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 #ifdef HAVE_ERRNO_H
15 #include <errno.h>
16 #endif
17
18
19 #ifdef HAVE_SYS_TYPES_H
20 #include <sys/types.h>
21 #endif
22 #ifdef HAVE_SYS_STAT_H
23 #include <sys/stat.h>
24 #endif
25 #ifdef HAVE_FCNTL_H
26 #include <fcntl.h>
27 #endif
28 #ifdef HAVE_UNISTD_H
29 #include <unistd.h>
30 #endif
31 #ifdef HAVE_STDLIB_H
32 #include <stdlib.h>
33 #endif
34 #ifdef LIBXML_ZLIB_ENABLED
35 #include <zlib.h>
36 #endif
37 #ifdef LIBXML_LZMA_ENABLED
38 #include <lzma.h>
39 #endif
40
41 #include "xzlib.h"
42 #include <libxml/xmlmemory.h>
43
44 /* values for xz_state how */
45 #define LOOK 0 /* look for a gzip/lzma header */
46 #define COPY 1 /* copy input directly */
47 #define GZIP 2 /* decompress a gzip stream */
48 #define LZMA 3 /* decompress a lzma stream */
49
50 /* internal lzma file state data structure */
51 typedef struct {
52 int mode; /* see lzma modes above */
53 int fd; /* file descriptor */
54 char *path; /* path or fd for error messages */
55 uint64_t pos; /* current position in uncompressed data */
56 unsigned int size; /* buffer size, zero if not allocated yet */
57 unsigned int want; /* requested buffer size, default is BUFSIZ */
58 unsigned char *in; /* input buffer */
59 unsigned char *out; /* output buffer (double-sized when reading) */
60 unsigned char *next; /* next output data to deliver or write */
61 unsigned int have; /* amount of output data unused at next */
62 int eof; /* true if end of input file reached */
63 uint64_t start; /* where the lzma data started, for rewinding */
64 uint64_t raw; /* where the raw data started, for seeking */
65 int how; /* 0: get header, 1: copy, 2: decompress */
66 int direct; /* true if last read direct, false if lzma */
67 /* seek request */
68 uint64_t skip; /* amount to skip (already rewound if backwards) */
69 int seek; /* true if seek request pending */
70 /* error information */
71 int err; /* error code */
72 char *msg; /* error message */
73 /* lzma stream */
74 int init; /* is the inflate stream initialized */
75 lzma_stream strm; /* stream structure in-place (not a pointer) */
76 char padding1[32]; /* padding allowing to cope with possible
77 extensions of above structure without
78 too much side effect */
79 #ifdef LIBXML_ZLIB_ENABLED
80 /* zlib inflate or deflate stream */
81 z_stream zstrm; /* stream structure in-place (not a pointer) */
82 #endif
83 char padding2[32]; /* padding allowing to cope with possible
84 extensions of above structure without
85 too much side effect */
86 } xz_state, *xz_statep;
87
88 static void
xz_error(xz_statep state,int err,const char * msg)89 xz_error(xz_statep state, int err, const char *msg)
90 {
91 /* free previously allocated message and clear */
92 if (state->msg != NULL) {
93 if (state->err != LZMA_MEM_ERROR)
94 xmlFree(state->msg);
95 state->msg = NULL;
96 }
97
98 /* set error code, and if no message, then done */
99 state->err = err;
100 if (msg == NULL)
101 return;
102
103 /* for an out of memory error, save as static string */
104 if (err == LZMA_MEM_ERROR) {
105 state->msg = (char *) msg;
106 return;
