/
physfs_archiver_zip.c
1710 lines (1380 loc) · 53 KB
1
2
3
/*
* ZIP support routines for PhysicsFS.
*
4
* Please see the file LICENSE.txt in the source's root directory.
5
*
6
7
* This file written by Ryan C. Gordon, with some peeking at "unzip.c"
* by Gilles Vollant.
8
9
*/
10
11
12
13
#define __PHYSICSFS_INTERNAL__
#include "physfs_internal.h"
#if PHYSFS_SUPPORTS_ZIP
14
15
#include <errno.h>
16
#include <time.h>
17
18
19
#include "physfs_miniz.h"
20
/*
21
22
* A buffer of ZIP_READBUFSIZE is allocated for each compressed file opened,
* and is freed when you close the file; compressed data is read into
23
24
25
26
27
28
29
30
31
32
33
* this buffer, and then is decompressed into the buffer passed to
* PHYSFS_read().
*
* Uncompressed entries in a zipfile do not allocate this buffer; they just
* read data directly into the buffer passed to PHYSFS_read().
*
* Depending on your speed and memory requirements, you should tweak this
* value.
*/
#define ZIP_READBUFSIZE (16 * 1024)
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
/*
* Entries are "unresolved" until they are first opened. At that time,
* local file headers parsed/validated, data offsets will be updated to look
* at the actual file data instead of the header, and symlinks will be
* followed and optimized. This means that we don't seek and read around the
* archive until forced to do so, and after the first time, we had to do
* less reading and parsing, which is very CD-ROM friendly.
*/
typedef enum
{
ZIP_UNRESOLVED_FILE,
ZIP_UNRESOLVED_SYMLINK,
ZIP_RESOLVING,
ZIP_RESOLVED,
49
ZIP_DIRECTORY,
50
ZIP_BROKEN_FILE,
51
ZIP_BROKEN_SYMLINK
52
53
54
} ZipResolveType;
55
56
57
/*
* One ZIPentry is kept for each file in an open ZIP archive.
*/
58
typedef struct _ZIPentry
59
{
60
__PHYSFS_DirTreeEntry tree; /* manages directory tree */
61
62
struct _ZIPentry *symlink; /* NULL or file we symlink to */
ZipResolveType resolved; /* Have we resolved file/symlink? */
63
PHYSFS_uint64 offset; /* offset of data in archive */
64
65
PHYSFS_uint16 version; /* version made by */
PHYSFS_uint16 version_needed; /* version needed to extract */
66
PHYSFS_uint16 general_bits; /* general purpose bits */
67
68
PHYSFS_uint16 compression_method; /* compression method */
PHYSFS_uint32 crc; /* crc-32 */
69
70
PHYSFS_uint64 compressed_size; /* compressed size */
PHYSFS_uint64 uncompressed_size; /* uncompressed size */
71
PHYSFS_sint64 last_mod_time; /* last file mod time */
72
PHYSFS_uint32 dos_mod_time; /* original MS-DOS style mod time */
73
74
} ZIPentry;
75
76
77
/*
* One ZIPinfo is kept for each open ZIP archive.
*/
78
79
typedef struct
{
80
__PHYSFS_DirTree tree; /* manages directory tree. */
81
82
83
PHYSFS_Io *io; /* the i/o interface for this archive. */
int zip64; /* non-zero if this is a Zip64 archive. */
int has_crypto; /* non-zero if any entry uses encryption. */
84
85
} ZIPinfo;
86
87
88
/*
* One ZIPfileinfo is kept for each open file in a ZIP archive.
