1 /* gzread.c -- zlib functions for reading gzip files
2 * Copyright (C) 2004, 2005, 2010 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_next4 OF((gz_statep, unsigned long *));
12 local int gz_head OF((gz_statep));
13 local int gz_decomp OF((gz_statep));
14 local int gz_make OF((gz_statep));
15 local int gz_skip OF((gz_statep, z_off64_t));
17 /* Use read() to load a buffer -- return -1 on error, otherwise 0. Read from
18 state->fd, and update state->eof, state->err, and state->msg as appropriate.
19 This function needs to loop on read(), since read() is not guaranteed to
20 read the number of bytes requested, depending on the type of descriptor. */
21 local int gz_load(state, buf, len, have)
31 ret = read(state->fd, buf + *have, len - *have);
35 } while (*have < len);
37 gz_error(state, Z_ERRNO, zstrerror());
45 /* Load up input buffer and set eof flag if last data loaded -- return -1 on
46 error, 0 otherwise. Note that the eof flag is set when the end of the input
47 file is reached, even though there may be unused data in the buffer. Once
48 that data has been used, no more attempts will be made to read the file.
49 gz_avail() assumes that strm->avail_in == 0. */
50 local int gz_avail(state)
53 z_streamp strm = &(state->strm);
55 if (state->err != Z_OK)
57 if (state->eof == 0) {
58 if (gz_load(state, state->in, state->size,
59 (unsigned *)&(strm->avail_in)) == -1)
61 strm->next_in = state->in;
66 /* Get next byte from input, or -1 if end or error. */
67 #define NEXT() ((strm->avail_in == 0 && gz_avail(state) == -1) ? -1 : \
68 (strm->avail_in == 0 ? -1 : \
69 (strm->avail_in--, *(strm->next_in)++)))
71 /* Get a four-byte little-endian integer and return 0 on success and the value
72 in *ret. Otherwise -1 is returned and *ret is not modified. */
73 local int gz_next4(state, ret)
79 z_streamp strm = &(state->strm);
82 val += (unsigned)NEXT() << 8;
83 val += (unsigned long)NEXT() << 16;
87 val += (unsigned long)ch << 24;
92 /* Look for gzip header, set up for inflate or copy. state->have must be zero.
93 If this is the first time in, allocate required memory. state->how will be
94 left unchanged if there is no more input data available, will be set to COPY
95 if there is no gzip header and direct copying will be performed, or it will
96 be set to GZIP for decompression, and the gzip header will be skipped so
97 that the next available input data is the raw deflate stream. If direct
98 copying, then leftover input data from the input buffer will be copied to
99 the output buffer. In that case, all further file reads will be directly to
100 either the output buffer or a user buffer. If decompressing, the inflate
101 state and the check value will be initialized. gz_head() will return 0 on
102 success or -1 on failure. Failures may include read errors or gzip header
104 local int gz_head(state)
107 z_streamp strm = &(state->strm);
111 /* allocate read buffers and inflate memory */
112 if (state->size == 0) {
113 /* allocate buffers */
114 state->in = malloc(state->want);
115 state->out = malloc(state->want << 1);
116 if (state->in == NULL || state->out == NULL) {
117 if (state->out != NULL)
119 if (state->in != NULL)
121 gz_error(state, Z_MEM_ERROR, "out of memory");
124 state->size = state->want;
126 /* allocate inflate memory */
127 state->strm.zalloc = Z_NULL;
128 state->strm.zfree = Z_NULL;
129 state->strm.opaque = Z_NULL;
130 state->strm.avail_in = 0;
131 state->strm.next_in = Z_NULL;
132 if (inflateInit2(&(state->strm), -15) != Z_OK) { /* raw inflate */
136 gz_error(state, Z_MEM_ERROR, "out of memory");
141 /* get some data in the input buffer */
142 if (strm->avail_in == 0) {
143 if (gz_avail(state) == -1)
145 if (strm->avail_in == 0)
149 /* look for the gzip magic header bytes 31 and 139 */
150 if (strm->next_in[0] == 31) {
