sqstdstring.cpp 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517
  1. /* see copyright notice in squirrel.h */
  2. #include <squirrel.h>
  3. #include <sqstdstring.h>
  4. #include <string.h>
  5. #include <stdlib.h>
  6. #include <stdio.h>
  7. #include <ctype.h>
  8. #include <assert.h>
  9. #include <stddef.h>
  10. #define MAX_FORMAT_LEN 20
  11. #define MAX_WFORMAT_LEN 3
  12. #define ADDITIONAL_FORMAT_SPACE (100*sizeof(SQChar))
  13. static SQRESULT validate_format(HSQUIRRELVM v, SQChar *fmt, const SQChar *src, SQInteger n,SQInteger &width)
  14. {
  15. SQChar swidth[MAX_WFORMAT_LEN];
  16. SQInteger wc = 0;
  17. SQInteger start = n;
  18. fmt[0] = '%';
  19. while (scstrchr(_SC("-+ #0"), src[n])) n++;
  20. while (scisdigit(src[n])) {
  21. swidth[wc] = src[n];
  22. n++;
  23. wc++;
  24. if(wc>=MAX_WFORMAT_LEN)
  25. return sq_throwerror(v,_SC("width format too long"));
  26. }
  27. swidth[wc] = '\0';
  28. if(wc > 0) {
  29. width = scatoi(swidth);
  30. }
  31. else
  32. width = 0;
  33. if (src[n] == '.') {
  34. n++;
  35. wc = 0;
  36. while (scisdigit(src[n])) {
  37. swidth[wc] = src[n];
  38. n++;
  39. wc++;
  40. if(wc>=MAX_WFORMAT_LEN)
  41. return sq_throwerror(v,_SC("precision format too long"));
  42. }
  43. swidth[wc] = '\0';
  44. if(wc > 0) {
  45. width += scatoi(swidth);
  46. }
  47. }
  48. if (n-start > MAX_FORMAT_LEN )
  49. return sq_throwerror(v,_SC("format too long"));
  50. memcpy(&fmt[1],&src[start],((n-start)+1)*sizeof(SQChar));
  51. fmt[(n-start)+2] = '\0';
  52. return n;
  53. }
  54. SQRESULT sqstd_format(HSQUIRRELVM v,SQInteger nformatstringidx,SQInteger *outlen,SQChar **output)
  55. {
  56. const SQChar *format;
  57. SQChar *dest;
  58. SQChar fmt[MAX_FORMAT_LEN];
  59. sq_getstring(v,nformatstringidx,&format);
  60. SQInteger format_size = sq_getsize(v,nformatstringidx);
  61. SQInteger allocated = (format_size+2)*sizeof(SQChar);
  62. dest = sq_getscratchpad(v,allocated);
  63. SQInteger n = 0,i = 0, nparam = nformatstringidx+1, w = 0;
  64. while(n < format_size) {
  65. if(format[n] != '%') {
  66. assert(i < allocated);
  67. dest[i++] = format[n];
  68. n++;
  69. }
  70. else if(format[n+1] == '%') { //handles %%
  71. dest[i++] = '%';
  72. n += 2;
  73. }
  74. else {
  75. n++;
  76. if( nparam > sq_gettop(v) )
  77. return sq_throwerror(v,_SC("not enough paramters for the given format string"));
  78. n = validate_format(v,fmt,format,n,w);
  79. if(n < 0) return -1;
  80. SQInteger addlen = 0;
  81. SQInteger valtype = 0;
  82. const SQChar *ts;
  83. SQInteger ti;
  84. SQFloat tf;
  85. SQInteger fc = format[n];
  86. switch(fc) {
  87. case 'q':
  88. case 's':
  89. if(SQ_FAILED(sq_getstring(v,nparam,&ts)))
  90. return sq_throwerror(v,_SC("string expected for the specified format"));
  91. if(fc == 'q'){
  92. addlen = 2; //quotes before and after
  93. SQInteger size = sq_getsize(v,nparam);
  94. SQChar *ts2 = (SQChar*)ts;
  95. while (size--) {
  96. ++addlen;
  97. if (*ts2 == '\r' && *(ts2+1) == '\n' ) {
