1/************************************************* 2* Perl-Compatible Regular Expressions * 3*************************************************/ 4 5/* PCRE is a library of functions to support regular expressions whose syntax 6and semantics are as close as possible to those of the Perl 5 language. 7 8 Written by Philip Hazel 9 Copyright (c) 1997-2010 University of Cambridge 10 11----------------------------------------------------------------------------- 12Redistribution and use in source and binary forms, with or without 13modification, are permitted provided that the following conditions are met: 14 15 * Redistributions of source code must retain the above copyright notice, 16 this list of conditions and the following disclaimer. 17 18 * Redistributions in binary form must reproduce the above copyright 19 notice, this list of conditions and the following disclaimer in the 20 documentation and/or other materials provided with the distribution. 21 22 * Neither the name of the University of Cambridge nor the names of its 23 contributors may be used to endorse or promote products derived from 24 this software without specific prior written permission. 25 26THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" 27AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 28IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 29ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE 30LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 31CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 32SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 33INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 34CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 35ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 36POSSIBILITY OF SUCH DAMAGE. 37----------------------------------------------------------------------------- 38*/ 39 40 41/* This module contains a PCRE private debugging function for printing out the 42internal form of a compiled regular expression, along with some supporting 43local functions. This source file is used in two places: 44 45(1) It is #included by pcre_compile.c when it is compiled in debugging mode 46(PCRE_DEBUG defined in pcre_internal.h). It is not included in production 47compiles. 48 49(2) It is always #included by pcretest.c, which can be asked to print out a 50compiled regex for debugging purposes. */ 51 52 53/* Macro that decides whether a character should be output as a literal or in 54hexadecimal. We don't use isprint() because that can vary from system to system 55(even without the use of locales) and we want the output always to be the same, 56for testing purposes. This macro is used in pcretest as well as in this file. */ 57 58#ifdef EBCDIC 59#define PRINTABLE(c) ((c) >= 64 && (c) < 255) 60#else 61#define PRINTABLE(c) ((c) >= 32 && (c) < 127) 62#endif 63 64/* The table of operator names. */ 65 66static const char *OP_names[] = { OP_NAME_LIST }; 67 68 69 70/************************************************* 71* Print single- or multi-byte character * 72*************************************************/ 73 74static int 75print_char(FILE *f, uschar *ptr, BOOL