scanf.cpp 20 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618
  1. /*
  2. * Copyright (c) 2021, the SerenityOS developers.
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include <AK/Assertions.h>
  7. #include <AK/Format.h>
  8. #include <AK/GenericLexer.h>
  9. #include <ctype.h>
  10. #include <stdarg.h>
  11. #include <stdio.h>
  12. #include <stdlib.h>
  13. #include <string.h>
  14. enum class LengthModifier {
  15. None,
  16. Default,
  17. Char,
  18. Short,
  19. Long,
  20. LongLong,
  21. IntMax,
  22. Size,
  23. PtrDiff,
  24. LongDouble,
  25. };
  26. enum class ConversionSpecifier {
  27. Unspecified,
  28. Decimal,
  29. Integer,
  30. Octal,
  31. Unsigned,
  32. Hex,
  33. Floating,
  34. String,
  35. UseScanList,
  36. Character,
  37. Pointer,
  38. OutputNumberOfBytes,
  39. Invalid,
  40. };
  41. enum class ReadKind {
  42. Normal,
  43. Octal,
  44. Hex,
  45. Infer,
  46. };
  47. template<typename T, typename ApT, ReadKind kind = ReadKind::Normal>
  48. struct ReadElementConcrete {
  49. bool operator()(GenericLexer&, va_list)
  50. {
  51. return false;
  52. }
  53. };
  54. template<typename ApT, ReadKind kind>
  55. struct ReadElementConcrete<int, ApT, kind> {
  56. bool operator()(GenericLexer& lexer, va_list* ap, bool suppress_assignment)
  57. {
  58. long value = 0;
  59. char* endptr = nullptr;
  60. auto nptr = lexer.remaining().characters_without_null_termination();
  61. if constexpr (kind == ReadKind::Normal)
  62. value = strtol(nptr, &endptr, 10);
  63. if constexpr (kind == ReadKind::Octal)
  64. value = strtol(nptr, &endptr, 8);
  65. if constexpr (kind == ReadKind::Hex)
  66. value = strtol(nptr, &endptr, 16);
  67. if constexpr (kind == ReadKind::Infer)
  68. value = strtol(nptr, &endptr, 0);
  69. if (!endptr)
  70. return false;
  71. if (endptr == nptr)
  72. return false;
  73. auto diff = endptr - nptr;
  74. VERIFY(diff > 0);
  75. lexer.ignore((size_t)diff);
  76. if (!suppress_assignment) {
  77. auto* ptr = va_arg(*ap, ApT*);
  78. *ptr = value;
  79. }
  80. return true;
  81. }
  82. };
  83. template<typename ApT, ReadKind kind>
  84. struct ReadElementConcrete<char, ApT, kind> {
  85. bool operator()(GenericLexer& lexer, va_list* ap, bool suppress_assignment)
  86. {
  87. static_assert(kind == ReadKind::Normal, "Can't read a non-normal character");
  88. if (lexer.is_eof())
  89. return false;
  90. auto ch = lexer.consume();
  91. if (!suppress_assignment) {
  92. auto* ptr = va_arg(*ap, ApT*);
  93. *ptr = ch;
  94. }
  95. return true;
  96. }
  97. };
  98. template<typename ApT, ReadKind kind>
  99. struct ReadElementConcrete<unsigned, ApT, kind> {
  100. bool operator()(GenericLexer& lexer, va_list* ap, bool suppress_assignment)
  101. {
  102. unsigned long value = 0;
  103. char* endptr = nullptr;
  104. auto nptr = lexer.remaining().characters_without_null_termination();
  105. if constexpr (kind == ReadKind::Normal)
  106. value = strtoul(nptr, &endptr, 10);
  107. if constexpr (kind == ReadKind::Octal)
  108. value = strtoul(nptr, &endptr, 8);
  109. if constexpr (kind == ReadKind::Hex)
  110. value = strtoul(nptr, &endptr, 16);
  111. if constexpr (kind == ReadKind::Infer)
  112. value = strtoul(nptr, &endptr, 0);
  113. if (!endptr)
  114. return false;
  115. if (endptr == nptr)
  116. return false;
  117. auto diff = endptr - nptr;
  118. VERIFY(diff > 0);
  119. lexer.ignore((size_t)diff);
  120. if (!suppress_assignment) {
  121. auto* ptr = va_arg(*ap, ApT*);
  122. *ptr = value;
  123. }
  124. return true;
  125. }
  126. };
  127. template<typename ApT, ReadKind kind>
  128. struct ReadElementConcrete<long long, ApT, kind> {
  129. bool operator()(GenericLexer& lexer, va_list* ap, bool suppress_assignment)
  130. {
  131. long long value = 0;
  132. char* endptr = nullptr;
  133. auto nptr = lexer.remaining().characters_without_null_termination();
  134. if constexpr (kind == ReadKind::Normal)
