1 /**
2 * xzlib.c: front end for the transparent support of lzma compression
3 * at the I/O layer, based on an example file from lzma project
4 *
5 * See Copyright for the status of this software.
6 *
7 * Anders F Bjorklund <afb@users.sourceforge.net>
8 */
9 #define IN_LIBXML
10 #include "libxml.h"
11 #ifdef LIBXML_LZMA_ENABLED
12
13 #include <string.h>
14 #include <stdlib.h>
15 #include <errno.h>
16
17 #ifdef HAVE_SYS_STAT_H
18 #include <sys/stat.h>
19 #endif
20 #ifdef HAVE_FCNTL_H
21 #include <fcntl.h>
22 #endif
23 #ifdef HAVE_UNISTD_H
24 #include <unistd.h>
25 #elif defined (_WIN32)
26 #include <io.h>
27 #endif
28 #ifdef LIBXML_ZLIB_ENABLED
29 #include <zlib.h>
30 #endif
31 #ifdef LIBXML_LZMA_ENABLED
32 #include <lzma.h>
33 #endif
34
35 #include "private/xzlib.h"
36 #include <libxml/xmlmemory.h>
37
38 /* values for xz_state how */
39 #define LOOK 0 /* look for a gzip/lzma header */
40 #define COPY 1 /* copy input directly */
41 #define GZIP 2 /* decompress a gzip stream */
42 #define LZMA 3 /* decompress a lzma stream */
43
44 /* internal lzma file state data structure */
45 typedef struct {
46 int mode; /* see lzma modes above */
47 int fd; /* file descriptor */
48 char *path; /* path or fd for error messages */
49 uint64_t pos; /* current position in uncompressed data */
50 unsigned int size; /* buffer size, zero if not allocated yet */
51 unsigned int want; /* requested buffer size, default is BUFSIZ */
52 unsigned char *in; /* input buffer */
53 unsigned char *out; /* output buffer (double-sized when reading) */
54 unsigned char *next; /* next output data to deliver or write */
55 unsigned int have; /* amount of output data unused at next */
56 int eof; /* true if end of input file reached */
57 uint64_t start; /* where the lzma data started, for rewinding */
58 uint64_t raw; /* where the raw data started, for seeking */
59 int how; /* 0: get header, 1: copy, 2: decompress */
60 int direct; /* true if last read direct, false if lzma */
61 /* seek request */
62 uint64_t skip; /* amount to skip (already rewound if backwards) */
63 int seek; /* true if seek request pending */
64 /* error information */
65 int err; /* error code */
66 char *msg; /* error message */
67 /* lzma stream */
68 int init; /* is the inflate stream initialized */
69 lzma_stream strm; /* stream structure in-place (not a pointer) */
70 char padding1[32]; /* padding allowing to cope with possible
71 extensions of above structure without
72 too much side effect */
73 #ifdef LIBXML_ZLIB_ENABLED
74 /* zlib inflate or deflate stream */
75 z_stream zstrm; /* stream structure in-place (not a pointer) */
76 #endif
77 char padding2[32]; /* padding allowing to cope with possible
78 extensions of above structure without
79 too much side effect */
80 } xz_state, *xz_statep;
81
82 static void
xz_error(xz_statep state,int err,const char * msg)83 xz_error(xz_statep state, int err, const char *msg)
84 {
85 /* free previously allocated message and clear */
86 if (state->msg != NULL) {
87 if (state->err != LZMA_MEM_ERROR)
88 xmlFree(state->msg);
89 state->msg = NULL;
90 }
91
92 /* set error code, and if no message, then done */
