hexdump.c

   1 /* depend: */
   2 /* cflags: */
   3 /* linker: debug.o */
   4
   5 #include <assert.h>
   6 #include <limits.h>
   7 #include <malloc.h>
   8 #include <stdio.h>
   9 #include <stdlib.h>
  10 #include <string.h>
  11
  12 #include "debug.h"
  13
  14 /* macros */
  15
  16 #define CEIL(x, y) (((x) + (y) - 1) / (y))
  17 #define MIN(x, y) (((x) < (y)) ? (x) : (y))
  18 #define MAX(x, y) (((x) > (y)) ? (x) : (y))
  19
  20 //#define BUFFERSIZE 4096
  21 #define BUFFERSIZE 256
  22 #define NBCOLS 8
  23 #define NBDIGITS 6
  24 #define SEQLEN 32
  25
  26 /* gobal variables */
  27
  28 int nbcols = NBCOLS;
  29 int nbdigits = NBDIGITS;
  30 int offset = 0;
  31
  32 char buffer[BUFFERSIZE] = {0};
  33 FILE *fin = NULL;
  34 long int addrfile = 0;
  35 FILE *fout = NULL;
  36 char *progname = NULL;
  37
  38 /* type definitions */
  39
  40 typedef struct {
  41     char *sequence;
  42     char bytes[SEQLEN];
  43     int length;
  44 } sequence_t;
  45
  46 /* help function */
  47
  48 int usage (int ret)
  49 {
  50     FILE *fd = ret ? stderr : stdout;
  51     fprintf (fd, "usage: %s [-i file] [-h] [-n nbcols] [-o file] [-v]\n", progname);
  52     fprintf (fd, " -i: input file\n");
  53     fprintf (fd, " -h: help message\n");
  54     fprintf (fd, " -n: number of columns\n");
  55     fprintf (fd, " -e: commands\n");
  56     fprintf (fd, " -o: output file\n");
  57     fprintf (fd, " -v: verbose level (%d)\n", verbose);
  58     fprintf (fd, "\n");
  59     fprintf (fd, "commands: [/hstr/|addr|+nb] [a hstr] [d nb|-] [i hstr] [p nb|-] [s/h1/h2/[g]]\n");
  60     fprintf (fd, " addr: move to address (0... octal, [1-9]... deci, 0x... hexa)\n");
  61     fprintf (fd, " +nb: move to offset (0... octal, [1-9]... deci, 0x... hexa)\n");
  62     fprintf (fd, " //: move to hexa stringi hstr\n");
  63     fprintf (fd, " a : append hexa string hstr to current address\n");
  64     fprintf (fd, " d : delete nb bytes (- until end file)\n");
  65     fprintf (fd, " i : insert hexa string hstr to current address\n");
  66     fprintf (fd, " p : print nb bytes (- until end file)\n");
  67     fprintf (fd, " s : substitute h1 by h2 (g for globally)\n");
  68
  69     return ret;
  70 }
  71
  72 /* get number of digits */
  73
  74 unsigned int getnbdigits (unsigned long int l) {
  75     int n = 0;
  76     while (l) {
  77         n += 2;
  78         l /= 256;
  79     }
  80     return n;
  81 }
  82
  83 /* print a line */
  84
  85 void printline (char *buffer, int nb, int addr) {
  86     int i;
  87
  88     printf ("0x%0*x:", nbdigits, addr);
  89     for (i = 0; i < nb; i++) {
  90         printf (" %02x", buffer[i]);
  91     }
  92     for (i = nb; i < nbcols; i++) {
  93         printf ("   ");
  94     }
  95     printf ("  ");
  96     for (i = 0; i < nb; i++) {
  97         char c = buffer[i];
  98         printf ("%c", (c > 31) && (c < 127) ? c : '.');
  99     }
 100     printf ("\n");
 101 }
 102
 103 /* write file function */
 104
 105 int writefile (char *pt, int nb) {
 106     if (fout) {
 107         fwrite (pt, 1, nb, fout);
 108     }
 109     return 1;
 110 }
 111
 112 /* search sequence function */
 113
 114 int searchseq (sequence_t *seq) {
 115     char *pt = buffer;
 116     int nb = 0;
 117     int i, j;
 118     int valid = 0;
 119
 120     VERBOSE (DEBUG, printf ("search sequence: %s\n", seq->sequence));
 121
 122     while (!feof (fin)) {
 123         int nbread = fread (pt, 1, BUFFERSIZE - (pt - buffer), fin);
 124         nb += nbread;
 125         pt = buffer;
 126         for (i = 0; i < nb - seq->length; i++) {
 127             valid = 1;
 128             for (j = 0; (j < seq->length) && (valid); j++) {
 129                 if (pt[i + j] != seq->bytes[j]) {
 130                     valid = 0;
 131                 }
