dlls/jscript/jsregexp.c

   1 /*
   2  * Copyright 2008 Jacek Caban for CodeWeavers
   3  *
   4  * This library is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU Lesser General Public
   6  * License as published by the Free Software Foundation; either
   7  * version 2.1 of the License, or (at your option) any later version.
   8  *
   9  * This library is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  12  * Lesser General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU Lesser General Public
  15  * License along with this library; if not, write to the Free Software
  16  * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
  17  */
  18
  19 #include <math.h>
  20
  21 #include "jscript.h"
  22 #include "regexp.h"
  23
  24 #include "wine/debug.h"
  25
  26 WINE_DEFAULT_DEBUG_CHANNEL(jscript);
  27
  28 typedef struct {
  29     jsdisp_t dispex;
  30
  31     regexp_t *jsregexp;
  32     jsstr_t *str;
  33     INT last_index;
  34     jsval_t last_index_val;
  35 } RegExpInstance;
  36
  37 static const WCHAR sourceW[] = {'s','o','u','r','c','e',0};
  38 static const WCHAR globalW[] = {'g','l','o','b','a','l',0};
  39 static const WCHAR ignoreCaseW[] = {'i','g','n','o','r','e','C','a','s','e',0};
  40 static const WCHAR multilineW[] = {'m','u','l','t','i','l','i','n','e',0};
  41 static const WCHAR lastIndexW[] = {'l','a','s','t','I','n','d','e','x',0};
  42 static const WCHAR toStringW[] = {'t','o','S','t','r','i','n','g',0};
  43 static const WCHAR execW[] = {'e','x','e','c',0};
  44 static const WCHAR testW[] = {'t','e','s','t',0};
  45
  46 static const WCHAR leftContextW[] =
  47     {'l','e','f','t','C','o','n','t','e','x','t',0};
  48 static const WCHAR rightContextW[] =
  49     {'r','i','g','h','t','C','o','n','t','e','x','t',0};
  50
  51 static const WCHAR idx1W[] = {'$','1',0};
  52 static const WCHAR idx2W[] = {'$','2',0};
  53 static const WCHAR idx3W[] = {'$','3',0};
  54 static const WCHAR idx4W[] = {'$','4',0};
  55 static const WCHAR idx5W[] = {'$','5',0};
  56 static const WCHAR idx6W[] = {'$','6',0};
  57 static const WCHAR idx7W[] = {'$','7',0};
  58 static const WCHAR idx8W[] = {'$','8',0};
  59 static const WCHAR idx9W[] = {'$','9',0};
  60
  61 static inline RegExpInstance *regexp_from_jsdisp(jsdisp_t *jsdisp)
  62 {
  63     return CONTAINING_RECORD(jsdisp, RegExpInstance, dispex);
  64 }
  65
  66 static inline RegExpInstance *regexp_from_vdisp(vdisp_t *vdisp)
  67 {
  68     return regexp_from_jsdisp(vdisp->u.jsdisp);
  69 }
  70
  71 static void set_last_index(RegExpInstance *This, DWORD last_index)
  72 {
  73     This->last_index = last_index;
  74     jsval_release(This->last_index_val);
  75     This->last_index_val = jsval_number(last_index);
  76 }
  77
  78 static HRESULT do_regexp_match_next(script_ctx_t *ctx, RegExpInstance *regexp,
  79         DWORD rem_flags, jsstr_t *jsstr, const WCHAR *str, match_state_t *ret)
  80 {
  81     HRESULT hres;
  82
  83     hres = regexp_execute(regexp->jsregexp, ctx, &ctx->tmp_heap,
  84             str, jsstr_length(jsstr), ret);
  85     if(FAILED(hres))
  86         return hres;
  87     if(hres == S_FALSE) {
  88         if(rem_flags & REM_RESET_INDEX)
  89             set_last_index(regexp, 0);
  90         return S_FALSE;
  91     }
  92
  93     if(!(rem_flags & REM_NO_CTX_UPDATE) && ctx->last_match != jsstr) {
  94         jsstr_release(ctx->last_match);
  95         ctx->last_match = jsstr_addref(jsstr);
  96     }
  97
  98     if(!(rem_flags & REM_NO_CTX_UPDATE)) {
  99         DWORD i, n = min(sizeof(ctx->match_parens)/sizeof(ctx->match_parens[0]), ret->paren_count);
 100
 101         for(i=0; i < n; i++) {
 102             if(ret->parens[i].index == -1) {
 103                 ctx->match_parens[i].index = 0;
 104                 ctx->match_parens[i].length = 0;
 105             }else {
 106                 ctx->match_parens[i].index = ret->parens[i].index;
 107                 ctx->match_parens[i].length = ret->parens[i].length;
 108             }
 109         }
 110
 111         if(n < sizeof(ctx->match_parens)/sizeof(ctx->match_parens[0]))
 112             memset(ctx->match_parens+n, 0, sizeof(ctx->match_parens) - n*sizeof(ctx->match_parens[0]));
