b24e7fce |
1 | /* gzread.c -- zlib functions for reading gzip files |
9e052883 |
2 | * Copyright (C) 2004-2017 Mark Adler |
b24e7fce |
3 | * For conditions of distribution and use, see copyright notice in zlib.h |
4 | */ |
5 | |
6 | #include "gzguts.h" |
7 | |
b24e7fce |
8 | /* Use read() to load a buffer -- return -1 on error, otherwise 0. Read from |
9 | state->fd, and update state->eof, state->err, and state->msg as appropriate. |
10 | This function needs to loop on read(), since read() is not guaranteed to |
11 | read the number of bytes requested, depending on the type of descriptor. */ |
648db22b |
12 | local int gz_load(gz_statep state, unsigned char *buf, unsigned len, |
13 | unsigned *have) { |
b24e7fce |
14 | int ret; |
15 | unsigned get, max = ((unsigned)-1 >> 2) + 1; |
16 | |
17 | *have = 0; |
18 | do { |
19 | get = len - *have; |
20 | if (get > max) |
21 | get = max; |
22 | ret = read(state->fd, buf + *have, get); |
23 | if (ret <= 0) |
24 | break; |
25 | *have += (unsigned)ret; |
26 | } while (*have < len); |
27 | if (ret < 0) { |
28 | gz_error(state, Z_ERRNO, zstrerror()); |
29 | return -1; |
30 | } |
31 | if (ret == 0) |
32 | state->eof = 1; |
33 | return 0; |
34 | } |
35 | |
36 | /* Load up input buffer and set eof flag if last data loaded -- return -1 on |
37 | error, 0 otherwise. Note that the eof flag is set when the end of the input |
38 | file is reached, even though there may be unused data in the buffer. Once |
39 | that data has been used, no more attempts will be made to read the file. |
40 | If strm->avail_in != 0, then the current data is moved to the beginning of |
41 | the input buffer, and then the remainder of the buffer is loaded with the |
42 | available data from the input file. */ |
648db22b |
43 | local int gz_avail(gz_statep state) { |
b24e7fce |
44 | unsigned got; |
45 | z_streamp strm = &(state->strm); |
46 | |
47 | if (state->err != Z_OK && state->err != Z_BUF_ERROR) |
48 | return -1; |
49 | if (state->eof == 0) { |
50 | if (strm->avail_in) { /* copy what's there to the start */ |
51 | unsigned char *p = state->in; |
52 | unsigned const char *q = strm->next_in; |
53 | unsigned n = strm->avail_in; |
54 | do { |
55 | *p++ = *q++; |
56 | } while (--n); |
57 | } |
58 | if (gz_load(state, state->in + strm->avail_in, |
59 | state->size - strm->avail_in, &got) == -1) |
60 | return -1; |
61 | strm->avail_in += got; |
62 | strm->next_in = state->in; |
63 | } |
64 | return 0; |
65 | } |
66 | |
67 | /* Look for gzip header, set up for inflate or copy. state->x.have must be 0. |
68 | If this is the first time in, allocate required memory. state->how will be |
69 | left unchanged if there is no more input data available, will be set to COPY |
70 | if there is no gzip header and direct copying will be performed, or it will |
71 | be set to GZIP for decompression. If direct copying, then leftover input |
72 | data from the input buffer will be copied to the output buffer. In that |
73 | case, all further file reads will be directly to either the output buffer or |
74 | a user buffer. If decompressing, the inflate state will be initialized. |
75 | gz_look() will return 0 on success or -1 on failure. */ |
648db22b |
76 | local int gz_look(gz_statep state) { |
b24e7fce |
77 | z_streamp strm = &(state->strm); |
78 | |
79 | /* allocate read buffers and inflate memory */ |
80 | if (state->size == 0) { |
81 | /* allocate buffers */ |
82 | state->in = (unsigned char *)malloc(state->want); |
83 | state->out = (unsigned char *)malloc(state->want << 1); |
84 | if (state->in == NULL || state->out == NULL) { |
