}@(?:[A-Za-z0-9-]{1,63}\\.){1,125}[A-Za-z]{2,63}$";
  880. regex_t regex;
  881. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_NOERR);
  882. EXPECT_EQ(regexec(&regex, "emanuel.sprung@gmail.com", 0, NULL, 0), REG_NOERR);
  883. EXPECT_EQ(regexec(&regex, "kling@serenityos.org", 0, NULL, 0), REG_NOERR);
  884. regfree(&regex);
  885. }
  886. TEST_CASE(error_message)
  887. {
  888. String pattern = "^[A-Z0-9[a-z._%+-]{1,64}@[A-Za-z0-9-]{1,63}\\.{1,125}[A-Za-z]{2,63}$";
  889. regex_t regex;
  890. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED), REG_EBRACK);
  891. EXPECT_EQ(regexec(&regex, "asdf@asdf.com", 0, NULL, 0), REG_EBRACK);
  892. char buf[1024];
  893. size_t buflen = 1024;
  894. auto len = regerror(0, &regex, buf, buflen);
  895. 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}$\n ^---- [ ] imbalance.";
  896. for (size_t i = 0; i < len; ++i) {
  897. EXPECT_EQ(buf[i], expected[i]);
  898. }
  899. regfree(&regex);
  900. }
  901. TEST_CASE(simple_ignorecase)
  902. {
  903. String pattern = "^hello friends";
  904. regex_t regex;
  905. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED | REG_NOSUB | REG_ICASE), REG_NOERR);
  906. EXPECT_EQ(regexec(&regex, "Hello Friends", 0, NULL, 0), REG_NOERR);
  907. EXPECT_EQ(regexec(&regex, "hello Friends", 0, NULL, 0), REG_NOERR);
  908. EXPECT_EQ(regexec(&regex, "hello Friends!", 0, NULL, 0), REG_NOMATCH);
  909. EXPECT_EQ(regexec(&regex, "hello Friends!", 0, NULL, REG_GLOBAL), REG_NOERR);
  910. EXPECT_EQ(regexec(&regex, "hell Friends", 0, NULL, 0), REG_NOMATCH);
  911. EXPECT_EQ(regexec(&regex, "hell Friends", 0, NULL, REG_GLOBAL), REG_NOMATCH);
  912. regfree(&regex);
  913. }
  914. TEST_CASE(simple_notbol_noteol)
  915. {
  916. String pattern = "^hello friends$";
  917. String pattern2 = "hello friends";
  918. regex_t regex, regex2;
  919. EXPECT_EQ(regcomp(&regex, pattern.characters(), REG_EXTENDED | REG_NOSUB | REG_ICASE), REG_NOERR);
  920. EXPECT_EQ(regcomp(&regex2, pattern2.characters(), REG_EXTENDED | REG_NOSUB | REG_ICASE), REG_NOERR);
  921. EXPECT_EQ(regexec(&regex, "hello friends", 0, NULL, REG_NOTBOL), REG_NOMATCH);
  922. EXPECT_EQ(regexec(&regex, "hello friends", 0, NULL, REG_NOTEOL), REG_NOMATCH);
  923. EXPECT_EQ(regexec(&regex, "hello friends", 0, NULL, REG_NOTBOL | REG_NOTEOL), REG_NOMATCH);
  924. EXPECT_EQ(regexec(&regex, "a hello friends b", 0, NULL, REG_NOTBOL), REG_NOMATCH);
  925. EXPECT_EQ(regexec(&regex, "a hello friends", 0, NULL, REG_NOTBOL), REG_NOMATCH);
  926. EXPECT_EQ(regexec(&regex, "a hello friends", 0, NULL, REG_NOTBOL | REG_SEARCH), REG_NOERR);
  927. EXPECT_EQ(regexec(&regex, "a hello friends b", 0, NULL, REG_NOTBOL | REG_SEARCH), REG_NOERR);
  928. EXPECT_EQ(regexec(&regex, "a hello friends b", 0, NULL, REG_NOTEOL), REG_NOMATCH);
  929. EXPECT_EQ(regexec(&regex, "hello friends b", 0, NULL, REG_NOTEOL), REG_NOMATCH);
  930. EXPECT_EQ(regexec(&regex, "hello friends b", 0, NULL, REG_NOTEOL | REG_SEARCH), REG_NOERR);
  931. EXPECT_EQ(regexec(&regex, "a hello friends b", 0, NULL, REG_NOTEOL | REG_SEARCH), REG_NOMATCH);
  932. EXPECT_EQ(regexec(&regex, "a hello friends b", 0, NULL, REG_NOTBOL | REG_NOTEOL), REG_NOMATCH);
  933. EXPECT_EQ(regexec(&regex, "a hello friends b", 0, NULL, REG_NOTBOL | REG_NOTEOL | REG_SEARCH), REG_NOMATCH);
  934. EXPECT_EQ(regexec(&regex2, "hello friends", 0, NULL, REG_NOTBOL), REG_NOMATCH);
  935. EXPECT_EQ(regexec(&regex2, "hello friends", 0, NULL, REG_NOTEOL), REG_NOMATCH);
  936. regfree(&regex);
  937. regfree(&regex2);
  938. }
  939. TEST_CASE(bre_basic)
  940. {
  941. regex_t regex;
  942. EXPECT_EQ(regcomp(&regex, "hello friends", REG_NOSUB | REG_ICASE), REG_NOERR);
  943. EXPECT_EQ(regexec(&regex, "hello friends", 0, NULL, 0), REG_NOERR);
  944. regfree(&regex);
  945. EXPECT_EQ(regcomp(&regex, "\\(15\\)\\1", REG_NOSUB | REG_ICASE), REG_NOERR);
  946. EXPECT_EQ(regexec(&regex, "1515", 0, NULL, 0), REG_NOERR);
  947. EXPECT_EQ(regexec(&regex, "55", 0, NULL, 0), REG_NOMATCH);
  948. regfree(&regex);
  949. EXPECT_EQ(regcomp(&regex, "15\\{1,2\\}", REG_NOSUB | REG_ICASE), REG_NOERR);
  950. EXPECT_EQ(regexec(&regex, "15", 0, NULL, 0), REG_NOERR);
  951. EXPECT_EQ(regexec(&regex, "1515", 0, NULL, 0), REG_NOMATCH);
  952. EXPECT_EQ(regexec(&regex, "55", 0, NULL, 0), REG_NOMATCH);
  953. regfree(&regex);
  954. EXPECT_EQ(regcomp(&regex, "15{1,2}", REG_NOSUB | REG_ICASE), REG_NOERR);
  955. EXPECT_EQ(regexec(&regex, "15{1,2}", 0, NULL, 0), REG_NOERR);
  956. regfree(&regex);
  957. EXPECT_EQ(regcomp(&regex, "1[56]", REG_NOSUB | REG_ICASE), REG_NOERR);
  958. EXPECT_EQ(regexec(&regex, "15", 0, NULL, 0), REG_NOERR);
  959. EXPECT_EQ(regexec(&regex, "16", 0, NULL, 0), REG_NOERR);
  960. EXPECT_EQ(regexec(&regex, "17", 0, NULL, 0), REG_NOMATCH);
  961. regfree(&regex);
  962. }