/*-----------------------------------------------------------*/ /*--- Block recoverer program for bzip2 ---*/ /*--- bzip2recover.c ---*/ /*-----------------------------------------------------------*/ /*-- Modified for use under GNO by Stephen Heumann --*/ /*-- This program is bzip2recover, a program to attempt data salvage from damaged files created by the accompanying bzip2-1.0 program. Copyright (C) 1996-2002 Julian R Seward. All rights reserved. Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met: 1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer. 2. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required. 3. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software. 4. The name of the author may not be used to endorse or promote products derived from this software without specific prior written permission. THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. Julian Seward, Cambridge, UK. jseward@acm.org bzip2/libbzip2 version 1.0 of 21 March 2000 --*/ /*-- This program is a complete hack and should be rewritten properly. It isn't very complicated. --*/ #include #include #include #include #include #ifdef __appleiigs__ #include #if defined(__GNO__) && defined(__STACK_CHECK__) #include #endif #endif /* This program records bit locations in the file to be recovered. That means that if 64-bit ints are not supported, we will not be able to recover .bz2 files over 512MB (2^32 bits) long. On GNU supported platforms, we take advantage of the 64-bit int support to circumvent this problem. Ditto MSVC. This change occurred in version 1.0.2; all prior versions have the 512MB limitation. */ #ifdef __GNUC__ typedef unsigned long long int MaybeUInt64; # define MaybeUInt64_FMT "%Lu" #else #ifdef _MSC_VER typedef unsigned __int64 MaybeUInt64; # define MaybeUInt64_FMT "%I64u" #else #ifdef __ORCAC__ typedef unsigned long MaybeUInt64; # define MaybeUInt64_FMT "%lu" #else typedef unsigned int MaybeUInt64; # define MaybeUInt64_FMT "%u" #endif #endif #endif #ifdef __ORCAC__ typedef unsigned long UInt32; typedef long Int32; # define Int32_FMT "%ld" # define size_t_FMT "%lu" #else typedef unsigned int UInt32; typedef int Int32; # define Int32_FMT "%d" # define size_t_FMT "%d" #endif /* defined __ORCAC__ */ typedef unsigned char UChar; typedef char Char; typedef unsigned char Bool; #define True ((Bool)1) #define False ((Bool)0) #define BZ_MAX_FILENAME 2000 Char inFileName[BZ_MAX_FILENAME]; Char outFileName[BZ_MAX_FILENAME]; Char progName[BZ_MAX_FILENAME]; MaybeUInt64 bytesOut = 0; MaybeUInt64 bytesIn = 0; /*---------------------------------------------------*/ /*--- Header bytes ---*/ /*---------------------------------------------------*/ #define BZ_HDR_B 0x42 /* 'B' */ #define BZ_HDR_Z 0x5a /* 'Z' */ #define BZ_HDR_h 0x68 /* 'h' */ #define BZ_HDR_0 0x30 /* '0' */ /*---------------------------------------------------*/ /*--- I/O errors ---*/ /*---------------------------------------------------*/ /*---------------------------------------------*/ void readError ( void ) { fprintf ( stderr, "%s: I/O error reading `%s', possible reason follows.\n", progName, inFileName ); perror ( progName ); fprintf ( stderr, "%s: warning: output file(s) may be incomplete.\n", progName ); exit ( 1 ); } /*---------------------------------------------*/ void writeError ( void ) { fprintf ( stderr, "%s: I/O error reading `%s', possible reason follows.\n", progName, inFileName ); perror ( progName ); fprintf ( stderr, "%s: warning: output file(s) may be incomplete.\n", progName ); exit ( 1 ); } /*---------------------------------------------*/ void mallocFail ( Int32 n ) { fprintf ( stderr, "%s: malloc failed on request for " Int32_FMT " bytes.\n", progName, n ); fprintf ( stderr, "%s: warning: output file(s) may be incomplete.