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
12 local
int gz_load
OF((gz_statep
, unsigned char *, unsigned, unsigned *));
13 local
int gz_avail
OF((gz_statep
));
14 local
int gz_look
OF((gz_statep
));
15 local
int gz_decomp
OF((gz_statep
));
16 local
int gz_fetch
OF((gz_statep
));
17 local
int gz_skip
OF((gz_statep
, z_off64_t
));
19 /* Use read() to load a buffer -- return -1 on error, otherwise 0. Read from
20 state->fd, and update state->eof, state->err, and state->msg as appropriate.
21 This function needs to loop on read(), since read() is not guaranteed to
22 read the number of bytes requested, depending on the type of descriptor. */
23 local
int gz_load(state
, buf
, len
, have
)
33 ret
= read(state
->fd
, buf
+ *have
, len
- *have
);
37 } while (*have
< len
);
39 gz_error(state
, Z_ERRNO
, zstrerror());
47 /* Load up input buffer and set eof flag if last data loaded -- return -1 on
48 error, 0 otherwise. Note that the eof flag is set when the end of the input
49 file is reached, even though there may be unused data in the buffer. Once
50 that data has been used, no more attempts will be made to read the file.
51 If strm->avail_in != 0, then the current data is moved to the beginning of
52 the input buffer, and then the remainder of the buffer is loaded with the
53 available data from the input file. */
54 local
int gz_avail(state
)
58 z_streamp strm
= &(state
->strm
);
60 if (state
->err
!= Z_OK
&& state
->err
!= Z_BUF_ERROR
)
62 if (state
->eof
== 0) {
63 if (strm
->avail_in
) { /* copy what's there to the start */
64 unsigned char *p
= state
->in
;
65 unsigned const char *q
= strm
->next_in
;
66 unsigned n
= strm
->avail_in
;
71 if (gz_load(state
, state
->in
+ strm
->avail_in
,
72 state
->size
- strm
->avail_in
, &got
) == -1)
74 strm
->avail_in
+= got
;
75 strm
->next_in
= state
->in
;
80 /* Look for gzip header, set up for inflate or copy. state->x.have must be 0.
81 If this is the first time in, allocate required memory. state->how will be
82 left unchanged if there is no more input data available, will be set to COPY
83 if there is no gzip header and direct copying will be performed, or it will
84 be set to GZIP for decompression. If direct copying, then leftover input
85 data from the input buffer will be copied to the output buffer. In that
86 case, all further file reads will be directly to either the output buffer or
87 a user buffer. If decompressing, the inflate state will be initialized.
88 gz_look() will return 0 on success or -1 on failure. */
89 local
int gz_look(state
)
92 z_streamp strm
= &(state
->strm
);
94 /* allocate read buffers and inflate memory */
95 if (state
->size
== 0) {
96 /* allocate buffers */
97 state
->in
= (unsigned char *)malloc(state
->want
);
98 state
->out
= (unsigned char *)malloc(state
->want
<< 1);
99 if (state
->in
== NULL
|| state
->out
== NULL
) {
100 if (state
->out
!= NULL
)
102 if (state
->in
!= NULL
)
104 gz_error(state
, Z_MEM_ERROR
, "out of memory");
107 state
->size
= state
->want
;
109 /* allocate inflate memory */
110 state
->strm
.zalloc
= Z_NULL
;
111 state
->strm
.zfree
= Z_NULL
;
112 state
->strm
.opaque
= Z_NULL
;
113 state
->strm
.avail_in
= 0;
114 state
->strm
.next_in
= Z_NULL
;
115 if (inflateInit2(&(state
->strm
), 15 + 16) != Z_OK
) { /* gunzip */
119 gz_error(state
, Z_MEM_ERROR
, "out of memory");
124 /* get at least the magic bytes in the input buffer */
125 if (strm
->avail_in
< 2) {
126 if (gz_avail(state
) == -1)
128 if (strm
->avail_in
== 0)
132 /* look for gzip magic bytes -- if there, do gzip decoding (note: there is
133 a logical dilemma here when considering the case of a partially written
134 gzip file, to wit, if a single 31 byte is written, then we cannot tell
