Commit | Line | Data |
---|---|---|
7795edd6 JAS |
1 | /* gzread.c -- zlib functions for reading gzip files |
2 | * Copyright (C) 2004, 2005, 2010, 2011, 2012, 2013 Mark Adler | |
3 | * For conditions of distribution and use, see copyright notice in zlib.h | |
4 | */ | |
5 | ||
6 | #include "gzguts.h" | |
7 | ||
8 | /* Local functions */ | |
9 | local int gz_load OF((gz_statep, unsigned char *, unsigned, unsigned *)); | |
10 | local int gz_avail OF((gz_statep)); | |
11 | local int gz_look OF((gz_statep)); | |
12 | local int gz_decomp OF((gz_statep)); | |
13 | local int gz_fetch OF((gz_statep)); | |
14 | local int gz_skip OF((gz_statep, z_off64_t)); | |
15 | ||
16 | int ZEXPORT gzgetc_(gzFile file); | |
17 | ||
18 | /* Use read() to load a buffer -- return -1 on error, otherwise 0. Read from | |
19 | state->fd, and update state->eof, state->err, and state->msg as appropriate. | |
20 | This function needs to loop on read(), since read() is not guaranteed to | |
21 | read the number of bytes requested, depending on the type of descriptor. */ | |
22 | local int gz_load(gz_statep state, unsigned char *buf, unsigned len, unsigned *have) | |
23 | { | |
24 | int ret; | |
25 | ||
26 | *have = 0; | |
27 | do { | |
28 | ret = read(state->fd, buf + *have, len - *have); | |
29 | if (ret <= 0) | |
30 | break; | |
31 | *have += ret; | |
32 | } while (*have < len); | |
33 | if (ret < 0) { | |
34 | gz_error(state, Z_ERRNO, zstrerror()); | |
35 | return -1; | |
36 | } | |
37 | if (ret == 0) | |
38 | state->eof = 1; | |
39 | return 0; | |
40 | } | |
41 | ||
42 | /* Load up input buffer and set eof flag if last data loaded -- return -1 on | |
43 | error, 0 otherwise. Note that the eof flag is set when the end of the input | |
44 | file is reached, even though there may be unused data in the buffer. Once | |
45 | that data has been used, no more attempts will be made to read the file. | |
46 | If strm->avail_in != 0, then the current data is moved to the beginning of | |
47 | the input buffer, and then the remainder of the buffer is loaded with the | |
48 | available data from the input file. */ | |
49 | local int gz_avail(gz_statep state) | |
50 | { | |
51 | unsigned got; | |
52 | z_streamp strm = &(state->strm); | |
53 | ||
54 | if (state->err != Z_OK && state->err != Z_BUF_ERROR) | |
55 | return -1; | |
56 | if (state->eof == 0) { | |
57 | if (strm->avail_in) { /* copy what's there to the start */ | |
58 | unsigned char *p = state->in; | |
59 | unsigned const char *q = strm->next_in; | |
60 | unsigned n = strm->avail_in; | |
61 | do { | |
62 | *p++ = *q++; | |
63 | } while (--n); | |
64 | } | |
65 | if (gz_load(state, state->in + strm->avail_in, | |
66 | state->size - strm->avail_in, &got) == -1) | |
67 | return -1; | |
68 | strm->avail_in += got; | |
69 | strm->next_in = state->in; | |
70 | } | |
71 | return 0; | |
72 | } | |
73 | ||
74 | /* Look for gzip header, set up for inflate or copy. state->x.have must be 0. | |
75 | If this is the first time in, allocate required memory. state->how will be | |
76 | left unchanged if there is no more input data available, will be set to COPY | |
77 | if there is no gzip header and direct copying will be performed, or it will | |
78 | be set to GZIP for decompression. If direct copying, then leftover input | |
79 | data from the input buffer will be copied to the output buffer. In that | |
80 | case, all further file reads will be directly to either the output buffer or | |