\n", progName ); exit ( 1 ); } /*---------------------------------------------*/ void tooManyBlocks ( Int32 max_handled_blocks ) { fprintf ( stderr, "%s: `%s' appears to contain more than " Int32_FMT " blocks\n", progName, inFileName, max_handled_blocks ); fprintf ( stderr, "%s: and cannot be handled. To fix, increase\n", progName ); fprintf ( stderr, "%s: BZ_MAX_HANDLED_BLOCKS in bzip2recover.c, and recompile.\n", progName ); exit ( 1 ); } /*---------------------------------------------------*/ /*--- Bit stream I/O ---*/ /*---------------------------------------------------*/ typedef struct { FILE* handle; Int32 buffer; Int32 buffLive; Char mode; } BitStream; /*---------------------------------------------*/ BitStream* bsOpenReadStream ( FILE* stream ) { BitStream *bs = malloc ( sizeof(BitStream) ); if (bs == NULL) mallocFail ( sizeof(BitStream) ); bs->handle = stream; bs->buffer = 0; bs->buffLive = 0; bs->mode = 'r'; return bs; } /*---------------------------------------------*/ BitStream* bsOpenWriteStream ( FILE* stream ) { BitStream *bs = malloc ( sizeof(BitStream) ); if (bs == NULL) mallocFail ( sizeof(BitStream) ); bs->handle = stream; bs->buffer = 0; bs->buffLive = 0; bs->mode = 'w'; return bs; } /*---------------------------------------------*/ void bsPutBit ( BitStream* bs, Int32 bit ) { if (bs->buffLive == 8) { Int32 retVal = putc ( (UChar) bs->buffer, bs->handle ); if (retVal == EOF) writeError(); bytesOut++; bs->buffLive = 1; bs->buffer = bit & 0x1; } else { bs->buffer = ( (bs->buffer << 1) | (bit & 0x1) ); bs->buffLive++; }; } /*---------------------------------------------*/ /*-- Returns 0 or 1, or 2 to indicate EOF. --*/ Int32 bsGetBit ( BitStream* bs ) { if (bs->buffLive > 0) { bs->buffLive --; return ( ((bs->buffer) >> (bs->buffLive)) & 0x1 ); } else { Int32 retVal = getc ( bs->handle ); if ( retVal == EOF ) { if (errno != 0) readError(); return 2; } bs->buffLive = 7; bs->buffer = retVal; return ( ((bs->buffer) >> 7) & 0x1 ); } } /*---------------------------------------------*/ void bsClose ( BitStream* bs ) { Int32 retVal; if ( bs->mode == 'w' ) { while ( bs->buffLive < 8 ) { bs->buffLive++; bs->buffer <<= 1; }; retVal = putc ( (UChar) (bs->buffer), bs->handle ); if (retVal == EOF) writeError(); bytesOut++; retVal = fflush ( bs->handle ); if (retVal == EOF) writeError(); } retVal = fclose ( bs->handle ); if (retVal == EOF) { if (bs->mode == 'w') writeError(); else readError(); } free ( bs ); } /*---------------------------------------------*/ void bsPutUChar ( BitStream* bs, UChar c ) { Int32 i; for (i = 7; i >= 0; i--) bsPutBit ( bs, (((UInt32) c) >> i) & 0x1 ); } /*---------------------------------------------*/ void bsPutUInt32 ( BitStream* bs, UInt32 c ) { Int32 i; for (i = 31; i >= 0; i--) bsPutBit ( bs, (c >> i) & 0x1 ); } /*---------------------------------------------*/ Bool endsInBz2 ( Char* name ) { Int32 n = strlen ( name ); if (n <= 4) return False; return (name[n-4] == '.' && #ifdef __GNO__ (name[n-3] == 'b' || name[n-3] == 'B') && (name[n-2] == 'z' || name[n-3] == 'Z') && #else name[n-3] == 'b' && name[n-2] == 'z' && #endif name[n-1] == '2'); } /*---------------------------------------------------*/ /*--- ---*/ /*---------------------------------------------------*/ /* This logic isn't really right when it comes to Cygwin. */ #ifdef _WIN32 # define BZ_SPLIT_SYM '\\' /* path splitter on Windows platform */ #else # define BZ_SPLIT_SYM '/' /* path splitter on Unix platform */ #endif #ifdef __appleiigs__ # define BZ_SPLIT_SYM_GS ':' /* possible path splitter on GS/OS */ #endif #define BLOCK_HEADER_HI 0x00003141UL #define BLOCK_HEADER_LO 0x59265359UL #define BLOCK_ENDMARK_HI 0x00001772UL #define BLOCK_ENDMARK_LO 0x45385090UL /* Increase if necessary. However, a .bz2 file with > 50000 blocks would have an