4 * Copyright 2000-2002 Stuart Caie
6 * This library is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * This library is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with this library; if not, write to the Free Software
18 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20 * Principal author: Stuart Caie <kyzer@4u.net>
22 * Based on specification documents from Microsoft Corporation
23 * Quantum decompression researched and implemented by Matthew Russoto
24 * Huffman code adapted from unlzx by Dave Tritscher.
25 * InfoZip team's INFLATE implementation adapted to MSZIP by Dirk Stoecker.
26 * Major LZX fixes by Jae Jung.
42 #include "wine/debug.h"
44 WINE_DEFAULT_DEBUG_CHANNEL(cabinet);
48 /* all the file IO is abstracted into these routines:
49 * cabinet_(open|close|read|seek|skip|getoffset)
50 * file_(open|close|write)
53 /* try to open a cabinet file, returns success */
54 static BOOL cabinet_open(struct cabinet *cab)
56 const char *name = cab->filename;
59 TRACE("(cab == ^%p)\n", cab);
61 if ((fh = CreateFileA( name, GENERIC_READ, FILE_SHARE_READ,
62 NULL, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL )) == INVALID_HANDLE_VALUE) {
63 ERR("Couldn't open %s\n", debugstr_a(name));
67 /* seek to end of file and get the length */
68 if ((cab->filelen = SetFilePointer(fh, 0, NULL, FILE_END)) == INVALID_SET_FILE_POINTER) {
69 if (GetLastError() != NO_ERROR) {
70 ERR("Seek END failed: %s\n", debugstr_a(name));
76 /* return to the start of the file */
77 if (SetFilePointer(fh, 0, NULL, FILE_BEGIN) == INVALID_SET_FILE_POINTER) {
78 ERR("Seek BEGIN failed: %s\n", debugstr_a(name));
87 /*******************************************************************
88 * cabinet_close (internal)
90 * close the file handle in a struct cabinet.
92 static void cabinet_close(struct cabinet *cab) {
93 TRACE("(cab == ^%p)\n", cab);
94 if (cab->fh) CloseHandle(cab->fh);
98 /*******************************************************
99 * ensure_filepath2 (internal)
101 static BOOL ensure_filepath2(char *path) {
106 new_path = HeapAlloc(GetProcessHeap(), 0, (strlen(path) + 1));
107 strcpy(new_path, path);
109 while((len = strlen(new_path)) && new_path[len - 1] == '\\')
110 new_path[len - 1] = 0;
112 TRACE("About to try to create directory %s\n", debugstr_a(new_path));
113 while(!CreateDirectoryA(new_path, NULL)) {
115 DWORD last_error = GetLastError();
117 if(last_error == ERROR_ALREADY_EXISTS)
120 if(last_error != ERROR_PATH_NOT_FOUND) {
125 if(!(slash = strrchr(new_path, '\\'))) {
130 len = slash - new_path;
132 if(! ensure_filepath2(new_path)) {
136 new_path[len] = '\\';
137 TRACE("New path in next iteration: %s\n", debugstr_a(new_path));
140 HeapFree(GetProcessHeap(), 0, new_path);
145 /**********************************************************************
146 * ensure_filepath (internal)
148 * ensure_filepath("a\b\c\d.txt") ensures a, a\b and a\b\c exist as dirs
150 static BOOL ensure_filepath(char *path) {
151 char new_path[MAX_PATH];
152 int len, i, lastslashpos = -1;
154 TRACE("(path == %s)\n", debugstr_a(path));
156 strcpy(new_path, path);
157 /* remove trailing slashes (shouldn't need to but wth...) */
158 while ((len = strlen(new_path)) && new_path[len - 1] == '\\')
159 new_path[len - 1] = 0;
160 /* find the position of the last '\\' */
161 for (i=0; i<MAX_PATH; i++) {
162 if (new_path[i] == 0) break;
163 if (new_path[i] == '\\')
166 if (lastslashpos > 0) {
167 new_path[lastslashpos] = 0;
168 /* may be trailing slashes but ensure_filepath2 will chop them */
169 return ensure_filepath2(new_path);
174 /*******************************************************************
175 * file_open (internal)
177 * opens a file for output, returns success
179 static BOOL file_open(struct cab_file *fi, BOOL lower, LPCSTR dir)
185 TRACE("(fi == ^%p, lower == %s, dir == %s)\n", fi, lower ? "TRUE" : "FALSE", debugstr_a(dir));
187 if (!(name = malloc(strlen(fi->filename) + (dir ? strlen(dir) : 0) + 2))) {
188 ERR("out of memory!\n");
192 /* start with blank name */
195 /* add output directory if needed */
201 /* remove leading slashes */
202 s = (char *) fi->filename;
203 while (*s == '\\') s++;
205 /* copy from fi->filename to new name.
206 * lowercases characters if needed.
208 d = &name[strlen(name)];
211 *d++ = (lower ? tolower((unsigned char) c) : c);
214 /* create directories if needed, attempt to write file */
215 if (ensure_filepath(name)) {
216 fi->fh = CreateFileA(name, GENERIC_WRITE, 0, NULL,
217 CREATE_ALWAYS, FILE_ATTRIBUTE_NORMAL, 0);
218 if (fi->fh != INVALID_HANDLE_VALUE)
221 ERR("CreateFileA returned INVALID_HANDLE_VALUE\n");
225 ERR("Couldn't ensure filepath for %s\n", debugstr_a(name));
228 ERR("Couldn't open file %s for writing\n", debugstr_a(name));
231 /* as full filename is no longer needed, free it */
237 /********************************************************
238 * close_file (internal)
240 * closes a completed file
242 static void file_close(struct cab_file *fi)
244 TRACE("(fi == ^%p)\n", fi);
252 /******************************************************************
253 * file_write (internal)
255 * writes from buf to a file specified as a cab_file struct.
256 * returns success/failure
258 static BOOL file_write(struct cab_file *fi, cab_UBYTE *buf, cab_off_t length)
262 TRACE("(fi == ^%p, buf == ^%p, length == %u)\n", fi, buf, length);
264 if ((!WriteFile( fi->fh, (LPCVOID) buf, length, &bytes_written, FALSE) ||
265 (bytes_written != length))) {
266 ERR("Error writing file: %s\n", debugstr_a(fi->filename));
273 /*******************************************************************
274 * cabinet_skip (internal)
276 * advance the file pointer associated with the cab structure
279 static void cabinet_skip(struct cabinet *cab, cab_off_t distance)
281 TRACE("(cab == ^%p, distance == %u)\n", cab, distance);
282 if (SetFilePointer(cab->fh, distance, NULL, FILE_CURRENT) == INVALID_SET_FILE_POINTER) {
283 if (distance != INVALID_SET_FILE_POINTER)
284 ERR("%s\n", debugstr_a(cab->filename));
288 /*******************************************************************
289 * cabinet_seek (internal)
291 * seek to the specified absolute offset in a cab
293 static void cabinet_seek(struct cabinet *cab, cab_off_t offset) {
294 TRACE("(cab == ^%p, offset == %u)\n", cab, offset);
295 if (SetFilePointer(cab->fh, offset, NULL, FILE_BEGIN) != offset)
296 ERR("%s seek failure\n", debugstr_a(cab->filename));
299 /*******************************************************************
300 * cabinet_getoffset (internal)
302 * returns the file pointer position of a cab
304 static cab_off_t cabinet_getoffset(struct cabinet *cab)
306 return SetFilePointer(cab->fh, 0, NULL, FILE_CURRENT);
309 /*******************************************************************
310 * cabinet_read (internal)
312 * read data from a cabinet, returns success
314 static BOOL cabinet_read(struct cabinet *cab, cab_UBYTE *buf, cab_off_t length)
317 cab_off_t avail = cab->filelen - cabinet_getoffset(cab);
319 TRACE("(cab == ^%p, buf == ^%p, length == %u)\n", cab, buf, length);
321 if (length > avail) {
322 WARN("%s: WARNING; cabinet is truncated\n", debugstr_a(cab->filename));
326 if (! ReadFile( cab->fh, (LPVOID) buf, length, &bytes_read, NULL )) {
327 ERR("%s read error\n", debugstr_a(cab->filename));
329 } else if (bytes_read != length) {
330 ERR("%s read size mismatch\n", debugstr_a(cab->filename));
337 /**********************************************************************
338 * cabinet_read_string (internal)
340 * allocate and read an aribitrarily long string from the cabinet
342 static char *cabinet_read_string(struct cabinet *cab)
344 cab_off_t len=256, base = cabinet_getoffset(cab), maxlen = cab->filelen - base;
347 cab_UBYTE *buf = NULL;
349 TRACE("(cab == ^%p)\n", cab);
352 if (len > maxlen) len = maxlen;
353 if (!(buf = realloc(buf, (size_t) len))) break;
354 if (!cabinet_read(cab, buf, (size_t) len)) break;
356 /* search for a null terminator in what we've just read */
357 for (i=0; i < len; i++) {
358 if (!buf[i]) {ok=TRUE; break;}
363 ERR("%s: WARNING; cabinet is truncated\n", debugstr_a(cab->filename));
367 cabinet_seek(cab, base);
375 ERR("out of memory!\n");
379 /* otherwise, set the stream to just after the string and return */
380 cabinet_seek(cab, base + ((cab_off_t) strlen((char *) buf)) + 1);
385 /******************************************************************
386 * cabinet_read_entries (internal)
388 * reads the header and all folder and file entries in this cabinet
390 static BOOL cabinet_read_entries(struct cabinet *cab)
392 int num_folders, num_files, header_resv, folder_resv = 0, i;
393 struct cab_folder *fol, *linkfol = NULL;
394 struct cab_file *file, *linkfile = NULL;
395 cab_off_t base_offset;
398 TRACE("(cab == ^%p)\n", cab);
400 /* read in the CFHEADER */
