sqstdstring.cpp 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516
  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. #define MAX_FORMAT_LEN 20
  10. #define MAX_WFORMAT_LEN 3
  11. #define ADDITIONAL_FORMAT_SPACE (100*sizeof(SQChar))
  12. static SQRESULT validate_format(HSQUIRRELVM v, SQChar *fmt, const SQChar *src, SQInteger n,SQInteger &width)
  13. {
  14. SQChar swidth[MAX_WFORMAT_LEN];
  15. SQInteger wc = 0;
  16. SQInteger start = n;
  17. fmt[0] = '%';
  18. while (scstrchr(_SC("-+ #0"), src[n])) n++;
  19. while (scisdigit(src[n])) {
  20. swidth[wc] = src[n];
  21. n++;
  22. wc++;
  23. if(wc>=MAX_WFORMAT_LEN)
  24. return sq_throwerror(v,_SC("width format too long"));
  25. }
  26. swidth[wc] = '\0';
  27. if(wc > 0) {
  28. width = scatoi(swidth);
  29. }
  30. else
  31. width = 0;
  32. if (src[n] == '.') {
  33. n++;
  34. wc = 0;
  35. while (scisdigit(src[n])) {
  36. swidth[wc] = src[n];
  37. n++;
  38. wc++;
  39. if(wc>=MAX_WFORMAT_LEN)
  40. return sq_throwerror(v,_SC("precision format too long"));
  41. }
  42. swidth[wc] = '\0';
  43. if(wc > 0) {
  44. width += scatoi(swidth);
  45. }
  46. }
  47. if (n-start > MAX_FORMAT_LEN )
  48. return sq_throwerror(v,_SC("format too long"));
  49. memcpy(&fmt[1],&src[start],((n-start)+1)*sizeof(SQChar));
  50. fmt[(n-start)+2] = '\0';
  51. return n;
  52. }
  53. SQRESULT sqstd_format(HSQUIRRELVM v,SQInteger nformatstringidx,SQInteger *outlen,SQChar **output)
  54. {
  55. const SQChar *format;
  56. SQChar *dest;
  57. SQChar fmt[MAX_FORMAT_LEN];
  58. sq_getstring(v,nformatstringidx,&format);
  59. SQInteger format_size = sq_getsize(v,nformatstringidx);
  60. SQInteger allocated = (format_size+2)*sizeof(SQChar);
  61. dest = sq_getscratchpad(v,allocated);
  62. SQInteger n = 0,i = 0, nparam = nformatstringidx+1, w = 0;
  63. while(n < format_size) {
  64. if(format[n] != '%') {
  65. assert(i < allocated);
  66. dest[i++] = format[n];
  67. n++;
  68. }
  69. else if(format[n+1] == '%') { //handles %%
  70. dest[i++] = '%';
  71. n += 2;
  72. }
  73. else {
  74. n++;
  75. if( nparam > sq_gettop(v) )
  76. return sq_throwerror(v,_SC("not enough paramters for the given format string"));
  77. n = validate_format(v,fmt,format,n,w);
  78. if(n < 0) return -1;
  79. SQInteger addlen = 0;
  80. SQInteger valtype = 0;
  81. const SQChar *ts;
  82. SQInteger ti;
  83. SQFloat tf;
  84. SQInteger fc = format[n];
  85. switch(fc) {
  86. case 'q':
  87. case 's':
  88. if(SQ_FAILED(sq_getstring(v,nparam,&ts)))
  89. return sq_throwerror(v,_SC("string expected for the specified format"));
  90. if(fc == 'q'){
  91. addlen = 2; //quotes before and after
  92. SQInteger size = sq_getsize(v,nparam);
  93. SQChar *ts2 = (SQChar*)ts;
  94. while (size--) {
  95. ++addlen;
  96. if (*ts2 == '\r' && *(ts2+1) == '\n' ) {
  97. ++addlen;
  98. ++ts2; //eat \r and output only \n
  99. }
  100. else if (*ts2 == '"' || *ts2 == '\\' || *ts2 == '\n') {
  101. ++addlen;
  102. }