*/
89
90
typedef struct
{
91
ZIPentry *entry; /* Info on file. */
92
PHYSFS_Io *io; /* physical file handle. */
93
94
95
PHYSFS_uint32 compressed_position; /* offset in compressed data. */
PHYSFS_uint32 uncompressed_position; /* tell() position. */
PHYSFS_uint8 *buffer; /* decompression buffer. */
96
97
PHYSFS_uint32 crypto_keys[3]; /* for "traditional" crypto. */
PHYSFS_uint32 initial_crypto_keys[3]; /* for "traditional" crypto. */
98
z_stream stream; /* zlib stream state. */
99
100
101
} ZIPfileinfo;
102
/* Magic numbers... */
103
104
105
106
107
108
#define ZIP_LOCAL_FILE_SIG 0x04034b50
#define ZIP_CENTRAL_DIR_SIG 0x02014b50
#define ZIP_END_OF_CENTRAL_DIR_SIG 0x06054b50
#define ZIP64_END_OF_CENTRAL_DIR_SIG 0x06064b50
#define ZIP64_END_OF_CENTRAL_DIRECTORY_LOCATOR_SIG 0x07064b50
#define ZIP64_EXTENDED_INFO_EXTRA_FIELD_SIG 0x0001
109
110
111
112
113
114
/* compression methods... */
#define COMPMETH_NONE 0
/* ...and others... */
115
116
117
#define UNIX_FILETYPE_MASK 0170000
#define UNIX_FILETYPE_SYMLINK 0120000
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
#define ZIP_GENERAL_BITS_TRADITIONAL_CRYPTO (1 << 0)
#define ZIP_GENERAL_BITS_IGNORE_LOCAL_HEADER (1 << 3)
/* support for "traditional" PKWARE encryption. */
static int zip_entry_is_tradional_crypto(const ZIPentry *entry)
{
return (entry->general_bits & ZIP_GENERAL_BITS_TRADITIONAL_CRYPTO) != 0;
} /* zip_entry_is_traditional_crypto */
static int zip_entry_ignore_local_header(const ZIPentry *entry)
{
return (entry->general_bits & ZIP_GENERAL_BITS_IGNORE_LOCAL_HEADER) != 0;
} /* zip_entry_is_traditional_crypto */
static PHYSFS_uint32 zip_crypto_crc32(const PHYSFS_uint32 crc, const PHYSFS_uint8 val)
{
int i;
PHYSFS_uint32 xorval = (crc ^ ((PHYSFS_uint32) val)) & 0xFF;
for (i = 0; i < 8; i++)
xorval = ((xorval & 1) ? (0xEDB88320 ^ (xorval >> 1)) : (xorval >> 1));
return xorval ^ (crc >> 8);
} /* zip_crc32 */
static void zip_update_crypto_keys(PHYSFS_uint32 *keys, const PHYSFS_uint8 val)
{
keys[0] = zip_crypto_crc32(keys[0], val);
keys[1] = keys[1] + (keys[0] & 0x000000FF);
keys[1] = (keys[1] * 134775813) + 1;
keys[2] = zip_crypto_crc32(keys[2], (PHYSFS_uint8) ((keys[1] >> 24) & 0xFF));
} /* zip_update_crypto_keys */
static PHYSFS_uint8 zip_decrypt_byte(const PHYSFS_uint32 *keys)
{
const PHYSFS_uint16 tmp = keys[2] | 2;
return (PHYSFS_uint8) ((tmp * (tmp ^ 1)) >> 8);
} /* zip_decrypt_byte */
static PHYSFS_sint64 zip_read_decrypt(ZIPfileinfo *finfo, void *buf, PHYSFS_uint64 len)
{
PHYSFS_Io *io = finfo->io;
const PHYSFS_sint64 br = io->read(io, buf, len);
/* Decompression the new data if necessary. */
if (zip_entry_is_tradional_crypto(finfo->entry) && (br > 0))
{
PHYSFS_uint32 *keys = finfo->crypto_keys;
PHYSFS_uint8 *ptr = (PHYSFS_uint8 *) buf;
PHYSFS_sint64 i;
for (i = 0; i < br; i++, ptr++)
{
const PHYSFS_uint8 ch = *ptr ^ zip_decrypt_byte(keys);
zip_update_crypto_keys(keys, ch);
*ptr = ch;
} /* for */
} /* if */
return br;
} /* zip_read_decrypt */
static int zip_prep_crypto_keys(ZIPfileinfo *finfo, const PHYSFS_uint8 *crypto_header, const PHYSFS_uint8 *password)
{
/* It doesn't appear to be documented in PKWare's APPNOTE.TXT, but you
need to use a different byte in the header to verify the password
if general purpose bit 3 is set. Discovered this from Info-Zip.
That's what the (verifier) value is doing, below. */
PHYSFS_uint32 *keys = finfo->crypto_keys;
const ZIPentry *entry = finfo->entry;
const int usedate = zip_entry_ignore_local_header(entry);
const PHYSFS_uint8 verifier = (PHYSFS_uint8) ((usedate ? (entry->dos_mod_time >> 8) : (entry->crc >> 24)) & 0xFF);
PHYSFS_uint8 finalbyte = 0;
int i = 0;
/* initialize vector with defaults, then password, then header. */
keys[0] = 305419896;
keys[1] = 591751049;
keys[2] = 878082192;
while (*password)
zip_update_crypto_keys(keys, *(password++));
for (i = 0; i < 12; i++)
{
const PHYSFS_uint8 c = crypto_header[i] ^ zip_decrypt_byte(keys);
zip_update_crypto_keys(keys, c);
finalbyte = c;
} /* for */
/* you have a 1/256 chance of passing this test incorrectly. :/ */
if (finalbyte != verifier)
208
BAIL(PHYSFS_ERR_BAD_PASSWORD, 0);
209
210
211
212
213
214
/* save the initial vector for seeking purposes. Not secure!! */
memcpy(finfo->initial_crypto_keys, finfo->crypto_keys, 12);
return 1;
} /* zip_prep_crypto_keys */
215
216
217
218
219
220
/*
* Bridge physfs allocation functions to zlib's format...
