508 lines
17 KiB
C++
508 lines
17 KiB
C++
/* Lziprecover - Data recovery tool for the lzip format
|
|
Copyright (C) 2009-2015 Antonio Diaz Diaz.
|
|
|
|
This program is free software: you can redistribute it and/or modify
|
|
it under the terms of the GNU General Public License as published by
|
|
the Free Software Foundation, either version 2 of the License, or
|
|
(at your option) any later version.
|
|
|
|
This program is distributed in the hope that it will be useful,
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
GNU General Public License for more details.
|
|
|
|
You should have received a copy of the GNU General Public License
|
|
along with this program. If not, see <http://www.gnu.org/licenses/>.
|
|
*/
|
|
|
|
#define _FILE_OFFSET_BITS 64
|
|
|
|
#include <cerrno>
|
|
#include <climits>
|
|
#include <cstdio>
|
|
#include <cstdlib>
|
|
#include <cstring>
|
|
#include <string>
|
|
#include <vector>
|
|
#include <stdint.h>
|
|
#include <unistd.h>
|
|
#include <sys/stat.h>
|
|
|
|
#include "lzip.h"
|
|
#include "decoder.h"
|
|
#include "block.h"
|
|
#include "file_index.h"
|
|
|
|
|
|
namespace {
|
|
|
|
// Add 'bv' to 'block_vector' splitting blocks as needed to keep all the
|
|
// edges (pos and end of every block).
|
|
// 'block_vector' contains the result. 'bv' is destroyed.
|
|
void combine( std::vector< Block > & block_vector, std::vector< Block > & bv )
|
|
{
|
|
if( block_vector.empty() ) { block_vector.swap( bv ); return; }
|
|
unsigned i1 = 0, i2 = 0;
|
|
while( i1 < block_vector.size() && i2 < bv.size() )
|
|
{
|
|
Block & b1 = block_vector[i1];
|
|
Block & b2 = bv[i2];
|
|
if( b1.overlaps( b2 ) )
|
|
{
|
|
if( b1 < b2 )
|
|
{
|
|
Block b = b1.split( b2.pos() );
|
|
block_vector.insert( block_vector.begin() + i1, b ); ++i1;
|
|
}
|
|
else if( b2 < b1 )
|
|
{
|
|
Block b( b2.pos(), b1.pos() - b2.pos() );
|
|
b2.split( b1.pos() );
|
|
block_vector.insert( block_vector.begin() + i1, b ); ++i1;
|
|
}
|
|
else if( b1.end() < b2.end() ) { b2.split( b1.end() ); ++i1; }
|
|
else if( b2.end() < b1.end() )
|
|
{
|
|
Block b = b1.split( b2.end() );
|
|
block_vector.insert( block_vector.begin() + i1, b ); ++i1; ++i2;
|
|
}
|
|
else { ++i1; ++i2; } // blocks are identical
|
|
}
|
|
else if( b1 < b2 ) ++i1;
|
|
else { block_vector.insert( block_vector.begin() + i1, b2 ); ++i1; ++i2; }
|
|
}
|
|
if( i2 < bv.size() ) // tail copy
|
|
block_vector.insert( block_vector.end(), bv.begin() + i2, bv.end() );
|
|
}
|
|
|
|
|
|
// positions in 'block_vector' are absolute file positions.
|
|
bool diff_member( const long long mpos, const long long msize,
|
|
const std::vector< int > & infd_vector,
|
|
std::vector< Block > & block_vector,
|
|
std::vector< int > & color_vector )
|
|
{
|
|
const int files = infd_vector.size();
|
|
const int buffer_size = 65536;
|
|
uint8_t * const buffer1 = new uint8_t[buffer_size];
|
|
uint8_t * const buffer2 = new uint8_t[buffer_size];
|
|
int next_color = 1;
|
|
|
|
bool error = false;
|
|
for( int i1 = 0; i1 < files && !error; ++i1 )
|
|
{
|
|
for( int i2 = i1 + 1; i2 < files && !error; ++i2 )
|
|
{
|
|
if( color_vector[i1] != 0 && color_vector[i1] == color_vector[i2] )
|
|
continue;
|
|
std::vector< Block > bv;
|
|
long long partial_pos = 0;
|
|
const int fd1 = infd_vector[i1], fd2 = infd_vector[i2];
|
|
int begin = -1; // begin of block. -1 means no block
|
|
bool prev_equal = true;
|
|
if( !safe_seek( fd1, mpos ) || !safe_seek( fd2, mpos ) )
|
|
{ error = true; break; }
|
|
|
|
while( partial_pos < msize )
|
|
{
|
|
const int size = std::min( (long long)buffer_size, msize - partial_pos );
|
|
const int rd = readblock( fd1, buffer1, size );
|
|
if( rd != size && errno )
|
|
{ show_error( "Error reading input file", errno ); error = true; break; }
|
|
if( rd > 0 )
|
|
{
|
|
if( readblock( fd2, buffer2, rd ) != rd )
|
|
{ show_error( "Error reading input file", errno );
|
|
error = true; break; }
|
|
for( int i = 0; i < rd; ++i )
|
|
{
