1 | /* gzread.c -- zlib functions for reading gzip files |
2 | * Copyright (C) 2004, 2005, 2010, 2011, 2012, 2013, 2016 Mark Adler |
3 | * For conditions of distribution and use, see copyright notice in zlib.h |
4 | */ |
5 | |
6 | #include "zbuild.h" |
7 | #include "gzguts.h" |
8 | |
9 | /* Local functions */ |
10 | static int gz_load(gz_state *, unsigned char *, unsigned, unsigned *); |
11 | static int gz_avail(gz_state *); |
12 | static int gz_look(gz_state *); |
13 | static int gz_decomp(gz_state *); |
14 | static int gz_fetch(gz_state *); |
15 | static int gz_skip(gz_state *, z_off64_t); |
16 | static size_t gz_read(gz_state *, void *, size_t); |
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 | static int gz_load(gz_state *state, unsigned char *buf, unsigned len, unsigned *have) { |
23 | ssize_t ret; |
24 | |
25 | *have = 0; |
26 | do { |
27 | ret = read(state->fd, buf + *have, len - *have); |
28 | if (ret <= 0) |
29 | break; |
30 | *have += (unsigned)ret; |
31 | } while (*have < len); |
32 | if (ret < 0) { |
33 | gz_error(state, Z_ERRNO, zstrerror()); |
34 | return -1; |
35 | } |
36 | if (ret == 0) |
37 | state->eof = 1; |
38 | return 0; |
39 | } |
40 | |
41 | /* Load up input buffer and set eof flag if last data loaded -- return -1 on |
42 | error, 0 otherwise. Note that the eof flag is set when the end of the input |
43 | file is reached, even though there may be unused data in the buffer. Once |
44 | that data has been used, no more attempts will be made to read the file. |
45 | If strm->avail_in != 0, then the current data is moved to the beginning of |
46 | the input buffer, and then the remainder of the buffer is loaded with the |
47 | available data from the input file. */ |
48 | static int gz_avail(gz_state *state) { |
49 | unsigned got; |
50 | PREFIX3(stream) *strm = &(state->strm); |
51 | |
52 | if (state->err != Z_OK && state->err != Z_BUF_ERROR) |
53 | return -1; |
54 | if (state->eof == 0) { |
55 | if (strm->avail_in) { /* copy what's there to the start */ |
56 | unsigned char *p = state->in; |
57 | unsigned const char *q = strm->next_in; |
58 | unsigned n = strm->avail_in; |
59 | do { |
60 | *p++ = *q++; |
61 | } while (--n); |
62 | } |
63 | if (gz_load(state, state->in + strm->avail_in, state->size - strm->avail_in, &got) == -1) |
64 | return -1; |
65 | strm->avail_in += got; |
66 | strm->next_in = state->in; |
67 | } |
68 | return 0; |
69 | } |
70 | |
71 | /* Look for gzip header, set up for inflate or copy. state->x.have must be 0. |
72 | If this is the first time in, allocate required memory. state->how will be |
73 | left unchanged if there is no more input data available, will be set to COPY |
74 | if there is no gzip header and direct copying will be performed, or it will |
75 | be set to GZIP for decompression. If direct copying, then leftover input |
76 | data from the input buffer will be copied to the output buffer. In that |
77 | case, all further file reads will be directly to either the output buffer or |
78 | a user buffer. If decompressing, the inflate state will be initialized. |
79 | gz_look() will return 0 on success or -1 on failure. */ |
80 | static int gz_look(gz_state *state) { |
81 | PREFIX3(stream) *strm = &(state->strm); |
82 | |
83 | /* allocate read buffers and inflate memory */ |
84 | if (state->size == 0) { |
85 | /* allocate buffers */ |
86 | state->in = (unsigned char *)malloc(state->want); |
87 | state->out = (unsigned char *)malloc(state->want << 1); |
88 | if (state->in == NULL || state->out == NULL) { |