uncompressed size of at least 40GB, so the chances are low you'll need to up this. */ /* STH - Values larger than 5369 (actually a bit less than that) are useless when MaybeUInt64 is 32 bits. */ #ifdef __ORCAC__ #define BZ_MAX_HANDLED_BLOCKS 5369 #else #define BZ_MAX_HANDLED_BLOCKS 50000 #endif #ifndef __ORCAC__ MaybeUInt64 bStart [BZ_MAX_HANDLED_BLOCKS]; MaybeUInt64 bEnd [BZ_MAX_HANDLED_BLOCKS]; MaybeUInt64 rbStart[BZ_MAX_HANDLED_BLOCKS]; MaybeUInt64 rbEnd [BZ_MAX_HANDLED_BLOCKS]; #else /* if defined __GNO__ */ MaybeUInt64 *bStart; MaybeUInt64 *bEnd; MaybeUInt64 *rbStart; MaybeUInt64 *rbEnd; #endif int main ( int argc, Char** argv ) { FILE* inFile; FILE* outFile; BitStream* bsIn, *bsWr; Int32 b, wrBlock, currBlock, rbCtr; MaybeUInt64 bitsRead; UInt32 buffHi, buffLo, blockCRC; Char* p; #if defined(__GNO__) && defined(__STACK_CHECK__) __REPORT_STACK(); #endif strcpy ( progName, argv[0] ); inFileName[0] = outFileName[0] = 0; fprintf ( stderr, #ifdef __GNO__ "bzip2recover 1.0.2gs1: extracts blocks from damaged .bz2 files.\n" ); #else "bzip2recover 1.0.2: extracts blocks from damaged .bz2 files.\n" ); #endif if (argc != 2) { fprintf ( stderr, "%s: usage is `%s damaged_file_name'.\n", progName, progName ); switch (sizeof(MaybeUInt64)) { case 8: fprintf(stderr, "\trestrictions on size of recovered file: None\n"); break; case 4: fprintf(stderr, "\trestrictions on size of recovered file: 512 MB\n"); #ifndef __ORCAC__ fprintf(stderr, "\tto circumvent, recompile with MaybeUInt64 as an\n" "\tunsigned 64-bit int.\n"); #endif break; default: fprintf(stderr, "\tsizeof(MaybeUInt64) is not 4 or 8 -- " "configuration error.\n"); break; } exit(1); } if (strlen(argv[1]) >= BZ_MAX_FILENAME-20) { fprintf ( stderr, "%s: supplied filename is suspiciously (>= " size_t_FMT " chars) long. Bye!\n", progName, strlen(argv[1]) ); exit(1); } strcpy ( inFileName, argv[1] ); inFile = fopen ( inFileName, "rb" ); if (inFile == NULL) { fprintf ( stderr, "%s: can't read `%s'\n", progName, inFileName ); exit(1); } /* Allocate big arrays dynamically so we can use small memory model. These aren't explicitly free()'d anywhere, but exist for the duration of the program. */ #ifdef __ORCAC__ bStart = malloc(BZ_MAX_HANDLED_BLOCKS * sizeof(MaybeUInt64)); bEnd = malloc(BZ_MAX_HANDLED_BLOCKS * sizeof(MaybeUInt64)); rbStart = malloc(BZ_MAX_HANDLED_BLOCKS * sizeof(MaybeUInt64)); rbEnd = malloc(BZ_MAX_HANDLED_BLOCKS * sizeof(MaybeUInt64)); if ((bStart == NULL) || (bEnd == NULL) || (rbStart == NULL) || (rbEnd == NULL)) { fprintf ( stderr, "%s: couldn't allocate enough memory\n", progName ); exit(1); } #endif bsIn = bsOpenReadStream ( inFile ); fprintf ( stderr, "%s: searching for block boundaries ...\n", progName ); bitsRead = 0; buffHi = buffLo = 0; currBlock = 0; bStart[currBlock] = 0; rbCtr = 0; while (True) { b = bsGetBit ( bsIn ); bitsRead++; if (b == 2) { if (bitsRead >= bStart[currBlock] && (bitsRead - bStart[currBlock]) >= 40) { bEnd[currBlock] = bitsRead-1; if (currBlock > 0) fprintf ( stderr, " block " Int32_FMT " runs from " MaybeUInt64_FMT " to " MaybeUInt64_FMT " (incomplete)\n", currBlock, bStart[currBlock], bEnd[currBlock] ); } else currBlock--; break; } buffHi = (buffHi << 1) | (buffLo >> 31); buffLo = (buffLo << 1) | (b & 1); if ( ( (buffHi & 0x0000ffff) == BLOCK_HEADER_HI && buffLo == BLOCK_HEADER_LO) || ( (buffHi & 0x0000ffff) == BLOCK_ENDMARK_HI && buffLo == BLOCK_ENDMARK_LO) ) { if (bitsRead > 49) { bEnd[currBlock] = bitsRead-49; } else { bEnd[currBlock] = 0; } if (currBlock > 0 && (bEnd[currBlock] - bStart[currBlock]) >= 130) { fprintf ( stderr, " block " Int32_FMT " runs from " MaybeUInt64_FMT " to " MaybeUInt64_FMT "\n", rbCtr+1, bStart[currBlock], bEnd[currBlock] ); rbStart[rbCtr] = bStart[currBlock]; rbEnd[rbCtr] = bEnd[currBlock]; rbCtr++; } if (currBlock >= BZ_MAX_HANDLED_BLOCKS) tooManyBlocks(BZ_MAX_HANDLED_BLOCKS); currBlock++; bStart[currBlock] = bitsRead; } } bsClose ( bsIn ); /*-- identified blocks run from 1 to rbCtr inclusive. --*/ if (rbCtr < 1) { fprintf ( stderr, "%s: sorry, I couldn't find any block boundaries.