93 state->err = err;
94 if (msg == NULL)
95 return;
96
97 /* for an out of memory error, save as static string */
98 if (err == LZMA_MEM_ERROR) {
99 state->msg = (char *) msg;
100 return;
101 }
102
103 /* construct error message with path */
104 if ((state->msg =
105 xmlMalloc(strlen(state->path) + strlen(msg) + 3)) == NULL) {
106 state->err = LZMA_MEM_ERROR;
107 state->msg = (char *) "out of memory";
108 return;
109 }
110 strcpy(state->msg, state->path);
111 strcat(state->msg, ": ");
112 strcat(state->msg, msg);
113 return;
114 }
115
116 static void
xz_reset(xz_statep state)117 xz_reset(xz_statep state)
118 {
119 state->have = 0; /* no output data available */
120 state->eof = 0; /* not at end of file */
121 state->how = LOOK; /* look for gzip header */
122 state->direct = 1; /* default for empty file */
123 state->seek = 0; /* no seek request pending */
124 xz_error(state, LZMA_OK, NULL); /* clear error */
125 state->pos = 0; /* no uncompressed data yet */
126 state->strm.avail_in = 0; /* no input data yet */
127 #ifdef LIBXML_ZLIB_ENABLED
128 state->zstrm.avail_in = 0; /* no input data yet */
129 #endif
130 }
131
132 static xzFile
xz_open(const char * path,int fd,const char * mode ATTRIBUTE_UNUSED)133 xz_open(const char *path, int fd, const char *mode ATTRIBUTE_UNUSED)
134 {
135 xz_statep state;
136 off_t offset;
137
138 /* allocate xzFile structure to return */
139 state = xmlMalloc(sizeof(xz_state));
140 if (state == NULL)
141 return NULL;
142 state->size = 0; /* no buffers allocated yet */
143 state->want = BUFSIZ; /* requested buffer size */
144 state->msg = NULL; /* no error message yet */
145 state->init = 0; /* initialization of zlib data */
146
147 /* save the path name for error messages */
148 state->path = xmlMalloc(strlen(path) + 1);
149 if (state->path == NULL) {
150 xmlFree(state);
151 return NULL;
152 }
153 strcpy(state->path, path);
154
155 /* open the file with the appropriate mode (or just use fd) */
156 state->fd = fd != -1 ? fd : open(path,
157 #ifdef O_LARGEFILE
158 O_LARGEFILE |
159 #endif
160 #ifdef O_BINARY
161 O_BINARY |
162 #endif
163 O_RDONLY, 0666);
164 if (state->fd == -1) {
165 xmlFree(state->path);
166 xmlFree(state);
167 return NULL;
168 }
169
170 /* save the current position for rewinding (only if reading) */
171 offset = lseek(state->fd, 0, SEEK_CUR);
172 if (offset == -1)
173 state->start = 0;
174 else
175 state->start = offset;
176
177 /* initialize stream */
178 xz_reset(state);
179
180 /* return stream */
181 return (xzFile) state;
182 }
183
184 static int
xz_compressed(xzFile f)185 xz_compressed(xzFile f) {
186 xz_statep state;
187
188 if (f == NULL)
189 return(-1);
190 state = (xz_statep) f;
191 if (state->init <= 0)
192 return(-1);
193
194 switch (state->how) {
195 case COPY:
196 return(0);
197 case GZIP:
198 #ifdef LIBXML_ZLIB_ENABLED
199 /* Don't use lzma for gzip */
200 return(0);
201 #else
202 return(1);
203 #endif
204 case LZMA:
205 return(1);
206 }
207 return(-1);
208 }
209
210 xzFile
__libxml2_xzopen(const char * path,const char * mode)211 __libxml2_xzopen(const char *path, const char *mode)
212 {
213 return xz_open(path, -1, mode);
