unlzw.c

   1 /* unlzw.c -- decompress files in LZW format.
   2  * The code in this file is directly derived from the public domain 'compress'
   3  * written by Spencer Thomas, Joe Orost, James Woods, Jim McKie, Steve Davies,
   4  * Ken Turkowski, Dave Mack and Peter Jannesen.
   5  *
   6  * This is a temporary version which will be rewritten in some future version
   7  * to accommodate in-memory decompression.
   8  */
   9
  10 #ifdef RCSID
  11 static char rcsid[] = "$Id$";
  12 #endif
  13
  14 #include <config.h>
  15 #include "tailor.h"
  16
  17 #ifdef HAVE_UNISTD_H
  18 #  include <unistd.h>
  19 #endif
  20 #ifdef HAVE_FCNTL_H
  21 #  include <fcntl.h>
  22 #endif
  23
  24 #include "gzip.h"
  25 #include "lzw.h"
  26
  27 typedef unsigned char   char_type;
  28 typedef          long   code_int;
  29 typedef unsigned long   count_int;
  30 typedef unsigned short  count_short;
  31 typedef unsigned long   cmp_code_int;
  32
  33 #define MAXCODE(n)      (1L << (n))
  34
  35 #ifndef REGISTERS
  36 #       define  REGISTERS       2
  37 #endif
  38 #define REG1
  39 #define REG2
  40 #define REG3
  41 #define REG4
  42 #define REG5
  43 #define REG6
  44 #define REG7
  45 #define REG8
  46 #define REG9
  47 #define REG10
  48 #define REG11
  49 #define REG12
  50 #define REG13
  51 #define REG14
  52 #define REG15
  53 #define REG16
  54 #if REGISTERS >= 1
  55 #       undef   REG1
  56 #       define  REG1    register
  57 #endif
  58 #if REGISTERS >= 2
  59 #       undef   REG2
  60 #       define  REG2    register
  61 #endif
  62 #if REGISTERS >= 3
  63 #       undef   REG3
  64 #       define  REG3    register
  65 #endif
  66 #if REGISTERS >= 4
  67 #       undef   REG4
  68 #       define  REG4    register
  69 #endif
  70 #if REGISTERS >= 5
  71 #       undef   REG5
  72 #       define  REG5    register
  73 #endif
  74 #if REGISTERS >= 6
  75 #       undef   REG6
  76 #       define  REG6    register
  77 #endif
  78 #if REGISTERS >= 7
  79 #       undef   REG7
  80 #       define  REG7    register
  81 #endif
  82 #if REGISTERS >= 8
  83 #       undef   REG8
  84 #       define  REG8    register
  85 #endif
  86 #if REGISTERS >= 9
  87 #       undef   REG9
  88 #       define  REG9    register
  89 #endif
  90 #if REGISTERS >= 10
  91 #       undef   REG10
  92 #       define  REG10   register
  93 #endif
  94 #if REGISTERS >= 11
  95 #       undef   REG11
  96 #       define  REG11   register
  97 #endif
  98 #if REGISTERS >= 12
  99 #       undef   REG12
 100 #       define  REG12   register
 101 #endif
 102 #if REGISTERS >= 13
 103 #       undef   REG13
 104 #       define  REG13   register
 105 #endif
 106 #if REGISTERS >= 14
 107 #       undef   REG14
 108 #       define  REG14   register
 109 #endif
 110 #if REGISTERS >= 15
 111 #       undef   REG15
 112 #       define  REG15   register
 113 #endif
 114 #if REGISTERS >= 16
 115 #       undef   REG16
 116 #       define  REG16   register
 117 #endif
 118
 119 #ifndef BYTEORDER
 120 #       define  BYTEORDER       0000
 121 #endif
 122
 123 #ifndef NOALLIGN
 124 #       define  NOALLIGN        0
 125 #endif
 126
 127
 128 union   bytes {
 129     long  word;
 130     struct {
 131 #if BYTEORDER == 4321
 132         char_type       b1;
 133         char_type       b2;
 134         char_type       b3;
 135         char_type       b4;
 136 #else
 137 #if BYTEORDER == 1234
 138         char_type       b4;
 139         char_type       b3;
 140         char_type       b2;
 141         char_type       b1;
 142 #else
 143 #       undef   BYTEORDER
 144         int  dummy;
 145 #endif
 146 #endif
 147     } bytes;
 148 };
 149
 150 #if BYTEORDER == 4321 && NOALLIGN == 1
 151 #  define input(b,o,c,n,m){ \
 152      (c) = (*(long *)(&(b)[(o)>>3])>>((o)&0x7))&(m); \
 153      (o) += (n); \
 154    }
 155 #else
 156 #  define input(b,o,c,n,m){ \
 157      REG1 char_type *p = &(b)[(o)>>3]; \
 158      (c) = ((((long)(p[0]))|((long)(p[1])<<8)| \
 159      ((long)(p[2])<<16))>>((o)&0x7))&(m); \
 160      (o) += (n); \
 161    }
 162 #endif
 163
 164 #ifndef MAXSEG_64K
 165    /* DECLARE(ush, tab_prefix, (1<<BITS)); -- prefix code */
 166 #  define tab_prefixof(i) tab_prefix[i]