 113     }
 114
 115     set_last_index(regexp, ret->cp-str);
 116
 117     if(!(rem_flags & REM_NO_CTX_UPDATE)) {
 118         ctx->last_match_index = ret->cp-str-ret->match_len;
 119         ctx->last_match_length = ret->match_len;
 120     }
 121
 122     return S_OK;
 123 }
 124
 125 HRESULT regexp_match_next(script_ctx_t *ctx, jsdisp_t *dispex,
 126         DWORD rem_flags, jsstr_t *jsstr, match_state_t **ret)
 127 {
 128     RegExpInstance *regexp = (RegExpInstance*)dispex;
 129     match_state_t *match;
 130     heap_pool_t *mark;
 131     const WCHAR *str;
 132     HRESULT hres;
 133
 134     if((rem_flags & REM_CHECK_GLOBAL) && !(regexp->jsregexp->flags & REG_GLOB)) {
 135         if(rem_flags & REM_ALLOC_RESULT)
 136             *ret = NULL;
 137         return S_FALSE;
 138     }
 139
 140     str = jsstr_flatten(jsstr);
 141     if(!str)
 142         return E_OUTOFMEMORY;
 143
 144     if(rem_flags & REM_ALLOC_RESULT) {
 145         match = alloc_match_state(regexp->jsregexp, NULL, str);
 146         if(!match)
 147             return E_OUTOFMEMORY;
 148         *ret = match;
 149     }
 150
 151     mark = heap_pool_mark(&ctx->tmp_heap);
 152
 153     if(rem_flags & REM_NO_PARENS) {
 154         match = alloc_match_state(regexp->jsregexp, &ctx->tmp_heap, NULL);
 155         if(!match) {
 156             heap_pool_clear(mark);
 157             return E_OUTOFMEMORY;
 158         }
 159         match->cp = (*ret)->cp;
 160         match->match_len = (*ret)->match_len;
 161     }else {
 162         match = *ret;
 163     }
 164
 165     hres = do_regexp_match_next(ctx, regexp, rem_flags, jsstr, str, match);
 166
 167     if(rem_flags & REM_NO_PARENS) {
 168         (*ret)->cp = match->cp;
 169         (*ret)->match_len = match->match_len;
 170     }
 171
 172     heap_pool_clear(mark);
 173
 174     if(hres != S_OK && (rem_flags & REM_ALLOC_RESULT)) {
 175         heap_free(match);
 176         *ret = NULL;
 177     }
 178
 179     return hres;
 180 }
 181
 182 static HRESULT regexp_match(script_ctx_t *ctx, jsdisp_t *dispex, jsstr_t *jsstr, BOOL gflag,
 183         match_result_t **match_result, DWORD *result_cnt)
 184 {
 185     RegExpInstance *This = (RegExpInstance*)dispex;
 186     match_result_t *ret = NULL;
 187     match_state_t *result;
 188     DWORD i=0, ret_size = 0;
 189     heap_pool_t *mark;
 190     const WCHAR *str;
 191     HRESULT hres;
 192
 193     mark = heap_pool_mark(&ctx->tmp_heap);
 194
 195     str = jsstr_flatten(jsstr);
 196     if(!str)
 197         return E_OUTOFMEMORY;
 198
 199     result = alloc_match_state(This->jsregexp, &ctx->tmp_heap, str);
 200     if(!result) {
 201         heap_pool_clear(mark);
 202         return E_OUTOFMEMORY;
 203     }
 204
 205     while(1) {
 206         hres = do_regexp_match_next(ctx, This, 0, jsstr, str, result);
 207         if(hres == S_FALSE) {
 208             hres = S_OK;
 209             break;
 210         }
 211
 212         if(FAILED(hres))
 213             break;
 214
 215         if(ret_size == i) {
 216             if(ret) {
 217                 match_result_t *old_ret = ret;
 218
 219                 ret = heap_realloc(old_ret, (ret_size <<= 1) * sizeof(match_result_t));
 220                 if(!ret)
 221                     heap_free(old_ret);
 222             }else {
 223                 ret = heap_alloc((ret_size=4) * sizeof(match_result_t));
 224             }
 225             if(!ret) {
 226                 hres = E_OUTOFMEMORY;
 227                 break;
 228             }
 229         }
 230
 231         ret[i].index = result->cp - str - result->match_len;
 232         ret[i++].length = result->match_len;
 233
 234         if(!gflag && !(This->jsregexp->flags & REG_GLOB)) {
 235             hres = S_OK;
 236             break;
 237         }
 238     }
 239
 240     heap_pool_clear(mark);
 241     if(FAILED(hres)) {
 242         heap_free(ret);
 243         return hres;
 244     }
 245
 246     *match_result = ret;
 247     *result_cnt = i;
 248     return S_OK;
 249 }
 250
 251 static HRESULT RegExp_get_source(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t *r)
 252 {
 253     TRACE("\n");
 254
 255     *r = jsval_string(jsstr_addref(regexp_from_jsdisp(jsthis)->str));
