1 /* gzread.c -- zlib functions for reading gzip files
2 * Copyright (C) 2004, 2005, 2010, 2011, 2012 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, *q = strm->next_in;
65 unsigned n = strm->avail_in;
70 if (gz_load(state, state->in + strm->avail_in,
71 state->size - strm->avail_in, &got) == -1)
73 strm->avail_in += got;
74 strm->next_in = state->in;
79 /* Look for gzip header, set up for inflate or copy. state->x.have must be 0.
80 If this is the first time in, allocate required memory. state->how will be
81 left unchanged if there is no more input data available, will be set to COPY
82 if there is no gzip header and direct copying will be performed, or it will
83 be set to GZIP for decompression. If direct copying, then leftover input
84 data from the input buffer will be copied to the output buffer. In that
85 case, all further file reads will be directly to either the output buffer or
86 a user buffer. If decompressing, the inflate state will be initialized.
87 gz_look() will return 0 on success or -1 on failure. */
88 local int gz_look(state)
91 z_streamp strm = &(state->strm);
93 /* allocate read buffers and inflate memory */
94 if (state->size == 0) {
95 /* allocate buffers */
96 state->in = malloc(state->want);
97 state->out = malloc(state->want << 1);
98 if (state->in == NULL || state->out == NULL) {
99 if (state->out != NULL)
101 if (state->in != NULL)
103 gz_error(state, Z_MEM_ERROR, "out of memory");
106 state->size = state->want;
108 /* allocate inflate memory */
109 state->strm.zalloc = Z_NULL;
110 state->strm.zfree = Z_NULL;
111 state->strm.opaque = Z_NULL;
112 state->strm.avail_in = 0;
113 state->strm.next_in = Z_NULL;
114 if (inflateInit2(&(state->strm), 15 + 16) != Z_OK) { /* gunzip */
118 gz_error(state, Z_MEM_ERROR, "out of memory");
123 /* get at least the magic bytes in the input buffer */
124 if (strm->avail_in < 2) {
125 if (gz_avail(state) == -1)
127 if (strm->avail_in == 0)
131 /* look for gzip magic bytes -- if there, do gzip decoding (note: there is
132 a logical dilemma here when considering the case of a partially written
133 gzip file, to wit, if a single 31 byte is written, then we cannot tell
134 whether this is a single-byte file, or just a partially written gzip
135 file -- for here we assume that if a gzip file is being written, then
136 the header will be written in a single operation, so that reading a
137 single byte is sufficient indication that it is not a gzip file) */
138 if (strm->avail_in > 1 &&
139 strm->next_in[0] == 31 && strm->next_in[1] == 139) {
146 /* no gzip header -- if we were decoding gzip before, then this is trailing
147 garbage. Ignore the trailing garbage and finish. */
148 if (state->direct == 0) {
155 /* doing raw i/o, copy any leftover input to output -- this assumes that
156 the output buffer is larger than the input buffer, which also assures
157 space for gzungetc() */
158 state->x.next = state->out;
159 if (strm->avail_in) {
160 memcpy(state->x.next, strm->next_in, strm->avail_in);
161 state->x.have = strm->avail_in;
169 /* Decompress from input to the provided next_out and avail_out in the state.
170 On return, state->x.have and state->x.next point to the just decompressed
171 data. If the gzip stream completes, state->how is reset to LOOK to look for
172 the next gzip stream or raw data, once state->x.have is depleted. Returns 0
173 on success, -1 on failure. */
174 local int gz_decomp(state)
179 z_streamp strm = &(state->strm);
181 /* fill output buffer up to end of deflate stream */
182 had = strm->avail_out;
184 /* get more input for inflate() */
185 if (strm->avail_in == 0 && gz_avail(state) == -1)
187 if (strm->avail_in == 0) {
188 gz_error(state, Z_BUF_ERROR, "unexpected end of file");
192 /* decompress and handle errors */
193 ret = inflate(strm, Z_NO_FLUSH);
194 if (ret == Z_STREAM_ERROR || ret == Z_NEED_DICT) {
195 gz_error(state, Z_STREAM_ERROR,
196 "internal error: inflate stream corrupt");
199 if (ret == Z_MEM_ERROR) {
200 gz_error(state, Z_MEM_ERROR, "out of memory");
203 if (ret == Z_DATA_ERROR) { /* deflate stream invalid */
204 gz_error(state, Z_DATA_ERROR,
205 strm->msg == NULL ? "compressed data error" : strm->msg);
208 } while (strm->avail_out && ret != Z_STREAM_END);
210 /* update available output */
211 state->x.have = had - strm->avail_out;
212 state->x.next = strm->next_out - state->x.have;
214 /* if the gzip stream completed successfully, look for another */
215 if (ret == Z_STREAM_END)
218 /* good decompression */
222 /* Fetch data and put it in the output buffer. Assumes state->x.have is 0.