214 }
215
216 xzFile
__libxml2_xzdopen(const char * path,int fd,const char * mode)217 __libxml2_xzdopen(const char *path, int fd, const char *mode)
218 {
219 return xz_open(path, fd, mode);
220 }
221
222 static int
xz_load(xz_statep state,unsigned char * buf,unsigned int len,unsigned int * have)223 xz_load(xz_statep state, unsigned char *buf, unsigned int len,
224 unsigned int *have)
225 {
226 int ret;
227
228 *have = 0;
229 do {
230 ret = read(state->fd, buf + *have, len - *have);
231 if (ret <= 0)
232 break;
233 *have += ret;
234 } while (*have < len);
235 if (ret < 0) {
236 xz_error(state, -1, strerror(errno));
237 return -1;
238 }
239 if (ret == 0)
240 state->eof = 1;
241 return 0;
242 }
243
244 static int
xz_avail(xz_statep state)245 xz_avail(xz_statep state)
246 {
247 lzma_stream *strm = &(state->strm);
248
249 if (state->err != LZMA_OK)
250 return -1;
251 if (state->eof == 0) {
252 /* avail_in is size_t, which is not necessary sizeof(unsigned) */
253 unsigned tmp = strm->avail_in;
254
255 if (xz_load(state, state->in, state->size, &tmp) == -1) {
256 strm->avail_in = tmp;
257 return -1;
258 }
259 strm->avail_in = tmp;
260 strm->next_in = state->in;
261 }
262 return 0;
263 }
264
265 #ifdef LIBXML_ZLIB_ENABLED
266 static int
xz_avail_zstrm(xz_statep state)267 xz_avail_zstrm(xz_statep state)
268 {
269 int ret;
270 state->strm.avail_in = state->zstrm.avail_in;
271 state->strm.next_in = state->zstrm.next_in;
272 ret = xz_avail(state);
273 state->zstrm.avail_in = (uInt) state->strm.avail_in;
274 state->zstrm.next_in = (Bytef *) state->strm.next_in;
275 return ret;
276 }
277 #endif
278
279 static int
is_format_xz(xz_statep state)280 is_format_xz(xz_statep state)
281 {
282 lzma_stream *strm = &(state->strm);
283
284 return strm->avail_in >= 6 && memcmp(state->in, "\3757zXZ", 6) == 0;
285 }
286
287 static int
is_format_lzma(xz_statep state)288 is_format_lzma(xz_statep state)
289 {
290 lzma_stream *strm = &(state->strm);
291
292 lzma_filter filter;
293 lzma_options_lzma *opt;
294 uint32_t dict_size;
295 uint64_t uncompressed_size;
296 size_t i;
297
298 if (strm->avail_in < 13)
299 return 0;
300
301 filter.id = LZMA_FILTER_LZMA1;
302 if (lzma_properties_decode(&filter, NULL, state->in, 5) != LZMA_OK)
303 return 0;
304
305 opt = filter.options;
306 dict_size = opt->dict_size;
307 free(opt); /* we can't use xmlFree on a string returned by zlib */
308
309 /* A hack to ditch tons of false positives: We allow only dictionary
310 * sizes that are 2^n or 2^n + 2^(n-1) or UINT32_MAX. LZMA_Alone
311 * created only files with 2^n, but accepts any dictionary size.
312 * If someone complains, this will be reconsidered.
313 */
314 if (dict_size != UINT32_MAX) {
315 uint32_t d;
316
317 if (dict_size == 0)
318 return 0;
319
320 d = dict_size - 1;
321 d |= d >> 2;
322 d |= d >> 3;
323 d |= d >> 4;
324 d |= d >> 8;
325 d |= d >> 16;
326 ++d;
327 if (d != dict_size || dict_size == 0)
328 return 0;
329 }
330
331 /* Another hack to ditch false positives: Assume that if the
332 * uncompressed size is known, it must be less than 256 GiB.
333 * Again, if someone complains, this will be reconsidered.