107 }
108
109 /* construct error message with path */
110 if ((state->msg =
111 xmlMalloc(strlen(state->path) + strlen(msg) + 3)) == NULL) {
112 state->err = LZMA_MEM_ERROR;
113 state->msg = (char *) "out of memory";
114 return;
115 }
116 strcpy(state->msg, state->path);
117 strcat(state->msg, ": ");
118 strcat(state->msg, msg);
119 return;
120 }
121
122 static void
xz_reset(xz_statep state)123 xz_reset(xz_statep state)
124 {
125 state->have = 0; /* no output data available */
126 state->eof = 0; /* not at end of file */
127 state->how = LOOK; /* look for gzip header */
128 state->direct = 1; /* default for empty file */
129 state->seek = 0; /* no seek request pending */
130 xz_error(state, LZMA_OK, NULL); /* clear error */
131 state->pos = 0; /* no uncompressed data yet */
132 state->strm.avail_in = 0; /* no input data yet */
133 #ifdef LIBXML_ZLIB_ENABLED
134 state->zstrm.avail_in = 0; /* no input data yet */
135 #endif
136 }
137
138 static xzFile
xz_open(const char * path,int fd,const char * mode ATTRIBUTE_UNUSED)139 xz_open(const char *path, int fd, const char *mode ATTRIBUTE_UNUSED)
140 {
141 xz_statep state;
142
143 /* allocate xzFile structure to return */
144 state = xmlMalloc(sizeof(xz_state));
145 if (state == NULL)
146 return NULL;
147 state->size = 0; /* no buffers allocated yet */
148 state->want = BUFSIZ; /* requested buffer size */
149 state->msg = NULL; /* no error message yet */
150 state->init = 0; /* initialization of zlib data */
151
152 /* save the path name for error messages */
153 state->path = xmlMalloc(strlen(path) + 1);
154 if (state->path == NULL) {
155 xmlFree(state);
156 return NULL;
157 }
158 strcpy(state->path, path);
159
160 /* open the file with the appropriate mode (or just use fd) */
161 state->fd = fd != -1 ? fd : open(path,
162 #ifdef O_LARGEFILE
163 O_LARGEFILE |
164 #endif
165 #ifdef O_BINARY
166 O_BINARY |
167 #endif
168 O_RDONLY, 0666);
169 if (state->fd == -1) {
170 xmlFree(state->path);
171 xmlFree(state);
172 return NULL;
173 }
174
175 /* save the current position for rewinding (only if reading) */
176 state->start = lseek(state->fd, 0, SEEK_CUR);
177 if (state->start == (uint64_t) - 1)
178 state->start = 0;
179
180 /* initialize stream */
181 xz_reset(state);
182
183 /* return stream */
184 return (xzFile) state;
185 }
186
187 static int
xz_compressed(xzFile f)188 xz_compressed(xzFile f) {
189 xz_statep state;
190
191 if (f == NULL)
192 return(-1);
193 state = (xz_statep) f;
194 if (state->init <= 0)
195 return(-1);
196
197 switch (state->how) {
198 case COPY:
199 return(0);
200 case GZIP:
201 case LZMA:
202 return(1);
203 }
204 return(-1);
205 }
206
207 xzFile
__libxml2_xzopen(const char * path,const char * mode)208 __libxml2_xzopen(const char *path, const char *mode)
209 {
210 return xz_open(path, -1, mode);
211 }
212
213 int
__libxml2_xzcompressed(xzFile f)214 __libxml2_xzcompressed(xzFile f) {
215 return xz_compressed(f);
216 }
217
218 xzFile
__libxml2_xzdopen(int fd,const char * mode)219 __libxml2_xzdopen(int fd, const char *mode)
220 {
221 char *path; /* identifier for error messages */
222 xzFile xz;
