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
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
));
16 /* Use read() to load a buffer -- return -1 on error, otherwise 0. Read from
17 state->fd, and update state->eof, state->err, and state->msg as appropriate.
18 This function needs to loop on read(), since read() is not guaranteed to
19 read the number of bytes requested, depending on the type of descriptor. */
20 local
int gz_load(state
, buf
, len
, have
)
30 ret
= read(state
->fd
, buf
+ *have
, len
- *have
);
34 } while (*have
< len
);
36 gz_error(state
, Z_ERRNO
, zstrerror());
44 /* Load up input buffer and set eof flag if last data loaded -- return -1 on
45 error, 0 otherwise. Note that the eof flag is set when the end of the input
46 file is reached, even though there may be unused data in the buffer. Once
47 that data has been used, no more attempts will be made to read the file.
48 If strm->avail_in != 0, then the current data is moved to the beginning of
49 the input buffer, and then the remainder of the buffer is loaded with the
50 available data from the input file. */
51 local
int gz_avail(state
)
55 z_streamp strm
= &(state
->strm
);
57 if (state
->err
!= Z_OK
&& state
->err
!= Z_BUF_ERROR
)
59 if (state
->eof
== 0) {
60 if (strm
->avail_in
) { /* copy what's there to the start */
61 unsigned char *p
= state
->in
;
62 unsigned const char *q
= strm
->next_in
;
63 unsigned n
= strm
->avail_in
;
68 if (gz_load(state
, state
->in
+ strm
->avail_in
,
69 state
->size
- strm
->avail_in
, &got
) == -1)
71 strm
->avail_in
+= got
;
72 strm
->next_in
= state
->in
;
77 /* Look for gzip header, set up for inflate or copy. state->x.have must be 0.
78 If this is the first time in, allocate required memory. state->how will be
79 left unchanged if there is no more input data available, will be set to COPY
80 if there is no gzip header and direct copying will be performed, or it will
81 be set to GZIP for decompression. If direct copying, then leftover input
82 data from the input buffer will be copied to the output buffer. In that
83 case, all further file reads will be directly to either the output buffer or
84 a user buffer. If decompressing, the inflate state will be initialized.
85 gz_look() will return 0 on success or -1 on failure. */
86 local
int gz_look(state
)
89 z_streamp strm
= &(state
->strm
);
91 /* allocate read buffers and inflate memory */
92 if (state
->size
== 0) {
93 /* allocate buffers */
94 state
->in
= (unsigned char *)malloc(state
->want
);
95 state
->out
= (unsigned char *)malloc(state
->want
<< 1);
96 if (state
->in
== NULL
|| state
->out
== NULL
) {
97 if (state
->out
!= NULL
)
99 if (state
->in
!= NULL
)
101 gz_error(state
, Z_MEM_ERROR
, "out of memory");
104 state
->size
= state
->want
;
106 /* allocate inflate memory */
107 state
->strm
.zalloc
= Z_NULL
;
108 state
->strm
.zfree
= Z_NULL
;
109 state
->strm
.opaque
= Z_NULL
;
110 state
->strm
.avail_in
= 0;
111 state
->strm
.next_in
= Z_NULL
;
112 if (inflateInit2(&(state
->strm
), 15 + 16) != Z_OK
) { /* gunzip */
116 gz_error(state
, Z_MEM_ERROR
, "out of memory");
121 /* get at least the magic bytes in the input buffer */
122 if (strm
->avail_in
< 2) {
123 if (gz_avail(state
) == -1)
125 if (strm
->avail_in
== 0)
129 /* look for gzip magic bytes -- if there, do gzip decoding (note: there is
130 a logical dilemma here when considering the case of a partially written
131 gzip file, to wit, if a single 31 byte is written, then we cannot tell
132 whether this is a single-byte file, or just a partially written gzip
133 file -- for here we assume that if a gzip file is being written, then
134 the header will be written in a single operation, so that reading a
135 single byte is sufficient indication that it is not a gzip file) */
136 if (strm
->avail_in
> 1 &&
137 strm
->next_in
[0] == 31 && strm
->next_in
[1] == 139) {
144 /* no gzip header -- if we were decoding gzip before, then this is trailing
145 garbage. Ignore the trailing garbage and finish. */
146 if (state
->direct
== 0) {
153 /* doing raw i/o, copy any leftover input to output -- this assumes that
154 the output buffer is larger than the input buffer, which also assures
155 space for gzungetc() */
156 state
->x
.next
= state
->out
;
157 if (strm
->avail_in
) {
158 memcpy(state
->x
.next
, strm
->next_in
, strm
->avail_in
);
159 state
->x
.have
= strm
->avail_in
;
167 /* Decompress from input to the provided next_out and avail_out in the state.