401 base_offset = cabinet_getoffset(cab);
402 if (!cabinet_read(cab, buf, cfhead_SIZEOF)) {
406 /* check basic MSCF signature */
407 if (EndGetI32(buf+cfhead_Signature) != 0x4643534d) {
408 ERR("%s: not a Microsoft cabinet file\n", debugstr_a(cab->filename));
412 /* get the number of folders */
413 num_folders = EndGetI16(buf+cfhead_NumFolders);
414 if (num_folders == 0) {
415 ERR("%s: no folders in cabinet\n", debugstr_a(cab->filename));
419 /* get the number of files */
420 num_files = EndGetI16(buf+cfhead_NumFiles);
421 if (num_files == 0) {
422 ERR("%s: no files in cabinet\n", debugstr_a(cab->filename));
426 /* just check the header revision */
427 if ((buf[cfhead_MajorVersion] > 1) ||
428 (buf[cfhead_MajorVersion] == 1 && buf[cfhead_MinorVersion] > 3))
430 WARN("%s: WARNING; cabinet format version > 1.3\n", debugstr_a(cab->filename));
433 /* read the reserved-sizes part of header, if present */
434 cab->flags = EndGetI16(buf+cfhead_Flags);
435 if (cab->flags & cfheadRESERVE_PRESENT) {
436 if (!cabinet_read(cab, buf, cfheadext_SIZEOF)) return FALSE;
437 header_resv = EndGetI16(buf+cfheadext_HeaderReserved);
438 folder_resv = buf[cfheadext_FolderReserved];
439 cab->block_resv = buf[cfheadext_DataReserved];
441 if (header_resv > 60000) {
442 WARN("%s: WARNING; header reserved space > 60000\n", debugstr_a(cab->filename));
445 /* skip the reserved header */
447 if (SetFilePointer(cab->fh, (cab_off_t) header_resv, NULL, FILE_CURRENT) == INVALID_SET_FILE_POINTER)
448 ERR("seek failure: %s\n", debugstr_a(cab->filename));
451 if (cab->flags & cfheadPREV_CABINET) {
452 cab->prevname = cabinet_read_string(cab);
453 if (!cab->prevname) return FALSE;
454 cab->previnfo = cabinet_read_string(cab);
457 if (cab->flags & cfheadNEXT_CABINET) {
458 cab->nextname = cabinet_read_string(cab);
459 if (!cab->nextname) return FALSE;
460 cab->nextinfo = cabinet_read_string(cab);
464 for (i = 0; i < num_folders; i++) {
465 if (!cabinet_read(cab, buf, cffold_SIZEOF)) return FALSE;
466 if (folder_resv) cabinet_skip(cab, folder_resv);
468 fol = (struct cab_folder *) calloc(1, sizeof(struct cab_folder));
470 ERR("out of memory!\n");
475 fol->offset[0] = base_offset + (cab_off_t) EndGetI32(buf+cffold_DataOffset);
476 fol->num_blocks = EndGetI16(buf+cffold_NumBlocks);
477 fol->comp_type = EndGetI16(buf+cffold_CompType);
488 for (i = 0; i < num_files; i++) {
489 if (!cabinet_read(cab, buf, cffile_SIZEOF))
492 file = (struct cab_file *) calloc(1, sizeof(struct cab_file));
494 ERR("out of memory!\n");
498 file->length = EndGetI32(buf+cffile_UncompressedSize);
499 file->offset = EndGetI32(buf+cffile_FolderOffset);
500 file->index = EndGetI16(buf+cffile_FolderIndex);
501 file->time = EndGetI16(buf+cffile_Time);
502 file->date = EndGetI16(buf+cffile_Date);
503 file->attribs = EndGetI16(buf+cffile_Attribs);
504 file->filename = cabinet_read_string(cab);
506 if (!file->filename) {
514 linkfile->next = file;
521 /***********************************************************
522 * load_cab_offset (internal)
524 * validates and reads file entries from a cabinet at offset [offset] in
525 * file [name]. Returns a cabinet structure if successful, or NULL
528 static struct cabinet *load_cab_offset(LPCSTR name, cab_off_t offset)
530 struct cabinet *cab = (struct cabinet *) calloc(1, sizeof(struct cabinet));
533 TRACE("(name == %s, offset == %u)\n", debugstr_a(name), offset);
535 if (!cab) return NULL;
537 cab->filename = name;
538 if ((ok = cabinet_open(cab))) {
539 cabinet_seek(cab, offset);
540 ok = cabinet_read_entries(cab);
549 /* MSZIP decruncher */
551 /* Dirk Stoecker wrote the ZIP decoder, based on the InfoZip deflate code */
553 /********************************************************
554 * Ziphuft_free (internal)
556 static void Ziphuft_free(struct Ziphuft *t)
558 register struct Ziphuft *p, *q;
560 /* Go through linked list, freeing from the allocated (t[-1]) address. */
562 while (p != (struct Ziphuft *)NULL)
570 /*********************************************************
571 * Ziphuft_build (internal)
573 static cab_LONG Ziphuft_build(cab_ULONG *b, cab_ULONG n, cab_ULONG s, cab_UWORD *d, cab_UWORD *e,
574 struct Ziphuft **t, cab_LONG *m, cab_decomp_state *decomp_state)
576 cab_ULONG a; /* counter for codes of length k */
577 cab_ULONG el; /* length of EOB code (value 256) */
578 cab_ULONG f; /* i repeats in table every f entries */
579 cab_LONG g; /* maximum code length */
580 cab_LONG h; /* table level */
581 register cab_ULONG i; /* counter, current code */
582 register cab_ULONG j; /* counter */
583 register cab_LONG k; /* number of bits in current code */
584 cab_LONG *l; /* stack of bits per table */
585 register cab_ULONG *p; /* pointer into ZIP(c)[],ZIP(b)[],ZIP(v)[] */
586 register struct Ziphuft *q; /* points to current table */
587 struct Ziphuft r; /* table entry for structure assignment */
588 register cab_LONG w; /* bits before this table == (l * h) */
589 cab_ULONG *xp; /* pointer into x */
590 cab_LONG y; /* number of dummy codes added */
591 cab_ULONG z; /* number of entries in current table */
595 /* Generate counts for each bit length */
596 el = n > 256 ? b[256] : ZIPBMAX; /* set length of EOB code, if any */
598 for(i = 0; i < ZIPBMAX+1; ++i)
603 ZIP(c)[*p]++; p++; /* assume all entries <= ZIPBMAX */
605 if (ZIP(c)[0] == n) /* null input--all zero length codes */
607 *t = (struct Ziphuft *)NULL;
612 /* Find minimum and maximum length, bound *m by those */
613 for (j = 1; j <= ZIPBMAX; j++)
616 k = j; /* minimum code length */
617 if ((cab_ULONG)*m < j)
619 for (i = ZIPBMAX; i; i--)
622 g = i; /* maximum code length */
623 if ((cab_ULONG)*m > i)
626 /* Adjust last length count to fill out codes, if needed */
627 for (y = 1 << j; j < i; j++, y <<= 1)
628 if ((y -= ZIP(c)[j]) < 0)
629 return 2; /* bad input: more codes than bits */
630 if ((y -= ZIP(c)[i]) < 0)
634 /* Generate starting offsets LONGo the value table for each length */
636 p = ZIP(c) + 1; xp = ZIP(x) + 2;
638 { /* note that i == g from above */
642 /* Make a table of values in order of bit lengths */
646 ZIP(v)[ZIP(x)[j]++] = i;
650 /* Generate the Huffman codes and for each, make the table entries */
651 ZIP(x)[0] = i = 0; /* first Huffman code is zero */
652 p = ZIP(v); /* grab values in bit order */
653 h = -1; /* no tables yet--level -1 */
654 w = l[-1] = 0; /* no bits decoded yet */
655 ZIP(u)[0] = (struct Ziphuft *)NULL; /* just to keep compilers happy */
656 q = (struct Ziphuft *)NULL; /* ditto */
659 /* go through the bit lengths (k already is bits in shortest code) */
665 /* here i is the Huffman code of length k bits for value *p */
666 /* make tables up to required level */
669 w += l[h++]; /* add bits already decoded */
671 /* compute minimum size table less than or equal to *m bits */
672 z = (z = g - w) > (cab_ULONG)*m ? *m : z; /* upper limit */
673 if ((f = 1 << (j = k - w)) > a + 1) /* try a k-w bit table */
674 { /* too few codes for k-w bit table */
675 f -= a + 1; /* deduct codes from patterns left */
677 while (++j < z) /* try smaller tables up to z bits */
679 if ((f <<= 1) <= *++xp)
680 break; /* enough codes to use up j bits */
681 f -= *xp; /* else deduct codes from patterns */
684 if ((cab_ULONG)w + j > el && (cab_ULONG)w < el)
685 j = el - w; /* make EOB code end at table */
686 z = 1 << j; /* table entries for j-bit table */
687 l[h] = j; /* set table size in stack */
689 /* allocate and link in new table */
690 if (!(q = (struct Ziphuft *) malloc((z + 1)*sizeof(struct Ziphuft))))
693 Ziphuft_free(ZIP(u)[0]);
694 return 3; /* not enough memory */
696 *t = q + 1; /* link to list for Ziphuft_free() */
697 *(t = &(q->v.t)) = (struct Ziphuft *)NULL;
698 ZIP(u)[h] = ++q; /* table starts after link */
700 /* connect to last table, if there is one */
703 ZIP(x)[h] = i; /* save pattern for backing up */
704 r.b = (cab_UBYTE)l[h-1]; /* bits to dump before this table */
705 r.e = (cab_UBYTE)(16 + j); /* bits in this table */
706 r.v.t = q; /* pointer to this table */
707 j = (i & ((1 << w) - 1)) >> (w - l[h-1]);
708 ZIP(u)[h-1][j] = r; /* connect to last table */
712 /* set up table entry in r */
713 r.b = (cab_UBYTE)(k - w);
715 r.e = 99; /* out of values--invalid code */
718 r.e = (cab_UBYTE)(*p < 256 ? 16 : 15); /* 256 is end-of-block code */
719 r.v.n = *p++; /* simple code is just the value */
723 r.e = (cab_UBYTE)e[*p - s]; /* non-simple--look up in lists */
727 /* fill code-like entries with r */
729 for (j = i >> w; j < z; j += f)
732 /* backwards increment the k-bit code i */
733 for (j = 1 << (k - 1); i & j; j >>= 1)
737 /* backup over finished tables */
738 while ((i & ((1 << w) - 1)) != ZIP(x)[h])
739 w -= l[--h]; /* don't need to update q */
743 /* return actual size of base table */
746 /* Return true (1) if we were given an incomplete table */