 256     return S_OK;
 257 }
 258
 259 static HRESULT RegExp_set_source(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t value)
 260 {
 261     FIXME("\n");
 262     return E_NOTIMPL;
 263 }
 264
 265 static HRESULT RegExp_get_global(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t *r)
 266 {
 267     FIXME("\n");
 268     return E_NOTIMPL;
 269 }
 270
 271 static HRESULT RegExp_set_global(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t value)
 272 {
 273     FIXME("\n");
 274     return E_NOTIMPL;
 275 }
 276
 277 static HRESULT RegExp_get_ignoreCase(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t *r)
 278 {
 279     FIXME("\n");
 280     return E_NOTIMPL;
 281 }
 282
 283 static HRESULT RegExp_set_ignoreCase(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t value)
 284 {
 285     FIXME("\n");
 286     return E_NOTIMPL;
 287 }
 288
 289 static HRESULT RegExp_get_multiline(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t *r)
 290 {
 291     FIXME("\n");
 292     return E_NOTIMPL;
 293 }
 294
 295 static HRESULT RegExp_set_multiline(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t value)
 296 {
 297     FIXME("\n");
 298     return E_NOTIMPL;
 299 }
 300
 301 static INT index_from_val(script_ctx_t *ctx, jsval_t v)
 302 {
 303     double n;
 304     HRESULT hres;
 305
 306     hres = to_number(ctx, v, &n);
 307     if(FAILED(hres)) {
 308         clear_ei(ctx); /* FIXME: Move ignoring exceptions to to_primitive */
 309         return 0;
 310     }
 311
 312     n = floor(n);
 313     return is_int32(n) ? n : 0;
 314 }
 315
 316 static HRESULT RegExp_get_lastIndex(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t *r)
 317 {
 318     RegExpInstance *regexp = regexp_from_jsdisp(jsthis);
 319
 320     TRACE("\n");
 321
 322     return jsval_copy(regexp->last_index_val, r);
 323 }
 324
 325 static HRESULT RegExp_set_lastIndex(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t value)
 326 {
 327     RegExpInstance *regexp = regexp_from_jsdisp(jsthis);
 328     HRESULT hres;
 329
 330     TRACE("\n");
 331
 332     hres = jsval_copy(value, &regexp->last_index_val);
 333     if(FAILED(hres))
 334         return hres;
 335
 336     regexp->last_index = index_from_val(ctx, value);
 337     return S_OK;
 338 }
 339
 340 static HRESULT RegExp_toString(script_ctx_t *ctx, vdisp_t *jsthis, WORD flags, unsigned argc, jsval_t *argv,
 341         jsval_t *r)
 342 {
 343     RegExpInstance *regexp;
 344     unsigned len, f;
 345     jsstr_t *ret;
 346     WCHAR *ptr;
 347
 348     TRACE("\n");
 349
 350     if(!is_vclass(jsthis, JSCLASS_REGEXP)) {
 351         FIXME("Not a RegExp\n");
 352         return E_NOTIMPL;
 353     }
 354
 355     regexp = regexp_from_vdisp(jsthis);
 356
 357     if(!r)
 358         return S_OK;
 359
 360     len = jsstr_length(regexp->str) + 2;
 361
 362     f = regexp->jsregexp->flags;
 363     if(f & REG_FOLD)
 364         len++;
 365     if(f & REG_GLOB)
 366         len++;
 367     if(f & REG_MULTILINE)
 368         len++;
 369
 370     ptr = jsstr_alloc_buf(len, &ret);
 371     if(!ptr)
 372         return E_OUTOFMEMORY;
 373
 374     *ptr++ = '/';
 375     ptr += jsstr_flush(regexp->str, ptr);
 376     *ptr++ = '/';
 377
 378     if(f & REG_FOLD)
 379         *ptr++ = 'i';
 380     if(f & REG_GLOB)
 381         *ptr++ = 'g';
 382     if(f & REG_MULTILINE)
 383         *ptr++ = 'm';
 384
 385     *r = jsval_string(ret);
 386     return S_OK;
 387 }
 388
 389 static HRESULT create_match_array(script_ctx_t *ctx, jsstr_t *input_str,
 390         const match_state_t *result, IDispatch **ret)
 391 {
 392     const WCHAR *input;
 393     jsdisp_t *array;
 394     jsstr_t *str;
 395     DWORD i;
 396     HRESULT hres = S_OK;
 397
 398     static const WCHAR indexW[] = {'i','n','d','e','x',0};
 399     static const WCHAR inputW[] = {'i','n','p','u','t',0};
 400     static const WCHAR lastIndexW[] = {'l','a','s','t','I','n','d','e','x',0};
 401     static const WCHAR zeroW[] = {'0',0};
 402
 403     input = jsstr_flatten(input_str);
 404     if(!input)
 405         return E_OUTOFMEMORY;
 406
 407     hres = create_array(ctx, result->paren_count+1, &array);
 408     if(FAILED(hres))
 409         return hres;