85 | free(state->out); |
86 | free(state->in); |
87 | gz_error(state, Z_MEM_ERROR, "out of memory"); |
88 | return -1; |
89 | } |
90 | state->size = state->want; |
91 | |
92 | /* allocate inflate memory */ |
93 | state->strm.zalloc = Z_NULL; |
94 | state->strm.zfree = Z_NULL; |
95 | state->strm.opaque = Z_NULL; |
96 | state->strm.avail_in = 0; |
97 | state->strm.next_in = Z_NULL; |
98 | if (inflateInit2(&(state->strm), 15 + 16) != Z_OK) { /* gunzip */ |
99 | free(state->out); |
100 | free(state->in); |
101 | state->size = 0; |
102 | gz_error(state, Z_MEM_ERROR, "out of memory"); |
103 | return -1; |
104 | } |
105 | } |
106 | |
107 | /* get at least the magic bytes in the input buffer */ |
108 | if (strm->avail_in < 2) { |
109 | if (gz_avail(state) == -1) |
110 | return -1; |
111 | if (strm->avail_in == 0) |
112 | return 0; |
113 | } |
114 | |
115 | /* look for gzip magic bytes -- if there, do gzip decoding (note: there is |
116 | a logical dilemma here when considering the case of a partially written |
117 | gzip file, to wit, if a single 31 byte is written, then we cannot tell |
118 | whether this is a single-byte file, or just a partially written gzip |
119 | file -- for here we assume that if a gzip file is being written, then |
120 | the header will be written in a single operation, so that reading a |
121 | single byte is sufficient indication that it is not a gzip file) */ |
122 | if (strm->avail_in > 1 && |
123 | strm->next_in[0] == 31 && strm->next_in[1] == 139) { |
124 | inflateReset(strm); |
125 | state->how = GZIP; |
126 | state->direct = 0; |
127 | return 0; |
128 | } |
129 | |
130 | /* no gzip header -- if we were decoding gzip before, then this is trailing |
131 | garbage. Ignore the trailing garbage and finish. */ |
132 | if (state->direct == 0) { |
133 | strm->avail_in = 0; |
134 | state->eof = 1; |
135 | state->x.have = 0; |
136 | return 0; |
137 | } |
138 | |
139 | /* doing raw i/o, copy any leftover input to output -- this assumes that |
140 | the output buffer is larger than the input buffer, which also assures |
141 | space for gzungetc() */ |
142 | state->x.next = state->out; |
9e052883 |
143 | memcpy(state->x.next, strm->next_in, strm->avail_in); |
144 | state->x.have = strm->avail_in; |
145 | strm->avail_in = 0; |
b24e7fce |
146 | state->how = COPY; |
147 | state->direct = 1; |
148 | return 0; |
149 | } |
150 | |
151 | /* Decompress from input to the provided next_out and avail_out in the state. |
152 | On return, state->x.have and state->x.next point to the just decompressed |
153 | data. If the gzip stream completes, state->how is reset to LOOK to look for |
154 | the next gzip stream or raw data, once state->x.have is depleted. Returns 0 |
155 | on success, -1 on failure. */ |
648db22b |
156 | local int gz_decomp(gz_statep state) { |
b24e7fce |
157 | int ret = Z_OK; |
158 | unsigned had; |
159 | z_streamp strm = &(state->strm); |
160 | |
161 | /* fill output buffer up to end of deflate stream */ |
162 | had = strm->avail_out; |
163 | do { |
164 | /* get more input for inflate() */ |
165 | if (strm->avail_in == 0 && gz_avail(state) == -1) |
166 | return -1; |
167 | if (strm->avail_in == 0) { |
168 | gz_error(state, Z_BUF_ERROR, "unexpected end of file"); |
169 | break; |
170 | } |
171 | |
172 | /* decompress and handle errors */ |
173 | ret = inflate(strm, Z_NO_FLUSH); |
174 | if (ret == Z_STREAM_ERROR || ret == Z_NEED_DICT) { |
175 | gz_error(state, Z_STREAM_ERROR, |
176 | "internal error: inflate stream corrupt"); |
177 | return -1; |
178 | } |
179 | if (ret == Z_MEM_ERROR) { |
180 | gz_error(state, Z_MEM_ERROR, "out of memory"); |
