hexdump.c

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