747 return y != 0 && g != 1;
750 /*********************************************************
751 * Zipinflate_codes (internal)
753 static cab_LONG Zipinflate_codes(struct Ziphuft *tl, struct Ziphuft *td,
754 cab_LONG bl, cab_LONG bd, cab_decomp_state *decomp_state)
756 register cab_ULONG e; /* table entry flag/number of extra bits */
757 cab_ULONG n, d; /* length and index for copy */
758 cab_ULONG w; /* current window position */
759 struct Ziphuft *t; /* pointer to table entry */
760 cab_ULONG ml, md; /* masks for bl and bd bits */
761 register cab_ULONG b; /* bit buffer */
762 register cab_ULONG k; /* number of bits in bit buffer */
764 /* make local copies of globals */
765 b = ZIP(bb); /* initialize bit buffer */
767 w = ZIP(window_posn); /* initialize window position */
769 /* inflate the coded data */
770 ml = Zipmask[bl]; /* precompute masks for speed */
775 ZIPNEEDBITS((cab_ULONG)bl)
776 if((e = (t = tl + ((cab_ULONG)b & ml))->e) > 16)
784 } while ((e = (t = t->v.t + ((cab_ULONG)b & Zipmask[e]))->e) > 16);
786 if (e == 16) /* then it's a literal */
787 CAB(outbuf)[w++] = (cab_UBYTE)t->v.n;
788 else /* it's an EOB or a length */
790 /* exit if end of block */
794 /* get length of block to copy */
796 n = t->v.n + ((cab_ULONG)b & Zipmask[e]);
799 /* decode distance of block to copy */
800 ZIPNEEDBITS((cab_ULONG)bd)
801 if ((e = (t = td + ((cab_ULONG)b & md))->e) > 16)
808 } while ((e = (t = t->v.t + ((cab_ULONG)b & Zipmask[e]))->e) > 16);
811 d = w - t->v.n - ((cab_ULONG)b & Zipmask[e]);
815 n -= (e = (e = ZIPWSIZE - ((d &= ZIPWSIZE-1) > w ? d : w)) > n ?n:e);
818 CAB(outbuf)[w++] = CAB(outbuf)[d++];
824 /* restore the globals from the locals */
825 ZIP(window_posn) = w; /* restore global window pointer */
826 ZIP(bb) = b; /* restore global bit buffer */
833 /***********************************************************
834 * Zipinflate_stored (internal)
836 static cab_LONG Zipinflate_stored(cab_decomp_state *decomp_state)
837 /* "decompress" an inflated type 0 (stored) block. */
839 cab_ULONG n; /* number of bytes in block */
840 cab_ULONG w; /* current window position */
841 register cab_ULONG b; /* bit buffer */
842 register cab_ULONG k; /* number of bits in bit buffer */
844 /* make local copies of globals */
845 b = ZIP(bb); /* initialize bit buffer */
847 w = ZIP(window_posn); /* initialize window position */
849 /* go to byte boundary */
853 /* get the length and its complement */
855 n = ((cab_ULONG)b & 0xffff);
858 if (n != (cab_ULONG)((~b) & 0xffff))
859 return 1; /* error in compressed data */
862 /* read and output the compressed data */
866 CAB(outbuf)[w++] = (cab_UBYTE)b;
870 /* restore the globals from the locals */
871 ZIP(window_posn) = w; /* restore global window pointer */
872 ZIP(bb) = b; /* restore global bit buffer */
877 /******************************************************
878 * Zipinflate_fixed (internal)
880 static cab_LONG Zipinflate_fixed(cab_decomp_state *decomp_state)
882 struct Ziphuft *fixed_tl;
883 struct Ziphuft *fixed_td;
884 cab_LONG fixed_bl, fixed_bd;
885 cab_LONG i; /* temporary variable */
891 for(i = 0; i < 144; i++)
897 for(; i < 288; i++) /* make a complete, but wrong code set */
900 if((i = Ziphuft_build(l, 288, 257, (cab_UWORD *) Zipcplens,
901 (cab_UWORD *) Zipcplext, &fixed_tl, &fixed_bl, decomp_state)))
905 for(i = 0; i < 30; i++) /* make an incomplete code set */
908 if((i = Ziphuft_build(l, 30, 0, (cab_UWORD *) Zipcpdist, (cab_UWORD *) Zipcpdext,
909 &fixed_td, &fixed_bd, decomp_state)) > 1)
911 Ziphuft_free(fixed_tl);
915 /* decompress until an end-of-block code */
916 i = Zipinflate_codes(fixed_tl, fixed_td, fixed_bl, fixed_bd, decomp_state);
918 Ziphuft_free(fixed_td);
919 Ziphuft_free(fixed_tl);
923 /**************************************************************
924 * Zipinflate_dynamic (internal)
926 static cab_LONG Zipinflate_dynamic(cab_decomp_state *decomp_state)
927 /* decompress an inflated type 2 (dynamic Huffman codes) block. */
929 cab_LONG i; /* temporary variables */
932 cab_ULONG l; /* last length */
933 cab_ULONG m; /* mask for bit lengths table */
934 cab_ULONG n; /* number of lengths to get */
935 struct Ziphuft *tl; /* literal/length code table */
936 struct Ziphuft *td; /* distance code table */
937 cab_LONG bl; /* lookup bits for tl */
938 cab_LONG bd; /* lookup bits for td */
939 cab_ULONG nb; /* number of bit length codes */
940 cab_ULONG nl; /* number of literal/length codes */
941 cab_ULONG nd; /* number of distance codes */
942 register cab_ULONG b; /* bit buffer */
943 register cab_ULONG k; /* number of bits in bit buffer */
945 /* make local bit buffer */
950 /* read in table lengths */
952 nl = 257 + ((cab_ULONG)b & 0x1f); /* number of literal/length codes */
955 nd = 1 + ((cab_ULONG)b & 0x1f); /* number of distance codes */
958 nb = 4 + ((cab_ULONG)b & 0xf); /* number of bit length codes */
960 if(nl > 288 || nd > 32)
961 return 1; /* bad lengths */
963 /* read in bit-length-code lengths */
964 for(j = 0; j < nb; j++)
967 ll[Zipborder[j]] = (cab_ULONG)b & 7;
971 ll[Zipborder[j]] = 0;
973 /* build decoding table for trees--single level, 7 bit lookup */
975 if((i = Ziphuft_build(ll, 19, 19, NULL, NULL, &tl, &bl, decomp_state)) != 0)
979 return i; /* incomplete code set */
982 /* read in literal and distance code lengths */
986 while((cab_ULONG)i < n)
988 ZIPNEEDBITS((cab_ULONG)bl)
989 j = (td = tl + ((cab_ULONG)b & m))->b;
992 if (j < 16) /* length of code in bits (0..15) */
993 ll[i++] = l = j; /* save last length in l */
994 else if (j == 16) /* repeat last length 3 to 6 times */
997 j = 3 + ((cab_ULONG)b & 3);
999 if((cab_ULONG)i + j > n)
1004 else if (j == 17) /* 3 to 10 zero length codes */
1007 j = 3 + ((cab_ULONG)b & 7);
1009 if ((cab_ULONG)i + j > n)
1015 else /* j == 18: 11 to 138 zero length codes */
1018 j = 11 + ((cab_ULONG)b & 0x7f);
1020 if ((cab_ULONG)i + j > n)
1028 /* free decoding table for trees */
1031 /* restore the global bit buffer */
1035 /* build the decoding tables for literal/length and distance codes */
1037 if((i = Ziphuft_build(ll, nl, 257, (cab_UWORD *) Zipcplens, (cab_UWORD *) Zipcplext,
1038 &tl, &bl, decomp_state)) != 0)
1042 return i; /* incomplete code set */
1045 Ziphuft_build(ll + nl, nd, 0, (cab_UWORD *) Zipcpdist, (cab_UWORD *) Zipcpdext,
1046 &td, &bd, decomp_state);
1048 /* decompress until an end-of-block code */
1049 if(Zipinflate_codes(tl, td, bl, bd, decomp_state))
1052 /* free the decoding tables, return */
1058 /*****************************************************
1059 * Zipinflate_block (internal)
1061 static cab_LONG Zipinflate_block(cab_LONG *e, cab_decomp_state *decomp_state) /* e == last block flag */
1062 { /* decompress an inflated block */
1063 cab_ULONG t; /* block type */
1064 register cab_ULONG b; /* bit buffer */
1065 register cab_ULONG k; /* number of bits in bit buffer */
1067 /* make local bit buffer */
1071 /* read in last block bit */
1073 *e = (cab_LONG)b & 1;
1076 /* read in block type */
1078 t = (cab_ULONG)b & 3;
1081 /* restore the global bit buffer */
1085 /* inflate that block type */
1087 return Zipinflate_dynamic(decomp_state);
1089 return Zipinflate_stored(decomp_state);
1091 return Zipinflate_fixed(decomp_state);
1092 /* bad block type */
1096 /****************************************************
1097 * ZIPdecompress (internal)
1099 static int ZIPdecompress(int inlen, int outlen, cab_decomp_state *decomp_state)
1101 cab_LONG e; /* last block flag */
1103 TRACE("(inlen == %d, outlen == %d)\n", inlen, outlen);
1105 ZIP(inpos) = CAB(inbuf);
1106 ZIP(bb) = ZIP(bk) = ZIP(window_posn) = 0;
1107 if(outlen > ZIPWSIZE)
1108 return DECR_DATAFORMAT;
1110 /* CK = Chris Kirmse, official Microsoft purloiner */
1111 if(ZIP(inpos)[0] != 0x43 || ZIP(inpos)[1] != 0x4B)
1112 return DECR_ILLEGALDATA;
1117 if(Zipinflate_block(&e, decomp_state))
1118 return DECR_ILLEGALDATA;
1121 /* return success */
1125 /* Quantum decruncher */
1127 /* This decruncher was researched and implemented by Matthew Russoto. */
1128 /* It has since been tidied up by Stuart Caie */
1130 /******************************************************************
1131 * QTMinitmodel (internal)
1133 * Initialise a model which decodes symbols from [s] to [s]+[n]-1
1135 static void QTMinitmodel(struct QTMmodel *m, struct QTMmodelsym *sym, int n, int s) {
1140 memset(m->tabloc, 0xFF, sizeof(m->tabloc)); /* clear out look-up table */
1141 for (i = 0; i < n; i++) {
1142 m->tabloc[i+s] = i; /* set up a look-up entry for symbol */
1143 m->syms[i].sym = i+s; /* actual symbol */
1144 m->syms[i].cumfreq = n-i; /* current frequency of that symbol */
1146 m->syms[n].cumfreq = 0;
1149 /******************************************************************
1150 * QTMinit (internal)
1152 static int QTMinit(int window, int level, cab_decomp_state *decomp_state) {