223
224 if (fd == -1 || (path = xmlMalloc(7 + 3 * sizeof(int))) == NULL)
225 return NULL;
226 sprintf(path, "<fd:%d>", fd); /* for debugging */
227 xz = xz_open(path, fd, mode);
228 xmlFree(path);
229 return xz;
230 }
231
232 static int
xz_load(xz_statep state,unsigned char * buf,unsigned int len,unsigned int * have)233 xz_load(xz_statep state, unsigned char *buf, unsigned int len,
234 unsigned int *have)
235 {
236 int ret;
237
238 *have = 0;
239 do {
240 ret = read(state->fd, buf + *have, len - *have);
241 if (ret <= 0)
242 break;
243 *have += ret;
244 } while (*have < len);
245 if (ret < 0) {
246 xz_error(state, -1, strerror(errno));
247 return -1;
248 }
249 if (ret == 0)
250 state->eof = 1;
251 return 0;
252 }
253
254 static int
xz_avail(xz_statep state)255 xz_avail(xz_statep state)
256 {
257 lzma_stream *strm = &(state->strm);
258
259 if (state->err != LZMA_OK)
260 return -1;
261 if (state->eof == 0) {
262 /* avail_in is size_t, which is not necessary sizeof(unsigned) */
263 unsigned tmp = strm->avail_in;
264
265 if (xz_load(state, state->in, state->size, &tmp) == -1) {
266 strm->avail_in = tmp;
267 return -1;
268 }
269 strm->avail_in = tmp;
270 strm->next_in = state->in;
271 }
272 return 0;
273 }
274
275 #ifdef LIBXML_ZLIB_ENABLED
276 static int
xz_avail_zstrm(xz_statep state)277 xz_avail_zstrm(xz_statep state)
278 {
279 int ret;
280 state->strm.avail_in = state->zstrm.avail_in;
281 state->strm.next_in = state->zstrm.next_in;
282 ret = xz_avail(state);
283 state->zstrm.avail_in = (uInt) state->strm.avail_in;
284 state->zstrm.next_in = (Bytef *) state->strm.next_in;
285 return ret;
286 }
287 #endif
288
289 static int
is_format_xz(xz_statep state)290 is_format_xz(xz_statep state)
291 {
292 lzma_stream *strm = &(state->strm);
293
294 return strm->avail_in >= 6 && memcmp(state->in, "\3757zXZ", 6) == 0;
295 }
296
297 static int
is_format_lzma(xz_statep state)298 is_format_lzma(xz_statep state)
299 {
300 lzma_stream *strm = &(state->strm);
301
302 lzma_filter filter;
303 lzma_options_lzma *opt;
304 uint32_t dict_size;
305 uint64_t uncompressed_size;
306 size_t i;
307
308 if (strm->avail_in < 13)
309 return 0;
310
311 filter.id = LZMA_FILTER_LZMA1;
312 if (lzma_properties_decode(&filter, NULL, state->in, 5) != LZMA_OK)
313 return 0;
314
315 opt = filter.options;
316 dict_size = opt->dict_size;
317 free(opt); /* we can't use xmlFree on a string returned by zlib */
318
319 /* A hack to ditch tons of false positives: We allow only dictionary
320 * sizes that are 2^n or 2^n + 2^(n-1) or UINT32_MAX. LZMA_Alone
321 * created only files with 2^n, but accepts any dictionary size.
322 * If someone complains, this will be reconsidered.
323 */
324 if (dict_size != UINT32_MAX) {
325 uint32_t d = dict_size - 1;
326
327 d |= d >> 2;
328 d |= d >> 3;
329 d |= d >> 4;
330 d |= d >> 8;
331 d |= d >> 16;
332 ++d;
333 if (d != dict_size || dict_size == 0)
334 return 0;
335 }
336
337 /* Another hack to ditch false positives: Assume that if the
338 * uncompressed size is known, it must be less than 256 GiB.
339 * Again, if someone complains, this will be reconsidered.