153 if (strm->avail_in == 0 && gz_avail(state) == -1)
155 if (strm->avail_in && strm->next_in[0] == 139) {
156 /* we have a gzip header, woo hoo! */
160 /* skip rest of header */
161 if (NEXT() != 8) { /* compression method */
162 gz_error(state, Z_DATA_ERROR, "unknown compression method");
166 if (flags & 0xe0) { /* reserved flag bits */
167 gz_error(state, Z_DATA_ERROR, "unknown header flags set");
170 NEXT(); /* modification time */
174 NEXT(); /* extra flags */
175 NEXT(); /* operating system */
176 if (flags & 4) { /* extra field */
177 len = (unsigned)NEXT();
178 len += (unsigned)NEXT() << 8;
183 if (flags & 8) /* file name */
186 if (flags & 16) /* comment */
189 if (flags & 2) { /* header crc */
193 /* an unexpected end of file is not checked for here -- it will be
194 noticed on the first request for uncompressed data */
196 /* set up for decompression */
198 strm->adler = crc32(0L, Z_NULL, 0);
204 /* not a gzip file -- save first byte (31) and fall to raw i/o */
210 /* no gzip header -- if we were decoding gzip before, then this is trailing
211 garbage. Ignore the trailing garbage and finish. */
212 if (state->direct == 0) {
219 /* doing raw i/o, save start of raw data for seeking, copy any leftover
220 input to output -- this assumes that the output buffer is larger than
221 the input buffer, which also assures space for gzungetc() */
222 state->raw = state->pos;
223 state->next = state->out;
224 if (strm->avail_in) {
225 memcpy(state->next + state->have, strm->next_in, strm->avail_in);
226 state->have += strm->avail_in;
234 /* Decompress from input to the provided next_out and avail_out in the state.
235 If the end of the compressed data is reached, then verify the gzip trailer
236 check value and length (modulo 2^32). state->have and state->next are set
237 to point to the just decompressed data, and the crc is updated. If the
238 trailer is verified, state->how is reset to LOOK to look for the next gzip
239 stream or raw data, once state->have is depleted. Returns 0 on success, -1
240 on failure. Failures may include invalid compressed data or a failed gzip
241 trailer verification. */
242 local int gz_decomp(state)
247 unsigned long crc, len;
248 z_streamp strm = &(state->strm);
250 /* fill output buffer up to end of deflate stream */
251 had = strm->avail_out;
253 /* get more input for inflate() */
254 if (strm->avail_in == 0 && gz_avail(state) == -1)
256 if (strm->avail_in == 0) {
257 gz_error(state, Z_DATA_ERROR, "unexpected end of file");
261 /* decompress and handle errors */
262 ret = inflate(strm, Z_NO_FLUSH);
263 if (ret == Z_STREAM_ERROR || ret == Z_NEED_DICT) {
264 gz_error(state, Z_STREAM_ERROR,
265 "internal error: inflate stream corrupt");
268 if (ret == Z_MEM_ERROR) {
269 gz_error(state, Z_MEM_ERROR, "out of memory");
272 if (ret == Z_DATA_ERROR) { /* deflate stream invalid */
273 gz_error(state, Z_DATA_ERROR,
274 strm->msg == NULL ? "compressed data error" : strm->msg);
277 } while (strm->avail_out && ret != Z_STREAM_END);
279 /* update available output and crc check value */
280 state->have = had - strm->avail_out;
281 state->next = strm->next_out - state->have;
282 strm->adler = crc32(strm->adler, state->next, state->have);
284 /* check gzip trailer if at end of deflate stream */
285 if (ret == Z_STREAM_END) {
286 if (gz_next4(state, &crc) == -1 || gz_next4(state, &len) == -1) {
287 gz_error(state, Z_DATA_ERROR, "unexpected end of file");
290 if (crc != strm->adler) {
291 gz_error(state, Z_DATA_ERROR, "incorrect data check");
294 if (len != (strm->total_out & 0xffffffffL)) {
295 gz_error(state, Z_DATA_ERROR, "incorrect length check");
298 state->how = LOOK; /* ready for next stream, once have is 0 (leave
299 state->direct unchanged to remember how) */
302 /* good decompression */
306 /* Make data and put in the output buffer. Assumes that state->have == 0.