1153 unsigned int wndsize = 1 << window;
1154 int msz = window * 2, i;
1157 /* QTM supports window sizes of 2^10 (1Kb) through 2^21 (2Mb) */
1158 /* if a previously allocated window is big enough, keep it */
1159 if (window < 10 || window > 21) return DECR_DATAFORMAT;
1160 if (QTM(actual_size) < wndsize) {
1161 if (QTM(window)) free(QTM(window));
1165 if (!(QTM(window) = malloc(wndsize))) return DECR_NOMEMORY;
1166 QTM(actual_size) = wndsize;
1168 QTM(window_size) = wndsize;
1169 QTM(window_posn) = 0;
1171 /* initialise static slot/extrabits tables */
1172 for (i = 0, j = 0; i < 27; i++) {
1173 CAB(q_length_extra)[i] = (i == 26) ? 0 : (i < 2 ? 0 : i - 2) >> 2;
1174 CAB(q_length_base)[i] = j; j += 1 << ((i == 26) ? 5 : CAB(q_length_extra)[i]);
1176 for (i = 0, j = 0; i < 42; i++) {
1177 CAB(q_extra_bits)[i] = (i < 2 ? 0 : i-2) >> 1;
1178 CAB(q_position_base)[i] = j; j += 1 << CAB(q_extra_bits)[i];
1181 /* initialise arithmetic coding models */
1183 QTMinitmodel(&QTM(model7), &QTM(m7sym)[0], 7, 0);
1185 QTMinitmodel(&QTM(model00), &QTM(m00sym)[0], 0x40, 0x00);
1186 QTMinitmodel(&QTM(model40), &QTM(m40sym)[0], 0x40, 0x40);
1187 QTMinitmodel(&QTM(model80), &QTM(m80sym)[0], 0x40, 0x80);
1188 QTMinitmodel(&QTM(modelC0), &QTM(mC0sym)[0], 0x40, 0xC0);
1190 /* model 4 depends on table size, ranges from 20 to 24 */
1191 QTMinitmodel(&QTM(model4), &QTM(m4sym)[0], (msz < 24) ? msz : 24, 0);
1192 /* model 5 depends on table size, ranges from 20 to 36 */
1193 QTMinitmodel(&QTM(model5), &QTM(m5sym)[0], (msz < 36) ? msz : 36, 0);
1194 /* model 6pos depends on table size, ranges from 20 to 42 */
1195 QTMinitmodel(&QTM(model6pos), &QTM(m6psym)[0], msz, 0);
1196 QTMinitmodel(&QTM(model6len), &QTM(m6lsym)[0], 27, 0);
1201 /****************************************************************
1202 * QTMupdatemodel (internal)
1204 void QTMupdatemodel(struct QTMmodel *model, int sym) {
1205 struct QTMmodelsym temp;
1208 for (i = 0; i < sym; i++) model->syms[i].cumfreq += 8;
1210 if (model->syms[0].cumfreq > 3800) {
1211 if (--model->shiftsleft) {
1212 for (i = model->entries - 1; i >= 0; i--) {
1213 /* -1, not -2; the 0 entry saves this */
1214 model->syms[i].cumfreq >>= 1;
1215 if (model->syms[i].cumfreq <= model->syms[i+1].cumfreq) {
1216 model->syms[i].cumfreq = model->syms[i+1].cumfreq + 1;
1221 model->shiftsleft = 50;
1222 for (i = 0; i < model->entries ; i++) {
1223 /* no -1, want to include the 0 entry */
1224 /* this converts cumfreqs into frequencies, then shifts right */
1225 model->syms[i].cumfreq -= model->syms[i+1].cumfreq;
1226 model->syms[i].cumfreq++; /* avoid losing things entirely */
1227 model->syms[i].cumfreq >>= 1;
1230 /* now sort by frequencies, decreasing order -- this must be an
1231 * inplace selection sort, or a sort with the same (in)stability
1234 for (i = 0; i < model->entries - 1; i++) {
1235 for (j = i + 1; j < model->entries; j++) {
1236 if (model->syms[i].cumfreq < model->syms[j].cumfreq) {
1237 temp = model->syms[i];
1238 model->syms[i] = model->syms[j];
1239 model->syms[j] = temp;
1244 /* then convert frequencies back to cumfreq */
1245 for (i = model->entries - 1; i >= 0; i--) {
1246 model->syms[i].cumfreq += model->syms[i+1].cumfreq;
1248 /* then update the other part of the table */
1249 for (i = 0; i < model->entries; i++) {
1250 model->tabloc[model->syms[i].sym] = i;
1256 /*******************************************************************
1257 * QTMdecompress (internal)
1259 static int QTMdecompress(int inlen, int outlen, cab_decomp_state *decomp_state)
1261 cab_UBYTE *inpos = CAB(inbuf);
1262 cab_UBYTE *window = QTM(window);
1263 cab_UBYTE *runsrc, *rundest;
1265 cab_ULONG window_posn = QTM(window_posn);
1266 cab_ULONG window_size = QTM(window_size);
1268 /* used by bitstream macros */
1269 register int bitsleft, bitrun, bitsneed;
1270 register cab_ULONG bitbuf;
1272 /* used by GET_SYMBOL */
1277 int extra, togo = outlen, match_length = 0, copy_length;
1278 cab_UBYTE selector, sym;
1279 cab_ULONG match_offset = 0;
1281 cab_UWORD H = 0xFFFF, L = 0, C;
1283 TRACE("(inlen == %d, outlen == %d)\n", inlen, outlen);
1285 /* read initial value of C */
1289 /* apply 2^x-1 mask */
1290 window_posn &= window_size - 1;
1291 /* runs can't straddle the window wraparound */
1292 if ((window_posn + togo) > window_size) {
1293 TRACE("straddled run\n");
1294 return DECR_DATAFORMAT;
1298 GET_SYMBOL(model7, selector);
1301 GET_SYMBOL(model00, sym); window[window_posn++] = sym; togo--;
1304 GET_SYMBOL(model40, sym); window[window_posn++] = sym; togo--;
1307 GET_SYMBOL(model80, sym); window[window_posn++] = sym; togo--;
1310 GET_SYMBOL(modelC0, sym); window[window_posn++] = sym; togo--;
1314 /* selector 4 = fixed length of 3 */
1315 GET_SYMBOL(model4, sym);
1316 Q_READ_BITS(extra, CAB(q_extra_bits)[sym]);
1317 match_offset = CAB(q_position_base)[sym] + extra + 1;
1322 /* selector 5 = fixed length of 4 */
1323 GET_SYMBOL(model5, sym);
1324 Q_READ_BITS(extra, CAB(q_extra_bits)[sym]);
1325 match_offset = CAB(q_position_base)[sym] + extra + 1;
1330 /* selector 6 = variable length */
1331 GET_SYMBOL(model6len, sym);
1332 Q_READ_BITS(extra, CAB(q_length_extra)[sym]);
1333 match_length = CAB(q_length_base)[sym] + extra + 5;
1334 GET_SYMBOL(model6pos, sym);
1335 Q_READ_BITS(extra, CAB(q_extra_bits)[sym]);
1336 match_offset = CAB(q_position_base)[sym] + extra + 1;
1340 TRACE("Selector is bogus\n");
1341 return DECR_ILLEGALDATA;
1344 /* if this is a match */
1345 if (selector >= 4) {
1346 rundest = window + window_posn;
1347 togo -= match_length;
1349 /* copy any wrapped around source data */
1350 if (window_posn >= match_offset) {
1352 runsrc = rundest - match_offset;
1354 runsrc = rundest + (window_size - match_offset);
1355 copy_length = match_offset - window_posn;
1356 if (copy_length < match_length) {
1357 match_length -= copy_length;
1358 window_posn += copy_length;
1359 while (copy_length-- > 0) *rundest++ = *runsrc++;
1363 window_posn += match_length;
1365 /* copy match data - no worries about destination wraps */
1366 while (match_length-- > 0) *rundest++ = *runsrc++;
1368 } /* while (togo > 0) */
1371 TRACE("Frame overflow, this_run = %d\n", togo);
1372 return DECR_ILLEGALDATA;
1375 memcpy(CAB(outbuf), window + ((!window_posn) ? window_size : window_posn) -
1378 QTM(window_posn) = window_posn;
1382 /* LZX decruncher */
1384 /* Microsoft's LZX document and their implementation of the
1385 * com.ms.util.cab Java package do not concur.
1387 * In the LZX document, there is a table showing the correlation between
1388 * window size and the number of position slots. It states that the 1MB
1389 * window = 40 slots and the 2MB window = 42 slots. In the implementation,
1390 * 1MB = 42 slots, 2MB = 50 slots. The actual calculation is 'find the
1391 * first slot whose position base is equal to or more than the required
1392 * window size'. This would explain why other tables in the document refer
1393 * to 50 slots rather than 42.
1395 * The constant NUM_PRIMARY_LENGTHS used in the decompression pseudocode
1396 * is not defined in the specification.
1398 * The LZX document does not state the uncompressed block has an
1399 * uncompressed length field. Where does this length field come from, so
1400 * we can know how large the block is? The implementation has it as the 24
1401 * bits following after the 3 blocktype bits, before the alignment
1404 * The LZX document states that aligned offset blocks have their aligned
1405 * offset huffman tree AFTER the main and length trees. The implementation
1406 * suggests that the aligned offset tree is BEFORE the main and length
1409 * The LZX document decoding algorithm states that, in an aligned offset
1410 * block, if an extra_bits value is 1, 2 or 3, then that number of bits
1411 * should be read and the result added to the match offset. This is
1412 * correct for 1 and 2, but not 3, where just a huffman symbol (using the
1413 * aligned tree) should be read.
1415 * Regarding the E8 preprocessing, the LZX document states 'No translation
1416 * may be performed on the last 6 bytes of the input block'. This is
1417 * correct. However, the pseudocode provided checks for the *E8 leader*
1418 * up to the last 6 bytes. If the leader appears between -10 and -7 bytes
1419 * from the end, this would cause the next four bytes to be modified, at
1420 * least one of which would be in the last 6 bytes, which is not allowed
1421 * according to the spec.
1423 * The specification states that the huffman trees must always contain at
1424 * least one element. However, many CAB files contain blocks where the
1425 * length tree is completely empty (because there are no matches), and
1426 * this is expected to succeed.
1430 /* LZX uses what it calls 'position slots' to represent match offsets.