89 | free(state->out); |
90 | free(state->in); |
91 | gz_error(state, Z_MEM_ERROR, "out of memory" ); |
92 | return -1; |
93 | } |
94 | state->size = state->want; |
95 | |
96 | /* allocate inflate memory */ |
97 | state->strm.zalloc = NULL; |
98 | state->strm.zfree = NULL; |
99 | state->strm.opaque = NULL; |
100 | state->strm.avail_in = 0; |
101 | state->strm.next_in = NULL; |
102 | if (PREFIX(inflateInit2)(&(state->strm), 15 + 16) != Z_OK) { /* gunzip */ |
103 | free(state->out); |
104 | free(state->in); |
105 | state->size = 0; |
106 | gz_error(state, Z_MEM_ERROR, "out of memory" ); |
107 | return -1; |
108 | } |
109 | } |
110 | |
111 | /* get at least the magic bytes in the input buffer */ |
112 | if (strm->avail_in < 2) { |
113 | if (gz_avail(state) == -1) |
114 | return -1; |
115 | if (strm->avail_in == 0) |
116 | return 0; |
117 | } |
118 | |
119 | /* look for gzip magic bytes -- if there, do gzip decoding (note: there is |
120 | a logical dilemma here when considering the case of a partially written |
121 | gzip file, to wit, if a single 31 byte is written, then we cannot tell |
122 | whether this is a single-byte file, or just a partially written gzip |
123 | file -- for here we assume that if a gzip file is being written, then |
124 | the header will be written in a single operation, so that reading a |
125 | single byte is sufficient indication that it is not a gzip file) */ |
126 | if (strm->avail_in > 1 && |
127 | strm->next_in[0] == 31 && strm->next_in[1] == 139) { |
128 | PREFIX(inflateReset)(strm); |
129 | state->how = GZIP; |
130 | state->direct = 0; |
131 | return 0; |
132 | } |
133 | |
134 | /* no gzip header -- if we were decoding gzip before, then this is trailing |
135 | garbage. Ignore the trailing garbage and finish. */ |
136 | if (state->direct == 0) { |
137 | strm->avail_in = 0; |
138 | state->eof = 1; |
139 | state->x.have = 0; |
140 | return 0; |
141 | } |
142 | |
143 | /* doing raw i/o, copy any leftover input to output -- this assumes that |
144 | the output buffer is larger than the input buffer, which also assures |
145 | space for gzungetc() */ |
146 | state->x.next = state->out; |
147 | if (strm->avail_in) { |
148 | memcpy(state->x.next, strm->next_in, strm->avail_in); |
149 | state->x.have = strm->avail_in; |
150 | strm->avail_in = 0; |
151 | } |
152 | state->how = COPY; |
153 | state->direct = 1; |
154 | return 0; |
155 | } |
156 | |
157 | /* Decompress from input to the provided next_out and avail_out in the state. |
158 | On return, state->x.have and state->x.next point to the just decompressed |
159 | data. If the gzip stream completes, state->how is reset to LOOK to look for |
160 | the next gzip stream or raw data, once state->x.have is depleted. Returns 0 |
161 | on success, -1 on failure. */ |
162 | static int gz_decomp(gz_state *state) { |
163 | int ret = Z_OK; |
164 | unsigned had; |
165 | PREFIX3(stream) *strm = &(state->strm); |
166 | |
167 | /* fill output buffer up to end of deflate stream */ |
168 | had = strm->avail_out; |
169 | do { |
170 | /* get more input for inflate() */ |
171 | if (strm->avail_in == 0 && gz_avail(state) == -1) |
172 | return -1; |
173 | if (strm->avail_in == 0) { |
174 | gz_error(state, Z_BUF_ERROR, "unexpected end of file" ); |
175 | break; |
176 | } |
177 | |
178 | /* decompress and handle errors */ |
179 | ret = PREFIX(inflate)(strm, Z_NO_FLUSH); |
180 | if (ret == Z_STREAM_ERROR || ret == Z_NEED_DICT) { |
181 | gz_error(state, Z_STREAM_ERROR, "internal error: inflate stream corrupt" ); |