utf8) 76{ 77int c = *ptr; 78 79#ifndef SUPPORT_UTF8 80utf8 = utf8; /* Avoid compiler warning */ 81if (PRINTABLE(c)) fprintf(f, "%c", c); else fprintf(f, "\\x%02x", c); 82return 0; 83 84#else 85if (!utf8 || (c & 0xc0) != 0xc0) 86 { 87 if (PRINTABLE(c)) fprintf(f, "%c", c); else fprintf(f, "\\x%02x", c); 88 return 0; 89 } 90else 91 { 92 int i; 93 int a = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ 94 int s = 6*a; 95 c = (c & _pcre_utf8_table3[a]) << s; 96 for (i = 1; i <= a; i++) 97 { 98 /* This is a check for malformed UTF-8; it should only occur if the sanity 99 check has been turned off. Rather than swallow random bytes, just stop if 100 we hit a bad one. Print it with \X instead of \x as an indication. */ 101 102 if ((ptr[i] & 0xc0) != 0x80) 103 { 104 fprintf(f, "\\X{%x}", c); 105 return i - 1; 106 } 107 108 /* The byte is OK */ 109 110 s -= 6; 111 c |= (ptr[i] & 0x3f) << s; 112 } 113 if (c < 128) fprintf(f, "\\x%02x", c); else fprintf(f, "\\x{%x}", c); 114 return a; 115 } 116#endif 117} 118 119 120 121/************************************************* 122* Find Unicode property name * 123*************************************************/ 124 125static const char * 126get_ucpname(int ptype, int pvalue) 127{ 128#ifdef SUPPORT_UCP 129int i; 130for (i = _pcre_utt_size - 1; i >= 0; i--) 131 { 132 if (ptype == _pcre_utt[i].type && pvalue == _pcre_utt[i].value) break; 133 } 134return (i >= 0)? _pcre_utt_names + _pcre_utt[i].name_offset : "??"; 135#else 136/* It gets harder and harder to shut off unwanted compiler warnings. */ 137ptype = ptype * pvalue; 138return (ptype == pvalue)? "??" : "??"; 139#endif 140} 141 142 143 144/************************************************* 145* Print compiled regex * 146*************************************************/ 147 148/* Make this function work for a regex with integers either byte order. 149However, we assume that what we are passed is a compiled regex. The 150print_lengths flag controls whether offsets and lengths of items are printed. 151They can be turned off from pcretest so that automatic tests on bytecode can be 152written that do not depend on the value of LINK_SIZE. */ 153 154static void 155pcre_printint(pcre *external_re, FILE *f, BOOL print_lengths) 156{ 157real_pcre *re = (real_pcre *)external_re; 158uschar *codestart, *code; 159BOOL utf8; 160 161unsigned int options = re->options; 162int offset = re->name_table_offset; 163int count = re->name_count; 164int size = re->name_entry_size; 165 166if (re->magic_number != MAGIC_NUMBER) 167 { 168 offset = ((offset << 8) & 0xff00) | ((offset >> 8) & 0xff); 169 count = ((count << 8) & 0xff00) | ((count >> 8) & 0xff); 170 size = ((size << 8) & 0xff00) | ((size >> 8) & 0xff); 171 options = ((options << 24) & 0xff000000) | 172 ((options << 8) & 0x00ff0000) | 173 ((options >> 8) & 0x0000ff00) | 174 ((options >> 24) & 0x000000ff); 175 } 176 177code = codestart = (uschar *)re + offset + count * size; 178utf8 = (options & PCRE_UTF8) != 0; 179 180for(;;) 181 { 182 uschar *ccode; 183 int c; 184 int extra = 0; 185 186 if (print_lengths) 187 fprintf(f, "%3d ", (int)(code - codestart)); 188 else 189 