| 1 | /* see copyright notice in squirrel.h */ |
| 2 | #include <squirrel.h> |
| 3 | #include <string.h> |
| 4 | #include <ctype.h> |
| 5 | #include <setjmp.h> |
| 6 | #include <sqstdstring.h> |
| 7 | |
| 8 | #ifdef _DEBUG |
| 9 | #include <stdio.h> |
| 10 | |
| 11 | static const SQChar *g_nnames[] = |
| 12 | { |
| 13 | _SC("NONE" ),_SC("OP_GREEDY" ), _SC("OP_OR" ), |
| 14 | _SC("OP_EXPR" ),_SC("OP_NOCAPEXPR" ),_SC("OP_DOT" ), _SC("OP_CLASS" ), |
| 15 | _SC("OP_CCLASS" ),_SC("OP_NCLASS" ),_SC("OP_RANGE" ),_SC("OP_CHAR" ), |
| 16 | _SC("OP_EOL" ),_SC("OP_BOL" ),_SC("OP_WB" ),_SC("OP_MB" ) |
| 17 | }; |
| 18 | |
| 19 | #endif |
| 20 | |
| 21 | #define OP_GREEDY (MAX_CHAR+1) // * + ? {n} |
| 22 | #define OP_OR (MAX_CHAR+2) |
| 23 | #define OP_EXPR (MAX_CHAR+3) //parentesis () |
| 24 | #define OP_NOCAPEXPR (MAX_CHAR+4) //parentesis (?:) |
| 25 | #define OP_DOT (MAX_CHAR+5) |
| 26 | #define OP_CLASS (MAX_CHAR+6) |
| 27 | #define OP_CCLASS (MAX_CHAR+7) |
| 28 | #define OP_NCLASS (MAX_CHAR+8) //negates class the [^ |
| 29 | #define OP_RANGE (MAX_CHAR+9) |
| 30 | #define OP_CHAR (MAX_CHAR+10) |
| 31 | #define OP_EOL (MAX_CHAR+11) |
| 32 | #define OP_BOL (MAX_CHAR+12) |
| 33 | #define OP_WB (MAX_CHAR+13) |
| 34 | #define OP_MB (MAX_CHAR+14) //match balanced |
| 35 | |
| 36 | #define SQREX_SYMBOL_ANY_CHAR ('.') |
| 37 | #define SQREX_SYMBOL_GREEDY_ONE_OR_MORE ('+') |
| 38 | #define SQREX_SYMBOL_GREEDY_ZERO_OR_MORE ('*') |
| 39 | #define SQREX_SYMBOL_GREEDY_ZERO_OR_ONE ('?') |
| 40 | #define SQREX_SYMBOL_BRANCH ('|') |
| 41 | #define SQREX_SYMBOL_END_OF_STRING ('$') |
| 42 | #define SQREX_SYMBOL_BEGINNING_OF_STRING ('^') |
| 43 | #define SQREX_SYMBOL_ESCAPE_CHAR ('\\') |
| 44 | |
| 45 | |
| 46 | typedef int SQRexNodeType; |
| 47 | |
| 48 | typedef struct tagSQRexNode{ |
| 49 | SQRexNodeType type; |
| 50 | SQInteger left; |
| 51 | SQInteger right; |
| 52 | SQInteger next; |
| 53 | }SQRexNode; |
| 54 | |
| 55 | struct SQRex{ |
| 56 | const SQChar *_eol; |
| 57 | const SQChar *_bol; |
| 58 | const SQChar *_p; |
| 59 | SQInteger _first; |
| 60 | SQInteger _op; |
| 61 | SQRexNode *_nodes; |
| 62 | SQInteger _nallocated; |
| 63 | SQInteger _nsize; |
| 64 | SQInteger _nsubexpr; |
| 65 | SQRexMatch *_matches; |
| 66 | SQInteger _currsubexp; |
| 67 | void *_jmpbuf; |
| 68 | const SQChar **_error; |
| 69 | }; |
| 70 | |
| 71 | static SQInteger sqstd_rex_list(SQRex *exp); |
| 72 | |
| 73 | static SQInteger sqstd_rex_newnode(SQRex *exp, SQRexNodeType type) |
| 74 | { |
| 75 | SQRexNode n; |
| 76 | n.type = type; |
| 77 | n.next = n.right = n.left = -1; |
| 78 | if(type == OP_EXPR) |
| 79 | n.right = exp->_nsubexpr++; |
| 80 | if(exp->_nallocated < (exp->_nsize + 1)) { |
| 81 | SQInteger oldsize = exp->_nallocated; |
| 82 | exp->_nallocated *= 2; |
| 83 | exp->_nodes = (SQRexNode *)sq_realloc(exp->_nodes, oldsize * sizeof(SQRexNode) ,exp->_nallocated * sizeof(SQRexNode)); |
| 84 | } |
| 85 | exp->_nodes[exp->_nsize++] = n; |