  103. else if (*ts2 == '\0' || iscntrl(uchar(*ts2))) {
  104. SQChar buff[10];
  105. addlen += scsprintf(buff, "\\x%x", (int)uchar(*ts2))-1; //-1 because we already added the original char to the sum
  106. }
  107. ++ts2;
  108. }
  109. ts2 = (SQChar*)i; //save the i position using pointer as integer
  110. i += addlen;
  111. allocated += addlen;
  112. dest = sq_getscratchpad(v,allocated);
  113. size = sq_getsize(v,nparam);
  114. ts2 = &dest[(int)ts2]; //use saved i position saved on pointer as integer
  115. *ts2++ = '"';
  116. while (size--) {
  117. if (*ts == '\r' && *(ts+1) == '\n' ) {
  118. ++ts; //eat \r and output only \n
  119. }
  120. if (*ts == '"' || *ts == '\\') {
  121. *ts2++ = '\\';
  122. *ts2++ = *ts;
  123. }
  124. else if (*ts == '\n') {
  125. *ts2++ = '\\';
  126. *ts2++ = 'n';
  127. }
  128. else if (*ts == '\0' || iscntrl(uchar(*ts))) {
  129. SQChar buff[10];
  130. int iw;
  131. iw = scsprintf(buff, "\\x%x", (int)uchar(*ts));
  132. for(int i=0; i< iw; ++i) *ts2++ = buff[i];
  133. }
  134. else
  135. *ts2++ = *ts;
  136. ++ts;
  137. }
  138. *ts2++ = '"';
  139. ++n;
  140. ++nparam;
  141. continue;
  142. }
  143. else
  144. {
  145. addlen = (sq_getsize(v,nparam)*sizeof(SQChar))+((w+1)*sizeof(SQChar));
  146. valtype = 's';
  147. }
  148. break;
  149. case 'i': case 'd': case 'o': case 'u': case 'x': case 'X':
  150. #ifdef _SQ64
  151. {
  152. size_t flen = scstrlen(fmt);
  153. SQInteger fpos = flen - 1;
  154. SQChar f = fmt[fpos];
  155. SQChar *prec = (SQChar *)_PRINT_INT_PREC;
  156. while(*prec != _SC('\0')) {
  157. fmt[fpos++] = *prec++;
  158. }
  159. fmt[fpos++] = f;
  160. fmt[fpos++] = _SC('\0');
  161. }
  162. #endif
  163. case 'c':
  164. if(SQ_FAILED(sq_getinteger(v,nparam,&ti)))
  165. return sq_throwerror(v,_SC("integer expected for the specified format"));
  166. addlen = (ADDITIONAL_FORMAT_SPACE)+((w+1)*sizeof(SQChar));
  167. valtype = 'i';
  168. break;
  169. case 'f': case 'g': case 'G': case 'e': case 'E':
  170. if(SQ_FAILED(sq_getfloat(v,nparam,&tf)))
  171. return sq_throwerror(v,_SC("float expected for the specified format"));
  172. addlen = (ADDITIONAL_FORMAT_SPACE)+((w+1)*sizeof(SQChar));
  173. valtype = 'f';
  174. break;
  175. default:
  176. return sq_throwerror(v,_SC("invalid format"));
  177. }
  178. n++;
  179. allocated += addlen + sizeof(SQChar);
  180. dest = sq_getscratchpad(v,allocated);
  181. switch(valtype) {
  182. case 's': i += scsprintf(&dest[i],fmt,ts); break;
  183. case 'i': i += scsprintf(&dest[i],fmt,ti); break;
  184. case 'f': i += scsprintf(&dest[i],fmt,tf); break;
  185. };
  186. nparam ++;
  187. }
  188. }
  189. *outlen = i;
  190. dest[i] = '\0';
  191. *output = dest;
  192. return SQ_OK;
  193. }
  194. static SQRESULT _string_format(HSQUIRRELVM v)
  195. {
  196. SQChar *dest = NULL;
  197. SQInteger length = 0;
  198. if(SQ_FAILED(sqstd_format(v,2,&length,&dest)))
  199. return -1;
  200. sq_pushstring(v,dest,length);
  201. return 1;
