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 #include <config.h>
  11
  12 #include <unistd.h>
  13 #include <fcntl.h>
  14
  15 #include "tailor.h"
  16 #include "gzip.h"
  17 #include "lzw.h"
  18
  19 typedef unsigned char char_type;
  20 typedef          long code_int;
  21 typedef unsigned long count_int;
  22 typedef unsigned short count_short;
  23 typedef unsigned long cmp_code_int;
  24
  25 #define MAXCODE(n)      (1L << (n))
  26
  27 #ifndef BYTEORDER
  28 #       define  BYTEORDER       0000
  29 #endif
  30
  31 #ifndef NOALLIGN
  32 #       define  NOALLIGN        0
  33 #endif
  34
  35
  36 union   bytes {
  37     long  word;
  38     struct {
  39 #if BYTEORDER == 4321
  40         char_type       b1;
  41         char_type       b2;
  42         char_type       b3;
  43         char_type       b4;
  44 #else
  45 #if BYTEORDER == 1234
  46         char_type       b4;
  47         char_type       b3;
  48         char_type       b2;
  49         char_type       b1;
  50 #else
  51 #       undef   BYTEORDER
  52         int  dummy;
  53 #endif
  54 #endif
  55     } bytes;
  56 };
  57
  58 #if BYTEORDER == 4321 && NOALLIGN == 1
  59 #  define input(b,o,c,n,m){ \
  60      (c) = (*(long *)(&(b)[(o)>>3])>>((o)&0x7))&(m); \
  61      (o) += (n); \
  62    }
  63 #else
  64 #  define input(b,o,c,n,m){ \
  65      char_type *p = &(b)[(o)>>3]; \
  66      (c) = ((((long)(p[0]))|((long)(p[1])<<8)| \
  67      ((long)(p[2])<<16))>>((o)&0x7))&(m); \
  68      (o) += (n); \
  69    }
  70 #endif
  71
  72 #ifndef MAXSEG_64K
  73    /* DECLARE(ush, tab_prefix, (1<<BITS)); -- prefix code */
  74 #  define tab_prefixof(i) tab_prefix[i]
  75 #  define clear_tab_prefixof()  memzero(tab_prefix, 256);
  76 #else
  77    /* DECLARE(ush, tab_prefix0, (1<<(BITS-1)); -- prefix for even codes */
  78    /* DECLARE(ush, tab_prefix1, (1<<(BITS-1)); -- prefix for odd  codes */
  79    ush *tab_prefix[2];
  80 #  define tab_prefixof(i) tab_prefix[(i)&1][(i)>>1]
  81 #  define clear_tab_prefixof()  \
  82       memzero(tab_prefix0, 128), \
  83       memzero(tab_prefix1, 128);
  84 #endif
  85 #define de_stack        ((char_type *)(&d_buf[DIST_BUFSIZE-1]))
  86 #define tab_suffixof(i) tab_suffix[i]
  87
  88 /* block compress mode -C compatible with 2.0 */
  89 static int block_mode = BLOCK_MODE;
  90
  91 /* ============================================================================
  92  * Decompress in to out.  This routine adapts to the codes in the
  93  * file building the "string" table on-the-fly; requiring no table to
  94  * be stored in the compressed file.
  95  * IN assertions: the buffer inbuf contains already the beginning of
  96  *   the compressed data, from offsets iptr to insize-1 included.
  97  *   The magic header has already been checked and skipped.
  98  *   bytes_in and bytes_out have been initialized.
  99  * 'in' and 'out' are the input and output file descriptors.
 100  */
 101 int
 102 unlzw (int in, int out)
 103 {
 104     char_type  *stackp;
 105     code_int   code;
 106     int        finchar;
 107     code_int   oldcode;
 108     code_int   incode;
 109     long       inbits;
 110     long       posbits;
 111     int        outpos;
 112 /*  int        insize; (global) */
 113     unsigned   bitmask;
 114     code_int   free_ent;
 115     code_int   maxcode;
 116     code_int   maxmaxcode;
 117     int        n_bits;
 118     int        rsize;
 119
 120 #ifdef MAXSEG_64K
 121     tab_prefix[0] = tab_prefix0;
 122     tab_prefix[1] = tab_prefix1;
 123 #endif
 124     maxbits = get_byte();
 125     block_mode = maxbits & BLOCK_MODE;
 126     if ((maxbits & LZW_RESERVED) != 0) {
 127         WARN((stderr, "\n%s: %s: warning, unknown flags 0x%x\n",
 128               program_name, ifname, (unsigned int) maxbits & LZW_RESERVED));
 129     }
 130     maxbits &= BIT_MASK;
 131     maxmaxcode = MAXCODE(maxbits);
 132
 133     if (maxbits > BITS) {
 134         fprintf(stderr,
 135                 "\n%s: %s: compressed with %d bits, can only handle %d bits\n",
 136                 program_name, ifname, maxbits, BITS);
 137         exit_code = ERROR;
 138         return ERROR;
 139     }
 140     rsize = insize;
 141     maxcode = MAXCODE(n_bits = INIT_BITS)-1;
 142     bitmask = (1<<n_bits)-1;
 143     oldcode = -1;
 144     finchar = 0;
 145     outpos = 0;
 146     posbits = inptr<<3;
 147
 148     free_ent = ((block_mode) ? FIRST : 256);
 149
 150     clear_tab_prefixof(); /* Initialize the first 256 entries in the table. */
 151
 152     for (code = 255 ; code >= 0 ; --code) {
 153         tab_suffixof(code) = (char_type)code;
 154     }
 155     do {
 156         int i;
 157         int  e;
 158         int  o;
 159
 160     resetbuf:
 161         o = posbits >> 3;
 162         e = o <= insize ? insize - o : 0;
 163
 164         for (i = 0 ; i < e ; ++i) {
