Dogcows Code - chaz/tar/blob - src/transform.c

   1 /* This file is part of GNU tar.
   2    Copyright 2006-2008, 2013 Free Software Foundation, Inc.
   3
   4    This program is free software; you can redistribute it and/or modify it
   5    under the terms of the GNU General Public License as published by the
   6    Free Software Foundation; either version 3, or (at your option) any later
   7    version.
   8
   9    This program is distributed in the hope that it will be useful, but
  10    WITHOUT ANY WARRANTY; without even the implied warranty of
  11    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General
  12    Public License for more details.
  13
  14    You should have received a copy of the GNU General Public License along
  15    with this program.  If not, see <http://www.gnu.org/licenses/>.  */
  16
  17 #include <system.h>
  18 #include <regex.h>
  19 #include "common.h"
  20
  21 enum transform_type
  22   {
  23     transform_first,
  24     transform_global
  25   };
  26
  27 enum replace_segm_type
  28   {
  29     segm_literal,   /* Literal segment */
  30     segm_backref,   /* Back-reference segment */
  31     segm_case_ctl   /* Case control segment (GNU extension) */
  32   };
  33
  34 enum case_ctl_type
  35   {
  36     ctl_stop,       /* Stop case conversion */
  37     ctl_upcase_next,/* Turn the next character to uppercase */
  38     ctl_locase_next,/* Turn the next character to lowercase */
  39     ctl_upcase,     /* Turn the replacement to uppercase until ctl_stop */
  40     ctl_locase      /* Turn the replacement to lowercase until ctl_stop */
  41   };
  42
  43 struct replace_segm
  44 {
  45   struct replace_segm *next;
  46   enum replace_segm_type type;
  47   union
  48   {
  49     struct
  50     {
  51       char *ptr;
  52       size_t size;
  53     } literal;                /* type == segm_literal */
  54     size_t ref;               /* type == segm_backref */
  55     enum case_ctl_type ctl;   /* type == segm_case_ctl */
  56   } v;
  57 };
  58
  59 struct transform
  60 {
  61   struct transform *next;
  62   enum transform_type transform_type;
  63   int flags;
  64   unsigned match_number;
  65   regex_t regex;
  66   /* Compiled replacement expression */
  67   struct replace_segm *repl_head, *repl_tail;
  68   size_t segm_count; /* Number of elements in the above list */
  69 };
  70
  71 \f
  72
  73 static int transform_flags = XFORM_ALL;
  74 static struct transform *transform_head, *transform_tail;
  75
  76 static struct transform *
  77 new_transform (void)
  78 {
  79   struct transform *p = xzalloc (sizeof *p);
  80   if (transform_tail)
  81     transform_tail->next = p;
  82   else
  83     transform_head = p;
  84   transform_tail = p;
  85   return p;
  86 }
  87
  88 static struct replace_segm *
  89 add_segment (struct transform *tf)
  90 {
  91   struct replace_segm *segm = xmalloc (sizeof *segm);
  92   segm->next = NULL;
  93   if (tf->repl_tail)
  94     tf->repl_tail->next = segm;
  95   else
  96     tf->repl_head = segm;
  97   tf->repl_tail = segm;
  98   tf->segm_count++;
  99   return segm;
 100 }
 101
 102 static void
 103 add_literal_segment (struct transform *tf, char *str, char *end)
 104 {
 105   size_t len = end - str;
 106   if (len)
 107     {
 108       struct replace_segm *segm = add_segment (tf);
 109       segm->type = segm_literal;
 110       segm->v.literal.ptr = xmalloc (len + 1);
 111       memcpy (segm->v.literal.ptr, str, len);
 112       segm->v.literal.ptr[len] = 0;
 113       segm->v.literal.size = len;
 114     }
 115 }
 116
 117 static void
 118 add_char_segment (struct transform *tf, int chr)
 119 {
 120   struct replace_segm *segm = add_segment (tf);
 121   segm->type = segm_literal;
 122   segm->v.literal.ptr = xmalloc (2);