81 | a user buffer. If decompressing, the inflate state will be initialized. | |
82 | gz_look() will return 0 on success or -1 on failure. */ | |
83 | local int gz_look(gz_statep state) | |
84 | { | |
85 | z_streamp strm = &(state->strm); | |
86 | ||
87 | /* allocate read buffers and inflate memory */ | |
88 | if (state->size == 0) { | |
89 | /* allocate buffers */ | |
90 | state->in = (unsigned char *)malloc(state->want); | |
91 | state->out = (unsigned char *)malloc(state->want << 1); | |
92 | if (state->in == NULL || state->out == NULL) { | |
93 | if (state->out != NULL) | |
94 | free(state->out); | |
95 | if (state->in != NULL) | |
96 | free(state->in); | |
97 | gz_error(state, Z_MEM_ERROR, "out of memory"); | |
98 | return -1; | |
99 | } | |
100 | state->size = state->want; | |
101 | ||
102 | /* allocate inflate memory */ | |
103 | state->strm.zalloc = Z_NULL; | |
104 | state->strm.zfree = Z_NULL; | |
105 | state->strm.opaque = Z_NULL; | |
106 | state->strm.avail_in = 0; | |
107 | state->strm.next_in = Z_NULL; | |
108 | if (inflateInit2(&(state->strm), 15 + 16) != Z_OK) { /* gunzip */ | |
109 | free(state->out); | |
110 | free(state->in); | |
111 | state->size = 0; | |
112 | gz_error(state, Z_MEM_ERROR, "out of memory"); | |
113 | return -1; | |
114 | } | |
115 | } | |
116 | ||
117 | /* get at least the magic bytes in the input buffer */ | |
118 | if (strm->avail_in < 2) { | |
119 | if (gz_avail(state) == -1) | |
120 | return -1; | |
121 | if (strm->avail_in == 0) | |
122 | return 0; | |
123 | } | |
124 | ||
125 | /* look for gzip magic bytes -- if there, do gzip decoding (note: there is | |
126 | a logical dilemma here when considering the case of a partially written | |
127 | gzip file, to wit, if a single 31 byte is written, then we cannot tell | |
128 | whether this is a single-byte file, or just a partially written gzip | |
129 | file -- for here we assume that if a gzip file is being written, then | |
130 | the header will be written in a single operation, so that reading a | |
131 | single byte is sufficient indication that it is not a gzip file) */ | |
132 | if (strm->avail_in > 1 && | |
133 | strm->next_in[0] == 31 && strm->next_in[1] == 139) { | |
134 | inflateReset(strm); | |
135 | state->how = MODE_GZIP; | |
136 | state->direct = 0; | |
137 | return 0; | |
138 | } | |
139 | ||
140 | /* no gzip header -- if we were decoding gzip before, then this is trailing | |
141 | garbage. Ignore the trailing garbage and finish. */ | |
142 | if (state->direct == 0) { | |
143 | strm->avail_in = 0; | |
144 | state->eof = 1; | |
145 | state->x.have = 0; | |
146 | return 0; | |
147 | } | |
148 | ||
149 | /* doing raw i/o, copy any leftover input to output -- this assumes that | |
150 | the output buffer is larger than the input buffer, which also assures | |
151 | space for gzungetc() */ | |
152 | state->x.next = state->out; | |
153 | if (strm->avail_in) { | |
154 | memcpy(state->x.next, strm->next_in, strm->avail_in); | |
155 | state->x.have = strm->avail_in; | |
156 | strm->avail_in = 0; | |
157 | } | |
158 | state->how = MODE_COPY; | |
159 | state->direct = 1; | |
160 | return 0; | |
161 | } | |
162 | ||
163 | /* Decompress from input to the provided next_out and avail_out in the state. | |
164 | On return, state->x.have and state->x.next point to the just decompressed | |
165 | data. If the gzip stream completes, state->how is reset to LOOK to look for | |
166 | the next gzip stream or raw data, once state->x.have is depleted. Returns 0 | |
