lstrlib.c 46 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579
  1. /*
  2. ** $Id: lstrlib.c,v 1.247 2016/04/22 16:36:30 roberto Exp roberto $
  3. ** Standard library for string operations and pattern-matching
  4. ** See Copyright Notice in lua.h
  5. */
  6. #define lstrlib_c
  7. #define LUA_LIB
  8. #include "lprefix.h"
  9. #include <ctype.h>
  10. #include <float.h>
  11. #include <limits.h>
  12. #include <locale.h>
  13. #include <stddef.h>
  14. #include <stdio.h>
  15. #include <stdlib.h>
  16. #include <string.h>
  17. #include "lua.h"
  18. #include "lauxlib.h"
  19. #include "lualib.h"
  20. /*
  21. ** maximum number of captures that a pattern can do during
  22. ** pattern-matching. This limit is arbitrary, but must fit in
  23. ** an unsigned char.
  24. */
  25. #if !defined(LUA_MAXCAPTURES)
  26. #define LUA_MAXCAPTURES 32
  27. #endif
  28. /* macro to 'unsign' a character */
  29. #define uchar(c) ((unsigned char)(c))
  30. /*
  31. ** Some sizes are better limited to fit in 'int', but must also fit in
  32. ** 'size_t'. (We assume that 'lua_Integer' cannot be smaller than 'int'.)
  33. */
  34. #define MAX_SIZET ((size_t)(~(size_t)0))
  35. #define MAXSIZE \
  36. (sizeof(size_t) < sizeof(int) ? MAX_SIZET : (size_t)(INT_MAX))
  37. static int str_len (lua_State *L) {
  38. size_t l;
  39. luaL_checklstring(L, 1, &l);
  40. lua_pushinteger(L, (lua_Integer)l);
  41. return 1;
  42. }
  43. /* translate a relative string position: negative means back from end */
  44. static lua_Integer posrelat (lua_Integer pos, size_t len) {
  45. if (pos >= 0) return pos;
  46. else if (0u - (size_t)pos > len) return 0;
  47. else return (lua_Integer)len + pos + 1;
  48. }
  49. static int str_sub (lua_State *L) {
  50. size_t l;
  51. const char *s = luaL_checklstring(L, 1, &l);
  52. lua_Integer start = posrelat(luaL_checkinteger(L, 2), l);
  53. lua_Integer end = posrelat(luaL_optinteger(L, 3, -1), l);
  54. if (start < 1) start = 1;
  55. if (end > (lua_Integer)l) end = l;
  56. if (start <= end)
  57. lua_pushlstring(L, s + start - 1, (size_t)(end - start) + 1);
  58. else lua_pushliteral(L, "");
  59. return 1;
  60. }
  61. static int str_reverse (lua_State *L) {
  62. size_t l, i;
  63. luaL_Buffer b;
  64. const char *s = luaL_checklstring(L, 1, &l);
  65. char *p = luaL_buffinitsize(L, &b, l);
  66. for (i = 0; i < l; i++)
  67. p[i] = s[l - i - 1];
  68. luaL_pushresultsize(&b, l);
  69. return 1;
  70. }
  71. static int str_lower (lua_State *L) {
  72. size_t l;
  73. size_t i;
  74. luaL_Buffer b;
  75. const char *s = luaL_checklstring(L, 1, &l);
  76. char *p = luaL_buffinitsize(L, &b, l);
  77. for (i=0; i<l; i++)
  78. p[i] = tolower(uchar(s[i]));
  79. luaL_pushresultsize(&b, l);
  80. return 1;
  81. }
  82. static int str_upper (lua_State *L) {
  83. size_t l;
  84. size_t i;
  85. luaL_Buffer b;
  86. const char *s = luaL_checklstring(L, 1, &l);
  87. char *p = luaL_buffinitsize(L, &b, l);
  88. for (i=0; i<l; i++)
  89. p[i] = toupper(uchar(s[i]));
  90. luaL_pushresultsize(&b, l);
  91. return 1;
  92. }
  93. static int str_rep (lua_State *L) {
  94. size_t l, lsep;
  95. const char *s = luaL_checklstring(L, 1, &l);
  96. lua_Integer n = luaL_checkinteger(L, 2);
  97. const char *sep = luaL_optlstring(L, 3, "", &lsep);
  98. if (n <= 0) lua_pushliteral(L, "");
  99. else if (l + lsep < l || l + lsep > MAXSIZE / n) /* may overflow? */
  100. return luaL_error(L, "resulting string too large");
  101. else {
  102. size_t totallen = (size_t)n * l + (size_t)(n - 1) * lsep;
  103. luaL_Buffer b;
  104. char *p = luaL_buffinitsize(L, &b, totallen);
  105. while (n-- > 1) { /* first n-1 copies (followed by separator) */
  106. memcpy(p, s, l * sizeof(char)); p += l;
  107. if (lsep > 0) { /* empty 'memcpy' is not that cheap */
  108. memcpy(p, sep, lsep * sizeof(char));
  109. p += lsep;
  110. }
  111. }
  112. memcpy(p, s, l * sizeof(char)); /* last copy (not followed by separator) */
  113. luaL_pushresultsize(&b, totallen);
  114. }
  115. return 1;
  116. }
  117. static int str_byte (lua_State *L) {
  118. size_t l;
  119. const char *s = luaL_checklstring(L, 1, &l);
  120. lua_Integer posi = posrelat(luaL_optinteger(L, 2, 1), l);
  121. lua_Integer pose = posrelat(luaL_optinteger(L, 3, posi), l);
  122. int n, i;
  123. if (posi < 1) posi = 1;
  124. if (pose > (lua_Integer)l) pose = l;
  125. if (posi > pose) return 0; /* empty interval; return no values */
  126. if (pose - posi >= INT_MAX) /* arithmetic overflow? */
  127. return luaL_error(L, "string slice too long");
  128. n = (int)(pose - posi) + 1;
  129. luaL_checkstack(L, n, "string slice too long");
  130. for (i=0; i<n; i++)
  131. lua_pushinteger(L, uchar(s[posi+i-1]));
  132. return n;
  133. }
  134. static int str_char (lua_State *L) {
  135. int n = lua_gettop(L); /* number of arguments */
  136. int i;
  137. luaL_Buffer b;
  138. char *p = luaL_buffinitsize(L, &b, n);
  139. for (i=1; i<=n; i++) {
  140. lua_Integer c = luaL_checkinteger(L, i);
  141. luaL_argcheck(L, uchar(c) == c, i, "value out of range");
  142. p[i - 1] = uchar(c);
  143. }
  144. luaL_pushresultsize(&b, n);
  145. return 1;
  146. }
  147. static int writer (lua_State *L, const void *b, size_t size, void *B) {
  148. (void)L;
  149. luaL_addlstring((luaL_Buffer *) B, (const char *)b, size);
  150. return 0;
  151. }
  152. static int str_dump (lua_State *L) {
  153. luaL_Buffer b;
  154. int strip = lua_toboolean(L, 2);
  155. luaL_checktype(L, 1, LUA_TFUNCTION);
  156. lua_settop(L, 1);
  157. luaL_buffinit(L,&b);
  158. if (lua_dump(L, writer, &b, strip) != 0)
  159. return luaL_error(L, "unable to dump given function");
  160. luaL_pushresult(&b);
  161. return 1;
  162. }
  163. /*
  164. ** {======================================================
  165. ** PATTERN MATCHING
  166. ** =======================================================
  167. */
  168. #define CAP_UNFINISHED (-1)
  169. #define CAP_POSITION (-2)
  170. typedef struct MatchState {
  171. const char *src_init; /* init of source string */
  172. const char *src_end; /* end ('\0') of source string */
  173. const char *p_end; /* end ('\0') of pattern */
  174. lua_State *L;