340 */
341 uncompressed_size = 0;
342 for (i = 0; i < 8; ++i)
343 uncompressed_size |= (uint64_t) (state->in[5 + i]) << (i * 8);
344
345 if (uncompressed_size != UINT64_MAX
346 && uncompressed_size > (UINT64_C(1) << 38))
347 return 0;
348
349 return 1;
350 }
351
352 #ifdef LIBXML_ZLIB_ENABLED
353
354 /* Get next byte from input, or -1 if end or error. */
355 #define NEXT() ((strm->avail_in == 0 && xz_avail(state) == -1) ? -1 : \
356 (strm->avail_in == 0 ? -1 : \
357 (strm->avail_in--, *(strm->next_in)++)))
358 /* Same thing, but from zstrm */
359 #define NEXTZ() ((strm->avail_in == 0 && xz_avail_zstrm(state) == -1) ? -1 : \
360 (strm->avail_in == 0 ? -1 : \
361 (strm->avail_in--, *(strm->next_in)++)))
362
363 /* Get a four-byte little-endian integer and return 0 on success and the value
364 in *ret. Otherwise -1 is returned and *ret is not modified. */
365 static int
gz_next4(xz_statep state,unsigned long * ret)366 gz_next4(xz_statep state, unsigned long *ret)
367 {
368 int ch;
369 unsigned long val;
370 z_streamp strm = &(state->zstrm);
371
372 val = NEXTZ();
373 val += (unsigned) NEXTZ() << 8;
374 val += (unsigned long) NEXTZ() << 16;
375 ch = NEXTZ();
376 if (ch == -1)
377 return -1;
378 val += (unsigned long) ch << 24;
379 *ret = val;
380 return 0;
381 }
382 #endif
383
384 static int
xz_head(xz_statep state)385 xz_head(xz_statep state)
386 {
387 lzma_stream *strm = &(state->strm);
388 lzma_stream init = LZMA_STREAM_INIT;
389 int flags;
390 unsigned len;
391
392 /* Avoid unused variable warning if features are disabled. */
393 (void) flags;
394 (void) len;
395
396 /* allocate read buffers and inflate memory */
397 if (state->size == 0) {
398 /* allocate buffers */
399 state->in = xmlMalloc(state->want);
400 state->out = xmlMalloc(state->want << 1);
401 if (state->in == NULL || state->out == NULL) {
402 if (state->out != NULL)
403 xmlFree(state->out);
404 if (state->in != NULL)
405 xmlFree(state->in);
406 xz_error(state, LZMA_MEM_ERROR, "out of memory");
407 return -1;
408 }
409 state->size = state->want;
410
411 /* allocate decoder memory */
412 state->strm = init;
413 state->strm.avail_in = 0;
414 state->strm.next_in = NULL;
415 if (lzma_auto_decoder(&state->strm, 100000000, 0) != LZMA_OK) {
416 xmlFree(state->out);
417 xmlFree(state->in);
418 state->size = 0;
419 xz_error(state, LZMA_MEM_ERROR, "out of memory");
420 return -1;
421 }
422 #ifdef LIBXML_ZLIB_ENABLED
423 /* allocate inflate memory */
424 state->zstrm.zalloc = Z_NULL;
425 state->zstrm.zfree = Z_NULL;
426 state->zstrm.opaque = Z_NULL;
427 state->zstrm.avail_in = 0;
428 state->zstrm.next_in = Z_NULL;
429 if (state->init == 0) {
430 if (inflateInit2(&(state->zstrm), -15) != Z_OK) {/* raw inflate */
431 xmlFree(state->out);
432 xmlFree(state->in);
433 state->size = 0;
434 xz_error(state, LZMA_MEM_ERROR, "out of memory");
435 return -1;
436 }
437 state->init = 1;
438 }
439 #endif
440 }
441
442 /* get some data in the input buffer */
443 if (strm->avail_in == 0) {
444 if (xz_avail(state) == -1)
445 return -1;
446 if (strm->avail_in == 0)
447 return 0;
448 }
449
450 /* look for the xz magic header bytes */
451 if (is_format_xz(state) || is_format_lzma(state)) {
452 state->how = LZMA;
453 state->direct = 0;
454 return 0;
455 }
456 #ifdef LIBXML_ZLIB_ENABLED
457 /* look for the gzip magic header bytes 31 and 139 */
458 if (strm->next_in[0] == 31) {
459 strm->avail_in--;