223 Data is either copied from the input file or decompressed from the input
224 file depending on state->how. If state->how is LOOK, then a gzip header is
225 looked for to determine whether to copy or decompress. Returns -1 on error,
226 otherwise 0. gz_fetch() will leave state->how as COPY or GZIP unless the
227 end of the input file has been reached and all data has been processed. */
228 local int gz_fetch(state)
231 z_streamp strm = &(state->strm);
235 case LOOK: /* -> LOOK, COPY (only if never GZIP), or GZIP */
236 if (gz_look(state) == -1)
238 if (state->how == LOOK)
241 case COPY: /* -> COPY */
242 if (gz_load(state, state->out, state->size << 1, &(state->x.have))
245 state->x.next = state->out;
247 case GZIP: /* -> GZIP or LOOK (if end of gzip stream) */
248 strm->avail_out = state->size << 1;
249 strm->next_out = state->out;
250 if (gz_decomp(state) == -1)
253 } while (state->x.have == 0 && (!state->eof || strm->avail_in));
257 /* Skip len uncompressed bytes of output. Return -1 on error, 0 on success. */
258 local int gz_skip(state, len)
264 /* skip over len bytes or reach end-of-file, whichever comes first */
266 /* skip over whatever is in output buffer */
268 n = GT_OFF(state->x.have) || (z_off64_t)state->x.have > len ?
269 (unsigned)len : state->x.have;
276 /* output buffer empty -- return if we're at the end of the input */
277 else if (state->eof && state->strm.avail_in == 0)
280 /* need more data to skip -- load up output buffer */
282 /* get more output, looking for header if required */
283 if (gz_fetch(state) == -1)
289 /* -- see zlib.h -- */
290 int ZEXPORT gzread(file, buf, len)
299 /* get internal structure */
302 state = (gz_statep)file;
303 strm = &(state->strm);
305 /* check that we're reading and that there's no (serious) error */
306 if (state->mode != GZ_READ ||
307 (state->err != Z_OK && state->err != Z_BUF_ERROR))
310 /* since an int is returned, make sure len fits in one, otherwise return
311 with an error (this avoids the flaw in the interface) */
313 gz_error(state, Z_DATA_ERROR, "requested length does not fit in int");
317 /* if len is zero, avoid unnecessary operations */
321 /* process a skip request */
324 if (gz_skip(state, state->skip) == -1)
328 /* get len bytes to buf, or less than len if at the end */
331 /* first just try copying data from the output buffer */
333 n = state->x.have > len ? len : state->x.have;
334 memcpy(buf, state->x.next, n);
339 /* output buffer empty -- return if we're at the end of the input */
340 else if (state->eof && strm->avail_in == 0) {
341 state->past = 1; /* tried to read past end */
345 /* need output data -- for small len or new stream load up our output
347 else if (state->how == LOOK || len < (state->size << 1)) {
348 /* get more output, looking for header if required */
349 if (gz_fetch(state) == -1)
351 continue; /* no progress yet -- go back to copy above */
352 /* the copy above assures that we will leave with space in the
353 output buffer, allowing at least one gzungetc() to succeed */
356 /* large len -- read directly into user buffer */
357 else if (state->how == COPY) { /* read directly */
358 if (gz_load(state, buf, len, &n) == -1)
362 /* large len -- decompress directly into user buffer */
363 else { /* state->how == GZIP */
364 strm->avail_out = len;
365 strm->next_out = buf;
366 if (gz_decomp(state) == -1)
372 /* update progress */
374 buf = (char *)buf + n;
379 /* return number of bytes read into user buffer (will fit in int) */
383 /* -- see zlib.h -- */
385 int ZEXPORT gzgetc(file)
389 unsigned char buf[1];
392 /* get internal structure */
395 state = (gz_statep)file;
