Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1 | /* Modified for busybox by Glenn McGrath <bug1@optushome.com.au> */ |
Glenn L McGrath | fff11f1 | 2001-11-18 14:20:25 +0000 | [diff] [blame] | 2 | /* Added support output to stdout by Thomas Lundquist <thomasez@zelow.no> */ |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 3 | /*-- |
| 4 | This file is a part of bzip2 and/or libbzip2, a program and |
| 5 | library for lossless, block-sorting data compression. |
| 6 | |
| 7 | Copyright (C) 1996-2000 Julian R Seward. All rights reserved. |
| 8 | |
| 9 | Redistribution and use in source and binary forms, with or without |
| 10 | modification, are permitted provided that the following conditions |
| 11 | are met: |
| 12 | |
| 13 | 1. Redistributions of source code must retain the above copyright |
| 14 | notice, this list of conditions and the following disclaimer. |
| 15 | |
| 16 | 2. The origin of this software must not be misrepresented; you must |
| 17 | not claim that you wrote the original software. If you use this |
| 18 | software in a product, an acknowledgment in the product |
| 19 | documentation would be appreciated but is not required. |
| 20 | |
| 21 | 3. Altered source versions must be plainly marked as such, and must |
| 22 | not be misrepresented as being the original software. |
| 23 | |
| 24 | 4. The name of the author may not be used to endorse or promote |
| 25 | products derived from this software without specific prior written |
| 26 | permission. |
| 27 | |
| 28 | THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS |
| 29 | OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED |
| 30 | WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
| 31 | ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY |
| 32 | DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
| 33 | DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE |
| 34 | GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS |
| 35 | INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, |
| 36 | WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING |
| 37 | NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS |
| 38 | SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
| 39 | |
| 40 | Julian Seward, Cambridge, UK. |
| 41 | jseward@acm.org |
| 42 | bzip2/libbzip2 version 1.0 of 21 March 2000 |
| 43 | |
| 44 | This program is based on (at least) the work of: |
| 45 | Mike Burrows |
| 46 | David Wheeler |
| 47 | Peter Fenwick |
| 48 | Alistair Moffat |
| 49 | Radford Neal |
| 50 | Ian H. Witten |
| 51 | Robert Sedgewick |
| 52 | Jon L. Bentley |
| 53 | |
| 54 | For more information on these sources, see the manual. |
| 55 | --*/ |
| 56 | |
| 57 | #include <stdlib.h> |
| 58 | #include <stdio.h> |
| 59 | #include <string.h> |
Glenn L McGrath | fff11f1 | 2001-11-18 14:20:25 +0000 | [diff] [blame] | 60 | #include <getopt.h> |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 61 | #include <busybox.h> |
| 62 | |
| 63 | //#define TRUE 1 |
| 64 | //#define FALSE 0 |
| 65 | |
| 66 | #define MTFA_SIZE 4096 |
| 67 | #define MTFL_SIZE 16 |
| 68 | #define BZ_N_GROUPS 6 |
| 69 | #define BZ_G_SIZE 50 |
| 70 | #define BZ_MAX_ALPHA_SIZE 258 |
| 71 | |
| 72 | #define BZ_OK 0 |
| 73 | #define BZ_RUN_OK 1 |
| 74 | #define BZ_FLUSH_OK 2 |
| 75 | #define BZ_FINISH_OK 3 |
| 76 | #define BZ_STREAM_END 4 |
| 77 | #define BZ_SEQUENCE_ERROR (-1) |
| 78 | #define BZ_PARAM_ERROR (-2) |
| 79 | #define BZ_MEM_ERROR (-3) |
| 80 | #define BZ_DATA_ERROR (-4) |
| 81 | #define BZ_DATA_ERROR_MAGIC (-5) |
| 82 | #define BZ_IO_ERROR (-6) |
| 83 | #define BZ_UNEXPECTED_EOF (-7) |
| 84 | #define BZ_OUTBUFF_FULL (-8) |
| 85 | #define BZ_CONFIG_ERROR (-9) |
| 86 | |
| 87 | #define BZ_RUNA 0 |
| 88 | #define BZ_RUNB 1 |
| 89 | |
| 90 | #define BZ_MAX_UNUSED 5000 |
| 91 | #define FILE_NAME_LEN 1034 |
| 92 | /*-- states for decompression. --*/ |
| 93 | |
| 94 | #define BZ_X_IDLE 1 |
| 95 | #define BZ_X_OUTPUT 2 |
| 96 | |
| 97 | #define BZ_X_MAGIC_1 10 |
| 98 | #define BZ_X_MAGIC_2 11 |
| 99 | #define BZ_X_MAGIC_3 12 |
| 100 | #define BZ_X_MAGIC_4 13 |
| 101 | #define BZ_X_BLKHDR_1 14 |
| 102 | #define BZ_X_BLKHDR_2 15 |
| 103 | #define BZ_X_BLKHDR_3 16 |
| 104 | #define BZ_X_BLKHDR_4 17 |
| 105 | #define BZ_X_BLKHDR_5 18 |
| 106 | #define BZ_X_BLKHDR_6 19 |
| 107 | #define BZ_X_BCRC_1 20 |
| 108 | #define BZ_X_BCRC_2 21 |
| 109 | #define BZ_X_BCRC_3 22 |
| 110 | #define BZ_X_BCRC_4 23 |
| 111 | #define BZ_X_RANDBIT 24 |
| 112 | #define BZ_X_ORIGPTR_1 25 |
| 113 | #define BZ_X_ORIGPTR_2 26 |
| 114 | #define BZ_X_ORIGPTR_3 27 |
| 115 | #define BZ_X_MAPPING_1 28 |
| 116 | #define BZ_X_MAPPING_2 29 |
| 117 | #define BZ_X_SELECTOR_1 30 |
| 118 | #define BZ_X_SELECTOR_2 31 |
| 119 | #define BZ_X_SELECTOR_3 32 |
| 120 | #define BZ_X_CODING_1 33 |
| 121 | #define BZ_X_CODING_2 34 |
| 122 | #define BZ_X_CODING_3 35 |
| 123 | #define BZ_X_MTF_1 36 |
| 124 | #define BZ_X_MTF_2 37 |
| 125 | #define BZ_X_MTF_3 38 |
| 126 | #define BZ_X_MTF_4 39 |
| 127 | #define BZ_X_MTF_5 40 |
| 128 | #define BZ_X_MTF_6 41 |
| 129 | #define BZ_X_ENDHDR_2 42 |
| 130 | #define BZ_X_ENDHDR_3 43 |
| 131 | #define BZ_X_ENDHDR_4 44 |
| 132 | #define BZ_X_ENDHDR_5 45 |
| 133 | #define BZ_X_ENDHDR_6 46 |
| 134 | #define BZ_X_CCRC_1 47 |
| 135 | #define BZ_X_CCRC_2 48 |
| 136 | #define BZ_X_CCRC_3 49 |
| 137 | #define BZ_X_CCRC_4 50 |
| 138 | |
| 139 | #define BZ_MAX_CODE_LEN 23 |
| 140 | #define BZ_VERSION "1.0.1, 23-June-2000" |
| 141 | #define OM_TEST 3 |
| 142 | #define SM_F2F 3 |
| 143 | |
| 144 | typedef struct { |
| 145 | char *next_in; |
| 146 | unsigned int avail_in; |
| 147 | unsigned int total_in_lo32; |
| 148 | unsigned int total_in_hi32; |
| 149 | |
| 150 | char *next_out; |
| 151 | unsigned int avail_out; |
| 152 | unsigned int total_out_lo32; |
| 153 | unsigned int total_out_hi32; |
| 154 | |
| 155 | void *state; |
| 156 | |
| 157 | void *(*bzalloc)(void *,int,int); |
| 158 | void (*bzfree)(void *,void *); |
| 159 | void *opaque; |
| 160 | } bz_stream; |
| 161 | |
| 162 | typedef struct { |
| 163 | bz_stream strm; |
| 164 | FILE *handle; |
| 165 | unsigned char initialisedOk; |
| 166 | unsigned char writing; |
| 167 | char buf[BZ_MAX_UNUSED]; |
| 168 | int lastErr; |
| 169 | int bufN; |
| 170 | } bzFile; |
| 171 | |
| 172 | /*-- Structure holding all the decompression-side stuff. --*/ |
| 173 | typedef struct { |
| 174 | /* pointer back to the struct bz_stream */ |
| 175 | bz_stream* strm; |
| 176 | |
| 177 | /* state indicator for this stream */ |
| 178 | int state; |
| 179 | |
| 180 | /* for doing the final run-length decoding */ |
| 181 | unsigned char state_out_ch; |
| 182 | int state_out_len; |
| 183 | unsigned char blockRandomised; |
| 184 | int rNToGo; |
| 185 | int rTPos; |
| 186 | |
| 187 | /* the buffer for bit stream reading */ |
| 188 | unsigned int bsBuff; |
| 189 | int bsLive; |
| 190 | |
| 191 | /* misc administratium */ |
| 192 | int blockSize100k; |
| 193 | unsigned char smallDecompress; |
| 194 | int currBlockNo; |
| 195 | int verbosity; |
| 196 | |
| 197 | /* for undoing the Burrows-Wheeler transform */ |
| 198 | int origPtr; |
| 199 | unsigned int tPos; |
| 200 | int k0; |
| 201 | int unzftab[256]; |
| 202 | int nblock_used; |
| 203 | int cftab[257]; |
| 204 | int cftabCopy[257]; |
| 205 | |
| 206 | /* for undoing the Burrows-Wheeler transform (FAST) */ |
| 207 | unsigned int *tt; |
| 208 | |
| 209 | /* for undoing the Burrows-Wheeler transform (SMALL) */ |
| 210 | unsigned short *ll16; |
| 211 | unsigned char *ll4; |
| 212 | |
| 213 | /* stored and calculated CRCs */ |
| 214 | unsigned int storedBlockCRC; |
| 215 | unsigned int storedCombinedCRC; |
| 216 | unsigned int calculatedBlockCRC; |
| 217 | unsigned int calculatedCombinedCRC; |
| 218 | |
| 219 | /* map of bytes used in block */ |
| 220 | int nInUse; |
| 221 | unsigned char inUse[256]; |
| 222 | unsigned char inUse16[16]; |
| 223 | unsigned char seqToUnseq[256]; |
| 224 | |
| 225 | /* for decoding the MTF values */ |
| 226 | unsigned char mtfa [MTFA_SIZE]; |
| 227 | unsigned char selector [2 + (900000 / BZ_G_SIZE)]; |
| 228 | unsigned char selectorMtf[2 + (900000 / BZ_G_SIZE)]; |
| 229 | unsigned char len [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE]; |
| 230 | int mtfbase[256 / MTFL_SIZE]; |
| 231 | |
| 232 | int limit [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE]; |
| 233 | int base [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE]; |
| 234 | int perm [BZ_N_GROUPS][BZ_MAX_ALPHA_SIZE]; |
| 235 | int minLens[BZ_N_GROUPS]; |
| 236 | |
| 237 | /* save area for scalars in the main decompress code */ |
| 238 | int save_i; |
| 239 | int save_j; |
| 240 | int save_t; |
| 241 | int save_alphaSize; |
| 242 | int save_nGroups; |
| 243 | int save_nSelectors; |
| 244 | int save_EOB; |
| 245 | int save_groupNo; |
| 246 | int save_groupPos; |
| 247 | int save_nextSym; |
| 248 | int save_nblockMAX; |
| 249 | int save_nblock; |
| 250 | int save_es; |
| 251 | int save_N; |
| 252 | int save_curr; |
| 253 | int save_zt; |
| 254 | int save_zn; |
| 255 | int save_zvec; |
| 256 | int save_zj; |
| 257 | int save_gSel; |
| 258 | int save_gMinlen; |
| 259 | int *save_gLimit; |
| 260 | int *save_gBase; |
| 261 | int *save_gPerm; |
| 262 | } DState; |
| 263 | |
| 264 | int BZ2_rNums[512]; |
| 265 | //int verbosity_level; |
| 266 | unsigned char smallMode; |
| 267 | unsigned char noisy; |
| 268 | char *progName; |
| 269 | char inName[FILE_NAME_LEN]; |
| 270 | char outName[FILE_NAME_LEN]; |
| 271 | int srcMode; |
| 272 | int opMode; |
| 273 | unsigned char deleteOutputOnInterrupt; |
| 274 | FILE *outputHandleJustInCase; |
| 275 | int numFileNames; |
| 276 | int numFilesProcessed; |
| 277 | int exitValue; |
| 278 | |