  135. value = strtoll(nptr, &endptr, 10);
  136. if constexpr (kind == ReadKind::Octal)
  137. value = strtoll(nptr, &endptr, 8);
  138. if constexpr (kind == ReadKind::Hex)
  139. value = strtoll(nptr, &endptr, 16);
  140. if constexpr (kind == ReadKind::Infer)
  141. value = strtoll(nptr, &endptr, 0);
  142. if (!endptr)
  143. return false;
  144. if (endptr == nptr)
  145. return false;
  146. auto diff = endptr - nptr;
  147. VERIFY(diff > 0);
  148. lexer.ignore((size_t)diff);
  149. if (!suppress_assignment) {
  150. auto* ptr = va_arg(*ap, ApT*);
  151. *ptr = value;
  152. }
  153. return true;
  154. }
  155. };
  156. template<typename ApT, ReadKind kind>
  157. struct ReadElementConcrete<unsigned long long, ApT, kind> {
  158. bool operator()(GenericLexer& lexer, va_list* ap, bool suppress_assignment)
  159. {
  160. unsigned long long value = 0;
  161. char* endptr = nullptr;
  162. auto nptr = lexer.remaining().characters_without_null_termination();
  163. if constexpr (kind == ReadKind::Normal)
  164. value = strtoull(nptr, &endptr, 10);
  165. if constexpr (kind == ReadKind::Octal)
  166. value = strtoull(nptr, &endptr, 8);
  167. if constexpr (kind == ReadKind::Hex)
  168. value = strtoull(nptr, &endptr, 16);
  169. if constexpr (kind == ReadKind::Infer)
  170. value = strtoull(nptr, &endptr, 0);
  171. if (!endptr)
  172. return false;
  173. if (endptr == nptr)
  174. return false;
  175. auto diff = endptr - nptr;
  176. VERIFY(diff > 0);
  177. lexer.ignore((size_t)diff);
  178. if (!suppress_assignment) {
  179. auto* ptr = va_arg(*ap, ApT*);
  180. *ptr = value;
  181. }
  182. return true;
  183. }
  184. };
  185. template<typename ApT, ReadKind kind>
  186. struct ReadElementConcrete<float, ApT, kind> {
  187. bool operator()(GenericLexer& lexer, va_list* ap, bool suppress_assignment)
  188. {
  189. double value = 0;
  190. char* endptr = nullptr;
  191. auto nptr = lexer.remaining().characters_without_null_termination();
  192. if constexpr (kind == ReadKind::Normal)
  193. value = strtod(nptr, &endptr);
  194. else
  195. return false;
  196. if (!endptr)
  197. return false;
  198. if (endptr == nptr)
  199. return false;
  200. auto diff = endptr - nptr;
  201. VERIFY(diff > 0);
  202. lexer.ignore((size_t)diff);
  203. if (!suppress_assignment) {
  204. auto* ptr = va_arg(*ap, ApT*);
  205. *ptr = value;
  206. }
  207. return true;
  208. }
  209. };
  210. template<typename T, ReadKind kind>
  211. struct ReadElement {
  212. bool operator()(LengthModifier length_modifier, GenericLexer& input_lexer, va_list* ap, bool suppress_assignment)
  213. {
  214. switch (length_modifier) {
  215. default:
  216. case LengthModifier::None:
  217. VERIFY_NOT_REACHED();
  218. case LengthModifier::Default:
  219. return ReadElementConcrete<T, T, kind> {}(input_lexer, ap, suppress_assignment);
  220. case LengthModifier::Char:
  221. return ReadElementConcrete<T, char, kind> {}(input_lexer, ap, suppress_assignment);
  222. case LengthModifier::Short:
  223. return ReadElementConcrete<T, short, kind> {}(input_lexer, ap, suppress_assignment);
  224. case LengthModifier::Long:
  225. if constexpr (IsSame<T, int>)
  226. return ReadElementConcrete<T, long, kind> {}(input_lexer, ap, suppress_assignment);
  227. if constexpr (IsSame<T, unsigned>)
  228. return ReadElementConcrete<T, unsigned long, kind> {}(input_lexer, ap, suppress_assignment);
  229. if constexpr (IsSame<T, float>)
  230. return ReadElementConcrete<int, double, kind> {}(input_lexer, ap, suppress_assignment);
  231. return false;
  232. case LengthModifier::LongLong:
  233. if constexpr (IsSame<T, int>)
  234. return ReadElementConcrete<long long, long long, kind> {}(input_lexer, ap, suppress_assignment);
  235. if constexpr (IsSame<T, unsigned>)
  236. return ReadElementConcrete<unsigned long long, unsigned long long, kind> {}(input_lexer, ap, suppress_assignment);
  237. if constexpr (IsSame<T, float>)