135 whether this is a single-byte file, or just a partially written gzip
136 file -- for here we assume that if a gzip file is being written, then
137 the header will be written in a single operation, so that reading a
138 single byte is sufficient indication that it is not a gzip file) */
139 if (strm
->avail_in
> 1 &&
140 strm
->next_in
[0] == 31 && strm
->next_in
[1] == 139) {
147 /* no gzip header -- if we were decoding gzip before, then this is trailing
148 garbage. Ignore the trailing garbage and finish. */
149 if (state
->direct
== 0) {
156 /* doing raw i/o, copy any leftover input to output -- this assumes that
157 the output buffer is larger than the input buffer, which also assures
158 space for gzungetc() */
159 state
->x
.next
= state
->out
;
160 if (strm
->avail_in
) {
161 memcpy(state
->x
.next
, strm
->next_in
, strm
->avail_in
);
162 state
->x
.have
= strm
->avail_in
;
170 /* Decompress from input to the provided next_out and avail_out in the state.
171 On return, state->x.have and state->x.next point to the just decompressed
172 data. If the gzip stream completes, state->how is reset to LOOK to look for
173 the next gzip stream or raw data, once state->x.have is depleted. Returns 0
174 on success, -1 on failure. */
175 local
int gz_decomp(state
)
180 z_streamp strm
= &(state
->strm
);
182 /* fill output buffer up to end of deflate stream */
183 had
= strm
->avail_out
;
185 /* get more input for inflate() */
186 if (strm
->avail_in
== 0 && gz_avail(state
) == -1)
188 if (strm
->avail_in
== 0) {
189 gz_error(state
, Z_BUF_ERROR
, "unexpected end of file");
193 /* decompress and handle errors */
194 ret
= inflate(strm
, Z_NO_FLUSH
);
195 if (ret
== Z_STREAM_ERROR
|| ret
== Z_NEED_DICT
) {
196 gz_error(state
, Z_STREAM_ERROR
,
197 "internal error: inflate stream corrupt");
200 if (ret
== Z_MEM_ERROR
) {
201 gz_error(state
, Z_MEM_ERROR
, "out of memory");
204 if (ret
== Z_DATA_ERROR
) { /* deflate stream invalid */
205 gz_error(state
, Z_DATA_ERROR
,
206 strm
->msg
== NULL
? "compressed data error" : strm
->msg
);
209 } while (strm
->avail_out
&& ret
!= Z_STREAM_END
);
211 /* update available output */
212 state
->x
.have
= had
- strm
->avail_out
;
213 state
->x
.next
= strm
->next_out
- state
->x
.have
;
215 /* if the gzip stream completed successfully, look for another */
216 if (ret
== Z_STREAM_END
)
219 /* good decompression */
223 /* Fetch data and put it in the output buffer. Assumes state->x.have is 0.
224 Data is either copied from the input file or decompressed from the input
225 file depending on state->how. If state->how is LOOK, then a gzip header is
226 looked for to determine whether to copy or decompress. Returns -1 on error,
227 otherwise 0. gz_fetch() will leave state->how as COPY or GZIP unless the
228 end of the input file has been reached and all data has been processed. */
229 local
int gz_fetch(state
)
232 z_streamp strm
= &(state
->strm
);
236 case LOOK
: /* -> LOOK, COPY (only if never GZIP), or GZIP */
237 if (gz_look(state
) == -1)
239 if (state
->how
== LOOK
)
242 case COPY
: /* -> COPY */
243 if (gz_load(state
, state
->out
, state
->size
<< 1, &(state
->x
.have
))
246 state
->x
.next
= state
->out
;
248 case GZIP
: /* -> GZIP or LOOK (if end of gzip stream) */
249 strm
->avail_out
= state
->size
<< 1;
250 strm
->next_out
= state
->out
;
251 if (gz_decomp(state
) == -1)
254 } while (state
->x
.have
== 0 && (!state
->eof
|| strm
->avail_in
));
258 /* Skip len uncompressed bytes of output. Return -1 on error, 0 on success. */
259 local
int gz_skip(state
, len
)
265 /* skip over len bytes or reach end-of-file, whichever comes first */
267 /* skip over whatever is in output buffer */
269 n
= GT_OFF(state
->x
.have
) || (z_off64_t
)state
->x
.have
> len
?