  202. }
  203. static SQRESULT _string_printf(HSQUIRRELVM v)
  204. {
  205. SQChar *dest = NULL;
  206. SQInteger length = 0;
  207. SQPRINTFUNCTION sqprint = sq_getprintfunc(v);
  208. if(sqprint){
  209. if(SQ_FAILED(sqstd_format(v,2,&length,&dest)))
  210. return -1;
  211. sqprint(v,_SC("%s"),dest);
  212. sq_pushinteger(v, length);
  213. return 1;
  214. }
  215. return 0;
  216. }
  217. #define SETUP_REX(v) \
  218. SQRex *self = NULL; \
  219. sq_getinstanceup(v,1,(SQUserPointer *)&self,0);
  220. static SQRESULT _rexobj_releasehook(SQUserPointer p, SQInteger size, HSQUIRRELVM v)
  221. {
  222. SQRex *self = ((SQRex *)p);
  223. sqstd_rex_free(self);
  224. return 1;
  225. }
  226. static SQRESULT _regexp_match(HSQUIRRELVM v)
  227. {
  228. SETUP_REX(v);
  229. const SQChar *str;
  230. sq_getstring(v,2,&str);
  231. if(sqstd_rex_match(self,str) == SQTrue)
  232. {
  233. sq_pushbool(v,SQTrue);
  234. return 1;
  235. }
  236. sq_pushbool(v,SQFalse);
  237. return 1;
  238. }
  239. static SQRESULT _regexp_gmatch(HSQUIRRELVM v)
  240. {
  241. SETUP_REX(v);
  242. const SQChar *str;
  243. SQInteger str_size;
  244. sq_getstring(v,2,&str);
  245. str_size = sq_getsize(v, 2);
  246. const SQChar *begin,*end;
  247. while(sqstd_rex_searchrange(self,str, str+str_size,&begin,&end)){
  248. SQInteger n = sqstd_rex_getsubexpcount(self);
  249. SQRexMatch match;
  250. sq_pushroottable(v); //this
  251. SQInteger i = 0;
  252. for(;i < n; i++) {
  253. sqstd_rex_getsubexp(self,i,&match);
  254. if(i > 0){ //skip whole match
  255. sq_pushstring(v, match.begin, match.len);
  256. }
  257. }
  258. i = sq_call(v, n, SQFalse, SQTrue);
  259. if(i < 0) return i;
  260. str_size -= end-str;
  261. str = end;
  262. }
  263. sq_pushbool(v,SQFalse);
  264. return 1;
  265. }
  266. #include "sqstdblobimpl.h"
  267. static SQRESULT _regexp_gsub(HSQUIRRELVM v)
  268. {
  269. SETUP_REX(v);
  270. const SQChar *str;
  271. SQInteger str_size;
  272. sq_getstring(v,2,&str);
  273. str_size = sq_getsize(v, 2);
  274. const SQChar *begin,*end;
  275. SQBlob blob(0,8192);
  276. SQObjectType ptype = sq_gettype(v, 3);
  277. const SQChar *replacement;
  278. SQInteger replacement_size;
  279. while(sqstd_rex_searchrange(self,str, str+str_size,&begin,&end)){
  280. blob.Write(str, begin-str);
  281. SQInteger n = sqstd_rex_getsubexpcount(self);
  282. SQRexMatch match;
  283. SQInteger i = 0;
  284. switch(ptype){
  285. case OT_CLOSURE:{
  286. sq_pushroottable(v); //this
  287. for(;i < n; i++) {
  288. sqstd_rex_getsubexp(self,i,&match);
  289. if(i > 0){ //skip whole match
  290. sq_pushstring(v, match.begin, match.len);
  291. }
  292. }
  293. i = sq_call(v, n, SQTrue, SQTrue);
  294. if(i < 0) return i;
  295. if(sq_gettype(v, -1) == OT_STRING){
  296. const SQChar *svalue;
  297. sq_getstring(v, -1, &svalue);
  298. blob.Write(svalue, sq_getsize(v, -1));
  299. }
  300. sq_poptop(v);
  301. }
  302. break;
  303. case OT_ARRAY:{
  304. for(;i < n; i++) {