  238. return ReadElementConcrete<long long, double, kind> {}(input_lexer, ap, suppress_assignment);
  239. return false;
  240. case LengthModifier::IntMax:
  241. return ReadElementConcrete<T, intmax_t, kind> {}(input_lexer, ap, suppress_assignment);
  242. case LengthModifier::Size:
  243. return ReadElementConcrete<T, size_t, kind> {}(input_lexer, ap, suppress_assignment);
  244. case LengthModifier::PtrDiff:
  245. return ReadElementConcrete<T, ptrdiff_t, kind> {}(input_lexer, ap, suppress_assignment);
  246. case LengthModifier::LongDouble:
  247. return ReadElementConcrete<T, long double, kind> {}(input_lexer, ap, suppress_assignment);
  248. }
  249. }
  250. };
  251. template<>
  252. struct ReadElement<char*, ReadKind::Normal> {
  253. ReadElement(StringView scan_set = {}, bool invert = false)
  254. : scan_set(scan_set.is_null() ? " \t\n\f\r"sv : scan_set)
  255. , invert(scan_set.is_null() ? true : invert)
  256. {
  257. }
  258. bool operator()(LengthModifier length_modifier, GenericLexer& input_lexer, va_list* ap, bool suppress_assignment)
  259. {
  260. // FIXME: Implement wide strings and such.
  261. if (length_modifier != LengthModifier::Default)
  262. return false;
  263. auto str = input_lexer.consume_while([this](auto c) { return this->matches(c); });
  264. if (str.is_empty())
  265. return false;
  266. if (!suppress_assignment) {
  267. auto* ptr = va_arg(*ap, char*);
  268. memcpy(ptr, str.characters_without_null_termination(), str.length());
  269. ptr[str.length()] = 0;
  270. }
  271. return true;
  272. }
  273. private:
  274. bool matches(char c) const
  275. {
  276. return invert ^ scan_set.contains(c);
  277. }
  278. StringView const scan_set;
  279. bool invert { false };
  280. };
  281. template<>
  282. struct ReadElement<void*, ReadKind::Normal> {
  283. bool operator()(LengthModifier length_modifier, GenericLexer& input_lexer, va_list* ap, bool suppress_assignment)
  284. {
  285. if (length_modifier != LengthModifier::Default)
  286. return false;
  287. auto str = input_lexer.consume_while([this](auto c) { return this->should_consume(c); });
  288. if (count != 8) {
  289. fail:;
  290. for (size_t i = 0; i < count; ++i)
  291. input_lexer.retreat();
  292. return false;
  293. }
  294. char buf[9] { 0 };
  295. memcpy(buf, str.characters_without_null_termination(), 8);
  296. buf[8] = 0;
  297. char* endptr = nullptr;
  298. auto value = strtoull(buf, &endptr, 16);
  299. if (endptr != &buf[8])
  300. goto fail;
  301. if (!suppress_assignment) {
  302. auto* ptr = va_arg(*ap, void**);
  303. memcpy(ptr, &value, sizeof(value));
  304. }
  305. return true;
  306. }
  307. private:
  308. bool should_consume(char c)
  309. {
  310. if (count == 8)
  311. return false;
  312. if (!isxdigit(c))
  313. return false;
  314. ++count;
  315. return true;
  316. }
  317. size_t count { 0 };
  318. };
  319. extern "C" int vsscanf(char const* input, char const* format, va_list ap)
  320. {
  321. GenericLexer format_lexer { { format, strlen(format) } };
  322. GenericLexer input_lexer { { input, strlen(input) } };
  323. int elements_matched = 0;
  324. va_list copy;
  325. __builtin_va_copy(copy, ap);
  326. while (!format_lexer.is_eof()) {
  327. if (format_lexer.next_is(isspace)) {
  328. format_lexer.ignore_while(isspace);
  329. input_lexer.ignore_while(isspace);
  330. }
  331. if (!format_lexer.next_is('%')) {
  332. read_one_literal:;
  333. if (format_lexer.is_eof())
  334. break;
  335. auto next_char = format_lexer.consume();
  336. if (!input_lexer.consume_specific(next_char))
  337. return elements_matched;
  338. continue;
  339. }
  340. if (format_lexer.next_is("%%")) {
  341. format_lexer.ignore();
  342. goto read_one_literal;
  343. }
  344. format_lexer.ignore(); // '%'
  345. bool suppress_assignment = false;
  346. if (format_lexer.next_is('*')) {
  347. suppress_assignment = true;
  348. format_lexer.ignore();
  349. }
  350. // Parse width specification
  351. [[maybe_unused]] int width_specifier = 0;