Aaron Lehmann | b9df470 | 2001-12-06 03:22:43 +0000 | [diff] [blame] | 279 | const unsigned int BZ2_crc32Table[256] = { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 280 | |
| 281 | /*-- Ugly, innit? --*/ |
| 282 | |
| 283 | 0x00000000L, 0x04c11db7L, 0x09823b6eL, 0x0d4326d9L, |
| 284 | 0x130476dcL, 0x17c56b6bL, 0x1a864db2L, 0x1e475005L, |
| 285 | 0x2608edb8L, 0x22c9f00fL, 0x2f8ad6d6L, 0x2b4bcb61L, |
| 286 | 0x350c9b64L, 0x31cd86d3L, 0x3c8ea00aL, 0x384fbdbdL, |
| 287 | 0x4c11db70L, 0x48d0c6c7L, 0x4593e01eL, 0x4152fda9L, |
| 288 | 0x5f15adacL, 0x5bd4b01bL, 0x569796c2L, 0x52568b75L, |
| 289 | 0x6a1936c8L, 0x6ed82b7fL, 0x639b0da6L, 0x675a1011L, |
| 290 | 0x791d4014L, 0x7ddc5da3L, 0x709f7b7aL, 0x745e66cdL, |
| 291 | 0x9823b6e0L, 0x9ce2ab57L, 0x91a18d8eL, 0x95609039L, |
| 292 | 0x8b27c03cL, 0x8fe6dd8bL, 0x82a5fb52L, 0x8664e6e5L, |
| 293 | 0xbe2b5b58L, 0xbaea46efL, 0xb7a96036L, 0xb3687d81L, |
| 294 | 0xad2f2d84L, 0xa9ee3033L, 0xa4ad16eaL, 0xa06c0b5dL, |
| 295 | 0xd4326d90L, 0xd0f37027L, 0xddb056feL, 0xd9714b49L, |
| 296 | 0xc7361b4cL, 0xc3f706fbL, 0xceb42022L, 0xca753d95L, |
| 297 | 0xf23a8028L, 0xf6fb9d9fL, 0xfbb8bb46L, 0xff79a6f1L, |
| 298 | 0xe13ef6f4L, 0xe5ffeb43L, 0xe8bccd9aL, 0xec7dd02dL, |
| 299 | 0x34867077L, 0x30476dc0L, 0x3d044b19L, 0x39c556aeL, |
| 300 | 0x278206abL, 0x23431b1cL, 0x2e003dc5L, 0x2ac12072L, |
| 301 | 0x128e9dcfL, 0x164f8078L, 0x1b0ca6a1L, 0x1fcdbb16L, |
| 302 | 0x018aeb13L, 0x054bf6a4L, 0x0808d07dL, 0x0cc9cdcaL, |
| 303 | 0x7897ab07L, 0x7c56b6b0L, 0x71159069L, 0x75d48ddeL, |
| 304 | 0x6b93dddbL, 0x6f52c06cL, 0x6211e6b5L, 0x66d0fb02L, |
| 305 | 0x5e9f46bfL, 0x5a5e5b08L, 0x571d7dd1L, 0x53dc6066L, |
| 306 | 0x4d9b3063L, 0x495a2dd4L, 0x44190b0dL, 0x40d816baL, |
| 307 | 0xaca5c697L, 0xa864db20L, 0xa527fdf9L, 0xa1e6e04eL, |
| 308 | 0xbfa1b04bL, 0xbb60adfcL, 0xb6238b25L, 0xb2e29692L, |
| 309 | 0x8aad2b2fL, 0x8e6c3698L, 0x832f1041L, 0x87ee0df6L, |
| 310 | 0x99a95df3L, 0x9d684044L, 0x902b669dL, 0x94ea7b2aL, |
| 311 | 0xe0b41de7L, 0xe4750050L, 0xe9362689L, 0xedf73b3eL, |
| 312 | 0xf3b06b3bL, 0xf771768cL, 0xfa325055L, 0xfef34de2L, |
| 313 | 0xc6bcf05fL, 0xc27dede8L, 0xcf3ecb31L, 0xcbffd686L, |
| 314 | 0xd5b88683L, 0xd1799b34L, 0xdc3abdedL, 0xd8fba05aL, |
| 315 | 0x690ce0eeL, 0x6dcdfd59L, 0x608edb80L, 0x644fc637L, |
| 316 | 0x7a089632L, 0x7ec98b85L, 0x738aad5cL, 0x774bb0ebL, |
| 317 | 0x4f040d56L, 0x4bc510e1L, 0x46863638L, 0x42472b8fL, |
| 318 | 0x5c007b8aL, 0x58c1663dL, 0x558240e4L, 0x51435d53L, |
| 319 | 0x251d3b9eL, 0x21dc2629L, 0x2c9f00f0L, 0x285e1d47L, |
| 320 | 0x36194d42L, 0x32d850f5L, 0x3f9b762cL, 0x3b5a6b9bL, |
| 321 | 0x0315d626L, 0x07d4cb91L, 0x0a97ed48L, 0x0e56f0ffL, |
| 322 | 0x1011a0faL, 0x14d0bd4dL, 0x19939b94L, 0x1d528623L, |
| 323 | 0xf12f560eL, 0xf5ee4bb9L, 0xf8ad6d60L, 0xfc6c70d7L, |
| 324 | 0xe22b20d2L, 0xe6ea3d65L, 0xeba91bbcL, 0xef68060bL, |
| 325 | 0xd727bbb6L, 0xd3e6a601L, 0xdea580d8L, 0xda649d6fL, |
| 326 | 0xc423cd6aL, 0xc0e2d0ddL, 0xcda1f604L, 0xc960ebb3L, |
| 327 | 0xbd3e8d7eL, 0xb9ff90c9L, 0xb4bcb610L, 0xb07daba7L, |
| 328 | 0xae3afba2L, 0xaafbe615L, 0xa7b8c0ccL, 0xa379dd7bL, |
| 329 | 0x9b3660c6L, 0x9ff77d71L, 0x92b45ba8L, 0x9675461fL, |
| 330 | 0x8832161aL, 0x8cf30badL, 0x81b02d74L, 0x857130c3L, |
| 331 | 0x5d8a9099L, 0x594b8d2eL, 0x5408abf7L, 0x50c9b640L, |
| 332 | 0x4e8ee645L, 0x4a4ffbf2L, 0x470cdd2bL, 0x43cdc09cL, |
| 333 | 0x7b827d21L, 0x7f436096L, 0x7200464fL, 0x76c15bf8L, |
| 334 | 0x68860bfdL, 0x6c47164aL, 0x61043093L, 0x65c52d24L, |
| 335 | 0x119b4be9L, 0x155a565eL, 0x18197087L, 0x1cd86d30L, |
| 336 | 0x029f3d35L, 0x065e2082L, 0x0b1d065bL, 0x0fdc1becL, |
| 337 | 0x3793a651L, 0x3352bbe6L, 0x3e119d3fL, 0x3ad08088L, |
| 338 | 0x2497d08dL, 0x2056cd3aL, 0x2d15ebe3L, 0x29d4f654L, |
| 339 | 0xc5a92679L, 0xc1683bceL, 0xcc2b1d17L, 0xc8ea00a0L, |
| 340 | 0xd6ad50a5L, 0xd26c4d12L, 0xdf2f6bcbL, 0xdbee767cL, |
| 341 | 0xe3a1cbc1L, 0xe760d676L, 0xea23f0afL, 0xeee2ed18L, |
| 342 | 0xf0a5bd1dL, 0xf464a0aaL, 0xf9278673L, 0xfde69bc4L, |
| 343 | 0x89b8fd09L, 0x8d79e0beL, 0x803ac667L, 0x84fbdbd0L, |
| 344 | 0x9abc8bd5L, 0x9e7d9662L, 0x933eb0bbL, 0x97ffad0cL, |
| 345 | 0xafb010b1L, 0xab710d06L, 0xa6322bdfL, 0xa2f33668L, |
| 346 | 0xbcb4666dL, 0xb8757bdaL, 0xb5365d03L, 0xb1f740b4L |
| 347 | }; |
| 348 | |
| 349 | void bz_rand_udp_mask(DState *s) |
| 350 | { |
| 351 | if (s->rNToGo == 0) { |
| 352 | s->rNToGo = BZ2_rNums[s->rTPos]; |
| 353 | s->rTPos++; |
| 354 | if (s->rTPos == 512) { |
| 355 | s->rTPos = 0; |
| 356 | } |
| 357 | } |
| 358 | s->rNToGo--; |
| 359 | } |
| 360 | |
| 361 | static unsigned char myfeof(FILE *f) |
| 362 | { |
| 363 | int c = fgetc(f); |
| 364 | if (c == EOF) { |
| 365 | return(TRUE); |
| 366 | } |
| 367 | ungetc(c, f); |
| 368 | return(FALSE); |
| 369 | } |
| 370 | |
| 371 | static void cleanUpAndFail(int ec) |
| 372 | { |
| 373 | int retVal; |
| 374 | |
| 375 | if ((srcMode == SM_F2F) && (opMode != OM_TEST) && deleteOutputOnInterrupt) { |
| 376 | if (noisy) { |
| 377 | error_msg("%s: Deleting output file %s, if it exists.\n", progName, outName); |
| 378 | } |
| 379 | if (outputHandleJustInCase != NULL) { |
| 380 | fclose(outputHandleJustInCase); |
| 381 | } |
| 382 | retVal = remove(outName); |
| 383 | if (retVal != 0) { |
| 384 | error_msg("%s: WARNING: deletion of output file (apparently) failed.\n", progName); |
| 385 | } |
| 386 | } |
| 387 | if (noisy && (numFileNames > 0) && (numFilesProcessed < numFileNames)) { |
| 388 | error_msg("%s: WARNING: some files have not been processed:\n" |
| 389 | "\t%d specified on command line, %d not processed yet.\n\n", |
| 390 | progName, numFileNames, numFileNames - numFilesProcessed ); |
| 391 | } |
| 392 | |
| 393 | exit(ec); |
| 394 | } |
| 395 | |
| 396 | |
| 397 | void panic(char *s) |
| 398 | { |
| 399 | error_msg("\n%s: PANIC -- internal consistency error:\n" |
| 400 | "\t%s\n" |
| 401 | "\tThis is a BUG. Please report it to me at:\n" |
| 402 | "\tjseward@acm.org\n", |
| 403 | progName, s); |
| 404 | cleanUpAndFail( 3 ); |
| 405 | } |
| 406 | |
| 407 | void BZ2_hbCreateDecodeTables(int *limit, int *base, int *perm, unsigned char *length, int minLen, int maxLen, int alphaSize ) |
| 408 | { |
| 409 | int pp, i, j, vec; |
| 410 | |
| 411 | pp = 0; |
| 412 | for (i = minLen; i <= maxLen; i++) { |
| 413 | for (j = 0; j < alphaSize; j++) { |
| 414 | if (length[j] == i) { |
| 415 | perm[pp] = j; |
| 416 | pp++; |
| 417 | } |
| 418 | } |
| 419 | } |
| 420 | |
| 421 | for (i = 0; i < BZ_MAX_CODE_LEN; i++) { |
| 422 | base[i] = 0; |
| 423 | } |
| 424 | |
| 425 | for (i = 0; i < alphaSize; i++) { |
| 426 | base[length[i]+1]++; |
| 427 | } |
| 428 | |
| 429 | for (i = 1; i < BZ_MAX_CODE_LEN; i++) { |
| 430 | base[i] += base[i-1]; |
| 431 | } |
| 432 | |
| 433 | for (i = 0; i < BZ_MAX_CODE_LEN; i++) { |
| 434 | limit[i] = 0; |
| 435 | } |
| 436 | vec = 0; |
| 437 | |
| 438 | for (i = minLen; i <= maxLen; i++) { |
| 439 | vec += (base[i+1] - base[i]); |
| 440 | limit[i] = vec-1; |
| 441 | vec <<= 1; |
| 442 | } |
| 443 | for (i = minLen + 1; i <= maxLen; i++) { |
| 444 | base[i] = ((limit[i-1] + 1) << 1) - base[i]; |
| 445 | } |
| 446 | } |
| 447 | |
| 448 | int bz_get_small(DState *s) |
| 449 | { |
| 450 | int cccc; |
| 451 | int nb, na, mid; |
| 452 | nb = 0; |
| 453 | na = 256; |
| 454 | do { |
| 455 | mid = (nb + na) >> 1; |
| 456 | if (s->tPos >= s->cftab[mid]) { |
| 457 | nb = mid; |
| 458 | } else { |
| 459 | na = mid; |
| 460 | } |
| 461 | } |
| 462 | while (na - nb != 1); |
| 463 | cccc = nb; |
| 464 | s->tPos = (((unsigned int)s->ll16[s->tPos]) | |
| 465 | (((((unsigned int)(s->ll4[(s->tPos) >> 1])) >> |
| 466 | (((s->tPos) << 2) & 0x4)) & 0xF) << 16)); |
| 467 | return(cccc); |
| 468 | } |
| 469 | |
| 470 | void assert_h(int errcode) |
| 471 | { |
| 472 | error_msg_and_die("\n\nbzip2/libbzip2: internal error number %d.\n" |
| 473 | "This is a bug in bzip2/libbzip2, %s.\n" |
| 474 | "Please report it to me at: jseward@acm.org. If this happened\n" |
| 475 | "when you were using some program which uses libbzip2 as a\n" |
| 476 | "component, you should also report this bug to the author(s)\n" |
| 477 | "of that program. Please make an effort to report this bug;\n" |
| 478 | "timely and accurate bug reports eventually lead to higher\n" |
| 479 | "quality software. Thanks. Julian Seward, 21 March 2000.\n\n", |
| 480 | errcode, BZ_VERSION); |
| 481 | } |
| 482 | |
| 483 | static int get_bits(DState *s, int *vvv, char nnn) |
| 484 | { |
| 485 | while (1) { |
| 486 | if (s->bsLive >= nnn) { |
| 487 | *vvv = (s->bsBuff >> (s->bsLive-nnn)) & ((1 << nnn)-1); |
| 488 | s->bsLive -= nnn; |
| 489 | break; |
| 490 | } |
| 491 | if (s->strm->avail_in == 0) { |
| 492 | return(FALSE); |
| 493 | } |
| 494 | s->bsBuff = (s->bsBuff << 8) | ((unsigned int) (*((unsigned char*)(s->strm->next_in)))); |
| 495 | s->bsLive += 8; |
| 496 | s->strm->next_in++; |
| 497 | s->strm->avail_in--; |
| 498 | s->strm->total_in_lo32++; |
| 499 | if (s->strm->total_in_lo32 == 0) { |
| 500 | s->strm->total_in_hi32++; |
| 501 | } |
| 502 | } |
| 503 | return(TRUE); |
| 504 | } |
| 505 | |
| 506 | int bz_get_fast(DState *s) |
| 507 | { |
| 508 | int cccc; |
| 509 | s->tPos = s->tt[s->tPos]; |
| 510 | cccc = (unsigned char)(s->tPos & 0xff); |
| 511 | s->tPos >>= 8; |
| 512 | return(cccc); |
| 513 | } |
| 514 | |
| 515 | /*---------------------------------------------------*/ |
| 516 | int BZ2_decompress(DState *s) |
| 517 | { |
| 518 | int uc = 0; |
| 519 | int retVal; |
| 520 | int minLen, maxLen; |
| 521 | bz_stream *strm = s->strm; |
| 522 | |
| 523 | /* stuff that needs to be saved/restored */ |
| 524 | int i; |
| 525 | int j; |
| 526 | int t; |
| 527 | int alphaSize; |
| 528 | int nGroups; |
| 529 | int nSelectors; |
| 530 | int EOB; |
| 531 | int groupNo; |
| 532 | int groupPos; |
| 533 | int nextSym; |
| 534 | int nblockMAX; |
| 535 | int nblock; |
| 536 | int es; |
| 537 | int N; |
| 538 | int curr; |
| 539 | int zt; |
| 540 | int zn; |