  305. sqstd_rex_getsubexp(self,i,&match);
  306. if(i > 0){ //skip whole match
  307. sq_pushinteger(v, i-1);
  308. if(SQ_SUCCEEDED(sq_get(v, 3)) &&
  309. SQ_SUCCEEDED(sq_getstr_and_size(v, -1, &replacement, &replacement_size))){
  310. blob.Write(replacement, replacement_size);
  311. sq_pop(v, 1); //remove value
  312. }
  313. }
  314. }
  315. }
  316. break;
  317. case OT_TABLE:{
  318. for(;i < n; i++) {
  319. sqstd_rex_getsubexp(self,i,&match);
  320. if(i > 0){ //skip whole match
  321. sq_pushstring(v, match.begin, match.len);
  322. if(SQ_SUCCEEDED(sq_get(v, 3)) &&
  323. SQ_SUCCEEDED(sq_getstr_and_size(v, -1, &replacement, &replacement_size))){
  324. blob.Write(replacement, replacement_size);
  325. sq_pop(v, 1); //remove value
  326. }
  327. }
  328. }
  329. }
  330. break;
  331. default:
  332. return sq_throwerror(v, _SC("gsub only works with closure, array, table for replacement"));
  333. }
  334. str_size -= end-str;
  335. str = end;
  336. }
  337. if(str_size) blob.Write(str, str_size);
  338. sq_pushstring(v, (const SQChar *)blob.GetBuf(), blob.Len());
  339. return 1;
  340. }
  341. static void _addrexmatch(HSQUIRRELVM v,const SQChar *str,const SQChar *begin,const SQChar *end)
  342. {
  343. sq_newtable(v);
  344. sq_pushstring(v,_SC("begin"),-1);
  345. sq_pushinteger(v,begin - str);
  346. sq_rawset(v,-3);
  347. sq_pushstring(v,_SC("end"),-1);
  348. sq_pushinteger(v,end - str);
  349. sq_rawset(v,-3);
  350. }
  351. static SQRESULT _regexp_search(HSQUIRRELVM v)
  352. {
  353. SETUP_REX(v);
  354. const SQChar *str,*begin,*end;
  355. SQInteger start = 0;
  356. sq_getstring(v,2,&str);
  357. if(sq_gettop(v) > 2) sq_getinteger(v,3,&start);
  358. if(sqstd_rex_search(self,str+start,&begin,&end) == SQTrue) {
  359. _addrexmatch(v,str,begin,end);
  360. return 1;
  361. }
  362. return 0;
  363. }
  364. static SQRESULT _regexp_capture(HSQUIRRELVM v)
  365. {
  366. SETUP_REX(v);
  367. const SQChar *str,*begin,*end;
  368. SQInteger start = 0;
  369. sq_getstring(v,2,&str);
  370. if(sq_gettop(v) > 2) sq_getinteger(v,3,&start);
  371. if(sqstd_rex_search(self,str+start,&begin,&end) == SQTrue) {
  372. SQInteger n = sqstd_rex_getsubexpcount(self);
  373. SQRexMatch match;
  374. sq_newarray(v,0);
  375. for(SQInteger i = 0;i < n; i++) {
  376. sqstd_rex_getsubexp(self,i,&match);
  377. if(match.len > 0)
  378. _addrexmatch(v,str,match.begin,match.begin+match.len);
  379. else
  380. _addrexmatch(v,str,str,str); //empty match
  381. sq_arrayappend(v,-2);
  382. }
  383. return 1;
  384. }
  385. return 0;
  386. }
  387. static SQRESULT _regexp_xcapture(HSQUIRRELVM v)
  388. {
  389. SETUP_REX(v);
  390. const SQChar *str,*begin,*end;
  391. SQInteger start = 0;
  392. sq_getstring(v,2,&str);
  393. if(sq_gettop(v) > 2) sq_getinteger(v,3,&start);
  394. if(sqstd_rex_search(self,str+start,&begin,&end) == SQTrue) {
  395. sq_pushbool(v, SQTrue);
  396. return 1;
  397. }
  398. return 0;
  399. }