  352. if (format_lexer.next_is(isdigit)) {
  353. auto width_digits = format_lexer.consume_while([](char c) { return isdigit(c); });
  354. width_specifier = width_digits.to_number<int>().value();
  355. // FIXME: Actually use width specifier
  356. }
  357. bool invert_scanlist = false;
  358. StringView scanlist;
  359. LengthModifier length_modifier { LengthModifier::None };
  360. ConversionSpecifier conversion_specifier { ConversionSpecifier::Unspecified };
  361. reread_lookahead:;
  362. auto format_lookahead = format_lexer.peek();
  363. if (length_modifier == LengthModifier::None) {
  364. switch (format_lookahead) {
  365. case 'h':
  366. if (format_lexer.peek(1) == 'h') {
  367. format_lexer.consume(2);
  368. length_modifier = LengthModifier::Char;
  369. } else {
  370. format_lexer.consume(1);
  371. length_modifier = LengthModifier::Short;
  372. }
  373. break;
  374. case 'l':
  375. if (format_lexer.peek(1) == 'l') {
  376. format_lexer.consume(2);
  377. length_modifier = LengthModifier::LongLong;
  378. } else {
  379. format_lexer.consume(1);
  380. length_modifier = LengthModifier::Long;
  381. }
  382. break;
  383. case 'j':
  384. format_lexer.consume();
  385. length_modifier = LengthModifier::IntMax;
  386. break;
  387. case 'z':
  388. format_lexer.consume();
  389. length_modifier = LengthModifier::Size;
  390. break;
  391. case 't':
  392. format_lexer.consume();
  393. length_modifier = LengthModifier::PtrDiff;
  394. break;
  395. case 'L':
  396. format_lexer.consume();
  397. length_modifier = LengthModifier::LongDouble;
  398. break;
  399. default:
  400. length_modifier = LengthModifier::Default;
  401. break;
  402. }
  403. goto reread_lookahead;
  404. }
  405. if (conversion_specifier == ConversionSpecifier::Unspecified) {
  406. switch (format_lookahead) {
  407. case 'd':
  408. format_lexer.consume();
  409. conversion_specifier = ConversionSpecifier::Decimal;
  410. break;
  411. case 'i':
  412. format_lexer.consume();
  413. conversion_specifier = ConversionSpecifier::Integer;
  414. break;
  415. case 'o':
  416. format_lexer.consume();
  417. conversion_specifier = ConversionSpecifier::Octal;
  418. break;
  419. case 'u':
  420. format_lexer.consume();
  421. conversion_specifier = ConversionSpecifier::Unsigned;
  422. break;
  423. case 'x':
  424. case 'X':
  425. format_lexer.consume();
  426. conversion_specifier = ConversionSpecifier::Hex;
  427. break;
  428. case 'a':
  429. case 'e':
  430. case 'f':
  431. case 'g':
  432. format_lexer.consume();
  433. conversion_specifier = ConversionSpecifier::Floating;
  434. break;
  435. case 's':
  436. format_lexer.consume();
  437. conversion_specifier = ConversionSpecifier::String;
  438. break;
  439. case '[':
  440. format_lexer.consume();
  441. scanlist = format_lexer.consume_until(']');
  442. format_lexer.ignore();
  443. if (scanlist.starts_with('^')) {
  444. scanlist = scanlist.substring_view(1);
  445. invert_scanlist = true;
  446. }
  447. conversion_specifier = ConversionSpecifier::UseScanList;
  448. break;
  449. case 'c':
  450. format_lexer.consume();
  451. conversion_specifier = ConversionSpecifier::Character;
  452. break;
  453. case 'p':
  454. format_lexer.consume();
  455. conversion_specifier = ConversionSpecifier::Pointer;
  456. break;
  457. case 'n':
  458. format_lexer.consume();
  459. conversion_specifier = ConversionSpecifier::OutputNumberOfBytes;
  460. break;
  461. case 'C':
  462. format_lexer.consume();
  463. length_modifier = LengthModifier::Long;
  464. conversion_specifier = ConversionSpecifier::Character;
  465. break;
  466. case 'S':
  467. format_lexer.consume();
  468. length_modifier = LengthModifier::Long;
  469. conversion_specifier = ConversionSpecifier::String;
  470. break;
  471. default:
  472. format_lexer.consume();
  473. conversion_specifier = ConversionSpecifier::Invalid;
  474. break;
  475. }
  476. }
  477. // Now try to read.