334 */
335 uncompressed_size = 0;
336 for (i = 0; i < 8; ++i)
337 uncompressed_size |= (uint64_t) (state->in[5 + i]) << (i * 8);
338
339 if (uncompressed_size != UINT64_MAX
340 && uncompressed_size > (UINT64_C(1) << 38))
341 return 0;
342
343 return 1;
344 }
345
346 #ifdef LIBXML_ZLIB_ENABLED
347
348 /* Get next byte from input, or -1 if end or error. */
349 #define NEXT() ((strm->avail_in == 0 && xz_avail(state) == -1) ? -1 : \
350 (strm->avail_in == 0 ? -1 : \
351 (strm->avail_in--, *(strm->next_in)++)))
352 /* Same thing, but from zstrm */
353 #define NEXTZ() ((strm->avail_in == 0 && xz_avail_zstrm(state) == -1) ? -1 : \
354 (strm->avail_in == 0 ? -1 : \
355 (strm->avail_in--, *(strm->next_in)++)))
356
357 /* Get a four-byte little-endian integer and return 0 on success and the value
358 in *ret. Otherwise -1 is returned and *ret is not modified. */
359 static int
gz_next4(xz_statep state,unsigned long * ret)360 gz_next4(xz_statep state, unsigned long *ret)
361 {
362 int ch;
363 unsigned long val;
364 z_streamp strm = &(state->zstrm);
365
366 val = NEXTZ();
367 val += (unsigned) NEXTZ() << 8;
368 val += (unsigned long) NEXTZ() << 16;
369 ch = NEXTZ();
370 if (ch == -1)
371 return -1;
372 val += (unsigned long) ch << 24;
373 *ret = val;
374 return 0;
375 }
376 #endif
377
378 static int
xz_head(xz_statep state)379 xz_head(xz_statep state)
380 {
381 lzma_stream *strm = &(state->strm);
382 lzma_stream init = LZMA_STREAM_INIT;
383 int flags;
384 unsigned len;
385
386 /* Avoid unused variable warning if features are disabled. */
387 (void) flags;
388 (void) len;
389
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)
397 xmlFree(state->out);
398 if (state->in != NULL)
399 xmlFree(state->in);
400 xz_error(state, LZMA_MEM_ERROR, "out of memory");
401 return -1;
402 }
403 state->size = state->want;
404
405 /* allocate decoder memory */
406 state->strm = init;
407 state->strm.avail_in = 0;
408 state->strm.next_in = NULL;
409 if (lzma_auto_decoder(&state->strm, 100000000, 0) != LZMA_OK) {
410 xmlFree(state->out);
411 xmlFree(state->in);
412 state->size = 0;
413 xz_error(state, LZMA_MEM_ERROR, "out of memory");
414 return -1;
415 }
416 #ifdef LIBXML_ZLIB_ENABLED
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 */
425 xmlFree(state->out);
426 xmlFree(state->in);
427 state->size = 0;
428 xz_error(state, LZMA_MEM_ERROR, "out of memory");
429 return -1;
430 }
431 state->init = 1;
432 }
433 #endif
434 }
435
436 /* get some data in the input buffer */
437 if (strm->avail_in == 0) {
438 if (xz_avail(state) == -1)
439 return -1;
440 if (strm->avail_in == 0)
441 return 0;
442 }
443
444 /* look for the xz magic header bytes */
445 if (is_format_xz(state) || is_format_lzma(state)) {
446 state->how = LZMA;
447 state->direct = 0;
448 return 0;
449 }
450 #ifdef LIBXML_ZLIB_ENABLED
451 /* look for the gzip magic header bytes 31 and 139 */
452 if (strm->next_in[0] == 31) {
453 strm->avail_in--;
454 strm->next_in++;
