Martin v. Löwis | ccbd427 | 2010-08-28 07:37:05 +0000 | [diff] [blame] | 1 | /* gzlib.c -- zlib functions common to reading and writing gzip files |
| 2 | * Copyright (C) 2004, 2010 Mark Adler |
| 3 | * For conditions of distribution and use, see copyright notice in zlib.h |
| 4 | */ |
| 5 | |
| 6 | #include "gzguts.h" |
| 7 | |
| 8 | #if defined(_LARGEFILE64_SOURCE) && _LFS64_LARGEFILE-0 |
| 9 | # define LSEEK lseek64 |
| 10 | #else |
| 11 | # define LSEEK lseek |
| 12 | #endif |
| 13 | |
| 14 | /* Local functions */ |
| 15 | local void gz_reset OF((gz_statep)); |
| 16 | local gzFile gz_open OF((const char *, int, const char *)); |
| 17 | |
| 18 | #if defined UNDER_CE |
| 19 | |
| 20 | /* Map the Windows error number in ERROR to a locale-dependent error message |
| 21 | string and return a pointer to it. Typically, the values for ERROR come |
| 22 | from GetLastError. |
| 23 | |
| 24 | The string pointed to shall not be modified by the application, but may be |
| 25 | overwritten by a subsequent call to gz_strwinerror |
| 26 | |
| 27 | The gz_strwinerror function does not change the current setting of |
| 28 | GetLastError. */ |
| 29 | char ZLIB_INTERNAL *gz_strwinerror (error) |
| 30 | DWORD error; |
| 31 | { |
| 32 | static char buf[1024]; |
| 33 | |
| 34 | wchar_t *msgbuf; |
| 35 | DWORD lasterr = GetLastError(); |
| 36 | DWORD chars = FormatMessage(FORMAT_MESSAGE_FROM_SYSTEM |
| 37 | | FORMAT_MESSAGE_ALLOCATE_BUFFER, |
| 38 | NULL, |
| 39 | error, |
| 40 | 0, /* Default language */ |
| 41 | (LPVOID)&msgbuf, |
| 42 | 0, |
| 43 | NULL); |
| 44 | if (chars != 0) { |
| 45 | /* If there is an \r\n appended, zap it. */ |
| 46 | if (chars >= 2 |
| 47 | && msgbuf[chars - 2] == '\r' && msgbuf[chars - 1] == '\n') { |
| 48 | chars -= 2; |
| 49 | msgbuf[chars] = 0; |
| 50 | } |
| 51 | |
| 52 | if (chars > sizeof (buf) - 1) { |
| 53 | chars = sizeof (buf) - 1; |
| 54 | msgbuf[chars] = 0; |
| 55 | } |
| 56 | |
| 57 | wcstombs(buf, msgbuf, chars + 1); |
| 58 | LocalFree(msgbuf); |
| 59 | } |
| 60 | else { |
| 61 | sprintf(buf, "unknown win32 error (%ld)", error); |
| 62 | } |
| 63 | |
| 64 | SetLastError(lasterr); |
| 65 | return buf; |
| 66 | } |
| 67 | |
| 68 | #endif /* UNDER_CE */ |
| 69 | |
| 70 | /* Reset gzip file state */ |
| 71 | local void gz_reset(state) |
| 72 | gz_statep state; |
| 73 | { |
| 74 | if (state->mode == GZ_READ) { /* for reading ... */ |
| 75 | state->have = 0; /* no output data available */ |
| 76 | state->eof = 0; /* not at end of file */ |
| 77 | state->how = LOOK; /* look for gzip header */ |
| 78 | state->direct = 1; /* default for empty file */ |
| 79 | } |
| 80 | state->seek = 0; /* no seek request pending */ |
| 81 | gz_error(state, Z_OK, NULL); /* clear error */ |
| 82 | state->pos = 0; /* no uncompressed data yet */ |
| 83 | state->strm.avail_in = 0; /* no input data yet */ |
| 84 | } |
| 85 | |
| 86 | /* Open a gzip file either by name or file descriptor. */ |
| 87 | local gzFile gz_open(path, fd, mode) |
| 88 | const char *path; |
| 89 | int fd; |
| 90 | const char *mode; |
| 91 | { |
| 92 | gz_statep state; |
| 93 | |
| 94 | /* allocate gzFile structure to return */ |
| 95 | state = malloc(sizeof(gz_state)); |
| 96 | if (state == NULL) |
| 97 | return NULL; |
| 98 | state->size = 0; /* no buffers allocated yet */ |
| 99 | state->want = GZBUFSIZE; /* requested buffer size */ |
