2 * xzlib.c: front end for the transparent suport of lzma compression
3 * at the I/O layer, based on an example file from lzma project
5 * See Copyright for the status of this software.
7 * Anders F Bjorklund <afb@users.sourceforge.net>
19 #ifdef HAVE_SYS_TYPES_H
20 #include <sys/types.h>
22 #ifdef HAVE_SYS_STAT_H
40 #include <libxml/xmlmemory.h>
42 /* values for xz_state how */
43 #define LOOK 0 /* look for a gzip/lzma header */
44 #define COPY 1 /* copy input directly */
45 #define GZIP 2 /* decompress a gzip stream */
46 #define LZMA 3 /* decompress a lzma stream */
48 /* internal lzma file state data structure */
50 int mode; /* see lzma modes above */
51 int fd; /* file descriptor */
52 char *path; /* path or fd for error messages */
53 uint64_t pos; /* current position in uncompressed data */
54 unsigned int size; /* buffer size, zero if not allocated yet */
55 unsigned int want; /* requested buffer size, default is BUFSIZ */
56 unsigned char *in; /* input buffer */
57 unsigned char *out; /* output buffer (double-sized when reading) */
58 unsigned char *next; /* next output data to deliver or write */
59 unsigned int have; /* amount of output data unused at next */
60 int eof; /* true if end of input file reached */
61 uint64_t start; /* where the lzma data started, for rewinding */
62 uint64_t raw; /* where the raw data started, for seeking */
63 int how; /* 0: get header, 1: copy, 2: decompress */
64 int direct; /* true if last read direct, false if lzma */
66 uint64_t skip; /* amount to skip (already rewound if backwards) */
67 int seek; /* true if seek request pending */
68 /* error information */
69 int err; /* error code */
70 char *msg; /* error message */
72 int init; /* is the iniflate stream initialized */
73 lzma_stream strm; /* stream structure in-place (not a pointer) */
74 char padding1[32]; /* padding allowing to cope with possible
75 extensions of above structure without
76 too much side effect */
78 /* zlib inflate or deflate stream */
79 z_stream zstrm; /* stream structure in-place (not a pointer) */
81 char padding2[32]; /* padding allowing to cope with possible
82 extensions of above structure without
83 too much side effect */
84 } xz_state, *xz_statep;
87 xz_error(xz_statep state, int err, const char *msg)
89 /* free previously allocated message and clear */
90 if (state->msg != NULL) {
91 if (state->err != LZMA_MEM_ERROR)
96 /* set error code, and if no message, then done */
101 /* for an out of memory error, save as static string */
102 if (err == LZMA_MEM_ERROR) {
103 state->msg = (char *) msg;
107 /* construct error message with path */
109 xmlMalloc(strlen(state->path) + strlen(msg) + 3)) == NULL) {
110 state->err = LZMA_MEM_ERROR;
111 state->msg = (char *) "out of memory";
114 strcpy(state->msg, state->path);
115 strcat(state->msg, ": ");
116 strcat(state->msg, msg);
121 xz_reset(xz_statep state)
123 state->have = 0; /* no output data available */
124 state->eof = 0; /* not at end of file */
125 state->how = LOOK; /* look for gzip header */
126 state->direct = 1; /* default for empty file */
127 state->seek = 0; /* no seek request pending */
128 xz_error(state, LZMA_OK, NULL); /* clear error */
129 state->pos = 0; /* no uncompressed data yet */
130 state->strm.avail_in = 0; /* no input data yet */
132 state->zstrm.avail_in = 0; /* no input data yet */
137 xz_open(const char *path, int fd, const char *mode ATTRIBUTE_UNUSED)
141 /* allocate xzFile structure to return */
142 state = xmlMalloc(sizeof(xz_state));
145 state->size = 0; /* no buffers allocated yet */
146 state->want = BUFSIZ; /* requested buffer size */
147 state->msg = NULL; /* no error message yet */
148 state->init = 0; /* initialization of zlib data */
150 /* save the path name for error messages */
151 state->path = xmlMalloc(strlen(path) + 1);
152 if (state->path == NULL) {
156 strcpy(state->path, path);
158 /* open the file with the appropriate mode (or just use fd) */
159 state->fd = fd != -1 ? fd : open(path,
167 if (state->fd == -1) {
168 xmlFree(state->path);
173 /* save the current position for rewinding (only if reading) */