1431 * What this means is that a small 'position slot' number and a small
1432 * offset from that slot are encoded instead of one large offset for
1434 * - lzx_position_base is an index to the position slot bases
1435 * - lzx_extra_bits states how many bits of offset-from-base data is needed.
1438 /************************************************************
1439 * LZXinit (internal)
1441 static int LZXinit(int window, cab_decomp_state *decomp_state) {
1442 cab_ULONG wndsize = 1 << window;
1443 int i, j, posn_slots;
1445 /* LZX supports window sizes of 2^15 (32Kb) through 2^21 (2Mb) */
1446 /* if a previously allocated window is big enough, keep it */
1447 if (window < 15 || window > 21) return DECR_DATAFORMAT;
1448 if (LZX(actual_size) < wndsize) {
1449 if (LZX(window)) free(LZX(window));
1453 if (!(LZX(window) = malloc(wndsize))) return DECR_NOMEMORY;
1454 LZX(actual_size) = wndsize;
1456 LZX(window_size) = wndsize;
1458 /* initialise static tables */
1459 for (i=0, j=0; i <= 50; i += 2) {
1460 CAB(extra_bits)[i] = CAB(extra_bits)[i+1] = j; /* 0,0,0,0,1,1,2,2,3,3... */
1461 if ((i != 0) && (j < 17)) j++; /* 0,0,1,2,3,4...15,16,17,17,17,17... */
1463 for (i=0, j=0; i <= 50; i++) {
1464 CAB(lzx_position_base)[i] = j; /* 0,1,2,3,4,6,8,12,16,24,32,... */
1465 j += 1 << CAB(extra_bits)[i]; /* 1,1,1,1,2,2,4,4,8,8,16,16,32,32,... */
1468 /* calculate required position slots */
1469 if (window == 20) posn_slots = 42;
1470 else if (window == 21) posn_slots = 50;
1471 else posn_slots = window << 1;
1473 /*posn_slots=i=0; while (i < wndsize) i += 1 << CAB(extra_bits)[posn_slots++]; */
1475 LZX(R0) = LZX(R1) = LZX(R2) = 1;
1476 LZX(main_elements) = LZX_NUM_CHARS + (posn_slots << 3);
1477 LZX(header_read) = 0;
1478 LZX(frames_read) = 0;
1479 LZX(block_remaining) = 0;
1480 LZX(block_type) = LZX_BLOCKTYPE_INVALID;
1481 LZX(intel_curpos) = 0;
1482 LZX(intel_started) = 0;
1483 LZX(window_posn) = 0;
1485 /* initialise tables to 0 (because deltas will be applied to them) */
1486 for (i = 0; i < LZX_MAINTREE_MAXSYMBOLS; i++) LZX(MAINTREE_len)[i] = 0;
1487 for (i = 0; i < LZX_LENGTH_MAXSYMBOLS; i++) LZX(LENGTH_len)[i] = 0;
1492 /*************************************************************************
1493 * make_decode_table (internal)
1495 * This function was coded by David Tritscher. It builds a fast huffman
1496 * decoding table out of just a canonical huffman code lengths table.
1499 * nsyms: total number of symbols in this huffman tree.
1500 * nbits: any symbols with a code length of nbits or less can be decoded
1501 * in one lookup of the table.
1502 * length: A table to get code lengths from [0 to syms-1]
1503 * table: The table to fill up with decoded symbols and pointers.
1509 int make_decode_table(cab_ULONG nsyms, cab_ULONG nbits, cab_UBYTE *length, cab_UWORD *table) {
1510 register cab_UWORD sym;
1511 register cab_ULONG leaf;
1512 register cab_UBYTE bit_num = 1;
1514 cab_ULONG pos = 0; /* the current position in the decode table */
1515 cab_ULONG table_mask = 1 << nbits;
1516 cab_ULONG bit_mask = table_mask >> 1; /* don't do 0 length codes */
1517 cab_ULONG next_symbol = bit_mask; /* base of allocation for long codes */
1519 /* fill entries for codes short enough for a direct mapping */
1520 while (bit_num <= nbits) {
1521 for (sym = 0; sym < nsyms; sym++) {
1522 if (length[sym] == bit_num) {
1525 if((pos += bit_mask) > table_mask) return 1; /* table overrun */
1527 /* fill all possible lookups of this symbol with the symbol itself */
1529 while (fill-- > 0) table[leaf++] = sym;
1536 /* if there are any codes longer than nbits */
1537 if (pos != table_mask) {
1538 /* clear the remainder of the table */
1539 for (sym = pos; sym < table_mask; sym++) table[sym] = 0;
1541 /* give ourselves room for codes to grow by up to 16 more bits */
1546 while (bit_num <= 16) {
1547 for (sym = 0; sym < nsyms; sym++) {
1548 if (length[sym] == bit_num) {
1550 for (fill = 0; fill < bit_num - nbits; fill++) {
1551 /* if this path hasn't been taken yet, 'allocate' two entries */
1552 if (table[leaf] == 0) {
1553 table[(next_symbol << 1)] = 0;
1554 table[(next_symbol << 1) + 1] = 0;
1555 table[leaf] = next_symbol++;
1557 /* follow the path and select either left or right for next bit */
1558 leaf = table[leaf] << 1;
1559 if ((pos >> (15-fill)) & 1) leaf++;
1563 if ((pos += bit_mask) > table_mask) return 1; /* table overflow */
1572 if (pos == table_mask) return 0;
1574 /* either erroneous table, or all elements are 0 - let's find out. */
1575 for (sym = 0; sym < nsyms; sym++) if (length[sym]) return 1;
1579 /************************************************************
1580 * lzx_read_lens (internal)
1582 static int lzx_read_lens(cab_UBYTE *lens, cab_ULONG first, cab_ULONG last, struct lzx_bits *lb,
1583 cab_decomp_state *decomp_state) {
1587 register cab_ULONG bitbuf = lb->bb;
1588 register int bitsleft = lb->bl;
1589 cab_UBYTE *inpos = lb->ip;
1592 for (x = 0; x < 20; x++) {
1594 LENTABLE(PRETREE)[x] = y;
1596 BUILD_TABLE(PRETREE);
1598 for (x = first; x < last; ) {
1599 READ_HUFFSYM(PRETREE, z);
1601 READ_BITS(y, 4); y += 4;
1602 while (y--) lens[x++] = 0;
1605 READ_BITS(y, 5); y += 20;
1606 while (y--) lens[x++] = 0;
1609 READ_BITS(y, 1); y += 4;
1610 READ_HUFFSYM(PRETREE, z);
1611 z = lens[x] - z; if (z < 0) z += 17;
1612 while (y--) lens[x++] = z;
1615 z = lens[x] - z; if (z < 0) z += 17;
1626 /*******************************************************
1627 * LZXdecompress (internal)
1629 static int LZXdecompress(int inlen, int outlen, cab_decomp_state *decomp_state) {
1630 cab_UBYTE *inpos = CAB(inbuf);
1631 cab_UBYTE *endinp = inpos + inlen;
1632 cab_UBYTE *window = LZX(window);
1633 cab_UBYTE *runsrc, *rundest;
1634 cab_UWORD *hufftbl; /* used in READ_HUFFSYM macro as chosen decoding table */
1636 cab_ULONG window_posn = LZX(window_posn);
1637 cab_ULONG window_size = LZX(window_size);
1638 cab_ULONG R0 = LZX(R0);
1639 cab_ULONG R1 = LZX(R1);
1640 cab_ULONG R2 = LZX(R2);
1642 register cab_ULONG bitbuf;
1643 register int bitsleft;
1644 cab_ULONG match_offset, i,j,k; /* ijk used in READ_HUFFSYM macro */
1645 struct lzx_bits lb; /* used in READ_LENGTHS macro */
1647 int togo = outlen, this_run, main_element, aligned_bits;
1648 int match_length, copy_length, length_footer, extra, verbatim_bits;
1650 TRACE("(inlen == %d, outlen == %d)\n", inlen, outlen);
1654 /* read header if necessary */
1655 if (!LZX(header_read)) {
1657 READ_BITS(k, 1); if (k) { READ_BITS(i,16); READ_BITS(j,16); }
1658 LZX(intel_filesize) = (i << 16) | j; /* or 0 if not encoded */
1659 LZX(header_read) = 1;
1662 /* main decoding loop */
1664 /* last block finished, new block expected */
1665 if (LZX(block_remaining) == 0) {
1666 if (LZX(block_type) == LZX_BLOCKTYPE_UNCOMPRESSED) {
1667 if (LZX(block_length) & 1) inpos++; /* realign bitstream to word */
1671 READ_BITS(LZX(block_type), 3);
1674 LZX(block_remaining) = LZX(block_length) = (i << 8) | j;
1676 switch (LZX(block_type)) {
1677 case LZX_BLOCKTYPE_ALIGNED:
1678 for (i = 0; i < 8; i++) { READ_BITS(j, 3); LENTABLE(ALIGNED)[i] = j; }
1679 BUILD_TABLE(ALIGNED);
1680 /* rest of aligned header is same as verbatim */
1682 case LZX_BLOCKTYPE_VERBATIM:
1683 READ_LENGTHS(MAINTREE, 0, 256, lzx_read_lens);
1684 READ_LENGTHS(MAINTREE, 256, LZX(main_elements), lzx_read_lens);
1685 BUILD_TABLE(MAINTREE);
1686 if (LENTABLE(MAINTREE)[0xE8] != 0) LZX(intel_started) = 1;
1688 READ_LENGTHS(LENGTH, 0, LZX_NUM_SECONDARY_LENGTHS, lzx_read_lens);
1689 BUILD_TABLE(LENGTH);
1692 case LZX_BLOCKTYPE_UNCOMPRESSED:
1693 LZX(intel_started) = 1; /* because we can't assume otherwise */
1694 ENSURE_BITS(16); /* get up to 16 pad bits into the buffer */
1695 if (bitsleft > 16) inpos -= 2; /* and align the bitstream! */
1696 R0 = inpos[0]|(inpos[1]<<8)|(inpos[2]<<16)|(inpos[3]<<24);inpos+=4;
1697 R1 = inpos[0]|(inpos[1]<<8)|(inpos[2]<<16)|(inpos[3]<<24);inpos+=4;
1698 R2 = inpos[0]|(inpos[1]<<8)|(inpos[2]<<16)|(inpos[3]<<24);inpos+=4;
1702 return DECR_ILLEGALDATA;
1706 /* buffer exhaustion check */
1707 if (inpos > endinp) {
1708 /* it's possible to have a file where the next run is less than
1709 * 16 bits in size. In this case, the READ_HUFFSYM() macro used
1710 * in building the tables will exhaust the buffer, so we should
1711 * allow for this, but not allow those accidentally read bits to
1712 * be used (so we check that there are at least 16 bits
1713 * remaining - in this boundary case they aren't really part of
1714 * the compressed data)
1716 if (inpos > (endinp+2) || bitsleft < 16) return DECR_ILLEGALDATA;
1719 while ((this_run = LZX(block_remaining)) > 0 && togo > 0) {
1720 if (this_run > togo) this_run = togo;
1722 LZX(block_remaining) -= this_run;
1724 /* apply 2^x-1 mask */
1725 window_posn &= window_size - 1;
1726 /* runs can't straddle the window wraparound */
1727 if ((window_posn + this_run) > window_size)
1728 return DECR_DATAFORMAT;