| 86 | SQInteger newid = exp->_nsize - 1; |
| 87 | return (SQInteger)newid; |
| 88 | } |
| 89 | |
| 90 | static void sqstd_rex_error(SQRex *exp,const SQChar *error) |
| 91 | { |
| 92 | if(exp->_error) *exp->_error = error; |
| 93 | longjmp(*((jmp_buf*)exp->_jmpbuf),-1); |
| 94 | } |
| 95 | |
| 96 | static void sqstd_rex_expect(SQRex *exp, SQInteger n){ |
| 97 | if((*exp->_p) != n) |
| 98 | sqstd_rex_error(exp, _SC("expected paren" )); |
| 99 | exp->_p++; |
| 100 | } |
| 101 | |
| 102 | static SQChar sqstd_rex_escapechar(SQRex *exp) |
| 103 | { |
| 104 | if(*exp->_p == SQREX_SYMBOL_ESCAPE_CHAR){ |
| 105 | exp->_p++; |
| 106 | switch(*exp->_p) { |
| 107 | case 'v': exp->_p++; return '\v'; |
| 108 | case 'n': exp->_p++; return '\n'; |
| 109 | case 't': exp->_p++; return '\t'; |
| 110 | case 'r': exp->_p++; return '\r'; |
| 111 | case 'f': exp->_p++; return '\f'; |
| 112 | default: return (*exp->_p++); |
| 113 | } |
| 114 | } else if(!scisprint(*exp->_p)) sqstd_rex_error(exp,_SC("letter expected" )); |
| 115 | return (*exp->_p++); |
| 116 | } |
| 117 | |
| 118 | static SQInteger sqstd_rex_charclass(SQRex *exp,SQInteger classid) |
| 119 | { |
| 120 | SQInteger n = sqstd_rex_newnode(exp,OP_CCLASS); |
| 121 | exp->_nodes[n].left = classid; |
| 122 | return n; |
| 123 | } |
| 124 | |
| 125 | static SQInteger sqstd_rex_charnode(SQRex *exp,SQBool isclass) |
| 126 | { |
| 127 | SQChar t; |
| 128 | if(*exp->_p == SQREX_SYMBOL_ESCAPE_CHAR) { |
| 129 | exp->_p++; |
| 130 | switch(*exp->_p) { |
| 131 | case 'n': exp->_p++; return sqstd_rex_newnode(exp,'\n'); |
| 132 | case 't': exp->_p++; return sqstd_rex_newnode(exp,'\t'); |
| 133 | case 'r': exp->_p++; return sqstd_rex_newnode(exp,'\r'); |
| 134 | case 'f': exp->_p++; return sqstd_rex_newnode(exp,'\f'); |
| 135 | case 'v': exp->_p++; return sqstd_rex_newnode(exp,'\v'); |
| 136 | case 'a': case 'A': case 'w': case 'W': case 's': case 'S': |
| 137 | case 'd': case 'D': case 'x': case 'X': case 'c': case 'C': |
| 138 | case 'p': case 'P': case 'l': case 'u': |
| 139 | { |
| 140 | t = *exp->_p; exp->_p++; |
| 141 | return sqstd_rex_charclass(exp,t); |
| 142 | } |
| 143 | case 'm': |
| 144 | { |
| 145 | SQChar cb, ce; //cb = character begin match ce = character end match |
| 146 | cb = *++exp->_p; //skip 'm' |
| 147 | ce = *++exp->_p; |
| 148 | exp->_p++; //points to the next char to be parsed |
| 149 | if ((!cb) || (!ce)) sqstd_rex_error(exp,_SC("balanced chars expected" )); |
| 150 | if ( cb == ce ) sqstd_rex_error(exp,_SC("open/close char can't be the same" )); |
| 151 | SQInteger node = sqstd_rex_newnode(exp,OP_MB); |
| 152 | exp->_nodes[node].left = cb; |
| 153 | exp->_nodes[node].right = ce; |
| 154 | return node; |
| 155 | } |
| 156 | case 0: |
| 157 | sqstd_rex_error(exp,_SC("letter expected for argument of escape sequence" )); |
| 158 | break; |
| 159 | case 'b': |
| 160 | case 'B': |
| 161 | if(!isclass) { |
| 162 | SQInteger node = sqstd_rex_newnode(exp,OP_WB); |