455 if (strm->avail_in == 0 && xz_avail(state) == -1)
456 return -1;
457 if (strm->avail_in && strm->next_in[0] == 139) {
458 /* we have a gzip header, woo hoo! */
459 strm->avail_in--;
460 strm->next_in++;
461
462 /* skip rest of header */
463 if (NEXT() != 8) { /* compression method */
464 xz_error(state, LZMA_DATA_ERROR,
465 "unknown compression method");
466 return -1;
467 }
468 flags = NEXT();
469 if (flags & 0xe0) { /* reserved flag bits */
470 xz_error(state, LZMA_DATA_ERROR,
471 "unknown header flags set");
472 return -1;
473 }
474 NEXT(); /* modification time */
475 NEXT();
476 NEXT();
477 NEXT();
478 NEXT(); /* extra flags */
479 NEXT(); /* operating system */
480 if (flags & 4) { /* extra field */
481 len = (unsigned) NEXT();
482 len += (unsigned) NEXT() << 8;
483 while (len--)
484 if (NEXT() < 0)
485 break;
486 }
487 if (flags & 8) /* file name */
488 while (NEXT() > 0) ;
489 if (flags & 16) /* comment */
490 while (NEXT() > 0) ;
491 if (flags & 2) { /* header crc */
492 NEXT();
493 NEXT();
494 }
495 /* an unexpected end of file is not checked for here -- it will be
496 * noticed on the first request for uncompressed data */
497
498 /* set up for decompression */
499 inflateReset(&state->zstrm);
500 state->zstrm.adler = crc32(0L, Z_NULL, 0);
501 state->how = GZIP;
502 state->direct = 0;
503 return 0;
504 } else {
505 /* not a gzip file -- save first byte (31) and fall to raw i/o */
506 state->out[0] = 31;
507 state->have = 1;
508 }
509 }
510 #endif
511
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;
520 strm->avail_in = 0;
521 }
522 state->how = COPY;
523 state->direct = 1;
524 return 0;
525 }
526
527 static int
xz_decomp(xz_statep state)528 xz_decomp(xz_statep state)
529 {
530 int ret;
531 unsigned had;
532 unsigned long crc, len;
533 lzma_stream *strm = &(state->strm);
534
535 lzma_action action = LZMA_RUN;
536
537 /* Avoid unused variable warning if features are disabled. */
538 (void) crc;
539 (void) len;
540
541 /* fill output buffer up to end of deflate stream */
542 had = strm->avail_out;
543 do {
544 /* get more input for inflate() */
545 if (strm->avail_in == 0 && xz_avail(state) == -1)
546 return -1;
547 if (strm->avail_in == 0) {
548 xz_error(state, LZMA_DATA_ERROR, "unexpected end of file");
549 return -1;
550 }
551 if (state->eof)
552 action = LZMA_FINISH;
553
554 /* decompress and handle errors */
555 #ifdef LIBXML_ZLIB_ENABLED
556 if (state->how == GZIP) {
557 state->zstrm.avail_in = (uInt) state->strm.avail_in;
558 state->zstrm.next_in = (Bytef *) state->strm.next_in;
559 state->zstrm.avail_out = (uInt) state->strm.avail_out;
560 state->zstrm.next_out = (Bytef *) state->strm.next_out;
561 ret = inflate(&state->zstrm, Z_NO_FLUSH);
562 if (ret == Z_STREAM_ERROR || ret == Z_NEED_DICT) {
563 xz_error(state, Z_STREAM_ERROR,
564 "internal error: inflate stream corrupt");
565 return -1;
566 }
567 /*
568 * FIXME: Remapping a couple of error codes and falling through
569 * to the LZMA error handling looks fragile.