1730 switch (LZX(block_type)) {
1732 case LZX_BLOCKTYPE_VERBATIM:
1733 while (this_run > 0) {
1734 READ_HUFFSYM(MAINTREE, main_element);
1736 if (main_element < LZX_NUM_CHARS) {
1737 /* literal: 0 to LZX_NUM_CHARS-1 */
1738 window[window_posn++] = main_element;
1742 /* match: LZX_NUM_CHARS + ((slot<<3) | length_header (3 bits)) */
1743 main_element -= LZX_NUM_CHARS;
1745 match_length = main_element & LZX_NUM_PRIMARY_LENGTHS;
1746 if (match_length == LZX_NUM_PRIMARY_LENGTHS) {
1747 READ_HUFFSYM(LENGTH, length_footer);
1748 match_length += length_footer;
1750 match_length += LZX_MIN_MATCH;
1752 match_offset = main_element >> 3;
1754 if (match_offset > 2) {
1755 /* not repeated offset */
1756 if (match_offset != 3) {
1757 extra = CAB(extra_bits)[match_offset];
1758 READ_BITS(verbatim_bits, extra);
1759 match_offset = CAB(lzx_position_base)[match_offset]
1760 - 2 + verbatim_bits;
1766 /* update repeated offset LRU queue */
1767 R2 = R1; R1 = R0; R0 = match_offset;
1769 else if (match_offset == 0) {
1772 else if (match_offset == 1) {
1774 R1 = R0; R0 = match_offset;
1776 else /* match_offset == 2 */ {
1778 R2 = R0; R0 = match_offset;
1781 rundest = window + window_posn;
1782 this_run -= match_length;
1784 /* copy any wrapped around source data */
1785 if (window_posn >= match_offset) {
1787 runsrc = rundest - match_offset;
1789 runsrc = rundest + (window_size - match_offset);
1790 copy_length = match_offset - window_posn;
1791 if (copy_length < match_length) {
1792 match_length -= copy_length;
1793 window_posn += copy_length;
1794 while (copy_length-- > 0) *rundest++ = *runsrc++;
1798 window_posn += match_length;
1800 /* copy match data - no worries about destination wraps */
1801 while (match_length-- > 0) *rundest++ = *runsrc++;
1806 case LZX_BLOCKTYPE_ALIGNED:
1807 while (this_run > 0) {
1808 READ_HUFFSYM(MAINTREE, main_element);
1810 if (main_element < LZX_NUM_CHARS) {
1811 /* literal: 0 to LZX_NUM_CHARS-1 */
1812 window[window_posn++] = main_element;
1816 /* match: LZX_NUM_CHARS + ((slot<<3) | length_header (3 bits)) */
1817 main_element -= LZX_NUM_CHARS;
1819 match_length = main_element & LZX_NUM_PRIMARY_LENGTHS;
1820 if (match_length == LZX_NUM_PRIMARY_LENGTHS) {
1821 READ_HUFFSYM(LENGTH, length_footer);
1822 match_length += length_footer;
1824 match_length += LZX_MIN_MATCH;
1826 match_offset = main_element >> 3;
1828 if (match_offset > 2) {
1829 /* not repeated offset */
1830 extra = CAB(extra_bits)[match_offset];
1831 match_offset = CAB(lzx_position_base)[match_offset] - 2;
1833 /* verbatim and aligned bits */
1835 READ_BITS(verbatim_bits, extra);
1836 match_offset += (verbatim_bits << 3);
1837 READ_HUFFSYM(ALIGNED, aligned_bits);
1838 match_offset += aligned_bits;
1840 else if (extra == 3) {
1841 /* aligned bits only */
1842 READ_HUFFSYM(ALIGNED, aligned_bits);
1843 match_offset += aligned_bits;
1845 else if (extra > 0) { /* extra==1, extra==2 */
1846 /* verbatim bits only */
1847 READ_BITS(verbatim_bits, extra);
1848 match_offset += verbatim_bits;
1850 else /* extra == 0 */ {
1855 /* update repeated offset LRU queue */
1856 R2 = R1; R1 = R0; R0 = match_offset;
1858 else if (match_offset == 0) {
1861 else if (match_offset == 1) {
1863 R1 = R0; R0 = match_offset;
1865 else /* match_offset == 2 */ {
1867 R2 = R0; R0 = match_offset;
1870 rundest = window + window_posn;
1871 this_run -= match_length;
1873 /* copy any wrapped around source data */
1874 if (window_posn >= match_offset) {
1876 runsrc = rundest - match_offset;
1878 runsrc = rundest + (window_size - match_offset);
1879 copy_length = match_offset - window_posn;
1880 if (copy_length < match_length) {
1881 match_length -= copy_length;
1882 window_posn += copy_length;
1883 while (copy_length-- > 0) *rundest++ = *runsrc++;
1887 window_posn += match_length;
1889 /* copy match data - no worries about destination wraps */
1890 while (match_length-- > 0) *rundest++ = *runsrc++;
1895 case LZX_BLOCKTYPE_UNCOMPRESSED:
1896 if ((inpos + this_run) > endinp) return DECR_ILLEGALDATA;
1897 memcpy(window + window_posn, inpos, (size_t) this_run);
1898 inpos += this_run; window_posn += this_run;
1902 return DECR_ILLEGALDATA; /* might as well */
1908 if (togo != 0) return DECR_ILLEGALDATA;
1909 memcpy(CAB(outbuf), window + ((!window_posn) ? window_size : window_posn) -
1910 outlen, (size_t) outlen);
1912 LZX(window_posn) = window_posn;
1917 /* intel E8 decoding */
1918 if ((LZX(frames_read)++ < 32768) && LZX(intel_filesize) != 0) {
1919 if (outlen <= 6 || !LZX(intel_started)) {
1920 LZX(intel_curpos) += outlen;
1923 cab_UBYTE *data = CAB(outbuf);
1924 cab_UBYTE *dataend = data + outlen - 10;
1925 cab_LONG curpos = LZX(intel_curpos);
1926 cab_LONG filesize = LZX(intel_filesize);
1927 cab_LONG abs_off, rel_off;
1929 LZX(intel_curpos) = curpos + outlen;
1931 while (data < dataend) {
1932 if (*data++ != 0xE8) { curpos++; continue; }
1933 abs_off = data[0] | (data[1]<<8) | (data[2]<<16) | (data[3]<<24);
1934 if ((abs_off >= -curpos) && (abs_off < filesize)) {
1935 rel_off = (abs_off >= 0) ? abs_off - curpos : abs_off + filesize;
1936 data[0] = (cab_UBYTE) rel_off;
1937 data[1] = (cab_UBYTE) (rel_off >> 8);
1938 data[2] = (cab_UBYTE) (rel_off >> 16);
1939 data[3] = (cab_UBYTE) (rel_off >> 24);
1949 /*********************************************************
1950 * find_cabs_in_file (internal)
1952 static struct cabinet *find_cabs_in_file(LPCSTR name, cab_UBYTE search_buf[])
1954 struct cabinet *cab, *cab2, *firstcab = NULL, *linkcab = NULL;
1955 cab_UBYTE *pstart = &search_buf[0], *pend, *p;
1956 cab_off_t offset, caboff, cablen = 0, foffset = 0, filelen, length;
1957 int state = 0, found = 0, ok = 0;
1959 TRACE("(name == %s)\n", debugstr_a(name));
1961 /* open the file and search for cabinet headers */
1962 if ((cab = (struct cabinet *) calloc(1, sizeof(struct cabinet)))) {
1963 cab->filename = name;
1964 if (cabinet_open(cab)) {
1965 filelen = cab->filelen;
1966 for (offset = 0; (offset < filelen); offset += length) {
1967 /* search length is either the full length of the search buffer,
1968 * or the amount of data remaining to the end of the file,
1969 * whichever is less.
1971 length = filelen - offset;
1972 if (length > CAB_SEARCH_SIZE) length = CAB_SEARCH_SIZE;
1974 /* fill the search buffer with data from disk */
1975 if (!cabinet_read(cab, search_buf, length)) break;
1977 /* read through the entire buffer. */
1979 pend = &search_buf[length];
1982 /* starting state */
1984 /* we spend most of our time in this while loop, looking for
1985 * a leading 'M' of the 'MSCF' signature
1987 while (*p++ != 0x4D && p < pend);
1988 if (p < pend) state = 1; /* if we found tht 'M', advance state */
1991 /* verify that the next 3 bytes are 'S', 'C' and 'F' */
1992 case 1: state = (*p++ == 0x53) ? 2 : 0; break;
1993 case 2: state = (*p++ == 0x43) ? 3 : 0; break;
1994 case 3: state = (*p++ == 0x46) ? 4 : 0; break;
1996 /* we don't care about bytes 4-7 */
1997 /* bytes 8-11 are the overall length of the cabinet */
1998 case 8: cablen = *p++; state++; break;
1999 case 9: cablen |= *p++ << 8; state++; break;
2000 case 10: cablen |= *p++ << 16; state++; break;
2001 case 11: cablen |= *p++ << 24; state++; break;
2003 /* we don't care about bytes 12-15 */
2004 /* bytes 16-19 are the offset within the cabinet of the filedata */
2005 case 16: foffset = *p++; state++; break;
2006 case 17: foffset |= *p++ << 8; state++; break;
2007 case 18: foffset |= *p++ << 16; state++; break;
2008 case 19: foffset |= *p++ << 24;
2009 /* now we have received 20 bytes of potential cab header. */
2010 /* work out the offset in the file of this potential cabinet */
2011 caboff = offset + (p-pstart) - 20;
2013 /* check that the files offset is less than the alleged length
2014 * of the cabinet, and that the offset + the alleged length are
2015 * 'roughly' within the end of overall file length
2017 if ((foffset < cablen) &&
2018 ((caboff + foffset) < (filelen + 32)) &&
2019 ((caboff + cablen) < (filelen + 32)) )
2021 /* found a potential result - try loading it */
2023 cab2 = load_cab_offset(name, caboff);
2028 /* cause the search to restart after this cab's data. */
2029 offset = caboff + cablen;
2030 if (offset < cab->filelen) cabinet_seek(cab, offset);
2034 /* link the cab into the list */
2035 if (linkcab == NULL) firstcab = cab2;
2036 else linkcab->next = cab2;
2043 p++, state++; break;
2052 /* if there were cabinets that were found but are not ok, point this out */
2054 WARN("%s: found %d bad cabinets\n", debugstr_a(name), found-ok);
2057 /* if no cabinets were found, let the user know */
2059 WARN("%s: not a Microsoft cabinet file.\n", debugstr_a(name));
2064 /***********************************************************************
2065 * find_cabinet_file (internal)
2067 * tries to find *cabname, from the directory path of origcab, correcting the
2068 * case of *cabname if necessary, If found, writes back to *cabname.