 132             }
 133             if (valid) {
 134                 break;
 135             }
 136         }
 137
 138         if (!valid) {
 139             writefile (buffer, nb - seq->length);
 140             offset = 0;
 141             addrfile += nb - seq->length;
 142             for (i = 0; i < seq->length; i++) {
 143                 buffer[i] = buffer[nb - seq->length + i];
 144             }
 145             pt = buffer + seq->length;
 146             nb = seq->length;
 147         } else {
 148             writefile (buffer, i);
 149             offset = seq->length;
 150             addrfile += i;
 151             fseek (fin, i - nb, SEEK_CUR);
 152             VERBOSE (DEBUG, printf ("found sequence (%d)\n", i - nb));
 153             return 0;
 154         }
 155     }
 156
 157     if (!valid) {
 158         writefile (buffer, nb);
 159         addrfile += seq->length;
 160     }
 161
 162     return 1;
 163 }
 164
 165 /* go to address function */
 166
 167 int gotoaddr (long int addr) {
 168     char buffer[BUFFERSIZE] = {0};
 169
 170     if (addr == -1) {
 171         addr = LONG_MAX;
 172     } else if (addrfile > addr) {
 173         return 1;
 174     }
 175
 176     VERBOSE (DEBUG, printf ("look for address: 0x%04lx\n", addr));
 177     while (!feof (fin)) {
 178         int nbtoread = (addrfile + BUFFERSIZE > addr) ? addr - addrfile : BUFFERSIZE;
 179         int nbread = fread (buffer, 1, nbtoread, fin);
 180         writefile (buffer, nbread);
 181         addrfile += nbread;
 182         if (addrfile == addr) {
 183             return 0;
 184         }
 185     }
 186
 187     return 1;
 188 }
 189
 190 /* insert sequence function */
 191
 192 int insertseq (sequence_t *seq) {
 193     char buffer[BUFFERSIZE] = {0};
 194
 195     VERBOSE (DEBUG, printf ("insert (%d): '%s'\n", offset, seq->sequence);
 196                     int i;
 197                     for (i = 0; i < seq->length; i++) {
 198                         char c = seq->bytes[i];
 199                         printf (" 0x%02x (%c)", c, ((c >= 32) && (c < 127)) ? c : '.');
 200                     };
 201                     printf ("\n"));
 202     if (offset > 0) {
 203         int nbread = fread (buffer, 1, offset, fin);
 204         if (nbread != offset) {
 205             return 1;
 206         }
 207         writefile (buffer, offset);
 208         offset = 0;
 209     }
 210     writefile (seq->bytes, seq->length);
 211
 212     return 0;
 213 }
 214
 215 /* hexadecimal dump function */
 216
 217 int hexdump (int len) {
 218     char buffer[BUFFERSIZE] = {0};
 219     int i;
 220
 221     char *pt = buffer;
 222
 223     int nb = 0;
 224     while (!feof (fin)) {
 225         int nbtoread = BUFFERSIZE - (pt - buffer);
 226         if ((len > 0) && (nbtoread > len)) {
 227             nbtoread = len;
 228         }
 229         int nbread = fread (pt, 1, nbtoread, fin);
 230         if (len > 0) {
 231             len -= nbread;
 232         }
 233         nb += nbread;
 234         pt = buffer;
 235
 236         /* print line */
 237         while ((nb - (int)(pt - buffer)) / nbcols > 0) {
 238             printline (pt, nbcols, addrfile);
 239             writefile (pt, nbcols);
 240             addrfile += nbcols;
 241             pt += nbcols;
 242         }
 243
 244         /* copy end buffer */
 245         nb -= pt - buffer;
 246         for (i = 0; i < nb; i++) {
 247             buffer[i] = pt[i];
 248         }
 249         pt = buffer + nb;
 250
 251         /* end partial reading */
 252         if (len == 0) {
 253             break;
 254         }
 255     }
 256
 257     /* last line */
 258     if (nb > 0) {
 259         printline (buffer, nb, addrfile);
 260         writefile (buffer, nb);
 261         addrfile += nb;
 262     }
