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_TYPES_H
18 #include <sys/types.h>
19 #endif
20 #ifdef HAVE_SYS_STAT_H
21 #include <sys/stat.h>
22 #endif
23 #ifdef HAVE_FCNTL_H
24 #include <fcntl.h>
25 #endif
26 #ifdef HAVE_UNISTD_H
27 #include <unistd.h>
28 #endif
29 #ifdef LIBXML_ZLIB_ENABLED
30 #include <zlib.h>
31 #endif
32 #ifdef LIBXML_LZMA_ENABLED
33 #include <lzma.h>
34 #endif
35
36 #include "xzlib.h"
37 #include <libxml/xmlmemory.h>
38
39 /* values for xz_state how */
40 #define LOOK 0 /* look for a gzip/lzma header */
41 #define COPY 1 /* copy input directly */
42 #define GZIP 2 /* decompress a gzip stream */
43 #define LZMA 3 /* decompress a lzma stream */
44
45 /* internal lzma file state data structure */
46 typedef struct {
47 int mode; /* see lzma modes above */
48 int fd; /* file descriptor */
49 char *path; /* path or fd for error messages */
50 uint64_t pos; /* current position in uncompressed data */
51 unsigned int size; /* buffer size, zero if not allocated yet */
52 unsigned int want; /* requested buffer size, default is BUFSIZ */
53 unsigned char *in; /* input buffer */
54 unsigned char *out; /* output buffer (double-sized when reading) */
55 unsigned char *next; /* next output data to deliver or write */
56 unsigned int have; /* amount of output data unused at next */
57 int eof; /* true if end of input file reached */
58 uint64_t start; /* where the lzma data started, for rewinding */
59 uint64_t raw; /* where the raw data started, for seeking */
60 int how; /* 0: get header, 1: copy, 2: decompress */
61 int direct; /* true if last read direct, false if lzma */
62 /* seek request */
63 uint64_t skip; /* amount to skip (already rewound if backwards) */
64 int seek; /* true if seek request pending */
65 /* error information */
66 int err; /* error code */
67 char *msg; /* error message */
68 /* lzma stream */
69 int init; /* is the inflate stream initialized */
70 lzma_stream strm; /* stream structure in-place (not a pointer) */
71 char padding1[32]; /* padding allowing to cope with possible
72 extensions of above structure without
73 too much side effect */
74 #ifdef LIBXML_ZLIB_ENABLED
75 /* zlib inflate or deflate stream */
76 z_stream zstrm; /* stream structure in-place (not a pointer) */
77 #endif
78 char padding2[32]; /* padding allowing to cope with possible
79 extensions of above structure without
80 too much side effect */
81 } xz_state, *xz_statep;
82
83 static void
xz_error(xz_statep state,int err,const char * msg)84 xz_error(xz_statep state, int err, const char *msg)
85 {
86 /* free previously allocated message and clear */
87 if (state->msg != NULL) {
88 if (state->err != LZMA_MEM_ERROR)
89 xmlFree(state->msg);
90 state->msg = NULL;
91 }
92
93 /* set error code, and if no message, then done */
94 state->err = err;
95 if (msg == NULL)
96 return;
97
98 /* for an out of memory error, save as static string */
99 if (err == LZMA_MEM_ERROR) {
100 state->msg = (char *) msg;
101 return;
102 }
103
104 /* construct error message with path */
105 if ((state->msg =
106 xmlMalloc(strlen(state->path) + strlen(msg) + 3)) == NULL) {
107 state->err = LZMA_MEM_ERROR;
108 state->msg = (char *) "out of memory";
109 return;
110 }
111 strcpy(state->msg, state->path);
112 strcat(state->msg, ": ");
113 strcat(state->msg, msg);
114 return;
115 }
116
117 static void
xz_reset(xz_statep state)118 xz_reset(xz_statep state)
119 {
120 state->have = 0; /* no output data available */
121 state->eof = 0; /* not at end of file */
122 state->how = LOOK; /* look for gzip header */
123 state->direct = 1; /* default for empty file */
124 state->seek = 0; /* no seek request pending */
125 xz_error(state, LZMA_OK, NULL); /* clear error */
126 state->pos = 0; /* no uncompressed data yet */
127 state->strm.avail_in = 0; /* no input data yet */
128 #ifdef LIBXML_ZLIB_ENABLED
129 state->zstrm.avail_in = 0; /* no input data yet */