570 */
571 if (ret == Z_MEM_ERROR)
572 ret = LZMA_MEM_ERROR;
573 if (ret == Z_DATA_ERROR)
574 ret = LZMA_DATA_ERROR;
575 if (ret == Z_STREAM_END)
576 ret = LZMA_STREAM_END;
577 state->strm.avail_in = state->zstrm.avail_in;
578 state->strm.next_in = state->zstrm.next_in;
579 state->strm.avail_out = state->zstrm.avail_out;
580 state->strm.next_out = state->zstrm.next_out;
581 } else /* state->how == LZMA */
582 #endif
583 ret = lzma_code(strm, action);
584 if (ret == LZMA_MEM_ERROR) {
585 xz_error(state, LZMA_MEM_ERROR, "out of memory");
586 return -1;
587 }
588 if (ret == LZMA_DATA_ERROR) {
589 xz_error(state, LZMA_DATA_ERROR, "compressed data error");
590 return -1;
591 }
592 if (ret == LZMA_PROG_ERROR) {
593 xz_error(state, LZMA_PROG_ERROR, "compression error");
594 return -1;
595 }
596 if ((state->how != GZIP) &&
597 (ret != LZMA_OK) && (ret != LZMA_STREAM_END)) {
598 xz_error(state, ret, "lzma error");
599 return -1;
600 }
601 } while (strm->avail_out && ret != LZMA_STREAM_END);
602
603 /* update available output and crc check value */
604 state->have = had - strm->avail_out;
605 state->next = strm->next_out - state->have;
606 #ifdef LIBXML_ZLIB_ENABLED
607 state->zstrm.adler =
608 crc32(state->zstrm.adler, state->next, state->have);
609 #endif
610
611 if (ret == LZMA_STREAM_END) {
612 #ifdef LIBXML_ZLIB_ENABLED
613 if (state->how == GZIP) {
614 if (gz_next4(state, &crc) == -1 || gz_next4(state, &len) == -1) {
615 xz_error(state, LZMA_DATA_ERROR, "unexpected end of file");
616 return -1;
617 }
618 if (crc != state->zstrm.adler) {
619 xz_error(state, LZMA_DATA_ERROR, "incorrect data check");
620 return -1;
621 }
622 if (len != (state->zstrm.total_out & 0xffffffffL)) {
623 xz_error(state, LZMA_DATA_ERROR, "incorrect length check");
624 return -1;
625 }
626 state->strm.avail_in = 0;
627 state->strm.next_in = NULL;
628 state->strm.avail_out = 0;
629 state->strm.next_out = NULL;
630 } else
631 #endif
632 if (strm->avail_in != 0 || !state->eof) {
633 xz_error(state, LZMA_DATA_ERROR, "trailing garbage");
634 return -1;
635 }
636 state->how = LOOK; /* ready for next stream, once have is 0 (leave
637 * state->direct unchanged to remember how) */
638 }
639
640 /* good decompression */
641 return 0;
642 }
643
644 static int
xz_make(xz_statep state)645 xz_make(xz_statep state)
646 {
647 lzma_stream *strm = &(state->strm);
648
649 if (state->how == LOOK) { /* look for lzma / gzip header */
650 if (xz_head(state) == -1)
651 return -1;
652 if (state->have) /* got some data from xz_head() */
653 return 0;
654 }
655 if (state->how == COPY) { /* straight copy */
656 if (xz_load(state, state->out, state->size << 1, &(state->have)) ==
657 -1)
658 return -1;
659 state->next = state->out;
660 } else if (state->how == LZMA || state->how == GZIP) { /* decompress */
661 strm->avail_out = state->size << 1;
662 strm->next_out = state->out;
663 if (xz_decomp(state) == -1)
664 return -1;
665 }
666 return 0;
667 }
668
669 static int
xz_skip(xz_statep state,uint64_t len)670 xz_skip(xz_statep state, uint64_t len)
671 {
672 unsigned n;
673
674 /* skip over len bytes or reach end-of-file, whichever comes first */
675 while (len)
676 /* skip over whatever is in output buffer */
677 if (state->have) {
678 n = (uint64_t) state->have > len ?