182 | return -1; |
183 | } |
184 | if (ret == Z_MEM_ERROR) { |
185 | gz_error(state, Z_MEM_ERROR, "out of memory" ); |
186 | return -1; |
187 | } |
188 | if (ret == Z_DATA_ERROR) { /* deflate stream invalid */ |
189 | gz_error(state, Z_DATA_ERROR, strm->msg == NULL ? "compressed data error" : strm->msg); |
190 | return -1; |
191 | } |
192 | } while (strm->avail_out && ret != Z_STREAM_END); |
193 | |
194 | /* update available output */ |
195 | state->x.have = had - strm->avail_out; |
196 | state->x.next = strm->next_out - state->x.have; |
197 | |
198 | /* if the gzip stream completed successfully, look for another */ |
199 | if (ret == Z_STREAM_END) |
200 | state->how = LOOK; |
201 | |
202 | /* good decompression */ |
203 | return 0; |
204 | } |
205 | |
206 | /* Fetch data and put it in the output buffer. Assumes state->x.have is 0. |
207 | Data is either copied from the input file or decompressed from the input |
208 | file depending on state->how. If state->how is LOOK, then a gzip header is |
209 | looked for to determine whether to copy or decompress. Returns -1 on error, |
210 | otherwise 0. gz_fetch() will leave state->how as COPY or GZIP unless the |
211 | end of the input file has been reached and all data has been processed. */ |
212 | static int gz_fetch(gz_state *state) { |
213 | PREFIX3(stream) *strm = &(state->strm); |
214 | |
215 | do { |
216 | switch (state->how) { |
217 | case LOOK: /* -> LOOK, COPY (only if never GZIP), or GZIP */ |
218 | if (gz_look(state) == -1) |
219 | return -1; |
220 | if (state->how == LOOK) |
221 | return 0; |
222 | break; |
223 | case COPY: /* -> COPY */ |
224 | if (gz_load(state, state->out, state->size << 1, &(state->x.have)) |
225 | == -1) |
226 | return -1; |
227 | state->x.next = state->out; |
228 | return 0; |
229 | case GZIP: /* -> GZIP or LOOK (if end of gzip stream) */ |
230 | strm->avail_out = state->size << 1; |
231 | strm->next_out = state->out; |
232 | if (gz_decomp(state) == -1) |
233 | return -1; |
234 | } |
235 | } while (state->x.have == 0 && (!state->eof || strm->avail_in)); |
236 | return 0; |
237 | } |
238 | |
239 | /* Skip len uncompressed bytes of output. Return -1 on error, 0 on success. */ |
240 | static int gz_skip(gz_state *state, z_off64_t len) { |
241 | unsigned n; |
242 | |
243 | /* skip over len bytes or reach end-of-file, whichever comes first */ |
244 | while (len) |
245 | /* skip over whatever is in output buffer */ |
246 | if (state->x.have) { |
247 | n = GT_OFF(state->x.have) || (z_off64_t)state->x.have > len ? |
248 | (unsigned)len : state->x.have; |
249 | state->x.have -= n; |
250 | state->x.next += n; |
251 | state->x.pos += n; |
252 | len -= n; |
253 | } else if (state->eof && state->strm.avail_in == 0) { |
254 | /* output buffer empty -- return if we're at the end of the input */ |
255 | break; |
256 | } else { |
257 | /* need more data to skip -- load up output buffer */ |
258 | /* get more output, looking for header if required */ |
259 | if (gz_fetch(state) == -1) |
260 | return -1; |
261 | } |
262 | return 0; |
263 | } |
264 | |
265 | /* Read len bytes into buf from file, or less than len up to the end of the |
266 | input. Return the number of bytes read. If zero is returned, either the |
267 | end of file was reached, or there was an error. state->err must be |
268 | consulted in that case to determine which. */ |
269 | static size_t gz_read(gz_state *state, void *buf, size_t len) { |
270 | size_t got; |
271 | unsigned n; |
272 | |
273 | /* if len is zero, avoid unnecessary operations */ |
