TestString.cpp 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473
  1. /*
  2. * Copyright (c) 2022, Andreas Kling <kling@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. // This is included first on purpose. We specifically do not want LibTest to override VERIFY here so
  7. // that we can actually test that some String factory methods cause a crash with invalid input.
  8. #include <AK/String.h>
  9. #include <LibTest/TestCase.h>
  10. #include <AK/StringBuilder.h>
  11. #include <AK/Try.h>
  12. #include <AK/Utf8View.h>
  13. #include <AK/Vector.h>
  14. TEST_CASE(construct_empty)
  15. {
  16. String empty;
  17. EXPECT(empty.is_empty());
  18. EXPECT_EQ(empty.bytes().size(), 0u);
  19. auto empty2 = MUST(String::from_utf8(""sv));
  20. EXPECT(empty2.is_empty());
  21. EXPECT_EQ(empty, empty2);
  22. EXPECT_EQ(empty, ""sv);
  23. }
  24. TEST_CASE(move_assignment)
  25. {
  26. String string1 = MUST(String::from_utf8("hello"sv));
  27. string1 = MUST(String::from_utf8("friends!"sv));
  28. EXPECT_EQ(string1, "friends!"sv);
  29. }
  30. TEST_CASE(short_strings)
  31. {
  32. #ifdef AK_ARCH_64_BIT
  33. auto string1 = MUST(String::from_utf8("abcdefg"sv));
  34. EXPECT_EQ(string1.is_short_string(), true);
  35. EXPECT_EQ(string1.bytes().size(), 7u);
  36. EXPECT_EQ(string1.bytes_as_string_view(), "abcdefg"sv);
  37. constexpr auto string2 = String::from_utf8_short_string("abcdefg"sv);
  38. EXPECT_EQ(string2.is_short_string(), true);
  39. EXPECT_EQ(string2.bytes().size(), 7u);
  40. EXPECT_EQ(string2, string1);
  41. #else
  42. auto string1 = MUST(String::from_utf8("abc"sv));
  43. EXPECT_EQ(string1.is_short_string(), true);
  44. EXPECT_EQ(string1.bytes().size(), 3u);
  45. EXPECT_EQ(string1.bytes_as_string_view(), "abc"sv);
  46. constexpr auto string2 = String::from_utf8_short_string("abc"sv);
  47. EXPECT_EQ(string2.is_short_string(), true);
  48. EXPECT_EQ(string2.bytes().size(), 3u);
  49. EXPECT_EQ(string2, string1);
  50. #endif
  51. }
  52. TEST_CASE(long_strings)
  53. {
  54. auto string = MUST(String::from_utf8("abcdefgh"sv));
  55. EXPECT_EQ(string.is_short_string(), false);
  56. EXPECT_EQ(string.bytes().size(), 8u);
  57. EXPECT_EQ(string.bytes_as_string_view(), "abcdefgh"sv);
  58. }
  59. TEST_CASE(from_code_points)
  60. {
  61. for (u32 code_point = 0; code_point < 0x80; ++code_point) {
  62. auto string = String::from_code_point(code_point);
  63. auto ch = static_cast<char>(code_point);
  64. StringView view { &ch, 1 };
  65. EXPECT_EQ(string, view);
  66. }
  67. auto string = String::from_code_point(0x10ffff);
  68. EXPECT_EQ(string, "\xF4\x8F\xBF\xBF"sv);
  69. EXPECT_CRASH("Creating a string from an invalid code point", [] {
  70. String::from_code_point(0xffffffff);
  71. return Test::Crash::Failure::DidNotCrash;
  72. });
  73. }
  74. TEST_CASE(substring)
  75. {
  76. auto superstring = MUST(String::from_utf8("Hello I am a long string"sv));
  77. auto short_substring = MUST(superstring.substring_from_byte_offset(0, 5));