2070 static void find_cabinet_file(char **cabname, LPCSTR origcab) {
2072 char *tail, *cab, *name, *nextpart, nametmp[MAX_PATH];
2075 TRACE("(*cabname == ^%p, origcab == %s)\n", cabname ? *cabname : NULL, debugstr_a(origcab));
2077 /* ensure we have a cabinet name at all */
2078 if (!(name = *cabname)) {
2079 WARN("no cabinet name at all\n");
2082 /* find if there's a directory path in the origcab */
2083 tail = origcab ? max(strrchr(origcab, '/'), strrchr(origcab, '\\')) : NULL;
2085 if ((cab = (char *) malloc(MAX_PATH))) {
2086 /* add the directory path from the original cabinet name */
2088 memcpy(cab, origcab, tail - origcab);
2089 cab[tail - origcab] = '\0';
2091 /* default directory path of '.' */
2097 TRACE("trying cab == %s\n", debugstr_a(cab));
2099 /* we don't want null cabinet filenames */
2100 if (name[0] == '\0') {
2101 WARN("null cab name\n");
2105 /* if there is a directory component in the cabinet name,
2106 * look for that alone first
2108 nextpart = strchr(name, '\\');
2109 if (nextpart) *nextpart = '\0';
2111 found = SearchPathA(cab, name, NULL, MAX_PATH, nametmp, NULL);
2113 /* if the component was not found, look for it in the current dir */
2115 found = SearchPathA(".", name, NULL, MAX_PATH, nametmp, NULL);
2119 TRACE("found: %s\n", debugstr_a(nametmp));
2121 TRACE("not found.\n");
2123 /* restore the real name and skip to the next directory component
2124 * or actual cabinet name
2126 if (nextpart) *nextpart = '\\', name = &nextpart[1];
2128 /* while there is another directory component, and while we
2129 * successfully found the current component
2131 } while (nextpart && found);
2133 /* if we found the cabinet, change the next cabinet's name.
2134 * otherwise, pretend nothing happened
2137 free((void *) *cabname);
2139 memcpy(cab, nametmp, found+1);
2140 TRACE("result: %s\n", debugstr_a(cab));
2143 TRACE("result: nothing\n");
2148 /************************************************************************
2149 * process_files (internal)
2151 * this does the tricky job of running through every file in the cabinet,
2152 * including spanning cabinets, and working out which file is in which
2153 * folder in which cabinet. It also throws out the duplicate file entries
2154 * that appear in spanning cabinets. There is memory leakage here because
2155 * those entries are not freed. See the XAD CAB client (function CAB_GetInfo
2156 * in CAB.c) for an implementation of this that correctly frees the discarded
2159 static struct cab_file *process_files(struct cabinet *basecab) {
2160 struct cabinet *cab;
2161 struct cab_file *outfi = NULL, *linkfi = NULL, *nextfi, *fi, *cfi;
2162 struct cab_folder *fol, *firstfol, *lastfol = NULL, *predfol;
2165 FIXME("(basecab == ^%p): Memory leak.\n", basecab);
2167 for (cab = basecab; cab; cab = cab->nextcab) {
2168 /* firstfol = first folder in this cabinet */
2169 /* lastfol = last folder in this cabinet */
2170 /* predfol = last folder in previous cabinet (or NULL if first cabinet) */
2172 firstfol = cab->folders;
2173 for (lastfol = firstfol; lastfol->next;) lastfol = lastfol->next;
2176 for (fi = cab->files; fi; fi = nextfi) {
2180 if (i < cffileCONTINUED_FROM_PREV) {
2181 for (fol = firstfol; fol && i--; ) fol = fol->next;
2182 fi->folder = fol; /* NULL if an invalid folder index */
2185 /* folder merging */
2186 if (i == cffileCONTINUED_TO_NEXT
2187 || i == cffileCONTINUED_PREV_AND_NEXT) {
2188 if (cab->nextcab && !lastfol->contfile) lastfol->contfile = fi;
2191 if (i == cffileCONTINUED_FROM_PREV
2192 || i == cffileCONTINUED_PREV_AND_NEXT) {
2193 /* these files are to be continued in yet another
2194 * cabinet, don't merge them in just yet */
2195 if (i == cffileCONTINUED_PREV_AND_NEXT) mergeok = 0;
2197 /* only merge once per cabinet */
2199 if ((cfi = predfol->contfile)
2200 && (cfi->offset == fi->offset)
2201 && (cfi->length == fi->length)
2202 && (strcmp(cfi->filename, fi->filename) == 0)
2203 && (predfol->comp_type == firstfol->comp_type)) {
2204 /* increase the number of splits */
2205 if ((i = ++(predfol->num_splits)) > CAB_SPLITMAX) {
2207 ERR("%s: internal error: CAB_SPLITMAX exceeded. please report this to wine-devel@winehq.org)\n",
2208 debugstr_a(basecab->filename));
2211 /* copy information across from the merged folder */
2212 predfol->offset[i] = firstfol->offset[0];
2213 predfol->cab[i] = firstfol->cab[0];
2214 predfol->next = firstfol->next;
2215 predfol->contfile = firstfol->contfile;
2217 if (firstfol == lastfol) lastfol = predfol;
2219 predfol = NULL; /* don't merge again within this cabinet */
2223 /* if the folders won't merge, don't add their files */
2228 if (mergeok) fi->folder = firstfol;
2233 if (linkfi) linkfi->next = fi; else outfi = fi;
2237 } /* for (cab= ...*/
2242 /****************************************************************
2243 * convertUTF (internal)
2245 * translate UTF -> ASCII
2247 * UTF translates two-byte unicode characters into 1, 2 or 3 bytes.
2248 * %000000000xxxxxxx -> %0xxxxxxx
2249 * %00000xxxxxyyyyyy -> %110xxxxx %10yyyyyy
2250 * %xxxxyyyyyyzzzzzz -> %1110xxxx %10yyyyyy %10zzzzzz
2252 * Therefore, the inverse is as follows:
2254 * 0x00 - 0x7F = one byte char
2255 * 0x80 - 0xBF = invalid
2256 * 0xC0 - 0xDF = 2 byte char (next char only 0x80-0xBF is valid)
2257 * 0xE0 - 0xEF = 3 byte char (next 2 chars only 0x80-0xBF is valid)
2258 * 0xF0 - 0xFF = invalid
2260 * FIXME: use a winapi to do this
2262 static int convertUTF(cab_UBYTE *in) {
2263 cab_UBYTE c, *out = in, *end = in + strlen((char *) in) + 1;
2267 /* read unicode character */
2268 if ((c = *in++) < 0x80) x = c;
2270 if (c < 0xC0) return 0;
2271 else if (c < 0xE0) {
2272 x = (c & 0x1F) << 6;
2273 if ((c = *in++) < 0x80 || c > 0xBF) return 0; else x |= (c & 0x3F);
2275 else if (c < 0xF0) {
2276 x = (c & 0xF) << 12;
2277 if ((c = *in++) < 0x80 || c > 0xBF) return 0; else x |= (c & 0x3F)<<6;
2278 if ((c = *in++) < 0x80 || c > 0xBF) return 0; else x |= (c & 0x3F);
2283 /* terrible unicode -> ASCII conversion */
2284 if (x > 127) x = '_';
2286 if (in > end) return 0; /* just in case */
2287 } while ((*out++ = (cab_UBYTE) x));
2291 /****************************************************
2292 * NONEdecompress (internal)
2294 static int NONEdecompress(int inlen, int outlen, cab_decomp_state *decomp_state)
2296 if (inlen != outlen) return DECR_ILLEGALDATA;
2297 memcpy(CAB(outbuf), CAB(inbuf), (size_t) inlen);
2301 /**************************************************
2302 * checksum (internal)
2304 cab_ULONG checksum(cab_UBYTE *data, cab_UWORD bytes, cab_ULONG csum) {
2308 for (len = bytes >> 2; len--; data += 4) {
2309 csum ^= ((data[0]) | (data[1]<<8) | (data[2]<<16) | (data[3]<<24));
2312 switch (bytes & 3) {
2313 case 3: ul |= *data++ << 16;
2314 case 2: ul |= *data++ << 8;
2315 case 1: ul |= *data;
2322 /**********************************************************
2323 * decompress (internal)
2325 static int decompress(struct cab_file *fi, int savemode, int fix, cab_decomp_state *decomp_state)
2327 cab_ULONG bytes = savemode ? fi->length : fi->offset - CAB(offset);
2328 struct cabinet *cab = CAB(current)->cab[CAB(split)];
2329 cab_UBYTE buf[cfdata_SIZEOF], *data;
2330 cab_UWORD inlen, len, outlen, cando;
2334 TRACE("(fi == ^%p, savemode == %d, fix == %d)\n", fi, savemode, fix);
2337 /* cando = the max number of bytes we can do */
2338 cando = CAB(outlen);
2339 if (cando > bytes) cando = bytes;
2342 if (cando && savemode)
2343 file_write(fi, CAB(outpos), cando);
2345 CAB(outpos) += cando;
2346 CAB(outlen) -= cando;
2347 bytes -= cando; if (!bytes) break;
2349 /* we only get here if we emptied the output buffer */
2351 /* read data header + data */
2353 while (outlen == 0) {
2354 /* read the block header, skip the reserved part */
2355 if (!cabinet_read(cab, buf, cfdata_SIZEOF)) return DECR_INPUT;
2356 cabinet_skip(cab, cab->block_resv);
2358 /* we shouldn't get blocks over CAB_INPUTMAX in size */