  175. int matchdepth; /* control for recursive depth (to avoid C stack overflow) */
  176. unsigned char level; /* total number of captures (finished or unfinished) */
  177. struct {
  178. const char *init;
  179. ptrdiff_t len;
  180. } capture[LUA_MAXCAPTURES];
  181. } MatchState;
  182. /* recursive function */
  183. static const char *match (MatchState *ms, const char *s, const char *p);
  184. /* maximum recursion depth for 'match' */
  185. #if !defined(MAXCCALLS)
  186. #define MAXCCALLS 200
  187. #endif
  188. #define L_ESC '%'
  189. #define SPECIALS "^$*+?.([%-"
  190. static int check_capture (MatchState *ms, int l) {
  191. l -= '1';
  192. if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED)
  193. return luaL_error(ms->L, "invalid capture index %%%d", l + 1);
  194. return l;
  195. }
  196. static int capture_to_close (MatchState *ms) {
  197. int level = ms->level;
  198. for (level--; level>=0; level--)
  199. if (ms->capture[level].len == CAP_UNFINISHED) return level;
  200. return luaL_error(ms->L, "invalid pattern capture");
  201. }
  202. static const char *classend (MatchState *ms, const char *p) {
  203. switch (*p++) {
  204. case L_ESC: {
  205. if (p == ms->p_end)
  206. luaL_error(ms->L, "malformed pattern (ends with '%%')");
  207. return p+1;
  208. }
  209. case '[': {
  210. if (*p == '^') p++;
  211. do { /* look for a ']' */
  212. if (p == ms->p_end)
  213. luaL_error(ms->L, "malformed pattern (missing ']')");
  214. if (*(p++) == L_ESC && p < ms->p_end)
  215. p++; /* skip escapes (e.g. '%]') */
  216. } while (*p != ']');
  217. return p+1;
  218. }
  219. default: {
  220. return p;
  221. }
  222. }
  223. }
  224. static int match_class (int c, int cl) {
  225. int res;
  226. switch (tolower(cl)) {
  227. case 'a' : res = isalpha(c); break;
  228. case 'c' : res = iscntrl(c); break;
  229. case 'd' : res = isdigit(c); break;
  230. case 'g' : res = isgraph(c); break;
  231. case 'l' : res = islower(c); break;
  232. case 'p' : res = ispunct(c); break;
  233. case 's' : res = isspace(c); break;
  234. case 'u' : res = isupper(c); break;
  235. case 'w' : res = isalnum(c); break;
  236. case 'x' : res = isxdigit(c); break;
  237. case 'z' : res = (c == 0); break; /* deprecated option */
  238. default: return (cl == c);
  239. }
  240. return (islower(cl) ? res : !res);
  241. }
  242. static int matchbracketclass (int c, const char *p, const char *ec) {
  243. int sig = 1;
  244. if (*(p+1) == '^') {
  245. sig = 0;
  246. p++; /* skip the '^' */
  247. }
  248. while (++p < ec) {
  249. if (*p == L_ESC) {
  250. p++;
  251. if (match_class(c, uchar(*p)))
  252. return sig;
  253. }
  254. else if ((*(p+1) == '-') && (p+2 < ec)) {
  255. p+=2;
  256. if (uchar(*(p-2)) <= c && c <= uchar(*p))
  257. return sig;
  258. }
  259. else if (uchar(*p) == c) return sig;
  260. }
  261. return !sig;
  262. }
  263. static int singlematch (MatchState *ms, const char *s, const char *p,
  264. const char *ep) {
  265. if (s >= ms->src_end)
  266. return 0;
  267. else {
  268. int c = uchar(*s);
  269. switch (*p) {
  270. case '.': return 1; /* matches any char */
  271. case L_ESC: return match_class(c, uchar(*(p+1)));
  272. case '[': return matchbracketclass(c, p, ep-1);
  273. default: return (uchar(*p) == c);
  274. }
  275. }
  276. }
  277. static const char *matchbalance (MatchState *ms, const char *s,
  278. const char *p) {
  279. if (p >= ms->p_end - 1)
  280. luaL_error(ms->L, "malformed pattern (missing arguments to '%%b')");
  281. if (*s != *p) return NULL;
  282. else {
  283. int b = *p;
  284. int e = *(p+1);
  285. int cont = 1;
  286. while (++s < ms->src_end) {
  287. if (*s == e) {
  288. if (--cont == 0) return s+1;
  289. }
  290. else if (*s == b) cont++;
  291. }
  292. }
  293. return NULL; /* string ends out of balance */
  294. }
  295. static const char *max_expand (MatchState *ms, const char *s,
  296. const char *p, const char *ep) {
  297. ptrdiff_t i = 0; /* counts maximum expand for item */
  298. while (singlematch(ms, s + i, p, ep))
  299. i++;
  300. /* keeps trying to match with the maximum repetitions */
  301. while (i>=0) {
  302. const char *res = match(ms, (s+i), ep+1);
  303. if (res) return res;
  304. i--; /* else didn't match; reduce 1 repetition to try again */
  305. }
  306. return NULL;
  307. }
  308. static const char *min_expand (MatchState *ms, const char *s,
  309. const char *p, const char *ep) {
  310. for (;;) {
  311. const char *res = match(ms, s, ep+1);
  312. if (res != NULL)
  313. return res;
  314. else if (singlematch(ms, s, p, ep))
  315. s++; /* try with one more repetition */
  316. else return NULL;
  317. }
  318. }
  319. static const char *start_capture (MatchState *ms, const char *s,
  320. const char *p, int what) {
  321. const char *res;
  322. int level = ms->level;
  323. if (level >= LUA_MAXCAPTURES) luaL_error(ms->L, "too many captures");
  324. ms->capture[level].init = s;
  325. ms->capture[level].len = what;
  326. ms->level = level+1;
  327. if ((res=match(ms, s, p)) == NULL) /* match failed? */
  328. ms->level--; /* undo capture */
  329. return res;
  330. }
  331. static const char *end_capture (MatchState *ms, const char *s,
  332. const char *p) {
  333. int l = capture_to_close(ms);
  334. const char *res;
  335. ms->capture[l].len = s - ms->capture[l].init; /* close capture */
  336. if ((res = match(ms, s, p)) == NULL) /* match failed? */
  337. ms->capture[l].len = CAP_UNFINISHED; /* undo capture */
  338. return res;
  339. }
  340. static const char *match_capture (MatchState *ms, const char *s, int l) {
  341. size_t len;
  342. l = check_capture(ms, l);
  343. len = ms->capture[l].len;
  344. if ((size_t)(ms->src_end-s) >= len &&
  345. memcmp(ms->capture[l].init, s, len) == 0)
  346. return s+len;
  347. else return NULL;
  348. }
  349. static const char *match (MatchState *ms, const char *s, const char *p) {
  350. if (ms->matchdepth-- == 0)
  351. luaL_error(ms->L, "pattern too complex");
  352. init: /* using goto's to optimize tail recursion */
  353. if (p != ms->p_end) { /* end of pattern? */
  354. switch (*p) {
  355. case '(': { /* start capture */