 410
 411     for(i=0; i < result->paren_count; i++) {
 412         if(result->parens[i].index != -1)
 413             str = jsstr_substr(input_str, result->parens[i].index, result->parens[i].length);
 414         else
 415             str = jsstr_empty();
 416         if(!str) {
 417             hres = E_OUTOFMEMORY;
 418             break;
 419         }
 420
 421         hres = jsdisp_propput_idx(array, i+1, jsval_string(str));
 422         jsstr_release(str);
 423         if(FAILED(hres))
 424             break;
 425     }
 426
 427     while(SUCCEEDED(hres)) {
 428         hres = jsdisp_propput_name(array, indexW, jsval_number(result->cp-input-result->match_len));
 429         if(FAILED(hres))
 430             break;
 431
 432         hres = jsdisp_propput_name(array, lastIndexW, jsval_number(result->cp-input));
 433         if(FAILED(hres))
 434             break;
 435
 436         hres = jsdisp_propput_name(array, inputW, jsval_string(jsstr_addref(input_str)));
 437         if(FAILED(hres))
 438             break;
 439
 440         str = jsstr_alloc_len(result->cp-result->match_len, result->match_len);
 441         if(!str) {
 442             hres = E_OUTOFMEMORY;
 443             break;
 444         }
 445         hres = jsdisp_propput_name(array, zeroW, jsval_string(str));
 446         jsstr_release(str);
 447         break;
 448     }
 449
 450     if(FAILED(hres)) {
 451         jsdisp_release(array);
 452         return hres;
 453     }
 454
 455     *ret = to_disp(array);
 456     return S_OK;
 457 }
 458
 459 static HRESULT run_exec(script_ctx_t *ctx, vdisp_t *jsthis, jsval_t arg,
 460         jsstr_t **input, match_state_t **result, BOOL *ret)
 461 {
 462     RegExpInstance *regexp;
 463     match_state_t *match;
 464     DWORD last_index = 0;
 465     const WCHAR *string;
 466     jsstr_t *jsstr;
 467     HRESULT hres;
 468
 469     if(!is_vclass(jsthis, JSCLASS_REGEXP)) {
 470         FIXME("Not a RegExp\n");
 471         return E_NOTIMPL;
 472     }
 473
 474     regexp = regexp_from_vdisp(jsthis);
 475
 476     hres = to_flat_string(ctx, arg, &jsstr, &string);
 477     if(FAILED(hres))
 478         return hres;
 479
 480     if(regexp->jsregexp->flags & REG_GLOB) {
 481         if(regexp->last_index < 0) {
 482             jsstr_release(jsstr);
 483             set_last_index(regexp, 0);
 484             *ret = FALSE;
 485             if(input)
 486                 *input = jsstr_empty();
 487             return S_OK;
 488         }
 489
 490         last_index = regexp->last_index;
 491     }
 492
 493     match = alloc_match_state(regexp->jsregexp, &ctx->tmp_heap, string+last_index);
 494     if(!match) {
 495         jsstr_release(jsstr);
 496         return E_OUTOFMEMORY;
 497     }
 498
 499     hres = regexp_match_next(ctx, &regexp->dispex, REM_RESET_INDEX, jsstr, &match);
 500     if(FAILED(hres)) {
 501         jsstr_release(jsstr);
 502         return hres;
 503     }
 504
 505     *result = match;
 506     *ret = hres == S_OK;
 507     if(input)
 508         *input = jsstr;
 509     else
 510         jsstr_release(jsstr);
 511     return S_OK;
 512 }
 513
 514 static HRESULT RegExp_exec(script_ctx_t *ctx, vdisp_t *jsthis, WORD flags, unsigned argc, jsval_t *argv,
 515         jsval_t *r)
 516 {
 517     match_state_t *match;
 518     heap_pool_t *mark;
 519     BOOL b;
 520     jsstr_t *string;
 521     HRESULT hres;
 522
 523     TRACE("\n");
 524
 525     mark = heap_pool_mark(&ctx->tmp_heap);
 526
 527     hres = run_exec(ctx, jsthis, argc ? argv[0] : jsval_string(jsstr_empty()), &string, &match, &b);
 528     if(FAILED(hres)) {
 529         heap_pool_clear(mark);
 530         return hres;
 531     }
 532
 533     if(r) {
 534         if(b) {
 535             IDispatch *ret;
 536
 537             hres = create_match_array(ctx, string, match, &ret);
 538             if(SUCCEEDED(hres))
 539                 *r = jsval_disp(ret);
 540         }else {
 541             *r = jsval_null();
 542         }
 543     }
 544
 545     heap_pool_clear(mark);
 546     jsstr_release(string);
 547     return hres;
 548 }
 549
 550 static HRESULT RegExp_test(script_ctx_t *ctx, vdisp_t *jsthis, WORD flags, unsigned argc, jsval_t *argv,