  78. EXPECT_EQ(short_substring, "Hello"sv);
  79. auto long_substring = MUST(superstring.substring_from_byte_offset(0, 10));
  80. EXPECT_EQ(long_substring, "Hello I am"sv);
  81. }
  82. TEST_CASE(code_points)
  83. {
  84. auto string = MUST(String::from_utf8("🦬🪒"sv));
  85. Vector<u32> code_points;
  86. for (auto code_point : string.code_points())
  87. code_points.append(code_point);
  88. EXPECT_EQ(code_points[0], 0x1f9acu);
  89. EXPECT_EQ(code_points[1], 0x1fa92u);
  90. }
  91. TEST_CASE(string_builder)
  92. {
  93. StringBuilder builder;
  94. builder.append_code_point(0x1f9acu);
  95. builder.append_code_point(0x1fa92u);
  96. auto string = MUST(builder.to_string());
  97. EXPECT_EQ(string, "🦬🪒"sv);
  98. EXPECT_EQ(string.bytes().size(), 8u);
  99. }
  100. TEST_CASE(ak_format)
  101. {
  102. auto foo = MUST(String::formatted("Hello {}", MUST(String::from_utf8("friends"sv))));
  103. EXPECT_EQ(foo, "Hello friends"sv);
  104. }
  105. TEST_CASE(replace)
  106. {
  107. {
  108. auto haystack = MUST(String::from_utf8("Hello enemies"sv));
  109. auto result = MUST(haystack.replace("enemies"sv, "friends"sv, ReplaceMode::All));
  110. EXPECT_EQ(result, "Hello friends"sv);
  111. }
  112. {
  113. auto base_title = MUST(String::from_utf8("anon@courage:~"sv));
  114. auto result = MUST(base_title.replace("[*]"sv, "(*)"sv, ReplaceMode::FirstOnly));
  115. EXPECT_EQ(result, "anon@courage:~"sv);
  116. }
  117. }
  118. TEST_CASE(reverse)
  119. {
  120. auto test_reverse = [](auto test, auto expected) {
  121. auto string = MUST(String::from_utf8(test));
  122. auto result = MUST(string.reverse());
  123. EXPECT_EQ(result, expected);
  124. };
  125. test_reverse(""sv, ""sv);
  126. test_reverse("a"sv, "a"sv);
  127. test_reverse("ab"sv, "ba"sv);
  128. test_reverse("ab cd ef"sv, "fe dc ba"sv);
  129. test_reverse("😀"sv, "😀"sv);
  130. test_reverse("ab😀cd"sv, "dc😀ba"sv);
  131. }
  132. TEST_CASE(to_lowercase)
  133. {
  134. {
  135. auto string = MUST(String::from_utf8("Aa"sv));
  136. auto result = MUST(string.to_lowercase());
  137. EXPECT_EQ(result, "aa"sv);
  138. }
  139. {
  140. auto string = MUST(String::from_utf8("Ωω"sv));
  141. auto result = MUST(string.to_lowercase());
  142. EXPECT_EQ(result, "ωω"sv);
  143. }
  144. {
  145. auto string = MUST(String::from_utf8("İi̇"sv));
  146. auto result = MUST(string.to_lowercase());
  147. EXPECT_EQ(result, "i̇i̇"sv);
  148. }
  149. }
  150. TEST_CASE(to_uppercase)
  151. {
  152. {
  153. auto string = MUST(String::from_utf8("Aa"sv));
  154. auto result = MUST(string.to_uppercase());
  155. EXPECT_EQ(result, "AA"sv);
  156. }
  157. {
  158. auto string = MUST(String::from_utf8("Ωω"sv));
  159. auto result = MUST(string.to_uppercase());
  160. EXPECT_EQ(result, "ΩΩ"sv);
  161. }
  162. {
  163. auto string = MUST(String::from_utf8("ʼn"sv));
  164. auto result = MUST(string.to_uppercase());
  165. EXPECT_EQ(result, "ʼN"sv);
  166. }
  167. }
  168. TEST_CASE(to_titlecase)
  169. {
  170. {
  171. auto string = MUST(String::from_utf8("foo bar baz"sv));