*/
static voidpf zlibPhysfsAlloc(voidpf opaque, uInt items, uInt size)
{
221
return ((PHYSFS_Allocator *) opaque)->Malloc(items * size);
222
223
224
225
226
227
228
} /* zlibPhysfsAlloc */
/*
* Bridge physfs allocation functions to zlib's format...
*/
static void zlibPhysfsFree(voidpf opaque, voidpf address)
{
229
((PHYSFS_Allocator *) opaque)->Free(address);
230
231
232
233
234
235
236
237
238
239
240
} /* zlibPhysfsFree */
/*
* Construct a new z_stream to a sane state.
*/
static void initializeZStream(z_stream *pstr)
{
memset(pstr, '\0', sizeof (z_stream));
pstr->zalloc = zlibPhysfsAlloc;
pstr->zfree = zlibPhysfsFree;
241
pstr->opaque = &allocator;
242
243
244
} /* initializeZStream */
245
static PHYSFS_ErrorCode zlib_error_code(int rc)
246
247
248
{
switch (rc)
{
249
250
251
252
253
case Z_OK: return PHYSFS_ERR_OK; /* not an error. */
case Z_STREAM_END: return PHYSFS_ERR_OK; /* not an error. */
case Z_ERRNO: return PHYSFS_ERR_IO;
case Z_MEM_ERROR: return PHYSFS_ERR_OUT_OF_MEMORY;
default: return PHYSFS_ERR_CORRUPT;
254
} /* switch */
255
256
} /* zlib_error_string */
257
258
259
260
/*
* Wrap all zlib calls in this, so the physfs error state is set appropriately.
*/
261
static int zlib_err(const int rc)
262
{
263
PHYSFS_setErrorCode(zlib_error_code(rc));
264
return rc;
265
266
} /* zlib_err */
267
268
269
270
271
272
/*
* Read an unsigned 64-bit int and swap to native byte order.
*/
static int readui64(PHYSFS_Io *io, PHYSFS_uint64 *val)
{
PHYSFS_uint64 v;
273
BAIL_IF_ERRPASS(!__PHYSFS_readAll(io, &v, sizeof (v)), 0);
274
275
276
277
*val = PHYSFS_swapULE64(v);
return 1;
} /* readui64 */
278
279
280
/*
* Read an unsigned 32-bit int and swap to native byte order.
*/
281
static int readui32(PHYSFS_Io *io, PHYSFS_uint32 *val)
282
283
{
PHYSFS_uint32 v;
284
BAIL_IF_ERRPASS(!__PHYSFS_readAll(io, &v, sizeof (v)), 0);
285
*val = PHYSFS_swapULE32(v);
286
return 1;
287
288
289
290
291
292
} /* readui32 */
/*
* Read an unsigned 16-bit int and swap to native byte order.
*/
293
static int readui16(PHYSFS_Io *io, PHYSFS_uint16 *val)
294
295
{
PHYSFS_uint16 v;
296
BAIL_IF_ERRPASS(!__PHYSFS_readAll(io, &v, sizeof (v)), 0);
297
*val = PHYSFS_swapULE16(v);
298
return 1;
299
300
301
} /* readui16 */
302
static PHYSFS_sint64 ZIP_read(PHYSFS_Io *_io, void *buf, PHYSFS_uint64 len)
303
{
304
ZIPfileinfo *finfo = (ZIPfileinfo *) _io->opaque;
305
306
ZIPentry *entry = finfo->entry;
PHYSFS_sint64 retval = 0;
307
PHYSFS_sint64 maxread = (PHYSFS_sint64) len;
308
309
310
311
PHYSFS_sint64 avail = entry->uncompressed_size -
finfo->uncompressed_position;
if (avail < maxread)
312
maxread = avail;
313
314
BAIL_IF_ERRPASS(maxread == 0, 0); /* quick rejection. */
315
316
if (entry->compression_method == COMPMETH_NONE)
317
retval = zip_read_decrypt(finfo, buf, maxread);
318
319
320
else
{
finfo->stream.next_out = buf;
321
finfo->stream.avail_out = (uInt) maxread;
322
323
324
while (retval < maxread)
{
325
const PHYSFS_uint32 before = (PHYSFS_uint32) finfo->stream.total_out;
326
327
328
329
330
331
332
333
334
335
336
337
int rc;
if (finfo->stream.avail_in == 0)
{
PHYSFS_sint64 br;
br = entry->compressed_size - finfo->compressed_position;
if (br > 0)
{
if (br > ZIP_READBUFSIZE)
br = ZIP_READBUFSIZE;
338
br = zip_read_decrypt(finfo, finfo->buffer, (PHYSFS_uint64) br);
339
340
341
if (br <= 0)
break;
342
finfo->compressed_position += (PHYSFS_uint32) br;
343
finfo->stream.next_in = finfo->buffer;
344
finfo->stream.avail_in = (unsigned int) br;
345
346
} /* if */
} /* if */
347
348
349
rc = zlib_err(inflate(&finfo->stream, Z_SYNC_FLUSH));
retval += (finfo->stream.total_out - before);