 551         jsval_t *r)
 552 {
 553     match_state_t *match;
 554     jsstr_t *undef_str;
 555     heap_pool_t *mark;
 556     BOOL b;
 557     HRESULT hres;
 558
 559     TRACE("\n");
 560
 561     mark = heap_pool_mark(&ctx->tmp_heap);
 562     hres = run_exec(ctx, jsthis, argc ? argv[0] : jsval_string(undef_str = jsstr_undefined()), NULL, &match, &b);
 563     heap_pool_clear(mark);
 564     if(!argc)
 565         jsstr_release(undef_str);
 566     if(FAILED(hres))
 567         return hres;
 568
 569     if(r)
 570         *r = jsval_bool(b);
 571     return S_OK;
 572 }
 573
 574 static HRESULT RegExp_value(script_ctx_t *ctx, vdisp_t *jsthis, WORD flags, unsigned argc, jsval_t *argv,
 575         jsval_t *r)
 576 {
 577     TRACE("\n");
 578
 579     switch(flags) {
 580     case INVOKE_FUNC:
 581         return throw_type_error(ctx, JS_E_FUNCTION_EXPECTED, NULL);
 582     default:
 583         FIXME("unimplemented flags %x\n", flags);
 584         return E_NOTIMPL;
 585     }
 586
 587     return S_OK;
 588 }
 589
 590 static void RegExp_destructor(jsdisp_t *dispex)
 591 {
 592     RegExpInstance *This = (RegExpInstance*)dispex;
 593
 594     if(This->jsregexp)
 595         regexp_destroy(This->jsregexp);
 596     jsval_release(This->last_index_val);
 597     jsstr_release(This->str);
 598     heap_free(This);
 599 }
 600
 601 static const builtin_prop_t RegExp_props[] = {
 602     {execW,                  RegExp_exec,                  PROPF_METHOD|1},
 603     {globalW,                NULL,0,                       RegExp_get_global,     RegExp_set_global},
 604     {ignoreCaseW,            NULL,0,                       RegExp_get_ignoreCase, RegExp_set_ignoreCase},
 605     {lastIndexW,             NULL,0,                       RegExp_get_lastIndex,  RegExp_set_lastIndex},
 606     {multilineW,             NULL,0,                       RegExp_get_multiline,  RegExp_set_multiline},
 607     {sourceW,                NULL,0,                       RegExp_get_source,     RegExp_set_source},
 608     {testW,                  RegExp_test,                  PROPF_METHOD|1},
 609     {toStringW,              RegExp_toString,              PROPF_METHOD}
 610 };
 611
 612 static const builtin_info_t RegExp_info = {
 613     JSCLASS_REGEXP,
 614     {NULL, RegExp_value, 0},
 615     sizeof(RegExp_props)/sizeof(*RegExp_props),
 616     RegExp_props,
 617     RegExp_destructor,
 618     NULL
 619 };
 620
 621 static const builtin_prop_t RegExpInst_props[] = {
 622     {globalW,                NULL,0,                       RegExp_get_global,     RegExp_set_global},
 623     {ignoreCaseW,            NULL,0,                       RegExp_get_ignoreCase, RegExp_set_ignoreCase},
 624     {lastIndexW,             NULL,0,                       RegExp_get_lastIndex,  RegExp_set_lastIndex},
 625     {multilineW,             NULL,0,                       RegExp_get_multiline,  RegExp_set_multiline},
 626     {sourceW,                NULL,0,                       RegExp_get_source,     RegExp_set_source}
 627 };
 628
 629 static const builtin_info_t RegExpInst_info = {
 630     JSCLASS_REGEXP,
 631     {NULL, RegExp_value, 0},
 632     sizeof(RegExpInst_props)/sizeof(*RegExpInst_props),
 633     RegExpInst_props,
 634     RegExp_destructor,
 635     NULL
 636 };
 637
 638 static HRESULT alloc_regexp(script_ctx_t *ctx, jsdisp_t *object_prototype, RegExpInstance **ret)
 639 {
 640     RegExpInstance *regexp;
 641     HRESULT hres;
 642
 643     regexp = heap_alloc_zero(sizeof(RegExpInstance));
 644     if(!regexp)
 645         return E_OUTOFMEMORY;
 646
 647     if(object_prototype)
 648         hres = init_dispex(&regexp->dispex, ctx, &RegExp_info, object_prototype);
 649     else
 650         hres = init_dispex_from_constr(&regexp->dispex, ctx, &RegExpInst_info, ctx->regexp_constr);
 651
 652     if(FAILED(hres)) {
 653         heap_free(regexp);
 654         return hres;
 655     }
 656
 657     *ret = regexp;
 658     return S_OK;
 659 }
 660
 661 HRESULT create_regexp(script_ctx_t *ctx, jsstr_t *src, DWORD flags, jsdisp_t **ret)
 662 {
 663     RegExpInstance *regexp;
 664     const WCHAR *str;
 665     HRESULT hres;
 666
 667     TRACE("%s %x\n", debugstr_jsstr(src), flags);