397 /* check that we're reading and that there's no (serious) error */
398 if (state->mode != GZ_READ ||
399 (state->err != Z_OK && state->err != Z_BUF_ERROR))
402 /* try output buffer (no need to check for skip request) */
406 return *(state->x.next)++;
409 /* nothing there -- try gzread() */
410 ret = gzread(file, buf, 1);
411 return ret < 1 ? -1 : buf[0];
414 int ZEXPORT gzgetc_(file)
420 /* -- see zlib.h -- */
421 int ZEXPORT gzungetc(c, file)
427 /* get internal structure */
430 state = (gz_statep)file;
432 /* check that we're reading and that there's no (serious) error */
433 if (state->mode != GZ_READ ||
434 (state->err != Z_OK && state->err != Z_BUF_ERROR))
437 /* process a skip request */
440 if (gz_skip(state, state->skip) == -1)
448 /* if output buffer empty, put byte at end (allows more pushing) */
449 if (state->x.have == 0) {
451 state->x.next = state->out + (state->size << 1) - 1;
452 state->x.next[0] = c;
458 /* if no room, give up (must have already done a gzungetc()) */
459 if (state->x.have == (state->size << 1)) {
460 gz_error(state, Z_DATA_ERROR, "out of room to push characters");
464 /* slide output data if needed and insert byte before existing data */
465 if (state->x.next == state->out) {
466 unsigned char *src = state->out + state->x.have;
467 unsigned char *dest = state->out + (state->size << 1);
468 while (src > state->out)
470 state->x.next = dest;
474 state->x.next[0] = c;
480 /* -- see zlib.h -- */
481 char * ZEXPORT gzgets(file, buf, len)
491 /* check parameters and get internal structure */
492 if (file == NULL || buf == NULL || len < 1)
494 state = (gz_statep)file;
496 /* check that we're reading and that there's no (serious) error */
497 if (state->mode != GZ_READ ||
498 (state->err != Z_OK && state->err != Z_BUF_ERROR))
501 /* process a skip request */
504 if (gz_skip(state, state->skip) == -1)
508 /* copy output bytes up to new line or len - 1, whichever comes first --
509 append a terminating zero to the string (we don't check for a zero in
510 the contents, let the user worry about that) */
512 left = (unsigned)len - 1;
514 /* assure that something is in the output buffer */
515 if (state->x.have == 0 && gz_fetch(state) == -1)
516 return NULL; /* error */
517 if (state->x.have == 0) { /* end of file */
518 state->past = 1; /* read past end */
519 break; /* return what we have */
522 /* look for end-of-line in current output buffer */
523 n = state->x.have > left ? left : state->x.have;
524 eol = memchr(state->x.next, '\n', n);
526 n = (unsigned)(eol - state->x.next) + 1;
528 /* copy through end-of-line, or remainder if not found */
529 memcpy(buf, state->x.next, n);
535 } while (left && eol == NULL);
537 /* return terminated string, or if nothing, end of file */
544 /* -- see zlib.h -- */
545 int ZEXPORT gzdirect(file)
550 /* get internal structure */
553 state = (gz_statep)file;
555 /* if the state is not known, but we can find out, then do so (this is
556 mainly for right after a gzopen() or gzdopen()) */
557 if (state->mode == GZ_READ && state->how == LOOK && state->x.have == 0)
558 (void)gz_look(state);
560 /* return 1 if transparent, 0 if processing a gzip stream */
561 return state->direct;
564 /* -- see zlib.h -- */
565 int ZEXPORT gzclose_r(file)
571 /* get internal structure */
573 return Z_STREAM_ERROR;
574 state = (gz_statep)file;
576 /* check that we're reading */
577 if (state->mode != GZ_READ)
578 return Z_STREAM_ERROR;
580 /* free memory and close file */
582 inflateEnd(&(state->strm));
586 err = state->err == Z_BUF_ERROR ? Z_BUF_ERROR : Z_OK;
587 gz_error(state, Z_OK, NULL);
589 ret = close(state->fd);
591 return ret ? Z_ERRNO : err;