  356. if (*(p + 1) == ')') /* position capture? */
  357. s = start_capture(ms, s, p + 2, CAP_POSITION);
  358. else
  359. s = start_capture(ms, s, p + 1, CAP_UNFINISHED);
  360. break;
  361. }
  362. case ')': { /* end capture */
  363. s = end_capture(ms, s, p + 1);
  364. break;
  365. }
  366. case '$': {
  367. if ((p + 1) != ms->p_end) /* is the '$' the last char in pattern? */
  368. goto dflt; /* no; go to default */
  369. s = (s == ms->src_end) ? s : NULL; /* check end of string */
  370. break;
  371. }
  372. case L_ESC: { /* escaped sequences not in the format class[*+?-]? */
  373. switch (*(p + 1)) {
  374. case 'b': { /* balanced string? */
  375. s = matchbalance(ms, s, p + 2);
  376. if (s != NULL) {
  377. p += 4; goto init; /* return match(ms, s, p + 4); */
  378. } /* else fail (s == NULL) */
  379. break;
  380. }
  381. case 'f': { /* frontier? */
  382. const char *ep; char previous;
  383. p += 2;
  384. if (*p != '[')
  385. luaL_error(ms->L, "missing '[' after '%%f' in pattern");
  386. ep = classend(ms, p); /* points to what is next */
  387. previous = (s == ms->src_init) ? '\0' : *(s - 1);
  388. if (!matchbracketclass(uchar(previous), p, ep - 1) &&
  389. matchbracketclass(uchar(*s), p, ep - 1)) {
  390. p = ep; goto init; /* return match(ms, s, ep); */
  391. }
  392. s = NULL; /* match failed */
  393. break;
  394. }
  395. case '0': case '1': case '2': case '3':
  396. case '4': case '5': case '6': case '7':
  397. case '8': case '9': { /* capture results (%0-%9)? */
  398. s = match_capture(ms, s, uchar(*(p + 1)));
  399. if (s != NULL) {
  400. p += 2; goto init; /* return match(ms, s, p + 2) */
  401. }
  402. break;
  403. }
  404. default: goto dflt;
  405. }
  406. break;
  407. }
  408. default: dflt: { /* pattern class plus optional suffix */
  409. const char *ep = classend(ms, p); /* points to optional suffix */
  410. /* does not match at least once? */
  411. if (!singlematch(ms, s, p, ep)) {
  412. if (*ep == '*' || *ep == '?' || *ep == '-') { /* accept empty? */
  413. p = ep + 1; goto init; /* return match(ms, s, ep + 1); */
  414. }
  415. else /* '+' or no suffix */
  416. s = NULL; /* fail */
  417. }
  418. else { /* matched once */
  419. switch (*ep) { /* handle optional suffix */
  420. case '?': { /* optional */
  421. const char *res;
  422. if ((res = match(ms, s + 1, ep + 1)) != NULL)
  423. s = res;
  424. else {
  425. p = ep + 1; goto init; /* else return match(ms, s, ep + 1); */
  426. }
  427. break;
  428. }
  429. case '+': /* 1 or more repetitions */
  430. s++; /* 1 match already done */
  431. /* FALLTHROUGH */
  432. case '*': /* 0 or more repetitions */
  433. s = max_expand(ms, s, p, ep);
  434. break;
  435. case '-': /* 0 or more repetitions (minimum) */
  436. s = min_expand(ms, s, p, ep);
  437. break;
  438. default: /* no suffix */
  439. s++; p = ep; goto init; /* return match(ms, s + 1, ep); */
  440. }
  441. }
  442. break;
  443. }
  444. }
  445. }
  446. ms->matchdepth++;
  447. return s;
  448. }
  449. static const char *lmemfind (const char *s1, size_t l1,
  450. const char *s2, size_t l2) {
  451. if (l2 == 0) return s1; /* empty strings are everywhere */
  452. else if (l2 > l1) return NULL; /* avoids a negative 'l1' */
  453. else {
  454. const char *init; /* to search for a '*s2' inside 's1' */
  455. l2--; /* 1st char will be checked by 'memchr' */
  456. l1 = l1-l2; /* 's2' cannot be found after that */
  457. while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) {
  458. init++; /* 1st char is already checked */
  459. if (memcmp(init, s2+1, l2) == 0)
  460. return init-1;
  461. else { /* correct 'l1' and 's1' to try again */
  462. l1 -= init-s1;
  463. s1 = init;
  464. }
  465. }
  466. return NULL; /* not found */
  467. }
  468. }
  469. static void push_onecapture (MatchState *ms, int i, const char *s,
  470. const char *e) {
  471. if (i >= ms->level) {
  472. if (i == 0) /* ms->level == 0, too */
  473. lua_pushlstring(ms->L, s, e - s); /* add whole match */
  474. else
  475. luaL_error(ms->L, "invalid capture index %%%d", i + 1);
  476. }
  477. else {
  478. ptrdiff_t l = ms->capture[i].len;
  479. if (l == CAP_UNFINISHED) luaL_error(ms->L, "unfinished capture");
  480. if (l == CAP_POSITION)
  481. lua_pushinteger(ms->L, (ms->capture[i].init - ms->src_init) + 1);
  482. else
  483. lua_pushlstring(ms->L, ms->capture[i].init, l);
  484. }
  485. }
  486. static int push_captures (MatchState *ms, const char *s, const char *e) {
  487. int i;
  488. int nlevels = (ms->level == 0 && s) ? 1 : ms->level;
  489. luaL_checkstack(ms->L, nlevels, "too many captures");
  490. for (i = 0; i < nlevels; i++)
  491. push_onecapture(ms, i, s, e);
  492. return nlevels; /* number of strings pushed */
  493. }
  494. /* check whether pattern has no special characters */
  495. static int nospecials (const char *p, size_t l) {
  496. size_t upto = 0;
  497. do {
  498. if (strpbrk(p + upto, SPECIALS))
  499. return 0; /* pattern has a special character */
  500. upto += strlen(p + upto) + 1; /* may have more after \0 */
  501. } while (upto <= l);
  502. return 1; /* no special chars found */
  503. }
  504. static void prepstate (MatchState *ms, lua_State *L,
  505. const char *s, size_t ls, const char *p, size_t lp) {
  506. ms->L = L;
  507. ms->matchdepth = MAXCCALLS;
  508. ms->src_init = s;
  509. ms->src_end = s + ls;
  510. ms->p_end = p + lp;
  511. }
  512. static void reprepstate (MatchState *ms) {
  513. ms->level = 0;
  514. lua_assert(ms->matchdepth == MAXCCALLS);
  515. }
  516. static int str_find_aux (lua_State *L, int find) {
  517. size_t ls, lp;
  518. const char *s = luaL_checklstring(L, 1, &ls);
  519. const char *p = luaL_checklstring(L, 2, &lp);
  520. lua_Integer init = posrelat(luaL_optinteger(L, 3, 1), ls);
  521. if (init < 1) init = 1;
  522. else if (init > (lua_Integer)ls + 1) { /* start after string's end? */
  523. lua_pushnil(L); /* cannot find anything */
  524. return 1;
  525. }
  526. /* explicit request or no special characters? */
  527. if (find && (lua_toboolean(L, 4) || nospecials(p, lp))) {
  528. /* do a plain search */
  529. const char *s2 = lmemfind(s + init - 1, ls - (size_t)init + 1, p, lp);