  98. ++addlen;
  99. ++ts2; //eat \r and output only \n
  100. }
  101. else if (*ts2 == '"' || *ts2 == '\\' || *ts2 == '\n') {
  102. ++addlen;
  103. }
  104. else if (*ts2 == '\0' || iscntrl(uchar(*ts2))) {
  105. SQChar buff[10];
  106. addlen += scsprintf(buff, "\\x%x", (int)uchar(*ts2))-1; //-1 because we already added the original char to the sum
  107. }
  108. ++ts2;
  109. }
  110. ts2 = (SQChar*)i; //save the i position using pointer as integer
  111. i += addlen;
  112. allocated += addlen;
  113. dest = sq_getscratchpad(v,allocated);
  114. size = sq_getsize(v,nparam);
  115. ts2 = &dest[(ptrdiff_t)ts2]; //use saved i position saved on pointer as integer
  116. *ts2++ = '"';
  117. while (size--) {
  118. if (*ts == '\r' && *(ts+1) == '\n' ) {
  119. ++ts; //eat \r and output only \n
  120. }
  121. if (*ts == '"' || *ts == '\\') {
  122. *ts2++ = '\\';
  123. *ts2++ = *ts;
  124. }
  125. else if (*ts == '\n') {
  126. *ts2++ = '\\';
  127. *ts2++ = 'n';
  128. }
  129. else if (*ts == '\0' || iscntrl(uchar(*ts))) {
  130. SQChar buff[10];
  131. int iw;
  132. iw = scsprintf(buff, "\\x%x", (int)uchar(*ts));
  133. for(int i=0; i< iw; ++i) *ts2++ = buff[i];
  134. }
  135. else
  136. *ts2++ = *ts;
  137. ++ts;
  138. }
  139. *ts2++ = '"';
  140. ++n;
  141. ++nparam;
  142. continue;
  143. }
  144. else
  145. {
  146. addlen = (sq_getsize(v,nparam)*sizeof(SQChar))+((w+1)*sizeof(SQChar));
  147. valtype = 's';
  148. }
  149. break;
  150. case 'i': case 'd': case 'o': case 'u': case 'x': case 'X':
  151. #ifdef _SQ64
  152. {
  153. size_t flen = scstrlen(fmt);
  154. SQInteger fpos = flen - 1;
  155. SQChar f = fmt[fpos];
  156. SQChar *prec = (SQChar *)_PRINT_INT_PREC;
  157. while(*prec != _SC('\0')) {
  158. fmt[fpos++] = *prec++;
  159. }
  160. fmt[fpos++] = f;
  161. fmt[fpos++] = _SC('\0');
  162. }
  163. #endif
  164. case 'c':
  165. if(SQ_FAILED(sq_getinteger(v,nparam,&ti)))
  166. return sq_throwerror(v,_SC("integer expected for the specified format"));
  167. addlen = (ADDITIONAL_FORMAT_SPACE)+((w+1)*sizeof(SQChar));
  168. valtype = 'i';
  169. break;
  170. case 'f': case 'g': case 'G': case 'e': case 'E':
  171. if(SQ_FAILED(sq_getfloat(v,nparam,&tf)))
  172. return sq_throwerror(v,_SC("float expected for the specified format"));
  173. addlen = (ADDITIONAL_FORMAT_SPACE)+((w+1)*sizeof(SQChar));
  174. valtype = 'f';
  175. break;
  176. default:
  177. return sq_throwerror(v,_SC("invalid format"));
  178. }
  179. n++;
  180. allocated += addlen + sizeof(SQChar);
  181. dest = sq_getscratchpad(v,allocated);
  182. switch(valtype) {
  183. case 's': i += scsprintf(&dest[i],fmt,ts); break;
  184. case 'i': i += scsprintf(&dest[i],fmt,ti); break;
  185. case 'f': i += scsprintf(&dest[i],fmt,tf); break;
  186. };
  187. nparam ++;
  188. }
  189. }
  190. *outlen = i;
  191. dest[i] = '\0';
  192. *output = dest;
  193. return SQ_OK;
  194. }
  195. static SQRESULT _string_format(HSQUIRRELVM v)
  196. {
  197. SQChar *dest = NULL;
  198. SQInteger length = 0;