 668
 669     str = jsstr_flatten(src);
 670     if(!str)
 671         return E_OUTOFMEMORY;
 672
 673     hres = alloc_regexp(ctx, NULL, &regexp);
 674     if(FAILED(hres))
 675         return hres;
 676
 677     regexp->str = jsstr_addref(src);
 678     regexp->last_index_val = jsval_number(0);
 679
 680     regexp->jsregexp = regexp_new(ctx, &ctx->tmp_heap, str, jsstr_length(regexp->str), flags, FALSE);
 681     if(!regexp->jsregexp) {
 682         WARN("regexp_new failed\n");
 683         jsdisp_release(&regexp->dispex);
 684         return E_FAIL;
 685     }
 686
 687     *ret = &regexp->dispex;
 688     return S_OK;
 689 }
 690
 691 HRESULT create_regexp_var(script_ctx_t *ctx, jsval_t src_arg, jsval_t *flags_arg, jsdisp_t **ret)
 692 {
 693     unsigned flags, opt_len = 0;
 694     const WCHAR *opt = NULL;
 695     jsstr_t *src;
 696     HRESULT hres;
 697
 698     if(is_object_instance(src_arg)) {
 699         jsdisp_t *obj;
 700
 701         obj = iface_to_jsdisp((IUnknown*)get_object(src_arg));
 702         if(obj) {
 703             if(is_class(obj, JSCLASS_REGEXP)) {
 704                 RegExpInstance *regexp = (RegExpInstance*)obj;
 705
 706                 hres = create_regexp(ctx, regexp->str, regexp->jsregexp->flags, ret);
 707                 jsdisp_release(obj);
 708                 return hres;
 709             }
 710
 711             jsdisp_release(obj);
 712         }
 713     }
 714
 715     if(!is_string(src_arg)) {
 716         FIXME("src_arg = %s\n", debugstr_jsval(src_arg));
 717         return E_NOTIMPL;
 718     }
 719
 720     src = get_string(src_arg);
 721
 722     if(flags_arg) {
 723         jsstr_t *opt_str;
 724
 725         if(!is_string(*flags_arg)) {
 726             FIXME("unimplemented for %s\n", debugstr_jsval(*flags_arg));
 727             return E_NOTIMPL;
 728         }
 729
 730         opt_str = get_string(*flags_arg);
 731         opt = jsstr_flatten(opt_str);
 732         if(!opt)
 733             return E_OUTOFMEMORY;
 734         opt_len = jsstr_length(opt_str);
 735     }
 736
 737     hres = parse_regexp_flags(opt, opt_len, &flags);
 738     if(FAILED(hres))
 739         return hres;
 740
 741     return create_regexp(ctx, src, flags, ret);
 742 }
 743
 744 HRESULT regexp_string_match(script_ctx_t *ctx, jsdisp_t *re, jsstr_t *jsstr, jsval_t *r)
 745 {
 746     static const WCHAR indexW[] = {'i','n','d','e','x',0};
 747     static const WCHAR inputW[] = {'i','n','p','u','t',0};
 748     static const WCHAR lastIndexW[] = {'l','a','s','t','I','n','d','e','x',0};
 749
 750     RegExpInstance *regexp = (RegExpInstance*)re;
 751     match_result_t *match_result;
 752     unsigned match_cnt, i;
 753     const WCHAR *str;
 754     jsdisp_t *array;
 755     HRESULT hres;
 756
 757     str = jsstr_flatten(jsstr);
 758     if(!str)
 759         return E_OUTOFMEMORY;
 760
 761     if(!(regexp->jsregexp->flags & REG_GLOB)) {
 762         match_state_t *match;
 763         heap_pool_t *mark;
 764
 765         mark = heap_pool_mark(&ctx->tmp_heap);
 766         match = alloc_match_state(regexp->jsregexp, &ctx->tmp_heap, str);
 767         if(!match) {
 768             heap_pool_clear(mark);
 769             return E_OUTOFMEMORY;
 770         }
 771
 772         hres = regexp_match_next(ctx, &regexp->dispex, 0, jsstr, &match);
 773         if(FAILED(hres)) {
 774             heap_pool_clear(mark);
 775             return hres;
 776         }
 777
 778         if(r) {
 779             if(hres == S_OK) {
 780                 IDispatch *ret;
 781
 782                 hres = create_match_array(ctx, jsstr, match, &ret);
 783                 if(SUCCEEDED(hres))
 784                     *r = jsval_disp(ret);
 785             }else {
 786                 *r = jsval_null();
 787             }
 788         }
 789
 790         heap_pool_clear(mark);
 791         return S_OK;
 792     }
 793
 794     hres = regexp_match(ctx, &regexp->dispex, jsstr, FALSE, &match_result, &match_cnt);
 795     if(FAILED(hres))
 796         return hres;
 797
 798     if(!match_cnt) {
 799         TRACE("no match\n");
 800
 801         if(r)
 802             *r = jsval_null();
 803         return S_OK;
 804     }
 805
 806     hres = create_array(ctx, match_cnt, &array);
