archival/libunarchive/decompress_uncompress.c

   1 #include "config.h"
   2 #include "libbb.h"
   3
   4 #ifdef CONFIG_UNCOMPRESS || defined CONFIG_FEATURE_GUNZIP_UNCOMPRESS
   5
   6 /* uncompress for busybox -- (c) 2002 Robert Griebl
   7  *
   8  * based on the original compress42.c source
   9  * (see disclaimer below)
  10  */
  11
  12
  13 /* (N)compress42.c - File compression ala IEEE Computer, Mar 1992.
  14  *
  15  * Authors:
  16  *   Spencer W. Thomas   (decvax!harpo!utah-cs!utah-gr!thomas)
  17  *   Jim McKie           (decvax!mcvax!jim)
  18  *   Steve Davies        (decvax!vax135!petsd!peora!srd)
  19  *   Ken Turkowski       (decvax!decwrl!turtlevax!ken)
  20  *   James A. Woods      (decvax!ihnp4!ames!jaw)
  21  *   Joe Orost           (decvax!vax135!petsd!joe)
  22  *   Dave Mack           (csu@alembic.acs.com)
  23  *   Peter Jannesen, Network Communication Systems
  24  *                       (peter@ncs.nl)
  25  *
  26  * marc@suse.de : a small security fix for a buffer overflow
  27  *
  28  * [... History snipped ...]
  29  *
  30  */
  31 #include <stdio.h>
  32 #include <string.h>
  33 #include <unistd.h>
  34
  35 #define IBUFSIZ 2048    /* Defailt input buffer size                                                    */
  36 #define OBUFSIZ 2048    /* Default output buffer size                                                   */
  37
  38                                                         /* Defines for third byte of header                                     */
  39 #define MAGIC_1         (char_type)'\037'/* First byte of compressed file                               */
  40 #define MAGIC_2         (char_type)'\235'/* Second byte of compressed file                              */
  41 #define BIT_MASK        0x1f                    /* Mask for 'number of compresssion bits'               */
  42                                                                         /* Masks 0x20 and 0x40 are free.                                */
  43                                                                         /* I think 0x20 should mean that there is               */
  44                                                                         /* a fourth header byte (for expansion).        */
  45 #define BLOCK_MODE      0x80                    /* Block compresssion if table is full and              */
  46                                                                         /* compression rate is dropping flush tables    */
  47
  48                         /* the next two codes should not be changed lightly, as they must not   */
  49                         /* lie within the contiguous general code space.                                                */
  50 #define FIRST   257                                     /* first free entry                                                     */
  51 #define CLEAR   256                                     /* table clear output code                                              */
  52
  53 #define INIT_BITS 9                     /* initial number of bits/code */
  54
  55
  56 /*
  57  * machine variants which require cc -Dmachine:  pdp11, z8000, DOS
  58  */
  59 #define FAST
  60
  61 #define HBITS           17                      /* 50% occupancy */
  62 #define HSIZE      (1<<HBITS)
  63 #define HMASK      (HSIZE-1)
  64 #define HPRIME           9941
  65 #define BITS               16
  66 #undef  MAXSEG_64K
  67
  68 typedef long int                        code_int;
  69
  70 typedef long int                        count_int;
  71 typedef long int                        cmp_code_int;
  72
  73 typedef unsigned char   char_type;
  74
  75 #define MAXCODE(n)      (1L << (n))
  76
  77
  78
  79 int                             block_mode = BLOCK_MODE;/* Block compress mode -C compatible with 2.0*/
  80 int                             maxbits = BITS;         /* user settable max # bits/code                                */
  81 int                             exit_code = -1;         /* Exitcode of compress (-1 no file compressed) */
  82
  83 char_type               inbuf[IBUFSIZ+64];      /* Input buffer                                                                 */
  84 char_type               outbuf[OBUFSIZ+2048];/* Output buffer                                                           */
  85
  86
  87 count_int               htab[HSIZE];
  88 unsigned short  codetab[HSIZE];
  89
  90 #define htabof(i)                               htab[i]
  91 #define codetabof(i)                    codetab[i]
  92 #define tab_prefixof(i)                 codetabof(i)
  93 #define tab_suffixof(i)                 ((char_type *)(htab))[i]
  94 #define de_stack                                ((char_type *)&(htab[HSIZE-1]))
  95 #define clear_htab()                    memset(htab, -1, sizeof(htab))
  96 #define clear_tab_prefixof()    memset(codetab, 0, 256);
  97
  98
  99 /*
 100  * Decompress stdin to stdout.  This routine adapts to the codes in the
 101  * file building the "string" table on-the-fly; requiring no table to
 102  * be stored in the compressed file.  The tables used herein are shared
 103  * with those of the compress() routine.  See the definitions above.