174 state->start = lseek(state->fd, 0, SEEK_CUR);
175 if (state->start == (uint64_t) - 1)
178 /* initialize stream */
182 return (xzFile) state;
186 xz_compressed(xzFile f) {
191 state = (xz_statep) f;
192 if (state->init <= 0)
195 switch (state->how) {
206 __libxml2_xzopen(const char *path, const char *mode)
208 return xz_open(path, -1, mode);
212 __libxml2_xzcompressed(xzFile f) {
213 return xz_compressed(f);
217 __libxml2_xzdopen(int fd, const char *mode)
219 char *path; /* identifier for error messages */
222 if (fd == -1 || (path = xmlMalloc(7 + 3 * sizeof(int))) == NULL)
224 sprintf(path, "<fd:%d>", fd); /* for debugging */
225 xz = xz_open(path, fd, mode);
231 xz_load(xz_statep state, unsigned char *buf, unsigned int len,
238 ret = read(state->fd, buf + *have, len - *have);
242 } while (*have < len);
244 xz_error(state, -1, strerror(errno));
253 xz_avail(xz_statep state)
255 lzma_stream *strm = &(state->strm);
257 if (state->err != LZMA_OK)
259 if (state->eof == 0) {
260 /* avail_in is size_t, which is not necessary sizeof(unsigned) */
261 unsigned tmp = strm->avail_in;
263 if (xz_load(state, state->in, state->size, &tmp) == -1) {
264 strm->avail_in = tmp;
267 strm->avail_in = tmp;
268 strm->next_in = state->in;
275 xz_avail_zstrm(xz_statep state)
278 state->strm.avail_in = state->zstrm.avail_in;
279 state->strm.next_in = state->zstrm.next_in;
280 ret = xz_avail(state);
281 state->zstrm.avail_in = (uInt) state->strm.avail_in;
282 state->zstrm.next_in = (Bytef *) state->strm.next_in;
288 is_format_xz(xz_statep state)
290 lzma_stream *strm = &(state->strm);
292 return strm->avail_in >= 6 && memcmp(state->in, "\3757zXZ", 6) == 0;
296 is_format_lzma(xz_statep state)
298 lzma_stream *strm = &(state->strm);
301 lzma_options_lzma *opt;
303 uint64_t uncompressed_size;
306 if (strm->avail_in < 13)
309 filter.id = LZMA_FILTER_LZMA1;
310 if (lzma_properties_decode(&filter, NULL, state->in, 5) != LZMA_OK)
313 opt = filter.options;
314 dict_size = opt->dict_size;
315 free(opt); /* we can't use xmlFree on a string returned by zlib */
317 /* A hack to ditch tons of false positives: We allow only dictionary
318 * sizes that are 2^n or 2^n + 2^(n-1) or UINT32_MAX. LZMA_Alone
319 * created only files with 2^n, but accepts any dictionary size.
320 * If someone complains, this will be reconsidered.
322 if (dict_size != UINT32_MAX) {
323 uint32_t d = dict_size - 1;
331 if (d != dict_size || dict_size == 0)
335 /* Another hack to ditch false positives: Assume that if the
336 * uncompressed size is known, it must be less than 256 GiB.
337 * Again, if someone complains, this will be reconsidered.
339 uncompressed_size = 0;
340 for (i = 0; i < 8; ++i)
341 uncompressed_size |= (uint64_t) (state->in[5 + i]) << (i * 8);
343 if (uncompressed_size != UINT64_MAX
344 && uncompressed_size > (UINT64_C(1) << 38))
352 /* Get next byte from input, or -1 if end or error. */
353 #define NEXT() ((strm->avail_in == 0 && xz_avail(state) == -1) ? -1 : \
354 (strm->avail_in == 0 ? -1 : \
355 (strm->avail_in--, *(strm->next_in)++)))
356 /* Same thing, but from zstrm */
357 #define NEXTZ() ((strm->avail_in == 0 && xz_avail_zstrm(state) == -1) ? -1 : \
358 (strm->avail_in == 0 ? -1 : \
359 (strm->avail_in--, *(strm->next_in)++)))
361 /* Get a four-byte little-endian integer and return 0 on success and the value
362 in *ret. Otherwise -1 is returned and *ret is not modified. */
364 gz_next4(xz_statep state, unsigned long *ret)
368 z_streamp strm = &(state->zstrm);
371 val += (unsigned) NEXTZ() << 8;
372 val += (unsigned long) NEXTZ() << 16;
376 val += (unsigned long) ch << 24;
383 xz_head(xz_statep state)
385 lzma_stream *strm = &(state->strm);
386 lzma_stream init = LZMA_STREAM_INIT;
390 /* allocate read buffers and inflate memory */
391 if (state->size == 0) {
392 /* allocate buffers */
393 state->in = xmlMalloc(state->want);
394 state->out = xmlMalloc(state->want << 1);
395 if (state->in == NULL || state->out == NULL) {
396 if (state->out != NULL)
398 if (state->in != NULL)
400 xz_error(state, LZMA_MEM_ERROR, "out of memory");
403 state->size = state->want;