130 #endif
131 }
132
133 static xzFile
xz_open(const char * path,int fd,const char * mode ATTRIBUTE_UNUSED)134 xz_open(const char *path, int fd, const char *mode ATTRIBUTE_UNUSED)
135 {
136 xz_statep state;
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 state->start = lseek(state->fd, 0, SEEK_CUR);
172 if (state->start == (uint64_t) - 1)
173 state->start = 0;
174
175 /* initialize stream */
176 xz_reset(state);
177
178 /* return stream */
179 return (xzFile) state;
180 }
181
182 static int
xz_compressed(xzFile f)183 xz_compressed(xzFile f) {
184 xz_statep state;
185
186 if (f == NULL)
187 return(-1);
188 state = (xz_statep) f;
189 if (state->init <= 0)
190 return(-1);
191
192 switch (state->how) {
193 case COPY:
194 return(0);
195 case GZIP:
196 case LZMA:
197 return(1);
198 }
199 return(-1);
200 }
201
202 xzFile
__libxml2_xzopen(const char * path,const char * mode)203 __libxml2_xzopen(const char *path, const char *mode)
204 {
205 return xz_open(path, -1, mode);
206 }
207
208 int
__libxml2_xzcompressed(xzFile f)209 __libxml2_xzcompressed(xzFile f) {
210 return xz_compressed(f);
211 }
212
213 xzFile
__libxml2_xzdopen(int fd,const char * mode)214 __libxml2_xzdopen(int fd, const char *mode)
215 {
216 char *path; /* identifier for error messages */
217 xzFile xz;
218
219 if (fd == -1 || (path = xmlMalloc(7 + 3 * sizeof(int))) == NULL)
220 return NULL;
221 sprintf(path, "<fd:%d>", fd); /* for debugging */
222 xz = xz_open(path, fd, mode);
223 xmlFree(path);
224 return xz;
225 }
226
227 static int
xz_load(xz_statep state,unsigned char * buf,unsigned int len,unsigned int * have)228 xz_load(xz_statep state, unsigned char *buf, unsigned int len,
229 unsigned int *have)
230 {
231 int ret;
232
233 *have = 0;
234 do {
235 ret = read(state->fd, buf + *have, len - *have);
236 if (ret <= 0)
237 break;
238 *have += ret;
239 } while (*have < len);
240 if (ret < 0) {
241 xz_error(state, -1, strerror(errno));
242 return -1;
243 }
244 if (ret == 0)
245 state->eof = 1;
246 return 0;
247 }
248
249 static int
xz_avail(xz_statep state)250 xz_avail(xz_statep state)
251 {
252 lzma_stream *strm = &(state->strm);
253
254 if (state->err != LZMA_OK)
255 return -1;
256 if (state->eof == 0) {
257 /* avail_in is size_t, which is not necessary sizeof(unsigned) */
258 unsigned tmp = strm->avail_in;
259
260 if (xz_load(state, state->in, state->size, &tmp) == -1) {
261 strm->avail_in = tmp;
262 return -1;
263 }
264 strm->avail_in = tmp;
265 strm->next_in = state->in;
266 }
267 return 0;
268 }
269
270 #ifdef LIBXML_ZLIB_ENABLED
271 static int
xz_avail_zstrm(xz_statep state)272 xz_avail_zstrm(xz_statep state)
273 {
274 int ret;
275 state->strm.avail_in = state->zstrm.avail_in;
276 state->strm.next_in = state->zstrm.next_in;
277 ret = xz_avail(state);
278 state->zstrm.avail_in = (uInt) state->strm.avail_in;
279 state->zstrm.next_in = (Bytef *) state->strm.next_in;
280 return ret;
281 }
282 #endif
283
284 static int
is_format_xz(xz_statep state)285 is_format_xz(xz_statep state)
286 {
287 lzma_stream *strm = &(state->strm);
288
289 return strm->avail_in >= 6 && memcmp(state->in, "\3757zXZ", 6) == 0;
290 }
291
292 static int
is_format_lzma(xz_statep state)293 is_format_lzma(xz_statep state)
294 {
295 lzma_stream *strm = &(state->strm);
296
297 lzma_filter filter;
298 lzma_options_lzma *opt;
299 uint32_t dict_size;
300 uint64_t uncompressed_size;
301 size_t i;
302
303 if (strm->avail_in < 13)
304 return 0;
305
306 filter.id = LZMA_FILTER_LZMA1;
307 if (lzma_properties_decode(&filter, NULL, state->in, 5) != LZMA_OK)
308 return 0;
309
310 opt = filter.options;
311 dict_size = opt->dict_size;
312 free(opt); /* we can't use xmlFree on a string returned by zlib */
313
314 /* A hack to ditch tons of false positives: We allow only dictionary
315 * sizes that are 2^n or 2^n + 2^(n-1) or UINT32_MAX. LZMA_Alone
316 * created only files with 2^n, but accepts any dictionary size.