 104  */
 105
 106 extern int uncompress(int fd_in, int fd_out)
 107 {
 108     char_type           *stackp;
 109     code_int             code;
 110         int                              finchar;
 111         code_int                 oldcode;
 112         code_int                 incode;
 113         int                              inbits;
 114         int                              posbits;
 115         int                              outpos;
 116         int                              insize;
 117         int                              bitmask;
 118         code_int                 free_ent;
 119         code_int                 maxcode;
 120         code_int                 maxmaxcode;
 121         int                              n_bits;
 122         int                              rsize = 0;
 123
 124         insize = 0;
 125
 126         inbuf [0] = xread_char(fd_in);
 127
 128         maxbits = inbuf[0] & BIT_MASK;
 129         block_mode = inbuf[0] & BLOCK_MODE;
 130         maxmaxcode = MAXCODE(maxbits);
 131
 132         if (maxbits > BITS)
 133         {
 134                 fprintf(stderr, "compressed with %d bits, can only handle %d bits\n", maxbits, BITS);
 135                 return -1;
 136         }
 137
 138         //fprintf(stderr, "Bits: %d, block_mode: %d\n", maxbits, block_mode );
 139
 140     maxcode = MAXCODE(n_bits = INIT_BITS)-1;
 141         bitmask = (1<<n_bits)-1;
 142         oldcode = -1;
 143         finchar = 0;
 144         outpos = 0;
 145         posbits = 0<<3;
 146
 147     free_ent = ((block_mode) ? FIRST : 256);
 148
 149         clear_tab_prefixof();   /* As above, initialize the first
 150                                                            256 entries in the table. */
 151
 152     for (code = 255 ; code >= 0 ; --code)
 153                 tab_suffixof(code) = (char_type)code;
 154
 155         do
 156         {
 157 resetbuf:       ;
 158                 {
 159                         int     i;
 160                         int     e;
 161                         int     o;
 162
 163                         e = insize-(o = (posbits>>3));
 164
 165                         for (i = 0 ; i < e ; ++i)
 166                                 inbuf[i] = inbuf[i+o];
 167
 168                         insize = e;
 169                         posbits = 0;
 170                 }
 171
 172                 if (insize < (int) sizeof(inbuf)-IBUFSIZ)
 173                 {
 174                         rsize = read(fd_in, inbuf+insize, IBUFSIZ);
 175                         insize += rsize;
 176                 }
 177
 178                 inbits = ((rsize > 0) ? (insize - insize%n_bits)<<3 :
 179                                                                 (insize<<3)-(n_bits-1));
 180
 181                 while (inbits > posbits)
 182                 {
 183                         if (free_ent > maxcode)
 184                         {
 185                                 posbits = ((posbits-1) + ((n_bits<<3) -
 186                                                                  (posbits-1+(n_bits<<3))%(n_bits<<3)));
 187
 188                                 ++n_bits;
 189                                 if (n_bits == maxbits)
 190                                         maxcode = maxmaxcode;
 191                                 else
 192                                     maxcode = MAXCODE(n_bits)-1;
 193
 194                                 bitmask = (1<<n_bits)-1;
 195                                 goto resetbuf;
 196                         }
 197
 198
 199                         {
 200                                 char_type *p = &inbuf[posbits>>3];
 201                                 code = ((((long)(p[0]))|((long)(p[1])<<8)|((long)(p[2])<<16))>>(posbits&0x7))&bitmask;
 202                         }
 203                         posbits += n_bits;
 204
 205
 206                         if (oldcode == -1)
 207                         {
 208                                 outbuf[outpos++] = (char_type)(finchar = (int)(oldcode = code));
 209                                 continue;
 210                         }
 211
 212                         if (code == CLEAR && block_mode)
 213                         {
 214                                 clear_tab_prefixof();