307 Data is either copied from the input file or decompressed from the input
308 file depending on state->how. If state->how is LOOK, then a gzip header is
309 looked for (and skipped if found) to discern whether to copy or decompress.
310 Returns -1 on error, otherwise 0. gz_make() will leave state->how as COPY
311 or GZIP unless the end of the input file has been reached and all data has
313 local int gz_make(state)
316 z_streamp strm = &(state->strm);
318 if (state->how == LOOK) { /* look for gzip header */
319 if (gz_head(state) == -1)
321 if (state->have) /* got some data from gz_head() */
324 if (state->how == COPY) { /* straight copy */
325 if (gz_load(state, state->out, state->size << 1, &(state->have)) == -1)
327 state->next = state->out;
329 else if (state->how == GZIP) { /* decompress */
330 strm->avail_out = state->size << 1;
331 strm->next_out = state->out;
332 if (gz_decomp(state) == -1)
338 /* Skip len uncompressed bytes of output. Return -1 on error, 0 on success. */
339 local int gz_skip(state, len)
345 /* skip over len bytes or reach end-of-file, whichever comes first */
347 /* skip over whatever is in output buffer */
349 n = GT_OFF(state->have) || (z_off64_t)state->have > len ?
350 (unsigned)len : state->have;
357 /* output buffer empty -- return if we're at the end of the input */
358 else if (state->eof && state->strm.avail_in == 0)
361 /* need more data to skip -- load up output buffer */
363 /* get more output, looking for header if required */
364 if (gz_make(state) == -1)
370 /* -- see zlib.h -- */
371 int ZEXPORT gzread(file, buf, len)
380 /* get internal structure */
383 state = (gz_statep)file;
384 strm = &(state->strm);
386 /* check that we're reading and that there's no error */
387 if (state->mode != GZ_READ || state->err != Z_OK)
390 /* since an int is returned, make sure len fits in one, otherwise return
391 with an error (this avoids the flaw in the interface) */
393 gz_error(state, Z_BUF_ERROR, "requested length does not fit in int");
397 /* if len is zero, avoid unnecessary operations */
401 /* process a skip request */
404 if (gz_skip(state, state->skip) == -1)
408 /* get len bytes to buf, or less than len if at the end */
411 /* first just try copying data from the output buffer */
413 n = state->have > len ? len : state->have;
414 memcpy(buf, state->next, n);
419 /* output buffer empty -- return if we're at the end of the input */
420 else if (state->eof && strm->avail_in == 0)
423 /* need output data -- for small len or new stream load up our output
425 else if (state->how == LOOK || len < (state->size << 1)) {
426 /* get more output, looking for header if required */
427 if (gz_make(state) == -1)
429 continue; /* no progress yet -- go back to memcpy() above */
430 /* the copy above assures that we will leave with space in the
431 output buffer, allowing at least one gzungetc() to succeed */
434 /* large len -- read directly into user buffer */
435 else if (state->how == COPY) { /* read directly */
436 if (gz_load(state, buf, len, &n) == -1)
440 /* large len -- decompress directly into user buffer */
441 else { /* state->how == GZIP */
442 strm->avail_out = len;
443 strm->next_out = buf;
444 if (gz_decomp(state) == -1)
450 /* update progress */
452 buf = (char *)buf + n;
457 /* return number of bytes read into user buffer (will fit in int) */
461 /* -- see zlib.h -- */
462 int ZEXPORT gzgetc(file)
466 unsigned char buf[1];
469 /* get internal structure */