350
351
352
353
354
355
356
if (rc != Z_OK)
break;
} /* while */
} /* else */
if (retval > 0)
357
finfo->uncompressed_position += (PHYSFS_uint32) retval;
358
359
return retval;
360
361
362
} /* ZIP_read */
363
static PHYSFS_sint64 ZIP_write(PHYSFS_Io *io, const void *b, PHYSFS_uint64 len)
364
{
365
BAIL(PHYSFS_ERR_READ_ONLY, -1);
366
367
368
} /* ZIP_write */
369
static PHYSFS_sint64 ZIP_tell(PHYSFS_Io *io)
370
{
371
return ((ZIPfileinfo *) io->opaque)->uncompressed_position;
372
373
374
} /* ZIP_tell */
375
static int ZIP_seek(PHYSFS_Io *_io, PHYSFS_uint64 offset)
376
{
377
ZIPfileinfo *finfo = (ZIPfileinfo *) _io->opaque;
378
ZIPentry *entry = finfo->entry;
379
PHYSFS_Io *io = finfo->io;
380
const int encrypted = zip_entry_is_tradional_crypto(entry);
381
382
BAIL_IF(offset > entry->uncompressed_size, PHYSFS_ERR_PAST_EOF, 0);
383
384
if (!encrypted && (entry->compression_method == COMPMETH_NONE))
385
{
386
PHYSFS_sint64 newpos = offset + entry->offset;
387
BAIL_IF_ERRPASS(!io->seek(io, newpos), 0);
388
finfo->uncompressed_position = (PHYSFS_uint32) offset;
389
} /* if */
390
391
else
392
{
393
394
395
396
/*
* If seeking backwards, we need to redecode the file
* from the start and throw away the compressed bits until we hit
* the offset we need. If seeking forward, we still need to
397
* decode, but we don't rewind first.
398
399
400
401
402
*/
if (offset < finfo->uncompressed_position)
{
/* we do a copy so state is sane if inflateInit2() fails. */
z_stream str;
403
initializeZStream(&str);
404
if (zlib_err(inflateInit2(&str, -MAX_WBITS)) != Z_OK)
405
return 0;
406
407
if (!io->seek(io, entry->offset + (encrypted ? 12 : 0)))
408
return 0;
409
410
411
412
inflateEnd(&finfo->stream);
memcpy(&finfo->stream, &str, sizeof (z_stream));
finfo->uncompressed_position = finfo->compressed_position = 0;
413
414
415
if (encrypted)
memcpy(finfo->crypto_keys, finfo->initial_crypto_keys, 12);
416
417
418
419
420
} /* if */
while (finfo->uncompressed_position != offset)
{
PHYSFS_uint8 buf[512];
421
422
423
PHYSFS_uint32 maxread;
maxread = (PHYSFS_uint32) (offset - finfo->uncompressed_position);
424
425
426
if (maxread > sizeof (buf))
maxread = sizeof (buf);
427
if (ZIP_read(_io, buf, maxread) != maxread)
428
return 0;
429
430
431
} /* while */
} /* else */
432
return 1;
433
434
435
} /* ZIP_seek */
436
static PHYSFS_sint64 ZIP_length(PHYSFS_Io *io)
437
{
438
const ZIPfileinfo *finfo = (ZIPfileinfo *) io->opaque;
439
return (PHYSFS_sint64) finfo->entry->uncompressed_size;
440
} /* ZIP_length */
441
442
443
444
445
static PHYSFS_Io *zip_get_io(PHYSFS_Io *io, ZIPinfo *inf, ZIPentry *entry);
static PHYSFS_Io *ZIP_duplicate(PHYSFS_Io *io)
446
{
447
448
449
ZIPfileinfo *origfinfo = (ZIPfileinfo *) io->opaque;
PHYSFS_Io *retval = (PHYSFS_Io *) allocator.Malloc(sizeof (PHYSFS_Io));
ZIPfileinfo *finfo = (ZIPfileinfo *) allocator.Malloc(sizeof (ZIPfileinfo));
450
451
GOTO_IF(!retval, PHYSFS_ERR_OUT_OF_MEMORY, failed);
GOTO_IF(!finfo, PHYSFS_ERR_OUT_OF_MEMORY, failed);
452
453
454
455
memset(finfo, '\0', sizeof (*finfo));
finfo->entry = origfinfo->entry;
finfo->io = zip_get_io(origfinfo->io, NULL, finfo->entry);
456
GOTO_IF_ERRPASS(!finfo->io, failed);
457
458
initializeZStream(&finfo->stream);
459
460
461
if (finfo->entry->compression_method != COMPMETH_NONE)
{
finfo->buffer = (PHYSFS_uint8 *) allocator.Malloc(ZIP_READBUFSIZE);
462
GOTO_IF(!finfo->buffer, PHYSFS_ERR_OUT_OF_MEMORY, failed);
463
464
if (zlib_err(inflateInit2(&finfo->stream, -MAX_WBITS)) != Z_OK)
goto failed;
465
466
467
468
469
470
} /* if */
memcpy(retval, io, sizeof (PHYSFS_Io));