  400. static SQRESULT _regexp_getxcapture(HSQUIRRELVM v)
  401. {
  402. SETUP_REX(v);
  403. SQInteger n, start;
  404. const SQChar *str;
  405. sq_getstring(v,2,&str);
  406. sq_getinteger(v,3,&n);
  407. SQRexMatch match;
  408. sqstd_rex_getsubexp(self,n,&match);
  409. if(match.len > 0){
  410. start = match.begin-str;
  411. sq_pushinteger(v, start);
  412. sq_arrayappend(v,-2);
  413. sq_pushinteger(v, start+match.len);
  414. sq_arrayappend(v,-2);
  415. sq_pushbool(v, SQTrue);
  416. return 1;
  417. }
  418. return 0;
  419. }
  420. static SQRESULT _regexp_subexpcount(HSQUIRRELVM v)
  421. {
  422. SETUP_REX(v);
  423. sq_pushinteger(v,sqstd_rex_getsubexpcount(self));
  424. return 1;
  425. }
  426. static SQRESULT _regexp_constructor(HSQUIRRELVM v)
  427. {
  428. const SQChar *error,*pattern;
  429. sq_getstring(v,2,&pattern);
  430. SQRex *rex = sqstd_rex_compile(pattern,&error);
  431. if(!rex) return sq_throwerror(v,error);
  432. sq_setinstanceup(v,1,rex);
  433. sq_setreleasehook(v,1,_rexobj_releasehook);
  434. return 0;
  435. }
  436. static SQRESULT _regexp__typeof(HSQUIRRELVM v)
  437. {
  438. sq_pushstring(v,_SC("regexp"),-1);
  439. return 1;
  440. }
  441. #define _DECL_REX_FUNC(name,nparams,pmask) {_SC(#name),_regexp_##name,nparams,pmask}
  442. static SQRegFunction rexobj_funcs[]={
  443. _DECL_REX_FUNC(constructor,2,_SC(".s")),
  444. _DECL_REX_FUNC(search,-2,_SC("xsn")),
  445. _DECL_REX_FUNC(match,2,_SC("xs")),
  446. _DECL_REX_FUNC(gmatch,3,_SC("xsc")),
  447. _DECL_REX_FUNC(gsub,3,_SC("xs c|a|t")),
  448. _DECL_REX_FUNC(capture,-2,_SC("xsn")),
  449. _DECL_REX_FUNC(xcapture,-2,_SC("xsn")),
  450. _DECL_REX_FUNC(getxcapture,4,_SC("xsna")),
  451. _DECL_REX_FUNC(subexpcount,1,_SC("x")),
  452. _DECL_REX_FUNC(_typeof,1,_SC("x")),
  453. {0,0}
  454. };
  455. #undef _DECL_REX_FUNC
  456. #define _DECL_FUNC(name,nparams,pmask) {_SC(#name),_string_##name,nparams,pmask}
  457. static SQRegFunction stringlib_funcs[]={
  458. _DECL_FUNC(printf,-2,_SC(".s")),
  459. _DECL_FUNC(format,-2,_SC(".s")),
  460. {0,0}
  461. };
  462. #undef _DECL_FUNC
  463. SQInteger sqstd_register_stringlib(HSQUIRRELVM v)
  464. {
  465. sq_pushstring(v,_SC("regexp"),-1);
  466. sq_newclass(v,SQFalse);
  467. SQInteger i = 0;
  468. while(rexobj_funcs[i].name != 0) {
  469. SQRegFunction &f = rexobj_funcs[i];
  470. sq_pushstring(v,f.name,-1);
  471. sq_newclosure(v,f.f,0);
  472. sq_setparamscheck(v,f.nparamscheck,f.typemask);
  473. sq_setnativeclosurename(v,-1,f.name);
  474. sq_newslot(v,-3,SQFalse);
  475. i++;
  476. }
  477. sq_newslot(v,-3,SQFalse);
  478. i = 0;
  479. while(stringlib_funcs[i].name!=0)
  480. {
  481. sq_pushstring(v,stringlib_funcs[i].name,-1);
  482. sq_newclosure(v,stringlib_funcs[i].f,0);
  483. sq_setparamscheck(v,stringlib_funcs[i].nparamscheck,stringlib_funcs[i].typemask);
  484. sq_setnativeclosurename(v,-1,stringlib_funcs[i].name);
  485. sq_newslot(v,-3,SQFalse);
  486. i++;
  487. }
  488. return 1;
  489. }