167 | on success, -1 on failure. */ | |
168 | local int gz_decomp(gz_statep state) | |
169 | { | |
170 | int ret = Z_OK; | |
171 | unsigned had; | |
172 | z_streamp strm = &(state->strm); | |
173 | ||
174 | /* fill output buffer up to end of deflate stream */ | |
175 | had = strm->avail_out; | |
176 | do { | |
177 | /* get more input for inflate() */ | |
178 | if (strm->avail_in == 0 && gz_avail(state) == -1) | |
179 | return -1; | |
180 | if (strm->avail_in == 0) { | |
181 | gz_error(state, Z_BUF_ERROR, "unexpected end of file"); | |
182 | break; | |
183 | } | |
184 | ||
185 | /* decompress and handle errors */ | |
186 | ret = inflate(strm, Z_NO_FLUSH); | |
187 | if (ret == Z_STREAM_ERROR || ret == Z_NEED_DICT) { | |
188 | gz_error(state, Z_STREAM_ERROR, | |
189 | "internal error: inflate stream corrupt"); | |
190 | return -1; | |
191 | } | |
192 | if (ret == Z_MEM_ERROR) { | |
193 | gz_error(state, Z_MEM_ERROR, "out of memory"); | |
194 | return -1; | |
195 | } | |
196 | if (ret == Z_DATA_ERROR) { /* deflate stream invalid */ | |
197 | gz_error(state, Z_DATA_ERROR, | |
198 | strm->msg == NULL ? "compressed data error" : strm->msg); | |
199 | return -1; | |
200 | } | |
201 | } while (strm->avail_out && ret != Z_STREAM_END); | |
202 | ||
203 | /* update available output */ | |
204 | state->x.have = had - strm->avail_out; | |
205 | state->x.next = strm->next_out - state->x.have; | |
206 | ||
207 | /* if the gzip stream completed successfully, look for another */ | |
208 | if (ret == Z_STREAM_END) | |
209 | state->how = LOOK; | |
210 | ||
211 | /* good decompression */ | |
212 | return 0; | |
213 | } | |
214 | ||
215 | /* Fetch data and put it in the output buffer. Assumes state->x.have is 0. | |
216 | Data is either copied from the input file or decompressed from the input | |
217 | file depending on state->how. If state->how is LOOK, then a gzip header is | |
218 | looked for to determine whether to copy or decompress. Returns -1 on error, | |
219 | otherwise 0. gz_fetch() will leave state->how as COPY or GZIP unless the | |
220 | end of the input file has been reached and all data has been processed. */ | |
221 | local int gz_fetch(gz_statep state) | |
222 | { | |
223 | z_streamp strm = &(state->strm); | |
224 | ||
225 | do { | |
226 | switch(state->how) { | |
227 | case LOOK: /* -> LOOK, MODE_COPY (only if never GZIP), or MODE_GZIP */ | |
228 | if (gz_look(state) == -1) | |
229 | return -1; | |
230 | if (state->how == LOOK) | |
231 | return 0; | |
232 | break; | |
233 | case MODE_COPY: /* -> MODE_COPY */ | |
234 | if (gz_load(state, state->out, state->size << 1, &(state->x.have)) | |
235 | == -1) | |
236 | return -1; | |
237 | state->x.next = state->out; | |
238 | return 0; | |
239 | case MODE_GZIP: /* -> GZIP or LOOK (if end of gzip stream) */ | |
240 | strm->avail_out = state->size << 1; | |
241 | strm->next_out = state->out; | |
242 | if (gz_decomp(state) == -1) | |
243 | return -1; | |
244 | } | |
245 | } while (state->x.have == 0 && (!state->eof || strm->avail_in)); | |
246 | return 0; | |
247 | } | |
248 | ||
249 | /* Skip len uncompressed bytes of output. Return -1 on error, 0 on success. */ | |
250 | local int gz_skip(gz_statep state, z_off64_t len) | |
251 | { | |
252 | unsigned n; | |
253 | ||
254 | /* skip over len bytes or reach end-of-file, whichever comes first */ | |
255 | while (len) | |
256 | /* skip over whatever is in output buffer */ | |
257 | if (state->x.have) { | |