405 /* allocate decoder memory */
407 state->strm.avail_in = 0;
408 state->strm.next_in = NULL;
409 if (lzma_auto_decoder(&state->strm, UINT64_MAX, 0) != LZMA_OK) {
413 xz_error(state, LZMA_MEM_ERROR, "out of memory");
417 /* allocate inflate memory */
418 state->zstrm.zalloc = Z_NULL;
419 state->zstrm.zfree = Z_NULL;
420 state->zstrm.opaque = Z_NULL;
421 state->zstrm.avail_in = 0;
422 state->zstrm.next_in = Z_NULL;
423 if (state->init == 0) {
424 if (inflateInit2(&(state->zstrm), -15) != Z_OK) {/* raw inflate */
428 xz_error(state, LZMA_MEM_ERROR, "out of memory");
436 /* get some data in the input buffer */
437 if (strm->avail_in == 0) {
438 if (xz_avail(state) == -1)
440 if (strm->avail_in == 0)
444 /* look for the xz magic header bytes */
445 if (is_format_xz(state) || is_format_lzma(state)) {
451 /* look for the gzip magic header bytes 31 and 139 */
452 if (strm->next_in[0] == 31) {
455 if (strm->avail_in == 0 && xz_avail(state) == -1)
457 if (strm->avail_in && strm->next_in[0] == 139) {
458 /* we have a gzip header, woo hoo! */
462 /* skip rest of header */
463 if (NEXT() != 8) { /* compression method */
464 xz_error(state, LZMA_DATA_ERROR,
465 "unknown compression method");
469 if (flags & 0xe0) { /* reserved flag bits */
470 xz_error(state, LZMA_DATA_ERROR,
471 "unknown header flags set");
474 NEXT(); /* modification time */
478 NEXT(); /* extra flags */
479 NEXT(); /* operating system */
480 if (flags & 4) { /* extra field */
481 len = (unsigned) NEXT();
482 len += (unsigned) NEXT() << 8;
487 if (flags & 8) /* file name */
489 if (flags & 16) /* comment */
491 if (flags & 2) { /* header crc */
495 /* an unexpected end of file is not checked for here -- it will be
496 * noticed on the first request for uncompressed data */
498 /* set up for decompression */
499 inflateReset(&state->zstrm);
500 state->zstrm.adler = crc32(0L, Z_NULL, 0);
505 /* not a gzip file -- save first byte (31) and fall to raw i/o */
512 /* doing raw i/o, save start of raw data for seeking, copy any leftover
513 * input to output -- this assumes that the output buffer is larger than
514 * the input buffer, which also assures space for gzungetc() */
515 state->raw = state->pos;
516 state->next = state->out;
517 if (strm->avail_in) {
518 memcpy(state->next + state->have, strm->next_in, strm->avail_in);
519 state->have += strm->avail_in;
528 xz_decomp(xz_statep state)
532 unsigned long crc, len;
533 lzma_stream *strm = &(state->strm);
535 lzma_action action = LZMA_RUN;
537 /* fill output buffer up to end of deflate stream */
538 had = strm->avail_out;
540 /* get more input for inflate() */
541 if (strm->avail_in == 0 && xz_avail(state) == -1)
543 if (strm->avail_in == 0) {
544 xz_error(state, LZMA_DATA_ERROR, "unexpected end of file");
548 action = LZMA_FINISH;
550 /* decompress and handle errors */
552 if (state->how == GZIP) {
553 state->zstrm.avail_in = (uInt) state->strm.avail_in;
554 state->zstrm.next_in = (Bytef *) state->strm.next_in;
555 state->zstrm.avail_out = (uInt) state->strm.avail_out;
556 state->zstrm.next_out = (Bytef *) state->strm.next_out;
557 ret = inflate(&state->zstrm, Z_NO_FLUSH);
558 if (ret == Z_STREAM_ERROR || ret == Z_NEED_DICT) {
559 xz_error(state, Z_STREAM_ERROR,
560 "internal error: inflate stream corrupt");
563 if (ret == Z_MEM_ERROR)
564 ret = LZMA_MEM_ERROR;
565 if (ret == Z_DATA_ERROR)
566 ret = LZMA_DATA_ERROR;
567 if (ret == Z_STREAM_END)
568 ret = LZMA_STREAM_END;
569 state->strm.avail_in = state->zstrm.avail_in;
570 state->strm.next_in = state->zstrm.next_in;
571 state->strm.avail_out = state->zstrm.avail_out;
572 state->strm.next_out = state->zstrm.next_out;
573 } else /* state->how == LZMA */
575 ret = lzma_code(strm, action);
576 if (ret == LZMA_MEM_ERROR) {
577 xz_error(state, LZMA_MEM_ERROR, "out of memory");
580 if (ret == LZMA_DATA_ERROR) {
581 xz_error(state, LZMA_DATA_ERROR, "compressed data error");
584 } while (strm->avail_out && ret != LZMA_STREAM_END);
586 /* update available output and crc check value */
587 state->have = had - strm->avail_out;
588 state->next = strm->next_out - state->have;
