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

   1 /* This file is part of GNU tar.
   2    Copyright (C) 2006 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 2, 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, write to the Free Software Foundation, Inc.,
  16    51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.  */
  17
  18 #include <system.h>
  19 #include <regex.h>
  20 #include "common.h"
  21
  22 enum transform_type
  23   {
  24     transform_none,
  25     transform_first,
  26     transform_global
  27   }
  28 transform_type = transform_none;
  29 static unsigned match_number = 0;
  30 static regex_t regex;
  31 static struct obstack stk;
  32
  33 enum replace_segm_type
  34   {
  35     segm_literal,   /* Literal segment */
  36     segm_backref,   /* Back-reference segment */
  37     segm_case_ctl   /* Case control segment (GNU extension) */
  38   };
  39
  40 enum case_ctl_type
  41   {
  42     ctl_stop,       /* Stop case conversion */
  43     ctl_upcase_next,/* Turn the next character to uppercase */
  44     ctl_locase_next,/* Turn the next character to lowercase */
  45     ctl_upcase,     /* Turn the replacement to uppercase until ctl_stop */
  46     ctl_locase      /* Turn the replacement to lowercase until ctl_stop */
  47   };
  48
  49 struct replace_segm
  50 {
  51   struct replace_segm *next;
  52   enum replace_segm_type type;
  53   union
  54   {
  55     struct
  56     {
  57       char *ptr;
  58       size_t size;
  59     } literal;                /* type == segm_literal */
  60     size_t ref;               /* type == segm_backref */
  61     enum case_ctl_type ctl;   /* type == segm_case_ctl */
  62   } v;
  63 };
  64
  65 /* Compiled replacement expression */
  66 static struct replace_segm *repl_head, *repl_tail;
  67 static segm_count; /* Number of elements in the above list */
  68
  69 static struct replace_segm *
  70 add_segment (void)
  71 {
  72   struct replace_segm *segm = xmalloc (sizeof *segm);
  73   segm->next = NULL;
  74   if (repl_tail)
  75     repl_tail->next = segm;
  76   else
  77     repl_head = segm;
  78   repl_tail = segm;
  79   segm_count++;
  80   return segm;
  81 }
  82
  83 static void
  84 add_literal_segment (char *str, char *end)
  85 {
  86   size_t len = end - str;
  87   if (len)
  88     {
  89       struct replace_segm *segm = add_segment ();
  90       segm->type = segm_literal;
  91       segm->v.literal.ptr = xmalloc (len + 1);
  92       memcpy (segm->v.literal.ptr, str, len);
  93       segm->v.literal.ptr[len] = 0;
  94       segm->v.literal.size = len;
  95     }
  96 }
  97
  98 static void
  99 add_char_segment (int chr)
 100 {
 101   struct replace_segm *segm = add_segment ();
 102   segm->type = segm_literal;
 103   segm->v.literal.ptr = xmalloc (2);
 104   segm->v.literal.ptr[0] = chr;
 105   segm->v.literal.ptr[1] = 0;
 106   segm->v.literal.size = 1;
 107 }
 108
 109 static void
 110 add_backref_segment (size_t ref)
 111 {
 112   struct replace_segm *segm = add_segment ();
 113   segm->type = segm_backref;
 114   segm->v.ref = ref;
 115 }
 116
 117 static void
 118 add_case_ctl_segment (enum case_ctl_type ctl)
 119 {
 120   struct replace_segm *segm = add_segment ();
 121   segm->type = segm_case_ctl;
 122   segm->v.ctl = ctl;
 123 }
 124
 125 void
 126 set_transform_expr (const char *expr)
 127 {
 128   int delim;
 129   int i, j, rc;
 130   char *str, *beg, *cur;
 131   const char *p;
 132   int cflags = 0;
 133
 134   if (transform_type == transform_none)
 135     obstack_init (&stk);
 136   else
 137     {
 138       /* Redefinition of the transform expression */
 139       regfree (&regex);
 140     }
 141
 142   if (expr[0] != 's')
 143     USAGE_ERROR ((0, 0, _("Invalid transform expression")));
 144
 145   delim = expr[1];
 146
 147   /* Scan regular expression */
 148   for (i = 2; expr[i] && expr[i] != delim; i++)
 149     if (expr[i] == '\\' && expr[i+1])
 150       i++;
 151