| 163 | exp->_nodes[node].left = *exp->_p; |
| 164 | exp->_p++; |
| 165 | return node; |
| 166 | } //else default |
| 167 | default: |
| 168 | t = *exp->_p; exp->_p++; |
| 169 | return sqstd_rex_newnode(exp,t); |
| 170 | } |
| 171 | } |
| 172 | else if(!scisprint(*exp->_p)) { |
| 173 | |
| 174 | sqstd_rex_error(exp,_SC("letter expected" )); |
| 175 | } |
| 176 | t = *exp->_p; exp->_p++; |
| 177 | return sqstd_rex_newnode(exp,t); |
| 178 | } |
| 179 | static SQInteger sqstd_rex_class(SQRex *exp) |
| 180 | { |
| 181 | SQInteger ret = -1; |
| 182 | SQInteger first = -1,chain; |
| 183 | if(*exp->_p == SQREX_SYMBOL_BEGINNING_OF_STRING){ |
| 184 | ret = sqstd_rex_newnode(exp,OP_NCLASS); |
| 185 | exp->_p++; |
| 186 | }else ret = sqstd_rex_newnode(exp,OP_CLASS); |
| 187 | |
| 188 | if(*exp->_p == ']') sqstd_rex_error(exp,_SC("empty class" )); |
| 189 | chain = ret; |
| 190 | while(*exp->_p != ']' && exp->_p != exp->_eol) { |
| 191 | if(*exp->_p == '-' && first != -1){ |
| 192 | SQInteger r; |
| 193 | if(*exp->_p++ == ']') sqstd_rex_error(exp,_SC("unfinished range" )); |
| 194 | r = sqstd_rex_newnode(exp,OP_RANGE); |
| 195 | if(exp->_nodes[first].type>*exp->_p) sqstd_rex_error(exp,_SC("invalid range" )); |
| 196 | if(exp->_nodes[first].type == OP_CCLASS) sqstd_rex_error(exp,_SC("cannot use character classes in ranges" )); |
| 197 | exp->_nodes[r].left = exp->_nodes[first].type; |
| 198 | SQInteger t = sqstd_rex_escapechar(exp); |
| 199 | exp->_nodes[r].right = t; |
| 200 | exp->_nodes[chain].next = r; |
| 201 | chain = r; |
| 202 | first = -1; |
| 203 | } |
| 204 | else{ |
| 205 | if(first!=-1){ |
| 206 | SQInteger c = first; |
| 207 | exp->_nodes[chain].next = c; |
| 208 | chain = c; |
| 209 | first = sqstd_rex_charnode(exp,SQTrue); |
| 210 | } |
| 211 | else{ |
| 212 | first = sqstd_rex_charnode(exp,SQTrue); |
| 213 | } |
| 214 | } |
| 215 | } |
| 216 | if(first!=-1){ |
| 217 | SQInteger c = first; |
| 218 | exp->_nodes[chain].next = c; |
| 219 | } |
| 220 | /* hack? */ |
| 221 | exp->_nodes[ret].left = exp->_nodes[ret].next; |
| 222 | exp->_nodes[ret].next = -1; |
| 223 | return ret; |
| 224 | } |
| 225 | |
| 226 | static SQInteger sqstd_rex_parsenumber(SQRex *exp) |
| 227 | { |
| 228 | SQInteger ret = *exp->_p-'0'; |
| 229 | SQInteger positions = 10; |
| 230 | exp->_p++; |
| 231 | while(isdigit(*exp->_p)) { |
| 232 | ret = ret*10+(*exp->_p++-'0'); |
| 233 | if(positions==1000000000) sqstd_rex_error(exp,_SC("overflow in numeric constant" )); |
| 234 | positions *= 10; |
| 235 | }; |
| 236 | return ret; |
| 237 | } |
| 238 | |
| 239 | static SQInteger sqstd_rex_element(SQRex *exp) |
| 240 | { |
| 241 | SQInteger ret = -1; |
| 242 | switch(*exp->_p) |
| 243 | { |
| 244 | case '(': { |
| 245 | SQInteger expr; |
| 246 | exp->_p++; |
| 247 | |
| 248 | |
| 249 | if(*exp->_p =='?') { |
| 250 | exp->_p++; |
| 251 | sqstd_rex_expect(exp,':'); |
| 252 | expr = sqstd_rex_newnode(exp,OP_NOCAPEXPR); |