 123   segm->v.literal.ptr[0] = chr;
 124   segm->v.literal.ptr[1] = 0;
 125   segm->v.literal.size = 1;
 126 }
 127
 128 static void
 129 add_backref_segment (struct transform *tf, size_t ref)
 130 {
 131   struct replace_segm *segm = add_segment (tf);
 132   segm->type = segm_backref;
 133   segm->v.ref = ref;
 134 }
 135
 136 static int
 137 parse_xform_flags (int *pflags, int c)
 138 {
 139   switch (c)
 140     {
 141     case 'r':
 142       *pflags |= XFORM_REGFILE;
 143       break;
 144
 145     case 'R':
 146       *pflags &= ~XFORM_REGFILE;
 147       break;
 148
 149     case 'h':
 150       *pflags |= XFORM_LINK;
 151       break;
 152
 153     case 'H':
 154       *pflags &= ~XFORM_LINK;
 155       break;
 156
 157     case 's':
 158       *pflags |= XFORM_SYMLINK;
 159       break;
 160
 161     case 'S':
 162       *pflags &= ~XFORM_SYMLINK;
 163       break;
 164
 165     default:
 166       return 1;
 167     }
 168   return 0;
 169 }
 170
 171 static void
 172 add_case_ctl_segment (struct transform *tf, enum case_ctl_type ctl)
 173 {
 174   struct replace_segm *segm = add_segment (tf);
 175   segm->type = segm_case_ctl;
 176   segm->v.ctl = ctl;
 177 }
 178
 179 static const char *
 180 parse_transform_expr (const char *expr)
 181 {
 182   int delim;
 183   int i, j, rc;
 184   char *str, *beg, *cur;
 185   const char *p;
 186   int cflags = 0;
 187   struct transform *tf = new_transform ();
 188
 189   if (expr[0] != 's')
 190     {
 191       if (strncmp (expr, "flags=", 6) == 0)
 192         {
 193           transform_flags = 0;
 194           for (expr += 6; *expr; expr++)
 195             {
 196               if (*expr == ';')
 197                 {
 198                   expr++;
 199                   break;
 200                 }
 201               if (parse_xform_flags (&transform_flags, *expr))
 202                 USAGE_ERROR ((0, 0, _("Unknown transform flag: %c"),
 203                               *expr));
 204             }
 205           return expr;
 206         }
 207       USAGE_ERROR ((0, 0, _("Invalid transform expression")));
 208     }
 209
 210   delim = expr[1];
 211
 212   /* Scan regular expression */
 213   for (i = 2; expr[i] && expr[i] != delim; i++)
 214     if (expr[i] == '\\' && expr[i+1])
 215       i++;
 216
 217   if (expr[i] != delim)
 218     USAGE_ERROR ((0, 0, _("Invalid transform expression")));
 219
 220   /* Scan replacement expression */
 221   for (j = i + 1; expr[j] && expr[j] != delim; j++)
 222     if (expr[j] == '\\' && expr[j+1])
 223       j++;
 224
 225   if (expr[j] != delim)
 226     USAGE_ERROR ((0, 0, _("Invalid transform expression")));
 227
 228   /* Check flags */
 229   tf->transform_type = transform_first;
 230   tf->flags = transform_flags;
 231   for (p = expr + j + 1; *p && *p != ';'; p++)
 232     switch (*p)
 233       {
 234       case 'g':
 235         tf->transform_type = transform_global;
 236         break;
 237
 238       case 'i':
 239         cflags |= REG_ICASE;
 240         break;
 241
 242       case 'x':
 243         cflags |= REG_EXTENDED;
 244         break;
 245
 246       case '0': case '1': case '2': case '3': case '4':
 247       case '5': case '6': case '7': case '8': case '9':
 248         tf->match_number = strtoul (p, (char**) &p, 0);
 249         p--;
 250         break;
 251
 252       default:
 253         if (parse_xform_flags (&tf->flags, *p))
 254           USAGE_ERROR ((0, 0, _("Unknown flag in transform expression: %c"),
 255                         *p));
 256       }
 257
 258   if (*p == ';')
 259     p++;
 260
 261   /* Extract and compile regex */
 262   str = xmalloc (i - 1);
 263   memcpy (str, expr + 2, i - 2);
 264   str[i - 2] = 0;
 265
 266   rc = regcomp (&tf->regex, str, cflags);
 267