  530. if (s2) {
  531. lua_pushinteger(L, (s2 - s) + 1);
  532. lua_pushinteger(L, (s2 - s) + lp);
  533. return 2;
  534. }
  535. }
  536. else {
  537. MatchState ms;
  538. const char *s1 = s + init - 1;
  539. int anchor = (*p == '^');
  540. if (anchor) {
  541. p++; lp--; /* skip anchor character */
  542. }
  543. prepstate(&ms, L, s, ls, p, lp);
  544. do {
  545. const char *res;
  546. reprepstate(&ms);
  547. if ((res=match(&ms, s1, p)) != NULL) {
  548. if (find) {
  549. lua_pushinteger(L, (s1 - s) + 1); /* start */
  550. lua_pushinteger(L, res - s); /* end */
  551. return push_captures(&ms, NULL, 0) + 2;
  552. }
  553. else
  554. return push_captures(&ms, s1, res);
  555. }
  556. } while (s1++ < ms.src_end && !anchor);
  557. }
  558. lua_pushnil(L); /* not found */
  559. return 1;
  560. }
  561. static int str_find (lua_State *L) {
  562. return str_find_aux(L, 1);
  563. }
  564. static int str_match (lua_State *L) {
  565. return str_find_aux(L, 0);
  566. }
  567. /* state for 'gmatch' */
  568. typedef struct GMatchState {
  569. const char *src; /* current position */
  570. const char *p; /* pattern */
  571. const char *lastmatch; /* end of last match */
  572. MatchState ms; /* match state */
  573. } GMatchState;
  574. static int gmatch_aux (lua_State *L) {
  575. GMatchState *gm = (GMatchState *)lua_touserdata(L, lua_upvalueindex(3));
  576. const char *src;
  577. gm->ms.L = L;
  578. for (src = gm->src; src <= gm->ms.src_end; src++) {
  579. const char *e;
  580. reprepstate(&gm->ms);
  581. if ((e = match(&gm->ms, src, gm->p)) != NULL && e != gm->lastmatch) {
  582. gm->src = gm->lastmatch = e;
  583. return push_captures(&gm->ms, src, e);
  584. }
  585. }
  586. return 0; /* not found */
  587. }
  588. static int gmatch (lua_State *L) {
  589. size_t ls, lp;
  590. const char *s = luaL_checklstring(L, 1, &ls);
  591. const char *p = luaL_checklstring(L, 2, &lp);
  592. GMatchState *gm;
  593. lua_settop(L, 2); /* keep them on closure to avoid being collected */
  594. gm = (GMatchState *)lua_newuserdata(L, sizeof(GMatchState));
  595. prepstate(&gm->ms, L, s, ls, p, lp);
  596. gm->src = s; gm->p = p; gm->lastmatch = NULL;
  597. lua_pushcclosure(L, gmatch_aux, 3);
  598. return 1;
  599. }
  600. static void add_s (MatchState *ms, luaL_Buffer *b, const char *s,
  601. const char *e) {
  602. size_t l, i;
  603. lua_State *L = ms->L;
  604. const char *news = lua_tolstring(L, 3, &l);
  605. for (i = 0; i < l; i++) {
  606. if (news[i] != L_ESC)
  607. luaL_addchar(b, news[i]);
  608. else {
  609. i++; /* skip ESC */
  610. if (!isdigit(uchar(news[i]))) {
  611. if (news[i] != L_ESC)
  612. luaL_error(L, "invalid use of '%c' in replacement string", L_ESC);
  613. luaL_addchar(b, news[i]);
  614. }
  615. else if (news[i] == '0')
  616. luaL_addlstring(b, s, e - s);
  617. else {
  618. push_onecapture(ms, news[i] - '1', s, e);
  619. luaL_tolstring(L, -1, NULL); /* if number, convert it to string */
  620. lua_remove(L, -2); /* remove original value */
  621. luaL_addvalue(b); /* add capture to accumulated result */
  622. }
  623. }
  624. }
  625. }
  626. static void add_value (MatchState *ms, luaL_Buffer *b, const char *s,
  627. const char *e, int tr) {
  628. lua_State *L = ms->L;
  629. switch (tr) {
  630. case LUA_TFUNCTION: {
  631. int n;
  632. lua_pushvalue(L, 3);
  633. n = push_captures(ms, s, e);
  634. lua_call(L, n, 1);
  635. break;
  636. }
  637. case LUA_TTABLE: {
  638. push_onecapture(ms, 0, s, e);
  639. lua_gettable(L, 3);
  640. break;
  641. }
  642. default: { /* LUA_TNUMBER or LUA_TSTRING */
  643. add_s(ms, b, s, e);
  644. return;
  645. }
  646. }
  647. if (!lua_toboolean(L, -1)) { /* nil or false? */
  648. lua_pop(L, 1);
  649. lua_pushlstring(L, s, e - s); /* keep original text */
  650. }
  651. else if (!lua_isstring(L, -1))
  652. luaL_error(L, "invalid replacement value (a %s)", luaL_typename(L, -1));
  653. luaL_addvalue(b); /* add result to accumulator */
  654. }
  655. static int str_gsub (lua_State *L) {
  656. size_t srcl, lp;
  657. const char *src = luaL_checklstring(L, 1, &srcl); /* subject */
  658. const char *p = luaL_checklstring(L, 2, &lp); /* pattern */
  659. const char *lastmatch = NULL; /* end of last match */
  660. int tr = lua_type(L, 3); /* replacement type */
  661. lua_Integer max_s = luaL_optinteger(L, 4, srcl + 1); /* max replacements */
  662. int anchor = (*p == '^');
  663. lua_Integer n = 0; /* replacement count */
  664. MatchState ms;
  665. luaL_Buffer b;
  666. luaL_argcheck(L, tr == LUA_TNUMBER || tr == LUA_TSTRING ||
  667. tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3,
  668. "string/function/table expected");
  669. luaL_buffinit(L, &b);
  670. if (anchor) {
  671. p++; lp--; /* skip anchor character */
  672. }
  673. prepstate(&ms, L, src, srcl, p, lp);
  674. while (n < max_s) {
  675. const char *e;
  676. reprepstate(&ms); /* (re)prepare state for new match */
  677. if ((e = match(&ms, src, p)) != NULL && e != lastmatch) { /* match? */
  678. n++;
  679. add_value(&ms, &b, src, e, tr); /* add replacement to buffer */
  680. src = lastmatch = e;
  681. }
  682. else if (src < ms.src_end) /* otherwise, skip one character */
  683. luaL_addchar(&b, *src++);
  684. else break; /* end of subject */
  685. if (anchor) break;
  686. }
  687. luaL_addlstring(&b, src, ms.src_end-src);
  688. luaL_pushresult(&b);
  689. lua_pushinteger(L, n); /* number of substitutions */
  690. return 2;
  691. }
  692. /* }====================================================== */
  693. /*
  694. ** {======================================================
  695. ** STRING FORMAT
  696. ** =======================================================
  697. */
  698. #if !defined(lua_number2strx) /* { */
  699. /*
  700. ** Hexadecimal floating-point formatter
  701. */
  702. #include <math.h>
  703. #define SIZELENMOD (sizeof(LUA_NUMBER_FRMLEN)/sizeof(char))
  704. /*
  705. ** Number of bits that goes into the first digit. It can be any value
  706. ** between 1 and 4; the following definition tries to align the number
  707. ** to nibble boundaries by making what is left after that first digit a
  708. ** multiple of 4.