\n", progName ); exit(1); }; fprintf ( stderr, "%s: splitting into blocks\n", progName ); inFile = fopen ( inFileName, "rb" ); if (inFile == NULL) { fprintf ( stderr, "%s: can't open `%s'\n", progName, inFileName ); exit(1); } bsIn = bsOpenReadStream ( inFile ); /*-- placate gcc's dataflow analyser --*/ blockCRC = 0; bsWr = 0; bitsRead = 0; outFile = NULL; wrBlock = 0; while (True) { b = bsGetBit(bsIn); if (b == 2) break; buffHi = (buffHi << 1) | (buffLo >> 31); buffLo = (buffLo << 1) | (b & 1); if (bitsRead == 47+rbStart[wrBlock]) blockCRC = (buffHi << 16) | (buffLo >> 16); if (outFile != NULL && bitsRead >= rbStart[wrBlock] && bitsRead <= rbEnd[wrBlock]) { bsPutBit ( bsWr, b ); } bitsRead++; if (bitsRead == rbEnd[wrBlock]+1) { if (outFile != NULL) { bsPutUChar ( bsWr, 0x17 ); bsPutUChar ( bsWr, 0x72 ); bsPutUChar ( bsWr, 0x45 ); bsPutUChar ( bsWr, 0x38 ); bsPutUChar ( bsWr, 0x50 ); bsPutUChar ( bsWr, 0x90 ); bsPutUInt32 ( bsWr, blockCRC ); bsClose ( bsWr ); } if (wrBlock >= rbCtr) break; wrBlock++; } else if (bitsRead == rbStart[wrBlock]) { /* Create the output file name, correctly handling leading paths. (31.10.2001 by Sergey E. Kusikov) */ /* Modified by STH to make it work better on GNO. It would still be confused by files with a '/' character in their names. */ Char* split; Int32 ofs, k; for (k = 0; k < BZ_MAX_FILENAME; k++) outFileName[k] = 0; strcpy (outFileName, inFileName); #ifdef __appleiigs__ split = ((strrchr (outFileName, BZ_SPLIT_SYM_GS) > strrchr (outFileName, BZ_SPLIT_SYM)) ? strrchr (outFileName, BZ_SPLIT_SYM_GS) : strrchr (outFileName, BZ_SPLIT_SYM)); #else split = strrchr (outFileName, BZ_SPLIT_SYM); #endif /* defined __appleiigs__ */ if (split == NULL) { split = outFileName; } else { ++split; } /* Now split points to the start of the basename. */ ofs = split - outFileName; /* On GS, max. block number is < 6000, and ProDOS filenames are short, so use only four digits in output file name */ #ifdef __ORCAC__ sprintf (split, "rec%4ld", wrBlock+1); #else sprintf (split, "rec%5d", wrBlock+1); #endif for (p = split; *p != 0; p++) if (*p == ' ') *p = '0'; strcat (outFileName, inFileName + ofs); if ( !endsInBz2(outFileName)) strcat ( outFileName, ".bz2" ); fprintf ( stderr, " writing block " Int32_FMT " to `%s' ...\n", wrBlock+1, outFileName ); outFile = fopen ( outFileName, "wb" ); if (outFile == NULL) { fprintf ( stderr, "%s: can't write `%s'\n", progName, outFileName ); exit(1); } #ifdef __appleiigs__ /* Set filetype to BIN if running on the GS */ { static GSString255 fileNameStringGS; static FileInfoRecGS infoRec = { 4, /* pCount */ &fileNameStringGS, /* Ptr to file name */ 0x00C3, /* access restrictions (none) */ 0x06, /* filetype (BIN) */ 0x0000 /* auxtype ($0000) */ }; if (strlen( outFileName ) <= 255) { strncpy( fileNameStringGS.text, outFileName, 255 ); fileNameStringGS.length = strlen( outFileName ); SetFileInfo( &infoRec ); /* Ignoring any errors produced by this call */ } } #endif bsWr = bsOpenWriteStream ( outFile ); bsPutUChar ( bsWr, BZ_HDR_B ); bsPutUChar ( bsWr, BZ_HDR_Z ); bsPutUChar ( bsWr, BZ_HDR_h ); bsPutUChar ( bsWr, BZ_HDR_0 + 9 ); bsPutUChar ( bsWr, 0x31 ); bsPutUChar ( bsWr, 0x41 ); bsPutUChar ( bsWr, 0x59 ); bsPutUChar ( bsWr, 0x26 ); bsPutUChar ( bsWr, 0x53 ); bsPutUChar ( bsWr, 0x59 ); } } fprintf ( stderr, "%s: finished\n", progName ); return 0; } /*-----------------------------------------------------------*/ /*--- end bzip2recover.c ---*/ /*-----------------------------------------------------------*/