181 | return -1; |
182 | } |
183 | if (ret == Z_DATA_ERROR) { /* deflate stream invalid */ |
184 | gz_error(state, Z_DATA_ERROR, |
185 | strm->msg == NULL ? "compressed data error" : strm->msg); |
186 | return -1; |
187 | } |
188 | } while (strm->avail_out && ret != Z_STREAM_END); |
189 | |
190 | /* update available output */ |
191 | state->x.have = had - strm->avail_out; |
192 | state->x.next = strm->next_out - state->x.have; |
193 | |
194 | /* if the gzip stream completed successfully, look for another */ |
195 | if (ret == Z_STREAM_END) |
196 | state->how = LOOK; |
197 | |
198 | /* good decompression */ |
199 | return 0; |
200 | } |
201 | |
202 | /* Fetch data and put it in the output buffer. Assumes state->x.have is 0. |
203 | Data is either copied from the input file or decompressed from the input |
204 | file depending on state->how. If state->how is LOOK, then a gzip header is |
205 | looked for to determine whether to copy or decompress. Returns -1 on error, |
206 | otherwise 0. gz_fetch() will leave state->how as COPY or GZIP unless the |
207 | end of the input file has been reached and all data has been processed. */ |
648db22b |
208 | local int gz_fetch(gz_statep state) { |
b24e7fce |
209 | z_streamp strm = &(state->strm); |
210 | |
211 | do { |
212 | switch(state->how) { |
213 | case LOOK: /* -> LOOK, COPY (only if never GZIP), or GZIP */ |
214 | if (gz_look(state) == -1) |
215 | return -1; |
216 | if (state->how == LOOK) |
217 | return 0; |
218 | break; |
219 | case COPY: /* -> COPY */ |
220 | if (gz_load(state, state->out, state->size << 1, &(state->x.have)) |
221 | == -1) |
222 | return -1; |
223 | state->x.next = state->out; |
224 | return 0; |
225 | case GZIP: /* -> GZIP or LOOK (if end of gzip stream) */ |
226 | strm->avail_out = state->size << 1; |
227 | strm->next_out = state->out; |
228 | if (gz_decomp(state) == -1) |
229 | return -1; |
230 | } |
231 | } while (state->x.have == 0 && (!state->eof || strm->avail_in)); |
232 | return 0; |
233 | } |
234 | |
235 | /* Skip len uncompressed bytes of output. Return -1 on error, 0 on success. */ |
648db22b |
236 | local int gz_skip(gz_statep state, z_off64_t len) { |
b24e7fce |
237 | unsigned n; |
238 | |
239 | /* skip over len bytes or reach end-of-file, whichever comes first */ |
240 | while (len) |
241 | /* skip over whatever is in output buffer */ |
242 | if (state->x.have) { |
243 | n = GT_OFF(state->x.have) || (z_off64_t)state->x.have > len ? |
244 | (unsigned)len : state->x.have; |
245 | state->x.have -= n; |
246 | state->x.next += n; |
247 | state->x.pos += n; |
248 | len -= n; |
249 | } |
250 | |
251 | /* output buffer empty -- return if we're at the end of the input */ |
252 | else if (state->eof && state->strm.avail_in == 0) |
253 | break; |
254 | |
255 | /* need more data to skip -- load up output buffer */ |
256 | else { |
257 | /* get more output, looking for header if required */ |
258 | if (gz_fetch(state) == -1) |
259 | return -1; |
260 | } |
261 | return 0; |
262 | } |
263 | |
264 | /* Read len bytes into buf from file, or less than len up to the end of the |
265 | input. Return the number of bytes read. If zero is returned, either the |
266 | end of file was reached, or there was an error. state->err must be |
267 | consulted in that case to determine which. */ |
648db22b |
268 | local z_size_t gz_read(gz_statep state, voidp buf, z_size_t len) { |
b24e7fce |
269 | z_size_t got; |
270 | unsigned n; |
271 | |
272 | /* if len is zero, avoid unnecessary operations */ |
273 | if (len == 0) |
274 | return 0; |
275 | |
276 | /* process a skip request */ |