168 On return, state->x.have and state->x.next point to the just decompressed
169 data. If the gzip stream completes, state->how is reset to LOOK to look for
170 the next gzip stream or raw data, once state->x.have is depleted. Returns 0
171 on success, -1 on failure. */
172 local
int gz_decomp(state
)
177 z_streamp strm
= &(state
->strm
);
179 /* fill output buffer up to end of deflate stream */
180 had
= strm
->avail_out
;
182 /* get more input for inflate() */
183 if (strm
->avail_in
== 0 && gz_avail(state
) == -1)
185 if (strm
->avail_in
== 0) {
186 gz_error(state
, Z_BUF_ERROR
, "unexpected end of file");
190 /* decompress and handle errors */
191 ret
= inflate(strm
, Z_NO_FLUSH
);
192 if (ret
== Z_STREAM_ERROR
|| ret
== Z_NEED_DICT
) {
193 gz_error(state
, Z_STREAM_ERROR
,
194 "internal error: inflate stream corrupt");
197 if (ret
== Z_MEM_ERROR
) {
198 gz_error(state
, Z_MEM_ERROR
, "out of memory");
201 if (ret
== Z_DATA_ERROR
) { /* deflate stream invalid */
202 gz_error(state
, Z_DATA_ERROR
,
203 strm
->msg
== NULL
? "compressed data error" : strm
->msg
);
206 } while (strm
->avail_out
&& ret
!= Z_STREAM_END
);
208 /* update available output */
209 state
->x
.have
= had
- strm
->avail_out
;
210 state
->x
.next
= strm
->next_out
- state
->x
.have
;
212 /* if the gzip stream completed successfully, look for another */
213 if (ret
== Z_STREAM_END
)
216 /* good decompression */
220 /* Fetch data and put it in the output buffer. Assumes state->x.have is 0.
221 Data is either copied from the input file or decompressed from the input
222 file depending on state->how. If state->how is LOOK, then a gzip header is
223 looked for to determine whether to copy or decompress. Returns -1 on error,
224 otherwise 0. gz_fetch() will leave state->how as COPY or GZIP unless the
225 end of the input file has been reached and all data has been processed. */
226 local
int gz_fetch(state
)
229 z_streamp strm
= &(state
->strm
);
233 case LOOK
: /* -> LOOK, COPY (only if never GZIP), or GZIP */
234 if (gz_look(state
) == -1)
236 if (state
->how
== LOOK
)
239 case COPY
: /* -> COPY */
240 if (gz_load(state
, state
->out
, state
->size
<< 1, &(state
->x
.have
))
243 state
->x
.next
= state
->out
;
245 case GZIP
: /* -> GZIP or LOOK (if end of gzip stream) */
246 strm
->avail_out
= state
->size
<< 1;
247 strm
->next_out
= state
->out
;
248 if (gz_decomp(state
) == -1)
251 } while (state
->x
.have
== 0 && (!state
->eof
|| strm
->avail_in
));
255 /* Skip len uncompressed bytes of output. Return -1 on error, 0 on success. */
256 local
int gz_skip(state
, len
)
262 /* skip over len bytes or reach end-of-file, whichever comes first */
264 /* skip over whatever is in output buffer */
266 n
= GT_OFF(state
->x
.have
) || (z_off64_t
)state
->x
.have
> len
?
267 (unsigned)len
: state
->x
.have
;
274 /* output buffer empty -- return if we're at the end of the input */
275 else if (state
->eof
&& state
->strm
.avail_in
== 0)
278 /* need more data to skip -- load up output buffer */
280 /* get more output, looking for header if required */
281 if (gz_fetch(state
) == -1)
287 /* -- see zlib.h -- */
288 int ZEXPORT
gzread(file
, buf
, len
)
297 /* get internal structure */
300 state
= (gz_statep
)file
;
301 strm
= &(state
->strm
);
303 /* check that we're reading and that there's no (serious) error */
304 if (state
->mode
!= GZ_READ
||
305 (state
->err
!= Z_OK
&& state
->err
!= Z_BUF_ERROR
))
308 /* since an int is returned, make sure len fits in one, otherwise return
309 with an error (this avoids the flaw in the interface) */
311 gz_error(state
, Z_DATA_ERROR
, "requested length does not fit in int");
315 /* if len is zero, avoid unnecessary operations */
319 /* process a skip request */
322 if (gz_skip(state
, state
->skip
) == -1)
326 /* get len bytes to buf, or less than len if at the end */
329 /* first just try copying data from the output buffer */
331 n
= state
->x
.have
> len
? len
: state
->x
.have
;
332 memcpy(buf
, state
->x
.next
, n
);
337 /* output buffer empty -- return if we're at the end of the input */
338 else if (state
->eof
&& strm
->avail_in
== 0) {
339 state
->past
= 1; /* tried to read past end */
343 /* need output data -- for small len or new stream load up our output
345 else if (state
->how
== LOOK
|| len
< (state
->size
<< 1)) {
346 /* get more output, looking for header if required */
347 if (gz_fetch(state
) == -1)
349 continue; /* no progress yet -- go back to copy above */
350 /* the copy above assures that we will leave with space in the