| 541 | int zvec; |
| 542 | int zj; |
| 543 | int gSel; |
| 544 | int gMinlen; |
| 545 | int *gLimit; |
| 546 | int *gBase; |
| 547 | int *gPerm; |
| 548 | int switch_val; |
| 549 | |
| 550 | int get_mtf_val_init(void) |
| 551 | { |
| 552 | if (groupPos == 0) { |
| 553 | groupNo++; |
| 554 | if (groupNo >= nSelectors) { |
| 555 | retVal = BZ_DATA_ERROR; |
| 556 | return(FALSE); |
| 557 | } |
| 558 | groupPos = BZ_G_SIZE; |
| 559 | gSel = s->selector[groupNo]; |
| 560 | gMinlen = s->minLens[gSel]; |
| 561 | gLimit = &(s->limit[gSel][0]); |
| 562 | gPerm = &(s->perm[gSel][0]); |
| 563 | gBase = &(s->base[gSel][0]); |
| 564 | } |
| 565 | groupPos--; |
| 566 | zn = gMinlen; |
| 567 | return(TRUE); |
| 568 | } |
| 569 | |
| 570 | if (s->state == BZ_X_MAGIC_1) { |
| 571 | /*initialise the save area*/ |
| 572 | s->save_i = 0; |
| 573 | s->save_j = 0; |
| 574 | s->save_t = 0; |
| 575 | s->save_alphaSize = 0; |
| 576 | s->save_nGroups = 0; |
| 577 | s->save_nSelectors = 0; |
| 578 | s->save_EOB = 0; |
| 579 | s->save_groupNo = 0; |
| 580 | s->save_groupPos = 0; |
| 581 | s->save_nextSym = 0; |
| 582 | s->save_nblockMAX = 0; |
| 583 | s->save_nblock = 0; |
| 584 | s->save_es = 0; |
| 585 | s->save_N = 0; |
| 586 | s->save_curr = 0; |
| 587 | s->save_zt = 0; |
| 588 | s->save_zn = 0; |
| 589 | s->save_zvec = 0; |
| 590 | s->save_zj = 0; |
| 591 | s->save_gSel = 0; |
| 592 | s->save_gMinlen = 0; |
| 593 | s->save_gLimit = NULL; |
| 594 | s->save_gBase = NULL; |
| 595 | s->save_gPerm = NULL; |
| 596 | } |
| 597 | |
| 598 | /*restore from the save area*/ |
| 599 | i = s->save_i; |
| 600 | j = s->save_j; |
| 601 | t = s->save_t; |
| 602 | alphaSize = s->save_alphaSize; |
| 603 | nGroups = s->save_nGroups; |
| 604 | nSelectors = s->save_nSelectors; |
| 605 | EOB = s->save_EOB; |
| 606 | groupNo = s->save_groupNo; |
| 607 | groupPos = s->save_groupPos; |
| 608 | nextSym = s->save_nextSym; |
| 609 | nblockMAX = s->save_nblockMAX; |
| 610 | nblock = s->save_nblock; |
| 611 | es = s->save_es; |
| 612 | N = s->save_N; |
| 613 | curr = s->save_curr; |
| 614 | zt = s->save_zt; |
| 615 | zn = s->save_zn; |
| 616 | zvec = s->save_zvec; |
| 617 | zj = s->save_zj; |
| 618 | gSel = s->save_gSel; |
| 619 | gMinlen = s->save_gMinlen; |
| 620 | gLimit = s->save_gLimit; |
| 621 | gBase = s->save_gBase; |
| 622 | gPerm = s->save_gPerm; |
| 623 | |
| 624 | retVal = BZ_OK; |
| 625 | switch_val = s->state; |
| 626 | switch (switch_val) { |
| 627 | case BZ_X_MAGIC_1: |
| 628 | s->state = BZ_X_MAGIC_1; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 629 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 630 | retVal = BZ_OK; |
| 631 | goto save_state_and_return; |
| 632 | } |
| 633 | if (uc != 'B') { |
| 634 | retVal = BZ_DATA_ERROR_MAGIC; |
| 635 | goto save_state_and_return; |
| 636 | } |
| 637 | |
| 638 | case BZ_X_MAGIC_2: |
| 639 | s->state = BZ_X_MAGIC_2; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 640 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 641 | retVal = BZ_OK; |
| 642 | goto save_state_and_return; |
| 643 | } |
| 644 | if (uc != 'Z') { |
| 645 | retVal = BZ_DATA_ERROR_MAGIC; |
| 646 | goto save_state_and_return; |
| 647 | } |
| 648 | |
| 649 | case BZ_X_MAGIC_3: |
| 650 | s->state = BZ_X_MAGIC_3; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 651 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 652 | retVal = BZ_OK; |
| 653 | goto save_state_and_return; |
| 654 | } |
| 655 | if (uc != 'h') { |
| 656 | retVal = BZ_DATA_ERROR_MAGIC; |
| 657 | goto save_state_and_return; |
| 658 | } |
| 659 | |
| 660 | case BZ_X_MAGIC_4: |
| 661 | s->state = BZ_X_MAGIC_4; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 662 | if (! get_bits(s, &s->blockSize100k, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 663 | retVal = BZ_OK; |
| 664 | goto save_state_and_return; |
| 665 | } |
| 666 | if ((s->blockSize100k < '1') || (s->blockSize100k > '9')) { |
| 667 | retVal = BZ_DATA_ERROR_MAGIC; |
| 668 | goto save_state_and_return; |
| 669 | } |
| 670 | s->blockSize100k -= '0'; |
| 671 | |
| 672 | if (s->smallDecompress) { |
| 673 | s->ll16 = (strm->bzalloc)(strm->opaque, s->blockSize100k * 100000 * sizeof(unsigned short), 1); |
| 674 | s->ll4 = (strm->bzalloc)(strm->opaque, ((1 + s->blockSize100k * 100000) >> 1) * sizeof(unsigned char), 1); |
| 675 | |
| 676 | if (s->ll16 == NULL || s->ll4 == NULL) { |
| 677 | retVal = BZ_MEM_ERROR; |
| 678 | goto save_state_and_return; |
| 679 | } |
| 680 | } else { |
| 681 | s->tt = (strm->bzalloc)(strm->opaque, s->blockSize100k * 100000 * sizeof(int), 1); |
| 682 | if (s->tt == NULL) { |
| 683 | retVal = BZ_MEM_ERROR; |
| 684 | goto save_state_and_return; |
| 685 | } |
| 686 | } |
| 687 | |
| 688 | case BZ_X_BLKHDR_1: |
| 689 | s->state = BZ_X_BLKHDR_1; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 690 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 691 | retVal = BZ_OK; |
| 692 | goto save_state_and_return; |
| 693 | } |
| 694 | |
| 695 | if (uc == 0x17) { |
| 696 | goto endhdr_2; |
| 697 | } |
| 698 | if (uc != 0x31) { |
| 699 | retVal = BZ_DATA_ERROR; |
| 700 | goto save_state_and_return; |
| 701 | } |
| 702 | |
| 703 | case BZ_X_BLKHDR_2: |
| 704 | s->state = BZ_X_BLKHDR_2; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 705 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 706 | retVal = BZ_OK; |
| 707 | goto save_state_and_return; |
| 708 | } |
| 709 | if (uc != 0x41) { |
| 710 | retVal = BZ_DATA_ERROR; |
| 711 | goto save_state_and_return; |
| 712 | } |
| 713 | |
| 714 | case BZ_X_BLKHDR_3: |
| 715 | s->state = BZ_X_BLKHDR_3; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 716 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 717 | retVal = BZ_OK; |
| 718 | goto save_state_and_return; |
| 719 | } |
| 720 | if (uc != 0x59) { |
| 721 | retVal = BZ_DATA_ERROR; |
| 722 | goto save_state_and_return; |
| 723 | } |
| 724 | |
| 725 | case BZ_X_BLKHDR_4: |
| 726 | s->state = BZ_X_BLKHDR_4; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 727 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 728 | retVal = BZ_OK; |
| 729 | goto save_state_and_return; |
| 730 | } |
| 731 | if (uc != 0x26) { |
| 732 | retVal = BZ_DATA_ERROR; |
| 733 | goto save_state_and_return; |
| 734 | } |
| 735 | |
| 736 | case BZ_X_BLKHDR_5: |
| 737 | s->state = BZ_X_BLKHDR_5; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 738 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 739 | retVal = BZ_OK; |
| 740 | goto save_state_and_return; |
| 741 | } |
| 742 | if (uc != 0x53) { |
| 743 | retVal = BZ_DATA_ERROR; |
| 744 | goto save_state_and_return; |
| 745 | } |
| 746 | |
| 747 | case BZ_X_BLKHDR_6: |
| 748 | s->state = BZ_X_BLKHDR_6; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 749 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 750 | retVal = BZ_OK; |
| 751 | goto save_state_and_return; |
| 752 | } |
| 753 | if (uc != 0x59) { |
| 754 | retVal = BZ_DATA_ERROR; |
| 755 | goto save_state_and_return; |
| 756 | } |
| 757 | |
| 758 | s->currBlockNo++; |
| 759 | if (s->verbosity >= 2) { |
| 760 | error_msg("\n [%d: huff+mtf ", s->currBlockNo); |
| 761 | } |
| 762 | s->storedBlockCRC = 0; |
| 763 | |
| 764 | case BZ_X_BCRC_1: |
| 765 | s->state = BZ_X_BCRC_1; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 766 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 767 | retVal = BZ_OK; |
| 768 | goto save_state_and_return; |
| 769 | } |
| 770 | s->storedBlockCRC = (s->storedBlockCRC << 8) | ((unsigned int)uc); |
| 771 | |
| 772 | case BZ_X_BCRC_2: |
| 773 | s->state = BZ_X_BCRC_2; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 774 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 775 | retVal = BZ_OK; |
| 776 | goto save_state_and_return; |
| 777 | } |
| 778 | s->storedBlockCRC = (s->storedBlockCRC << 8) | ((unsigned int)uc); |
| 779 | |
| 780 | case BZ_X_BCRC_3: |
| 781 | s->state = BZ_X_BCRC_3; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 782 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 783 | retVal = BZ_OK; |
| 784 | goto save_state_and_return; |
| 785 | } |
| 786 | s->storedBlockCRC = (s->storedBlockCRC << 8) | ((unsigned int)uc); |
| 787 | |
| 788 | case BZ_X_BCRC_4: |
| 789 | s->state = BZ_X_BCRC_4; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 790 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 791 | retVal = BZ_OK; |
| 792 | goto save_state_and_return; |
| 793 | } |
| 794 | s->storedBlockCRC = (s->storedBlockCRC << 8) | ((unsigned int)uc); |
| 795 | |
| 796 | case BZ_X_RANDBIT: |
| 797 | s->state = BZ_X_RANDBIT; |
| 798 | { |
| 799 | int tmp = s->blockRandomised; |
| 800 | const int ret = get_bits(s, &tmp, 1); |
| 801 | s->blockRandomised = tmp; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 802 | if (! ret) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 803 | retVal = BZ_OK; |
| 804 | goto save_state_and_return; |
| 805 | } |
| 806 | } |
| 807 | |
| 808 | s->origPtr = 0; |
| 809 | |
| 810 | case BZ_X_ORIGPTR_1: |
| 811 | s->state = BZ_X_ORIGPTR_1; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 812 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 813 | retVal = BZ_OK; |
| 814 | goto save_state_and_return; |
| 815 | } |
| 816 | s->origPtr = (s->origPtr << 8) | ((int)uc); |
| 817 | |
| 818 | case BZ_X_ORIGPTR_2: |
| 819 | s->state = BZ_X_ORIGPTR_2; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 820 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 821 | retVal = BZ_OK; |
| 822 | goto save_state_and_return; |
| 823 | } |
| 824 | s->origPtr = (s->origPtr << 8) | ((int)uc); |
| 825 | |
| 826 | case BZ_X_ORIGPTR_3: |
| 827 | s->state = BZ_X_ORIGPTR_3; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 828 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 829 | retVal = BZ_OK; |