retval->opaque = finfo;
return retval;
471
failed:
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
if (finfo != NULL)
{
if (finfo->io != NULL)
finfo->io->destroy(finfo->io);
if (finfo->buffer != NULL)
{
allocator.Free(finfo->buffer);
inflateEnd(&finfo->stream);
} /* if */
allocator.Free(finfo);
} /* if */
if (retval != NULL)
allocator.Free(retval);
return NULL;
} /* ZIP_duplicate */
static int ZIP_flush(PHYSFS_Io *io) { return 1; /* no write support. */ }
static void ZIP_destroy(PHYSFS_Io *io)
{
ZIPfileinfo *finfo = (ZIPfileinfo *) io->opaque;
finfo->io->destroy(finfo->io);
498
499
500
501
502
if (finfo->entry->compression_method != COMPMETH_NONE)
inflateEnd(&finfo->stream);
if (finfo->buffer != NULL)
503
allocator.Free(finfo->buffer);
504
505
allocator.Free(finfo);
506
507
508
509
510
511
allocator.Free(io);
} /* ZIP_destroy */
static const PHYSFS_Io ZIP_Io =
{
512
CURRENT_PHYSFS_IO_API_VERSION, NULL,
513
514
515
516
517
518
519
ZIP_read,
ZIP_write,
ZIP_seek,
ZIP_tell,
ZIP_length,
ZIP_duplicate,
ZIP_flush,
520
ZIP_destroy
521
522
};
523
524
525
static PHYSFS_sint64 zip_find_end_of_central_dir(PHYSFS_Io *io, PHYSFS_sint64 *len)
526
{
527
PHYSFS_uint8 buf[256];
528
PHYSFS_uint8 extra[4] = { 0, 0, 0, 0 };
529
PHYSFS_sint32 i = 0;
530
531
532
PHYSFS_sint64 filelen;
PHYSFS_sint64 filepos;
PHYSFS_sint32 maxread;
533
534
PHYSFS_sint32 totalread = 0;
int found = 0;
535
536
filelen = io->length(io);
537
BAIL_IF_ERRPASS(filelen == -1, -1);
538
539
540
541
542
543
544
545
546
/*
* Jump to the end of the file and start reading backwards.
* The last thing in the file is the zipfile comment, which is variable
* length, and the field that specifies its size is before it in the
* file (argh!)...this means that we need to scan backwards until we
* hit the end-of-central-dir signature. We can then sanity check that
* the comment was as big as it should be to make sure we're in the
* right place. The comment length field is 16 bits, so we can stop
547
548
* searching for that signature after a little more than 64k at most,
* and call it a corrupted zipfile.
549
550
551
552
553
554
555
556
557
558
*/
if (sizeof (buf) < filelen)
{
filepos = filelen - sizeof (buf);
maxread = sizeof (buf);
} /* if */
else
{
filepos = 0;
559
maxread = (PHYSFS_uint32) filelen;
560
561
} /* else */
562
while ((totalread < filelen) && (totalread < 65557))
563
{
564
BAIL_IF_ERRPASS(!io->seek(io, filepos), -1);
565
566
567
/* make sure we catch a signature between buffers. */
if (totalread != 0)
568
{
569
if (!__PHYSFS_readAll(io, buf, maxread - 4))
570
return -1;
571
memcpy(&buf[maxread - 4], &extra, sizeof (extra));
572
totalread += maxread - 4;
573
} /* if */
574
575
else
{
576
if (!__PHYSFS_readAll(io, buf, maxread))
577
return -1;
578
579
580
totalread += maxread;
} /* else */
581
memcpy(&extra, buf, sizeof (extra));
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
for (i = maxread - 4; i > 0; i--)
{
if ((buf[i + 0] == 0x50) &&
(buf[i + 1] == 0x4B) &&
(buf[i + 2] == 0x05) &&
(buf[i + 3] == 0x06) )
{
found = 1; /* that's the signature! */
break;
} /* if */
} /* for */
if (found)
break;
filepos -= (maxread - 4);
599
600
if (filepos < 0)
filepos = 0;
601
602
} /* while */
603
BAIL_IF(!found, PHYSFS_ERR_UNSUPPORTED, -1);
604
605
606
607
if (len != NULL)
*len = filelen;
608
return (filepos + i);
609
} /* zip_find_end_of_central_dir */
610
611
612
static int isZip(PHYSFS_Io *io)
613
{
614
PHYSFS_uint32 sig = 0;
615
int retval = 0;
616
617
618
619
620
/*
* The first thing in a zip file might be the signature of the
* first local file record, so it makes for a quick determination.
