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_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->have must be zero.
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, save start of raw data for seeking, copy any leftover
148 input to output -- this assumes that the output buffer is larger than
149 the input buffer, which also assures space for gzungetc() */
150 state->raw = state->pos;
151 state->next = state->out;
152 if (strm->avail_in) {
153 memcpy(state->next, strm->next_in, strm->avail_in);
154 state->have = strm->avail_in;
162 /* Decompress from input to the provided next_out and avail_out in the state.
163 state->have and state->next are set to point to the just decompressed data,
164 If the gzip stream completes, state->how is reset to LOOK to look for the
165 next gzip stream or raw data, once state->have is depleted. Returns 0 on
166 success, -1 on failure. */
167 local int gz_decomp(state)
172 z_streamp strm = &(state->strm);
174 /* fill output buffer up to end of deflate stream */
175 had = strm->avail_out;
177 /* get more input for inflate() */
178 if (strm->avail_in == 0 && gz_avail(state) == -1)
180 if (strm->avail_in == 0) {
181 gz_error(state, Z_BUF_ERROR, "unexpected end of file");
185 /* decompress and handle errors */
186 ret = inflate(strm, Z_NO_FLUSH);
187 if (ret == Z_STREAM_ERROR || ret == Z_NEED_DICT) {
188 gz_error(state, Z_STREAM_ERROR,
189 "internal error: inflate stream corrupt");
192 if (ret == Z_MEM_ERROR) {
193 gz_error(state, Z_MEM_ERROR, "out of memory");
196 if (ret == Z_DATA_ERROR) { /* deflate stream invalid */
197 gz_error(state, Z_DATA_ERROR,
198 strm->msg == NULL ? "compressed data error" : strm->msg);
201 } while (strm->avail_out && ret != Z_STREAM_END);
203 /* update available output */
204 state->have = had - strm->avail_out;
205 state->next = strm->next_out - state->have;
207 /* if the gzip stream completed successfully, look for another */
208 if (ret == Z_STREAM_END)
211 /* good decompression */
215 /* Fetch data and put it in the output buffer. Assumes that state->have == 0.
216 Data is either copied from the input file or decompressed from the input
217 file depending on state->how. If state->how is LOOK, then a gzip header is
218 looked for to determine whether to copy or decompress. Returns -1 on error,
219 otherwise 0. gz_fetch() will leave state->how as COPY or GZIP unless the
220 end of the input file has been reached and all data has been processed. */
221 local int gz_fetch(state)
224 z_streamp strm = &(state->strm);
228 case LOOK: /* -> LOOK, COPY (only if never GZIP), or GZIP */
229 if (gz_look(state) == -1)
231 if (state->how == LOOK)
234 case COPY: /* -> COPY */
235 if (gz_load(state, state->out, state->size << 1, &(state->have))
238 state->next = state->out;
240 case GZIP: /* -> GZIP or LOOK (if end of gzip stream) */
241 strm->avail_out = state->size << 1;
242 strm->next_out = state->out;
243 if (gz_decomp(state) == -1)
246 } while (state->have == 0);
250 /* Skip len uncompressed bytes of output. Return -1 on error, 0 on success. */
251 local int gz_skip(state, len)
257 /* skip over len bytes or reach end-of-file, whichever comes first */
259 /* skip over whatever is in output buffer */
261 n = GT_OFF(state->have) || (z_off64_t)state->have > len ?