258 | n = GT_OFF(state->x.have) || (z_off64_t)state->x.have > len ? | |
259 | (unsigned)len : state->x.have; | |
260 | state->x.have -= n; | |
261 | state->x.next += n; | |
262 | state->x.pos += n; | |
263 | len -= n; | |
264 | } | |
265 | ||
266 | /* output buffer empty -- return if we're at the end of the input */ | |
267 | else if (state->eof && state->strm.avail_in == 0) | |
268 | break; | |
269 | ||
270 | /* need more data to skip -- load up output buffer */ | |
271 | else { | |
272 | /* get more output, looking for header if required */ | |
273 | if (gz_fetch(state) == -1) | |
274 | return -1; | |
275 | } | |
276 | return 0; | |
277 | } | |
278 | ||
279 | /* -- see zlib.h -- */ | |
280 | int ZEXPORT gzread(gzFile file, voidp buf, unsigned len) | |
281 | { | |
282 | unsigned got, n; | |
283 | gz_statep state; | |
284 | z_streamp strm; | |
285 | ||
286 | /* get internal structure */ | |
287 | if (file == NULL) | |
288 | return -1; | |
289 | state = (gz_statep)file; | |
290 | strm = &(state->strm); | |
291 | ||
292 | /* check that we're reading and that there's no (serious) error */ | |
293 | if (state->mode != GZ_READ || | |
294 | (state->err != Z_OK && state->err != Z_BUF_ERROR)) | |
295 | return -1; | |
296 | ||
297 | /* since an int is returned, make sure len fits in one, otherwise return | |
298 | with an error (this avoids the flaw in the interface) */ | |
299 | if ((int)len < 0) { | |
300 | gz_error(state, Z_DATA_ERROR, "requested length does not fit in int"); | |
301 | return -1; | |
302 | } | |
303 | ||
304 | /* if len is zero, avoid unnecessary operations */ | |
305 | if (len == 0) | |
306 | return 0; | |
307 | ||
308 | /* process a skip request */ | |
309 | if (state->seek) { | |
310 | state->seek = 0; | |
311 | if (gz_skip(state, state->skip) == -1) | |
312 | return -1; | |
313 | } | |
314 | ||
315 | /* get len bytes to buf, or less than len if at the end */ | |
316 | got = 0; | |
317 | n = 0; | |
318 | do { | |
319 | /* first just try copying data from the output buffer */ | |
320 | if (state->x.have) { | |
321 | n = state->x.have > len ? len : state->x.have; | |
322 | memcpy(buf, state->x.next, n); | |
323 | state->x.next += n; | |
324 | state->x.have -= n; | |
325 | } | |
326 | ||
327 | /* output buffer empty -- return if we're at the end of the input */ | |
328 | else if (state->eof && strm->avail_in == 0) { | |
329 | state->past = 1; /* tried to read past end */ | |
330 | break; | |
331 | } | |
332 | ||
333 | /* need output data -- for small len or new stream load up our output | |
334 | buffer */ | |
335 | else if (state->how == LOOK || len < (state->size << 1)) { | |
336 | /* get more output, looking for header if required */ | |
337 | if (gz_fetch(state) == -1) | |
338 | return -1; | |
339 | continue; /* no progress yet -- go back to copy above */ | |
340 | /* the copy above assures that we will leave with space in the | |
341 | output buffer, allowing at least one gzungetc() to succeed */ | |
342 | } | |
343 | ||
344 | /* large len -- read directly into user buffer */ | |
345 | else if (state->how == MODE_COPY) { /* read directly */ | |
346 | if (gz_load(state, (unsigned char *)buf, len, &n) == -1) | |
347 | return -1; | |
348 | } | |
349 | ||
350 | /* large len -- decompress directly into user buffer */ | |
351 | else { /* state->how == GZIP */ | |
352 | strm->avail_out = len; | |
353 | strm->next_out = (unsigned char *)buf; | |
354 | if (gz_decomp(state) == -1) | |
355 | return -1; | |