| 830 | goto save_state_and_return; |
| 831 | } |
| 832 | s->origPtr = (s->origPtr << 8) | ((int)uc); |
| 833 | |
| 834 | if (s->origPtr < 0) { |
| 835 | retVal = BZ_DATA_ERROR; |
| 836 | goto save_state_and_return; |
| 837 | } |
| 838 | if (s->origPtr > 10 + 100000*s->blockSize100k) { |
| 839 | retVal = BZ_DATA_ERROR; |
| 840 | goto save_state_and_return; |
| 841 | } |
| 842 | |
| 843 | /*--- Receive the mapping table ---*/ |
| 844 | case BZ_X_MAPPING_1: |
| 845 | for (i = 0; i < 16; i++) { |
| 846 | s->state = BZ_X_MAPPING_1; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 847 | if (! get_bits(s, &uc, 1)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 848 | retVal = BZ_OK; |
| 849 | goto save_state_and_return; |
| 850 | } |
| 851 | if (uc == 1) { |
| 852 | s->inUse16[i] = TRUE; |
| 853 | } else { |
| 854 | s->inUse16[i] = FALSE; |
| 855 | } |
| 856 | } |
| 857 | |
| 858 | for (i = 0; i < 256; i++) { |
| 859 | s->inUse[i] = FALSE; |
| 860 | } |
| 861 | |
| 862 | for (i = 0; i < 16; i++) { |
| 863 | if (s->inUse16[i]) { |
| 864 | for (j = 0; j < 16; j++) { |
| 865 | case BZ_X_MAPPING_2: |
| 866 | s->state = BZ_X_MAPPING_2; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 867 | if (! get_bits(s, &uc, 1)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 868 | retVal = BZ_OK; |
| 869 | goto save_state_and_return; |
| 870 | } |
| 871 | if (uc == 1) { |
| 872 | s->inUse[i * 16 + j] = TRUE; |
| 873 | } |
| 874 | } |
| 875 | } |
| 876 | } |
| 877 | |
| 878 | s->nInUse = 0; |
| 879 | for (i = 0; i < 256; i++) { |
| 880 | if (s->inUse[i]) { |
| 881 | s->seqToUnseq[s->nInUse] = i; |
| 882 | s->nInUse++; |
| 883 | } |
| 884 | } |
| 885 | if (s->nInUse == 0) { |
| 886 | retVal = BZ_DATA_ERROR; |
| 887 | goto save_state_and_return; |
| 888 | } |
| 889 | alphaSize = s->nInUse+2; |
| 890 | |
| 891 | /*--- Now the selectors ---*/ |
| 892 | case BZ_X_SELECTOR_1: |
| 893 | s->state = BZ_X_SELECTOR_1; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 894 | if (! get_bits(s, &nGroups, 3)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 895 | retVal = BZ_OK; |
| 896 | goto save_state_and_return; |
| 897 | } |
| 898 | if (nGroups < 2 || nGroups > 6) { |
| 899 | retVal = BZ_DATA_ERROR; |
| 900 | goto save_state_and_return; |
| 901 | } |
| 902 | |
| 903 | case BZ_X_SELECTOR_2: |
| 904 | s->state = BZ_X_SELECTOR_2; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 905 | if (! get_bits(s, &nSelectors, 15)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 906 | retVal = BZ_OK; |
| 907 | goto save_state_and_return; |
| 908 | } |
| 909 | if (nSelectors < 1) { |
| 910 | retVal = BZ_DATA_ERROR; |
| 911 | goto save_state_and_return; |
| 912 | } |
| 913 | |
| 914 | |
| 915 | |
| 916 | for (i = 0; i < nSelectors; i++) { |
| 917 | j = 0; |
| 918 | while (1) { |
| 919 | case BZ_X_SELECTOR_3: |
| 920 | s->state = BZ_X_SELECTOR_3; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 921 | if (! get_bits(s, &uc, 1)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 922 | retVal = BZ_OK; |
| 923 | goto save_state_and_return; |
| 924 | } |
| 925 | if (uc == 0) { |
| 926 | break; |
| 927 | } |
| 928 | j++; |
| 929 | if (j >= nGroups) { |
| 930 | retVal = BZ_DATA_ERROR; |
| 931 | goto save_state_and_return; |
| 932 | } |
| 933 | } |
| 934 | s->selectorMtf[i] = j; |
| 935 | } |
| 936 | |
| 937 | /*--- Undo the MTF values for the selectors. ---*/ |
| 938 | { |
| 939 | unsigned char pos[BZ_N_GROUPS], tmp, v; |
| 940 | for (v = 0; v < nGroups; v++) { |
| 941 | pos[v] = v; |
| 942 | } |
| 943 | for (i = 0; i < nSelectors; i++) { |
| 944 | v = s->selectorMtf[i]; |
| 945 | tmp = pos[v]; |
| 946 | while (v > 0) { |
| 947 | pos[v] = pos[v-1]; |
| 948 | v--; |
| 949 | } |
| 950 | pos[0] = tmp; |
| 951 | s->selector[i] = tmp; |
| 952 | } |
| 953 | } |
| 954 | |
| 955 | /*--- Now the coding tables ---*/ |
| 956 | for (t = 0; t < nGroups; t++) { |
| 957 | case BZ_X_CODING_1: |
| 958 | s->state = BZ_X_CODING_1; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 959 | if (! get_bits(s, &curr, 5)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 960 | retVal = BZ_OK; |
| 961 | goto save_state_and_return; |
| 962 | } |
| 963 | for (i = 0; i < alphaSize; i++) { |
| 964 | while (TRUE) { |
| 965 | if (curr < 1 || curr > 20) { |
| 966 | retVal = BZ_DATA_ERROR; |
| 967 | goto save_state_and_return; |
| 968 | } |
| 969 | |
| 970 | case BZ_X_CODING_2: |
| 971 | s->state = BZ_X_CODING_2; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 972 | if (! get_bits(s, &uc, 1)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 973 | retVal = BZ_OK; |
| 974 | goto save_state_and_return; |
| 975 | } |
| 976 | if (uc == 0) { |
| 977 | break; |
| 978 | } |
| 979 | |
| 980 | case BZ_X_CODING_3: |
| 981 | s->state = BZ_X_CODING_3; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 982 | if (! get_bits(s, &uc, 1)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 983 | retVal = BZ_OK; |
| 984 | goto save_state_and_return; |
| 985 | } |
| 986 | if (uc == 0) { |
| 987 | curr++; |
| 988 | } else { |
| 989 | curr--; |
| 990 | } |
| 991 | } |
| 992 | s->len[t][i] = curr; |
| 993 | } |
| 994 | } |
| 995 | |
| 996 | /*--- Create the Huffman decoding tables ---*/ |
| 997 | for (t = 0; t < nGroups; t++) { |
| 998 | minLen = 32; |
| 999 | maxLen = 0; |
| 1000 | for (i = 0; i < alphaSize; i++) { |
| 1001 | if (s->len[t][i] > maxLen) { |
| 1002 | maxLen = s->len[t][i]; |
| 1003 | } |
| 1004 | if (s->len[t][i] < minLen) { |
| 1005 | minLen = s->len[t][i]; |
| 1006 | } |
| 1007 | } |
| 1008 | |
| 1009 | BZ2_hbCreateDecodeTables ( |
| 1010 | &(s->limit[t][0]), |
| 1011 | &(s->base[t][0]), |
| 1012 | &(s->perm[t][0]), |
| 1013 | &(s->len[t][0]), |
| 1014 | minLen, maxLen, alphaSize |
| 1015 | ); |
| 1016 | |
| 1017 | |
| 1018 | s->minLens[t] = minLen; |
| 1019 | } |
| 1020 | |
| 1021 | /*--- Now the MTF values ---*/ |
| 1022 | |
| 1023 | EOB = s->nInUse+1; |
| 1024 | nblockMAX = 100000 * s->blockSize100k; |
| 1025 | groupNo = -1; |
| 1026 | groupPos = 0; |
| 1027 | |
| 1028 | for (i = 0; i <= 255; i++) { |
| 1029 | s->unzftab[i] = 0; |
| 1030 | } |
| 1031 | /*-- MTF init --*/ |
| 1032 | { |
| 1033 | int ii, jj, kk; |
| 1034 | kk = MTFA_SIZE-1; |
| 1035 | for (ii = 256 / MTFL_SIZE - 1; ii >= 0; ii--) { |
| 1036 | for (jj = MTFL_SIZE-1; jj >= 0; jj--) { |
| 1037 | s->mtfa[kk] = (unsigned char)(ii * MTFL_SIZE + jj); |
| 1038 | kk--; |
| 1039 | } |
| 1040 | s->mtfbase[ii] = kk + 1; |
| 1041 | } |
| 1042 | } |
| 1043 | /*-- end MTF init --*/ |
| 1044 | |
| 1045 | nblock = 0; |
| 1046 | |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1047 | if (! get_mtf_val_init()) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1048 | goto save_state_and_return; |
| 1049 | } |
| 1050 | case BZ_X_MTF_1: |
| 1051 | s->state = BZ_X_MTF_1; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1052 | if (! get_bits(s, &zvec, zn)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1053 | retVal = BZ_OK; |
| 1054 | goto save_state_and_return; |
| 1055 | } |
| 1056 | while (1) { |
| 1057 | if (zn > 20 /* the longest code */) { |
| 1058 | retVal = BZ_DATA_ERROR; |
| 1059 | goto save_state_and_return; |
| 1060 | } |
| 1061 | if (zvec <= gLimit[zn]) { |
| 1062 | break; |
| 1063 | } |
| 1064 | zn++; |
| 1065 | |
| 1066 | case BZ_X_MTF_2: |
| 1067 | s->state = BZ_X_MTF_2; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1068 | if (! get_bits(s, &zj, 1)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1069 | retVal = BZ_OK; |
| 1070 | goto save_state_and_return; |
| 1071 | } |
| 1072 | zvec = (zvec << 1) | zj; |
| 1073 | } |
| 1074 | if (zvec - gBase[zn] < 0 || zvec - gBase[zn] >= BZ_MAX_ALPHA_SIZE) { |
| 1075 | retVal = BZ_DATA_ERROR; |
| 1076 | goto save_state_and_return; |
| 1077 | } |
| 1078 | nextSym = gPerm[zvec - gBase[zn]]; |
| 1079 | |
| 1080 | while (1) { |
| 1081 | if (nextSym == EOB) { |
| 1082 | break; |
| 1083 | } |
| 1084 | |
| 1085 | if (nextSym == BZ_RUNA || nextSym == BZ_RUNB) { |
| 1086 | es = -1; |
| 1087 | N = 1; |
| 1088 | do { |
| 1089 | if (nextSym == BZ_RUNA) { |
| 1090 | es = es + (0+1) * N; |
| 1091 | } else { |
| 1092 | if (nextSym == BZ_RUNB) { |
| 1093 | es = es + (1+1) * N; |
| 1094 | } |
| 1095 | } |
| 1096 | N = N * 2; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1097 | if (! get_mtf_val_init()) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1098 | goto save_state_and_return; |
| 1099 | } |
| 1100 | case BZ_X_MTF_3: |
| 1101 | s->state = BZ_X_MTF_3; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1102 | if (! get_bits(s, &zvec, zn)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1103 | retVal = BZ_OK; |
| 1104 | goto save_state_and_return; |
| 1105 | } |
| 1106 | while (1) { |
| 1107 | if (zn > 20 /* the longest code */) { |
| 1108 | retVal = BZ_DATA_ERROR; |
| 1109 | goto save_state_and_return; |
| 1110 | } |
| 1111 | if (zvec <= gLimit[zn]) { |
| 1112 | break; |
| 1113 | } |
| 1114 | zn++; |
| 1115 | |
| 1116 | case BZ_X_MTF_4: |
| 1117 | s->state = BZ_X_MTF_4; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1118 | if (! get_bits(s, &zj, 1)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1119 | retVal = BZ_OK; |
| 1120 | goto save_state_and_return; |
| 1121 | } |
| 1122 | zvec = (zvec << 1) | zj; |
| 1123 | } |
| 1124 | if (zvec - gBase[zn] < 0 || zvec - gBase[zn] >= BZ_MAX_ALPHA_SIZE) { |
| 1125 | retVal = BZ_DATA_ERROR; |
| 1126 | goto save_state_and_return; |
| 1127 | |
| 1128 | } |
| 1129 | nextSym = gPerm[zvec - gBase[zn]]; |
| 1130 | } |
| 1131 | while (nextSym == BZ_RUNA || nextSym == BZ_RUNB); |
| 1132 | |
| 1133 | es++; |
| 1134 | uc = s->seqToUnseq[ s->mtfa[s->mtfbase[0]] ]; |
| 1135 | s->unzftab[uc] += es; |
| 1136 | |
| 1137 | if (s->smallDecompress) { |
| 1138 | while (es > 0) { |
| 1139 | if (nblock >= nblockMAX) { |