277 | if (state->seek) { |
278 | state->seek = 0; |
279 | if (gz_skip(state, state->skip) == -1) |
280 | return 0; |
281 | } |
282 | |
283 | /* get len bytes to buf, or less than len if at the end */ |
284 | got = 0; |
285 | do { |
286 | /* set n to the maximum amount of len that fits in an unsigned int */ |
9e052883 |
287 | n = (unsigned)-1; |
b24e7fce |
288 | if (n > len) |
9e052883 |
289 | n = (unsigned)len; |
b24e7fce |
290 | |
291 | /* first just try copying data from the output buffer */ |
292 | if (state->x.have) { |
293 | if (state->x.have < n) |
294 | n = state->x.have; |
295 | memcpy(buf, state->x.next, n); |
296 | state->x.next += n; |
297 | state->x.have -= n; |
298 | } |
299 | |
300 | /* output buffer empty -- return if we're at the end of the input */ |
301 | else if (state->eof && state->strm.avail_in == 0) { |
302 | state->past = 1; /* tried to read past end */ |
303 | break; |
304 | } |
305 | |
306 | /* need output data -- for small len or new stream load up our output |
307 | buffer */ |
308 | else if (state->how == LOOK || n < (state->size << 1)) { |
309 | /* get more output, looking for header if required */ |
310 | if (gz_fetch(state) == -1) |
311 | return 0; |
312 | continue; /* no progress yet -- go back to copy above */ |
313 | /* the copy above assures that we will leave with space in the |
314 | output buffer, allowing at least one gzungetc() to succeed */ |
315 | } |
316 | |
317 | /* large len -- read directly into user buffer */ |
318 | else if (state->how == COPY) { /* read directly */ |
319 | if (gz_load(state, (unsigned char *)buf, n, &n) == -1) |
320 | return 0; |
321 | } |
322 | |
323 | /* large len -- decompress directly into user buffer */ |
324 | else { /* state->how == GZIP */ |
325 | state->strm.avail_out = n; |
326 | state->strm.next_out = (unsigned char *)buf; |
327 | if (gz_decomp(state) == -1) |
328 | return 0; |
329 | n = state->x.have; |
330 | state->x.have = 0; |
331 | } |
332 | |
333 | /* update progress */ |
334 | len -= n; |
335 | buf = (char *)buf + n; |
336 | got += n; |
337 | state->x.pos += n; |
338 | } while (len); |
339 | |
340 | /* return number of bytes read into user buffer */ |
341 | return got; |
342 | } |
343 | |
344 | /* -- see zlib.h -- */ |
648db22b |
345 | int ZEXPORT gzread(gzFile file, voidp buf, unsigned len) { |
b24e7fce |
346 | gz_statep state; |
347 | |
348 | /* get internal structure */ |
349 | if (file == NULL) |
350 | return -1; |
351 | state = (gz_statep)file; |
352 | |
353 | /* check that we're reading and that there's no (serious) error */ |
354 | if (state->mode != GZ_READ || |
355 | (state->err != Z_OK && state->err != Z_BUF_ERROR)) |
356 | return -1; |
357 | |
358 | /* since an int is returned, make sure len fits in one, otherwise return |
359 | with an error (this avoids a flaw in the interface) */ |
360 | if ((int)len < 0) { |
361 | gz_error(state, Z_STREAM_ERROR, "request does not fit in an int"); |
362 | return -1; |
363 | } |
364 | |
365 | /* read len or fewer bytes to buf */ |
9e052883 |
366 | len = (unsigned)gz_read(state, buf, len); |
b24e7fce |
367 | |
368 | /* check for an error */ |
369 | if (len == 0 && state->err != Z_OK && state->err != Z_BUF_ERROR) |
370 | return -1; |
371 | |
372 | /* return the number of bytes read (this is assured to fit in an int) */ |
373 | return (int)len; |
374 | } |
375 | |
376 | /* -- see zlib.h -- */ |
648db22b |
377 | z_size_t ZEXPORT gzfread(voidp buf, z_size_t size, z_size_t nitems, gzFile file) { |
b24e7fce |
378 | z_size_t len; |
379 | gz_statep state; |
380 | |
381 | /* get internal structure */ |