 152   if (expr[i] != delim)
 153     USAGE_ERROR ((0, 0, _("Invalid transform expression")));
 154
 155   /* Scan replacement expression */
 156   for (j = i + 1; expr[j] && expr[j] != delim; j++)
 157     if (expr[j] == '\\' && expr[j+1])
 158       j++;
 159
 160   if (expr[j] != delim)
 161     USAGE_ERROR ((0, 0, _("Invalid transform expression")));
 162
 163   /* Check flags */
 164   transform_type = transform_first;
 165   for (p = expr + j + 1; *p; p++)
 166     switch (*p)
 167       {
 168       case 'g':
 169         transform_type = transform_global;
 170         break;
 171
 172       case 'i':
 173         cflags |= REG_ICASE;
 174         break;
 175
 176       case 'x':
 177         cflags |= REG_EXTENDED;
 178         break;
 179
 180       case '0': case '1': case '2': case '3': case '4':
 181       case '5': case '6': case '7': case '8': case '9':
 182         match_number = strtoul (p, (char**) &p, 0);
 183         p--;
 184         break;
 185
 186       default:
 187         USAGE_ERROR ((0, 0, _("Unknown flag in transform expression")));
 188       }
 189
 190   /* Extract and compile regex */
 191   str = xmalloc (i - 1);
 192   memcpy (str, expr + 2, i - 2);
 193   str[i - 2] = 0;
 194
 195   rc = regcomp (&regex, str, cflags);
 196
 197   if (rc)
 198     {
 199       char errbuf[512];
 200       regerror (rc, &regex, errbuf, sizeof (errbuf));
 201       USAGE_ERROR ((0, 0, _("Invalid transform expression: %s"), errbuf));
 202     }
 203
 204   if (str[0] == '^' || str[strlen (str) - 1] == '$')
 205     transform_type = transform_first;
 206
 207   free (str);
 208
 209   /* Extract and compile replacement expr */
 210   i++;
 211   str = xmalloc (j - i + 1);
 212   memcpy (str, expr + i, j - i);
 213   str[j - i] = 0;
 214
 215   for (cur = beg = str; *cur;)
 216     {
 217       if (*cur == '\\')
 218         {
 219           size_t n;
 220
 221           add_literal_segment (beg, cur);
 222           switch (*++cur)
 223             {
 224             case '0': case '1': case '2': case '3': case '4':
 225             case '5': case '6': case '7': case '8': case '9':
 226               n = strtoul (cur, &cur, 10);
 227               if (n > regex.re_nsub)
 228                 USAGE_ERROR ((0, 0, _("Invalid transform replacement: back reference out of range")));
 229               add_backref_segment (n);
 230               break;
 231
 232             case '\\':
 233               add_char_segment ('\\');
 234               cur++;
 235               break;
 236
 237             case 'a':
 238               add_char_segment ('\a');
 239               cur++;
 240               break;
 241
 242             case 'b':
 243               add_char_segment ('\b');
 244               cur++;
 245               break;
 246
 247             case 'f':
 248               add_char_segment ('\f');
 249               cur++;
 250               break;
 251
 252             case 'n':
 253               add_char_segment ('\n');
 254               cur++;
 255               break;
 256
 257             case 'r':
 258               add_char_segment ('\r');
 259               cur++;
 260               break;
 261
 262             case 't':
 263               add_char_segment ('\t');
 264               cur++;
 265               break;
 266
 267             case 'v':
 268               add_char_segment ('\v');
 269               cur++;
 270               break;
 271
 272             case '&':
 273               add_char_segment ('&');
 274               cur++;
 275               break;
 276
 277             case 'L':
 278               /* Turn the replacement to lowercase until a `\U' or `\E'
 279                  is found, */
 280               add_case_ctl_segment (ctl_locase);
 281               cur++;
 282               break;
 283
 284             case 'l':
 285               /* Turn the next character to lowercase, */
 286               add_case_ctl_segment (ctl_locase_next);
 287               cur++;
 288               break;