| 100 | state->msg = NULL; /* no error message yet */ |
| 101 | |
| 102 | /* interpret mode */ |
| 103 | state->mode = GZ_NONE; |
| 104 | state->level = Z_DEFAULT_COMPRESSION; |
| 105 | state->strategy = Z_DEFAULT_STRATEGY; |
| 106 | while (*mode) { |
| 107 | if (*mode >= '0' && *mode <= '9') |
| 108 | state->level = *mode - '0'; |
| 109 | else |
| 110 | switch (*mode) { |
| 111 | case 'r': |
| 112 | state->mode = GZ_READ; |
| 113 | break; |
| 114 | #ifndef NO_GZCOMPRESS |
| 115 | case 'w': |
| 116 | state->mode = GZ_WRITE; |
| 117 | break; |
| 118 | case 'a': |
| 119 | state->mode = GZ_APPEND; |
| 120 | break; |
| 121 | #endif |
| 122 | case '+': /* can't read and write at the same time */ |
| 123 | free(state); |
| 124 | return NULL; |
| 125 | case 'b': /* ignore -- will request binary anyway */ |
| 126 | break; |
| 127 | case 'f': |
| 128 | state->strategy = Z_FILTERED; |
| 129 | break; |
| 130 | case 'h': |
| 131 | state->strategy = Z_HUFFMAN_ONLY; |
| 132 | break; |
| 133 | case 'R': |
| 134 | state->strategy = Z_RLE; |
| 135 | break; |
| 136 | case 'F': |
| 137 | state->strategy = Z_FIXED; |
| 138 | default: /* could consider as an error, but just ignore */ |
| 139 | ; |
| 140 | } |
| 141 | mode++; |
| 142 | } |
| 143 | |
| 144 | /* must provide an "r", "w", or "a" */ |
| 145 | if (state->mode == GZ_NONE) { |
| 146 | free(state); |
| 147 | return NULL; |
| 148 | } |
| 149 | |
| 150 | /* save the path name for error messages */ |
| 151 | state->path = malloc(strlen(path) + 1); |
| 152 | if (state->path == NULL) { |
| 153 | free(state); |
| 154 | return NULL; |
| 155 | } |
| 156 | strcpy(state->path, path); |
| 157 | |
| 158 | /* open the file with the appropriate mode (or just use fd) */ |
| 159 | state->fd = fd != -1 ? fd : |
| 160 | open(path, |
| 161 | #ifdef O_LARGEFILE |
| 162 | O_LARGEFILE | |
| 163 | #endif |
| 164 | #ifdef O_BINARY |
| 165 | O_BINARY | |
| 166 | #endif |
| 167 | (state->mode == GZ_READ ? |
| 168 | O_RDONLY : |
| 169 | (O_WRONLY | O_CREAT | ( |
| 170 | state->mode == GZ_WRITE ? |
| 171 | O_TRUNC : |
| 172 | O_APPEND))), |
| 173 | 0666); |
| 174 | if (state->fd == -1) { |
| 175 | free(state->path); |
| 176 | free(state); |
| 177 | return NULL; |
| 178 | } |
| 179 | if (state->mode == GZ_APPEND) |
| 180 | state->mode = GZ_WRITE; /* simplify later checks */ |
| 181 | |
| 182 | /* save the current position for rewinding (only if reading) */ |
| 183 | if (state->mode == GZ_READ) { |
| 184 | state->start = LSEEK(state->fd, 0, SEEK_CUR); |
| 185 | if (state->start == -1) state->start = 0; |
| 186 | } |
| 187 | |
| 188 | /* initialize stream */ |
| 189 | gz_reset(state); |
| 190 | |
| 191 | /* return stream */ |
| 192 | return (gzFile)state; |
| 193 | } |
| 194 | |
| 195 | /* -- see zlib.h -- */ |
| 196 | gzFile ZEXPORT gzopen(path, mode) |
| 197 | const char *path; |
| 198 | const char *mode; |
| 199 | { |
| 200 | return gz_open(path, -1, mode); |
| 201 | } |
| 202 | |
| 203 | /* -- see zlib.h -- */ |
| 204 | gzFile ZEXPORT gzopen64(path, mode) |
| 205 | const char *path; |
| 206 | const char *mode; |
| 207 | { |
| 208 | return gz_open(path, -1, mode); |
| 209 | } |
| 210 | |
| 211 | /* -- see zlib.h -- */ |
| 212 | gzFile ZEXPORT gzdopen(fd, mode) |
| 213 | int fd; |
| 214 | const char *mode; |
| 215 | { |
| 216 | char *path; /* identifier for error messages */ |