  172. auto result = MUST(string.to_titlecase());
  173. EXPECT_EQ(result, "Foo Bar Baz"sv);
  174. }
  175. {
  176. auto string = MUST(String::from_utf8("foo \n \r bar \t baz"sv));
  177. auto result = MUST(string.to_titlecase());
  178. EXPECT_EQ(result, "Foo \n \r Bar \t Baz"sv);
  179. }
  180. {
  181. auto string = MUST(String::from_utf8("f\"oo\" b'ar'"sv));
  182. auto result = MUST(string.to_titlecase());
  183. EXPECT_EQ(result, "F\"Oo\" B'Ar'"sv);
  184. }
  185. {
  186. auto string = MUST(String::from_utf8("123dollars"sv));
  187. auto result = MUST(string.to_titlecase());
  188. EXPECT_EQ(result, "123Dollars"sv);
  189. }
  190. }
  191. TEST_CASE(equals_ignoring_case)
  192. {
  193. {
  194. String string1 {};
  195. String string2 {};
  196. EXPECT(MUST(string1.equals_ignoring_case(string2)));
  197. }
  198. {
  199. auto string1 = MUST(String::from_utf8("abcd"sv));
  200. auto string2 = MUST(String::from_utf8("ABCD"sv));
  201. auto string3 = MUST(String::from_utf8("AbCd"sv));
  202. auto string4 = MUST(String::from_utf8("dcba"sv));
  203. EXPECT(MUST(string1.equals_ignoring_case(string2)));
  204. EXPECT(MUST(string1.equals_ignoring_case(string3)));
  205. EXPECT(!MUST(string1.equals_ignoring_case(string4)));
  206. EXPECT(MUST(string2.equals_ignoring_case(string1)));
  207. EXPECT(MUST(string2.equals_ignoring_case(string3)));
  208. EXPECT(!MUST(string2.equals_ignoring_case(string4)));
  209. EXPECT(MUST(string3.equals_ignoring_case(string1)));
  210. EXPECT(MUST(string3.equals_ignoring_case(string2)));
  211. EXPECT(!MUST(string3.equals_ignoring_case(string4)));
  212. }
  213. {
  214. auto string1 = MUST(String::from_utf8("\u00DF"sv)); // LATIN SMALL LETTER SHARP S
  215. auto string2 = MUST(String::from_utf8("SS"sv));
  216. auto string3 = MUST(String::from_utf8("Ss"sv));
  217. auto string4 = MUST(String::from_utf8("ss"sv));
  218. auto string5 = MUST(String::from_utf8("S"sv));
  219. auto string6 = MUST(String::from_utf8("s"sv));
  220. EXPECT(MUST(string1.equals_ignoring_case(string2)));
  221. EXPECT(MUST(string1.equals_ignoring_case(string3)));
  222. EXPECT(MUST(string1.equals_ignoring_case(string4)));
  223. EXPECT(!MUST(string1.equals_ignoring_case(string5)));
  224. EXPECT(!MUST(string1.equals_ignoring_case(string6)));
  225. EXPECT(MUST(string2.equals_ignoring_case(string1)));
  226. EXPECT(MUST(string2.equals_ignoring_case(string3)));
  227. EXPECT(MUST(string2.equals_ignoring_case(string4)));
  228. EXPECT(!MUST(string2.equals_ignoring_case(string5)));
  229. EXPECT(!MUST(string2.equals_ignoring_case(string6)));
  230. EXPECT(MUST(string3.equals_ignoring_case(string1)));
  231. EXPECT(MUST(string3.equals_ignoring_case(string2)));
  232. EXPECT(MUST(string3.equals_ignoring_case(string4)));
  233. EXPECT(!MUST(string3.equals_ignoring_case(string5)));
  234. EXPECT(!MUST(string3.equals_ignoring_case(string6)));
  235. EXPECT(MUST(string4.equals_ignoring_case(string1)));
  236. EXPECT(MUST(string4.equals_ignoring_case(string2)));
