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
11 local int gz_load OF((gz_statep, unsigned char *, unsigned, unsigned *));
12 local int gz_avail OF((gz_statep));
13 local int gz_next4 OF((gz_statep, unsigned long *));
14 local int gz_head OF((gz_statep));
15 local int gz_decomp OF((gz_statep));
16 local int gz_make OF((gz_statep));
17 local int gz_skip OF((gz_statep, z_off_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 gz_avail() assumes that strm->avail_in == 0. */
52 local int gz_avail(state)
55 z_streamp strm = &(state->strm);
57 if (state->err != Z_OK)
59 if (state->eof == 0) {
60 if (gz_load(state, state->in, state->size, &(strm->avail_in)) == -1)
62 strm->next_in = state->in;
67 /* Get next byte from input, or -1 if end or error. */
68 #define NEXT() ((strm->avail_in == 0 && gz_avail(state) == -1) ? -1 : \
69 (strm->avail_in == 0 ? -1 : \
70 (strm->avail_in--, *(strm->next_in)++)))
72 /* Get a four-byte little-endian integer and return 0 on success and the value
73 in *ret. Otherwise -1 is returned and *ret is not modified. */
74 local int gz_next4(state, ret)
80 z_streamp strm = &(state->strm);
83 val += (unsigned)NEXT() << 8;
84 val += (unsigned long)NEXT() << 16;
88 val += (unsigned long)ch << 24;
93 /* Look for gzip header, set up for inflate or copy. state->have must be zero.
94 If this is the first time in, allocate required memory. state->how will be
95 left unchanged if there is no more input data available, will be set to 1 if
96 there is no gzip header and direct copying will be performed, or it will be
97 set to 2 for decompression, and the gzip header will be skipped so that the
98 next available input data is the raw deflate stream. If direct copying,
99 then leftover input data from the input buffer will be copied to the output
100 buffer. In that case, all further file reads will be directly to either the
101 output buffer or a user buffer. If decompressing, the inflate state and the
102 check value will be initialized. gz_head() will return 0 on success or -1
103 on failure. Failures may include read errors or gzip header errors. */
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);
203 /* not a gzip file -- save first byte (31) and fall to raw i/o */
209 /* doing raw i/o, save start of raw data for seeking, copy any leftover
210 input to output -- this assumes that the output buffer is larger than
211 the input buffer, which also assures space for gzungetc() */
212 state->raw = state->pos;
213 state->next = state->out;
214 if (strm->avail_in) {
215 memcpy(state->next + state->have, strm->next_in, strm->avail_in);
216 state->have += strm->avail_in;
223 /* Decompress from input to the provided next_out and avail_out in the state.
224 If the end of the compressed data is reached, then verify the gzip trailer
225 check value and length (modulo 2^32). state->have and state->next are set
226 to point to the just decompressed data, and the crc is updated. If the
227 trailer is verified, state->how is reset to zero to look for the next gzip
228 stream or raw data, once state->have is depleted. Returns 0 on success, -1
229 on failure. Failures may include invalid compressed data or a failed gzip
230 trailer verification. */
231 local int gz_decomp(state)
236 unsigned long crc, len;
237 z_streamp strm = &(state->strm);
239 /* fill output buffer up to end of deflate stream */
240 had = strm->avail_out;
242 /* get more input for inflate() */
243 if (strm->avail_in == 0 && gz_avail(state) == -1)
245 if (strm->avail_in == 0) {
246 gz_error(state, Z_DATA_ERROR, "unexpected end of file");
250 /* decompress and handle errors */
251 ret = inflate(strm, Z_NO_FLUSH);
252 if (ret == Z_STREAM_ERROR || ret == Z_NEED_DICT) {
253 gz_error(state, Z_STREAM_ERROR,
254 "internal error: inflate stream corrupt");
257 if (ret == Z_MEM_ERROR) {
258 gz_error(state, Z_MEM_ERROR, "out of memory");
261 if (ret == Z_DATA_ERROR) { /* deflate stream invalid */
262 gz_error(state, Z_DATA_ERROR,
263 strm->msg == NULL ? "compressed data error" : strm->msg);
266 } while (strm->avail_out && ret != Z_STREAM_END);
268 /* update available output and crc check value */
269 state->have = had - strm->avail_out;
270 state->next = strm->next_out - state->have;
271 strm->adler = crc32(strm->adler, state->next, state->have);
273 /* check gzip trailer if at end of deflate stream */
274 if (ret == Z_STREAM_END) {
275 if (gz_next4(state, &crc) == -1 || gz_next4(state, &len) == -1) {
276 gz_error(state, Z_DATA_ERROR, "unexpected end of file");
279 if (crc != strm->adler) {
280 gz_error(state, Z_DATA_ERROR, "incorrect data check");
283 if (len != (strm->total_out & 0xffffffffL)) {
284 gz_error(state, Z_DATA_ERROR, "incorrect length check");
287 state->how = 0; /* ready for next stream, once have is 0 */
290 /* good decompression */
294 /* Make data and put in the output buffer. Assumes that state->have == 0.