| 1140 | retVal = BZ_DATA_ERROR; |
| 1141 | goto save_state_and_return; |
| 1142 | } |
| 1143 | s->ll16[nblock] = (unsigned short)uc; |
| 1144 | nblock++; |
| 1145 | es--; |
| 1146 | } |
| 1147 | } else { |
| 1148 | while (es > 0) { |
| 1149 | if (nblock >= nblockMAX) { |
| 1150 | retVal = BZ_DATA_ERROR; |
| 1151 | goto save_state_and_return; |
| 1152 | } |
| 1153 | s->tt[nblock] = (unsigned int)uc; |
| 1154 | nblock++; |
| 1155 | es--; |
| 1156 | } |
| 1157 | } |
| 1158 | continue; |
| 1159 | } else { |
| 1160 | if (nblock >= nblockMAX) { |
| 1161 | retVal = BZ_DATA_ERROR; |
| 1162 | goto save_state_and_return; |
| 1163 | } |
| 1164 | /*-- uc = MTF ( nextSym-1 ) --*/ |
| 1165 | { |
| 1166 | int ii, jj, kk, pp, lno, off; |
| 1167 | unsigned int nn; |
| 1168 | nn = (unsigned int)(nextSym - 1); |
| 1169 | |
| 1170 | if (nn < MTFL_SIZE) { |
| 1171 | /* avoid general-case expense */ |
| 1172 | pp = s->mtfbase[0]; |
| 1173 | uc = s->mtfa[pp+nn]; |
| 1174 | while (nn > 3) { |
| 1175 | int z = pp+nn; |
| 1176 | s->mtfa[(z) ] = s->mtfa[(z)-1]; |
| 1177 | s->mtfa[(z)-1] = s->mtfa[(z)-2]; |
| 1178 | s->mtfa[(z)-2] = s->mtfa[(z)-3]; |
| 1179 | s->mtfa[(z)-3] = s->mtfa[(z)-4]; |
| 1180 | nn -= 4; |
| 1181 | } |
| 1182 | while (nn > 0) { |
| 1183 | s->mtfa[(pp+nn)] = s->mtfa[(pp+nn)-1]; nn--; |
| 1184 | } |
| 1185 | s->mtfa[pp] = uc; |
| 1186 | } else { |
| 1187 | /* general case */ |
| 1188 | lno = nn / MTFL_SIZE; |
| 1189 | off = nn % MTFL_SIZE; |
| 1190 | pp = s->mtfbase[lno] + off; |
| 1191 | uc = s->mtfa[pp]; |
| 1192 | while (pp > s->mtfbase[lno]) { |
| 1193 | s->mtfa[pp] = s->mtfa[pp-1]; |
| 1194 | pp--; |
| 1195 | } |
| 1196 | s->mtfbase[lno]++; |
| 1197 | while (lno > 0) { |
| 1198 | s->mtfbase[lno]--; |
| 1199 | s->mtfa[s->mtfbase[lno]] = s->mtfa[s->mtfbase[lno-1] + MTFL_SIZE - 1]; |
| 1200 | lno--; |
| 1201 | } |
| 1202 | s->mtfbase[0]--; |
| 1203 | s->mtfa[s->mtfbase[0]] = uc; |
| 1204 | if (s->mtfbase[0] == 0) { |
| 1205 | kk = MTFA_SIZE-1; |
| 1206 | for (ii = 256 / MTFL_SIZE-1; ii >= 0; ii--) { |
| 1207 | for (jj = MTFL_SIZE-1; jj >= 0; jj--) { |
| 1208 | s->mtfa[kk] = s->mtfa[s->mtfbase[ii] + jj]; |
| 1209 | kk--; |
| 1210 | } |
| 1211 | s->mtfbase[ii] = kk + 1; |
| 1212 | } |
| 1213 | } |
| 1214 | } |
| 1215 | } |
| 1216 | /*-- end uc = MTF ( nextSym-1 ) --*/ |
| 1217 | |
| 1218 | s->unzftab[s->seqToUnseq[uc]]++; |
| 1219 | if (s->smallDecompress) { |
| 1220 | s->ll16[nblock] = (unsigned short)(s->seqToUnseq[uc]); |
| 1221 | } else { |
| 1222 | s->tt[nblock] = (unsigned int)(s->seqToUnseq[uc]); |
| 1223 | } |
| 1224 | nblock++; |
| 1225 | |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1226 | if (! get_mtf_val_init()) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1227 | goto save_state_and_return; |
| 1228 | } |
| 1229 | case BZ_X_MTF_5: |
| 1230 | s->state = BZ_X_MTF_5; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1231 | if (! get_bits(s, &zvec, zn)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1232 | retVal = BZ_OK; |
| 1233 | goto save_state_and_return; |
| 1234 | } |
| 1235 | while (1) { |
| 1236 | if (zn > 20 /* the longest code */) { |
| 1237 | retVal = BZ_DATA_ERROR; |
| 1238 | goto save_state_and_return; |
| 1239 | } |
| 1240 | if (zvec <= gLimit[zn]) { |
| 1241 | break; |
| 1242 | } |
| 1243 | zn++; |
| 1244 | |
| 1245 | case BZ_X_MTF_6: |
| 1246 | s->state = BZ_X_MTF_6; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1247 | if (! get_bits(s, &zj, 1)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1248 | retVal = BZ_OK; |
| 1249 | goto save_state_and_return; |
| 1250 | } |
| 1251 | zvec = (zvec << 1) | zj; |
| 1252 | } |
| 1253 | if (zvec - gBase[zn] < 0 || zvec - gBase[zn] >= BZ_MAX_ALPHA_SIZE) { |
| 1254 | retVal = BZ_DATA_ERROR; |
| 1255 | goto save_state_and_return; |
| 1256 | } |
| 1257 | nextSym = gPerm[zvec - gBase[zn]]; |
| 1258 | continue; |
| 1259 | } |
| 1260 | } |
| 1261 | |
| 1262 | /* Now we know what nblock is, we can do a better sanity |
| 1263 | check on s->origPtr. |
| 1264 | */ |
| 1265 | if (s->origPtr < 0 || s->origPtr >= nblock) { |
| 1266 | retVal = BZ_DATA_ERROR; |
| 1267 | goto save_state_and_return; |
| 1268 | } |
| 1269 | s->state_out_len = 0; |
| 1270 | s->state_out_ch = 0; |
| 1271 | s->calculatedBlockCRC = 0xffffffffL; |
| 1272 | s->state = BZ_X_OUTPUT; |
| 1273 | if (s->verbosity >= 2) { |
| 1274 | error_msg("rt+rld"); |
| 1275 | } |
| 1276 | |
| 1277 | /*-- Set up cftab to facilitate generation of T^(-1) --*/ |
| 1278 | s->cftab[0] = 0; |
| 1279 | for (i = 1; i <= 256; i++) { |
| 1280 | s->cftab[i] = s->unzftab[i-1]; |
| 1281 | } |
| 1282 | for (i = 1; i <= 256; i++) { |
| 1283 | s->cftab[i] += s->cftab[i-1]; |
| 1284 | } |
| 1285 | |
| 1286 | if (s->smallDecompress) { |
| 1287 | |
| 1288 | /*-- Make a copy of cftab, used in generation of T --*/ |
| 1289 | for (i = 0; i <= 256; i++) { |
| 1290 | s->cftabCopy[i] = s->cftab[i]; |
| 1291 | } |
| 1292 | |
| 1293 | /*-- compute the T vector --*/ |
| 1294 | for (i = 0; i < nblock; i++) { |
| 1295 | uc = (unsigned char)(s->ll16[i]); |
| 1296 | s->ll16[i] = (unsigned short)(s->cftabCopy[uc] & 0x0000ffff); |
| 1297 | if (((i) & 0x1) == 0) { |
| 1298 | s->ll4[(i) >> 1] = (s->ll4[(i) >> 1] & 0xf0) | (s->cftabCopy[uc] >> 16); |
| 1299 | } else { |
| 1300 | s->ll4[(i) >> 1] = (s->ll4[(i) >> 1] & 0x0f) | ((s->cftabCopy[uc] >> 16) << 4); |
| 1301 | } |
| 1302 | s->cftabCopy[uc]++; |
| 1303 | } |
| 1304 | |
| 1305 | /*-- Compute T^(-1) by pointer reversal on T --*/ |
| 1306 | i = s->origPtr; |
| 1307 | j = (((unsigned int)s->ll16[i]) | |
| 1308 | (((((unsigned int)(s->ll4[(i) >> 1])) >> |
| 1309 | (((i) << 2) & 0x4)) & 0xF) << 16)); |
| 1310 | |
| 1311 | do { |
| 1312 | const int tmp = (((unsigned int)s->ll16[j]) | |
| 1313 | (((((unsigned int)(s->ll4[(j) >> 1])) >> |
| 1314 | (((j) << 2) & 0x4)) & 0xF) << 16)); |
| 1315 | |
| 1316 | s->ll16[j] = (unsigned short)(i & 0x0000ffff); |
| 1317 | if (((j) & 0x1) == 0) { |
| 1318 | s->ll4[(j) >> 1] = (s->ll4[(j) >> 1] & 0xf0) | (i >> 16); |
| 1319 | } else { |
| 1320 | s->ll4[(j) >> 1] = (s->ll4[(j) >> 1] & 0x0f) | ((i >> 16) << 4); |
| 1321 | } |
| 1322 | i = j; |
| 1323 | j = tmp; |
| 1324 | } |
| 1325 | while (i != s->origPtr); |
| 1326 | s->tPos = s->origPtr; |
| 1327 | s->nblock_used = 0; |
| 1328 | if (s->blockRandomised) { |
| 1329 | s->rNToGo = 0; |
| 1330 | s->rTPos = 0; |
| 1331 | s->k0 = bz_get_small(s); |
| 1332 | s->nblock_used++; |
| 1333 | bz_rand_udp_mask(s); |
| 1334 | s->k0 ^= ((s->rNToGo == 1) ? 1 : 0); |
| 1335 | } else { |
| 1336 | s->k0 = bz_get_small(s); |
| 1337 | s->nblock_used++; |
| 1338 | } |
| 1339 | } else { |
| 1340 | /*-- compute the T^(-1) vector --*/ |
| 1341 | for (i = 0; i < nblock; i++) { |
| 1342 | uc = (unsigned char)(s->tt[i] & 0xff); |
| 1343 | s->tt[s->cftab[uc]] |= (i << 8); |
| 1344 | s->cftab[uc]++; |
| 1345 | } |
| 1346 | |
| 1347 | s->tPos = s->tt[s->origPtr] >> 8; |
| 1348 | s->nblock_used = 0; |
| 1349 | if (s->blockRandomised) { |
| 1350 | s->rNToGo = 0; |
| 1351 | s->rTPos = 0; |
| 1352 | s->k0 = bz_get_fast(s); |
| 1353 | |
| 1354 | s->nblock_used++; |
| 1355 | bz_rand_udp_mask(s); |
| 1356 | s->k0 ^= ((s->rNToGo == 1) ? 1 : 0); |
| 1357 | } else { |
| 1358 | s->k0 = bz_get_fast(s); |
| 1359 | s->nblock_used++; |
| 1360 | } |
| 1361 | } |
| 1362 | |
| 1363 | retVal = BZ_OK; |
| 1364 | goto save_state_and_return; |
| 1365 | |
| 1366 | endhdr_2: |
| 1367 | case BZ_X_ENDHDR_2: |
| 1368 | s->state = BZ_X_ENDHDR_2; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1369 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1370 | retVal = BZ_OK; |
| 1371 | goto save_state_and_return; |
| 1372 | } |
| 1373 | if (uc != 0x72) { |
| 1374 | retVal = BZ_DATA_ERROR; |
| 1375 | goto save_state_and_return; |
| 1376 | } |
| 1377 | |
| 1378 | case BZ_X_ENDHDR_3: |
| 1379 | s->state = BZ_X_ENDHDR_3; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1380 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1381 | retVal = BZ_OK; |
| 1382 | goto save_state_and_return; |
| 1383 | } |
| 1384 | if (uc != 0x45) { |
| 1385 | retVal = BZ_DATA_ERROR; |
| 1386 | goto save_state_and_return; |
| 1387 | } |
| 1388 | |
| 1389 | case BZ_X_ENDHDR_4: |
| 1390 | s->state = BZ_X_ENDHDR_4; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1391 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1392 | retVal = BZ_OK; |
| 1393 | goto save_state_and_return; |
| 1394 | } |
| 1395 | if (uc != 0x38) { |
| 1396 | retVal = BZ_DATA_ERROR; |
| 1397 | goto save_state_and_return; |
| 1398 | } |
| 1399 | |
| 1400 | case BZ_X_ENDHDR_5: |
| 1401 | s->state = BZ_X_ENDHDR_5; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1402 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1403 | retVal = BZ_OK; |
| 1404 | goto save_state_and_return; |
| 1405 | } |
| 1406 | if (uc != 0x50) { |
| 1407 | retVal = BZ_DATA_ERROR; |
| 1408 | goto save_state_and_return; |
| 1409 | } |
| 1410 | |
| 1411 | case BZ_X_ENDHDR_6: |
| 1412 | s->state = BZ_X_ENDHDR_6; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1413 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1414 | retVal = BZ_OK; |
| 1415 | goto save_state_and_return; |
| 1416 | } |
| 1417 | if (uc != 0x90) { |
| 1418 | retVal = BZ_DATA_ERROR; |
| 1419 | goto save_state_and_return; |
| 1420 | } |
| 1421 | s->storedCombinedCRC = 0; |
| 1422 | |
| 1423 | case BZ_X_CCRC_1: |
| 1424 | s->state = BZ_X_CCRC_1; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1425 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1426 | retVal = BZ_OK; |
| 1427 | goto save_state_and_return; |
| 1428 | } |
| 1429 | s->storedCombinedCRC = (s->storedCombinedCRC << 8) | ((unsigned int)uc); |
| 1430 | case BZ_X_CCRC_2: |
| 1431 | s->state = BZ_X_CCRC_2; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1432 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1433 | retVal = BZ_OK; |