| 253 | } |
| 254 | else |
| 255 | expr = sqstd_rex_newnode(exp,OP_EXPR); |
| 256 | SQInteger newn = sqstd_rex_list(exp); |
| 257 | exp->_nodes[expr].left = newn; |
| 258 | ret = expr; |
| 259 | sqstd_rex_expect(exp,')'); |
| 260 | } |
| 261 | break; |
| 262 | case '[': |
| 263 | exp->_p++; |
| 264 | ret = sqstd_rex_class(exp); |
| 265 | sqstd_rex_expect(exp,']'); |
| 266 | break; |
| 267 | case SQREX_SYMBOL_END_OF_STRING: exp->_p++; ret = sqstd_rex_newnode(exp,OP_EOL);break; |
| 268 | case SQREX_SYMBOL_ANY_CHAR: exp->_p++; ret = sqstd_rex_newnode(exp,OP_DOT);break; |
| 269 | default: |
| 270 | ret = sqstd_rex_charnode(exp,SQFalse); |
| 271 | break; |
| 272 | } |
| 273 | |
| 274 | |
| 275 | SQBool isgreedy = SQFalse; |
| 276 | unsigned short p0 = 0, p1 = 0; |
| 277 | switch(*exp->_p){ |
| 278 | case SQREX_SYMBOL_GREEDY_ZERO_OR_MORE: p0 = 0; p1 = 0xFFFF; exp->_p++; isgreedy = SQTrue; break; |
| 279 | case SQREX_SYMBOL_GREEDY_ONE_OR_MORE: p0 = 1; p1 = 0xFFFF; exp->_p++; isgreedy = SQTrue; break; |
| 280 | case SQREX_SYMBOL_GREEDY_ZERO_OR_ONE: p0 = 0; p1 = 1; exp->_p++; isgreedy = SQTrue; break; |
| 281 | case '{': |
| 282 | exp->_p++; |
| 283 | if(!isdigit(*exp->_p)) sqstd_rex_error(exp,_SC("number expected" )); |
| 284 | p0 = (unsigned short)sqstd_rex_parsenumber(exp); |
| 285 | /*******************************/ |
| 286 | switch(*exp->_p) { |
| 287 | case '}': |
| 288 | p1 = p0; exp->_p++; |
| 289 | break; |
| 290 | case ',': |
| 291 | exp->_p++; |
| 292 | p1 = 0xFFFF; |
| 293 | if(isdigit(*exp->_p)){ |
| 294 | p1 = (unsigned short)sqstd_rex_parsenumber(exp); |
| 295 | } |
| 296 | sqstd_rex_expect(exp,'}'); |
| 297 | break; |
| 298 | default: |
| 299 | sqstd_rex_error(exp,_SC(", or } expected" )); |
| 300 | } |
| 301 | /*******************************/ |
| 302 | isgreedy = SQTrue; |
| 303 | break; |
| 304 | |
| 305 | } |
| 306 | if(isgreedy) { |
| 307 | SQInteger nnode = sqstd_rex_newnode(exp,OP_GREEDY); |
| 308 | exp->_nodes[nnode].left = ret; |
| 309 | exp->_nodes[nnode].right = ((p0)<<16)|p1; |
| 310 | ret = nnode; |
| 311 | } |
| 312 | |
| 313 | if((*exp->_p != SQREX_SYMBOL_BRANCH) && (*exp->_p != ')') && (*exp->_p != SQREX_SYMBOL_GREEDY_ZERO_OR_MORE) && (*exp->_p != SQREX_SYMBOL_GREEDY_ONE_OR_MORE) && (*exp->_p != '\0')) { |
| 314 | SQInteger nnode = sqstd_rex_element(exp); |
| 315 | exp->_nodes[ret].next = nnode; |
| 316 | } |
| 317 | |
| 318 | return ret; |
| 319 | } |
| 320 | |
| 321 | static SQInteger sqstd_rex_list(SQRex *exp) |
| 322 | { |
| 323 | SQInteger ret=-1,e; |
| 324 | if(*exp->_p == SQREX_SYMBOL_BEGINNING_OF_STRING) { |
| 325 | exp->_p++; |
| 326 | ret = sqstd_rex_newnode(exp,OP_BOL); |
| 327 | } |
| 328 | e = sqstd_rex_element(exp); |
| 329 | if(ret != -1) { |
| 330 | exp->_nodes[ret].next = e; |
| 331 | } |
| 332 | else ret = e; |
| 333 | |
| 334 | if(*exp->_p == SQREX_SYMBOL_BRANCH) { |
| 335 | SQInteger temp,tright; |
| 336 | exp->_p++; |