317 * If someone complains, this will be reconsidered.
318 */
319 if (dict_size != UINT32_MAX) {
320 uint32_t d = dict_size - 1;
321
322 d |= d >> 2;
323 d |= d >> 3;
324 d |= d >> 4;
325 d |= d >> 8;
326 d |= d >> 16;
327 ++d;
328 if (d != dict_size || dict_size == 0)
329 return 0;
330 }
331
332 /* Another hack to ditch false positives: Assume that if the
333 * uncompressed size is known, it must be less than 256 GiB.
334 * Again, if someone complains, this will be reconsidered.
335 */
336 uncompressed_size = 0;
337 for (i = 0; i < 8; ++i)
338 uncompressed_size |= (uint64_t) (state->in[5 + i]) << (i * 8);
339
340 if (uncompressed_size != UINT64_MAX
341 && uncompressed_size > (UINT64_C(1) << 38))
342 return 0;
343
344 return 1;
345 }
346
347 #ifdef LIBXML_ZLIB_ENABLED
348
349 /* Get next byte from input, or -1 if end or error. */
350 #define NEXT() ((strm->avail_in == 0 && xz_avail(state) == -1) ? -1 : \
351 (strm->avail_in == 0 ? -1 : \
352 (strm->avail_in--, *(strm->next_in)++)))
353 /* Same thing, but from zstrm */
354 #define NEXTZ() ((strm->avail_in == 0 && xz_avail_zstrm(state) == -1) ? -1 : \
355 (strm->avail_in == 0 ? -1 : \
356 (strm->avail_in--, *(strm->next_in)++)))
357
358 /* Get a four-byte little-endian integer and return 0 on success and the value
359 in *ret. Otherwise -1 is returned and *ret is not modified. */
360 static int
gz_next4(xz_statep state,unsigned long * ret)361 gz_next4(xz_statep state, unsigned long *ret)
362 {
363 int ch;
364 unsigned long val;
365 z_streamp strm = &(state->zstrm);
366
367 val = NEXTZ();
368 val += (unsigned) NEXTZ() << 8;
369 val += (unsigned long) NEXTZ() << 16;
370 ch = NEXTZ();
371 if (ch == -1)
372 return -1;
373 val += (unsigned long) ch << 24;
374 *ret = val;
375 return 0;
376 }
377 #endif
378
379 static int
xz_head(xz_statep state)380 xz_head(xz_statep state)
381 {
382 lzma_stream *strm = &(state->strm);
383 lzma_stream init = LZMA_STREAM_INIT;
384 int flags;
385 unsigned len;
386
387 /* Avoid unused variable warning if features are disabled. */
388 (void) flags;
389 (void) len;
390
391 /* allocate read buffers and inflate memory */
392 if (state->size == 0) {
393 /* allocate buffers */
394 state->in = xmlMalloc(state->want);
395 state->out = xmlMalloc(state->want << 1);
396 if (state->in == NULL || state->out == NULL) {
397 if (state->out != NULL)
398 xmlFree(state->out);
399 if (state->in != NULL)
400 xmlFree(state->in);
401 xz_error(state, LZMA_MEM_ERROR, "out of memory");
402 return -1;
403 }
404 state->size = state->want;
405
406 /* allocate decoder memory */
407 state->strm = init;
408 state->strm.avail_in = 0;
409 state->strm.next_in = NULL;
410 if (lzma_auto_decoder(&state->strm, 100000000, 0) != LZMA_OK) {
411 xmlFree(state->out);
412 xmlFree(state->in);
413 state->size = 0;
414 xz_error(state, LZMA_MEM_ERROR, "out of memory");
415 return -1;
416 }
417 #ifdef LIBXML_ZLIB_ENABLED
418 /* allocate inflate memory */
419 state->zstrm.zalloc = Z_NULL;
420 state->zstrm.zfree = Z_NULL;
421 state->zstrm.opaque = Z_NULL;
422 state->zstrm.avail_in = 0;
423 state->zstrm.next_in = Z_NULL;
424 if (state->init == 0) {
425 if (inflateInit2(&(state->zstrm), -15) != Z_OK) {/* raw inflate */
426 xmlFree(state->out);
427 xmlFree(state->in);
428 state->size = 0;
429 xz_error(state, LZMA_MEM_ERROR, "out of memory");
430 return -1;
431 }
432 state->init = 1;
433 }
434 #endif
435 }
436
437 /* get some data in the input buffer */
438 if (strm->avail_in == 0) {
439 if (xz_avail(state) == -1)
440 return -1;
441 if (strm->avail_in == 0)
442 return 0;
443 }
444
445 /* look for the xz magic header bytes */