| 217 | gzFile gz; |
| 218 | |
| 219 | if (fd == -1 || (path = malloc(7 + 3 * sizeof(int))) == NULL) |
| 220 | return NULL; |
| 221 | sprintf(path, "<fd:%d>", fd); /* for debugging */ |
| 222 | gz = gz_open(path, fd, mode); |
| 223 | free(path); |
| 224 | return gz; |
| 225 | } |
| 226 | |
| 227 | /* -- see zlib.h -- */ |
| 228 | int ZEXPORT gzbuffer(file, size) |
| 229 | gzFile file; |
| 230 | unsigned size; |
| 231 | { |
| 232 | gz_statep state; |
| 233 | |
| 234 | /* get internal structure and check integrity */ |
| 235 | if (file == NULL) |
| 236 | return -1; |
| 237 | state = (gz_statep)file; |
| 238 | if (state->mode != GZ_READ && state->mode != GZ_WRITE) |
| 239 | return -1; |
| 240 | |
| 241 | /* make sure we haven't already allocated memory */ |
| 242 | if (state->size != 0) |
| 243 | return -1; |
| 244 | |
| 245 | /* check and set requested size */ |
| 246 | if (size == 0) |
| 247 | return -1; |
| 248 | state->want = size; |
| 249 | return 0; |
| 250 | } |
| 251 | |
| 252 | /* -- see zlib.h -- */ |
| 253 | int ZEXPORT gzrewind(file) |
| 254 | gzFile file; |
| 255 | { |
| 256 | gz_statep state; |
| 257 | |
| 258 | /* get internal structure */ |
| 259 | if (file == NULL) |
| 260 | return -1; |
| 261 | state = (gz_statep)file; |
| 262 | |
| 263 | /* check that we're reading and that there's no error */ |
| 264 | if (state->mode != GZ_READ || state->err != Z_OK) |
| 265 | return -1; |
| 266 | |
| 267 | /* back up and start over */ |
| 268 | if (LSEEK(state->fd, state->start, SEEK_SET) == -1) |
| 269 | return -1; |
| 270 | gz_reset(state); |
| 271 | return 0; |
| 272 | } |
| 273 | |
| 274 | /* -- see zlib.h -- */ |
| 275 | z_off64_t ZEXPORT gzseek64(file, offset, whence) |
| 276 | gzFile file; |
| 277 | z_off64_t offset; |
| 278 | int whence; |
| 279 | { |
| 280 | unsigned n; |
| 281 | z_off64_t ret; |
| 282 | gz_statep state; |
| 283 | |
| 284 | /* get internal structure and check integrity */ |
| 285 | if (file == NULL) |
| 286 | return -1; |
| 287 | state = (gz_statep)file; |
| 288 | if (state->mode != GZ_READ && state->mode != GZ_WRITE) |
| 289 | return -1; |
| 290 | |
| 291 | /* check that there's no error */ |
| 292 | if (state->err != Z_OK) |
| 293 | return -1; |
| 294 | |
| 295 | /* can only seek from start or relative to current position */ |
| 296 | if (whence != SEEK_SET && whence != SEEK_CUR) |
| 297 | return -1; |
| 298 | |
| 299 | /* normalize offset to a SEEK_CUR specification */ |
| 300 | if (whence == SEEK_SET) |
| 301 | offset -= state->pos; |
| 302 | else if (state->seek) |
| 303 | offset += state->skip; |
| 304 | state->seek = 0; |
| 305 | |
| 306 | /* if within raw area while reading, just go there */ |
| 307 | if (state->mode == GZ_READ && state->how == COPY && |
| 308 | state->pos + offset >= state->raw) { |
| 309 | ret = LSEEK(state->fd, offset - state->have, SEEK_CUR); |
| 310 | if (ret == -1) |
| 311 | return -1; |
| 312 | state->have = 0; |
| 313 | state->eof = 0; |
| 314 | state->seek = 0; |
| 315 | gz_error(state, Z_OK, NULL); |
| 316 | state->strm.avail_in = 0; |
| 317 | state->pos += offset; |
| 318 | return state->pos; |
| 319 | } |
| 320 | |
| 321 | /* calculate skip amount, rewinding if needed for back seek when reading */ |
| 322 | if (offset < 0) { |
| 323 | if (state->mode != GZ_READ) /* writing -- can't go backwards */ |
| 324 | return -1; |
