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