 289
 290             case 'U':
 291               /* Turn the replacement to uppercase until a `\L' or `\E'
 292                  is found, */
 293               add_case_ctl_segment (ctl_upcase);
 294               cur++;
 295               break;
 296
 297             case 'u':
 298               /* Turn the next character to uppercase, */
 299               add_case_ctl_segment (ctl_upcase_next);
 300               cur++;
 301               break;
 302
 303             case 'E':
 304               /* Stop case conversion started by `\L' or `\U'. */
 305               add_case_ctl_segment (ctl_stop);
 306               cur++;
 307               break;
 308
 309             default:
 310               /* Try to be nice */
 311               {
 312                 char buf[2];
 313                 buf[0] = '\\';
 314                 buf[1] = *cur;
 315                 add_literal_segment (buf, buf + 2);
 316               }
 317               cur++;
 318               break;
 319             }
 320           beg = cur;
 321         }
 322       else if (*cur == '&')
 323         {
 324           add_literal_segment (beg, cur);
 325           add_backref_segment (0);
 326           beg = ++cur;
 327         }
 328       else
 329         cur++;
 330     }
 331   add_literal_segment (beg, cur);
 332
 333 }
 334
 335 /* Run case conversion specified by CASE_CTL on array PTR of SIZE
 336    characters. Returns pointer to statically allocated storage. */
 337 static char *
 338 run_case_conv (enum case_ctl_type case_ctl, char *ptr, size_t size)
 339 {
 340   static char *case_ctl_buffer;
 341   static size_t case_ctl_bufsize;
 342   char *p;
 343
 344   if (case_ctl_bufsize < size)
 345     {
 346       case_ctl_bufsize = size;
 347       case_ctl_buffer = xrealloc (case_ctl_buffer, case_ctl_bufsize);
 348     }
 349   memcpy (case_ctl_buffer, ptr, size);
 350   switch (case_ctl)
 351     {
 352     case ctl_upcase_next:
 353       case_ctl_buffer[0] = toupper (case_ctl_buffer[0]);
 354       break;
 355
 356     case ctl_locase_next:
 357       case_ctl_buffer[0] = tolower (case_ctl_buffer[0]);
 358       break;
 359
 360     case ctl_upcase:
 361       for (p = case_ctl_buffer; p < case_ctl_buffer + size; p++)
 362         *p = toupper (*p);
 363       break;
 364
 365     case ctl_locase:
 366       for (p = case_ctl_buffer; p < case_ctl_buffer + size; p++)
 367         *p = tolower (*p);
 368       break;
 369
 370     case ctl_stop:
 371       break;
 372     }
 373   return case_ctl_buffer;
 374 }
 375
 376 bool
 377 _transform_name_to_obstack (char *input)
 378 {
 379   regmatch_t *rmp;
 380   char *p;
 381   int rc;
 382   size_t nmatches = 0;
 383   enum case_ctl_type case_ctl = ctl_stop,  /* Current case conversion op */
 384                      save_ctl = ctl_stop;  /* Saved case_ctl for \u and \l */
 385
 386   /* Reset case conversion after a single-char operation */
 387 #define CASE_CTL_RESET()  if (case_ctl == ctl_upcase_next     \
 388                               || case_ctl == ctl_locase_next) \
 389                             {                                 \
 390                               case_ctl = save_ctl;            \
 391                               save_ctl = ctl_stop;            \
 392                             }
 393
 394   if (transform_type == transform_none)
 395     return false;
 396
 397   rmp = xmalloc ((regex.re_nsub + 1) * sizeof (*rmp));
 398
 399   while (*input)
 400     {
 401       size_t disp;
 402       char *ptr;
 403
 404       rc = regexec (&regex, input, regex.re_nsub + 1, rmp, 0);
 405
 406       if (rc == 0)
 407         {
 408           struct replace_segm *segm;
 409
 410           disp = rmp[0].rm_eo;
 411
 412           if (rmp[0].rm_so)
 413             obstack_grow (&stk, input, rmp[0].rm_so);
 414
 415           nmatches++;
 416           if (match_number && nmatches < match_number)
 417             {
 418               obstack_grow (&stk, input, disp);