  199. if(SQ_FAILED(sqstd_format(v,2,&length,&dest)))
  200. return -1;
  201. sq_pushstring(v,dest,length);
  202. return 1;
  203. }
  204. static SQRESULT _string_printf(HSQUIRRELVM v)
  205. {
  206. SQChar *dest = NULL;
  207. SQInteger length = 0;
  208. SQPRINTFUNCTION sqprint = sq_getprintfunc(v);
  209. if(sqprint){
  210. if(SQ_FAILED(sqstd_format(v,2,&length,&dest)))
  211. return -1;
  212. sqprint(v,_SC("%s"),dest);
  213. sq_pushinteger(v, length);
  214. return 1;
  215. }
  216. return 0;
  217. }
  218. #define SETUP_REX(v) \
  219. SQRex *self = NULL; \
  220. sq_getinstanceup(v,1,(SQUserPointer *)&self,0);
  221. static SQRESULT _rexobj_releasehook(SQUserPointer p, SQInteger size, HSQUIRRELVM v)
  222. {
  223. SQRex *self = ((SQRex *)p);
  224. sqstd_rex_free(self);
  225. return 1;
  226. }
  227. static SQRESULT _regexp_match(HSQUIRRELVM v)
  228. {
  229. SETUP_REX(v);
  230. const SQChar *str;
  231. sq_getstring(v,2,&str);
  232. if(sqstd_rex_match(self,str) == SQTrue)
  233. {
  234. sq_pushbool(v,SQTrue);
  235. return 1;
  236. }
  237. sq_pushbool(v,SQFalse);
  238. return 1;
  239. }
  240. static SQRESULT _regexp_gmatch(HSQUIRRELVM v)
  241. {
  242. SETUP_REX(v);
  243. const SQChar *str;
  244. SQInteger str_size;
  245. sq_getstring(v,2,&str);
  246. str_size = sq_getsize(v, 2);
  247. const SQChar *begin,*end;
  248. while(sqstd_rex_searchrange(self,str, str+str_size,&begin,&end)){
  249. SQInteger n = sqstd_rex_getsubexpcount(self);
  250. SQRexMatch match;
  251. sq_pushroottable(v); //this
  252. SQInteger i = 0;
  253. for(;i < n; i++) {
  254. sqstd_rex_getsubexp(self,i,&match);
  255. if(i > 0){ //skip whole match
  256. sq_pushstring(v, match.begin, match.len);
  257. }
  258. }
  259. i = sq_call(v, n, SQFalse, SQTrue);
  260. if(i < 0) return i;
  261. str_size -= end-str;
  262. str = end;
  263. }
  264. sq_pushbool(v,SQFalse);
  265. return 1;
  266. }
  267. #include "sqstdblobimpl.h"
  268. static SQRESULT _regexp_gsub(HSQUIRRELVM v)
  269. {
  270. SETUP_REX(v);
  271. const SQChar *str;
  272. SQInteger str_size;
  273. sq_getstring(v,2,&str);
  274. str_size = sq_getsize(v, 2);
  275. const SQChar *begin,*end;
  276. SQBlob blob(0,8192);
  277. SQObjectType ptype = sq_gettype(v, 3);
  278. const SQChar *replacement;
  279. SQInteger replacement_size;
  280. while(sqstd_rex_searchrange(self,str, str+str_size,&begin,&end)){
  281. blob.Write(str, begin-str);
  282. SQInteger n = sqstd_rex_getsubexpcount(self);
  283. SQRexMatch match;
  284. SQInteger i = 0;
  285. switch(ptype){
  286. case OT_CLOSURE:{
  287. sq_pushroottable(v); //this
  288. for(;i < n; i++) {
  289. sqstd_rex_getsubexp(self,i,&match);
  290. if(i > 0){ //skip whole match
  291. sq_pushstring(v, match.begin, match.len);
  292. }
  293. }
  294. i = sq_call(v, n, SQTrue, SQTrue);
  295. if(i < 0) return i;
  296. if(sq_gettype(v, -1) == OT_STRING){
  297. const SQChar *svalue;
  298. sq_getstring(v, -1, &svalue);
  299. blob.Write(svalue, sq_getsize(v, -1));
  300. }
