1 /* gzread.c -- zlib functions for reading gzip files
2 * Copyright (C) 2004, 2005, 2010, 2011 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) {
61 memmove(state->in, strm->next_in, strm->avail_in);
62 if (gz_load(state, state->in + strm->avail_in,
63 state->size - strm->avail_in, &got) == -1)
65 strm->avail_in += got;
66 strm->next_in = state->in;
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 local int gz_look(state)
83 z_streamp strm = &(state->strm);
85 /* allocate read buffers and inflate memory */
86 if (state->size == 0) {
87 /* allocate buffers */
88 state->in = malloc(state->want);
89 state->out = malloc(state->want << 1);
90 if (state->in == NULL || state->out == NULL) {
91 if (state->out != NULL)
93 if (state->in != NULL)
95 gz_error(state, Z_MEM_ERROR, "out of memory");
98 state->size = state->want;
100 /* allocate inflate memory */
101 state->strm.zalloc = Z_NULL;
102 state->strm.zfree = Z_NULL;
103 state->strm.opaque = Z_NULL;
104 state->strm.avail_in = 0;
105 state->strm.next_in = Z_NULL;
106 if (inflateInit2(&(state->strm), 15 + 16) != Z_OK) { /* gunzip */
110 gz_error(state, Z_MEM_ERROR, "out of memory");
115 /* get at least the magic bytes in the input buffer */
116 if (strm->avail_in < 2) {
117 if (gz_avail(state) == -1)
119 if (strm->avail_in == 0)
123 /* look for gzip magic bytes -- if there, do gzip decoding (note: there is
124 a logical dilemma here when considering the case of a partially written
125 gzip file, to wit, if a single 31 byte is written, then we cannot tell
126 whether this is a single-byte file, or just a partially written gzip
127 file -- for here we assume that if a gzip file is being written, then
128 the header will be written in a single operation, so that reading a
129 single byte is sufficient indication that it is not a gzip file) */
130 if (strm->avail_in > 1 &&
131 strm->next_in[0] == 31 && strm->next_in[1] == 139) {
138 /* no gzip header -- if we were decoding gzip before, then this is trailing
139 garbage. Ignore the trailing garbage and finish. */
140 if (state->direct == 0) {
147 /* doing raw i/o, copy any leftover input to output -- this assumes that
148 the output buffer is larger than the input buffer, which also assures
149 space for gzungetc() */
150 state->x.next = state->out;
151 if (strm->avail_in) {
152 memcpy(state->x.next, strm->next_in, strm->avail_in);
153 state->x.have = strm->avail_in;
161 /* Decompress from input to the provided next_out and avail_out in the state.
162 On return, state->x.have and state->x.next point to the just decompressed
163 data. If the gzip stream completes, state->how is reset to LOOK to look for
164 the next gzip stream or raw data, once state->x.have is depleted. Returns 0
165 on success, -1 on failure. */
166 local int gz_decomp(state)
171 z_streamp strm = &(state->strm);
173 /* fill output buffer up to end of deflate stream */
174 had = strm->avail_out;
176 /* get more input for inflate() */
177 if (strm->avail_in == 0 && gz_avail(state) == -1)
179 if (strm->avail_in == 0) {
180 gz_error(state, Z_BUF_ERROR, "unexpected end of file");
184 /* decompress and handle errors */
185 ret = inflate(strm, Z_NO_FLUSH);
186 if (ret == Z_STREAM_ERROR || ret == Z_NEED_DICT) {
187 gz_error(state, Z_STREAM_ERROR,
188 "internal error: inflate stream corrupt");
191 if (ret == Z_MEM_ERROR) {
192 gz_error(state, Z_MEM_ERROR, "out of memory");
195 if (ret == Z_DATA_ERROR) { /* deflate stream invalid */
196 gz_error(state, Z_DATA_ERROR,
197 strm->msg == NULL ? "compressed data error" : strm->msg);
200 } while (strm->avail_out && ret != Z_STREAM_END);
202 /* update available output */
203 state->x.have = had - strm->avail_out;
204 state->x.next = strm->next_out - state->x.have;
206 /* if the gzip stream completed successfully, look for another */
207 if (ret == Z_STREAM_END)
210 /* good decompression */
214 /* Fetch data and put it in the output buffer. Assumes state->x.have is 0.