446 if (is_format_xz(state) || is_format_lzma(state)) {
447 state->how = LZMA;
448 state->direct = 0;
449 return 0;
450 }
451 #ifdef LIBXML_ZLIB_ENABLED
452 /* look for the gzip magic header bytes 31 and 139 */
453 if (strm->next_in[0] == 31) {
454 strm->avail_in--;
455 strm->next_in++;
456 if (strm->avail_in == 0 && xz_avail(state) == -1)
457 return -1;
458 if (strm->avail_in && strm->next_in[0] == 139) {
459 /* we have a gzip header, woo hoo! */
460 strm->avail_in--;
461 strm->next_in++;
462
463 /* skip rest of header */
464 if (NEXT() != 8) { /* compression method */
465 xz_error(state, LZMA_DATA_ERROR,
466 "unknown compression method");
467 return -1;
468 }
469 flags = NEXT();
470 if (flags & 0xe0) { /* reserved flag bits */
471 xz_error(state, LZMA_DATA_ERROR,
472 "unknown header flags set");
473 return -1;
474 }
475 NEXT(); /* modification time */
476 NEXT();
477 NEXT();
478 NEXT();
479 NEXT(); /* extra flags */
480 NEXT(); /* operating system */
481 if (flags & 4) { /* extra field */
482 len = (unsigned) NEXT();
483 len += (unsigned) NEXT() << 8;
484 while (len--)
485 if (NEXT() < 0)
486 break;
487 }
488 if (flags & 8) /* file name */
489 while (NEXT() > 0) ;
490 if (flags & 16) /* comment */
491 while (NEXT() > 0) ;
492 if (flags & 2) { /* header crc */
493 NEXT();
494 NEXT();
495 }
496 /* an unexpected end of file is not checked for here -- it will be
497 * noticed on the first request for uncompressed data */
498
499 /* set up for decompression */
500 inflateReset(&state->zstrm);
501 state->zstrm.adler = crc32(0L, Z_NULL, 0);
502 state->how = GZIP;
503 state->direct = 0;
504 return 0;
505 } else {
506 /* not a gzip file -- save first byte (31) and fall to raw i/o */
507 state->out[0] = 31;
508 state->have = 1;
509 }
510 }
511 #endif
512
513 /* doing raw i/o, save start of raw data for seeking, copy any leftover
514 * input to output -- this assumes that the output buffer is larger than
515 * the input buffer, which also assures space for gzungetc() */
516 state->raw = state->pos;
517 state->next = state->out;
518 if (strm->avail_in) {
519 memcpy(state->next + state->have, strm->next_in, strm->avail_in);
520 state->have += strm->avail_in;
521 strm->avail_in = 0;
522 }
523 state->how = COPY;
524 state->direct = 1;
525 return 0;
526 }
527
528 static int
xz_decomp(xz_statep state)529 xz_decomp(xz_statep state)
530 {
531 int ret;
532 unsigned had;
533 unsigned long crc, len;
534 lzma_stream *strm = &(state->strm);
535
536 lzma_action action = LZMA_RUN;
537
538 /* Avoid unused variable warning if features are disabled. */
539 (void) crc;
540 (void) len;
541
542 /* fill output buffer up to end of deflate stream */
543 had = strm->avail_out;
544 do {
545 /* get more input for inflate() */
546 if (strm->avail_in == 0 && xz_avail(state) == -1)
547 return -1;
548 if (strm->avail_in == 0) {
549 xz_error(state, LZMA_DATA_ERROR, "unexpected end of file");
550 return -1;
551 }
552 if (state->eof)
553 action = LZMA_FINISH;
554
555 /* decompress and handle errors */
556 #ifdef LIBXML_ZLIB_ENABLED
557 if (state->how == GZIP) {
558 state->zstrm.avail_in = (uInt) state->strm.avail_in;
559 state->zstrm.next_in = (Bytef *) state->strm.next_in;
560 state->zstrm.avail_out = (uInt) state->strm.avail_out;
561 state->zstrm.next_out = (Bytef *) state->strm.next_out;
562 ret = inflate(&state->zstrm, Z_NO_FLUSH);
563 if (ret == Z_STREAM_ERROR || ret == Z_NEED_DICT) {
564 xz_error(state, Z_STREAM_ERROR,
565 "internal error: inflate stream corrupt");
566 return -1;
567 }
568 /*
569 * FIXME: Remapping a couple of error codes and falling through
570 * to the LZMA error handling looks fragile.