| 1434 | goto save_state_and_return; |
| 1435 | } |
| 1436 | s->storedCombinedCRC = (s->storedCombinedCRC << 8) | ((unsigned int)uc); |
| 1437 | |
| 1438 | case BZ_X_CCRC_3: |
| 1439 | s->state = BZ_X_CCRC_3; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1440 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1441 | retVal = BZ_OK; |
| 1442 | goto save_state_and_return; |
| 1443 | } |
| 1444 | s->storedCombinedCRC = (s->storedCombinedCRC << 8) | ((unsigned int)uc); |
| 1445 | |
| 1446 | case BZ_X_CCRC_4: |
| 1447 | s->state = BZ_X_CCRC_4; |
Matt Kraai | 1f0c436 | 2001-12-20 23:13:26 +0000 | [diff] [blame] | 1448 | if (! get_bits(s, &uc, 8)) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 1449 | retVal = BZ_OK; |
| 1450 | goto save_state_and_return; |
| 1451 | } |
| 1452 | s->storedCombinedCRC = (s->storedCombinedCRC << 8) | ((unsigned int)uc); |
| 1453 | |
| 1454 | s->state = BZ_X_IDLE; |
| 1455 | retVal = BZ_STREAM_END; |
| 1456 | goto save_state_and_return; |
| 1457 | |
| 1458 | default: |
| 1459 | printf("switch val is %d\n", switch_val); |
| 1460 | assert_h(4001); |
| 1461 | } |
| 1462 | |
| 1463 | assert_h(4002); |
| 1464 | |
| 1465 | save_state_and_return: |
| 1466 | s->save_i = i; |
| 1467 | s->save_j = j; |
| 1468 | s->save_t = t; |
| 1469 | s->save_alphaSize = alphaSize; |
| 1470 | s->save_nGroups = nGroups; |
| 1471 | s->save_nSelectors = nSelectors; |
| 1472 | s->save_EOB = EOB; |
| 1473 | s->save_groupNo = groupNo; |
| 1474 | s->save_groupPos = groupPos; |
| 1475 | s->save_nextSym = nextSym; |
| 1476 | s->save_nblockMAX = nblockMAX; |
| 1477 | s->save_nblock = nblock; |
| 1478 | s->save_es = es; |
| 1479 | s->save_N = N; |
| 1480 | s->save_curr = curr; |
| 1481 | s->save_zt = zt; |
| 1482 | s->save_zn = zn; |
| 1483 | s->save_zvec = zvec; |
| 1484 | s->save_zj = zj; |
| 1485 | s->save_gSel = gSel; |
| 1486 | s->save_gMinlen = gMinlen; |
| 1487 | s->save_gLimit = gLimit; |
| 1488 | s->save_gBase = gBase; |
| 1489 | s->save_gPerm = gPerm; |
| 1490 | |
| 1491 | return retVal; |
| 1492 | } |
| 1493 | |
| 1494 | static void *default_bzalloc(void *opaque, int items, int size) |
| 1495 | { |
| 1496 | void *v = xmalloc(items *size); |
| 1497 | return v; |
| 1498 | } |
| 1499 | |
| 1500 | static void default_bzfree(void *opaque, void *addr) |
| 1501 | { |
| 1502 | if (addr != NULL) { |
| 1503 | free(addr); |
| 1504 | } |
| 1505 | } |
| 1506 | |
| 1507 | //int BZ2_bzDecompressInit(bz_stream* strm, int verbosity_level, int small) |
| 1508 | int BZ2_bzDecompressInit(bz_stream* strm, int small) |
| 1509 | { |
| 1510 | DState* s; |
| 1511 | |
| 1512 | if (sizeof(int) != 4) { |
| 1513 | return BZ_CONFIG_ERROR; |
| 1514 | } |
| 1515 | if (sizeof(short) != 2) { |
| 1516 | return BZ_CONFIG_ERROR; |
| 1517 | } |
| 1518 | if (sizeof(char) != 1) { |
| 1519 | return BZ_CONFIG_ERROR; |
| 1520 | } |
| 1521 | if (strm == NULL) { |
| 1522 | return BZ_PARAM_ERROR; |
| 1523 | } |
| 1524 | if (small != 0 && small != 1) { |
| 1525 | return BZ_PARAM_ERROR; |
| 1526 | } |
| 1527 | // if (verbosity_level < 0 || verbosity_level > 4) { |
| 1528 | // return BZ_PARAM_ERROR; |
| 1529 | // } |
| 1530 | if (strm->bzalloc == NULL) { |
| 1531 | strm->bzalloc = default_bzalloc; |
| 1532 | } |
| 1533 | if (strm->bzfree == NULL) { |
| 1534 | strm->bzfree = default_bzfree; |
| 1535 | } |
| 1536 | s = (strm->bzalloc)(strm->opaque, sizeof(DState), 1); |
| 1537 | if (s == NULL) { |
| 1538 | return BZ_MEM_ERROR; |
| 1539 | } |
| 1540 | s->strm = strm; |
| 1541 | strm->state = s; |
| 1542 | s->state = BZ_X_MAGIC_1; |
| 1543 | s->bsLive = 0; |
| 1544 | s->bsBuff = 0; |
| 1545 | s->calculatedCombinedCRC = 0; |
| 1546 | strm->total_in_lo32 = 0; |
| 1547 | strm->total_in_hi32 = 0; |
| 1548 | strm->total_out_lo32 = 0; |
| 1549 | strm->total_out_hi32 = 0; |
| 1550 | s->smallDecompress = (unsigned char)small; |
| 1551 | s->ll4 = NULL; |
| 1552 | s->ll16 = NULL; |
| 1553 | s->tt = NULL; |
| 1554 | s->currBlockNo = 0; |
| 1555 | // s->verbosity = verbosity_level; |
| 1556 | |
| 1557 | return BZ_OK; |
| 1558 | } |
| 1559 | |
| 1560 | void bz_seterr(int eee, int *bzerror, bzFile **bzf) |
| 1561 | { |
| 1562 | if (bzerror != NULL) { |
| 1563 | *bzerror = eee; |
| 1564 | } |
| 1565 | if (*bzf != NULL) { |
| 1566 | (*bzf)->lastErr = eee; |
| 1567 | } |
| 1568 | } |
| 1569 | |
| 1570 | void BZ2_bzReadClose(int *bzerror, void *b) |
| 1571 | { |
| 1572 | bzFile* bzf = (bzFile*)b; |
| 1573 | |
| 1574 | bz_seterr(BZ_OK, bzerror, &bzf); |
| 1575 | if (bzf == NULL) { |
| 1576 | bz_seterr(BZ_OK, bzerror, &bzf); |
| 1577 | return; |
| 1578 | } |
| 1579 | |
| 1580 | if (bzf->writing) { |
| 1581 | bz_seterr(BZ_SEQUENCE_ERROR, bzerror, &bzf); |
| 1582 | return; |
| 1583 | } |
| 1584 | |
| 1585 | if (bzf->initialisedOk) { |
| 1586 | bz_stream *strm = &(bzf->strm); |
| 1587 | DState *s; |
| 1588 | if (strm == NULL) { |
| 1589 | return; |
| 1590 | } |
| 1591 | s = strm->state; |
| 1592 | if ((s == NULL) || (s->strm != strm)) { |
| 1593 | return; |
| 1594 | } |
| 1595 | if (s->tt != NULL) { |
| 1596 | (strm->bzfree)(strm->opaque,(s->tt)); |
| 1597 | } |
| 1598 | if (s->ll16 != NULL) { |
| 1599 | (strm->bzfree)(strm->opaque,(s->ll16)); |
| 1600 | } |
| 1601 | if (s->ll4 != NULL) { |
| 1602 | (strm->bzfree)(strm->opaque,(s->ll4)); |
| 1603 | } |
| 1604 | (strm->bzfree)(strm->opaque,(strm->state)); |
| 1605 | strm->state = NULL; |
| 1606 | return; |
| 1607 | } |
| 1608 | free(bzf); |
| 1609 | } |
| 1610 | |
| 1611 | static void unRLE_obuf_to_output_FAST(DState *s) |
| 1612 | { |
| 1613 | unsigned char k1; |
| 1614 | |
| 1615 | if (s->blockRandomised) { |
| 1616 | while (1) { |
| 1617 | /* try to finish existing run */ |
| 1618 | while (1) { |
| 1619 | if (s->strm->avail_out == 0) { |
| 1620 | return; |
| 1621 | } |
| 1622 | if (s->state_out_len == 0) { |
| 1623 | break; |
| 1624 | } |
| 1625 | *((unsigned char *)(s->strm->next_out)) = s->state_out_ch; |
| 1626 | s->calculatedBlockCRC = (s->calculatedBlockCRC << 8) ^ |
| 1627 | BZ2_crc32Table[(s->calculatedBlockCRC >> 24) ^ |
| 1628 | ((unsigned char)s->state_out_ch)]; |
| 1629 | s->state_out_len--; |
| 1630 | s->strm->next_out++; |
| 1631 | s->strm->avail_out--; |
| 1632 | s->strm->total_out_lo32++; |
| 1633 | if (s->strm->total_out_lo32 == 0) { |
| 1634 | s->strm->total_out_hi32++; |
| 1635 | } |
| 1636 | } |
| 1637 | |
| 1638 | /* can a new run be started? */ |
| 1639 | if (s->nblock_used == s->save_nblock+1) { |
| 1640 | return; |
| 1641 | } |
| 1642 | s->state_out_len = 1; |
| 1643 | s->state_out_ch = s->k0; |
| 1644 | k1 = bz_get_fast(s); |
| 1645 | bz_rand_udp_mask(s); |
| 1646 | k1 ^= ((s->rNToGo == 1) ? 1 : 0); |
| 1647 | s->nblock_used++; |
| 1648 | if (s->nblock_used == s->save_nblock+1) { |
| 1649 | continue; |
| 1650 | } |
| 1651 | if (k1 != s->k0) { |
| 1652 | s->k0 = k1; |
| 1653 | continue; |
| 1654 | } |
| 1655 | |
| 1656 | s->state_out_len = 2; |
| 1657 | k1 = bz_get_fast(s); |
| 1658 | bz_rand_udp_mask(s); |
| 1659 | k1 ^= ((s->rNToGo == 1) ? 1 : 0); |
| 1660 | s->nblock_used++; |
| 1661 | if (s->nblock_used == s->save_nblock+1) { |
| 1662 | continue; |
| 1663 | } |
| 1664 | if (k1 != s->k0) { |
| 1665 | s->k0 = k1; |
| 1666 | continue; |
| 1667 | } |
| 1668 | s->state_out_len = 3; |
| 1669 | k1 = bz_get_fast(s); |
| 1670 | bz_rand_udp_mask(s); |
| 1671 | k1 ^= ((s->rNToGo == 1) ? 1 : 0); |
| 1672 | s->nblock_used++; |
| 1673 | if (s->nblock_used == s->save_nblock+1) { |
| 1674 | continue; |
| 1675 | } |
| 1676 | if (k1 != s->k0) { |
| 1677 | s->k0 = k1; |
| 1678 | continue; |
| 1679 | } |
| 1680 | |
| 1681 | k1 = bz_get_fast(s); |
| 1682 | bz_rand_udp_mask(s); |
| 1683 | k1 ^= ((s->rNToGo == 1) ? 1 : 0); |
| 1684 | s->nblock_used++; |
| 1685 | s->state_out_len = ((int)k1) + 4; |
| 1686 | s->k0 = bz_get_fast(s); |
| 1687 | bz_rand_udp_mask(s); |
| 1688 | s->k0 ^= ((s->rNToGo == 1) ? 1 : 0); |
| 1689 | s->nblock_used++; |
| 1690 | } |
| 1691 | } else { |
| 1692 | /* restore */ |
| 1693 | unsigned int c_calculatedBlockCRC = s->calculatedBlockCRC; |
| 1694 | unsigned char c_state_out_ch = s->state_out_ch; |
| 1695 | int c_state_out_len = s->state_out_len; |
| 1696 | int c_nblock_used = s->nblock_used; |
| 1697 | int c_k0 = s->k0; |
| 1698 | unsigned int *c_tt = s->tt; |
| 1699 | unsigned int c_tPos = s->tPos; |
| 1700 | char *cs_next_out = s->strm->next_out; |
| 1701 | unsigned int cs_avail_out = s->strm->avail_out; |
| 1702 | /* end restore */ |
| 1703 | |
| 1704 | unsigned int avail_out_INIT = cs_avail_out; |
| 1705 | int s_save_nblockPP = s->save_nblock+1; |
| 1706 | unsigned int total_out_lo32_old; |
| 1707 | |
| 1708 | while (1) { |
| 1709 | /* try to finish existing run */ |
| 1710 | if (c_state_out_len > 0) { |
| 1711 | while (TRUE) { |
| 1712 | if (cs_avail_out == 0) { |
| 1713 | goto return_notr; |
| 1714 | } |
| 1715 | if (c_state_out_len == 1) { |
| 1716 | break; |
| 1717 | } |
| 1718 | *((unsigned char *)(cs_next_out)) = c_state_out_ch; |
| 1719 | c_calculatedBlockCRC = (c_calculatedBlockCRC << 8) ^ |
| 1720 | BZ2_crc32Table[(c_calculatedBlockCRC >> 24) ^ |
| 1721 | ((unsigned char)c_state_out_ch)]; |
| 1722 | c_state_out_len--; |
| 1723 | cs_next_out++; |
| 1724 | cs_avail_out--; |
| 1725 | } |
| 1726 | s_state_out_len_eq_one: |
| 1727 | { |
| 1728 | if (cs_avail_out == 0) { |
| 1729 | c_state_out_len = 1; |
| 1730 | goto return_notr; |
| 1731 | } |
| 1732 | *((unsigned char *)(cs_next_out)) = c_state_out_ch; |
| 1733 | c_calculatedBlockCRC = (c_calculatedBlockCRC << 8) ^ |
| 1734 | BZ2_crc32Table[(c_calculatedBlockCRC >> 24) ^ |
| 1735 | ((unsigned char)c_state_out_ch)]; |
| 1736 | cs_next_out++; |
| 1737 | cs_avail_out--; |
| 1738 | } |
| 1739 | } |
| 1740 | /* can a new run be started? */ |
| 1741 | if (c_nblock_used == s_save_nblockPP) { |
| 1742 | c_state_out_len = 0; goto return_notr; |