2359 data = CAB(inbuf) + inlen;
2360 len = EndGetI16(buf+cfdata_CompressedSize);
2362 if (inlen > CAB_INPUTMAX) return DECR_INPUT;
2363 if (!cabinet_read(cab, data, len)) return DECR_INPUT;
2365 /* clear two bytes after read-in data */
2366 data[len+1] = data[len+2] = 0;
2368 /* perform checksum test on the block (if one is stored) */
2369 cksum = EndGetI32(buf+cfdata_CheckSum);
2370 if (cksum && cksum != checksum(buf+4, 4, checksum(data, len, 0))) {
2371 /* checksum is wrong */
2372 if (fix && ((fi->folder->comp_type & cffoldCOMPTYPE_MASK)
2373 == cffoldCOMPTYPE_MSZIP))
2375 WARN("%s: checksum failed\n", debugstr_a(fi->filename));
2378 return DECR_CHECKSUM;
2382 /* outlen=0 means this block was part of a split block */
2383 outlen = EndGetI16(buf+cfdata_UncompressedSize);
2386 cab = CAB(current)->cab[++CAB(split)];
2387 if (!cabinet_open(cab)) return DECR_INPUT;
2388 cabinet_seek(cab, CAB(current)->offset[CAB(split)]);
2392 /* decompress block */
2393 if ((err = CAB(decompress)(inlen, outlen, decomp_state))) {
2394 if (fix && ((fi->folder->comp_type & cffoldCOMPTYPE_MASK)
2395 == cffoldCOMPTYPE_MSZIP))
2397 ERR("%s: failed decrunching block\n", debugstr_a(fi->filename));
2403 CAB(outlen) = outlen;
2404 CAB(outpos) = CAB(outbuf);
2410 /****************************************************************
2411 * extract_file (internal)
2413 * workhorse to extract a particular file from a cab
2415 static void extract_file(struct cab_file *fi, int lower, int fix, LPCSTR dir, cab_decomp_state *decomp_state)
2417 struct cab_folder *fol = fi->folder, *oldfol = CAB(current);
2418 cab_LONG err = DECR_OK;
2420 TRACE("(fi == ^%p, lower == %d, fix == %d, dir == %s)\n", fi, lower, fix, debugstr_a(dir));
2422 /* is a change of folder needed? do we need to reset the current folder? */
2423 if (fol != oldfol || fi->offset < CAB(offset)) {
2424 cab_UWORD comptype = fol->comp_type;
2425 int ct1 = comptype & cffoldCOMPTYPE_MASK;
2426 int ct2 = oldfol ? (oldfol->comp_type & cffoldCOMPTYPE_MASK) : 0;
2428 /* if the archiver has changed, call the old archiver's free() function */
2431 case cffoldCOMPTYPE_LZX:
2437 case cffoldCOMPTYPE_QUANTUM:
2447 case cffoldCOMPTYPE_NONE:
2448 CAB(decompress) = NONEdecompress;
2451 case cffoldCOMPTYPE_MSZIP:
2452 CAB(decompress) = ZIPdecompress;
2455 case cffoldCOMPTYPE_QUANTUM:
2456 CAB(decompress) = QTMdecompress;
2457 err = QTMinit((comptype >> 8) & 0x1f, (comptype >> 4) & 0xF, decomp_state);
2460 case cffoldCOMPTYPE_LZX:
2461 CAB(decompress) = LZXdecompress;
2462 err = LZXinit((comptype >> 8) & 0x1f, decomp_state);
2466 err = DECR_DATAFORMAT;
2468 if (err) goto exit_handler;
2470 /* initialisation OK, set current folder and reset offset */
2471 if (oldfol) cabinet_close(oldfol->cab[CAB(split)]);
2472 if (!cabinet_open(fol->cab[0])) goto exit_handler;
2473 cabinet_seek(fol->cab[0], fol->offset[0]);
2476 CAB(outlen) = 0; /* discard existing block */
2480 if (fi->offset > CAB(offset)) {
2481 /* decode bytes and send them to /dev/null */
2482 if ((err = decompress(fi, 0, fix, decomp_state))) goto exit_handler;
2483 CAB(offset) = fi->offset;
2486 if (!file_open(fi, lower, dir)) return;
2487 err = decompress(fi, 1, fix, decomp_state);
2488 if (err) CAB(current) = NULL; else CAB(offset) += fi->length;
2494 const char *cabname;
2497 errmsg = "out of memory!\n"; break;
2498 case DECR_ILLEGALDATA:
2499 errmsg = "%s: illegal or corrupt data\n"; break;
2500 case DECR_DATAFORMAT:
2501 errmsg = "%s: unsupported data format\n"; break;
2503 errmsg = "%s: checksum error\n"; break;
2505 errmsg = "%s: input error\n"; break;
2507 errmsg = "%s: output error\n"; break;
2509 errmsg = "%s: unknown error (BUG)\n";
2513 cabname = (CAB(current)->cab[CAB(split)]->filename);
2516 cabname = (fi->folder->cab[0]->filename);
2519 ERR(errmsg, cabname);
2523 /*********************************************************
2524 * print_fileinfo (internal)
2526 static void print_fileinfo(struct cab_file *fi) {
2529 if (fi->attribs & cffile_A_NAME_IS_UTF) {
2530 fname = malloc(strlen(fi->filename) + 1);
2532 strcpy(fname, fi->filename);
2533 convertUTF((cab_UBYTE *) fname);
2537 TRACE("%9u | %02d.%02d.%04d %02d:%02d:%02d | %s\n",
2539 fi->date & 0x1f, (fi->date>>5) & 0xf, (fi->date>>9) + 1980,
2540 fi->time >> 11, (fi->time>>5) & 0x3f, (fi->time << 1) & 0x3e,
2541 fname ? fname : fi->filename
2544 if (fname) free(fname);
2547 /****************************************************************************
2548 * process_cabinet (internal)
2550 * called to simply "extract" a cabinet file. Will find every cabinet file
2551 * in that file, search for every chained cabinet attached to those cabinets,
2552 * and will either extract the cabinets, or ? (call a callback?)
2555 * cabname [I] name of the cabinet file to extract
2556 * dir [I] directory to extract to
2557 * fix [I] attempt to process broken cabinets
2558 * lower [I] ? (lower case something or other?)
2565 BOOL process_cabinet(LPCSTR cabname, LPCSTR dir, BOOL fix, BOOL lower, EXTRACTdest *dest)
2567 struct cabinet *basecab, *cab, *cab1, *cab2;
2568 struct cab_file *filelist, *fi;
2569 struct ExtractFileList **destlistptr = &(dest->filelist);
2571 /* The first result of a search will be returned, and
2572 * the remaining results will be chained to it via the cab->next structure
2575 cab_UBYTE search_buf[CAB_SEARCH_SIZE];
2577 cab_decomp_state decomp_state_local;
2578 cab_decomp_state *decomp_state = &decomp_state_local;
2580 /* has the list-mode header been seen before? */
2583 ZeroMemory(decomp_state, sizeof(cab_decomp_state));
2585 TRACE("Extract %s\n", debugstr_a(cabname));
2587 /* load the file requested */
2588 basecab = find_cabs_in_file(cabname, search_buf);
2589 if (!basecab) return FALSE;
2591 /* iterate over all cabinets found in that file */
2592 for (cab = basecab; cab; cab=cab->next) {
2594 /* bi-directionally load any spanning cabinets -- backwards */
2595 for (cab1 = cab; cab1->flags & cfheadPREV_CABINET; cab1 = cab1->prevcab) {
2596 TRACE("%s: extends backwards to %s (%s)\n", debugstr_a(cabname),
2597 debugstr_a(cab1->prevname), debugstr_a(cab1->previnfo));
2598 find_cabinet_file(&(cab1->prevname), cabname);
2599 if (!(cab1->prevcab = load_cab_offset(cab1->prevname, 0))) {
2600 ERR("%s: can't read previous cabinet %s\n", debugstr_a(cabname), debugstr_a(cab1->prevname));
2603 cab1->prevcab->nextcab = cab1;
2606 /* bi-directionally load any spanning cabinets -- forwards */
2607 for (cab2 = cab; cab2->flags & cfheadNEXT_CABINET; cab2 = cab2->nextcab) {
2608 TRACE("%s: extends to %s (%s)\n", debugstr_a(cabname),
2609 debugstr_a(cab2->nextname), debugstr_a(cab2->nextinfo));
2610 find_cabinet_file(&(cab2->nextname), cabname);
2611 if (!(cab2->nextcab = load_cab_offset(cab2->nextname, 0))) {
2612 ERR("%s: can't read next cabinet %s\n", debugstr_a(cabname), debugstr_a(cab2->nextname));
2615 cab2->nextcab->prevcab = cab2;
2618 filelist = process_files(cab1);
2619 CAB(current) = NULL;
2622 TRACE("File size | Date Time | Name\n");
2623 TRACE("----------+---------------------+-------------\n");
2626 for (fi = filelist; fi; fi = fi->next) {
2630 TRACE("Beginning Extraction...\n");
2631 for (fi = filelist; fi; fi = fi->next) {
2632 TRACE(" extracting: %s\n", debugstr_a(fi->filename));
2633 extract_file(fi, lower, fix, dir, decomp_state);
2634 sprintf(dest->lastfile, "%s%s%s",
2635 strlen(dest->directory) ? dest->directory : "",
2636 strlen(dest->directory) ? "\\": "",
2638 *destlistptr = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY,
2639 sizeof(struct ExtractFileList));
2641 (*destlistptr)->unknown = TRUE; /* FIXME: were do we get the value? */
2642 (*destlistptr)->filename = HeapAlloc(GetProcessHeap(), 0, (
2643 strlen(fi->filename)+1));
2644 if((*destlistptr)->filename)
2645 lstrcpyA((*destlistptr)->filename, fi->filename);
2646 destlistptr = &((*destlistptr)->next);
2651 TRACE("Finished processing cabinet.\n");