 807     if(FAILED(hres))
 808         return hres;
 809
 810     for(i=0; i < match_cnt; i++) {
 811         jsstr_t *tmp_str;
 812
 813         tmp_str = jsstr_substr(jsstr, match_result[i].index, match_result[i].length);
 814         if(!tmp_str) {
 815             hres = E_OUTOFMEMORY;
 816             break;
 817         }
 818
 819         hres = jsdisp_propput_idx(array, i, jsval_string(tmp_str));
 820         jsstr_release(tmp_str);
 821         if(FAILED(hres))
 822             break;
 823     }
 824
 825     while(SUCCEEDED(hres)) {
 826         hres = jsdisp_propput_name(array, indexW, jsval_number(match_result[match_cnt-1].index));
 827         if(FAILED(hres))
 828             break;
 829
 830         hres = jsdisp_propput_name(array, lastIndexW,
 831                 jsval_number(match_result[match_cnt-1].index + match_result[match_cnt-1].length));
 832         if(FAILED(hres))
 833             break;
 834
 835         hres = jsdisp_propput_name(array, inputW, jsval_string(jsstr));
 836         break;
 837     }
 838
 839     heap_free(match_result);
 840
 841     if(SUCCEEDED(hres) && r)
 842         *r = jsval_obj(array);
 843     else
 844         jsdisp_release(array);
 845     return hres;
 846 }
 847
 848 static HRESULT global_idx(script_ctx_t *ctx, DWORD idx, jsval_t *r)
 849 {
 850     jsstr_t *ret;
 851
 852     ret = jsstr_substr(ctx->last_match, ctx->match_parens[idx].index, ctx->match_parens[idx].length);
 853     if(!ret)
 854         return E_OUTOFMEMORY;
 855
 856     *r = jsval_string(ret);
 857     return S_OK;
 858 }
 859
 860 static HRESULT RegExpConstr_get_idx1(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t *r)
 861 {
 862     TRACE("\n");
 863     return global_idx(ctx, 0, r);
 864 }
 865
 866 static HRESULT RegExpConstr_get_idx2(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t *r)
 867 {
 868     TRACE("\n");
 869     return global_idx(ctx, 1, r);
 870 }
 871
 872 static HRESULT RegExpConstr_get_idx3(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t *r)
 873 {
 874     TRACE("\n");
 875     return global_idx(ctx, 2, r);
 876 }
 877
 878 static HRESULT RegExpConstr_get_idx4(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t *r)
 879 {
 880     TRACE("\n");
 881     return global_idx(ctx, 3, r);
 882 }
 883
 884 static HRESULT RegExpConstr_get_idx5(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t *r)
 885 {
 886     TRACE("\n");
 887     return global_idx(ctx, 4, r);
 888 }
 889
 890 static HRESULT RegExpConstr_get_idx6(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t *r)
 891 {
 892     TRACE("\n");
 893     return global_idx(ctx, 5, r);
 894 }
 895
 896 static HRESULT RegExpConstr_get_idx7(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t *r)
 897 {
 898     TRACE("\n");
 899     return global_idx(ctx, 6, r);
 900 }
 901
 902 static HRESULT RegExpConstr_get_idx8(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t *r)
 903 {
 904     TRACE("\n");
 905     return global_idx(ctx, 7, r);
 906 }
 907
 908 static HRESULT RegExpConstr_get_idx9(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t *r)
 909 {
 910     TRACE("\n");
 911     return global_idx(ctx, 8, r);
 912 }
 913
 914 static HRESULT RegExpConstr_get_leftContext(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t *r)
 915 {
 916     jsstr_t *ret;
 917
 918     TRACE("\n");
 919
 920     ret = jsstr_substr(ctx->last_match, 0, ctx->last_match_index);
 921     if(!ret)
 922         return E_OUTOFMEMORY;
 923
 924     *r = jsval_string(ret);
 925     return S_OK;
 926 }
 927
 928 static HRESULT RegExpConstr_get_rightContext(script_ctx_t *ctx, jsdisp_t *jsthis, jsval_t *r)
 929 {
 930     jsstr_t *ret;
 931
 932     TRACE("\n");
 933
 934     ret = jsstr_substr(ctx->last_match, ctx->last_match_index+ctx->last_match_length,
 935             jsstr_length(ctx->last_match) - ctx->last_match_index - ctx->last_match_length);
 936     if(!ret)
 937         return E_OUTOFMEMORY;
 938
 939     *r = jsval_string(ret);
 940     return S_OK;
 941 }
 942
 943 static HRESULT RegExpConstr_value(script_ctx_t *ctx, vdisp_t *jsthis, WORD flags, unsigned argc, jsval_t *argv,
 944         jsval_t *r)
 945 {
 946     TRACE("\n");