270 (unsigned)len
: state
->x
.have
;
277 /* output buffer empty -- return if we're at the end of the input */
278 else if (state
->eof
&& state
->strm
.avail_in
== 0)
281 /* need more data to skip -- load up output buffer */
283 /* get more output, looking for header if required */
284 if (gz_fetch(state
) == -1)
290 /* -- see zlib.h -- */
291 int ZEXPORT
gzread(file
, buf
, len
)
300 /* get internal structure */
303 state
= (gz_statep
)file
;
304 strm
= &(state
->strm
);
306 /* check that we're reading and that there's no (serious) error */
307 if (state
->mode
!= GZ_READ
||
308 (state
->err
!= Z_OK
&& state
->err
!= Z_BUF_ERROR
))
311 /* since an int is returned, make sure len fits in one, otherwise return
312 with an error (this avoids the flaw in the interface) */
314 gz_error(state
, Z_DATA_ERROR
, "requested length does not fit in int");
318 /* if len is zero, avoid unnecessary operations */
322 /* process a skip request */
325 if (gz_skip(state
, state
->skip
) == -1)
329 /* get len bytes to buf, or less than len if at the end */
332 /* first just try copying data from the output buffer */
334 n
= state
->x
.have
> len
? len
: state
->x
.have
;
335 memcpy(buf
, state
->x
.next
, n
);
340 /* output buffer empty -- return if we're at the end of the input */
341 else if (state
->eof
&& strm
->avail_in
== 0) {
342 state
->past
= 1; /* tried to read past end */
346 /* need output data -- for small len or new stream load up our output
348 else if (state
->how
== LOOK
|| len
< (state
->size
<< 1)) {
349 /* get more output, looking for header if required */
350 if (gz_fetch(state
) == -1)
352 continue; /* no progress yet -- go back to copy above */
353 /* the copy above assures that we will leave with space in the
354 output buffer, allowing at least one gzungetc() to succeed */
357 /* large len -- read directly into user buffer */
358 else if (state
->how
== COPY
) { /* read directly */
359 if (gz_load(state
, (unsigned char *)buf
, len
, &n
) == -1)
363 /* large len -- decompress directly into user buffer */
364 else { /* state->how == GZIP */
365 strm
->avail_out
= len
;
366 strm
->next_out
= (unsigned char *)buf
;
367 if (gz_decomp(state
) == -1)
373 /* update progress */
375 buf
= (char *)buf
+ n
;
380 /* return number of bytes read into user buffer (will fit in int) */
384 /* -- see zlib.h -- */
390 int ZEXPORT
gzgetc(file
)
394 unsigned char buf
[1];
397 /* get internal structure */
400 state
= (gz_statep
)file
;
402 /* check that we're reading and that there's no (serious) error */
403 if (state
->mode
!= GZ_READ
||
404 (state
->err
!= Z_OK
&& state
->err
!= Z_BUF_ERROR
))
407 /* try output buffer (no need to check for skip request) */
411 return *(state
->x
.next
)++;
414 /* nothing there -- try gzread() */
415 ret
= gzread(file
, buf
, 1);
416 return ret
< 1 ? -1 : buf
[0];
419 int ZEXPORT
gzgetc_(file
)
425 /* -- see zlib.h -- */
426 int ZEXPORT
gzungetc(c
, file
)
432 /* get internal structure */
435 state
= (gz_statep
)file
;
437 /* check that we're reading and that there's no (serious) error */
438 if (state
->mode
!= GZ_READ
||
439 (state
->err
!= Z_OK