*/
621
if (readui32(io, &sig))
622
{
623
624
625
626
627
628
629
630
retval = (sig == ZIP_LOCAL_FILE_SIG);
if (!retval)
{
/*
* No sig...might be a ZIP with data at the start
* (a self-extracting executable, etc), so we'll have to do
* it the hard way...
*/
631
retval = (zip_find_end_of_central_dir(io, NULL) != -1);
632
} /* if */
633
634
} /* if */
635
return retval;
636
} /* isZip */
637
638
639
/* Convert paths from old, buggy DOS zippers... */
640
static void zip_convert_dos_path(const PHYSFS_uint16 entryversion, char *path)
641
{
642
const PHYSFS_uint8 hosttype = (PHYSFS_uint8) ((entryversion >> 8) & 0xFF);
643
644
645
646
647
648
649
650
651
652
if (hosttype == 0) /* FS_FAT_ */
{
while (*path)
{
if (*path == '\\')
*path = '/';
path++;
} /* while */
} /* if */
} /* zip_convert_dos_path */
653
654
655
static void zip_expand_symlink_path(char *path)
656
{
657
658
char *ptr = path;
char *prevptr = path;
659
660
while (1)
661
{
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
ptr = strchr(ptr, '/');
if (ptr == NULL)
break;
if (*(ptr + 1) == '.')
{
if (*(ptr + 2) == '/')
{
/* current dir in middle of string: ditch it. */
memmove(ptr, ptr + 2, strlen(ptr + 2) + 1);
} /* else if */
else if (*(ptr + 2) == '\0')
{
/* current dir at end of string: ditch it. */
*ptr = '\0';
} /* else if */
else if (*(ptr + 2) == '.')
{
if (*(ptr + 3) == '/')
{
/* parent dir in middle: move back one, if possible. */
memmove(prevptr, ptr + 4, strlen(ptr + 4) + 1);
ptr = prevptr;
while (prevptr != path)
{
prevptr--;
if (*prevptr == '/')
{
prevptr++;
break;
} /* if */
} /* while */
} /* if */
697
698
699
700
701
702
703
704
705
706
707
if (*(ptr + 3) == '\0')
{
/* parent dir at end: move back one, if possible. */
*prevptr = '\0';
} /* if */
} /* if */
} /* if */
else
{
prevptr = ptr;
708
ptr++;
709
710
711
} /* else */
} /* while */
} /* zip_expand_symlink_path */
712
713
714
715
716
717
718
static inline ZIPentry *zip_find_entry(ZIPinfo *info, const char *path)
{
return (ZIPentry *) __PHYSFS_DirTreeFind(&info->tree, path);
} /* zip_find_entry */
719
/* (forward reference: zip_follow_symlink and zip_resolve call each other.) */
720
static int zip_resolve(PHYSFS_Io *io, ZIPinfo *info, ZIPentry *entry);
721
722
/*
723
724
725
* Look for the entry named by (path). If it exists, resolve it, and return
* a pointer to that entry. If it's another symlink, keep resolving until you
* hit a real file and then return a pointer to the final non-symlink entry.
726
* If there's a problem, return NULL.
727
*/
728
static ZIPentry *zip_follow_symlink(PHYSFS_Io *io, ZIPinfo *info, char *path)
729
{
730
731
732
ZIPentry *entry;
zip_expand_symlink_path(path);
733
entry = zip_find_entry(info, path);
734
735
if (entry != NULL)
{
736
if (!zip_resolve(io, info, entry)) /* recursive! */
737
738
739
740
741
742
743
744
entry = NULL;
else
{
if (entry->symlink != NULL)
entry = entry->symlink;
} /* else */
} /* if */
745
return entry;
746
} /* zip_follow_symlink */
747
748
749
static int zip_resolve_symlink(PHYSFS_Io *io, ZIPinfo *info, ZIPentry *entry)
750
{
751
const size_t size = (size_t) entry->uncompressed_size;
752
char *path = NULL;
753
754
int rc = 0;
755
756
757
758
759
760
/*
* We've already parsed the local file header of the symlink at this
* point. Now we need to read the actual link from the file data and
* follow it.