fprintf(f, " "); 190 191 switch(*code) 192 { 193/* ========================================================================== */ 194 /* These cases are never obeyed. This is a fudge that causes a compile- 195 time error if the vectors OP_names or _pcre_OP_lengths, which are indexed 196 by opcode, are not the correct length. It seems to be the only way to do 197 such a check at compile time, as the sizeof() operator does not work in 198 the C preprocessor. We do this while compiling pcretest, because that 199 #includes pcre_tables.c, which holds _pcre_OP_lengths. We can't do this 200 when building pcre_compile.c with PCRE_DEBUG set, because it doesn't then 201 know the size of _pcre_OP_lengths. */ 202 203#ifdef COMPILING_PCRETEST 204 case OP_TABLE_LENGTH: 205 case OP_TABLE_LENGTH + 206 ((sizeof(OP_names)/sizeof(const char *) == OP_TABLE_LENGTH) && 207 (sizeof(_pcre_OP_lengths) == OP_TABLE_LENGTH)): 208 break; 209#endif 210/* ========================================================================== */ 211 212 case OP_END: 213 fprintf(f, " %s\n", OP_names[*code]); 214 fprintf(f, "------------------------------------------------------------------\n"); 215 return; 216 217 case OP_OPT: 218 fprintf(f, " %.2x %s", code[1], OP_names[*code]); 219 break; 220 221 case OP_CHAR: 222 fprintf(f, " "); 223 do 224 { 225 code++; 226 code += 1 + print_char(f, code, utf8); 227 } 228 while (*code == OP_CHAR); 229 fprintf(f, "\n"); 230 continue; 231 232 case OP_CHARNC: 233 fprintf(f, " NC "); 234 do 235 { 236 code++; 237 code += 1 + print_char(f, code, utf8); 238 } 239 while (*code == OP_CHARNC); 240 fprintf(f, "\n"); 241 continue; 242 243 case OP_CBRA: 244 case OP_SCBRA: 245 if (print_lengths) fprintf(f, "%3d ", GET(code, 1)); 246 else fprintf(f, " "); 247 fprintf(f, "%s %d", OP_names[*code], GET2(code, 1+LINK_SIZE)); 248 break; 249 250 case OP_BRA: 251 case OP_SBRA: 252 case OP_KETRMAX: 253 case OP_KETRMIN: 254 case OP_ALT: 255 case OP_KET: 256 case OP_ASSERT: 257 case OP_ASSERT_NOT: 258 case OP_ASSERTBACK: 259 case OP_ASSERTBACK_NOT: 260 case OP_ONCE: 261 case OP_COND: 262 case OP_SCOND: 263 case OP_REVERSE: 264 if (print_lengths) fprintf(f, "%3d ", GET(code, 1)); 265 else fprintf(f, " "); 266 fprintf(f, "%s", OP_names[*code]); 267 break; 268 269 case OP_CLOSE: 270 fprintf(f, " %s %d", OP_names[*code], GET2(code, 1)); 271 break; 272 273 case OP_CREF: 274 case OP_NCREF: 275 fprintf(f, "%3d %s", GET2(code,1), OP_names[*code]); 276 break; 277 278 case OP_RREF: 279 c = GET2(code, 1); 280 if (c == RREF_ANY) 281 fprintf(f, " Cond recurse any"); 282 else 283 fprintf(f, " Cond recurse %d", c); 284 break; 285 286 case OP_NRREF: 287 c = GET2(code, 1); 288 if (c == RREF_ANY) 289 fprintf(f, " Cond nrecurse any"); 290 else 291 fprintf(f, " Cond nrecurse %d", c); 292 break; 293 294 case OP_DEF: 295 fprintf(f, " Cond def"); 296 break; 297 298 case OP_STAR: 299 case OP_MINSTAR: 300 case OP_POSSTAR: 301 case OP_PLUS: 302 case OP_MINPLUS: 303 case OP_POSPLUS: 304 case OP_QUERY: 305 case OP_MINQUERY: 306 case OP_POSQUERY: 307 case OP_TYPESTAR: 308 case OP_TYPEMINSTAR: 309 case OP_TYPEPOSSTAR: 310 case OP_TYPEPLUS: 311 case OP_TYPEMINPLUS: 