472 state = (gz_statep)file;
474 /* check that we're reading and that there's no error */
475 if (state->mode != GZ_READ || state->err != Z_OK)
478 /* try output buffer (no need to check for skip request) */
482 return *(state->next)++;
485 /* nothing there -- try gzread() */
486 ret = gzread(file, buf, 1);
487 return ret < 1 ? -1 : buf[0];
490 /* -- see zlib.h -- */
491 int ZEXPORT gzungetc(c, file)
497 /* get internal structure */
500 state = (gz_statep)file;
502 /* check that we're reading and that there's no error */
503 if (state->mode != GZ_READ || state->err != Z_OK)
506 /* process a skip request */
509 if (gz_skip(state, state->skip) == -1)
517 /* if output buffer empty, put byte at end (allows more pushing) */
518 if (state->have == 0) {
520 state->next = state->out + (state->size << 1) - 1;
526 /* if no room, give up (must have already done a gzungetc()) */
527 if (state->have == (state->size << 1)) {
528 gz_error(state, Z_BUF_ERROR, "out of room to push characters");
532 /* slide output data if needed and insert byte before existing data */
533 if (state->next == state->out) {
534 unsigned char *src = state->out + state->have;
535 unsigned char *dest = state->out + (state->size << 1);
536 while (src > state->out)
547 /* -- see zlib.h -- */
548 char * ZEXPORT gzgets(file, buf, len)
558 /* check parameters and get internal structure */
559 if (file == NULL || buf == NULL || len < 1)
561 state = (gz_statep)file;
563 /* check that we're reading and that there's no error */
564 if (state->mode != GZ_READ || state->err != Z_OK)
567 /* process a skip request */
570 if (gz_skip(state, state->skip) == -1)
574 /* copy output bytes up to new line or len - 1, whichever comes first --
575 append a terminating zero to the string (we don't check for a zero in
576 the contents, let the user worry about that) */
578 left = (unsigned)len - 1;
580 /* assure that something is in the output buffer */
581 while (state->have == 0 && (state->strm.avail_in || !state->eof)) {
582 if (gz_make(state) == -1)
583 return NULL; /* error */
585 if (state->have == 0) { /* end of file */
586 if (buf == str) /* got bupkus */
588 break; /* got something -- return it */
591 /* look for end-of-line in current output buffer */
592 n = state->have > left ? left : state->have;
593 eol = memchr(state->next, '\n', n);
595 n = (unsigned)(eol - state->next) + 1;
597 /* copy through end-of-line, or remainder if not found */
598 memcpy(buf, state->next, n);
604 } while (left && eol == NULL);
606 /* found end-of-line or out of space -- terminate string and return it */
611 /* -- see zlib.h -- */
612 int ZEXPORT gzdirect(file)
617 /* get internal structure */
620 state = (gz_statep)file;
622 /* check that we're reading */
623 if (state->mode != GZ_READ)
626 /* if the state is not known, but we can find out, then do so (this is
627 mainly for right after a gzopen() or gzdopen()) */
628 if (state->how == LOOK && state->have == 0)
629 (void)gz_head(state);
631 /* return 1 if reading direct, 0 if decompressing a gzip stream */
632 return state->direct;
635 /* -- see zlib.h -- */
636 int ZEXPORT gzclose_r(file)
642 /* get internal structure */
644 return Z_STREAM_ERROR;
645 state = (gz_statep)file;
647 /* check that we're reading */
648 if (state->mode != GZ_READ)
649 return Z_STREAM_ERROR;
651 /* free memory and close file */
653 inflateEnd(&(state->strm));
657 gz_error(state, Z_OK, NULL);
659 ret = close(state->fd);
661 return ret ? Z_ERRNO : Z_OK;