  709. */
  710. #define L_NBFD ((l_mathlim(MANT_DIG) - 1)%4 + 1)
  711. /*
  712. ** Add integer part of 'x' to buffer and return new 'x'
  713. */
  714. static lua_Number adddigit (char *buff, int n, lua_Number x) {
  715. lua_Number dd = l_mathop(floor)(x); /* get integer part from 'x' */
  716. int d = (int)dd;
  717. buff[n] = (d < 10 ? d + '0' : d - 10 + 'a'); /* add to buffer */
  718. return x - dd; /* return what is left */
  719. }
  720. static int num2straux (char *buff, int sz, lua_Number x) {
  721. if (x != x || x == HUGE_VAL || x == -HUGE_VAL) /* inf or NaN? */
  722. return l_sprintf(buff, sz, LUA_NUMBER_FMT, x); /* equal to '%g' */
  723. else if (x == 0) { /* can be -0... */
  724. /* create "0" or "-0" followed by exponent */
  725. return l_sprintf(buff, sz, LUA_NUMBER_FMT "x0p+0", x);
  726. }
  727. else {
  728. int e;
  729. lua_Number m = l_mathop(frexp)(x, &e); /* 'x' fraction and exponent */
  730. int n = 0; /* character count */
  731. if (m < 0) { /* is number negative? */
  732. buff[n++] = '-'; /* add signal */
  733. m = -m; /* make it positive */
  734. }
  735. buff[n++] = '0'; buff[n++] = 'x'; /* add "0x" */
  736. m = adddigit(buff, n++, m * (1 << L_NBFD)); /* add first digit */
  737. e -= L_NBFD; /* this digit goes before the radix point */
  738. if (m > 0) { /* more digits? */
  739. buff[n++] = lua_getlocaledecpoint(); /* add radix point */
  740. do { /* add as many digits as needed */
  741. m = adddigit(buff, n++, m * 16);
  742. } while (m > 0);
  743. }
  744. n += l_sprintf(buff + n, sz - n, "p%+d", e); /* add exponent */
  745. lua_assert(n < sz);
  746. return n;
  747. }
  748. }
  749. static int lua_number2strx (lua_State *L, char *buff, int sz,
  750. const char *fmt, lua_Number x) {
  751. int n = num2straux(buff, sz, x);
  752. if (fmt[SIZELENMOD] == 'A') {
  753. int i;
  754. for (i = 0; i < n; i++)
  755. buff[i] = toupper(uchar(buff[i]));
  756. }
  757. else if (fmt[SIZELENMOD] != 'a')
  758. luaL_error(L, "modifiers for format '%%a'/'%%A' not implemented");
  759. return n;
  760. }
  761. #endif /* } */
  762. /*
  763. ** Maximum size of each formatted item. This maximum size is produced
  764. ** by format('%.99f', -maxfloat), and is equal to 99 + 3 ('-', '.',
  765. ** and '\0') + number of decimal digits to represent maxfloat (which
  766. ** is maximum exponent + 1). (99+3+1 then rounded to 120 for "extra
  767. ** expenses", such as locale-dependent stuff)
  768. */
  769. #define MAX_ITEM (120 + l_mathlim(MAX_10_EXP))
  770. /* valid flags in a format specification */
  771. #define FLAGS "-+ #0"
  772. /*
  773. ** maximum size of each format specification (such as "%-099.99d")
  774. */
  775. #define MAX_FORMAT 32
  776. static void addquoted (luaL_Buffer *b, const char *s, size_t len) {
  777. luaL_addchar(b, '"');
  778. while (len--) {
  779. if (*s == '"' || *s == '\\' || *s == '\n') {
  780. luaL_addchar(b, '\\');
  781. luaL_addchar(b, *s);
  782. }
  783. else if (iscntrl(uchar(*s))) {
  784. char buff[10];
  785. if (!isdigit(uchar(*(s+1))))
  786. l_sprintf(buff, sizeof(buff), "\\%d", (int)uchar(*s));
  787. else
  788. l_sprintf(buff, sizeof(buff), "\\%03d", (int)uchar(*s));
  789. luaL_addstring(b, buff);
  790. }
  791. else
  792. luaL_addchar(b, *s);
  793. s++;
  794. }
  795. luaL_addchar(b, '"');
  796. }
  797. /*
  798. ** Convert a Lua number to a string in floating-point hexadecimal
  799. ** format. Ensures the resulting string uses a dot as the radix
  800. ** character.
  801. */
  802. static void addliteralnum (lua_State *L, luaL_Buffer *b, lua_Number n) {
  803. char *buff = luaL_prepbuffsize(b, MAX_ITEM);
  804. int nb = lua_number2strx(L, buff, MAX_ITEM,
  805. "%" LUA_NUMBER_FRMLEN "a", n);
  806. if (memchr(buff, '.', nb) == NULL) { /* no dot? */
  807. char point = lua_getlocaledecpoint(); /* try locale point */
  808. char *ppoint = memchr(buff, point, nb);
  809. if (ppoint) *ppoint = '.'; /* change it to a dot */
  810. }
  811. luaL_addsize(b, nb);
  812. }
  813. static void addliteral (lua_State *L, luaL_Buffer *b, int arg) {
  814. switch (lua_type(L, arg)) {
  815. case LUA_TSTRING: {
  816. size_t len;
  817. const char *s = lua_tolstring(L, arg, &len);
  818. addquoted(b, s, len);
  819. break;
  820. }
  821. case LUA_TNUMBER: {
  822. if (!lua_isinteger(L, arg)) { /* write floats as hexa ('%a') */
  823. addliteralnum(L, b, lua_tonumber(L, arg));
  824. break;
  825. }
  826. /* else integers; write in "native" format *//* FALLTHROUGH */
  827. }
  828. case LUA_TNIL: case LUA_TBOOLEAN: {
  829. luaL_tolstring(L, arg, NULL);
  830. luaL_addvalue(b);
  831. break;
  832. }
  833. default: {
  834. luaL_argerror(L, arg, "value has no literal form");
  835. }
  836. }
  837. }
  838. static const char *scanformat (lua_State *L, const char *strfrmt, char *form) {
  839. const char *p = strfrmt;
  840. while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++; /* skip flags */
  841. if ((size_t)(p - strfrmt) >= sizeof(FLAGS)/sizeof(char))
  842. luaL_error(L, "invalid format (repeated flags)");
  843. if (isdigit(uchar(*p))) p++; /* skip width */
  844. if (isdigit(uchar(*p))) p++; /* (2 digits at most) */
  845. if (*p == '.') {
  846. p++;
  847. if (isdigit(uchar(*p))) p++; /* skip precision */
  848. if (isdigit(uchar(*p))) p++; /* (2 digits at most) */
  849. }
  850. if (isdigit(uchar(*p)))
  851. luaL_error(L, "invalid format (width or precision too long)");
  852. *(form++) = '%';
  853. memcpy(form, strfrmt, ((p - strfrmt) + 1) * sizeof(char));
  854. form += (p - strfrmt) + 1;
  855. *form = '\0';
  856. return p;
  857. }
  858. /*
  859. ** add length modifier into formats
  860. */
  861. static void addlenmod (char *form, const char *lenmod) {
  862. size_t l = strlen(form);
  863. size_t lm = strlen(lenmod);
  864. char spec = form[l - 1];
  865. strcpy(form + l - 1, lenmod);
  866. form[l + lm - 1] = spec;
  867. form[l + lm] = '\0';
  868. }
  869. static int str_format (lua_State *L) {
  870. int top = lua_gettop(L);
  871. int arg = 1;
  872. size_t sfl;
  873. const char *strfrmt = luaL_checklstring(L, arg, &sfl);
  874. const char *strfrmt_end = strfrmt+sfl;
  875. luaL_Buffer b;
  876. luaL_buffinit(L, &b);
  877. while (strfrmt < strfrmt_end) {
  878. if (*strfrmt != L_ESC)
  879. luaL_addchar(&b, *strfrmt++);
  880. else if (*++strfrmt == L_ESC)
  881. luaL_addchar(&b, *strfrmt++); /* %% */
  882. else { /* format item */
  883. char form[MAX_FORMAT]; /* to store the format ('%...') */
  884. char *buff = luaL_prepbuffsize(&b, MAX_ITEM); /* to put formatted item */
  885. int nb = 0; /* number of bytes in added item */
  886. if (++arg > top)
  887. luaL_argerror(L, arg, "no value");
  888. strfrmt = scanformat(L, strfrmt, form);