 263
 264     return 0;
 265 }
 266
 267 /* parse octal string */
 268
 269 long int octal (char *s, int n) {
 270     int i;
 271     unsigned long int l = 0;
 272     for (i = 0; i < n; i++) {
 273         if ((s[i] >= '0') && (s[i] <= '9')) {
 274             l = l * 8 + s[i] - '0';
 275         } else {
 276             return -1;
 277         }
 278     }
 279     return l;
 280 }
 281
 282 /* parse hexa string */
 283
 284 long int hexa (char *s, int n) {
 285     int i;
 286     unsigned long int l = 0;
 287     for (i = 0; i < n; i++) {
 288         l *= 16;
 289         if ((s[i] >= '0') && (s[i] <= '9')) {
 290             l += s[i] - '0';
 291         } else if ((s[i] >= 'A') && (s[i] <= 'F')) {
 292             l += s[i] + 10 - 'A';
 293         } else if ((s[i] >= 'a') && (s[i] <= 'f')) {
 294             l += s[i] + 10 - 'a';
 295         } else {
 296             return -1;
 297         }
 298     }
 299     return l;
 300 }
 301
 302 /* special character function */
 303
 304 int specialchar (char *s, char *b) {
 305     int i = 0, j = 0;
 306     while (s[i] != 0) {
 307         if (j == SEQLEN) {
 308             return 0;
 309         }
 310         if (s[i] != '\\') {
 311             b[j++] = s[i++];
 312             continue;
 313         }
 314
 315         int l = -1;
 316         switch (s[i + 1]) {
 317         case 'a': l = 0x07; i += 2; break;
 318         case 'b': l = 0x08; i += 2; break;
 319         case 'e': l = 0x1b; i += 2; break;
 320         case 'f': l = 0x0c; i += 2; break;
 321         case 'n': l = 0x0a; i += 2; break;
 322         case 'r': l = 0x0d; i += 2; break;
 323         case 't': l = 0x09; i += 2; break;
 324         case 'v': l = 0x0b; i += 2; break;
 325         case '/': l = '/'; i += 2; break;
 326         case '\\': l = '\\'; i += 2; break;
 327         case '\'': l = '\''; i += 2; break;
 328         case '"': l = '"'; i += 2; break;
 329         case '0':
 330         case '1':
 331         case '2':
 332         case '3':
 333             l = octal (s + i + 1, 3);
 334             if (l != -1) {
 335                 i += 4;
 336             }
 337             break;
 338         case 'x':
 339             l = hexa (s + i + 2, 2);
 340             if (l != -1) {
 341                 i += 4;
 342             }
 343             break;
 344         default:
 345             break;
 346         }
 347         if (l != -1) {
 348             VERBOSE (DEBUG, printf("l: 0x%02x '%c'\n", l, l));
 349         }
 350         b[j++] = (l != -1) ? l : s[i++];
 351     }
 352
 353     return j;
 354 }
 355
 356 /* main function */
 357
 358 int main (int argc, char *argv[])
 359 {
 360     int i, rc = 0;
 361     char *input = NULL;
 362     char *output = NULL;
 363     char *commands = NULL;
 364     long int length = -1;
 365     sequence_t seq = {0};
 366     long int addr = 0;
 367     char c;
 368
 369     /* get basename */
 370     char *pt = progname = argv[0];
 371     while (*pt) {
 372         if ((*pt == '/') || (*pt == '\\')) {
 373            progname = pt + 1;
 374         }
 375         pt++;
 376     }
 377
 378      while (argc-- > 1) {
 379         char *arg = *(++argv);
 380         if (arg[0] != '-') {
 381             VERBOSE (ERROR, fprintf (stderr, "%s: invalid option -- %s\n", progname, arg));
 382             return usage (1);
 383         }
 384         char c = arg[1];
 385         switch (c) {
 386         case 'e':
 387             arg = (arg[2]) ? arg + 2 : (--argc > 0) ? *(++argv) : NULL;
 388             if (arg) {
 389                 if (commands == NULL) {
 390                     commands = arg;
 391                 } else {
 392                     strcat (commands, " ");
 393                     strcat (commands, arg);
 394                 }
 395             }