571 */
572 if (ret == Z_MEM_ERROR)
573 ret = LZMA_MEM_ERROR;
574 if (ret == Z_DATA_ERROR)
575 ret = LZMA_DATA_ERROR;
576 if (ret == Z_STREAM_END)
577 ret = LZMA_STREAM_END;
578 state->strm.avail_in = state->zstrm.avail_in;
579 state->strm.next_in = state->zstrm.next_in;
580 state->strm.avail_out = state->zstrm.avail_out;
581 state->strm.next_out = state->zstrm.next_out;
582 } else /* state->how == LZMA */
583 #endif
584 ret = lzma_code(strm, action);
585 if (ret == LZMA_MEM_ERROR) {
586 xz_error(state, LZMA_MEM_ERROR, "out of memory");
587 return -1;
588 }
589 if (ret == LZMA_DATA_ERROR) {
590 xz_error(state, LZMA_DATA_ERROR, "compressed data error");
591 return -1;
592 }
593 if (ret == LZMA_PROG_ERROR) {
594 xz_error(state, LZMA_PROG_ERROR, "compression error");
595 return -1;
596 }
597 if ((state->how != GZIP) &&
598 (ret != LZMA_OK) && (ret != LZMA_STREAM_END)) {
599 xz_error(state, ret, "lzma error");
600 return -1;
601 }
602 } while (strm->avail_out && ret != LZMA_STREAM_END);
603
604 /* update available output and crc check value */
605 state->have = had - strm->avail_out;
606 state->next = strm->next_out - state->have;
607 #ifdef LIBXML_ZLIB_ENABLED
608 state->zstrm.adler =
609 crc32(state->zstrm.adler, state->next, state->have);
610 #endif
611
612 if (ret == LZMA_STREAM_END) {
613 #ifdef LIBXML_ZLIB_ENABLED
614 if (state->how == GZIP) {
615 if (gz_next4(state, &crc) == -1 || gz_next4(state, &len) == -1) {
616 xz_error(state, LZMA_DATA_ERROR, "unexpected end of file");
617 return -1;
618 }
619 if (crc != state->zstrm.adler) {
620 xz_error(state, LZMA_DATA_ERROR, "incorrect data check");
621 return -1;
622 }
623 if (len != (state->zstrm.total_out & 0xffffffffL)) {
624 xz_error(state, LZMA_DATA_ERROR, "incorrect length check");
625 return -1;
626 }
627 state->strm.avail_in = 0;
628 state->strm.next_in = NULL;
629 state->strm.avail_out = 0;
630 state->strm.next_out = NULL;
631 } else
632 #endif
633 if (strm->avail_in != 0 || !state->eof) {
634 xz_error(state, LZMA_DATA_ERROR, "trailing garbage");
635 return -1;
636 }
637 state->how = LOOK; /* ready for next stream, once have is 0 (leave
638 * state->direct unchanged to remember how) */
639 }
640
641 /* good decompression */
642 return 0;
643 }
644
645 static int
xz_make(xz_statep state)646 xz_make(xz_statep state)
647 {
648 lzma_stream *strm = &(state->strm);
649
650 if (state->how == LOOK) { /* look for lzma / gzip header */
651 if (xz_head(state) == -1)
652 return -1;
653 if (state->have) /* got some data from xz_head() */
654 return 0;
655 }
656 if (state->how == COPY) { /* straight copy */
657 if (xz_load(state, state->out, state->size << 1, &(state->have)) ==
658 -1)
659 return -1;
660 state->next = state->out;
661 } else if (state->how == LZMA || state->how == GZIP) { /* decompress */
662 strm->avail_out = state->size << 1;
663 strm->next_out = state->out;
664 if (xz_decomp(state) == -1)
665 return -1;
666 }
667 return 0;
668 }
669
670 static int
xz_skip(xz_statep state,uint64_t len)671 xz_skip(xz_statep state, uint64_t len)
672 {
673 unsigned n;
674
675 /* skip over len bytes or reach end-of-file, whichever comes first */
676 while (len)
677 /* skip over whatever is in output buffer */
678 if (state->have) {
679 n = (uint64_t) state->have > len ?