 268   if (rc)
 269     {
 270       char errbuf[512];
 271       regerror (rc, &tf->regex, errbuf, sizeof (errbuf));
 272       USAGE_ERROR ((0, 0, _("Invalid transform expression: %s"), errbuf));
 273     }
 274
 275   if (str[0] == '^' || str[strlen (str) - 1] == '$')
 276     tf->transform_type = transform_first;
 277
 278   free (str);
 279
 280   /* Extract and compile replacement expr */
 281   i++;
 282   str = xmalloc (j - i + 1);
 283   memcpy (str, expr + i, j - i);
 284   str[j - i] = 0;
 285
 286   for (cur = beg = str; *cur;)
 287     {
 288       if (*cur == '\\')
 289         {
 290           size_t n;
 291
 292           add_literal_segment (tf, beg, cur);
 293           switch (*++cur)
 294             {
 295             case '0': case '1': case '2': case '3': case '4':
 296             case '5': case '6': case '7': case '8': case '9':
 297               n = strtoul (cur, &cur, 10);
 298               if (n > tf->regex.re_nsub)
 299                 USAGE_ERROR ((0, 0, _("Invalid transform replacement: back reference out of range")));
 300               add_backref_segment (tf, n);
 301               break;
 302
 303             case '\\':
 304               add_char_segment (tf, '\\');
 305               cur++;
 306               break;
 307
 308             case 'a':
 309               add_char_segment (tf, '\a');
 310               cur++;
 311               break;
 312
 313             case 'b':
 314               add_char_segment (tf, '\b');
 315               cur++;
 316               break;
 317
 318             case 'f':
 319               add_char_segment (tf, '\f');
 320               cur++;
 321               break;
 322
 323             case 'n':
 324               add_char_segment (tf, '\n');
 325               cur++;
 326               break;
 327
 328             case 'r':
 329               add_char_segment (tf, '\r');
 330               cur++;
 331               break;
 332
 333             case 't':
 334               add_char_segment (tf, '\t');
 335               cur++;
 336               break;
 337
 338             case 'v':
 339               add_char_segment (tf, '\v');
 340               cur++;
 341               break;
 342
 343             case '&':
 344               add_char_segment (tf, '&');
 345               cur++;
 346               break;
 347
 348             case 'L':
 349               /* Turn the replacement to lowercase until a '\U' or '\E'
 350                  is found, */
 351               add_case_ctl_segment (tf, ctl_locase);
 352               cur++;
 353               break;
 354
 355             case 'l':
 356               /* Turn the next character to lowercase, */
 357               add_case_ctl_segment (tf, ctl_locase_next);
 358               cur++;
 359               break;
 360
 361             case 'U':
 362               /* Turn the replacement to uppercase until a '\L' or '\E'
 363                  is found, */
 364               add_case_ctl_segment (tf, ctl_upcase);
 365               cur++;
 366               break;
 367
 368             case 'u':
 369               /* Turn the next character to uppercase, */
 370               add_case_ctl_segment (tf, ctl_upcase_next);
 371               cur++;
 372               break;
 373
 374             case 'E':
 375               /* Stop case conversion started by '\L' or '\U'. */
 376               add_case_ctl_segment (tf, ctl_stop);
 377               cur++;
 378               break;
 379
 380             default:
 381               /* Try to be nice */
 382               {
 383                 char buf[2];
 384                 buf[0] = '\\';
 385                 buf[1] = *cur;
 386                 add_literal_segment (tf, buf, buf + 2);
 387               }
 388               cur++;
 389               break;