295 Data is either copied from the input file or decompressed from the input
296 file depending on state->how. If state->how is zero, then a gzip header is
297 looked for (and skipped if found) to determine wither to copy or decompress.
298 Returns -1 on error, otherwise 0. gz_make() will leave state->have non-zero
299 unless the end of the input file has been reached and all data has been
301 local int gz_make(state)
304 z_streamp strm = &(state->strm);
306 if (state->how == 0) { /* look for gzip header */
307 if (gz_head(state) == -1)
309 if (state->have) /* got some data from gz_head() */
312 if (state->how == 1) { /* straight copy */
313 if (gz_load(state, state->out, state->size << 1, &(state->have)) == -1)
315 state->next = state->out;
317 else if (state->how == 2) { /* decompress */
318 strm->avail_out = state->size << 1;
319 strm->next_out = state->out;
320 if (gz_decomp(state) == -1)
326 /* Skip len uncompressed bytes of output. Return -1 on error, 0 on success. */
327 local int gz_skip(state, len)
333 /* skip over len bytes or reach end-of-file, whichever comes first */
335 /* skip over whatever is in output buffer */
337 n = state->have > len ? (unsigned)len : state->have;
344 /* output buffer empty -- return if we're at the end of the input */
345 else if (state->eof && state->strm.avail_in == 0)
348 /* need more data to skip -- load up output buffer */
350 /* get more output, looking for header if required */
351 if (gz_make(state) == -1)
357 /* -- see zlib.h -- */
358 int ZEXPORT gzread(file, buf, len)
367 /* get internal structure */
370 state = (gz_statep)file;
371 strm = &(state->strm);
373 /* check that we're reading and that there's no error */
374 if (state->mode != GZ_READ || state->err != Z_OK)
377 /* since an int is returned, make sure len fits in one, otherwise return
378 with an error (this avoids the flaw in the interface) */
380 gz_error(state, Z_BUF_ERROR, "requested length does not fit in int");
384 /* if len is zero, avoid unnecessary operations */
388 /* process a skip request */
391 if (gz_skip(state, state->skip) == -1)
395 /* get len bytes to buf, or less than len if at the end */
398 /* first just try copying data from the output buffer */
400 n = state->have > len ? len : state->have;
401 memcpy(buf, state->next, n);
406 /* output buffer empty -- return if we're at the end of the input */
407 else if (state->eof && strm->avail_in == 0)
410 /* need output data -- for small len or new stream load up our output
412 else if (state->how == 0 || len < (state->size << 1)) {
413 /* get more output, looking for header if required */
414 if (gz_make(state) == -1)
416 continue; /* no progress yet -- go back to memcpy() above */
417 /* the copy above assures that we will leave with space in the
418 output buffer, allowing at least one gzungetc() to succeed */
421 /* large len -- read directly into user buffer */
422 else if (state->how == 1) { /* read directly */
423 if (gz_load(state, buf, len, &n) == -1)
427 /* large len -- decompress directly into user buffer */
428 else { /* state->how == 2 */
429 strm->avail_out = len;
430 strm->next_out = buf;
431 if (gz_decomp(state) == -1)
437 /* update progress */
439 buf = (char *)buf + n;
444 /* return number of bytes read into user buffer (will fit in int) */