 396             break;
 397         case 'i':
 398             input = (arg[2]) ? arg + 2 : (--argc > 0 ) ? *(++argv) : NULL;
 399             break;
 400         case 'n':
 401             arg = (arg[2]) ? arg + 2 : (--argc > 0) ? *(++argv) : NULL;
 402             if (arg == NULL) {
 403                 VERBOSE (ERROR, fprintf (stderr, "%s: missing number of columns\n", progname));
 404                 return usage (1);
 405             }
 406             nbcols = atoi (arg);
 407             break;
 408         case 'o':
 409             output = (arg[2]) ? arg + 2 : (--argc > 0 ) ? *(++argv) : NULL;
 410             break;
 411         case 'v':
 412             arg = (arg[2]) ? arg + 2 : (--argc > 0) ? *(++argv) : NULL;
 413             if (arg == NULL) {
 414                 VERBOSE (ERROR, fprintf (stderr, "%s: missing verbose level\n", progname));
 415                 return usage (1);
 416             }
 417             verbose = atoi (arg);
 418             break;
 419         case 'h':
 420         default:
 421             return usage (c != 'h');
 422         }
 423     }
 424
 425     /* check input */
 426     if (input) {
 427         fin = fopen (input, "rb");
 428         if (!fin) {
 429             VERBOSE (ERROR, fprintf (stderr, "error: can't open file '%s'\n", input));
 430             return 1;
 431         }
 432     } else {
 433         fin = stdin;
 434     }
 435
 436     /* check output */
 437     if (output) {
 438         fout = fopen (output, "wb");
 439         if (!fout) {
 440             VERBOSE (ERROR, fprintf (stderr, "error: can't open file '%s'\n", output));
 441             fclose (fin);
 442             return 1;
 443         }
 444     } else {
 445         //fout = stdout;
 446     }
 447
 448     /* get file size */
 449     if (fin != stdin) {
 450         fseek (fin, 0 , SEEK_END);
 451         unsigned long int filesize = ftell (fin);
 452         fseek (fin, 0 , SEEK_SET);
 453         nbdigits = getnbdigits (filesize);
 454     }
 455
 456     if (commands == NULL) {
 457         VERBOSE (DEBUG, printf ("no command\n"));
 458         hexdump (-1);
 459     } else {
 460         VERBOSE (DEBUG, printf ("commands: %s\n", commands));
 461         while ((*commands != '\0') && (rc == 0)) {
 462             switch (c = *commands++) {
 463             case ' ':
 464             case '\t':
 465                 break;
 466
 467             case '/': /* read patern */
 468                 seq.sequence = commands;
 469                 seq.length = 0;
 470                 while (*commands) {
 471                     if ((*commands == '\\') &&
 472                         ((commands[1] == '/') || (commands[1] == '\\'))) {
 473                         commands++;
 474                     } else if (*commands == '/') {
 475                         *commands++ = 0;
 476                         break;
 477                     }
 478                     commands++;
 479                 }
 480                 seq.length = specialchar (seq.sequence, seq.bytes);
 481                 if (seq.length != 0) {
 482                     rc = searchseq (&seq);
 483                 } else {
 484                     VERBOSE (ERROR, fprintf (stderr, "incorrect sequence (%s)\n", seq.sequence));
 485                     rc = 1;
 486                 }
 487                 break;
 488
 489             case '0': /* read address */
 490                 if (*commands == 'x') {
 491                     commands++;
 492                     addr = strtol (commands, &commands, 16);
 493                 } else {
 494                     addr = strtol (commands, &commands, 8);
 495                 }
 496                 if (addr) {
 497                     rc = gotoaddr (addr);
 498                 } else {
 499                     VERBOSE (ERROR, fprintf (stderr, "erroneous address\n"));