 390             }
 391           beg = cur;
 392         }
 393       else if (*cur == '&')
 394         {
 395           add_literal_segment (tf, beg, cur);
 396           add_backref_segment (tf, 0);
 397           beg = ++cur;
 398         }
 399       else
 400         cur++;
 401     }
 402   add_literal_segment (tf, beg, cur);
 403
 404   return p;
 405 }
 406
 407 void
 408 set_transform_expr (const char *expr)
 409 {
 410   while (*expr)
 411     expr = parse_transform_expr (expr);
 412 }
 413
 414 /* Run case conversion specified by CASE_CTL on array PTR of SIZE
 415    characters. Returns pointer to statically allocated storage. */
 416 static char *
 417 run_case_conv (enum case_ctl_type case_ctl, char *ptr, size_t size)
 418 {
 419   static char *case_ctl_buffer;
 420   static size_t case_ctl_bufsize;
 421   char *p;
 422
 423   if (case_ctl_bufsize < size)
 424     {
 425       case_ctl_bufsize = size;
 426       case_ctl_buffer = xrealloc (case_ctl_buffer, case_ctl_bufsize);
 427     }
 428   memcpy (case_ctl_buffer, ptr, size);
 429   switch (case_ctl)
 430     {
 431     case ctl_upcase_next:
 432       case_ctl_buffer[0] = toupper ((unsigned char) case_ctl_buffer[0]);
 433       break;
 434
 435     case ctl_locase_next:
 436       case_ctl_buffer[0] = tolower ((unsigned char) case_ctl_buffer[0]);
 437       break;
 438
 439     case ctl_upcase:
 440       for (p = case_ctl_buffer; p < case_ctl_buffer + size; p++)
 441         *p = toupper ((unsigned char) *p);
 442       break;
 443
 444     case ctl_locase:
 445       for (p = case_ctl_buffer; p < case_ctl_buffer + size; p++)
 446         *p = tolower ((unsigned char) *p);
 447       break;
 448
 449     case ctl_stop:
 450       break;
 451     }
 452   return case_ctl_buffer;
 453 }
 454
 455 \f
 456 static struct obstack stk;
 457 static bool stk_init;
 458
 459 static void
 460 _single_transform_name_to_obstack (struct transform *tf, char *input)
 461 {
 462   regmatch_t *rmp;
 463   int rc;
 464   size_t nmatches = 0;
 465   enum case_ctl_type case_ctl = ctl_stop,  /* Current case conversion op */
 466                      save_ctl = ctl_stop;  /* Saved case_ctl for \u and \l */
 467
 468   /* Reset case conversion after a single-char operation */
 469 #define CASE_CTL_RESET()  if (case_ctl == ctl_upcase_next     \
 470                               || case_ctl == ctl_locase_next) \
 471                             {                                 \
 472                               case_ctl = save_ctl;            \
 473                               save_ctl = ctl_stop;            \
 474                             }
 475
 476   rmp = xmalloc ((tf->regex.re_nsub + 1) * sizeof (*rmp));
 477
 478   while (*input)
 479     {
 480       size_t disp;
 481       char *ptr;
 482
 483       rc = regexec (&tf->regex, input, tf->regex.re_nsub + 1, rmp, 0);
 484
 485       if (rc == 0)
 486         {
 487           struct replace_segm *segm;
 488
 489           disp = rmp[0].rm_eo;
 490
 491           if (rmp[0].rm_so)
 492             obstack_grow (&stk, input, rmp[0].rm_so);
 493
 494           nmatches++;
 495           if (tf->match_number && nmatches < tf->match_number)
 496             {
 497               obstack_grow (&stk, input, disp);
 498               input += disp;
 499               continue;
 500             }
 501
 502           for (segm = tf->repl_head; segm; segm = segm->next)
 503             {
 504               switch (segm->type)
 505                 {
 506                 case segm_literal:    /* Literal segment */
 507                   if (case_ctl == ctl_stop)
 508                     ptr = segm->v.literal.ptr;
 509                   else
 510                     {
 511                       ptr = run_case_conv (case_ctl,