448 /* -- see zlib.h -- */
449 int ZEXPORT gzgetc(file)
453 unsigned char buf[1];
456 /* get internal structure */
459 state = (gz_statep)file;
461 /* check that we're reading and that there's no error */
462 if (state->mode != GZ_READ || state->err != Z_OK)
465 /* try output buffer (no need to check for skip request) */
469 return *(state->next)++;
472 /* nothing there -- try gzread() */
473 ret = gzread(file, buf, 1);
474 return ret < 1 ? -1 : buf[0];
477 /* -- see zlib.h -- */
478 int ZEXPORT gzungetc(c, file)
484 /* get internal structure */
487 state = (gz_statep)file;
489 /* check that we're reading and that there's no error */
490 if (state->mode != GZ_READ || state->err != Z_OK)
493 /* process a skip request */
496 if (gz_skip(state, state->skip) == -1)
504 /* if output buffer empty, put byte at end (allows more pushing) */
505 if (state->have == 0) {
507 state->next = state->out + (state->size << 1) - 1;
513 /* if no room, give up (must have already done a gzungetc()) */
514 if (state->have == (state->size << 1)) {
515 gz_error(state, Z_BUF_ERROR, "out of room to push characters");
519 /* slide output data if needed and insert byte before existing data */
520 if (state->next == state->out) {
521 unsigned char *src = state->out + state->have;
522 unsigned char *dest = state->out + (state->size << 1);
523 while (src > state->out)
534 /* -- see zlib.h -- */
535 char * ZEXPORT gzgets(file, buf, len)
545 /* get internal structure */
548 state = (gz_statep)file;
550 /* check that we're reading and that there's no error */
551 if (state->mode != GZ_READ || state->err != Z_OK)
554 /* process a skip request */
557 if (gz_skip(state, state->skip) == -1)
561 /* check for a dumb length */
565 /* copy output bytes up to new line or len - 1, whichever comes first --
566 append a terminating zero to the string (we don't check for a zero in
567 the contents, let the user worry about that) */
569 left = (unsigned)len - 1;
571 /* assure that something is in the output buffer */
572 if (state->have == 0) {
573 if (gz_make(state) == -1)
574 return NULL; /* error */
575 if (state->have == 0) { /* end of file */
576 if (buf == str) /* got bupkus */
578 break; /* got something -- return it */
582 /* look for end-of-line in current output buffer */
583 n = state->have > left ? left : state->have;
584 eol = memchr(state->next, '\n', n);
586 n = (eol - state->next) + 1;
588 /* copy through end-of-line, or remainder if not found */
589 memcpy(buf, state->next, n);
595 } while (left && eol == NULL);
597 /* found end-of-line or out of space -- terminate string and return it */
602 /* -- see zlib.h -- */
603 int ZEXPORT gzdirect(file)
608 /* get internal structure */
611 state = (gz_statep)file;
613 /* check that we're reading */
614 if (state->mode != GZ_READ)
617 /* return true if reading without decompression */
618 return state->how == 1;
621 /* -- see zlib.h -- */
622 int ZEXPORT gzclose_r(file)
627 /* get internal structure */
629 return Z_STREAM_ERROR;
630 state = (gz_statep)file;
632 /* check that we're reading */
633 if (state->mode != GZ_READ)
634 return Z_STREAM_ERROR;
636 /* free memory and close file */
638 inflateEnd(&(state->strm));
642 gz_error(state, Z_OK, NULL);
648 #endif /* !OLD_GZIO */