&& state
->err
!= Z_BUF_ERROR
))
442 /* process a skip request */
445 if (gz_skip(state
, state
->skip
) == -1)
453 /* if output buffer empty, put byte at end (allows more pushing) */
454 if (state
->x
.have
== 0) {
456 state
->x
.next
= state
->out
+ (state
->size
<< 1) - 1;
457 state
->x
.next
[0] = c
;
463 /* if no room, give up (must have already done a gzungetc()) */
464 if (state
->x
.have
== (state
->size
<< 1)) {
465 gz_error(state
, Z_DATA_ERROR
, "out of room to push characters");
469 /* slide output data if needed and insert byte before existing data */
470 if (state
->x
.next
== state
->out
) {
471 unsigned char *src
= state
->out
+ state
->x
.have
;
472 unsigned char *dest
= state
->out
+ (state
->size
<< 1);
473 while (src
> state
->out
)
475 state
->x
.next
= dest
;
479 state
->x
.next
[0] = c
;
485 /* -- see zlib.h -- */
486 char * ZEXPORT
gzgets(file
, buf
, len
)
496 /* check parameters and get internal structure */
497 if (file
== NULL
|| buf
== NULL
|| len
< 1)
499 state
= (gz_statep
)file
;
501 /* check that we're reading and that there's no (serious) error */
502 if (state
->mode
!= GZ_READ
||
503 (state
->err
!= Z_OK
&& state
->err
!= Z_BUF_ERROR
))
506 /* process a skip request */
509 if (gz_skip(state
, state
->skip
) == -1)
513 /* copy output bytes up to new line or len - 1, whichever comes first --
514 append a terminating zero to the string (we don't check for a zero in
515 the contents, let the user worry about that) */
517 left
= (unsigned)len
- 1;
519 /* assure that something is in the output buffer */
520 if (state
->x
.have
== 0 && gz_fetch(state
) == -1)
521 return NULL
; /* error */
522 if (state
->x
.have
== 0) { /* end of file */
523 state
->past
= 1; /* read past end */
524 break; /* return what we have */
527 /* look for end-of-line in current output buffer */
528 n
= state
->x
.have
> left
? left
: state
->x
.have
;
529 eol
= (unsigned char *)memchr(state
->x
.next
, '\n', n
);
531 n
= (unsigned)(eol
- state
->x
.next
) + 1;
533 /* copy through end-of-line, or remainder if not found */
534 memcpy(buf
, state
->x
.next
, n
);
540 } while (left
&& eol
== NULL
);
542 /* return terminated string, or if nothing, end of file */
549 /* -- see zlib.h -- */
550 int ZEXPORT
gzdirect(file
)
555 /* get internal structure */
558 state
= (gz_statep
)file
;
560 /* if the state is not known, but we can find out, then do so (this is
561 mainly for right after a gzopen() or gzdopen()) */
562 if (state
->mode
== GZ_READ
&& state
->how
== LOOK
&& state
->x
.have
== 0)
563 (void)gz_look(state
);
565 /* return 1 if transparent, 0 if processing a gzip stream */
566 return state
->direct
;
569 /* -- see zlib.h -- */
570 int ZEXPORT
gzclose_r(file
)
576 /* get internal structure */
578 return Z_STREAM_ERROR
;
579 state
= (gz_statep
)file
;
581 /* check that we're reading */
582 if (state
->mode
!= GZ_READ
)
583 return Z_STREAM_ERROR
;
585 /* free memory and close file */
587 inflateEnd(&(state
->strm
));
591 err
= state
->err
== Z_BUF_ERROR
? Z_BUF_ERROR
: Z_OK
;
592 gz_error(state
, Z_OK
, NULL
);
594 ret
= close(state
->fd
);
596 return ret
? Z_ERRNO
: err
;