460 strm->next_in++;
461 if (strm->avail_in == 0 && xz_avail(state) == -1)
462 return -1;
463 if (strm->avail_in && strm->next_in[0] == 139) {
464 /* we have a gzip header, woo hoo! */
465 strm->avail_in--;
466 strm->next_in++;
467
468 /* skip rest of header */
469 if (NEXT() != 8) { /* compression method */
470 xz_error(state, LZMA_DATA_ERROR,
471 "unknown compression method");
472 return -1;
473 }
474 flags = NEXT();
475 if (flags & 0xe0) { /* reserved flag bits */
476 xz_error(state, LZMA_DATA_ERROR,
477 "unknown header flags set");
478 return -1;
479 }
480 NEXT(); /* modification time */
481 NEXT();
482 NEXT();
483 NEXT();
484 NEXT(); /* extra flags */
485 NEXT(); /* operating system */
486 if (flags & 4) { /* extra field */
487 len = (unsigned) NEXT();
488 len += (unsigned) NEXT() << 8;
489 while (len--)
490 if (NEXT() < 0)
491 break;
492 }
493 if (flags & 8) /* file name */
494 while (NEXT() > 0) ;
495 if (flags & 16) /* comment */
496 while (NEXT() > 0) ;
497 if (flags & 2) { /* header crc */
498 NEXT();
499 NEXT();
500 }
501 /* an unexpected end of file is not checked for here -- it will be
502 * noticed on the first request for uncompressed data */
503
504 /* set up for decompression */
505 inflateReset(&state->zstrm);
506 state->zstrm.adler = crc32(0L, Z_NULL, 0);
507 state->how = GZIP;
508 state->direct = 0;
509 return 0;
510 } else {
511 /* not a gzip file -- save first byte (31) and fall to raw i/o */
512 state->out[0] = 31;
513 state->have = 1;
514 }
515 }
516 #endif
517
518 /* doing raw i/o, save start of raw data for seeking, copy any leftover
519 * input to output -- this assumes that the output buffer is larger than
520 * the input buffer, which also assures space for gzungetc() */
521 state->raw = state->pos;
522 state->next = state->out;
523 if (strm->avail_in) {
524 memcpy(state->next + state->have, strm->next_in, strm->avail_in);
525 state->have += strm->avail_in;
526 strm->avail_in = 0;
527 }
528 state->how = COPY;
529 state->direct = 1;
530 return 0;
531 }
532
533 static int
xz_decomp(xz_statep state)534 xz_decomp(xz_statep state)
535 {
536 int ret;
537 unsigned had;
538 unsigned long crc, len;
539 lzma_stream *strm = &(state->strm);
540
541 lzma_action action = LZMA_RUN;
542
543 /* Avoid unused variable warning if features are disabled. */
544 (void) crc;
545 (void) len;
546
547 /* fill output buffer up to end of deflate stream */
548 had = strm->avail_out;
549 do {
550 /* get more input for inflate() */
551 if (strm->avail_in == 0 && xz_avail(state) == -1)
552 return -1;
553 if (strm->avail_in == 0) {
554 xz_error(state, LZMA_DATA_ERROR, "unexpected end of file");
555 return -1;
556 }
557 if (state->eof)
558 action = LZMA_FINISH;
559
560 /* decompress and handle errors */
561 #ifdef LIBXML_ZLIB_ENABLED
562 if (state->how == GZIP) {
563 state->zstrm.avail_in = (uInt) state->strm.avail_in;
564 state->zstrm.next_in = (Bytef *) state->strm.next_in;
565 state->zstrm.avail_out = (uInt) state->strm.avail_out;
566 state->zstrm.next_out = (Bytef *) state->strm.next_out;
567 ret = inflate(&state->zstrm, Z_NO_FLUSH);
568 if (ret == Z_STREAM_ERROR || ret == Z_NEED_DICT) {
569 xz_error(state, Z_STREAM_ERROR,
570 "internal error: inflate stream corrupt");
571 return -1;
572 }
573 /*
574 * FIXME: Remapping a couple of error codes and falling through
575 * to the LZMA error handling looks fragile.