| 325 | offset += state->pos; |
| 326 | if (offset < 0) /* before start of file! */ |
| 327 | return -1; |
| 328 | if (gzrewind(file) == -1) /* rewind, then skip to offset */ |
| 329 | return -1; |
| 330 | } |
| 331 | |
| 332 | /* if reading, skip what's in output buffer (one less gzgetc() check) */ |
| 333 | if (state->mode == GZ_READ) { |
| 334 | n = GT_OFF(state->have) || (z_off64_t)state->have > offset ? |
| 335 | (unsigned)offset : state->have; |
| 336 | state->have -= n; |
| 337 | state->next += n; |
| 338 | state->pos += n; |
| 339 | offset -= n; |
| 340 | } |
| 341 | |
| 342 | /* request skip (if not zero) */ |
| 343 | if (offset) { |
| 344 | state->seek = 1; |
| 345 | state->skip = offset; |
| 346 | } |
| 347 | return state->pos + offset; |
| 348 | } |
| 349 | |
| 350 | /* -- see zlib.h -- */ |
| 351 | z_off_t ZEXPORT gzseek(file, offset, whence) |
| 352 | gzFile file; |
| 353 | z_off_t offset; |
| 354 | int whence; |
| 355 | { |
| 356 | z_off64_t ret; |
| 357 | |
| 358 | ret = gzseek64(file, (z_off64_t)offset, whence); |
| 359 | return ret == (z_off_t)ret ? (z_off_t)ret : -1; |
| 360 | } |
| 361 | |
| 362 | /* -- see zlib.h -- */ |
| 363 | z_off64_t ZEXPORT gztell64(file) |
| 364 | gzFile file; |
| 365 | { |
| 366 | gz_statep state; |
| 367 | |
| 368 | /* get internal structure and check integrity */ |
| 369 | if (file == NULL) |
| 370 | return -1; |
| 371 | state = (gz_statep)file; |
| 372 | if (state->mode != GZ_READ && state->mode != GZ_WRITE) |
| 373 | return -1; |
| 374 | |
| 375 | /* return position */ |
| 376 | return state->pos + (state->seek ? state->skip : 0); |
| 377 | } |
| 378 | |
| 379 | /* -- see zlib.h -- */ |
| 380 | z_off_t ZEXPORT gztell(file) |
| 381 | gzFile file; |
| 382 | { |
| 383 | z_off64_t ret; |
| 384 | |
| 385 | ret = gztell64(file); |
| 386 | return ret == (z_off_t)ret ? (z_off_t)ret : -1; |
| 387 | } |
| 388 | |
| 389 | /* -- see zlib.h -- */ |
| 390 | z_off64_t ZEXPORT gzoffset64(file) |
| 391 | gzFile file; |
| 392 | { |
| 393 | z_off64_t offset; |
| 394 | gz_statep state; |
| 395 | |
| 396 | /* get internal structure and check integrity */ |
| 397 | if (file == NULL) |
| 398 | return -1; |
| 399 | state = (gz_statep)file; |
| 400 | if (state->mode != GZ_READ && state->mode != GZ_WRITE) |
| 401 | return -1; |
| 402 | |
| 403 | /* compute and return effective offset in file */ |
| 404 | offset = LSEEK(state->fd, 0, SEEK_CUR); |
| 405 | if (offset == -1) |
| 406 | return -1; |
| 407 | if (state->mode == GZ_READ) /* reading */ |
| 408 | offset -= state->strm.avail_in; /* don't count buffered input */ |
| 409 | return offset; |
| 410 | } |
| 411 | |
| 412 | /* -- see zlib.h -- */ |
| 413 | z_off_t ZEXPORT gzoffset(file) |
| 414 | gzFile file; |
| 415 | { |
| 416 | z_off64_t ret; |
| 417 | |
| 418 | ret = gzoffset64(file); |
| 419 | return ret == (z_off_t)ret ? (z_off_t)ret : -1; |
| 420 | } |
| 421 | |
| 422 | /* -- see zlib.h -- */ |
| 423 | int ZEXPORT gzeof(file) |
| 424 | gzFile file; |
| 425 | { |
| 426 | gz_statep state; |
| 427 | |
| 428 | /* get internal structure and check integrity */ |
| 429 | if (file == NULL) |
| 430 | return 0; |
| 431 | state = (gz_statep)file; |
| 432 | if (state->mode != GZ_READ && state->mode != GZ_WRITE) |
| 433 | return 0; |
| 434 | |
| 435 | /* return end-of-file state */ |
| 436 | return state->mode == GZ_READ ? |