 500                     rc = 1;
 501                 }
 502                 offset = 0;
 503                 break;
 504
 505             case '1':
 506             case '2':
 507             case '3':
 508             case '4':
 509             case '5':
 510             case '6':
 511             case '7':
 512             case '8':
 513             case '9': /* read address */
 514                 commands--;
 515                 addr = strtol (commands, &commands, 10);
 516                 rc = gotoaddr (addr);
 517                 offset = 0;
 518                 break;
 519
 520             case 'a': /* append mode */
 521                 offset = 0;
 522                 /* fall through */
 523
 524             case 'i': /* insert mode */
 525                 while (*commands) {
 526                     if ((*commands == ' ') || (*commands == '\t')) {
 527                         commands++;
 528                     } else {
 529                         break;
 530                     }
 531                 }
 532                 seq.sequence = commands;
 533                 seq.length = 0;
 534                 i = 0;
 535                 while (*commands) {
 536                     if ((*commands == ' ') || (*commands == '\t')) {
 537                        *commands++ = '\0';
 538                        break;
 539                     } else {
 540                         commands++;
 541                         i++;
 542                         if (i % 2 == 0) {
 543                             seq.bytes[seq.length] = hexa (seq.sequence + 2 * seq.length, 2);
 544                             if (seq.bytes[seq.length] == -1) {
 545                                 rc = 1;
 546                                 break;
 547                             }
 548                             seq.length++;
 549                         }
 550                     }
 551                 }
 552                 if ((seq.length > 0) && (rc == 0) && (i % 2 == 0)) {
 553                     rc = insertseq (&seq);
 554                 } else {
 555                     VERBOSE (ERROR, fprintf (stderr, "erroneous sequence '%s'\n", seq.sequence));
 556                     rc = 1;
 557                 }
 558                 offset = 0;
 559                 break;
 560
 561             case '+': /* relative move */
 562                 /* fall through */
 563
 564             case 'd': /* delete mode */
 565                 /* fall through */
 566
 567             case 'p': /* print mode */
 568                 length = -1;
 569                 while (*commands != '\0') {
 570                     if ((*commands == ' ') || (*commands == '\t')) {
 571                         commands++;
 572                     } else if ((*commands >= '0') && (*commands <= '9')) {
 573                         length = strtol (commands, &commands, 10);
 574                         break;
 575                     } else if (*commands == '-') {
 576                         length = -1;
 577                         commands++;
 578                         break;
 579                     } else {
 580                         VERBOSE (ERROR, fprintf (stderr, "unknown length (%s)\n", commands));
 581                         rc = 1;
 582                         break;
 583                     }
 584                 }
 585                 if (rc == 0) {
 586                     switch (c) {
 587                     case '+':
 588                         rc = gotoaddr ((length > 0) ? addrfile + length : -1);
 589                         break;
 590
 591                     case 'd':
 592                         fseek (fin, length, SEEK_CUR);
 593                         break;
 594
 595                     case 'p':
 596                         hexdump (length);