274 | if (len == 0) |
275 | return 0; |
276 | |
277 | /* process a skip request */ |
278 | if (state->seek) { |
279 | state->seek = 0; |
280 | if (gz_skip(state, state->skip) == -1) |
281 | return 0; |
282 | } |
283 | |
284 | /* get len bytes to buf, or less than len if at the end */ |
285 | got = 0; |
286 | do { |
287 | /* set n to the maximum amount of len that fits in an unsigned int */ |
288 | n = (unsigned)-1; |
289 | if (n > len) |
290 | n = (unsigned)len; |
291 | |
292 | /* first just try copying data from the output buffer */ |
293 | if (state->x.have) { |
294 | if (state->x.have < n) |
295 | n = state->x.have; |
296 | memcpy(buf, state->x.next, n); |
297 | state->x.next += n; |
298 | state->x.have -= n; |
299 | } |
300 | |
301 | /* output buffer empty -- return if we're at the end of the input */ |
302 | else if (state->eof && state->strm.avail_in == 0) { |
303 | state->past = 1; /* tried to read past end */ |
304 | break; |
305 | } |
306 | |
307 | /* need output data -- for small len or new stream load up our output |
308 | buffer */ |
309 | else if (state->how == LOOK || n < (state->size << 1)) { |
310 | /* get more output, looking for header if required */ |
311 | if (gz_fetch(state) == -1) |
312 | return 0; |
313 | continue; /* no progress yet -- go back to copy above */ |
314 | /* the copy above assures that we will leave with space in the |
315 | output buffer, allowing at least one gzungetc() to succeed */ |
316 | } |
317 | |
318 | /* large len -- read directly into user buffer */ |
319 | else if (state->how == COPY) { /* read directly */ |
320 | if (gz_load(state, (unsigned char *)buf, n, &n) == -1) |
321 | return 0; |
322 | } |
323 | |
324 | /* large len -- decompress directly into user buffer */ |
325 | else { /* state->how == GZIP */ |
326 | state->strm.avail_out = n; |
327 | state->strm.next_out = (unsigned char *)buf; |
328 | if (gz_decomp(state) == -1) |
329 | return 0; |
330 | n = state->x.have; |
331 | state->x.have = 0; |
332 | } |
333 | |
334 | /* update progress */ |
335 | len -= n; |
336 | buf = (char *)buf + n; |
337 | got += n; |
338 | state->x.pos += n; |
339 | } while (len); |
340 | |
341 | /* return number of bytes read into user buffer */ |
342 | return got; |
343 | } |
344 | |
345 | /* -- see zlib.h -- */ |
346 | int ZEXPORT PREFIX(gzread)(gzFile file, void *buf, unsigned len) { |
347 | gz_state *state; |
348 | |
349 | /* get internal structure */ |
350 | if (file == NULL) |
351 | return -1; |
352 | state = (gz_state *)file; |
353 | |
354 | /* check that we're reading and that there's no (serious) error */ |
355 | if (state->mode != GZ_READ || |
356 | (state->err != Z_OK && state->err != Z_BUF_ERROR)) |
357 | return -1; |
358 | |
359 | /* since an int is returned, make sure len fits in one, otherwise return |
360 | with an error (this avoids a flaw in the interface) */ |
361 | if ((int)len < 0) { |
362 | gz_error(state, Z_STREAM_ERROR, "request does not fit in an int" ); |
363 | return -1; |
364 | } |
365 | |
366 | /* read len or fewer bytes to buf */ |
367 | len = (unsigned)gz_read(state, buf, len); |
368 | |
369 | /* check for an error */ |
370 | if (len == 0 && state->err != Z_OK && state->err != Z_BUF_ERROR) |
371 | return -1; |
372 | |
373 | /* return the number of bytes read (this is assured to fit in an int) */ |
374 | return (int)len; |
375 | } |
376 | |
377 | /* -- see zlib.h -- */ |
378 | size_t ZEXPORT PREFIX(gzfread)(void *buf, size_t size, size_t nitems, gzFile file) { |
379 | size_t len; |