| 1743 | } |
| 1744 | c_state_out_ch = c_k0; |
| 1745 | c_tPos = c_tt[c_tPos]; |
| 1746 | k1 = (unsigned char)(c_tPos & 0xff); |
| 1747 | c_tPos >>= 8; |
| 1748 | |
| 1749 | c_nblock_used++; |
| 1750 | |
| 1751 | if (k1 != c_k0) { |
| 1752 | c_k0 = k1; |
| 1753 | goto s_state_out_len_eq_one; |
| 1754 | } |
| 1755 | |
| 1756 | if (c_nblock_used == s_save_nblockPP) { |
| 1757 | goto s_state_out_len_eq_one; |
| 1758 | } |
| 1759 | |
| 1760 | c_state_out_len = 2; |
| 1761 | c_tPos = c_tt[c_tPos]; |
| 1762 | k1 = (unsigned char)(c_tPos & 0xff); |
| 1763 | c_tPos >>= 8; |
| 1764 | |
| 1765 | c_nblock_used++; |
| 1766 | if (c_nblock_used == s_save_nblockPP) { |
| 1767 | continue; |
| 1768 | } |
| 1769 | if (k1 != c_k0) { |
| 1770 | c_k0 = k1; |
| 1771 | continue; |
| 1772 | } |
| 1773 | |
| 1774 | c_state_out_len = 3; |
| 1775 | c_tPos = c_tt[c_tPos]; |
| 1776 | k1 = (unsigned char)(c_tPos & 0xff); |
| 1777 | c_tPos >>= 8; |
| 1778 | |
| 1779 | c_nblock_used++; |
| 1780 | if (c_nblock_used == s_save_nblockPP) { |
| 1781 | continue; |
| 1782 | } |
| 1783 | if (k1 != c_k0) { |
| 1784 | c_k0 = k1; |
| 1785 | continue; |
| 1786 | } |
| 1787 | |
| 1788 | c_tPos = c_tt[c_tPos]; |
| 1789 | k1 = (unsigned char)(c_tPos & 0xff); |
| 1790 | c_tPos >>= 8; |
| 1791 | |
| 1792 | c_nblock_used++; |
| 1793 | c_state_out_len = ((int)k1) + 4; |
| 1794 | |
| 1795 | c_tPos = c_tt[c_tPos]; |
| 1796 | c_k0 = (unsigned char)(c_tPos & 0xff); |
| 1797 | c_tPos >>= 8; |
| 1798 | |
| 1799 | c_nblock_used++; |
| 1800 | } |
| 1801 | |
| 1802 | return_notr: |
| 1803 | total_out_lo32_old = s->strm->total_out_lo32; |
| 1804 | s->strm->total_out_lo32 += (avail_out_INIT - cs_avail_out); |
| 1805 | if (s->strm->total_out_lo32 < total_out_lo32_old) { |
| 1806 | s->strm->total_out_hi32++; |
| 1807 | } |
| 1808 | |
| 1809 | /* save */ |
| 1810 | s->calculatedBlockCRC = c_calculatedBlockCRC; |
| 1811 | s->state_out_ch = c_state_out_ch; |
| 1812 | s->state_out_len = c_state_out_len; |
| 1813 | s->nblock_used = c_nblock_used; |
| 1814 | s->k0 = c_k0; |
| 1815 | s->tt = c_tt; |
| 1816 | s->tPos = c_tPos; |
| 1817 | s->strm->next_out = cs_next_out; |
| 1818 | s->strm->avail_out = cs_avail_out; |
| 1819 | /* end save */ |
| 1820 | } |
| 1821 | } |
| 1822 | |
| 1823 | static void unRLE_obuf_to_output_SMALL(DState *s) |
| 1824 | { |
| 1825 | unsigned char k1; |
| 1826 | |
| 1827 | if (s->blockRandomised) { |
| 1828 | while (1) { |
| 1829 | /* try to finish existing run */ |
| 1830 | while (1) { |
| 1831 | if (s->strm->avail_out == 0) { |
| 1832 | return; |
| 1833 | } |
| 1834 | if (s->state_out_len == 0) { |
| 1835 | break; |
| 1836 | } |
| 1837 | *((unsigned char *)(s->strm->next_out)) = s->state_out_ch; |
| 1838 | s->calculatedBlockCRC = (s->calculatedBlockCRC << 8) ^ |
| 1839 | BZ2_crc32Table[(s->calculatedBlockCRC >> 24) ^ |
| 1840 | ((unsigned char)s->state_out_ch)]; |
| 1841 | s->state_out_len--; |
| 1842 | s->strm->next_out++; |
| 1843 | s->strm->avail_out--; |
| 1844 | s->strm->total_out_lo32++; |
| 1845 | if (s->strm->total_out_lo32 == 0) { |
| 1846 | s->strm->total_out_hi32++; |
| 1847 | } |
| 1848 | } |
| 1849 | |
| 1850 | /* can a new run be started? */ |
| 1851 | if (s->nblock_used == s->save_nblock+1) { |
| 1852 | return; |
| 1853 | } |
| 1854 | |
| 1855 | s->state_out_len = 1; |
| 1856 | s->state_out_ch = s->k0; |
| 1857 | k1 = bz_get_small(s); |
| 1858 | bz_rand_udp_mask(s); |
| 1859 | k1 ^= ((s->rNToGo == 1) ? 1 : 0); |
| 1860 | s->nblock_used++; |
| 1861 | if (s->nblock_used == s->save_nblock+1) { |
| 1862 | continue; |
| 1863 | } |
| 1864 | if (k1 != s->k0) { |
| 1865 | s->k0 = k1; |
| 1866 | continue; |
| 1867 | } |
| 1868 | |
| 1869 | s->state_out_len = 2; |
| 1870 | k1 = bz_get_small(s); |
| 1871 | bz_rand_udp_mask(s); |
| 1872 | k1 ^= ((s->rNToGo == 1) ? 1 : 0); |
| 1873 | s->nblock_used++; |
| 1874 | if (s->nblock_used == s->save_nblock+1) { |
| 1875 | continue; |
| 1876 | } |
| 1877 | if (k1 != s->k0) { |
| 1878 | s->k0 = k1; |
| 1879 | continue; |
| 1880 | } |
| 1881 | |
| 1882 | s->state_out_len = 3; |
| 1883 | k1 = bz_get_small(s); |
| 1884 | bz_rand_udp_mask(s); |
| 1885 | k1 ^= ((s->rNToGo == 1) ? 1 : 0); |
| 1886 | s->nblock_used++; |
| 1887 | if (s->nblock_used == s->save_nblock+1) { |
| 1888 | continue; |
| 1889 | } |
| 1890 | if (k1 != s->k0) { |
| 1891 | s->k0 = k1; |
| 1892 | continue; |
| 1893 | } |
| 1894 | k1 = bz_get_small(s); |
| 1895 | bz_rand_udp_mask(s); |
| 1896 | k1 ^= ((s->rNToGo == 1) ? 1 : 0); |
| 1897 | s->nblock_used++; |
| 1898 | s->state_out_len = ((int)k1) + 4; |
| 1899 | s->k0 = bz_get_small(s); |
| 1900 | bz_rand_udp_mask(s); |
| 1901 | s->k0 ^= ((s->rNToGo == 1) ? 1 : 0); |
| 1902 | s->nblock_used++; |
| 1903 | } |
| 1904 | } else { |
| 1905 | while (1) { |
| 1906 | /* try to finish existing run */ |
| 1907 | while (1) { |
| 1908 | if (s->strm->avail_out == 0) { |
| 1909 | return; |
| 1910 | } |
| 1911 | if (s->state_out_len == 0) { |
| 1912 | break; |
| 1913 | } |
| 1914 | *((unsigned char *)(s->strm->next_out)) = s->state_out_ch; |
| 1915 | s->calculatedBlockCRC = (s->calculatedBlockCRC << 8) ^ |
| 1916 | BZ2_crc32Table[(s->calculatedBlockCRC >> 24) ^ |
| 1917 | ((unsigned char)s->state_out_ch)]; |
| 1918 | s->state_out_len--; |
| 1919 | s->strm->next_out++; |
| 1920 | s->strm->avail_out--; |
| 1921 | s->strm->total_out_lo32++; |
| 1922 | if (s->strm->total_out_lo32 == 0) { |
| 1923 | s->strm->total_out_hi32++; |
| 1924 | } |
| 1925 | } |
| 1926 | |
| 1927 | /* can a new run be started? */ |
| 1928 | if (s->nblock_used == s->save_nblock+1) { |
| 1929 | return; |
| 1930 | } |
| 1931 | |
| 1932 | s->state_out_len = 1; |
| 1933 | s->state_out_ch = s->k0; |
| 1934 | k1 = bz_get_small(s); |
| 1935 | s->nblock_used++; |
| 1936 | if (s->nblock_used == s->save_nblock+1) { |
| 1937 | continue; |
| 1938 | } |
| 1939 | if (k1 != s->k0) { |
| 1940 | s->k0 = k1; |
| 1941 | continue; |
| 1942 | } |
| 1943 | |
| 1944 | s->state_out_len = 2; |
| 1945 | k1 = bz_get_small(s); |
| 1946 | s->nblock_used++; |
| 1947 | if (s->nblock_used == s->save_nblock+1) { |
| 1948 | continue; |
| 1949 | } |
| 1950 | if (k1 != s->k0) { |
| 1951 | s->k0 = k1; |
| 1952 | continue; |
| 1953 | } |
| 1954 | |
| 1955 | s->state_out_len = 3; |
| 1956 | k1 = bz_get_small(s); |
| 1957 | s->nblock_used++; |
| 1958 | if (s->nblock_used == s->save_nblock+1) { |
| 1959 | continue; |
| 1960 | } |
| 1961 | if (k1 != s->k0) { |
| 1962 | s->k0 = k1; |
| 1963 | continue; |
| 1964 | } |
| 1965 | |
| 1966 | k1 = bz_get_small(s); |
| 1967 | s->nblock_used++; |
| 1968 | s->state_out_len = ((int)k1) + 4; |
| 1969 | s->k0 = bz_get_small(s); |
| 1970 | s->nblock_used++; |
| 1971 | } |
| 1972 | } |
| 1973 | } |
| 1974 | |
| 1975 | int BZ2_bzDecompress(bz_stream *strm) |
| 1976 | { |
| 1977 | DState* s; |
| 1978 | if (strm == NULL) { |
| 1979 | return BZ_PARAM_ERROR; |
| 1980 | } |
| 1981 | s = strm->state; |
| 1982 | if (s == NULL) { |
| 1983 | return BZ_PARAM_ERROR; |
| 1984 | } |
| 1985 | if (s->strm != strm) { |
| 1986 | return BZ_PARAM_ERROR; |
| 1987 | } |
| 1988 | |
| 1989 | while (1) { |
| 1990 | if (s->state == BZ_X_IDLE) { |
| 1991 | return BZ_SEQUENCE_ERROR; |
| 1992 | } |
| 1993 | if (s->state == BZ_X_OUTPUT) { |
| 1994 | if (s->smallDecompress) { |
| 1995 | unRLE_obuf_to_output_SMALL(s); |
| 1996 | } else { |
| 1997 | unRLE_obuf_to_output_FAST(s); |
| 1998 | } |
| 1999 | if (s->nblock_used == s->save_nblock+1 && s->state_out_len == 0) { |
| 2000 | s->calculatedBlockCRC = ~(s->calculatedBlockCRC); |
| 2001 | if (s->verbosity >= 3) { |
| 2002 | error_msg("{0x%x, 0x%x}", s->storedBlockCRC, s->calculatedBlockCRC); |
| 2003 | } |
| 2004 | if (s->verbosity >= 2) { |
| 2005 | error_msg("]"); |
| 2006 | } |
| 2007 | if (s->calculatedBlockCRC != s->storedBlockCRC) { |
| 2008 | return BZ_DATA_ERROR; |
| 2009 | } |
| 2010 | s->calculatedCombinedCRC = (s->calculatedCombinedCRC << 1) | (s->calculatedCombinedCRC >> 31); |
| 2011 | s->calculatedCombinedCRC ^= s->calculatedBlockCRC; |
| 2012 | s->state = BZ_X_BLKHDR_1; |
| 2013 | } else { |
| 2014 | return BZ_OK; |
| 2015 | } |
| 2016 | } |
| 2017 | if (s->state >= BZ_X_MAGIC_1) { |
| 2018 | int r = BZ2_decompress(s); |
| 2019 | if (r == BZ_STREAM_END) { |
| 2020 | if (s->verbosity >= 3) { |
| 2021 | error_msg("\n combined CRCs: stored = 0x%x, computed = 0x%x", |
| 2022 | s->storedCombinedCRC, s->calculatedCombinedCRC ); |
| 2023 | } |
| 2024 | if (s->calculatedCombinedCRC != s->storedCombinedCRC) { |
| 2025 | return BZ_DATA_ERROR; |
| 2026 | } |
| 2027 | return r; |
| 2028 | } |
| 2029 | if (s->state != BZ_X_OUTPUT) { |
| 2030 | return r; |
| 2031 | } |
| 2032 | } |
| 2033 | } |
| 2034 | |
| 2035 | assert_h(6001); |
| 2036 | |
| 2037 | return(0); /*NOTREACHED*/ |
| 2038 | } |
| 2039 | |
| 2040 | int BZ2_bzRead(int *bzerror, void *b, void *buf, int len) |
| 2041 | { |
| 2042 | int n, ret; |
| 2043 | bzFile *bzf = (bzFile*)b; |
| 2044 | |
| 2045 | bz_seterr(BZ_OK, bzerror, &bzf); |
| 2046 | |
| 2047 | if (bzf == NULL || buf == NULL || len < 0) { |
| 2048 | bz_seterr(BZ_PARAM_ERROR, bzerror, &bzf); |
| 2049 | return 0; |
| 2050 | } |
| 2051 | |
| 2052 | if (bzf->writing) { |
| 2053 | bz_seterr(BZ_SEQUENCE_ERROR, bzerror, &bzf); |
| 2054 | return 0; |
| 2055 | } |
| 2056 | |
| 2057 | if (len == 0) { |
| 2058 | bz_seterr(BZ_OK, bzerror, &bzf); |
| 2059 | return 0; |
| 2060 | } |
| 2061 | |
| 2062 | bzf->strm.avail_out = len; |
| 2063 | bzf->strm.next_out = buf; |
| 2064 | |
| 2065 | while (1) { |
| 2066 | if (ferror(bzf->handle)) { |
| 2067 | bz_seterr(BZ_IO_ERROR, bzerror, &bzf); |
| 2068 | return 0; |
| 2069 | } |
| 2070 | if ((bzf->strm.avail_in == 0) && !myfeof(bzf->handle)) { |
| 2071 | n = fread(bzf->buf, sizeof(unsigned char), BZ_MAX_UNUSED, bzf->handle); |
| 2072 | if (ferror(bzf->handle)) { |