*/
761
BAIL_IF_ERRPASS(!io->seek(io, entry->offset), 0);
762
763
path = (char *) __PHYSFS_smallAlloc(size + 1);
764
BAIL_IF(!path, PHYSFS_ERR_OUT_OF_MEMORY, 0);
765
766
if (entry->compression_method == COMPMETH_NONE)
767
rc = __PHYSFS_readAll(io, path, size);
768
769
770
771
else /* symlink target path is compressed... */
{
z_stream stream;
772
const size_t complen = (size_t) entry->compressed_size;
773
PHYSFS_uint8 *compressed = (PHYSFS_uint8*) __PHYSFS_smallAlloc(complen);
774
775
if (compressed != NULL)
{
776
if (__PHYSFS_readAll(io, compressed, complen))
777
{
778
initializeZStream(&stream);
779
stream.next_in = compressed;
780
stream.avail_in = (unsigned int) complen;
781
stream.next_out = (unsigned char *) path;
782
stream.avail_out = (unsigned int) size;
783
784
if (zlib_err(inflateInit2(&stream, -MAX_WBITS)) == Z_OK)
{
785
rc = zlib_err(inflate(&stream, Z_FINISH));
786
inflateEnd(&stream);
787
788
789
/* both are acceptable outcomes... */
rc = ((rc == Z_OK) || (rc == Z_STREAM_END));
790
791
} /* if */
} /* if */
792
__PHYSFS_smallFree(compressed);
793
794
} /* if */
} /* else */
795
796
if (rc)
797
798
{
path[entry->uncompressed_size] = '\0'; /* null-terminate it. */
799
zip_convert_dos_path(entry->version, path);
800
entry->symlink = zip_follow_symlink(io, info, path);
801
802
} /* else */
803
804
__PHYSFS_smallFree(path);
805
return (entry->symlink != NULL);
806
807
808
809
810
811
} /* zip_resolve_symlink */
/*
* Parse the local file header of an entry, and update entry->offset.
*/
812
static int zip_parse_local(PHYSFS_Io *io, ZIPentry *entry)
813
814
815
816
817
818
{
PHYSFS_uint32 ui32;
PHYSFS_uint16 ui16;
PHYSFS_uint16 fnamelen;
PHYSFS_uint16 extralen;
819
820
821
822
823
/*
* crc and (un)compressed_size are always zero if this is a "JAR"
* archive created with Sun's Java tools, apparently. We only
* consider this archive corrupted if those entries don't match and
* aren't zero. That seems to work well.
824
825
* We also ignore a mismatch if the value is 0xFFFFFFFF here, since it's
* possible that's a Zip64 thing.
826
827
*/
828
829
830
831
/* !!! FIXME: apparently these are zero if general purpose bit 3 is set,
!!! FIXME: which is probably true for Jar files, fwiw, but we don't
!!! FIXME: care about these values anyhow. */
832
833
834
835
836
837
838
839
840
841
842
843
844
845
BAIL_IF_ERRPASS(!io->seek(io, entry->offset), 0);
BAIL_IF_ERRPASS(!readui32(io, &ui32), 0);
BAIL_IF(ui32 != ZIP_LOCAL_FILE_SIG, PHYSFS_ERR_CORRUPT, 0);
BAIL_IF_ERRPASS(!readui16(io, &ui16), 0);
BAIL_IF(ui16 != entry->version_needed, PHYSFS_ERR_CORRUPT, 0);
BAIL_IF_ERRPASS(!readui16(io, &ui16), 0); /* general bits. */
BAIL_IF_ERRPASS(!readui16(io, &ui16), 0);
BAIL_IF(ui16 != entry->compression_method, PHYSFS_ERR_CORRUPT, 0);
BAIL_IF_ERRPASS(!readui32(io, &ui32), 0); /* date/time */
BAIL_IF_ERRPASS(!readui32(io, &ui32), 0);
BAIL_IF(ui32 && (ui32 != entry->crc), PHYSFS_ERR_CORRUPT, 0);
BAIL_IF_ERRPASS(!readui32(io, &ui32), 0);
BAIL_IF(ui32 && (ui32 != 0xFFFFFFFF) &&
846
847
(ui32 != entry->compressed_size), PHYSFS_ERR_CORRUPT, 0);
848
849
BAIL_IF_ERRPASS(!readui32(io, &ui32), 0);
BAIL_IF(ui32 && (ui32 != 0xFFFFFFFF) &&
850
851
(ui32 != entry->uncompressed_size), PHYSFS_ERR_CORRUPT, 0);
852
853
BAIL_IF_ERRPASS(!readui16(io, &fnamelen), 0);
BAIL_IF_ERRPASS(!readui16(io, &extralen), 0);
854
855
entry->offset += fnamelen + extralen + 30;
856
return 1;
857
858
859
} /* zip_parse_local */
860
static int zip_resolve(PHYSFS_Io *io, ZIPinfo *info, ZIPentry *entry)
861
862
{
int retval = 1;
863
864
865
866
const ZipResolveType resolve_type = entry->resolved;
if (resolve_type == ZIP_DIRECTORY)
return 1; /* we're good. */
867
868
/* Don't bother if we've failed to resolve this entry before. */
869
870
BAIL_IF(resolve_type == ZIP_BROKEN_FILE, PHYSFS_ERR_CORRUPT, 0);
BAIL_IF(resolve_type == ZIP_BROKEN_SYMLINK, PHYSFS_ERR_CORRUPT, 0);
871
872
/* uhoh...infinite symlink loop! */
873
BAIL_IF(resolve_type == ZIP_RESOLVING, PHYSFS_ERR_SYMLINK_LOOP, 0);
874
875
876
877
878
879
880
881
882
883
/*
* We fix up the offset to point to the actual data on the
* first open, since we don't want to seek across the whole file on
* archive open (can be SLOW on large, CD-stored files), but we
* need to check the local file header...not just for corruption,
* but since it stores offset info the central directory does not.