215 Data is either copied from the input file or decompressed from the input
216 file depending on state->how. If state->how is LOOK, then a gzip header is
217 looked for to determine whether to copy or decompress. Returns -1 on error,
218 otherwise 0. gz_fetch() will leave state->how as COPY or GZIP unless the
219 end of the input file has been reached and all data has been processed. */
220 local int gz_fetch(state)
223 z_streamp strm = &(state->strm);
227 case LOOK: /* -> LOOK, COPY (only if never GZIP), or GZIP */
228 if (gz_look(state) == -1)
230 if (state->how == LOOK)
233 case COPY: /* -> COPY */
234 if (gz_load(state, state->out, state->size << 1, &(state->x.have))
237 state->x.next = state->out;
239 case GZIP: /* -> GZIP or LOOK (if end of gzip stream) */
240 strm->avail_out = state->size << 1;
241 strm->next_out = state->out;
242 if (gz_decomp(state) == -1)
245 } while (state->x.have == 0 && (!state->eof || strm->avail_in));
249 /* Skip len uncompressed bytes of output. Return -1 on error, 0 on success. */
250 local int gz_skip(state, len)
256 /* skip over len bytes or reach end-of-file, whichever comes first */
258 /* skip over whatever is in output buffer */
260 n = GT_OFF(state->x.have) || (z_off64_t)state->x.have > len ?
261 (unsigned)len : state->x.have;
268 /* output buffer empty -- return if we're at the end of the input */
269 else if (state->eof && state->strm.avail_in == 0)
272 /* need more data to skip -- load up output buffer */
274 /* get more output, looking for header if required */
275 if (gz_fetch(state) == -1)
281 /* -- see zlib.h -- */
282 int ZEXPORT gzread(file, buf, len)
291 /* get internal structure */
294 state = (gz_statep)file;
295 strm = &(state->strm);
297 /* check that we're reading and that there's no (serious) error */
298 if (state->mode != GZ_READ ||
299 (state->err != Z_OK && state->err != Z_BUF_ERROR))
302 /* since an int is returned, make sure len fits in one, otherwise return
303 with an error (this avoids the flaw in the interface) */
305 gz_error(state, Z_DATA_ERROR, "requested length does not fit in int");
309 /* if len is zero, avoid unnecessary operations */
313 /* process a skip request */
316 if (gz_skip(state, state->skip) == -1)
320 /* get len bytes to buf, or less than len if at the end */
323 /* first just try copying data from the output buffer */
325 n = state->x.have > len ? len : state->x.have;
326 memcpy(buf, state->x.next, n);
331 /* output buffer empty -- return if we're at the end of the input */
332 else if (state->eof && strm->avail_in == 0) {
333 state->past = 1; /* tried to read past end */
337 /* need output data -- for small len or new stream load up our output
339 else if (state->how == LOOK || len < (state->size << 1)) {
340 /* get more output, looking for header if required */
341 if (gz_fetch(state) == -1)
343 continue; /* no progress yet -- go back to memcpy() above */
344 /* the copy above assures that we will leave with space in the
345 output buffer, allowing at least one gzungetc() to succeed */
348 /* large len -- read directly into user buffer */
349 else if (state->how == COPY) { /* read directly */
350 if (gz_load(state, buf, len, &n) == -1)
354 /* large len -- decompress directly into user buffer */
355 else { /* state->how == GZIP */
356 strm->avail_out = len;
357 strm->next_out = buf;
358 if (gz_decomp(state) == -1)
364 /* update progress */
366 buf = (char *)buf + n;
371 /* return number of bytes read into user buffer (will fit in int) */
375 /* -- see zlib.h -- */
376 int ZEXPORT gzgetc_(file)
380 unsigned char buf[1];
383 /* get internal structure */
386 state = (gz_statep)file;