| 337 | temp = sqstd_rex_newnode(exp,OP_OR); |
| 338 | exp->_nodes[temp].left = ret; |
| 339 | tright = sqstd_rex_list(exp); |
| 340 | exp->_nodes[temp].right = tright; |
| 341 | ret = temp; |
| 342 | } |
| 343 | return ret; |
| 344 | } |
| 345 | |
| 346 | static SQBool sqstd_rex_matchcclass(SQInteger cclass,SQChar c) |
| 347 | { |
| 348 | switch(cclass) { |
| 349 | case 'a': return isalpha(c)?SQTrue:SQFalse; |
| 350 | case 'A': return !isalpha(c)?SQTrue:SQFalse; |
| 351 | case 'w': return (isalnum(c) || c == '_')?SQTrue:SQFalse; |
| 352 | case 'W': return (!isalnum(c) && c != '_')?SQTrue:SQFalse; |
| 353 | case 's': return isspace(c)?SQTrue:SQFalse; |
| 354 | case 'S': return !isspace(c)?SQTrue:SQFalse; |
| 355 | case 'd': return isdigit(c)?SQTrue:SQFalse; |
| 356 | case 'D': return !isdigit(c)?SQTrue:SQFalse; |
| 357 | case 'x': return isxdigit(c)?SQTrue:SQFalse; |
| 358 | case 'X': return !isxdigit(c)?SQTrue:SQFalse; |
| 359 | case 'c': return iscntrl(c)?SQTrue:SQFalse; |
| 360 | case 'C': return !iscntrl(c)?SQTrue:SQFalse; |
| 361 | case 'p': return ispunct(c)?SQTrue:SQFalse; |
| 362 | case 'P': return !ispunct(c)?SQTrue:SQFalse; |
| 363 | case 'l': return islower(c)?SQTrue:SQFalse; |
| 364 | case 'u': return isupper(c)?SQTrue:SQFalse; |
| 365 | } |
| 366 | return SQFalse; /*cannot happen*/ |
| 367 | } |
| 368 | |
| 369 | static SQBool sqstd_rex_matchclass(SQRex* exp,SQRexNode *node,SQChar c) |
| 370 | { |
| 371 | do { |
| 372 | switch(node->type) { |
| 373 | case OP_RANGE: |
| 374 | if(c >= node->left && c <= node->right) return SQTrue; |
| 375 | break; |
| 376 | case OP_CCLASS: |
| 377 | if(sqstd_rex_matchcclass(node->left,c)) return SQTrue; |
| 378 | break; |
| 379 | default: |
| 380 | if(c == node->type)return SQTrue; |
| 381 | } |
| 382 | } while((node->next != -1) && (node = &exp->_nodes[node->next])); |
| 383 | return SQFalse; |
| 384 | } |
| 385 | |
| 386 | static const SQChar *sqstd_rex_matchnode(SQRex* exp,SQRexNode *node,const SQChar *str,SQRexNode *next) |
| 387 | { |
| 388 | |
| 389 | SQRexNodeType type = node->type; |
| 390 | switch(type) { |
| 391 | case OP_GREEDY: { |
| 392 | //SQRexNode *greedystop = (node->next != -1) ? &exp->_nodes[node->next] : NULL; |
| 393 | SQRexNode *greedystop = NULL; |
| 394 | SQInteger p0 = (node->right >> 16)&0x0000FFFF, p1 = node->right&0x0000FFFF, nmaches = 0; |
| 395 | const SQChar *s=str, *good = str; |
| 396 | |
| 397 | if(node->next != -1) { |
| 398 | greedystop = &exp->_nodes[node->next]; |
| 399 | } |
| 400 | else { |
| 401 | greedystop = next; |
| 402 | } |
| 403 | |
| 404 | while((nmaches == 0xFFFF || nmaches < p1)) { |
| 405 | |
| 406 | const SQChar *stop; |
| 407 | if(!(s = sqstd_rex_matchnode(exp,&exp->_nodes[node->left],s,greedystop))) |
| 408 | break; |
| 409 | nmaches++; |
| 410 | good=s; |
| 411 | if(greedystop) { |
| 412 | //checks that 0 matches satisfy the expression(if so skips) |
| 413 | //if not would always stop(for instance if is a '?') |