680 (unsigned) len : state->have;
681 state->have -= n;
682 state->next += n;
683 state->pos += n;
684 len -= n;
685 }
686
687 /* output buffer empty -- return if we're at the end of the input */
688 else if (state->eof && state->strm.avail_in == 0)
689 break;
690
691 /* need more data to skip -- load up output buffer */
692 else {
693 /* get more output, looking for header if required */
694 if (xz_make(state) == -1)
695 return -1;
696 }
697 return 0;
698 }
699
700 int
__libxml2_xzread(xzFile file,void * buf,unsigned len)701 __libxml2_xzread(xzFile file, void *buf, unsigned len)
702 {
703 unsigned got, n;
704 xz_statep state;
705 lzma_stream *strm;
706
707 /* get internal structure */
708 if (file == NULL)
709 return -1;
710 state = (xz_statep) file;
711 strm = &(state->strm);
712
713 /* check that we're reading and that there's no error */
714 if (state->err != LZMA_OK)
715 return -1;
716
717 /* since an int is returned, make sure len fits in one, otherwise return
718 * with an error (this avoids the flaw in the interface) */
719 if ((int) len < 0) {
720 xz_error(state, LZMA_BUF_ERROR,
721 "requested length does not fit in int");
722 return -1;
723 }
724
725 /* if len is zero, avoid unnecessary operations */
726 if (len == 0)
727 return 0;
728
729 /* process a skip request */
730 if (state->seek) {
731 state->seek = 0;
732 if (xz_skip(state, state->skip) == -1)
733 return -1;
734 }
735
736 /* get len bytes to buf, or less than len if at the end */
737 got = 0;
738 do {
739 /* first just try copying data from the output buffer */
740 if (state->have) {
741 n = state->have > len ? len : state->have;
742 memcpy(buf, state->next, n);
743 state->next += n;
744 state->have -= n;
745 }
746
747 /* output buffer empty -- return if we're at the end of the input */
748 else if (state->eof && strm->avail_in == 0)
749 break;
750
751 /* need output data -- for small len or new stream load up our output
752 * buffer */
753 else if (state->how == LOOK || len < (state->size << 1)) {
754 /* get more output, looking for header if required */
755 if (xz_make(state) == -1)
756 return -1;
757 continue; /* no progress yet -- go back to memcpy() above */
758 /* the copy above assures that we will leave with space in the
759 * output buffer, allowing at least one gzungetc() to succeed */
760 }
761
762 /* large len -- read directly into user buffer */
763 else if (state->how == COPY) { /* read directly */
764 if (xz_load(state, buf, len, &n) == -1)
765 return -1;
766 }
767
768 /* large len -- decompress directly into user buffer */
769 else { /* state->how == LZMA */
770 strm->avail_out = len;
771 strm->next_out = buf;
772 if (xz_decomp(state) == -1)
773 return -1;
774 n = state->have;
775 state->have = 0;
776 }
777
778 /* update progress */
779 len -= n;
780 buf = (char *) buf + n;
781 got += n;
782 state->pos += n;
783 } while (len);
784
785 /* return number of bytes read into user buffer (will fit in int) */
786 return (int) got;
787 }
788
789 int
__libxml2_xzclose(xzFile file)790 __libxml2_xzclose(xzFile file)
791 {
792 int ret;
793 xz_statep state;
794
795 /* get internal structure */
796 if (file == NULL)
797 return LZMA_DATA_ERROR;
798 state = (xz_statep) file;
799
800 /* free memory and close file */
801 if (state->size) {
802 lzma_end(&(state->strm));
803 #ifdef LIBXML_ZLIB_ENABLED
804 if (state->init == 1)
805 inflateEnd(&(state->zstrm));
806 state->init = 0;
807 #endif
808 xmlFree(state->out);
809 xmlFree(state->in);
810 }
811 xmlFree(state->path);
812 if ((state->msg != NULL) && (state->err != LZMA_MEM_ERROR))
813 xmlFree(state->msg);
814 ret = close(state->fd);
815 xmlFree(state);
816 return ret ? ret : LZMA_OK;
817 }
818 #endif /* LIBXML_LZMA_ENABLED */
819