  237. EXPECT(MUST(string4.equals_ignoring_case(string3)));
  238. EXPECT(!MUST(string4.equals_ignoring_case(string5)));
  239. EXPECT(!MUST(string4.equals_ignoring_case(string6)));
  240. }
  241. }
  242. TEST_CASE(is_one_of)
  243. {
  244. auto foo = MUST(String::from_utf8("foo"sv));
  245. auto bar = MUST(String::from_utf8("bar"sv));
  246. EXPECT(foo.is_one_of(foo));
  247. EXPECT(foo.is_one_of(foo, bar));
  248. EXPECT(foo.is_one_of(bar, foo));
  249. EXPECT(!foo.is_one_of(bar));
  250. EXPECT(!bar.is_one_of("foo"sv));
  251. EXPECT(bar.is_one_of("foo"sv, "bar"sv));
  252. EXPECT(bar.is_one_of("bar"sv, "foo"sv));
  253. EXPECT(bar.is_one_of("bar"sv));
  254. }
  255. TEST_CASE(split)
  256. {
  257. {
  258. auto test = MUST(String::from_utf8("foo bar baz"sv));
  259. auto parts = MUST(test.split(' '));
  260. EXPECT_EQ(parts.size(), 3u);
  261. EXPECT_EQ(parts[0], "foo");
  262. EXPECT_EQ(parts[1], "bar");
  263. EXPECT_EQ(parts[2], "baz");
  264. }
  265. {
  266. auto test = MUST(String::from_utf8("ωΣ2ωΣω"sv));
  267. auto parts = MUST(test.split(0x03A3u));
  268. EXPECT_EQ(parts.size(), 3u);
  269. EXPECT_EQ(parts[0], "ω"sv);
  270. EXPECT_EQ(parts[1], "2ω"sv);
  271. EXPECT_EQ(parts[2], "ω"sv);
  272. }
  273. }
  274. TEST_CASE(find_byte_offset)
  275. {
  276. {
  277. String string {};
  278. auto index1 = string.find_byte_offset(0);
  279. EXPECT(!index1.has_value());
  280. auto index2 = string.find_byte_offset(""sv);
  281. EXPECT(!index2.has_value());
  282. }
  283. {
  284. auto string = MUST(String::from_utf8("foo"sv));
  285. auto index1 = string.find_byte_offset('f');
  286. EXPECT_EQ(index1, 0u);
  287. auto index2 = string.find_byte_offset('o');
  288. EXPECT_EQ(index2, 1u);
  289. auto index3 = string.find_byte_offset('o', *index2 + 1);
  290. EXPECT_EQ(index3, 2u);
  291. auto index4 = string.find_byte_offset('b');
  292. EXPECT(!index4.has_value());
  293. }
  294. {
  295. auto string = MUST(String::from_utf8("foo"sv));
  296. auto index1 = string.find_byte_offset("fo"sv);
  297. EXPECT_EQ(index1, 0u);
  298. auto index2 = string.find_byte_offset("oo"sv);
  299. EXPECT_EQ(index2, 1u);
  300. auto index3 = string.find_byte_offset("o"sv, *index2 + 1);
  301. EXPECT_EQ(index3, 2u);
  302. auto index4 = string.find_byte_offset("fooo"sv);
  303. EXPECT(!index4.has_value());
  304. }
  305. {
  306. auto string = MUST(String::from_utf8("ωΣωΣω"sv));
  307. auto index1 = string.find_byte_offset(0x03C9U);
  308. EXPECT_EQ(index1, 0u);
  309. auto index2 = string.find_byte_offset(0x03A3u);
  310. EXPECT_EQ(index2, 2u);
  311. auto index3 = string.find_byte_offset(0x03C9U, 2);
  312. EXPECT_EQ(index3, 4u);
  313. auto index4 = string.find_byte_offset(0x03A3u, 4);
  314. EXPECT_EQ(index4, 6u);
  315. auto index5 = string.find_byte_offset(0x03C9U, 6);
  316. EXPECT_EQ(index5, 8u);
  317. }
  318. {
  319. auto string = MUST(String::from_utf8("ωΣωΣω"sv));
  320. auto index1 = string.find_byte_offset("ω"sv);
  321. EXPECT_EQ(index1, 0u);