| 437 | (state->eof && state->strm.avail_in == 0 && state->have == 0) : 0; |
| 438 | } |
| 439 | |
| 440 | /* -- see zlib.h -- */ |
| 441 | const char * ZEXPORT gzerror(file, errnum) |
| 442 | gzFile file; |
| 443 | int *errnum; |
| 444 | { |
| 445 | gz_statep state; |
| 446 | |
| 447 | /* get internal structure and check integrity */ |
| 448 | if (file == NULL) |
| 449 | return NULL; |
| 450 | state = (gz_statep)file; |
| 451 | if (state->mode != GZ_READ && state->mode != GZ_WRITE) |
| 452 | return NULL; |
| 453 | |
| 454 | /* return error information */ |
| 455 | if (errnum != NULL) |
| 456 | *errnum = state->err; |
| 457 | return state->msg == NULL ? "" : state->msg; |
| 458 | } |
| 459 | |
| 460 | /* -- see zlib.h -- */ |
| 461 | void ZEXPORT gzclearerr(file) |
| 462 | gzFile file; |
| 463 | { |
| 464 | gz_statep state; |
| 465 | |
| 466 | /* get internal structure and check integrity */ |
| 467 | if (file == NULL) |
| 468 | return; |
| 469 | state = (gz_statep)file; |
| 470 | if (state->mode != GZ_READ && state->mode != GZ_WRITE) |
| 471 | return; |
| 472 | |
| 473 | /* clear error and end-of-file */ |
| 474 | if (state->mode == GZ_READ) |
| 475 | state->eof = 0; |
| 476 | gz_error(state, Z_OK, NULL); |
| 477 | } |
| 478 | |
| 479 | /* Create an error message in allocated memory and set state->err and |
| 480 | state->msg accordingly. Free any previous error message already there. Do |
| 481 | not try to free or allocate space if the error is Z_MEM_ERROR (out of |
| 482 | memory). Simply save the error message as a static string. If there is an |
| 483 | allocation failure constructing the error message, then convert the error to |
| 484 | out of memory. */ |
| 485 | void ZLIB_INTERNAL gz_error(state, err, msg) |
| 486 | gz_statep state; |
| 487 | int err; |
| 488 | const char *msg; |
| 489 | { |
| 490 | /* free previously allocated message and clear */ |
| 491 | if (state->msg != NULL) { |
| 492 | if (state->err != Z_MEM_ERROR) |
| 493 | free(state->msg); |
| 494 | state->msg = NULL; |
| 495 | } |
| 496 | |
| 497 | /* set error code, and if no message, then done */ |
| 498 | state->err = err; |
| 499 | if (msg == NULL) |
| 500 | return; |
| 501 | |
| 502 | /* for an out of memory error, save as static string */ |
| 503 | if (err == Z_MEM_ERROR) { |
| 504 | state->msg = (char *)msg; |
| 505 | return; |
| 506 | } |
| 507 | |
| 508 | /* construct error message with path */ |
| 509 | if ((state->msg = malloc(strlen(state->path) + strlen(msg) + 3)) == NULL) { |
| 510 | state->err = Z_MEM_ERROR; |
| 511 | state->msg = (char *)"out of memory"; |
| 512 | return; |
| 513 | } |
| 514 | strcpy(state->msg, state->path); |
| 515 | strcat(state->msg, ": "); |
| 516 | strcat(state->msg, msg); |
| 517 | return; |
| 518 | } |
| 519 | |
| 520 | #ifndef INT_MAX |
| 521 | /* portably return maximum value for an int (when limits.h presumed not |
| 522 | available) -- we need to do this to cover cases where 2's complement not |
| 523 | used, since C standard permits 1's complement and sign-bit representations, |
| 524 | otherwise we could just use ((unsigned)-1) >> 1 */ |
| 525 | unsigned ZLIB_INTERNAL gz_intmax() |
| 526 | { |
| 527 | unsigned p, q; |
| 528 | |
| 529 | p = 1; |
| 530 | do { |
| 531 | q = p; |
| 532 | p <<= 1; |
| 533 | p++; |
| 534 | } while (p > q); |
| 535 | return q >> 1; |
| 536 | } |
| 537 | #endif |