591 crc32(state->zstrm.adler, state->next, state->have);
594 if (ret == LZMA_STREAM_END) {
596 if (state->how == GZIP) {
597 if (gz_next4(state, &crc) == -1 || gz_next4(state, &len) == -1) {
598 xz_error(state, LZMA_DATA_ERROR, "unexpected end of file");
601 if (crc != state->zstrm.adler) {
602 xz_error(state, LZMA_DATA_ERROR, "incorrect data check");
605 if (len != (state->zstrm.total_out & 0xffffffffL)) {
606 xz_error(state, LZMA_DATA_ERROR, "incorrect length check");
609 state->strm.avail_in = 0;
610 state->strm.next_in = NULL;
611 state->strm.avail_out = 0;
612 state->strm.next_out = NULL;
615 if (strm->avail_in != 0 || !state->eof) {
616 xz_error(state, LZMA_DATA_ERROR, "trailing garbage");
619 state->how = LOOK; /* ready for next stream, once have is 0 (leave
620 * state->direct unchanged to remember how) */
623 /* good decompression */
628 xz_make(xz_statep state)
630 lzma_stream *strm = &(state->strm);
632 if (state->how == LOOK) { /* look for lzma / gzip header */
633 if (xz_head(state) == -1)
635 if (state->have) /* got some data from xz_head() */
638 if (state->how == COPY) { /* straight copy */
639 if (xz_load(state, state->out, state->size << 1, &(state->have)) ==
642 state->next = state->out;
643 } else if (state->how == LZMA || state->how == GZIP) { /* decompress */
644 strm->avail_out = state->size << 1;
645 strm->next_out = state->out;
646 if (xz_decomp(state) == -1)
653 xz_skip(xz_statep state, uint64_t len)
657 /* skip over len bytes or reach end-of-file, whichever comes first */
659 /* skip over whatever is in output buffer */
661 n = (uint64_t) state->have > len ?
662 (unsigned) len : state->have;
669 /* output buffer empty -- return if we're at the end of the input */
670 else if (state->eof && state->strm.avail_in == 0)
673 /* need more data to skip -- load up output buffer */
675 /* get more output, looking for header if required */
676 if (xz_make(state) == -1)
683 __libxml2_xzread(xzFile file, void *buf, unsigned len)
689 /* get internal structure */
692 state = (xz_statep) file;
693 strm = &(state->strm);
695 /* check that we're reading and that there's no error */
696 if (state->err != LZMA_OK)
699 /* since an int is returned, make sure len fits in one, otherwise return
700 * with an error (this avoids the flaw in the interface) */
702 xz_error(state, LZMA_BUF_ERROR,
703 "requested length does not fit in int");
707 /* if len is zero, avoid unnecessary operations */
711 /* process a skip request */
714 if (xz_skip(state, state->skip) == -1)
718 /* get len bytes to buf, or less than len if at the end */
721 /* first just try copying data from the output buffer */
723 n = state->have > len ? len : state->have;
724 memcpy(buf, state->next, n);
729 /* output buffer empty -- return if we're at the end of the input */
730 else if (state->eof && strm->avail_in == 0)
733 /* need output data -- for small len or new stream load up our output
735 else if (state->how == LOOK || len < (state->size << 1)) {
736 /* get more output, looking for header if required */
737 if (xz_make(state) == -1)
739 continue; /* no progress yet -- go back to memcpy() above */
740 /* the copy above assures that we will leave with space in the
741 * output buffer, allowing at least one gzungetc() to succeed */
744 /* large len -- read directly into user buffer */
745 else if (state->how == COPY) { /* read directly */
746 if (xz_load(state, buf, len, &n) == -1)
750 /* large len -- decompress directly into user buffer */
751 else { /* state->how == LZMA */
752 strm->avail_out = len;
753 strm->next_out = buf;
754 if (xz_decomp(state) == -1)
760 /* update progress */
762 buf = (char *) buf + n;
767 /* return number of bytes read into user buffer (will fit in int) */
772 __libxml2_xzclose(xzFile file)
777 /* get internal structure */
779 return LZMA_DATA_ERROR;
780 state = (xz_statep) file;
782 /* free memory and close file */
784 lzma_end(&(state->strm));
786 if (state->init == 1)
787 inflateEnd(&(state->zstrm));
793 xmlFree(state->path);
794 ret = close(state->fd);
796 return ret ? ret : LZMA_OK;
798 #endif /* HAVE_LZMA_H */