312 case OP_TYPEPOSPLUS: 313 case OP_TYPEQUERY: 314 case OP_TYPEMINQUERY: 315 case OP_TYPEPOSQUERY: 316 fprintf(f, " "); 317 if (*code >= OP_TYPESTAR) 318 { 319 fprintf(f, "%s", OP_names[code[1]]); 320 if (code[1] == OP_PROP || code[1] == OP_NOTPROP) 321 { 322 fprintf(f, " %s ", get_ucpname(code[2], code[3])); 323 extra = 2; 324 } 325 } 326 else extra = print_char(f, code+1, utf8); 327 fprintf(f, "%s", OP_names[*code]); 328 break; 329 330 case OP_EXACT: 331 case OP_UPTO: 332 case OP_MINUPTO: 333 case OP_POSUPTO: 334 fprintf(f, " "); 335 extra = print_char(f, code+3, utf8); 336 fprintf(f, "{"); 337 if (*code != OP_EXACT) fprintf(f, "0,"); 338 fprintf(f, "%d}", GET2(code,1)); 339 if (*code == OP_MINUPTO) fprintf(f, "?"); 340 else if (*code == OP_POSUPTO) fprintf(f, "+"); 341 break; 342 343 case OP_TYPEEXACT: 344 case OP_TYPEUPTO: 345 case OP_TYPEMINUPTO: 346 case OP_TYPEPOSUPTO: 347 fprintf(f, " %s", OP_names[code[3]]); 348 if (code[3] == OP_PROP || code[3] == OP_NOTPROP) 349 { 350 fprintf(f, " %s ", get_ucpname(code[4], code[5])); 351 extra = 2; 352 } 353 fprintf(f, "{"); 354 if (*code != OP_TYPEEXACT) fprintf(f, "0,"); 355 fprintf(f, "%d}", GET2(code,1)); 356 if (*code == OP_TYPEMINUPTO) fprintf(f, "?"); 357 else if (*code == OP_TYPEPOSUPTO) fprintf(f, "+"); 358 break; 359 360 case OP_NOT: 361 c = code[1]; 362 if (PRINTABLE(c)) fprintf(f, " [^%c]", c); 363 else fprintf(f, " [^\\x%02x]", c); 364 break; 365 366 case OP_NOTSTAR: 367 case OP_NOTMINSTAR: 368 case OP_NOTPOSSTAR: 369 case OP_NOTPLUS: 370 case OP_NOTMINPLUS: 371 case OP_NOTPOSPLUS: 372 case OP_NOTQUERY: 373 case OP_NOTMINQUERY: 374 case OP_NOTPOSQUERY: 375 c = code[1]; 376 if (PRINTABLE(c)) fprintf(f, " [^%c]", c); 377 else fprintf(f, " [^\\x%02x]", c); 378 fprintf(f, "%s", OP_names[*code]); 379 break; 380 381 case OP_NOTEXACT: 382 case OP_NOTUPTO: 383 case OP_NOTMINUPTO: 384 case OP_NOTPOSUPTO: 385 c = code[3]; 386 if (PRINTABLE(c)) fprintf(f, " [^%c]{", c); 387 else fprintf(f, " [^\\x%02x]{", c); 388 if (*code != OP_NOTEXACT) fprintf(f, "0,"); 389 fprintf(f, "%d}", GET2(code,1)); 390 if (*code == OP_NOTMINUPTO) fprintf(f, "?"); 391 else if (*code == OP_NOTPOSUPTO) fprintf(f, "+"); 392 break; 393 394 case OP_RECURSE: 395 if (print_lengths) fprintf(f, "%3d ", GET(code, 1)); 396 else fprintf(f, " "); 397 fprintf(f, "%s", OP_names[*code]); 398 break; 399 400 case OP_REF: 401 fprintf(f, " \\%d", GET2(code,1)); 402 ccode = code + _pcre_OP_lengths[*code]; 403 goto CLASS_REF_REPEAT; 404 405 case OP_CALLOUT: 406 fprintf(f, " %s %d %d %d", OP_names[*code], code[1], GET(code,2), 407 GET(code, 2 + LINK_SIZE)); 408 break; 409 410 case OP_PROP: 411 case OP_NOTPROP: 412 fprintf(f, " %s %s", OP_names[*code], get_ucpname(code[1], code[2])); 413 break; 414 415 /* OP_XCLASS can only occur in UTF-8 mode. However, there's no harm in 416 having this code always here, and it makes it less messy without all those 417 #ifdefs. */ 418 419 case OP_CLASS: 420 case OP_NCLASS: 421 case OP_XCLASS: 422 { 423 int i, min, max; 424 BOOL printmap; 425 426 fprintf(f, " ["); 427 428 if (*code == OP_XCLASS) 429 { 430 extra = GET(code, 1); 431 ccode = code + LINK_SIZE + 1; 