351 output buffer, allowing at least one gzungetc() to succeed */
354 /* large len -- read directly into user buffer */
355 else if (state
->how
== COPY
) { /* read directly */
356 if (gz_load(state
, (unsigned char *)buf
, len
, &n
) == -1)
360 /* large len -- decompress directly into user buffer */
361 else { /* state->how == GZIP */
362 strm
->avail_out
= len
;
363 strm
->next_out
= (unsigned char *)buf
;
364 if (gz_decomp(state
) == -1)
370 /* update progress */
372 buf
= (char *)buf
+ n
;
377 /* return number of bytes read into user buffer (will fit in int) */
381 /* -- see zlib.h -- */
387 int ZEXPORT
gzgetc(file
)
391 unsigned char buf
[1];
394 /* get internal structure */
397 state
= (gz_statep
)file
;
399 /* check that we're reading and that there's no (serious) error */
400 if (state
->mode
!= GZ_READ
||
401 (state
->err
!= Z_OK
&& state
->err
!= Z_BUF_ERROR
))
404 /* try output buffer (no need to check for skip request) */
408 return *(state
->x
.next
)++;
411 /* nothing there -- try gzread() */
412 ret
= gzread(file
, buf
, 1);
413 return ret
< 1 ? -1 : buf
[0];
416 int ZEXPORT
gzgetc_(file
)
422 /* -- see zlib.h -- */
423 int ZEXPORT
gzungetc(c
, file
)
429 /* get internal structure */
432 state
= (gz_statep
)file
;
434 /* check that we're reading and that there's no (serious) error */
435 if (state
->mode
!= GZ_READ
||
436 (state
->err
!= Z_OK
&& state
->err
!= Z_BUF_ERROR
))
439 /* process a skip request */
442 if (gz_skip(state
, state
->skip
) == -1)
450 /* if output buffer empty, put byte at end (allows more pushing) */
451 if (state
->x
.have
== 0) {
453 state
->x
.next
= state
->out
+ (state
->size
<< 1) - 1;
454 state
->x
.next
[0] = c
;
460 /* if no room, give up (must have already done a gzungetc()) */
461 if (state
->x
.have
== (state
->size
<< 1)) {
462 gz_error(state
, Z_DATA_ERROR
, "out of room to push characters");
466 /* slide output data if needed and insert byte before existing data */
467 if (state
->x
.next
== state
->out
) {
468 unsigned char *src
= state
->out
+ state
->x
.have
;
469 unsigned char *dest
= state
->out
+ (state
->size
<< 1);
470 while (src
> state
->out
)
472 state
->x
.next
= dest
;
476 state
->x
.next
[0] = c
;
482 /* -- see zlib.h -- */
483 char * ZEXPORT
gzgets(file
, buf
, len
)
493 /* check parameters and get internal structure */
494 if (file
== NULL
|| buf
== NULL
|| len
< 1)
496 state
= (gz_statep
)file
;
498 /* check that we're reading and that there's no (serious) error */
499 if (state
->mode
!= GZ_READ
||
500 (state
->err
!= Z_OK
&& state
->err
!= Z_BUF_ERROR
))
503 /* process a skip request */
506 if (gz_skip(state
, state
->skip
) == -1)
510 /* copy output bytes up to new line or len - 1, whichever comes first --
511 append a terminating zero to the string (we don't check for a zero in
512 the contents, let the user worry about that) */
514 left
= (unsigned)len
- 1;
516 /* assure that something is in the output buffer */
517 if (state
->x
.have
== 0 && gz_fetch(state
) == -1)
518 return NULL
; /* error */
519 if (state
->x
.have
== 0) { /* end of file */
520 state
->past
= 1; /* read past end */
521 break; /* return what we have */
524 /* look for end-of-line in current output buffer */
525 n
= state
->x
.have
> left
? left
: state
->x
.have
;
526 eol
= (unsigned char *)memchr(state
->x
.next
, '\n', n
);
528 n
= (unsigned)(eol
- state
->x
.next
) + 1;
530 /* copy through end-of-line, or remainder if not found */
531 memcpy(buf
, state
->x
.next
, n
);
537 } while (left
&& eol
== NULL
);
539 /* return terminated string, or if nothing, end of file */
546 /* -- see zlib.h -- */
547 int ZEXPORT
gzdirect(file
)
552 /* get internal structure */
555 state
= (gz_statep
)file
;
557 /* if the state is not known, but we can find out, then do so (this is
558 mainly for right after a gzopen() or gzdopen()) */
559 if (state
->mode
== GZ_READ
&& state
->how
== LOOK
&& state
->x
.have
== 0)
560 (void)gz_look(state
);
562 /* return 1 if transparent, 0 if processing a gzip stream */
563 return state
->direct
;
566 /* -- see zlib.h -- */
567 int ZEXPORT
gzclose_r(file
)
573 /* get internal structure */
575 return Z_STREAM_ERROR
;
576 state
= (gz_statep
)file
;
578 /* check that we're reading */
579 if (state
->mode
!= GZ_READ
)
580 return Z_STREAM_ERROR
;
582 /* free memory and close file */
584 inflateEnd(&(state
->strm
));
588 err
= state
->err
== Z_BUF_ERROR
? Z_BUF_ERROR
: Z_OK
;
589 gz_error(state
, Z_OK
, NULL
);
591 ret
= close(state
->fd
);
593 return ret
? Z_ERRNO
: err
;