  889. switch (*strfrmt++) {
  890. case 'c': {
  891. nb = l_sprintf(buff, MAX_ITEM, form, (int)luaL_checkinteger(L, arg));
  892. break;
  893. }
  894. case 'd': case 'i':
  895. case 'o': case 'u': case 'x': case 'X': {
  896. lua_Integer n = luaL_checkinteger(L, arg);
  897. addlenmod(form, LUA_INTEGER_FRMLEN);
  898. nb = l_sprintf(buff, MAX_ITEM, form, n);
  899. break;
  900. }
  901. case 'a': case 'A':
  902. addlenmod(form, LUA_NUMBER_FRMLEN);
  903. nb = lua_number2strx(L, buff, MAX_ITEM, form,
  904. luaL_checknumber(L, arg));
  905. break;
  906. case 'e': case 'E': case 'f':
  907. case 'g': case 'G': {
  908. addlenmod(form, LUA_NUMBER_FRMLEN);
  909. nb = l_sprintf(buff, MAX_ITEM, form, luaL_checknumber(L, arg));
  910. break;
  911. }
  912. case 'q': {
  913. addliteral(L, &b, arg);
  914. break;
  915. }
  916. case 's': {
  917. size_t l;
  918. const char *s = luaL_tolstring(L, arg, &l);
  919. if (form[2] == '\0') /* no modifiers? */
  920. luaL_addvalue(&b); /* keep entire string */
  921. else {
  922. luaL_argcheck(L, l == strlen(s), arg, "string contains zeros");
  923. if (!strchr(form, '.') && l >= 100) {
  924. /* no precision and string is too long to be formatted */
  925. luaL_addvalue(&b); /* keep entire string */
  926. }
  927. else { /* format the string into 'buff' */
  928. nb = l_sprintf(buff, MAX_ITEM, form, s);
  929. lua_pop(L, 1); /* remove result from 'luaL_tolstring' */
  930. }
  931. }
  932. break;
  933. }
  934. default: { /* also treat cases 'pnLlh' */
  935. return luaL_error(L, "invalid option '%%%c' to 'format'",
  936. *(strfrmt - 1));
  937. }
  938. }
  939. lua_assert(nb < MAX_ITEM);
  940. luaL_addsize(&b, nb);
  941. }
  942. }
  943. luaL_pushresult(&b);
  944. return 1;
  945. }
  946. /* }====================================================== */
  947. /*
  948. ** {======================================================
  949. ** PACK/UNPACK
  950. ** =======================================================
  951. */
  952. /* value used for padding */
  953. #if !defined(LUAL_PACKPADBYTE)
  954. #define LUAL_PACKPADBYTE 0x00
  955. #endif
  956. /* maximum size for the binary representation of an integer */
  957. #define MAXINTSIZE 16
  958. /* number of bits in a character */
  959. #define NB CHAR_BIT
  960. /* mask for one character (NB 1's) */
  961. #define MC ((1 << NB) - 1)
  962. /* size of a lua_Integer */
  963. #define SZINT ((int)sizeof(lua_Integer))
  964. /* dummy union to get native endianness */
  965. static const union {
  966. int dummy;
  967. char little; /* true iff machine is little endian */
  968. } nativeendian = {1};
  969. /* dummy structure to get native alignment requirements */
  970. struct cD {
  971. char c;
  972. union { double d; void *p; lua_Integer i; lua_Number n; } u;
  973. };
  974. #define MAXALIGN (offsetof(struct cD, u))
  975. /*
  976. ** Union for serializing floats
  977. */
  978. typedef union Ftypes {
  979. float f;
  980. double d;
  981. lua_Number n;
  982. char buff[5 * sizeof(lua_Number)]; /* enough for any float type */
  983. } Ftypes;
  984. /*
  985. ** information to pack/unpack stuff
  986. */
  987. typedef struct Header {
  988. lua_State *L;
  989. int islittle;
  990. int maxalign;
  991. } Header;
  992. /*
  993. ** options for pack/unpack
  994. */
  995. typedef enum KOption {
  996. Kint, /* signed integers */
  997. Kuint, /* unsigned integers */
  998. Kfloat, /* floating-point numbers */
  999. Kchar, /* fixed-length strings */
  1000. Kstring, /* strings with prefixed length */
  1001. Kzstr, /* zero-terminated strings */
  1002. Kpadding, /* padding */
  1003. Kpaddalign, /* padding for alignment */
  1004. Knop /* no-op (configuration or spaces) */
  1005. } KOption;
  1006. /*
  1007. ** Read an integer numeral from string 'fmt' or return 'df' if
  1008. ** there is no numeral
  1009. */
  1010. static int digit (int c) { return '0' <= c && c <= '9'; }
  1011. static int getnum (const char **fmt, int df) {
  1012. if (!digit(**fmt)) /* no number? */
  1013. return df; /* return default value */
  1014. else {
  1015. int a = 0;
  1016. do {
  1017. a = a*10 + (*((*fmt)++) - '0');
  1018. } while (digit(**fmt) && a <= ((int)MAXSIZE - 9)/10);
  1019. return a;
  1020. }
  1021. }
  1022. /*
  1023. ** Read an integer numeral and raises an error if it is larger
  1024. ** than the maximum size for integers.
  1025. */
  1026. static int getnumlimit (Header *h, const char **fmt, int df) {
  1027. int sz = getnum(fmt, df);
  1028. if (sz > MAXINTSIZE || sz <= 0)
  1029. luaL_error(h->L, "integral size (%d) out of limits [1,%d]",
  1030. sz, MAXINTSIZE);
  1031. return sz;
  1032. }
  1033. /*
  1034. ** Initialize Header
  1035. */
  1036. static void initheader (lua_State *L, Header *h) {
  1037. h->L = L;
  1038. h->islittle = nativeendian.little;
  1039. h->maxalign = 1;
  1040. }
  1041. /*
  1042. ** Read and classify next option. 'size' is filled with option's size.
  1043. */
  1044. static KOption getoption (Header *h, const char **fmt, int *size) {
  1045. int opt = *((*fmt)++);
  1046. *size = 0; /* default */
  1047. switch (opt) {
  1048. case 'b': *size = sizeof(char); return Kint;
  1049. case 'B': *size = sizeof(char); return Kuint;
  1050. case 'h': *size = sizeof(short); return Kint;
  1051. case 'H': *size = sizeof(short); return Kuint;
  1052. case 'l': *size = sizeof(long); return Kint;
  1053. case 'L': *size = sizeof(long); return Kuint;
  1054. case 'j': *size = sizeof(lua_Integer); return Kint;
  1055. case 'J': *size = sizeof(lua_Integer); return Kuint;
  1056. case 'T': *size = sizeof(size_t); return Kuint;
  1057. case 'f': *size = sizeof(float); return Kfloat;
  1058. case 'd': *size = sizeof(double); return Kfloat;
  1059. case 'n': *size = sizeof(lua_Number); return Kfloat;
  1060. case 'i': *size = getnumlimit(h, fmt, sizeof(int)); return Kint;
  1061. case 'I': *size = getnumlimit(h, fmt, sizeof(int)); return Kuint;
  1062. case 's': *size = getnumlimit(h, fmt, sizeof(size_t)); return Kstring;
  1063. case 'c':
  1064. *size = getnum(fmt, -1);
  1065. if (*size == -1)
  1066. luaL_error(h->L, "missing size for format option 'c'");
  1067. return Kchar;
  1068. case 'z': return Kzstr;
  1069. case 'x': *size = 1; return Kpadding;
  1070. case 'X': return Kpaddalign;
  1071. case ' ': break;
  1072. case '<': h->islittle = 1; break;
  1073. case '>': h->islittle = 0; break;
  1074. case '=': h->islittle = nativeendian.little; break;
  1075. case '!': h->maxalign = getnumlimit(h, fmt, MAXALIGN); break;
  1076. default: luaL_error(h->L, "invalid format option '%c'", opt);
  1077. }
  1078. return Knop;
  1079. }
  1080. /*
  1081. ** Read, classify, and fill other details about the next option.