| 2073 | bz_seterr(BZ_IO_ERROR, bzerror, &bzf); |
| 2074 | return 0; |
| 2075 | } |
| 2076 | bzf->bufN = n; |
| 2077 | bzf->strm.avail_in = bzf->bufN; |
| 2078 | bzf->strm.next_in = bzf->buf; |
| 2079 | } |
| 2080 | |
| 2081 | ret = BZ2_bzDecompress(&(bzf->strm)); |
| 2082 | |
| 2083 | if ((ret != BZ_OK) && (ret != BZ_STREAM_END)) { |
| 2084 | bz_seterr(ret, bzerror, &bzf); |
| 2085 | return 0; |
| 2086 | } |
| 2087 | |
| 2088 | if ((ret == BZ_OK) && myfeof(bzf->handle) && |
| 2089 | (bzf->strm.avail_in == 0) && (bzf->strm.avail_out > 0)) { |
| 2090 | bz_seterr(BZ_UNEXPECTED_EOF, bzerror, &bzf); |
| 2091 | return(0); |
| 2092 | } |
| 2093 | |
| 2094 | if (ret == BZ_STREAM_END) { |
| 2095 | bz_seterr(BZ_STREAM_END, bzerror, &bzf); |
| 2096 | return(len - bzf->strm.avail_out); |
| 2097 | } |
| 2098 | if (bzf->strm.avail_out == 0) { |
| 2099 | bz_seterr(BZ_OK, bzerror, &bzf); |
| 2100 | return(len); |
| 2101 | } |
| 2102 | } |
| 2103 | return(0); /*not reached*/ |
| 2104 | } |
| 2105 | |
| 2106 | void BZ2_bzReadGetUnused(int *bzerror, void *b, void **unused, int *nUnused) |
| 2107 | { |
| 2108 | bzFile *bzf = (bzFile*)b; |
| 2109 | if (bzf == NULL) { |
| 2110 | bz_seterr(BZ_PARAM_ERROR, bzerror, &bzf); |
| 2111 | return; |
| 2112 | } |
| 2113 | if (bzf->lastErr != BZ_STREAM_END) { |
| 2114 | bz_seterr(BZ_SEQUENCE_ERROR, bzerror, &bzf); |
| 2115 | return; |
| 2116 | } |
| 2117 | if (unused == NULL || nUnused == NULL) { |
| 2118 | bz_seterr(BZ_PARAM_ERROR, bzerror, &bzf); |
| 2119 | return; |
| 2120 | } |
| 2121 | |
| 2122 | bz_seterr(BZ_OK, bzerror, &bzf); |
| 2123 | *nUnused = bzf->strm.avail_in; |
| 2124 | *unused = bzf->strm.next_in; |
| 2125 | } |
| 2126 | |
| 2127 | void *BZ2_bzReadOpen(int *bzerror, FILE *f, int small, void *unused, int nUnused) |
| 2128 | { |
| 2129 | bzFile *bzf = NULL; |
| 2130 | int ret; |
| 2131 | |
| 2132 | bz_seterr(BZ_OK, bzerror, &bzf); |
| 2133 | |
| 2134 | if (f == NULL || (small != 0 && small != 1) || |
| 2135 | (unused != NULL && (nUnused < 0 || nUnused > BZ_MAX_UNUSED)) || |
| 2136 | // (verbosity_level < 0 || verbosity_level > 4) || |
| 2137 | (unused == NULL && nUnused != 0)) { |
| 2138 | bz_seterr(BZ_PARAM_ERROR, bzerror, &bzf); |
| 2139 | return NULL; |
| 2140 | } |
| 2141 | |
| 2142 | if (ferror(f)) { |
| 2143 | bz_seterr(BZ_IO_ERROR, bzerror, &bzf); |
| 2144 | return NULL; |
| 2145 | } |
| 2146 | |
| 2147 | bzf = xmalloc(sizeof(bzFile)); |
| 2148 | if (bzf == NULL) { |
| 2149 | bz_seterr(BZ_MEM_ERROR, bzerror, &bzf); |
| 2150 | return NULL; |
| 2151 | } |
| 2152 | bz_seterr(BZ_OK, bzerror, &bzf); |
| 2153 | |
| 2154 | bzf->initialisedOk = FALSE; |
| 2155 | bzf->handle = f; |
| 2156 | bzf->bufN = 0; |
| 2157 | bzf->writing = FALSE; |
| 2158 | bzf->strm.bzalloc = NULL; |
| 2159 | bzf->strm.bzfree = NULL; |
| 2160 | bzf->strm.opaque = NULL; |
| 2161 | |
| 2162 | while (nUnused > 0) { |
| 2163 | bzf->buf[bzf->bufN] = *((unsigned char *)(unused)); bzf->bufN++; |
| 2164 | unused = ((void *)( 1 + ((unsigned char *)(unused)) )); |
| 2165 | nUnused--; |
| 2166 | } |
| 2167 | |
| 2168 | ret = BZ2_bzDecompressInit(&(bzf->strm), small); |
| 2169 | if (ret != BZ_OK) { |
| 2170 | bz_seterr(ret, bzerror, &bzf); |
| 2171 | free(bzf); |
| 2172 | return NULL; |
| 2173 | } |
| 2174 | |
| 2175 | bzf->strm.avail_in = bzf->bufN; |
| 2176 | bzf->strm.next_in = bzf->buf; |
| 2177 | |
| 2178 | bzf->initialisedOk = TRUE; |
| 2179 | return bzf; |
| 2180 | } |
| 2181 | |
| 2182 | static unsigned char uncompressStream(FILE *zStream, FILE *stream) |
| 2183 | { |
| 2184 | unsigned char unused[BZ_MAX_UNUSED]; |
| 2185 | unsigned char *unusedTmp; |
| 2186 | unsigned char obuf[5000]; |
| 2187 | void *bzf = NULL; |
| 2188 | int bzerr_dummy; |
| 2189 | int bzerr; |
| 2190 | int nread; |
| 2191 | int nUnused; |
| 2192 | int streamNo; |
| 2193 | int ret; |
| 2194 | int i; |
| 2195 | |
| 2196 | nUnused = 0; |
| 2197 | streamNo = 0; |
| 2198 | |
| 2199 | if (ferror(stream)) { |
| 2200 | goto errhandler_io; |
| 2201 | } |
| 2202 | if (ferror(zStream)) { |
| 2203 | goto errhandler_io; |
| 2204 | } |
| 2205 | |
| 2206 | while(1) { |
| 2207 | bzf = BZ2_bzReadOpen(&bzerr, zStream, (int)smallMode, unused, nUnused); |
| 2208 | if (bzf == NULL || bzerr != BZ_OK) { |
| 2209 | goto errhandler; |
| 2210 | } |
| 2211 | streamNo++; |
| 2212 | |
| 2213 | while (bzerr == BZ_OK) { |
| 2214 | nread = BZ2_bzRead(&bzerr, bzf, obuf, 5000); |
| 2215 | if (bzerr == BZ_DATA_ERROR_MAGIC) { |
| 2216 | goto errhandler; |
| 2217 | } |
| 2218 | if ((bzerr == BZ_OK || bzerr == BZ_STREAM_END) && nread > 0) { |
| 2219 | fwrite(obuf, sizeof(unsigned char), nread, stream); |
| 2220 | } |
| 2221 | if (ferror(stream)) { |
| 2222 | goto errhandler_io; |
| 2223 | } |
| 2224 | } |
| 2225 | if (bzerr != BZ_STREAM_END) { |
| 2226 | goto errhandler; |
| 2227 | } |
| 2228 | BZ2_bzReadGetUnused(&bzerr, bzf, (void **)(&unusedTmp), &nUnused); |
| 2229 | if (bzerr != BZ_OK) { |
| 2230 | panic("decompress:bzReadGetUnused"); |
| 2231 | } |
| 2232 | for (i = 0; i < nUnused; i++) { |
| 2233 | unused[i] = unusedTmp[i]; |
| 2234 | } |
| 2235 | BZ2_bzReadClose(&bzerr, bzf); |
| 2236 | if (bzerr != BZ_OK) { |
| 2237 | panic("decompress:bzReadGetUnused"); |
| 2238 | } |
| 2239 | if ((nUnused == 0) && myfeof(zStream)) { |
| 2240 | break; |
| 2241 | } |
| 2242 | } |
| 2243 | |
| 2244 | if (ferror(zStream)) { |
| 2245 | goto errhandler_io; |
| 2246 | } |
| 2247 | ret = fclose(zStream); |
| 2248 | if (ret == EOF) { |
| 2249 | goto errhandler_io; |
| 2250 | } |
| 2251 | if (ferror(stream)) { |
| 2252 | goto errhandler_io; |
| 2253 | } |
| 2254 | ret = fflush(stream); |
| 2255 | if (ret != 0) { |
| 2256 | goto errhandler_io; |
| 2257 | } |
| 2258 | if (stream != stdout) { |
| 2259 | ret = fclose(stream); |
| 2260 | if (ret == EOF) { |
| 2261 | goto errhandler_io; |
| 2262 | } |
| 2263 | } |
| 2264 | // if (verbosity_level >= 2) { |
| 2265 | // fprintf(stderr,"\n "); |
| 2266 | // } |
| 2267 | return TRUE; |
| 2268 | |
| 2269 | errhandler: |
| 2270 | BZ2_bzReadClose ( &bzerr_dummy, bzf ); |
| 2271 | switch (bzerr) { |
| 2272 | case BZ_CONFIG_ERROR: |
| 2273 | error_msg("bzip2: I'm not configured correctly for this platform!\n" |
| 2274 | "\tI require Int32, Int16 and Char to have sizes\n" |
| 2275 | "\tof 4, 2 and 1 bytes to run properly, and they don't.\n" |
| 2276 | "\tProbably you can fix this by defining them correctly,\n" |
| 2277 | "\tand recompiling. Bye!\n" ); |
| 2278 | exit(3); |
| 2279 | case BZ_IO_ERROR: |
| 2280 | errhandler_io: |
| 2281 | error_msg("\n%s: I/O or other error, bailing out. " |
| 2282 | "Possible reason follows.\n", progName); |
| 2283 | perror(progName); |
| 2284 | cleanUpAndFail(1); |
| 2285 | case BZ_DATA_ERROR: |
| 2286 | error_msg("\n%s: Data integrity error when decompressing.\n", progName); |
| 2287 | cleanUpAndFail(2); |
| 2288 | case BZ_MEM_ERROR: |
| 2289 | error_msg("\n%s: couldn't allocate enough memory\n", progName); |
| 2290 | cleanUpAndFail(1); |
| 2291 | case BZ_UNEXPECTED_EOF: |
| 2292 | error_msg("\n%s: Compressed file ends unexpectedly;\n\t" |
| 2293 | "perhaps it is corrupted? *Possible* reason follows.\n", progName); |
| 2294 | perror(progName); |
| 2295 | cleanUpAndFail(2); |
| 2296 | case BZ_DATA_ERROR_MAGIC: |
| 2297 | if (zStream != stdin) { |
| 2298 | fclose(zStream); |
| 2299 | } |
| 2300 | if (stream != stdout) { |
| 2301 | fclose(stream); |
| 2302 | } |
| 2303 | if (streamNo == 1) { |
| 2304 | return FALSE; |
| 2305 | } else { |
| 2306 | if (noisy) { |
| 2307 | error_msg("\n%s: %s: trailing garbage after EOF ignored\n", progName, inName ); |
| 2308 | } |
| 2309 | return TRUE; |
| 2310 | } |
| 2311 | default: |
| 2312 | panic ( "decompress:unexpected error" ); |
| 2313 | } |
| 2314 | |
| 2315 | panic("decompress:end"); |
| 2316 | return(TRUE); /*notreached*/ |
| 2317 | } |
| 2318 | |
| 2319 | int bunzip2_main(int argc, char **argv) |
| 2320 | { |
Glenn L McGrath | fff11f1 | 2001-11-18 14:20:25 +0000 | [diff] [blame] | 2321 | const int bunzip_to_stdout = 1; |
| 2322 | int flags = 0; |
| 2323 | int opt = 0; |
| 2324 | |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 2325 | FILE *src_stream; |
| 2326 | FILE *dst_stream; |
| 2327 | char *save_name; |
| 2328 | char *save_name_ptr; |
Glenn L McGrath | fff11f1 | 2001-11-18 14:20:25 +0000 | [diff] [blame] | 2329 | |
| 2330 | /* if called as bzcat */ |
| 2331 | if (strcmp(applet_name, "bzcat") == 0) |
| 2332 | flags |= bunzip_to_stdout; |
| 2333 | |
| 2334 | while ((opt = getopt(argc, argv, "ch")) != -1) { |
| 2335 | switch (opt) { |
| 2336 | case 'c': |
| 2337 | flags |= bunzip_to_stdout; |
| 2338 | break; |
| 2339 | case 'h': |
| 2340 | default: |
| 2341 | show_usage(); /* exit's inside usage */ |
| 2342 | } |
| 2343 | } |
| 2344 | |
| 2345 | save_name = xstrdup(argv[optind]); |
| 2346 | |
| 2347 | if (save_name == NULL) { |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 2348 | show_usage(); |
| 2349 | } |
Glenn L McGrath | fff11f1 | 2001-11-18 14:20:25 +0000 | [diff] [blame] | 2350 | |
| 2351 | src_stream = xfopen(argv[optind], "r"); |
| 2352 | |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 2353 | save_name_ptr = strrchr(save_name, '.'); |
| 2354 | if (save_name_ptr == NULL) { |
| 2355 | return(FALSE); |
| 2356 | } |
| 2357 | if (strcmp(save_name_ptr, ".bz2") != 0) { |
| 2358 | error_msg("Invalid extension, expected .bz2"); |
| 2359 | } |
| 2360 | *save_name_ptr = '\0'; |
Glenn L McGrath | fff11f1 | 2001-11-18 14:20:25 +0000 | [diff] [blame] | 2361 | |
| 2362 | if (flags & bunzip_to_stdout) { |
| 2363 | dst_stream = stdout; |
| 2364 | } else { |
| 2365 | dst_stream = xfopen(save_name, "w"); |
| 2366 | } |
Glenn L McGrath | 24e2833 | 2001-10-05 03:48:57 +0000 | [diff] [blame] | 2367 | uncompressStream(src_stream, dst_stream); |
| 2368 | |
| 2369 | return(TRUE); |
| 2370 | } |