 167 #  define clear_tab_prefixof()  memzero(tab_prefix, 256);
 168 #else
 169    /* DECLARE(ush, tab_prefix0, (1<<(BITS-1)); -- prefix for even codes */
 170    /* DECLARE(ush, tab_prefix1, (1<<(BITS-1)); -- prefix for odd  codes */
 171    ush *tab_prefix[2];
 172 #  define tab_prefixof(i) tab_prefix[(i)&1][(i)>>1]
 173 #  define clear_tab_prefixof()  \
 174       memzero(tab_prefix0, 128), \
 175       memzero(tab_prefix1, 128);
 176 #endif
 177 #define de_stack        ((char_type *)(&d_buf[DIST_BUFSIZE-1]))
 178 #define tab_suffixof(i) tab_suffix[i]
 179
 180 int block_mode = BLOCK_MODE; /* block compress mode -C compatible with 2.0 */
 181
 182 /* ============================================================================
 183  * Decompress in to out.  This routine adapts to the codes in the
 184  * file building the "string" table on-the-fly; requiring no table to
 185  * be stored in the compressed file.
 186  * IN assertions: the buffer inbuf contains already the beginning of
 187  *   the compressed data, from offsets iptr to insize-1 included.
 188  *   The magic header has already been checked and skipped.
 189  *   bytes_in and bytes_out have been initialized.
 190  */
 191 int unlzw(in, out)
 192     int in, out;    /* input and output file descriptors */
 193 {
 194     REG2   char_type  *stackp;
 195     REG3   code_int   code;
 196     REG4   int        finchar;
 197     REG5   code_int   oldcode;
 198     REG6   code_int   incode;
 199     REG7   long       inbits;
 200     REG8   long       posbits;
 201     REG9   int        outpos;
 202 /*  REG10  int        insize; (global) */
 203     REG11  unsigned   bitmask;
 204     REG12  code_int   free_ent;
 205     REG13  code_int   maxcode;
 206     REG14  code_int   maxmaxcode;
 207     REG15  int        n_bits;
 208     REG16  int        rsize;
 209
 210 #ifdef MAXSEG_64K
 211     tab_prefix[0] = tab_prefix0;
 212     tab_prefix[1] = tab_prefix1;
 213 #endif
 214     maxbits = get_byte();
 215     block_mode = maxbits & BLOCK_MODE;
 216     if ((maxbits & LZW_RESERVED) != 0) {
 217         WARN((stderr, "\n%s: %s: warning, unknown flags 0x%x\n",
 218               program_name, ifname, maxbits & LZW_RESERVED));
 219     }
 220     maxbits &= BIT_MASK;
 221     maxmaxcode = MAXCODE(maxbits);
 222
 223     if (maxbits > BITS) {
 224         fprintf(stderr,
 225                 "\n%s: %s: compressed with %d bits, can only handle %d bits\n",
 226                 program_name, ifname, maxbits, BITS);
 227         exit_code = ERROR;
 228         return ERROR;
 229     }
 230     rsize = insize;
 231     maxcode = MAXCODE(n_bits = INIT_BITS)-1;
 232     bitmask = (1<<n_bits)-1;
 233     oldcode = -1;
 234     finchar = 0;
 235     outpos = 0;
 236     posbits = inptr<<3;
 237
 238     free_ent = ((block_mode) ? FIRST : 256);
 239
 240     clear_tab_prefixof(); /* Initialize the first 256 entries in the table. */
 241
 242     for (code = 255 ; code >= 0 ; --code) {
 243         tab_suffixof(code) = (char_type)code;
 244     }
 245     do {
 246         REG1 int i;
 247         int  e;
 248         int  o;
 249
 250     resetbuf:
 251         e = insize-(o = (posbits>>3));
 252
 253         for (i = 0 ; i < e ; ++i) {
 254             inbuf[i] = inbuf[i+o];
 255         }
 256         insize = e;
 257         posbits = 0;
 258
 259         if (insize < INBUF_EXTRA) {
 260             if ((rsize = read(in, (char*)inbuf+insize, INBUFSIZ)) == -1) {
 261                 read_error();
 262             }
 263             insize += rsize;
 264             bytes_in += (off_t)rsize;
 265         }
 266         inbits = ((rsize != 0) ? ((long)insize - insize%n_bits)<<3 :
 267                   ((long)insize<<3)-(n_bits-1));
 268
 269         while (inbits > posbits) {
 270             if (free_ent > maxcode) {
 271                 posbits = ((posbits-1) +
 272                            ((n_bits<<3)-(posbits-1+(n_bits<<3))%(n_bits<<3)));
 273                 ++n_bits;
 274                 if (n_bits == maxbits) {
 275                     maxcode = maxmaxcode;
 276                 } else {
 277                     maxcode = MAXCODE(n_bits)-1;
 278                 }