  322. auto index2 = string.find_byte_offset("Σ"sv);
  323. EXPECT_EQ(index2, 2u);
  324. auto index3 = string.find_byte_offset("ω"sv, 2);
  325. EXPECT_EQ(index3, 4u);
  326. auto index4 = string.find_byte_offset("Σ"sv, 4);
  327. EXPECT_EQ(index4, 6u);
  328. auto index5 = string.find_byte_offset("ω"sv, 6);
  329. EXPECT_EQ(index5, 8u);
  330. }
  331. }
  332. TEST_CASE(repeated)
  333. {
  334. {
  335. auto string1 = MUST(String::repeated('a', 0));
  336. EXPECT(string1.is_short_string());
  337. EXPECT(string1.is_empty());
  338. auto string2 = MUST(String::repeated(0x03C9U, 0));
  339. EXPECT(string2.is_short_string());
  340. EXPECT(string2.is_empty());
  341. auto string3 = MUST(String::repeated(0x10300, 0));
  342. EXPECT(string3.is_short_string());
  343. EXPECT(string3.is_empty());
  344. }
  345. {
  346. auto string1 = MUST(String::repeated('a', 1));
  347. EXPECT(string1.is_short_string());
  348. EXPECT_EQ(string1.bytes_as_string_view().length(), 1u);
  349. EXPECT_EQ(string1, "a"sv);
  350. auto string2 = MUST(String::repeated(0x03C9U, 1));
  351. EXPECT(string2.is_short_string());
  352. EXPECT_EQ(string2.bytes_as_string_view().length(), 2u);
  353. EXPECT_EQ(string2, "ω"sv);
  354. auto string3 = MUST(String::repeated(0x10300, 1));
  355. #ifdef AK_ARCH_64_BIT
  356. EXPECT(string3.is_short_string());
  357. #else
  358. EXPECT(!string3.is_short_string());
  359. #endif
  360. EXPECT_EQ(string3.bytes_as_string_view().length(), 4u);
  361. EXPECT_EQ(string3, "𐌀"sv);
  362. }
  363. {
  364. auto string1 = MUST(String::repeated('a', 3));
  365. EXPECT(string1.is_short_string());
  366. EXPECT_EQ(string1.bytes_as_string_view().length(), 3u);
  367. EXPECT_EQ(string1, "aaa"sv);
  368. auto string2 = MUST(String::repeated(0x03C9U, 3));
  369. #ifdef AK_ARCH_64_BIT
  370. EXPECT(string2.is_short_string());
  371. #else
  372. EXPECT(!string2.is_short_string());
  373. #endif
  374. EXPECT_EQ(string2.bytes_as_string_view().length(), 6u);
  375. EXPECT_EQ(string2, "ωωω"sv);
  376. auto string3 = MUST(String::repeated(0x10300, 3));
  377. EXPECT(!string3.is_short_string());
  378. EXPECT_EQ(string3.bytes_as_string_view().length(), 12u);
  379. EXPECT_EQ(string3, "𐌀𐌀𐌀"sv);
  380. }
  381. {
  382. auto string1 = MUST(String::repeated('a', 10));
  383. EXPECT(!string1.is_short_string());
  384. EXPECT_EQ(string1.bytes_as_string_view().length(), 10u);
  385. EXPECT_EQ(string1, "aaaaaaaaaa"sv);
  386. auto string2 = MUST(String::repeated(0x03C9U, 10));
  387. EXPECT(!string2.is_short_string());
  388. EXPECT_EQ(string2.bytes_as_string_view().length(), 20u);
  389. EXPECT_EQ(string2, "ωωωωωωωωωω"sv);
  390. auto string3 = MUST(String::repeated(0x10300, 10));
  391. EXPECT(!string3.is_short_string());
  392. EXPECT_EQ(string3.bytes_as_string_view().length(), 40u);
  393. EXPECT_EQ(string3, "𐌀𐌀𐌀𐌀𐌀𐌀𐌀𐌀𐌀𐌀"sv);
  394. }
  395. EXPECT_CRASH("Creating a string from an invalid code point", [] {
  396. (void)String::repeated(0xffffffff, 1);
  397. return Test::Crash::Failure::DidNotCrash;
  398. });
  399. }