679 (unsigned) len : state->have;
680 state->have -= n;
681 state->next += n;
682 state->pos += n;
683 len -= n;
684 }
685
686 /* output buffer empty -- return if we're at the end of the input */
687 else if (state->eof && state->strm.avail_in == 0)
688 break;
689
690 /* need more data to skip -- load up output buffer */
691 else {
692 /* get more output, looking for header if required */
693 if (xz_make(state) == -1)
694 return -1;
695 }
696 return 0;
697 }
698
699 int
__libxml2_xzcompressed(xzFile f)700 __libxml2_xzcompressed(xzFile f) {
701 xz_head(f);
702
703 return xz_compressed(f);
704 }
705
706 int
__libxml2_xzread(xzFile file,void * buf,unsigned len)707 __libxml2_xzread(xzFile file, void *buf, unsigned len)
708 {
709 unsigned got, n;
710 xz_statep state;
711 lzma_stream *strm;
712
713 /* get internal structure */
714 if (file == NULL)
715 return -1;
716 state = (xz_statep) file;
717 strm = &(state->strm);
718
719 /* check that we're reading and that there's no error */
720 if (state->err != LZMA_OK)
721 return -1;
722
723 /* since an int is returned, make sure len fits in one, otherwise return
724 * with an error (this avoids the flaw in the interface) */
725 if ((int) len < 0) {
726 xz_error(state, LZMA_BUF_ERROR,
727 "requested length does not fit in int");
728 return -1;
729 }
730
731 /* if len is zero, avoid unnecessary operations */
732 if (len == 0)
733 return 0;
734
735 /* process a skip request */
736 if (state->seek) {
737 state->seek = 0;
738 if (xz_skip(state, state->skip) == -1)
739 return -1;
740 }
741
742 /* get len bytes to buf, or less than len if at the end */
743 got = 0;
744 do {
745 /* first just try copying data from the output buffer */
746 if (state->have) {
747 n = state->have > len ? len : state->have;
748 memcpy(buf, state->next, n);
749 state->next += n;
750 state->have -= n;
751 }
752
753 /* output buffer empty -- return if we're at the end of the input */
754 else if (state->eof && strm->avail_in == 0)
755 break;
756
757 /* need output data -- for small len or new stream load up our output
758 * buffer */
759 else if (state->how == LOOK || len < (state->size << 1)) {
760 /* get more output, looking for header if required */
761 if (xz_make(state) == -1)
762 return -1;
763 continue; /* no progress yet -- go back to memcpy() above */
764 /* the copy above assures that we will leave with space in the
765 * output buffer, allowing at least one gzungetc() to succeed */
766 }
767
768 /* large len -- read directly into user buffer */
769 else if (state->how == COPY) { /* read directly */
770 if (xz_load(state, buf, len, &n) == -1)
771 return -1;
772 }
773
774 /* large len -- decompress directly into user buffer */
775 else { /* state->how == LZMA */
776 strm->avail_out = len;
777 strm->next_out = buf;
778 if (xz_decomp(state) == -1)
779 return -1;
780 n = state->have;
781 state->have = 0;
782 }
783
784 /* update progress */
785 len -= n;
786 buf = (char *) buf + n;
787 got += n;
788 state->pos += n;
789 } while (len);
790
791 /* return number of bytes read into user buffer (will fit in int) */
792 return (int) got;
793 }
794
795 int
__libxml2_xzclose(xzFile file)796 __libxml2_xzclose(xzFile file)
797 {
798 int ret;
799 xz_statep state;
800
801 /* get internal structure */
802 if (file == NULL)
803 return LZMA_DATA_ERROR;
804 state = (xz_statep) file;
805
806 /* free memory and close file */
807 if (state->size) {
808 lzma_end(&(state->strm));
809 #ifdef LIBXML_ZLIB_ENABLED
810 if (state->init == 1)
811 inflateEnd(&(state->zstrm));
812 state->init = 0;
813 #endif
814 xmlFree(state->out);
815 xmlFree(state->in);
816 }
817 xmlFree(state->path);
818 if ((state->msg != NULL) && (state->err != LZMA_MEM_ERROR))
819 xmlFree(state->msg);
820 ret = close(state->fd);
821 xmlFree(state);
822 return ret ? ret : LZMA_OK;
823 }
824 #endif /* LIBXML_LZMA_ENABLED */
825