356 | n = state->x.have; | |
357 | state->x.have = 0; | |
358 | } | |
359 | ||
360 | /* update progress */ | |
361 | len -= n; | |
362 | buf = (char *)buf + n; | |
363 | got += n; | |
364 | state->x.pos += n; | |
365 | } while (len); | |
366 | ||
367 | /* return number of bytes read into user buffer (will fit in int) */ | |
368 | return (int)got; | |
369 | } | |
370 | ||
371 | /* -- see zlib.h -- */ | |
372 | #ifdef Z_PREFIX_SET | |
373 | # undef z_gzgetc | |
374 | #else | |
375 | # undef gzgetc | |
376 | #endif | |
377 | int ZEXPORT gzgetc(gzFile file) | |
378 | { | |
379 | int ret; | |
380 | unsigned char buf[1]; | |
381 | gz_statep state; | |
382 | ||
383 | /* get internal structure */ | |
384 | if (file == NULL) | |
385 | return -1; | |
386 | state = (gz_statep)file; | |
387 | ||
388 | /* check that we're reading and that there's no (serious) error */ | |
389 | if (state->mode != GZ_READ || | |
390 | (state->err != Z_OK && state->err != Z_BUF_ERROR)) | |
391 | return -1; | |
392 | ||
393 | /* try output buffer (no need to check for skip request) */ | |
394 | if (state->x.have) { | |
395 | state->x.have--; | |
396 | state->x.pos++; | |
397 | return *(state->x.next)++; | |
398 | } | |
399 | ||
400 | /* nothing there -- try gzread() */ | |
401 | ret = gzread(file, buf, 1); | |
402 | return ret < 1 ? -1 : buf[0]; | |
403 | } | |
404 | ||
405 | int ZEXPORT gzgetc_(gzFile file) | |
406 | { | |
407 | return gzgetc(file); | |
408 | } | |
409 | ||
410 | /* -- see zlib.h -- */ | |
411 | int ZEXPORT gzungetc(int c, gzFile file) | |
412 | { | |
413 | gz_statep state; | |
414 | ||
415 | /* get internal structure */ | |
416 | if (file == NULL) | |
417 | return -1; | |
418 | state = (gz_statep)file; | |
419 | ||
420 | /* check that we're reading and that there's no (serious) error */ | |
421 | if (state->mode != GZ_READ || | |
422 | (state->err != Z_OK && state->err != Z_BUF_ERROR)) | |
423 | return -1; | |
424 | ||
425 | /* process a skip request */ | |
426 | if (state->seek) { | |
427 | state->seek = 0; | |
428 | if (gz_skip(state, state->skip) == -1) | |
429 | return -1; | |
430 | } | |
431 | ||
432 | /* can't push EOF */ | |
433 | if (c < 0) | |
434 | return -1; | |
435 | ||
436 | /* if output buffer empty, put byte at end (allows more pushing) */ | |
437 | if (state->x.have == 0) { | |
438 | state->x.have = 1; | |
439 | state->x.next = state->out + (state->size << 1) - 1; | |
440 | state->x.next[0] = c; | |
441 | state->x.pos--; | |
442 | state->past = 0; | |
443 | return c; | |
444 | } | |
445 | ||
446 | /* if no room, give up (must have already done a gzungetc()) */ | |
447 | if (state->x.have == (state->size << 1)) { | |
448 | gz_error(state, Z_DATA_ERROR, "out of room to push characters"); | |
449 | return -1; | |
450 | } | |
451 | ||
452 | /* slide output data if needed and insert byte before existing data */ | |
453 | if (state->x.next == state->out) { | |
454 | unsigned char *src = state->out + state->x.have; | |
455 | unsigned char *dest = state->out + (state->size << 1); | |
456 | while (src > state->out) | |
457 | *--dest = *--src; | |
458 | state->x.next = dest; | |
459 | } | |
460 | state->x.have++; | |
461 | state->x.next--; | |
462 | state->x.next[0] = c; | |
463 | state->x.pos--; | |
464 | state->past = 0; | |
465 | return c; | |
466 | } | |
467 | ||
468 | /* -- see zlib.h -- */ | |
469 | char * ZEXPORT gzgets(gzFile file, char *buf, int len) | |
470 | { | |
471 | unsigned left, n; | |
472 | char *str; | |