 279                 bitmask = (1<<n_bits)-1;
 280                 goto resetbuf;
 281             }
 282             input(inbuf,posbits,code,n_bits,bitmask);
 283             Tracev((stderr, "%d ", code));
 284
 285             if (oldcode == -1) {
 286                 if (256 <= code)
 287                   gzip_error ("corrupt input.");
 288                 outbuf[outpos++] = (char_type)(finchar = (int)(oldcode=code));
 289                 continue;
 290             }
 291             if (code == CLEAR && block_mode) {
 292                 clear_tab_prefixof();
 293                 free_ent = FIRST - 1;
 294                 posbits = ((posbits-1) +
 295                            ((n_bits<<3)-(posbits-1+(n_bits<<3))%(n_bits<<3)));
 296                 maxcode = MAXCODE(n_bits = INIT_BITS)-1;
 297                 bitmask = (1<<n_bits)-1;
 298                 goto resetbuf;
 299             }
 300             incode = code;
 301             stackp = de_stack;
 302
 303             if (code >= free_ent) { /* Special case for KwKwK string. */
 304                 if (code > free_ent) {
 305 #ifdef DEBUG
 306                     char_type *p;
 307
 308                     posbits -= n_bits;
 309                     p = &inbuf[posbits>>3];
 310                     fprintf(stderr,
 311                             "code:%ld free_ent:%ld n_bits:%d insize:%u\n",
 312                             code, free_ent, n_bits, insize);
 313                     fprintf(stderr,
 314                             "posbits:%ld inbuf:%02X %02X %02X %02X %02X\n",
 315                             posbits, p[-1],p[0],p[1],p[2],p[3]);
 316 #endif
 317                     if (!test && outpos > 0) {
 318                         write_buf(out, (char*)outbuf, outpos);
 319                         bytes_out += (off_t)outpos;
 320                     }
 321                     gzip_error (to_stdout
 322                                 ? "corrupt input."
 323                                 : "corrupt input. Use zcat to recover some data.");
 324                 }
 325                 *--stackp = (char_type)finchar;
 326                 code = oldcode;
 327             }
 328
 329             while ((cmp_code_int)code >= (cmp_code_int)256) {
 330                 /* Generate output characters in reverse order */
 331                 *--stackp = tab_suffixof(code);
 332                 code = tab_prefixof(code);
 333             }
 334             *--stackp = (char_type)(finchar = tab_suffixof(code));
 335
 336             /* And put them out in forward order */
 337             {
 338                 REG1 int        i;
 339
 340                 if (outpos+(i = (de_stack-stackp)) >= OUTBUFSIZ) {
 341                     do {
 342                         if (i > OUTBUFSIZ-outpos) i = OUTBUFSIZ-outpos;
 343
 344                         if (i > 0) {
 345                             memcpy(outbuf+outpos, stackp, i);
 346                             outpos += i;
 347                         }
 348                         if (outpos >= OUTBUFSIZ) {
 349                             if (!test) {
 350                                 write_buf(out, (char*)outbuf, outpos);
 351                                 bytes_out += (off_t)outpos;
 352                             }
 353                             outpos = 0;
 354                         }
 355                         stackp+= i;
 356                     } while ((i = (de_stack-stackp)) > 0);
 357                 } else {
 358                     memcpy(outbuf+outpos, stackp, i);
 359                     outpos += i;
 360                 }
 361             }
 362
 363             if ((code = free_ent) < maxmaxcode) { /* Generate the new entry. */
 364
 365                 tab_prefixof(code) = (unsigned short)oldcode;
 366                 tab_suffixof(code) = (char_type)finchar;
 367                 free_ent = code+1;
 368             }
 369             oldcode = incode;   /* Remember previous code.      */
 370         }
 371     } while (rsize != 0);
 372
 373     if (!test && outpos > 0) {
 374         write_buf(out, (char*)outbuf, outpos);
 375         bytes_out += (off_t)outpos;
 376     }
 377     return OK;
 378 }