 215                         free_ent = FIRST - 1;
 216                                 posbits = ((posbits-1) + ((n_bits<<3) -
 217                                                         (posbits-1+(n_bits<<3))%(n_bits<<3)));
 218                             maxcode = MAXCODE(n_bits = INIT_BITS)-1;
 219                                 bitmask = (1<<n_bits)-1;
 220                                 goto resetbuf;
 221                         }
 222
 223                         incode = code;
 224                     stackp = de_stack;
 225
 226                         if (code >= free_ent)   /* Special case for KwKwK string.       */
 227                         {
 228                                 if (code > free_ent)
 229                                 {
 230                                         char_type               *p;
 231
 232                                         posbits -= n_bits;
 233                                         p = &inbuf[posbits>>3];
 234
 235                                         fprintf(stderr, "insize:%d posbits:%d inbuf:%02X %02X %02X %02X %02X (%d)\n", insize, posbits,
 236                                                         p[-1],p[0],p[1],p[2],p[3], (posbits&07));
 237                                 fprintf(stderr, "uncompress: corrupt input\n");
 238                                         return -1;
 239                                 }
 240
 241                 *--stackp = (char_type)finchar;
 242                         code = oldcode;
 243                         }
 244
 245                         while ((cmp_code_int)code >= (cmp_code_int)256)
 246                         { /* Generate output characters in reverse order */
 247                         *--stackp = tab_suffixof(code);
 248                         code = tab_prefixof(code);
 249                         }
 250
 251                         *--stackp =     (char_type)(finchar = tab_suffixof(code));
 252
 253                 /* And put them out in forward order */
 254
 255                         {
 256                                 int     i;
 257
 258                                 if (outpos+(i = (de_stack-stackp)) >= OBUFSIZ)
 259                                 {
 260                                         do
 261                                         {
 262                                                 if (i > OBUFSIZ-outpos) i = OBUFSIZ-outpos;
 263
 264                                                 if (i > 0)
 265                                                 {
 266                                                         memcpy(outbuf+outpos, stackp, i);
 267                                                         outpos += i;
 268                                                 }
 269
 270                                                 if (outpos >= OBUFSIZ)
 271                                                 {
 272                                                         write(fd_out, outbuf, outpos);
 273                                                         outpos = 0;
 274                                                 }
 275                                                 stackp+= i;
 276                                         }
 277                                         while ((i = (de_stack-stackp)) > 0);
 278                                 }
 279                                 else
 280                                 {
 281                                         memcpy(outbuf+outpos, stackp, i);
 282                                         outpos += i;
 283                                 }
 284                         }
 285
 286                         if ((code = free_ent) < maxmaxcode) /* Generate the new entry. */
 287                         {
 288                         tab_prefixof(code) = (unsigned short)oldcode;
 289                         tab_suffixof(code) = (char_type)finchar;
 290                         free_ent = code+1;
 291                         }
 292
 293                         oldcode = incode;       /* Remember previous code.      */
 294                 }
 295
 296     }
 297         while (rsize > 0);
 298
 299         if (outpos > 0) {
 300                 write(fd_out, outbuf, outpos);
 301         }
 302
 303         return 0;
 304 }
 305
 306
 307 #endif