473 | unsigned char *eol; | |
474 | gz_statep state; | |
475 | ||
476 | /* check parameters and get internal structure */ | |
477 | if (file == NULL || buf == NULL || len < 1) | |
478 | return NULL; | |
479 | state = (gz_statep)file; | |
480 | ||
481 | /* check that we're reading and that there's no (serious) error */ | |
482 | if (state->mode != GZ_READ || | |
483 | (state->err != Z_OK && state->err != Z_BUF_ERROR)) | |
484 | return NULL; | |
485 | ||
486 | /* process a skip request */ | |
487 | if (state->seek) { | |
488 | state->seek = 0; | |
489 | if (gz_skip(state, state->skip) == -1) | |
490 | return NULL; | |
491 | } | |
492 | ||
493 | /* copy output bytes up to new line or len - 1, whichever comes first -- | |
494 | append a terminating zero to the string (we don't check for a zero in | |
495 | the contents, let the user worry about that) */ | |
496 | str = buf; | |
497 | left = (unsigned)len - 1; | |
498 | if (left) do { | |
499 | /* assure that something is in the output buffer */ | |
500 | if (state->x.have == 0 && gz_fetch(state) == -1) | |
501 | return NULL; /* error */ | |
502 | if (state->x.have == 0) { /* end of file */ | |
503 | state->past = 1; /* read past end */ | |
504 | break; /* return what we have */ | |
505 | } | |
506 | ||
507 | /* look for end-of-line in current output buffer */ | |
508 | n = state->x.have > left ? left : state->x.have; | |
509 | eol = (unsigned char *)memchr(state->x.next, '\n', n); | |
510 | if (eol != NULL) | |
511 | n = (unsigned)(eol - state->x.next) + 1; | |
512 | ||
513 | /* copy through end-of-line, or remainder if not found */ | |
514 | memcpy(buf, state->x.next, n); | |
515 | state->x.have -= n; | |
516 | state->x.next += n; | |
517 | state->x.pos += n; | |
518 | left -= n; | |
519 | buf += n; | |
520 | } while (left && eol == NULL); | |
521 | ||
522 | /* return terminated string, or if nothing, end of file */ | |
523 | if (buf == str) | |
524 | return NULL; | |
525 | buf[0] = 0; | |
526 | return str; | |
527 | } | |
528 | ||
529 | /* -- see zlib.h -- */ | |
530 | int ZEXPORT gzdirect(gzFile file) | |
531 | { | |
532 | gz_statep state; | |
533 | ||
534 | /* get internal structure */ | |
535 | if (file == NULL) | |
536 | return 0; | |
537 | state = (gz_statep)file; | |
538 | ||
539 | /* if the state is not known, but we can find out, then do so (this is | |
540 | mainly for right after a gzopen() or gzdopen()) */ | |
541 | if (state->mode == GZ_READ && state->how == LOOK && state->x.have == 0) | |
542 | (void)gz_look(state); | |
543 | ||
544 | /* return 1 if transparent, 0 if processing a gzip stream */ | |
545 | return state->direct; | |
546 | } | |
547 | ||
548 | /* -- see zlib.h -- */ | |
549 | int gzclose_r(gzFile file) | |
550 | { | |
551 | int ret, err; | |
552 | gz_statep state; | |
553 | ||
554 | /* get internal structure */ | |
555 | if (file == NULL) | |
556 | return Z_STREAM_ERROR; | |
557 | state = (gz_statep)file; | |
558 | ||
559 | /* check that we're reading */ | |
560 | if (state->mode != GZ_READ) | |
561 | return Z_STREAM_ERROR; | |
562 | ||
563 | /* free memory and close file */ | |
564 | if (state->size) { | |
565 | inflateEnd(&(state->strm)); | |
566 | free(state->out); | |
567 | free(state->in); | |
568 | } | |
569 | err = state->err == Z_BUF_ERROR ? Z_BUF_ERROR : Z_OK; | |
570 | gz_error(state, Z_OK, NULL); | |
571 | free(state->path); | |
572 | ret = close(state->fd); | |
573 | free(state); | |
574 | return ret ? Z_ERRNO : err; | |
575 | } |