  301. sq_poptop(v);
  302. }
  303. break;
  304. case OT_ARRAY:{
  305. for(;i < n; i++) {
  306. sqstd_rex_getsubexp(self,i,&match);
  307. if(i > 0){ //skip whole match
  308. sq_pushinteger(v, i-1);
  309. if(SQ_SUCCEEDED(sq_get(v, 3)) &&
  310. SQ_SUCCEEDED(sq_getstr_and_size(v, -1, &replacement, &replacement_size))){
  311. blob.Write(replacement, replacement_size);
  312. sq_pop(v, 1); //remove value
  313. }
  314. }
  315. }
  316. }
  317. break;
  318. case OT_TABLE:{
  319. for(;i < n; i++) {
  320. sqstd_rex_getsubexp(self,i,&match);
  321. if(i > 0){ //skip whole match
  322. sq_pushstring(v, match.begin, match.len);
  323. if(SQ_SUCCEEDED(sq_get(v, 3)) &&
  324. SQ_SUCCEEDED(sq_getstr_and_size(v, -1, &replacement, &replacement_size))){
  325. blob.Write(replacement, replacement_size);
  326. sq_pop(v, 1); //remove value
  327. }
  328. }
  329. }
  330. }
  331. break;
  332. default:
  333. return sq_throwerror(v, _SC("gsub only works with closure, array, table for replacement"));
  334. }
  335. str_size -= end-str;
  336. str = end;
  337. }
  338. if(str_size) blob.Write(str, str_size);
  339. sq_pushstring(v, (const SQChar *)blob.GetBuf(), blob.Len());
  340. return 1;
  341. }
  342. static void _addrexmatch(HSQUIRRELVM v,const SQChar *str,const SQChar *begin,const SQChar *end)
  343. {
  344. sq_newtable(v);
  345. sq_pushstring(v,_SC("begin"),-1);
  346. sq_pushinteger(v,begin - str);
  347. sq_rawset(v,-3);
  348. sq_pushstring(v,_SC("end"),-1);
  349. sq_pushinteger(v,end - str);
  350. sq_rawset(v,-3);
  351. }
  352. static SQRESULT _regexp_search(HSQUIRRELVM v)
  353. {
  354. SETUP_REX(v);
  355. const SQChar *str,*begin,*end;
  356. SQInteger start = 0;
  357. sq_getstring(v,2,&str);
  358. if(sq_gettop(v) > 2) sq_getinteger(v,3,&start);
  359. if(sqstd_rex_search(self,str+start,&begin,&end) == SQTrue) {
  360. _addrexmatch(v,str,begin,end);
  361. return 1;
  362. }
  363. return 0;
  364. }
  365. static SQRESULT _regexp_capture(HSQUIRRELVM v)
  366. {
  367. SETUP_REX(v);
  368. const SQChar *str,*begin,*end;
  369. SQInteger start = 0;
  370. sq_getstring(v,2,&str);
  371. if(sq_gettop(v) > 2) sq_getinteger(v,3,&start);
  372. if(sqstd_rex_search(self,str+start,&begin,&end) == SQTrue) {
  373. SQInteger n = sqstd_rex_getsubexpcount(self);
  374. SQRexMatch match;
  375. sq_newarray(v,0);
  376. for(SQInteger i = 0;i < n; i++) {
  377. sqstd_rex_getsubexp(self,i,&match);
  378. if(match.len > 0)
  379. _addrexmatch(v,str,match.begin,match.begin+match.len);
  380. else
  381. _addrexmatch(v,str,str,str); //empty match
  382. sq_arrayappend(v,-2);
  383. }
  384. return 1;
  385. }
  386. return 0;
  387. }
  388. static SQRESULT _regexp_xcapture(HSQUIRRELVM v)
  389. {
  390. SETUP_REX(v);
  391. const SQChar *str,*begin,*end;
  392. SQInteger start = 0;
  393. sq_getstring(v,2,&str);
  394. if(sq_gettop(v) > 2) sq_getinteger(v,3,&start);
  395. if(sqstd_rex_search(self,str+start,&begin,&end) == SQTrue) {
  396. sq_pushbool(v, SQTrue);
  397. return 1;