380 | gz_state *state; |
381 | |
382 | /* Exit early if size is zero, also prevents potential division by zero */ |
383 | if (size == 0) |
384 | return 0; |
385 | |
386 | /* get internal structure */ |
387 | if (file == NULL) |
388 | return 0; |
389 | state = (gz_state *)file; |
390 | |
391 | /* check that we're reading and that there's no (serious) error */ |
392 | if (state->mode != GZ_READ || |
393 | (state->err != Z_OK && state->err != Z_BUF_ERROR)) |
394 | return 0; |
395 | |
396 | /* compute bytes to read -- error on overflow */ |
397 | len = nitems * size; |
398 | if (size && len / size != nitems) { |
399 | gz_error(state, Z_STREAM_ERROR, "request does not fit in a size_t" ); |
400 | return 0; |
401 | } |
402 | |
403 | /* read len or fewer bytes to buf, return the number of full items read */ |
404 | return len ? gz_read(state, buf, len) / size : 0; |
405 | } |
406 | |
407 | /* -- see zlib.h -- */ |
408 | #undef gzgetc |
409 | #undef zng_gzgetc |
410 | int ZEXPORT PREFIX(gzgetc)(gzFile file) { |
411 | unsigned char buf[1]; |
412 | gz_state *state; |
413 | |
414 | /* get internal structure */ |
415 | if (file == NULL) |
416 | return -1; |
417 | state = (gz_state *)file; |
418 | |
419 | /* check that we're reading and that there's no (serious) error */ |
420 | if (state->mode != GZ_READ || (state->err != Z_OK && state->err != Z_BUF_ERROR)) |
421 | return -1; |
422 | |
423 | /* try output buffer (no need to check for skip request) */ |
424 | if (state->x.have) { |
425 | state->x.have--; |
426 | state->x.pos++; |
427 | return *(state->x.next)++; |
428 | } |
429 | |
430 | /* nothing there -- try gz_read() */ |
431 | return gz_read(state, buf, 1) < 1 ? -1 : buf[0]; |
432 | } |
433 | |
434 | int ZEXPORT PREFIX(gzgetc_)(gzFile file) { |
435 | return PREFIX(gzgetc)(file); |
436 | } |
437 | |
438 | /* -- see zlib.h -- */ |
439 | int ZEXPORT PREFIX(gzungetc)(int c, gzFile file) { |
440 | gz_state *state; |
441 | |
442 | /* get internal structure */ |
443 | if (file == NULL) |
444 | return -1; |
445 | state = (gz_state *)file; |
446 | |
447 | /* check that we're reading and that there's no (serious) error */ |
448 | if (state->mode != GZ_READ || (state->err != Z_OK && state->err != Z_BUF_ERROR)) |
449 | return -1; |
450 | |
451 | /* process a skip request */ |
452 | if (state->seek) { |
453 | state->seek = 0; |
454 | if (gz_skip(state, state->skip) == -1) |
455 | return -1; |
456 | } |
457 | |
458 | /* can't push EOF */ |
459 | if (c < 0) |
460 | return -1; |
461 | |
462 | /* if output buffer empty, put byte at end (allows more pushing) */ |
463 | if (state->x.have == 0) { |
464 | state->x.have = 1; |
465 | state->x.next = state->out + (state->size << 1) - 1; |
466 | state->x.next[0] = (unsigned char)c; |
467 | state->x.pos--; |
468 | state->past = 0; |
469 | return c; |
470 | } |
471 | |
472 | /* if no room, give up (must have already done a gzungetc()) */ |
473 | if (state->x.have == (state->size << 1)) { |
474 | gz_error(state, Z_DATA_ERROR, "out of room to push characters" ); |
475 | return -1; |
476 | } |
477 | |
478 | /* slide output data if needed and insert byte before existing data */ |
479 | if (state->x.next == state->out) { |
480 | unsigned char *src = state->out + state->x.have; |
481 | unsigned char *dest = state->out + (state->size << 1); |
482 | while (src > state->out) |
483 | *--dest = *--src; |
484 | state->x.next = dest; |
485 | } |
486 | state->x.have++; |
487 | state->x.next--; |
488 | state->x.next[0] = (unsigned char)c; |
489 | state->x.pos--; |