*/
if (resolve_type != ZIP_RESOLVED)
{
884
885
886
887
888
if (entry->tree.isdir) /* an ancestor dir that DirTree filled in? */
{
entry->resolved = ZIP_DIRECTORY;
return 1;
} /* if */
889
890
retval = zip_parse_local(io, entry);
891
892
893
894
895
896
897
898
if (retval)
{
/*
* If it's a symlink, find the original file. This will cause
* resolution of other entries (other symlinks and, eventually,
* the real file) if all goes well.
*/
if (resolve_type == ZIP_UNRESOLVED_SYMLINK)
899
retval = zip_resolve_symlink(io, info, entry);
900
901
902
903
904
905
} /* if */
if (resolve_type == ZIP_UNRESOLVED_SYMLINK)
entry->resolved = ((retval) ? ZIP_RESOLVED : ZIP_BROKEN_SYMLINK);
else if (resolve_type == ZIP_UNRESOLVED_FILE)
entry->resolved = ((retval) ? ZIP_RESOLVED : ZIP_BROKEN_FILE);
906
} /* if */
907
908
return retval;
909
} /* zip_resolve */
910
911
912
913
914
915
916
917
918
919
static int zip_entry_is_symlink(const ZIPentry *entry)
{
return ((entry->resolved == ZIP_UNRESOLVED_SYMLINK) ||
(entry->resolved == ZIP_BROKEN_SYMLINK) ||
(entry->symlink));
} /* zip_entry_is_symlink */
920
static int zip_version_does_symlinks(PHYSFS_uint32 version)
921
922
{
int retval = 0;
923
PHYSFS_uint8 hosttype = (PHYSFS_uint8) ((version >> 8) & 0xFF);
924
925
926
switch (hosttype)
{
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
/*
* These are the platforms that can NOT build an archive with
* symlinks, according to the Info-ZIP project.
*/
case 0: /* FS_FAT_ */
case 1: /* AMIGA_ */
case 2: /* VMS_ */
case 4: /* VM_CSM_ */
case 6: /* FS_HPFS_ */
case 11: /* FS_NTFS_ */
case 14: /* FS_VFAT_ */
case 13: /* ACORN_ */
case 15: /* MVS_ */
case 18: /* THEOS_ */
break; /* do nothing. */
default: /* assume the rest to be unix-like. */
944
945
946
947
retval = 1;
break;
} /* switch */
948
return retval;
949
950
951
} /* zip_version_does_symlinks */
952
953
static inline int zip_has_symlink_attr(const ZIPentry *entry,
const PHYSFS_uint32 extern_attr)
954
{
955
PHYSFS_uint16 xattr = ((extern_attr >> 16) & 0xFFFF);
956
957
958
return ( (zip_version_does_symlinks(entry->version)) &&
(entry->uncompressed_size > 0) &&
((xattr & UNIX_FILETYPE_MASK) == UNIX_FILETYPE_SYMLINK) );
959
} /* zip_has_symlink_attr */
960
961
962
static PHYSFS_sint64 zip_dos_time_to_physfs_time(PHYSFS_uint32 dostime)
963
{
964
PHYSFS_uint32 dosdate;
965
966
struct tm unixtime;
memset(&unixtime, '\0', sizeof (unixtime));
967
968
969
dosdate = (PHYSFS_uint32) ((dostime >> 16) & 0xFFFF);
dostime &= 0xFFFF;
970
971
972
973
974
975
976
977
978
979
980
981
982
/* dissect date */
unixtime.tm_year = ((dosdate >> 9) & 0x7F) + 80;
unixtime.tm_mon = ((dosdate >> 5) & 0x0F) - 1;
unixtime.tm_mday = ((dosdate ) & 0x1F);
/* dissect time */
unixtime.tm_hour = ((dostime >> 11) & 0x1F);
unixtime.tm_min = ((dostime >> 5) & 0x3F);
unixtime.tm_sec = ((dostime << 1) & 0x3E);
/* let mktime calculate daylight savings time. */
unixtime.tm_isdst = -1;
983
984
return ((PHYSFS_sint64) mktime(&unixtime));
985
} /* zip_dos_time_to_physfs_time */
986
987
988
static ZIPentry *zip_load_entry(ZIPinfo *info, const int zip64,
989
const PHYSFS_uint64 ofs_fixup)
990
{
991
PHYSFS_Io *io = info->io;
992
993
ZIPentry entry;
ZIPentry *retval = NULL;
994
995
PHYSFS_uint16 fnamelen, extralen, commentlen;
PHYSFS_uint32 external_attr;
996
997
PHYSFS_uint32 starting_disk;
PHYSFS_uint64 offset;
998
999
1000
PHYSFS_uint16 ui16;
PHYSFS_uint32 ui32;
PHYSFS_sint64 si64;