| 414 | if(greedystop->type != OP_GREEDY || |
| 415 | (greedystop->type == OP_GREEDY && ((greedystop->right >> 16)&0x0000FFFF) != 0)) |
| 416 | { |
| 417 | SQRexNode *gnext = NULL; |
| 418 | if(greedystop->next != -1) { |
| 419 | gnext = &exp->_nodes[greedystop->next]; |
| 420 | }else if(next && next->next != -1){ |
| 421 | gnext = &exp->_nodes[next->next]; |
| 422 | } |
| 423 | stop = sqstd_rex_matchnode(exp,greedystop,s,gnext); |
| 424 | if(stop) { |
| 425 | //if satisfied stop it |
| 426 | if(p0 == p1 && p0 == nmaches) break; |
| 427 | else if(nmaches >= p0 && p1 == 0xFFFF) break; |
| 428 | else if(nmaches >= p0 && nmaches <= p1) break; |
| 429 | } |
| 430 | } |
| 431 | } |
| 432 | |
| 433 | if(s >= exp->_eol) |
| 434 | break; |
| 435 | } |
| 436 | if(p0 == p1 && p0 == nmaches) return good; |
| 437 | else if(nmaches >= p0 && p1 == 0xFFFF) return good; |
| 438 | else if(nmaches >= p0 && nmaches <= p1) return good; |
| 439 | return NULL; |
| 440 | } |
| 441 | case OP_OR: { |
| 442 | const SQChar *asd = str; |
| 443 | SQRexNode *temp=&exp->_nodes[node->left]; |
| 444 | while( (asd = sqstd_rex_matchnode(exp,temp,asd,NULL)) ) { |
| 445 | if(temp->next != -1) |
| 446 | temp = &exp->_nodes[temp->next]; |
| 447 | else |
| 448 | return asd; |
| 449 | } |
| 450 | asd = str; |
| 451 | temp = &exp->_nodes[node->right]; |
| 452 | while( (asd = sqstd_rex_matchnode(exp,temp,asd,NULL)) ) { |
| 453 | if(temp->next != -1) |
| 454 | temp = &exp->_nodes[temp->next]; |
| 455 | else |
| 456 | return asd; |
| 457 | } |
| 458 | return NULL; |
| 459 | break; |
| 460 | } |
| 461 | case OP_EXPR: |
| 462 | case OP_NOCAPEXPR:{ |
| 463 | SQRexNode *n = &exp->_nodes[node->left]; |
| 464 | const SQChar *cur = str; |
| 465 | SQInteger capture = -1; |
| 466 | if(node->type != OP_NOCAPEXPR && node->right == exp->_currsubexp) { |
| 467 | capture = exp->_currsubexp; |
| 468 | exp->_matches[capture].begin = cur; |
| 469 | exp->_currsubexp++; |
| 470 | } |
| 471 | SQInteger tempcap = exp->_currsubexp; |
| 472 | do { |
| 473 | SQRexNode *subnext = NULL; |
| 474 | if(n->next != -1) { |
| 475 | subnext = &exp->_nodes[n->next]; |
| 476 | }else { |
| 477 | subnext = next; |
| 478 | } |
| 479 | if(!(cur = sqstd_rex_matchnode(exp,n,cur,subnext))) { |
| 480 | if(capture != -1){ |
| 481 | exp->_matches[capture].begin = 0; |
| 482 | exp->_matches[capture].len = 0; |
| 483 | } |
| 484 | return NULL; |
| 485 | } |
| 486 | } while((n->next != -1) && (n = &exp->_nodes[n->next])); |
| 487 | |
| 488 | exp->_currsubexp = tempcap; |
| 489 | if(capture != -1) |
| 490 | exp->_matches[capture].len = cur - exp->_matches[capture].begin; |
| 491 | return cur; |
| 492 | } |
| 493 | case OP_WB: |
| 494 | if((str == exp->_bol && !isspace(*str)) |
| 495 | || (str == exp->_eol && !isspace(*(str-1))) |
| 496 | || (!isspace(*str) && isspace(*(str+1))) |
| 497 | || (isspace(*str) && !isspace(*(str+1))) ) { |
| 498 | return (node->left == 'b')?str:NULL; |
| 499 | } |