490 | state->past = 0; |
491 | return c; |
492 | } |
493 | |
494 | /* -- see zlib.h -- */ |
495 | char * ZEXPORT PREFIX(gzgets)(gzFile file, char *buf, int len) { |
496 | unsigned left, n; |
497 | char *str; |
498 | unsigned char *eol; |
499 | gz_state *state; |
500 | |
501 | /* check parameters and get internal structure */ |
502 | if (file == NULL || buf == NULL || len < 1) |
503 | return NULL; |
504 | state = (gz_state *)file; |
505 | |
506 | /* check that we're reading and that there's no (serious) error */ |
507 | if (state->mode != GZ_READ || (state->err != Z_OK && state->err != Z_BUF_ERROR)) |
508 | return NULL; |
509 | |
510 | /* process a skip request */ |
511 | if (state->seek) { |
512 | state->seek = 0; |
513 | if (gz_skip(state, state->skip) == -1) |
514 | return NULL; |
515 | } |
516 | |
517 | /* copy output bytes up to new line or len - 1, whichever comes first -- |
518 | append a terminating zero to the string (we don't check for a zero in |
519 | the contents, let the user worry about that) */ |
520 | str = buf; |
521 | left = (unsigned)len - 1; |
522 | if (left) do { |
523 | /* assure that something is in the output buffer */ |
524 | if (state->x.have == 0 && gz_fetch(state) == -1) |
525 | return NULL; /* error */ |
526 | if (state->x.have == 0) { /* end of file */ |
527 | state->past = 1; /* read past end */ |
528 | break; /* return what we have */ |
529 | } |
530 | |
531 | /* look for end-of-line in current output buffer */ |
532 | n = state->x.have > left ? left : state->x.have; |
533 | eol = (unsigned char *)memchr(state->x.next, '\n', n); |
534 | if (eol != NULL) |
535 | n = (unsigned)(eol - state->x.next) + 1; |
536 | |
537 | /* copy through end-of-line, or remainder if not found */ |
538 | memcpy(buf, state->x.next, n); |
539 | state->x.have -= n; |
540 | state->x.next += n; |
541 | state->x.pos += n; |
542 | left -= n; |
543 | buf += n; |
544 | } while (left && eol == NULL); |
545 | |
546 | /* return terminated string, or if nothing, end of file */ |
547 | if (buf == str) |
548 | return NULL; |
549 | buf[0] = 0; |
550 | return str; |
551 | } |
552 | |
553 | /* -- see zlib.h -- */ |
554 | int ZEXPORT PREFIX(gzdirect)(gzFile file) { |
555 | gz_state *state; |
556 | |
557 | /* get internal structure */ |
558 | if (file == NULL) |
559 | return 0; |
560 | |
561 | state = (gz_state *)file; |
562 | |
563 | /* if the state is not known, but we can find out, then do so (this is |
564 | mainly for right after a gzopen() or gzdopen()) */ |
565 | if (state->mode == GZ_READ && state->how == LOOK && state->x.have == 0) |
566 | (void)gz_look(state); |
567 | |
568 | /* return 1 if transparent, 0 if processing a gzip stream */ |
569 | return state->direct; |
570 | } |
571 | |
572 | /* -- see zlib.h -- */ |
573 | int ZEXPORT PREFIX(gzclose_r)(gzFile file) { |
574 | int ret, err; |
575 | gz_state *state; |
576 | |
577 | /* get internal structure */ |
578 | if (file == NULL) |
579 | return Z_STREAM_ERROR; |
580 | |
581 | state = (gz_state *)file; |
582 | |
583 | /* check that we're reading */ |
584 | if (state->mode != GZ_READ) |
585 | return Z_STREAM_ERROR; |
586 | |
587 | /* free memory and close file */ |
588 | if (state->size) { |
589 | PREFIX(inflateEnd)(&(state->strm)); |
590 | free(state->out); |
591 | free(state->in); |
592 | } |
593 | err = state->err == Z_BUF_ERROR ? Z_BUF_ERROR : Z_OK; |
594 | gz_error(state, Z_OK, NULL); |
595 | free(state->path); |
596 | ret = close(state->fd); |
597 | free(state); |
598 | return ret ? Z_ERRNO : err; |
599 | } |
600 | |