262 (unsigned)len : state->have;
269 /* output buffer empty -- return if we're at the end of the input */
270 else if (state->eof && state->strm.avail_in == 0)
273 /* need more data to skip -- load up output buffer */
275 /* get more output, looking for header if required */
276 if (gz_fetch(state) == -1)
282 /* -- see zlib.h -- */
283 int ZEXPORT gzread(file, buf, len)
292 /* get internal structure */
295 state = (gz_statep)file;
296 strm = &(state->strm);
298 /* check that we're reading and that there's no (serious) error */
299 if (state->mode != GZ_READ ||
300 (state->err != Z_OK && state->err != Z_BUF_ERROR))
303 /* since an int is returned, make sure len fits in one, otherwise return
304 with an error (this avoids the flaw in the interface) */
306 gz_error(state, Z_BUF_ERROR, "requested length does not fit in int");
310 /* if len is zero, avoid unnecessary operations */
314 /* process a skip request */
317 if (gz_skip(state, state->skip) == -1)
321 /* get len bytes to buf, or less than len if at the end */
324 /* first just try copying data from the output buffer */
326 n = state->have > len ? len : state->have;
327 memcpy(buf, state->next, n);
332 /* output buffer empty -- return if we're at the end of the input */
333 else if (state->eof && strm->avail_in == 0)
336 /* need output data -- for small len or new stream load up our output
338 else if (state->how == LOOK || len < (state->size << 1)) {
339 /* get more output, looking for header if required */
340 if (gz_fetch(state) == -1)
342 continue; /* no progress yet -- go back to memcpy() above */
343 /* the copy above assures that we will leave with space in the
344 output buffer, allowing at least one gzungetc() to succeed */
347 /* large len -- read directly into user buffer */
348 else if (state->how == COPY) { /* read directly */
349 if (gz_load(state, buf, len, &n) == -1)
353 /* large len -- decompress directly into user buffer */
354 else { /* state->how == GZIP */
355 strm->avail_out = len;
356 strm->next_out = buf;
357 if (gz_decomp(state) == -1)
363 /* update progress */
365 buf = (char *)buf + n;
370 /* return number of bytes read into user buffer (will fit in int) */
374 /* -- see zlib.h -- */
375 int ZEXPORT gzgetc(file)
379 unsigned char buf[1];
382 /* get internal structure */
385 state = (gz_statep)file;
387 /* check that we're reading and that there's no (serious) error */
388 if (state->mode != GZ_READ ||
389 (state->err != Z_OK && state->err != Z_BUF_ERROR))
392 /* try output buffer (no need to check for skip request) */
396 return *(state->next)++;
399 /* nothing there -- try gzread() */
400 ret = gzread(file, buf, 1);
401 return ret < 1 ? -1 : buf[0];
404 /* -- see zlib.h -- */
405 int ZEXPORT gzungetc(c, file)
411 /* get internal structure */
414 state = (gz_statep)file;
416 /* check that we're reading and that there's no (serious) error */
417 if (state->mode != GZ_READ ||
418 (state->err != Z_OK && state->err != Z_BUF_ERROR))
421 /* process a skip request */
424 if (gz_skip(state, state->skip) == -1)
432 /* if output buffer empty, put byte at end (allows more pushing) */
433 if (state->have == 0) {
435 state->next = state->out + (state->size << 1) - 1;
441 /* if no room, give up (must have already done a gzungetc()) */
442 if (state->have == (state->size << 1)) {
443 gz_error(state, Z_BUF_ERROR, "out of room to push characters");
447 /* slide output data if needed and insert byte before existing data */
448 if (state->next == state->out) {
449 unsigned char *src = state->out + state->have;
450 unsigned char *dest = state->out + (state->size << 1);
451 while (src > state->out)
462 /* -- see zlib.h -- */
463 char * ZEXPORT gzgets(file, buf, len)
473 /* check parameters and get internal structure */
474 if (file == NULL || buf == NULL || len < 1)
476 state = (gz_statep)file;
478 /* check that we're reading and that there's no (serious) error */
479 if (state->mode != GZ_READ ||
480 (state->err != Z_OK && state->err != Z_BUF_ERROR))
483 /* process a skip request */
486 if (gz_skip(state, state->skip) == -1)
490 /* copy output bytes up to new line or len - 1, whichever comes first --
491 append a terminating zero to the string (we don't check for a zero in
492 the contents, let the user worry about that) */
494 left = (unsigned)len - 1;
496 /* assure that something is in the output buffer */
497 if (state->have == 0 && gz_fetch(state) == -1)
498 return NULL; /* error */
499 if (state->have == 0) { /* end of file */
500 if (buf == str) /* got bupkus */
502 break; /* got something -- return it */
505 /* look for end-of-line in current output buffer */
506 n = state->have > left ? left : state->have;
507 eol = memchr(state->next, '\n', n);
509 n = (unsigned)(eol - state->next) + 1;
511 /* copy through end-of-line, or remainder if not found */
512 memcpy(buf, state->next, n);
518 } while (left && eol == NULL);
520 /* found end-of-line or out of space -- terminate string and return it */
525 /* -- see zlib.h -- */
526 int ZEXPORT gzdirect(file)
531 /* get internal structure */
534 state = (gz_statep)file;
536 /* check that we're reading */
537 if (state->mode != GZ_READ)
540 /* if the state is not known, but we can find out, then do so (this is
541 mainly for right after a gzopen() or gzdopen()) */
542 if (state->how == LOOK && state->have == 0)
543 (void)gz_look(state);
545 /* return 1 if reading direct, 0 if decompressing a gzip stream */
546 return state->direct;
549 /* -- see zlib.h -- */
550 int ZEXPORT gzclose_r(file)
556 /* get internal structure */
558 return Z_STREAM_ERROR;
559 state = (gz_statep)file;
561 /* check that we're reading */
562 if (state->mode != GZ_READ)
563 return Z_STREAM_ERROR;
565 /* free memory and close file */
567 inflateEnd(&(state->strm));
571 gz_error(state, Z_OK, NULL);
573 ret = close(state->fd);
575 return ret ? Z_ERRNO : Z_OK;