 419               input += disp;
 420               continue;
 421             }
 422
 423           for (segm = repl_head; segm; segm = segm->next)
 424             {
 425               switch (segm->type)
 426                 {
 427                 case segm_literal:    /* Literal segment */
 428                   if (case_ctl == ctl_stop)
 429                     ptr = segm->v.literal.ptr;
 430                   else
 431                     {
 432                       ptr = run_case_conv (case_ctl,
 433                                            segm->v.literal.ptr,
 434                                            segm->v.literal.size);
 435                       CASE_CTL_RESET();
 436                     }
 437                   obstack_grow (&stk, ptr, segm->v.literal.size);
 438                   break;
 439
 440                 case segm_backref:    /* Back-reference segment */
 441                   if (rmp[segm->v.ref].rm_so != -1
 442                       && rmp[segm->v.ref].rm_eo != -1)
 443                     {
 444                       size_t size = rmp[segm->v.ref].rm_eo
 445                                       - rmp[segm->v.ref].rm_so;
 446                       ptr = input + rmp[segm->v.ref].rm_so;
 447                       if (case_ctl != ctl_stop)
 448                         {
 449                           ptr = run_case_conv (case_ctl, ptr, size);
 450                           CASE_CTL_RESET();
 451                         }
 452
 453                       obstack_grow (&stk, ptr, size);
 454                     }
 455                   break;
 456
 457                 case segm_case_ctl:
 458                   switch (segm->v.ctl)
 459                     {
 460                     case ctl_upcase_next:
 461                     case ctl_locase_next:
 462                       switch (save_ctl)
 463                         {
 464                         case ctl_stop:
 465                         case ctl_upcase:
 466                         case ctl_locase:
 467                           save_ctl = case_ctl;
 468                         default:
 469                           break;
 470                         }
 471                       /*FALL THROUGH*/
 472
 473                     case ctl_upcase:
 474                     case ctl_locase:
 475                     case ctl_stop:
 476                       case_ctl = segm->v.ctl;
 477                     }
 478                 }
 479             }
 480         }
 481       else
 482         {
 483           disp = strlen (input);
 484           obstack_grow (&stk, input, disp);
 485         }
 486
 487       input += disp;
 488
 489       if (transform_type == transform_first)
 490         {
 491           obstack_grow (&stk, input, strlen (input));
 492           break;
 493         }
 494     }
 495
 496   obstack_1grow (&stk, 0);
 497   free (rmp);
 498   return true;
 499 }
 500
 501 bool
 502 transform_name_fp (char **pinput, char *(*fun)(char *))
 503 {
 504     char *str, *p;
 505     bool ret = _transform_name_to_obstack (*pinput);
 506     if (ret)
 507       {
 508         str = obstack_finish (&stk);
 509         assign_string (pinput, fun ? fun (str) : str);
 510         obstack_free (&stk, str);
 511       }
 512     else if (fun)
 513       {
 514         str = *pinput;
 515         *pinput = NULL;
 516         assign_string (pinput, fun (str));
 517         free (str);
 518         ret = true;
 519       }
 520     return ret;
 521 }
 522
 523 bool
 524 transform_name (char **pinput)
 525 {
 526   return transform_name_fp (pinput, NULL);
 527 }
 528
 529 #if 0
 530 void
 531 read_and_transform_loop ()
 532 {
 533   char buf[512];
 534   while (fgets (buf, sizeof buf, stdin))
 535     {
 536       char *p = buf + strlen (buf);
 537       if (p[-1] == '\n')
 538         p[-1] = 0;
 539       if (transform_name (buf, &p))
 540         printf ("=> %s\n", p);
 541       else
 542         printf ("=\n");
 543     }
 544 }
 545 #endif