  1082. ** 'psize' is filled with option's size, 'notoalign' with its
  1083. ** alignment requirements.
  1084. ** Local variable 'size' gets the size to be aligned. (Kpadal option
  1085. ** always gets its full alignment, other options are limited by
  1086. ** the maximum alignment ('maxalign'). Kchar option needs no alignment
  1087. ** despite its size.
  1088. */
  1089. static KOption getdetails (Header *h, size_t totalsize,
  1090. const char **fmt, int *psize, int *ntoalign) {
  1091. KOption opt = getoption(h, fmt, psize);
  1092. int align = *psize; /* usually, alignment follows size */
  1093. if (opt == Kpaddalign) { /* 'X' gets alignment from following option */
  1094. if (**fmt == '\0' || getoption(h, fmt, &align) == Kchar || align == 0)
  1095. luaL_argerror(h->L, 1, "invalid next option for option 'X'");
  1096. }
  1097. if (align <= 1 || opt == Kchar) /* need no alignment? */
  1098. *ntoalign = 0;
  1099. else {
  1100. if (align > h->maxalign) /* enforce maximum alignment */
  1101. align = h->maxalign;
  1102. if ((align & (align - 1)) != 0) /* is 'align' not a power of 2? */
  1103. luaL_argerror(h->L, 1, "format asks for alignment not power of 2");
  1104. *ntoalign = (align - (int)(totalsize & (align - 1))) & (align - 1);
  1105. }
  1106. return opt;
  1107. }
  1108. /*
  1109. ** Pack integer 'n' with 'size' bytes and 'islittle' endianness.
  1110. ** The final 'if' handles the case when 'size' is larger than
  1111. ** the size of a Lua integer, correcting the extra sign-extension
  1112. ** bytes if necessary (by default they would be zeros).
  1113. */
  1114. static void packint (luaL_Buffer *b, lua_Unsigned n,
  1115. int islittle, int size, int neg) {
  1116. char *buff = luaL_prepbuffsize(b, size);
  1117. int i;
  1118. buff[islittle ? 0 : size - 1] = (char)(n & MC); /* first byte */
  1119. for (i = 1; i < size; i++) {
  1120. n >>= NB;
  1121. buff[islittle ? i : size - 1 - i] = (char)(n & MC);
  1122. }
  1123. if (neg && size > SZINT) { /* negative number need sign extension? */
  1124. for (i = SZINT; i < size; i++) /* correct extra bytes */
  1125. buff[islittle ? i : size - 1 - i] = (char)MC;
  1126. }
  1127. luaL_addsize(b, size); /* add result to buffer */
  1128. }
  1129. /*
  1130. ** Copy 'size' bytes from 'src' to 'dest', correcting endianness if
  1131. ** given 'islittle' is different from native endianness.
  1132. */
  1133. static void copywithendian (volatile char *dest, volatile const char *src,
  1134. int size, int islittle) {
  1135. if (islittle == nativeendian.little) {
  1136. while (size-- != 0)
  1137. *(dest++) = *(src++);
  1138. }
  1139. else {
  1140. dest += size - 1;
  1141. while (size-- != 0)
  1142. *(dest--) = *(src++);
  1143. }
  1144. }
  1145. static int str_pack (lua_State *L) {
  1146. luaL_Buffer b;
  1147. Header h;
  1148. const char *fmt = luaL_checkstring(L, 1); /* format string */
  1149. int arg = 1; /* current argument to pack */
  1150. size_t totalsize = 0; /* accumulate total size of result */
  1151. initheader(L, &h);
  1152. lua_pushnil(L); /* mark to separate arguments from string buffer */
  1153. luaL_buffinit(L, &b);
  1154. while (*fmt != '\0') {
  1155. int size, ntoalign;
  1156. KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign);
  1157. totalsize += ntoalign + size;
  1158. while (ntoalign-- > 0)
  1159. luaL_addchar(&b, LUAL_PACKPADBYTE); /* fill alignment */
  1160. arg++;
  1161. switch (opt) {
  1162. case Kint: { /* signed integers */
  1163. lua_Integer n = luaL_checkinteger(L, arg);
  1164. if (size < SZINT) { /* need overflow check? */
  1165. lua_Integer lim = (lua_Integer)1 << ((size * NB) - 1);
  1166. luaL_argcheck(L, -lim <= n && n < lim, arg, "integer overflow");
  1167. }
  1168. packint(&b, (lua_Unsigned)n, h.islittle, size, (n < 0));
  1169. break;
  1170. }
  1171. case Kuint: { /* unsigned integers */
  1172. lua_Integer n = luaL_checkinteger(L, arg);
  1173. if (size < SZINT) /* need overflow check? */
  1174. luaL_argcheck(L, (lua_Unsigned)n < ((lua_Unsigned)1 << (size * NB)),
  1175. arg, "unsigned overflow");
  1176. packint(&b, (lua_Unsigned)n, h.islittle, size, 0);
  1177. break;
  1178. }
  1179. case Kfloat: { /* floating-point options */
  1180. volatile Ftypes u;
  1181. char *buff = luaL_prepbuffsize(&b, size);
  1182. lua_Number n = luaL_checknumber(L, arg); /* get argument */
  1183. if (size == sizeof(u.f)) u.f = (float)n; /* copy it into 'u' */
  1184. else if (size == sizeof(u.d)) u.d = (double)n;
  1185. else u.n = n;
  1186. /* move 'u' to final result, correcting endianness if needed */
  1187. copywithendian(buff, u.buff, size, h.islittle);
  1188. luaL_addsize(&b, size);
  1189. break;
  1190. }
  1191. case Kchar: { /* fixed-size string */
  1192. size_t len;
  1193. const char *s = luaL_checklstring(L, arg, &len);
  1194. if ((size_t)size <= len) /* string larger than (or equal to) needed? */
  1195. luaL_addlstring(&b, s, size); /* truncate string to asked size */
  1196. else { /* string smaller than needed */
  1197. luaL_addlstring(&b, s, len); /* add it all */
  1198. while (len++ < (size_t)size) /* pad extra space */
  1199. luaL_addchar(&b, LUAL_PACKPADBYTE);
  1200. }
  1201. break;
  1202. }
  1203. case Kstring: { /* strings with length count */
  1204. size_t len;
  1205. const char *s = luaL_checklstring(L, arg, &len);
  1206. luaL_argcheck(L, size >= (int)sizeof(size_t) ||
  1207. len < ((size_t)1 << (size * NB)),
  1208. arg, "string length does not fit in given size");
  1209. packint(&b, (lua_Unsigned)len, h.islittle, size, 0); /* pack length */
  1210. luaL_addlstring(&b, s, len);
  1211. totalsize += len;
  1212. break;
  1213. }
  1214. case Kzstr: { /* zero-terminated string */
  1215. size_t len;
  1216. const char *s = luaL_checklstring(L, arg, &len);
  1217. luaL_argcheck(L, strlen(s) == len, arg, "string contains zeros");
  1218. luaL_addlstring(&b, s, len);
  1219. luaL_addchar(&b, '\0'); /* add zero at the end */
  1220. totalsize += len + 1;
  1221. break;
  1222. }
  1223. case Kpadding: luaL_addchar(&b, LUAL_PACKPADBYTE); /* FALLTHROUGH */
  1224. case Kpaddalign: case Knop:
  1225. arg--; /* undo increment */
  1226. break;
  1227. }
  1228. }
  1229. luaL_pushresult(&b);
  1230. return 1;
  1231. }
  1232. static int str_packsize (lua_State *L) {
  1233. Header h;
  1234. const char *fmt = luaL_checkstring(L, 1); /* format string */
  1235. size_t totalsize = 0; /* accumulate total size of result */
  1236. initheader(L, &h);
  1237. while (*fmt != '\0') {
  1238. int size, ntoalign;
  1239. KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign);
  1240. size += ntoalign; /* total space used by option */
  1241. luaL_argcheck(L, totalsize <= MAXSIZE - size, 1,
  1242. "format result too large");
  1243. totalsize += size;
  1244. switch (opt) {
  1245. case Kstring: /* strings with length count */
  1246. case Kzstr: /* zero-terminated string */
  1247. luaL_argerror(L, 1, "variable-length format");
  1248. /* call never return, but to avoid warnings: *//* FALLTHROUGH */
  1249. default: break;
  1250. }
  1251. }
  1252. lua_pushinteger(L, (lua_Integer)totalsize);
  1253. return 1;
  1254. }
  1255. /*
  1256. ** Unpack an integer with 'size' bytes and 'islittle' endianness.