576 */
577 if (ret == Z_MEM_ERROR)
578 ret = LZMA_MEM_ERROR;
579 if (ret == Z_DATA_ERROR)
580 ret = LZMA_DATA_ERROR;
581 if (ret == Z_STREAM_END)
582 ret = LZMA_STREAM_END;
583 state->strm.avail_in = state->zstrm.avail_in;
584 state->strm.next_in = state->zstrm.next_in;
585 state->strm.avail_out = state->zstrm.avail_out;
586 state->strm.next_out = state->zstrm.next_out;
587 } else /* state->how == LZMA */
588 #endif
589 ret = lzma_code(strm, action);
590 if (ret == LZMA_MEM_ERROR) {
591 xz_error(state, LZMA_MEM_ERROR, "out of memory");
592 return -1;
593 }
594 if (ret == LZMA_DATA_ERROR) {
595 xz_error(state, LZMA_DATA_ERROR, "compressed data error");
596 return -1;
597 }
598 if (ret == LZMA_PROG_ERROR) {
599 xz_error(state, LZMA_PROG_ERROR, "compression error");
600 return -1;
601 }
602 if ((state->how != GZIP) &&
603 (ret != LZMA_OK) && (ret != LZMA_STREAM_END)) {
604 xz_error(state, ret, "lzma error");
605 return -1;
606 }
607 } while (strm->avail_out && ret != LZMA_STREAM_END);
608
609 /* update available output and crc check value */
610 state->have = had - strm->avail_out;
611 state->next = strm->next_out - state->have;
612 #ifdef LIBXML_ZLIB_ENABLED
613 state->zstrm.adler =
614 crc32(state->zstrm.adler, state->next, state->have);
615 #endif
616
617 if (ret == LZMA_STREAM_END) {
618 #ifdef LIBXML_ZLIB_ENABLED
619 if (state->how == GZIP) {
620 if (gz_next4(state, &crc) == -1 || gz_next4(state, &len) == -1) {
621 xz_error(state, LZMA_DATA_ERROR, "unexpected end of file");
622 return -1;
623 }
624 if (crc != state->zstrm.adler) {
625 xz_error(state, LZMA_DATA_ERROR, "incorrect data check");
626 return -1;
627 }
628 if (len != (state->zstrm.total_out & 0xffffffffL)) {
629 xz_error(state, LZMA_DATA_ERROR, "incorrect length check");
630 return -1;
631 }
632 state->strm.avail_in = 0;
633 state->strm.next_in = NULL;
634 state->strm.avail_out = 0;
635 state->strm.next_out = NULL;
636 } else
637 #endif
638 if (strm->avail_in != 0 || !state->eof) {
639 xz_error(state, LZMA_DATA_ERROR, "trailing garbage");
640 return -1;
641 }
642 state->how = LOOK; /* ready for next stream, once have is 0 (leave
643 * state->direct unchanged to remember how) */
644 }
645
646 /* good decompression */
647 return 0;
648 }
649
650 static int
xz_make(xz_statep state)651 xz_make(xz_statep state)
652 {
653 lzma_stream *strm = &(state->strm);
654
655 if (state->how == LOOK) { /* look for lzma / gzip header */
656 if (xz_head(state) == -1)
657 return -1;
658 if (state->have) /* got some data from xz_head() */
659 return 0;
660 }
661 if (state->how == COPY) { /* straight copy */
662 if (xz_load(state, state->out, state->size << 1, &(state->have)) ==
663 -1)
664 return -1;
665 state->next = state->out;
666 } else if (state->how == LZMA || state->how == GZIP) { /* decompress */
667 strm->avail_out = state->size << 1;
668 strm->next_out = state->out;
669 if (xz_decomp(state) == -1)
670 return -1;
671 }
672 return 0;
673 }
674
675 static int
xz_skip(xz_statep state,uint64_t len)676 xz_skip(xz_statep state, uint64_t len)
677 {
678 unsigned n;
679
680 /* skip over len bytes or reach end-of-file, whichever comes first */
681 while (len)
682 /* skip over whatever is in output buffer */
683 if (state->have) {
684 n = (uint64_t) state->have > len ?