432 printmap = (*ccode & XCL_MAP) != 0; 433 if ((*ccode++ & XCL_NOT) != 0) fprintf(f, "^"); 434 } 435 else 436 { 437 printmap = TRUE; 438 ccode = code + 1; 439 } 440 441 /* Print a bit map */ 442 443 if (printmap) 444 { 445 for (i = 0; i < 256; i++) 446 { 447 if ((ccode[i/8] & (1 << (i&7))) != 0) 448 { 449 int j; 450 for (j = i+1; j < 256; j++) 451 if ((ccode[j/8] & (1 << (j&7))) == 0) break; 452 if (i == '-' || i == ']') fprintf(f, "\\"); 453 if (PRINTABLE(i)) fprintf(f, "%c", i); 454 else fprintf(f, "\\x%02x", i); 455 if (--j > i) 456 { 457 if (j != i + 1) fprintf(f, "-"); 458 if (j == '-' || j == ']') fprintf(f, "\\"); 459 if (PRINTABLE(j)) fprintf(f, "%c", j); 460 else fprintf(f, "\\x%02x", j); 461 } 462 i = j; 463 } 464 } 465 ccode += 32; 466 } 467 468 /* For an XCLASS there is always some additional data */ 469 470 if (*code == OP_XCLASS) 471 { 472 int ch; 473 while ((ch = *ccode++) != XCL_END) 474 { 475 if (ch == XCL_PROP) 476 { 477 int ptype = *ccode++; 478 int pvalue = *ccode++; 479 fprintf(f, "\\p{%s}", get_ucpname(ptype, pvalue)); 480 } 481 else if (ch == XCL_NOTPROP) 482 { 483 int ptype = *ccode++; 484 int pvalue = *ccode++; 485 fprintf(f, "\\P{%s}", get_ucpname(ptype, pvalue)); 486 } 487 else 488 { 489 ccode += 1 + print_char(f, ccode, TRUE); 490 if (ch == XCL_RANGE) 491 { 492 fprintf(f, "-"); 493 ccode += 1 + print_char(f, ccode, TRUE); 494 } 495 } 496 } 497 } 498 499 /* Indicate a non-UTF8 class which was created by negation */ 500 501 fprintf(f, "]%s", (*code == OP_NCLASS)? " (neg)" : ""); 502 503 /* Handle repeats after a class or a back reference */ 504 505 CLASS_REF_REPEAT: 506 switch(*ccode) 507 { 508 case OP_CRSTAR: 509 case OP_CRMINSTAR: 510 case OP_CRPLUS: 511 case OP_CRMINPLUS: 512 case OP_CRQUERY: 513 case OP_CRMINQUERY: 514 fprintf(f, "%s", OP_names[*ccode]); 515 extra += _pcre_OP_lengths[*ccode]; 516 break; 517 518 case OP_CRRANGE: 519 case OP_CRMINRANGE: 520 min = GET2(ccode,1); 521 max = GET2(ccode,3); 522 if (max == 0) fprintf(f, "{%d,}", min); 523 else fprintf(f, "{%d,%d}", min, max); 524 if (*ccode == OP_CRMINRANGE) fprintf(f, "?"); 525 extra += _pcre_OP_lengths[*ccode]; 526 break; 527 528 /* Do nothing if it's not a repeat; this code stops picky compilers 529 warning about the lack of a default code path. */ 530 531 default: 532 break; 533 } 534 } 535 break; 536 537 case OP_MARK: 538 case OP_PRUNE_ARG: 539 case OP_SKIP_ARG: 540 fprintf(f, " %s %s", OP_names[*code], code + 2); 541 extra += code[1]; 542 break; 543 544 case OP_THEN: 545 if (print_lengths) 546 fprintf(f, " %s %d", OP_names[*code], GET(code, 1)); 547 else 548 fprintf(f, " %s", OP_names[*code]); 549 break; 550 551 case OP_THEN_ARG: 552 if (print_lengths) 553 fprintf(f, " %s %d %s", OP_names[*code], GET(code, 1), 554 code + 2 + LINK_SIZE); 555 else 556 fprintf(f, " %s %s", OP_names[*code], code + 2 + LINK_SIZE); 557 extra += code[1+LINK_SIZE]; 558 break; 559 560 /* Anything else is just an item with no data*/ 561 562 default: 563 fprintf(f, " %s", OP_names[*code]); 564 break; 565 } 566 567 code += _pcre_OP_lengths[*code] + extra; 568 fprintf(f, "\n"); 569 } 570} 571 572/* End of pcre_printint.src */ 573