 165             inbuf[i] = inbuf[i+o];
 166         }
 167         insize = e;
 168         posbits = 0;
 169
 170         if (insize < INBUF_EXTRA) {
 171             rsize = read_buffer (in, (char *) inbuf + insize, INBUFSIZ);
 172             if (rsize == -1) {
 173                 read_error();
 174             }
 175             insize += rsize;
 176             bytes_in += (off_t)rsize;
 177         }
 178         inbits = ((rsize != 0) ? ((long)insize - insize%n_bits)<<3 :
 179                   ((long)insize<<3)-(n_bits-1));
 180
 181         while (inbits > posbits) {
 182             if (free_ent > maxcode) {
 183                 posbits = ((posbits-1) +
 184                            ((n_bits<<3)-(posbits-1+(n_bits<<3))%(n_bits<<3)));
 185                 ++n_bits;
 186                 if (n_bits == maxbits) {
 187                     maxcode = maxmaxcode;
 188                 } else {
 189                     maxcode = MAXCODE(n_bits)-1;
 190                 }
 191                 bitmask = (1<<n_bits)-1;
 192                 goto resetbuf;
 193             }
 194             input(inbuf,posbits,code,n_bits,bitmask);
 195             Tracev((stderr, "%ld ", code));
 196
 197             if (oldcode == -1) {
 198                 if (256 <= code)
 199                   gzip_error ("corrupt input.");
 200                 outbuf[outpos++] = (char_type)(finchar = (int)(oldcode=code));
 201                 continue;
 202             }
 203             if (code == CLEAR && block_mode) {
 204                 clear_tab_prefixof();
 205                 free_ent = FIRST - 1;
 206                 posbits = ((posbits-1) +
 207                            ((n_bits<<3)-(posbits-1+(n_bits<<3))%(n_bits<<3)));
 208                 maxcode = MAXCODE(n_bits = INIT_BITS)-1;
 209                 bitmask = (1<<n_bits)-1;
 210                 goto resetbuf;
 211             }
 212             incode = code;
 213             stackp = de_stack;
 214
 215             if (code >= free_ent) { /* Special case for KwKwK string. */
 216                 if (code > free_ent) {
 217 #ifdef DEBUG
 218                     char_type *p;
 219
 220                     posbits -= n_bits;
 221                     p = &inbuf[posbits>>3];
 222                     fprintf(stderr,
 223                             "code:%ld free_ent:%ld n_bits:%d insize:%u\n",
 224                             code, free_ent, n_bits, insize);
 225                     fprintf(stderr,
 226                             "posbits:%ld inbuf:%02X %02X %02X %02X %02X\n",
 227                             posbits, p[-1],p[0],p[1],p[2],p[3]);
 228 #endif
 229                     if (outpos > 0)
 230                       write_buf (out, outbuf, outpos);
 231                     gzip_error (to_stdout
 232                                 ? "corrupt input."
 233                                 : "corrupt input. Use zcat to recover some data.");
 234                 }
 235                 *--stackp = (char_type)finchar;
 236                 code = oldcode;
 237             }
 238
 239             while ((cmp_code_int)code >= (cmp_code_int)256) {
 240                 /* Generate output characters in reverse order */
 241                 *--stackp = tab_suffixof(code);
 242                 code = tab_prefixof(code);
 243             }
 244             *--stackp = (char_type)(finchar = tab_suffixof(code));
 245
 246             /* And put them out in forward order */
 247             {
 248                 int i;
 249
 250                 if (outpos+(i = (de_stack-stackp)) >= OUTBUFSIZ) {
 251                     do {
 252                         if (i > OUTBUFSIZ-outpos) i = OUTBUFSIZ-outpos;
 253
 254                         if (i > 0) {
 255                             memcpy(outbuf+outpos, stackp, i);
 256                             outpos += i;
 257                         }
 258                         if (outpos >= OUTBUFSIZ) {
 259                             write_buf (out, outbuf, outpos);
 260                             outpos = 0;
 261                         }
 262                         stackp+= i;
 263                     } while ((i = (de_stack-stackp)) > 0);
 264                 } else {
 265                     memcpy(outbuf+outpos, stackp, i);
 266                     outpos += i;
 267                 }
 268             }
 269
 270             if ((code = free_ent) < maxmaxcode) { /* Generate the new entry. */
 271
 272                 tab_prefixof(code) = (unsigned short)oldcode;
 273                 tab_suffixof(code) = (char_type)finchar;
 274                 free_ent = code+1;
 275             }
 276             oldcode = incode;   /* Remember previous code.      */
 277         }
 278     } while (rsize != 0);
 279
 280     if (outpos > 0)
 281       write_buf (out, outbuf, outpos);
 282     return OK;
 283 }