685 (unsigned) len : state->have;
686 state->have -= n;
687 state->next += n;
688 state->pos += n;
689 len -= n;
690 }
691
692 /* output buffer empty -- return if we're at the end of the input */
693 else if (state->eof && state->strm.avail_in == 0)
694 break;
695
696 /* need more data to skip -- load up output buffer */
697 else {
698 /* get more output, looking for header if required */
699 if (xz_make(state) == -1)
700 return -1;
701 }
702 return 0;
703 }
704
705 int
__libxml2_xzread(xzFile file,void * buf,unsigned len)706 __libxml2_xzread(xzFile file, void *buf, unsigned len)
707 {
708 unsigned got, n;
709 xz_statep state;
710 lzma_stream *strm;
711
712 /* get internal structure */
713 if (file == NULL)
714 return -1;
715 state = (xz_statep) file;
716 strm = &(state->strm);
717
718 /* check that we're reading and that there's no error */
719 if (state->err != LZMA_OK)
720 return -1;
721
722 /* since an int is returned, make sure len fits in one, otherwise return
723 * with an error (this avoids the flaw in the interface) */
724 if ((int) len < 0) {
725 xz_error(state, LZMA_BUF_ERROR,
726 "requested length does not fit in int");
727 return -1;
728 }
729
730 /* if len is zero, avoid unnecessary operations */
731 if (len == 0)
732 return 0;
733
734 /* process a skip request */
735 if (state->seek) {
736 state->seek = 0;
737 if (xz_skip(state, state->skip) == -1)
738 return -1;
739 }
740
741 /* get len bytes to buf, or less than len if at the end */
742 got = 0;
743 do {
744 /* first just try copying data from the output buffer */
745 if (state->have) {
746 n = state->have > len ? len : state->have;
747 memcpy(buf, state->next, n);
748 state->next += n;
749 state->have -= n;
750 }
751
752 /* output buffer empty -- return if we're at the end of the input */
753 else if (state->eof && strm->avail_in == 0)
754 break;
755
756 /* need output data -- for small len or new stream load up our output
757 * buffer */
758 else if (state->how == LOOK || len < (state->size << 1)) {
759 /* get more output, looking for header if required */
760 if (xz_make(state) == -1)
761 return -1;
762 continue; /* no progress yet -- go back to memcpy() above */
763 /* the copy above assures that we will leave with space in the
764 * output buffer, allowing at least one gzungetc() to succeed */
765 }
766
767 /* large len -- read directly into user buffer */
768 else if (state->how == COPY) { /* read directly */
769 if (xz_load(state, buf, len, &n) == -1)
770 return -1;
771 }
772
773 /* large len -- decompress directly into user buffer */
774 else { /* state->how == LZMA */
775 strm->avail_out = len;
776 strm->next_out = buf;
777 if (xz_decomp(state) == -1)
778 return -1;
779 n = state->have;
780 state->have = 0;
781 }
782
783 /* update progress */
784 len -= n;
785 buf = (char *) buf + n;
786 got += n;
787 state->pos += n;
788 } while (len);
789
790 /* return number of bytes read into user buffer (will fit in int) */
791 return (int) got;
792 }
793
794 int
__libxml2_xzclose(xzFile file)795 __libxml2_xzclose(xzFile file)
796 {
797 int ret;
798 xz_statep state;
799
800 /* get internal structure */
801 if (file == NULL)
802 return LZMA_DATA_ERROR;
803 state = (xz_statep) file;
804
805 /* free memory and close file */
806 if (state->size) {
807 lzma_end(&(state->strm));
808 #ifdef LIBXML_ZLIB_ENABLED
809 if (state->init == 1)
810 inflateEnd(&(state->zstrm));
811 state->init = 0;
812 #endif
813 xmlFree(state->out);
814 xmlFree(state->in);
815 }
816 xmlFree(state->path);
817 if ((state->msg != NULL) && (state->err != LZMA_MEM_ERROR))
818 xmlFree(state->msg);
819 ret = close(state->fd);
820 xmlFree(state);
821 return ret ? ret : LZMA_OK;
822 }
823 #endif /* LIBXML_LZMA_ENABLED */
824