  478. switch (conversion_specifier) {
  479. case ConversionSpecifier::Invalid:
  480. case ConversionSpecifier::Unspecified:
  481. default:
  482. // "undefined behavior", let's be nice and crash.
  483. dbgln("Invalid conversion specifier {} in scanf!", (int)conversion_specifier);
  484. VERIFY_NOT_REACHED();
  485. case ConversionSpecifier::Decimal:
  486. if (!ReadElement<int, ReadKind::Normal> {}(length_modifier, input_lexer, &copy, suppress_assignment))
  487. format_lexer.consume_all();
  488. else if (!suppress_assignment)
  489. ++elements_matched;
  490. break;
  491. case ConversionSpecifier::Integer:
  492. if (!ReadElement<int, ReadKind::Infer> {}(length_modifier, input_lexer, &copy, suppress_assignment))
  493. format_lexer.consume_all();
  494. else if (!suppress_assignment)
  495. ++elements_matched;
  496. break;
  497. case ConversionSpecifier::Octal:
  498. if (!ReadElement<unsigned, ReadKind::Octal> {}(length_modifier, input_lexer, &copy, suppress_assignment))
  499. format_lexer.consume_all();
  500. else if (!suppress_assignment)
  501. ++elements_matched;
  502. break;
  503. case ConversionSpecifier::Unsigned:
  504. if (!ReadElement<unsigned, ReadKind::Normal> {}(length_modifier, input_lexer, &copy, suppress_assignment))
  505. format_lexer.consume_all();
  506. else if (!suppress_assignment)
  507. ++elements_matched;
  508. break;
  509. case ConversionSpecifier::Hex:
  510. if (!ReadElement<unsigned, ReadKind::Hex> {}(length_modifier, input_lexer, &copy, suppress_assignment))
  511. format_lexer.consume_all();
  512. else if (!suppress_assignment)
  513. ++elements_matched;
  514. break;
  515. case ConversionSpecifier::Floating:
  516. if (!ReadElement<float, ReadKind::Normal> {}(length_modifier, input_lexer, &copy, suppress_assignment))
  517. format_lexer.consume_all();
  518. else if (!suppress_assignment)
  519. ++elements_matched;
  520. break;
  521. case ConversionSpecifier::String:
  522. if (!ReadElement<char*, ReadKind::Normal> {}(length_modifier, input_lexer, &copy, suppress_assignment))
  523. format_lexer.consume_all();
  524. else if (!suppress_assignment)
  525. ++elements_matched;
  526. break;
  527. case ConversionSpecifier::UseScanList:
  528. if (!ReadElement<char*, ReadKind::Normal> { scanlist, invert_scanlist }(length_modifier, input_lexer, &copy, suppress_assignment))
  529. format_lexer.consume_all();
  530. else if (!suppress_assignment)
  531. ++elements_matched;
  532. break;
  533. case ConversionSpecifier::Character:
  534. if (!ReadElement<char, ReadKind::Normal> {}(length_modifier, input_lexer, &copy, suppress_assignment))
  535. format_lexer.consume_all();
  536. else if (!suppress_assignment)
  537. ++elements_matched;
  538. break;
  539. case ConversionSpecifier::Pointer:
  540. if (!ReadElement<void*, ReadKind::Normal> {}(length_modifier, input_lexer, &copy, suppress_assignment))
  541. format_lexer.consume_all();
  542. else if (!suppress_assignment)
  543. ++elements_matched;
  544. break;
  545. case ConversionSpecifier::OutputNumberOfBytes: {
  546. if (!suppress_assignment) {
  547. auto* ptr = va_arg(copy, int*);
  548. *ptr = input_lexer.tell();
  549. }
  550. break;
  551. }
  552. }
  553. }
  554. va_end(copy);
  555. return elements_matched;
  556. }