  1257. ** If size is smaller than the size of a Lua integer and integer
  1258. ** is signed, must do sign extension (propagating the sign to the
  1259. ** higher bits); if size is larger than the size of a Lua integer,
  1260. ** it must check the unread bytes to see whether they do not cause an
  1261. ** overflow.
  1262. */
  1263. static lua_Integer unpackint (lua_State *L, const char *str,
  1264. int islittle, int size, int issigned) {
  1265. lua_Unsigned res = 0;
  1266. int i;
  1267. int limit = (size <= SZINT) ? size : SZINT;
  1268. for (i = limit - 1; i >= 0; i--) {
  1269. res <<= NB;
  1270. res |= (lua_Unsigned)(unsigned char)str[islittle ? i : size - 1 - i];
  1271. }
  1272. if (size < SZINT) { /* real size smaller than lua_Integer? */
  1273. if (issigned) { /* needs sign extension? */
  1274. lua_Unsigned mask = (lua_Unsigned)1 << (size*NB - 1);
  1275. res = ((res ^ mask) - mask); /* do sign extension */
  1276. }
  1277. }
  1278. else if (size > SZINT) { /* must check unread bytes */
  1279. int mask = (!issigned || (lua_Integer)res >= 0) ? 0 : MC;
  1280. for (i = limit; i < size; i++) {
  1281. if ((unsigned char)str[islittle ? i : size - 1 - i] != mask)
  1282. luaL_error(L, "%d-byte integer does not fit into Lua Integer", size);
  1283. }
  1284. }
  1285. return (lua_Integer)res;
  1286. }
  1287. static int str_unpack (lua_State *L) {
  1288. Header h;
  1289. const char *fmt = luaL_checkstring(L, 1);
  1290. size_t ld;
  1291. const char *data = luaL_checklstring(L, 2, &ld);
  1292. size_t pos = (size_t)posrelat(luaL_optinteger(L, 3, 1), ld) - 1;
  1293. int n = 0; /* number of results */
  1294. luaL_argcheck(L, pos <= ld, 3, "initial position out of string");
  1295. initheader(L, &h);
  1296. while (*fmt != '\0') {
  1297. int size, ntoalign;
  1298. KOption opt = getdetails(&h, pos, &fmt, &size, &ntoalign);
  1299. if ((size_t)ntoalign + size > ~pos || pos + ntoalign + size > ld)
  1300. luaL_argerror(L, 2, "data string too short");
  1301. pos += ntoalign; /* skip alignment */
  1302. /* stack space for item + next position */
  1303. luaL_checkstack(L, 2, "too many results");
  1304. n++;
  1305. switch (opt) {
  1306. case Kint:
  1307. case Kuint: {
  1308. lua_Integer res = unpackint(L, data + pos, h.islittle, size,
  1309. (opt == Kint));
  1310. lua_pushinteger(L, res);
  1311. break;
  1312. }
  1313. case Kfloat: {
  1314. volatile Ftypes u;
  1315. lua_Number num;
  1316. copywithendian(u.buff, data + pos, size, h.islittle);
  1317. if (size == sizeof(u.f)) num = (lua_Number)u.f;
  1318. else if (size == sizeof(u.d)) num = (lua_Number)u.d;
  1319. else num = u.n;
  1320. lua_pushnumber(L, num);
  1321. break;
  1322. }
  1323. case Kchar: {
  1324. lua_pushlstring(L, data + pos, size);
  1325. break;
  1326. }
  1327. case Kstring: {
  1328. size_t len = (size_t)unpackint(L, data + pos, h.islittle, size, 0);
  1329. luaL_argcheck(L, pos + len + size <= ld, 2, "data string too short");
  1330. lua_pushlstring(L, data + pos + size, len);
  1331. pos += len; /* skip string */
  1332. break;
  1333. }
  1334. case Kzstr: {
  1335. size_t len = (int)strlen(data + pos);
  1336. lua_pushlstring(L, data + pos, len);
  1337. pos += len + 1; /* skip string plus final '\0' */
  1338. break;
  1339. }
  1340. case Kpaddalign: case Kpadding: case Knop:
  1341. n--; /* undo increment */
  1342. break;
  1343. }
  1344. pos += size;
  1345. }
  1346. lua_pushinteger(L, pos + 1); /* next position */
  1347. return n + 1;
  1348. }
  1349. /* }====================================================== */
  1350. static const luaL_Reg strlib[] = {
  1351. {"byte", str_byte},
  1352. {"char", str_char},
  1353. {"dump", str_dump},
  1354. {"find", str_find},
  1355. {"format", str_format},
  1356. {"gmatch", gmatch},
  1357. {"gsub", str_gsub},
  1358. {"len", str_len},
  1359. {"lower", str_lower},
  1360. {"match", str_match},
  1361. {"rep", str_rep},
  1362. {"reverse", str_reverse},
  1363. {"sub", str_sub},
  1364. {"upper", str_upper},
  1365. {"pack", str_pack},
  1366. {"packsize", str_packsize},
  1367. {"unpack", str_unpack},
  1368. {NULL, NULL}
  1369. };
  1370. static void createmetatable (lua_State *L) {
  1371. lua_createtable(L, 0, 1); /* table to be metatable for strings */
  1372. lua_pushliteral(L, ""); /* dummy string */
  1373. lua_pushvalue(L, -2); /* copy table */
  1374. lua_setmetatable(L, -2); /* set table as metatable for strings */
  1375. lua_pop(L, 1); /* pop dummy string */
  1376. lua_pushvalue(L, -2); /* get string library */
  1377. lua_setfield(L, -2, "__index"); /* metatable.__index = string */
  1378. lua_pop(L, 1); /* pop metatable */
  1379. }
  1380. /*
  1381. ** Open string library
  1382. */
  1383. LUAMOD_API int luaopen_string (lua_State *L) {
  1384. luaL_newlib(L, strlib);
  1385. createmetatable(L);
  1386. return 1;
  1387. }