 512                                            segm->v.literal.ptr,
 513                                            segm->v.literal.size);
 514                       CASE_CTL_RESET();
 515                     }
 516                   obstack_grow (&stk, ptr, segm->v.literal.size);
 517                   break;
 518
 519                 case segm_backref:    /* Back-reference segment */
 520                   if (rmp[segm->v.ref].rm_so != -1
 521                       && rmp[segm->v.ref].rm_eo != -1)
 522                     {
 523                       size_t size = rmp[segm->v.ref].rm_eo
 524                                       - rmp[segm->v.ref].rm_so;
 525                       ptr = input + rmp[segm->v.ref].rm_so;
 526                       if (case_ctl != ctl_stop)
 527                         {
 528                           ptr = run_case_conv (case_ctl, ptr, size);
 529                           CASE_CTL_RESET();
 530                         }
 531
 532                       obstack_grow (&stk, ptr, size);
 533                     }
 534                   break;
 535
 536                 case segm_case_ctl:
 537                   switch (segm->v.ctl)
 538                     {
 539                     case ctl_upcase_next:
 540                     case ctl_locase_next:
 541                       switch (save_ctl)
 542                         {
 543                         case ctl_stop:
 544                         case ctl_upcase:
 545                         case ctl_locase:
 546                           save_ctl = case_ctl;
 547                         default:
 548                           break;
 549                         }
 550                       /*FALL THROUGH*/
 551
 552                     case ctl_upcase:
 553                     case ctl_locase:
 554                     case ctl_stop:
 555                       case_ctl = segm->v.ctl;
 556                     }
 557                 }
 558             }
 559         }
 560       else
 561         {
 562           disp = strlen (input);
 563           obstack_grow (&stk, input, disp);
 564         }
 565
 566       input += disp;
 567
 568       if (tf->transform_type == transform_first)
 569         {
 570           obstack_grow (&stk, input, strlen (input));
 571           break;
 572         }
 573     }
 574
 575   obstack_1grow (&stk, 0);
 576   free (rmp);
 577 }
 578
 579 static bool
 580 _transform_name_to_obstack (int flags, char *input, char **output)
 581 {
 582   struct transform *tf;
 583   bool alloced = false;
 584
 585   if (!stk_init)
 586     {
 587       obstack_init (&stk);
 588       stk_init = true;
 589     }
 590
 591   for (tf = transform_head; tf; tf = tf->next)
 592     {
 593       if (tf->flags & flags)
 594         {
 595           _single_transform_name_to_obstack (tf, input);
 596           input = obstack_finish (&stk);
 597           alloced = true;
 598         }
 599     }
 600   *output = input;
 601   return alloced;
 602 }
 603
 604 bool
 605 transform_name_fp (char **pinput, int flags,
 606                    char *(*fun)(char *, void *), void *dat)
 607 {
 608     char *str;
 609     bool ret = _transform_name_to_obstack (flags, *pinput, &str);
 610     if (ret)
 611       {
 612         assign_string (pinput, fun ? fun (str, dat) : str);
 613         obstack_free (&stk, str);
 614       }
 615     else if (fun)
 616       {
 617         *pinput = NULL;
 618         assign_string (pinput, fun (str, dat));
 619         free (str);
 620         ret = true;
 621       }
 622     return ret;
 623 }
 624
 625 bool
 626 transform_name (char **pinput, int type)
 627 {
 628   return transform_name_fp (pinput, type, NULL, NULL);
 629 }
 630
 631 bool
 632 transform_program_p (void)
 633 {
 634   return transform_head != NULL;
 635 }