| 500 | return (node->left == 'b')?NULL:str; |
| 501 | case OP_BOL: |
| 502 | if(str == exp->_bol) return str; |
| 503 | return NULL; |
| 504 | case OP_EOL: |
| 505 | if(str == exp->_eol) return str; |
| 506 | return NULL; |
| 507 | case OP_DOT:{ |
| 508 | if (str == exp->_eol) return NULL; |
| 509 | str++; |
| 510 | } |
| 511 | return str; |
| 512 | case OP_NCLASS: |
| 513 | case OP_CLASS: |
| 514 | if (str == exp->_eol) return NULL; |
| 515 | if(sqstd_rex_matchclass(exp,&exp->_nodes[node->left],*str)?(type == OP_CLASS?SQTrue:SQFalse):(type == OP_NCLASS?SQTrue:SQFalse)) { |
| 516 | str++; |
| 517 | return str; |
| 518 | } |
| 519 | return NULL; |
| 520 | case OP_CCLASS: |
| 521 | if (str == exp->_eol) return NULL; |
| 522 | if(sqstd_rex_matchcclass(node->left,*str)) { |
| 523 | str++; |
| 524 | return str; |
| 525 | } |
| 526 | return NULL; |
| 527 | case OP_MB: |
| 528 | { |
| 529 | SQInteger cb = node->left; //char that opens a balanced expression |
| 530 | if(*str != cb) return NULL; // string doesnt start with open char |
| 531 | SQInteger ce = node->right; //char that closes a balanced expression |
| 532 | SQInteger cont = 1; |
| 533 | const SQChar *streol = exp->_eol; |
| 534 | while (++str < streol) { |
| 535 | if (*str == ce) { |
| 536 | if (--cont == 0) { |
| 537 | return ++str; |
| 538 | } |
| 539 | } |
| 540 | else if (*str == cb) cont++; |
| 541 | } |
| 542 | } |
| 543 | return NULL; // string ends out of balance |
| 544 | default: /* char */ |
| 545 | if (str == exp->_eol) return NULL; |
| 546 | if(*str != node->type) return NULL; |
| 547 | str++; |
| 548 | return str; |
| 549 | } |
| 550 | return NULL; |
| 551 | } |
| 552 | |
| 553 | /* public api */ |
| 554 | SQRex *sqstd_rex_compile(const SQChar *pattern,const SQChar **error) |
| 555 | { |
| 556 | SQRex * volatile exp = (SQRex *)sq_malloc(sizeof(SQRex)); // "volatile" is needed for setjmp() |
| 557 | exp->_eol = exp->_bol = NULL; |
| 558 | exp->_p = pattern; |
| 559 | exp->_nallocated = (SQInteger)scstrlen(pattern) * sizeof(SQChar); |
| 560 | exp->_nodes = (SQRexNode *)sq_malloc(exp->_nallocated * sizeof(SQRexNode)); |
| 561 | exp->_nsize = 0; |
| 562 | exp->_matches = 0; |
| 563 | exp->_nsubexpr = 0; |
| 564 | exp->_first = sqstd_rex_newnode(exp,OP_EXPR); |
| 565 | exp->_error = error; |
| 566 | exp->_jmpbuf = sq_malloc(sizeof(jmp_buf)); |
| 567 | if(setjmp(*((jmp_buf*)exp->_jmpbuf)) == 0) { |
| 568 | SQInteger res = sqstd_rex_list(exp); |
| 569 | exp->_nodes[exp->_first].left = res; |
| 570 | if(*exp->_p!='\0') |
| 571 | sqstd_rex_error(exp,_SC("unexpected character" )); |
| 572 | #ifdef _DEBUG |
| 573 | { |
| 574 | SQInteger nsize,i; |
| 575 | SQRexNode *t; |
| 576 | nsize = exp->_nsize; |
| 577 | t = &exp->_nodes[0]; |
| 578 | scprintf(_SC("\n" )); |
| 579 | for(i = 0;i < nsize; i++) { |
| 580 | if(exp->_nodes[i].type>MAX_CHAR) |
| 581 | scprintf(_SC("[%02d] %10s " ), (SQInt32)i,g_nnames[exp->_nodes[i].type-MAX_CHAR]); |