 947
 948     switch(flags) {
 949     case DISPATCH_METHOD:
 950         if(argc) {
 951             if(is_object_instance(argv[0])) {
 952                 jsdisp_t *jsdisp = iface_to_jsdisp((IUnknown*)get_object(argv[0]));
 953                 if(jsdisp) {
 954                     if(is_class(jsdisp, JSCLASS_REGEXP)) {
 955                         if(argc > 1 && !is_undefined(argv[1])) {
 956                             jsdisp_release(jsdisp);
 957                             return throw_regexp_error(ctx, JS_E_REGEXP_SYNTAX, NULL);
 958                         }
 959
 960                         if(r)
 961                             *r = jsval_obj(jsdisp);
 962                         else
 963                             jsdisp_release(jsdisp);
 964                         return S_OK;
 965                     }
 966                     jsdisp_release(jsdisp);
 967                 }
 968             }
 969         }
 970         /* fall through */
 971     case DISPATCH_CONSTRUCT: {
 972         jsdisp_t *ret;
 973         HRESULT hres;
 974
 975         if(!argc) {
 976             FIXME("no args\n");
 977             return E_NOTIMPL;
 978         }
 979
 980         hres = create_regexp_var(ctx, argv[0], argc > 1 ? argv+1 : NULL, &ret);
 981         if(FAILED(hres))
 982             return hres;
 983
 984         if(r)
 985             *r = jsval_obj(ret);
 986         else
 987             jsdisp_release(ret);
 988         return S_OK;
 989     }
 990     default:
 991         FIXME("unimplemented flags: %x\n", flags);
 992         return E_NOTIMPL;
 993     }
 994
 995     return S_OK;
 996 }
 997
 998 static const builtin_prop_t RegExpConstr_props[] = {
 999     {idx1W,           NULL,0,  RegExpConstr_get_idx1,         builtin_set_const},
1000     {idx2W,           NULL,0,  RegExpConstr_get_idx2,         builtin_set_const},
1001     {idx3W,           NULL,0,  RegExpConstr_get_idx3,         builtin_set_const},
1002     {idx4W,           NULL,0,  RegExpConstr_get_idx4,         builtin_set_const},
1003     {idx5W,           NULL,0,  RegExpConstr_get_idx5,         builtin_set_const},
1004     {idx6W,           NULL,0,  RegExpConstr_get_idx6,         builtin_set_const},
1005     {idx7W,           NULL,0,  RegExpConstr_get_idx7,         builtin_set_const},
1006     {idx8W,           NULL,0,  RegExpConstr_get_idx8,         builtin_set_const},
1007     {idx9W,           NULL,0,  RegExpConstr_get_idx9,         builtin_set_const},
1008     {leftContextW,    NULL,0,  RegExpConstr_get_leftContext,  builtin_set_const},
1009     {rightContextW,   NULL,0,  RegExpConstr_get_rightContext, builtin_set_const}
1010 };
1011
1012 static const builtin_info_t RegExpConstr_info = {
1013     JSCLASS_FUNCTION,
1014     DEFAULT_FUNCTION_VALUE,
1015     sizeof(RegExpConstr_props)/sizeof(*RegExpConstr_props),
1016     RegExpConstr_props,
1017     NULL,
1018     NULL
1019 };
1020
1021 HRESULT create_regexp_constr(script_ctx_t *ctx, jsdisp_t *object_prototype, jsdisp_t **ret)
1022 {
1023     RegExpInstance *regexp;
1024     HRESULT hres;
1025
1026     static const WCHAR RegExpW[] = {'R','e','g','E','x','p',0};
1027
1028     hres = alloc_regexp(ctx, object_prototype, &regexp);
1029     if(FAILED(hres))
1030         return hres;
1031
1032     hres = create_builtin_constructor(ctx, RegExpConstr_value, RegExpW, &RegExpConstr_info,
1033             PROPF_CONSTR|2, &regexp->dispex, ret);
1034
1035     jsdisp_release(&regexp->dispex);
1036     return hres;
1037 }
1038
1039 HRESULT parse_regexp_flags(const WCHAR *str, DWORD str_len, DWORD *ret)
1040 {
1041     const WCHAR *p;
1042     DWORD flags = 0;
1043
1044     for (p = str; p < str+str_len; p++) {
1045         switch (*p) {
1046         case 'g':
1047             flags |= REG_GLOB;
1048             break;
1049         case 'i':
1050             flags |= REG_FOLD;
1051             break;
1052         case 'm':
1053             flags |= REG_MULTILINE;
1054             break;
1055         case 'y':
1056             flags |= REG_STICKY;
1057             break;
1058         default:
1059             WARN("wrong flag %c\n", *p);
1060             return E_FAIL;
1061         }
1062     }
1063
1064     *ret = flags;
1065     return S_OK;
1066 }