382 | if (file == NULL) |
383 | return 0; |
384 | state = (gz_statep)file; |
385 | |
386 | /* check that we're reading and that there's no (serious) error */ |
387 | if (state->mode != GZ_READ || |
388 | (state->err != Z_OK && state->err != Z_BUF_ERROR)) |
389 | return 0; |
390 | |
391 | /* compute bytes to read -- error on overflow */ |
392 | len = nitems * size; |
393 | if (size && len / size != nitems) { |
394 | gz_error(state, Z_STREAM_ERROR, "request does not fit in a size_t"); |
395 | return 0; |
396 | } |
397 | |
398 | /* read len or fewer bytes to buf, return the number of full items read */ |
399 | return len ? gz_read(state, buf, len) / size : 0; |
400 | } |
401 | |
402 | /* -- see zlib.h -- */ |
403 | #ifdef Z_PREFIX_SET |
404 | # undef z_gzgetc |
405 | #else |
406 | # undef gzgetc |
407 | #endif |
648db22b |
408 | int ZEXPORT gzgetc(gzFile file) { |
b24e7fce |
409 | unsigned char buf[1]; |
410 | gz_statep state; |
411 | |
412 | /* get internal structure */ |
413 | if (file == NULL) |
414 | return -1; |
415 | state = (gz_statep)file; |
416 | |
417 | /* check that we're reading and that there's no (serious) error */ |
418 | if (state->mode != GZ_READ || |
419 | (state->err != Z_OK && state->err != Z_BUF_ERROR)) |
420 | return -1; |
421 | |
422 | /* try output buffer (no need to check for skip request) */ |
423 | if (state->x.have) { |
424 | state->x.have--; |
425 | state->x.pos++; |
426 | return *(state->x.next)++; |
427 | } |
428 | |
429 | /* nothing there -- try gz_read() */ |
9e052883 |
430 | return gz_read(state, buf, 1) < 1 ? -1 : buf[0]; |
b24e7fce |
431 | } |
432 | |
648db22b |
433 | int ZEXPORT gzgetc_(gzFile file) { |
b24e7fce |
434 | return gzgetc(file); |
435 | } |
436 | |
437 | /* -- see zlib.h -- */ |
648db22b |
438 | int ZEXPORT gzungetc(int c, gzFile file) { |
b24e7fce |
439 | gz_statep state; |
440 | |
441 | /* get internal structure */ |
442 | if (file == NULL) |
443 | return -1; |
444 | state = (gz_statep)file; |
445 | |
648db22b |
446 | /* in case this was just opened, set up the input buffer */ |
447 | if (state->mode == GZ_READ && state->how == LOOK && state->x.have == 0) |
448 | (void)gz_look(state); |
449 | |
b24e7fce |
450 | /* check that we're reading and that there's no (serious) error */ |
451 | if (state->mode != GZ_READ || |
452 | (state->err != Z_OK && state->err != Z_BUF_ERROR)) |
453 | return -1; |
454 | |
455 | /* process a skip request */ |
456 | if (state->seek) { |
457 | state->seek = 0; |
458 | if (gz_skip(state, state->skip) == -1) |
459 | return -1; |
460 | } |
461 | |
462 | /* can't push EOF */ |
463 | if (c < 0) |
464 | return -1; |
465 | |
466 | /* if output buffer empty, put byte at end (allows more pushing) */ |
467 | if (state->x.have == 0) { |
468 | state->x.have = 1; |
469 | state->x.next = state->out + (state->size << 1) - 1; |
470 | state->x.next[0] = (unsigned char)c; |
471 | state->x.pos--; |
472 | state->past = 0; |
473 | return c; |
474 | } |
475 | |
476 | /* if no room, give up (must have already done a gzungetc()) */ |
477 | if (state->x.have == (state->size << 1)) { |
478 | gz_error(state, Z_DATA_ERROR, "out of room to push characters"); |
479 | return -1; |
480 | } |
481 | |
482 | /* slide output data if needed and insert byte before existing data */ |
483 | if (state->x.next == state->out) { |
484 | unsigned char *src = state->out + state->x.have; |
485 | unsigned char *dest = state->out + (state->size << 1); |
486 | while (src > state->out) |
487 | *--dest = *--src; |
488 | state->x.next = dest; |
489 | } |
490 | state->x.have++; |
491 | state->x.next--; |