  398. }
  399. return 0;
  400. }
  401. static SQRESULT _regexp_getxcapture(HSQUIRRELVM v)
  402. {
  403. SETUP_REX(v);
  404. SQInteger n, start;
  405. const SQChar *str;
  406. sq_getstring(v,2,&str);
  407. sq_getinteger(v,3,&n);
  408. SQRexMatch match;
  409. sqstd_rex_getsubexp(self,n,&match);
  410. if(match.len > 0){
  411. start = match.begin-str;
  412. sq_pushinteger(v, start);
  413. sq_arrayappend(v,-2);
  414. sq_pushinteger(v, start+match.len);
  415. sq_arrayappend(v,-2);
  416. sq_pushbool(v, SQTrue);
  417. return 1;
  418. }
  419. return 0;
  420. }
  421. static SQRESULT _regexp_subexpcount(HSQUIRRELVM v)
  422. {
  423. SETUP_REX(v);
  424. sq_pushinteger(v,sqstd_rex_getsubexpcount(self));
  425. return 1;
  426. }
  427. static SQRESULT _regexp_constructor(HSQUIRRELVM v)
  428. {
  429. const SQChar *error,*pattern;
  430. sq_getstring(v,2,&pattern);
  431. SQRex *rex = sqstd_rex_compile(pattern,&error);
  432. if(!rex) return sq_throwerror(v,error);
  433. sq_setinstanceup(v,1,rex);
  434. sq_setreleasehook(v,1,_rexobj_releasehook);
  435. return 0;
  436. }
  437. static SQRESULT _regexp__typeof(HSQUIRRELVM v)
  438. {
  439. sq_pushstring(v,_SC("regexp"),-1);
  440. return 1;
  441. }
  442. #define _DECL_REX_FUNC(name,nparams,pmask) {_SC(#name),_regexp_##name,nparams,pmask}
  443. static SQRegFunction rexobj_funcs[]={
  444. _DECL_REX_FUNC(constructor,2,_SC(".s")),
  445. _DECL_REX_FUNC(search,-2,_SC("xsn")),
  446. _DECL_REX_FUNC(match,2,_SC("xs")),
  447. _DECL_REX_FUNC(gmatch,3,_SC("xsc")),
  448. _DECL_REX_FUNC(gsub,3,_SC("xs c|a|t")),
  449. _DECL_REX_FUNC(capture,-2,_SC("xsn")),
  450. _DECL_REX_FUNC(xcapture,-2,_SC("xsn")),
  451. _DECL_REX_FUNC(getxcapture,4,_SC("xsna")),
  452. _DECL_REX_FUNC(subexpcount,1,_SC("x")),
  453. _DECL_REX_FUNC(_typeof,1,_SC("x")),
  454. {0,0}
  455. };
  456. #undef _DECL_REX_FUNC
  457. #define _DECL_FUNC(name,nparams,pmask) {_SC(#name),_string_##name,nparams,pmask}
  458. static SQRegFunction stringlib_funcs[]={
  459. _DECL_FUNC(printf,-2,_SC(".s")),
  460. _DECL_FUNC(format,-2,_SC(".s")),
  461. {0,0}
  462. };
  463. #undef _DECL_FUNC
  464. SQInteger sqstd_register_stringlib(HSQUIRRELVM v)
  465. {
  466. sq_pushstring(v,_SC("regexp"),-1);
  467. sq_newclass(v,SQFalse);
  468. SQInteger i = 0;
  469. while(rexobj_funcs[i].name != 0) {
  470. SQRegFunction &f = rexobj_funcs[i];
  471. sq_pushstring(v,f.name,-1);
  472. sq_newclosure(v,f.f,0);
  473. sq_setparamscheck(v,f.nparamscheck,f.typemask);
  474. sq_setnativeclosurename(v,-1,f.name);
  475. sq_newslot(v,-3,SQFalse);
  476. i++;
  477. }
  478. sq_newslot(v,-3,SQFalse);
  479. i = 0;
  480. while(stringlib_funcs[i].name!=0)
  481. {
  482. sq_pushstring(v,stringlib_funcs[i].name,-1);
  483. sq_newclosure(v,stringlib_funcs[i].f,0);
  484. sq_setparamscheck(v,stringlib_funcs[i].nparamscheck,stringlib_funcs[i].typemask);
  485. sq_setnativeclosurename(v,-1,stringlib_funcs[i].name);
  486. sq_newslot(v,-3,SQFalse);
  487. i++;
  488. }
  489. return 1;
  490. }