388 /* check that we're reading and that there's no (serious) error */
389 if (state->mode != GZ_READ ||
390 (state->err != Z_OK && state->err != Z_BUF_ERROR))
393 /* try output buffer (no need to check for skip request) */
397 return *(state->x.next)++;
400 /* nothing there -- try gzread() */
401 ret = gzread(file, buf, 1);
402 return ret < 1 ? -1 : buf[0];
406 int ZEXPORT gzgetc(file)
409 return gzgetc_(file);
412 /* -- see zlib.h -- */
413 int ZEXPORT gzungetc(c, file)
419 /* get internal structure */
422 state = (gz_statep)file;
424 /* check that we're reading and that there's no (serious) error */
425 if (state->mode != GZ_READ ||
426 (state->err != Z_OK && state->err != Z_BUF_ERROR))
429 /* process a skip request */
432 if (gz_skip(state, state->skip) == -1)
440 /* if output buffer empty, put byte at end (allows more pushing) */
441 if (state->x.have == 0) {
443 state->x.next = state->out + (state->size << 1) - 1;
444 state->x.next[0] = c;
450 /* if no room, give up (must have already done a gzungetc()) */
451 if (state->x.have == (state->size << 1)) {
452 gz_error(state, Z_DATA_ERROR, "out of room to push characters");
456 /* slide output data if needed and insert byte before existing data */
457 if (state->x.next == state->out) {
458 unsigned char *src = state->out + state->x.have;
459 unsigned char *dest = state->out + (state->size << 1);
460 while (src > state->out)
462 state->x.next = dest;
466 state->x.next[0] = c;
472 /* -- see zlib.h -- */
473 char * ZEXPORT gzgets(file, buf, len)
483 /* check parameters and get internal structure */
484 if (file == NULL || buf == NULL || len < 1)
486 state = (gz_statep)file;
488 /* check that we're reading and that there's no (serious) error */
489 if (state->mode != GZ_READ ||
490 (state->err != Z_OK && state->err != Z_BUF_ERROR))
493 /* process a skip request */
496 if (gz_skip(state, state->skip) == -1)
500 /* copy output bytes up to new line or len - 1, whichever comes first --
501 append a terminating zero to the string (we don't check for a zero in
502 the contents, let the user worry about that) */
504 left = (unsigned)len - 1;
506 /* assure that something is in the output buffer */
507 if (state->x.have == 0 && gz_fetch(state) == -1)
508 return NULL; /* error */
509 if (state->x.have == 0) { /* end of file */
510 state->past = 1; /* read past end */
511 break; /* return what we have */
514 /* look for end-of-line in current output buffer */
515 n = state->x.have > left ? left : state->x.have;
516 eol = memchr(state->x.next, '\n', n);
518 n = (unsigned)(eol - state->x.next) + 1;
520 /* copy through end-of-line, or remainder if not found */
521 memcpy(buf, state->x.next, n);
527 } while (left && eol == NULL);
529 /* return terminated string, or if nothing, end of file */
536 /* -- see zlib.h -- */
537 int ZEXPORT gzdirect(file)
542 /* get internal structure */
545 state = (gz_statep)file;
547 /* if the state is not known, but we can find out, then do so (this is
548 mainly for right after a gzopen() or gzdopen()) */
549 if (state->mode == GZ_READ && state->how == LOOK && state->x.have == 0)
550 (void)gz_look(state);
552 /* return 1 if transparent, 0 if processing a gzip stream */
553 return state->direct;
556 /* -- see zlib.h -- */
557 int ZEXPORT gzclose_r(file)
563 /* get internal structure */
565 return Z_STREAM_ERROR;
566 state = (gz_statep)file;
568 /* check that we're reading */
569 if (state->mode != GZ_READ)
570 return Z_STREAM_ERROR;
572 /* free memory and close file */
574 inflateEnd(&(state->strm));
578 err = state->err == Z_BUF_ERROR ? Z_BUF_ERROR : Z_OK;
579 gz_error(state, Z_OK, NULL);
581 ret = close(state->fd);
583 return ret ? Z_ERRNO : err;