 597                         break;
 598                     }
 599                 }
 600                 offset = 0;
 601                 break;
 602
 603             case 's': /* substitute mode */
 604                 offset = 0;
 605                 break;
 606
 607             default:
 608                 VERBOSE (ERROR, fprintf (stderr, "unknown command (%c)\n", commands[-1]));
 609                 rc = 1;
 610             }
 611         }
 612     }
 613
 614     /* end of file */
 615     if ((rc == 0) && (fout != NULL)) {
 616         while (!feof (fin)) {
 617             int nbread = fread (buffer, 1, BUFFERSIZE, fin);
 618             if (nbread) {
 619                 fwrite (buffer, 1, nbread, fout);
 620             }
 621         }
 622     }
 623
 624     /* close all */
 625     if (fin) fclose (fin);
 626     if (fout) fclose (fout);
 627
 628     return rc;
 629 }
 630
 631 // test: hexdump.exe -h | awk '/usage:/ { rc=1 } END { exit (1-rc) }'
 632 // test: hexdump.exe foo 2>&1 | grep -q 'invalid option'
 633 // test: hexdump.exe -n 2>&1 | grep -q 'missing number of columns'
 634 // test: hexdump.exe -v 2>&1 | grep -q 'missing verbose level'
 635 // test: hexdump.exe -_ 2> /dev/null | awk 'END { if (NR == 0) { exit(0) } else exit (1) }'
 636 // test: hexdump.exe -_ 2>&1 | awk '/usage:/ { rc=1 } END { exit (1-rc) }'
 637 // test: hexdump.exe -i hexdump.c | grep -q '0x[0-9a-f]*: '
 638 // test: hexdump.exe -i hexdump.ko 2>&1 | grep -q "can't open file"
 639 // test: hexdump.exe -i hexdump.c -o ko/test.c 2>&1 | grep -q "can't open file"
 640 // test: cat hexdump.c | hexdump.exe -n 3 | head -2 | tail -1 | grep -q '0x000003: 64 65 70  dep'
 641 // test: hexdump.exe -i hexdump.c -n 3 | head -2 | tail -1 | grep -q '0x0003: 64 65 70  dep'
 642 // test: hexdump.exe -i hexdump.c -o test.c -e 'p 200' | tail -1 | grep -q '0x00c0:'
 643 // test: cmp hexdump.c test.c; x=$?; rm test.c; test x$x = x0
 644 // test: hexdump.exe -i hexdump.c -e ' /cflags/ p 17 /debug/ p 8' | grep -q '0x0019: 2a 2f 0a 2f 2a 20 6c 69  \*\/\./\* li'
 645 // test: hexdump.exe -i hexdump.c -o test.c -e ' /cfl\x61gs/ p 16 /d\145bug/ p 8' | grep -q '0x0027: 64 65 62 75 67 2e 6f 20  debug.o'
 646 // test: cmp hexdump.c test.c; x=$?; rm test.c; test x$x = x0
 647 // test: hexdump.exe -i hexdump.c -e ' /\n/ p 8' | grep -q '0x000d: 0a 2f 2a 20 63 66 6c 61  \./\* cfla'
 648 // test: hexdump.exe -i hexdump.c -o test.c -e ' /\a\b\e\f\r\t\v/ p 8'; x=$?; test x$x = x1
 649 // test: cmp hexdump.c test.c; x=$?; rm test.c; test x$x = x0
 650 // test: hexdump.exe -i hexdump.c -v 3 -e " /\'/" -e ' /\"/' -e ' /\\/' -e ' /\x2a/' -e ' s/\x3A/' | grep l: | wc -l | xargs test 5 =
 651 // test: hexdump.exe -i hexdump.c -e ' /\n\/* vim:/ p -' | grep -q ': 74 3a 20 2a 2f 0a *t: \*\/\.'
 652 // test: hexdump.exe -i hexdump.c -e 'p go_to_end' 2>&1 | grep -q 'unknown length'
 653 // test: hexdump.exe -i hexdump.c -e ' //' 2>&1 | grep -q 'incorrect sequence'
 654 // test: hexdump.exe -i hexdump.c -e 'foo' 2>&1 | grep -q 'unknown command'
 655 // test: hexdump.exe -i hexdump.c -e '0x20 p 8 64 p 8 0200 p 16' | grep -q '0x0080:'
 656 // test: hexdump.exe -i hexdump.c -e '0xg' 2>&1 | grep -q 'erroneous address'
 657 // test: hexdump.exe -i hexdump.c -o test.c -e ' /cflags/ a 414e5a /link/ i 2F333B'
 658 // test: grep -q ANZcflags test.c && grep -q 'link/3;er' test.c; x=$?; rm test.c; test x$x = x0
 659 // test: hexdump.exe -i hexdump.c -e ' /cflags/ a 414e5' 2>&1 | grep 'erroneous sequence'
 660 // test: hexdump.exe -i hexdump.c -o test.c -e ' /lags/ d 2'
 661 // test: grep -q cfgs test.c; x=$?; rm test.c; test x$x = x0
 662 // test: hexdump.exe -i hexdump.c -o test.c -e ' /lags/ +2 i 2041'
 663 // test: grep -q 'cf Ags' test.c; x=$?; rm test.c; test x$x = x0
 664
 665 /* vim: set ts=4 sw=4 et: */