| 582 | else |
| 583 | scprintf(_SC("[%02d] %10c " ), (SQInt32)i,exp->_nodes[i].type); |
| 584 | scprintf(_SC("left %02d right %02d next %02d\n" ), (SQInt32)exp->_nodes[i].left, (SQInt32)exp->_nodes[i].right, (SQInt32)exp->_nodes[i].next); |
| 585 | } |
| 586 | scprintf(_SC("\n" )); |
| 587 | } |
| 588 | #endif |
| 589 | exp->_matches = (SQRexMatch *) sq_malloc(exp->_nsubexpr * sizeof(SQRexMatch)); |
| 590 | memset(exp->_matches,0,exp->_nsubexpr * sizeof(SQRexMatch)); |
| 591 | } |
| 592 | else{ |
| 593 | sqstd_rex_free(exp); |
| 594 | return NULL; |
| 595 | } |
| 596 | return exp; |
| 597 | } |
| 598 | |
| 599 | void sqstd_rex_free(SQRex *exp) |
| 600 | { |
| 601 | if(exp) { |
| 602 | if(exp->_nodes) sq_free(exp->_nodes,exp->_nallocated * sizeof(SQRexNode)); |
| 603 | if(exp->_jmpbuf) sq_free(exp->_jmpbuf,sizeof(jmp_buf)); |
| 604 | if(exp->_matches) sq_free(exp->_matches,exp->_nsubexpr * sizeof(SQRexMatch)); |
| 605 | sq_free(exp,sizeof(SQRex)); |
| 606 | } |
| 607 | } |
| 608 | |
| 609 | SQBool sqstd_rex_match(SQRex* exp,const SQChar* text) |
| 610 | { |
| 611 | const SQChar* res = NULL; |
| 612 | exp->_bol = text; |
| 613 | exp->_eol = text + scstrlen(text); |
| 614 | exp->_currsubexp = 0; |
| 615 | res = sqstd_rex_matchnode(exp,exp->_nodes,text,NULL); |
| 616 | if(res == NULL || res != exp->_eol) |
| 617 | return SQFalse; |
| 618 | return SQTrue; |
| 619 | } |
| 620 | |
| 621 | SQBool sqstd_rex_searchrange(SQRex* exp,const SQChar* text_begin,const SQChar* text_end,const SQChar** out_begin, const SQChar** out_end) |
| 622 | { |
| 623 | const SQChar *cur = NULL; |
| 624 | SQInteger node = exp->_first; |
| 625 | if(text_begin >= text_end) return SQFalse; |
| 626 | exp->_bol = text_begin; |
| 627 | exp->_eol = text_end; |
| 628 | do { |
| 629 | cur = text_begin; |
| 630 | while(node != -1) { |
| 631 | exp->_currsubexp = 0; |
| 632 | cur = sqstd_rex_matchnode(exp,&exp->_nodes[node],cur,NULL); |
| 633 | if(!cur) |
| 634 | break; |
| 635 | node = exp->_nodes[node].next; |
| 636 | } |
| 637 | text_begin++; |
| 638 | } while(cur == NULL && text_begin != text_end); |
| 639 | |
| 640 | if(cur == NULL) |
| 641 | return SQFalse; |
| 642 | |
| 643 | --text_begin; |
| 644 | |
| 645 | if(out_begin) *out_begin = text_begin; |
| 646 | if(out_end) *out_end = cur; |
| 647 | return SQTrue; |
| 648 | } |
| 649 | |
| 650 | SQBool sqstd_rex_search(SQRex* exp,const SQChar* text, const SQChar** out_begin, const SQChar** out_end) |
| 651 | { |
| 652 | return sqstd_rex_searchrange(exp,text,text + scstrlen(text),out_begin,out_end); |
| 653 | } |
| 654 | |
| 655 | SQInteger sqstd_rex_getsubexpcount(SQRex* exp) |
| 656 | { |
| 657 | return exp->_nsubexpr; |
| 658 | } |
| 659 | |
| 660 | SQBool sqstd_rex_getsubexp(SQRex* exp, SQInteger n, SQRexMatch *subexp) |
| 661 | { |
| 662 | if( n<0 || n >= exp->_nsubexpr) return SQFalse; |
| 663 | *subexp = exp->_matches[n]; |
| 664 | return SQTrue; |
| 665 | } |
| 666 | |
| 667 | |