492 | state->x.next[0] = (unsigned char)c; |
493 | state->x.pos--; |
494 | state->past = 0; |
495 | return c; |
496 | } |
497 | |
498 | /* -- see zlib.h -- */ |
648db22b |
499 | char * ZEXPORT gzgets(gzFile file, char *buf, int len) { |
b24e7fce |
500 | unsigned left, n; |
501 | char *str; |
502 | unsigned char *eol; |
503 | gz_statep state; |
504 | |
505 | /* check parameters and get internal structure */ |
506 | if (file == NULL || buf == NULL || len < 1) |
507 | return NULL; |
508 | state = (gz_statep)file; |
509 | |
510 | /* check that we're reading and that there's no (serious) error */ |
511 | if (state->mode != GZ_READ || |
512 | (state->err != Z_OK && state->err != Z_BUF_ERROR)) |
513 | return NULL; |
514 | |
515 | /* process a skip request */ |
516 | if (state->seek) { |
517 | state->seek = 0; |
518 | if (gz_skip(state, state->skip) == -1) |
519 | return NULL; |
520 | } |
521 | |
522 | /* copy output bytes up to new line or len - 1, whichever comes first -- |
523 | append a terminating zero to the string (we don't check for a zero in |
524 | the contents, let the user worry about that) */ |
525 | str = buf; |
526 | left = (unsigned)len - 1; |
527 | if (left) do { |
528 | /* assure that something is in the output buffer */ |
529 | if (state->x.have == 0 && gz_fetch(state) == -1) |
530 | return NULL; /* error */ |
531 | if (state->x.have == 0) { /* end of file */ |
532 | state->past = 1; /* read past end */ |
533 | break; /* return what we have */ |
534 | } |
535 | |
536 | /* look for end-of-line in current output buffer */ |
537 | n = state->x.have > left ? left : state->x.have; |
538 | eol = (unsigned char *)memchr(state->x.next, '\n', n); |
539 | if (eol != NULL) |
540 | n = (unsigned)(eol - state->x.next) + 1; |
541 | |
542 | /* copy through end-of-line, or remainder if not found */ |
543 | memcpy(buf, state->x.next, n); |
544 | state->x.have -= n; |
545 | state->x.next += n; |
546 | state->x.pos += n; |
547 | left -= n; |
548 | buf += n; |
549 | } while (left && eol == NULL); |
550 | |
551 | /* return terminated string, or if nothing, end of file */ |
552 | if (buf == str) |
553 | return NULL; |
554 | buf[0] = 0; |
555 | return str; |
556 | } |
557 | |
558 | /* -- see zlib.h -- */ |
648db22b |
559 | int ZEXPORT gzdirect(gzFile file) { |
b24e7fce |
560 | gz_statep state; |
561 | |
562 | /* get internal structure */ |
563 | if (file == NULL) |
564 | return 0; |
565 | state = (gz_statep)file; |
566 | |
567 | /* if the state is not known, but we can find out, then do so (this is |
568 | mainly for right after a gzopen() or gzdopen()) */ |
569 | if (state->mode == GZ_READ && state->how == LOOK && state->x.have == 0) |
570 | (void)gz_look(state); |
571 | |
572 | /* return 1 if transparent, 0 if processing a gzip stream */ |
573 | return state->direct; |
574 | } |
575 | |
576 | /* -- see zlib.h -- */ |
648db22b |
577 | int ZEXPORT gzclose_r(gzFile file) { |
b24e7fce |
578 | int ret, err; |
579 | gz_statep state; |
580 | |
581 | /* get internal structure */ |
582 | if (file == NULL) |
583 | return Z_STREAM_ERROR; |
584 | state = (gz_statep)file; |
585 | |
586 | /* check that we're reading */ |
587 | if (state->mode != GZ_READ) |
588 | return Z_STREAM_ERROR; |
589 | |
590 | /* free memory and close file */ |
591 | if (state->size) { |
592 | inflateEnd(&(state->strm)); |
593 | free(state->out); |
594 | free(state->in); |
595 | } |
596 | err = state->err == Z_BUF_ERROR ? Z_BUF_ERROR : Z_OK; |
597 | gz_error(state, Z_OK, NULL); |
598 | free(state->path); |
599 | ret = close(state->fd); |
600 | free(state); |
601 | return ret ? Z_ERRNO : err; |
602 | } |