|
|
if( buffer1[i] != buffer2[i] )
|
|
{
|
|
prev_equal = false;
|
|
if( begin < 0 ) begin = partial_pos + i; // begin block
|
|
}
|
|
else if( !prev_equal ) prev_equal = true;
|
|
else if( begin >= 0 ) // end block
|
|
{
|
|
Block b( mpos + begin, partial_pos + i - 1 - begin );
|
|
begin = -1;
|
|
bv.push_back( b );
|
|
}
|
|
}
|
|
partial_pos += rd;
|
|
}
|
|
if( rd < buffer_size ) break; // EOF
|
|
}
|
|
if( begin >= 0 ) // finish last block
|
|
{
|
|
Block b( mpos + begin, partial_pos - prev_equal - begin );
|
|
bv.push_back( b );
|
|
}
|
|
if( bv.empty() ) // members are identical, set to same color
|
|
{
|
|
if( color_vector[i1] == 0 )
|
|
{
|
|
if( color_vector[i2] != 0 ) color_vector[i1] = color_vector[i2];
|
|
else color_vector[i1] = color_vector[i2] = next_color++;
|
|
}
|
|
else if( color_vector[i2] == 0 ) color_vector[i2] = color_vector[i1];
|
|
else internal_error( "different colors assigned to identical members." );
|
|
}
|
|
combine( block_vector, bv );
|
|
}
|
|
if( color_vector[i1] == 0 ) color_vector[i1] = next_color++;
|
|
}
|
|
delete[] buffer2; delete[] buffer1;
|
|
return !error;
|
|
}
|
|
|
|
|
|
long ipow( const unsigned base, const unsigned exponent )
|
|
{
|
|
unsigned long result = 1;
|
|
for( unsigned i = 0; i < exponent; ++i )
|
|
{
|
|
if( LONG_MAX / base >= result ) result *= base;
|
|
else { result = LONG_MAX; break; }
|
|
}
|
|
return result;
|
|
}
|
|
|
|
|
|
int open_input_files( const std::vector< std::string > & filenames,
|
|
std::vector< int > & infd_vector,
|
|
File_index & file_index, const int verbosity )
|
|
{
|
|
const int files = filenames.size();
|
|
bool identical = false;
|
|
for( int i = 1; i < files; ++i )
|
|
if( filenames[0] == filenames[i] )
|
|
{ identical = true; break; }
|
|
if( !identical )
|
|
for( int i = 0; i < files; ++i )
|
|
{
|
|
struct stat in_stats;
|
|
ino_t st_ino0 = 0;
|
|
dev_t st_dev0 = 0;
|
|
infd_vector[i] = open_instream( filenames[i].c_str(), &in_stats, true, true );
|
|
if( infd_vector[i] < 0 ) return 1;
|
|
if( i == 0 ) { st_ino0 = in_stats.st_ino; st_dev0 = in_stats.st_dev; }
|
|
else if( st_ino0 == in_stats.st_ino && st_dev0 == in_stats.st_dev )
|
|
{ identical = true; break; }
|
|
}
|
|
if( identical ) { show_error( "Two input files are the same." ); return 2; }
|
|
|
|
long long isize = 0;
|
|
for( int i = 0; i < files; ++i )
|
|
{
|
|
long long tmp;
|
|
const File_index fi( infd_vector[i] );
|
|
if( fi.retval() == 0 ) // file format is intact
|
|
{
|
|
if( file_index.retval() != 0 ) file_index = fi;
|
|
else if( file_index != fi )
|
|
{ show_error( "Input files are different." ); return 2; }
|
|
tmp = file_index.file_size();
|
|
}
|
|
else // file format is damaged
|
|
{
|
|
tmp = lseek( infd_vector[i], 0, SEEK_END );
|
|
if( tmp < 0 )
|
|
{
|
|
show_error2( "Input file", filenames[i].c_str(), "is not seekable." );
|
|
return 1;
|
|
}
|
|
}
|
|
if( i == 0 )
|
|
{
|
|
isize = tmp;
|
|
if( isize < min_member_size )
|
|
{ show_error2( "Input file", filenames[i].c_str(), "is too short." );
|
|
return 2; }
|
|
}
|
|
else if( isize != tmp )
|
|
{ show_error( "Sizes of input files are different." ); return 2; }
|
|
}
|
|
|
|
if( file_index.retval() != 0 )
|
|
{
|
|
const File_index fi( infd_vector, isize );
|
|
if( fi.retval() == 0 ) // file format could be recovered
|
|
file_index = fi;
|
|
else
|
|
{ show_error( "Format damaged in all input files." ); return 2; }
|
|
}
|
|
|
|
for( int i = 0; i < files; ++i )
|
|
{
|
|
const int infd = infd_vector[i];
|
|
bool error = false;
|
|
for( long j = 0; j < file_index.members(); ++j )
|
|
{
|
|
const long long mpos = file_index.mblock( j ).pos();
|
|
const long long msize = file_index.mblock( j ).size();
|
|
if( !safe_seek( infd, mpos ) ) return 1;
|
|
if( !try_decompress_member( infd, msize ) ) { error = true; break; }
|
|
}
|
|
if( !error )
|
|
{
|
|
if( verbosity >= 1 )
|
|
std::printf( "File '%s' has no errors. Recovery is not needed.\n",
|
|
filenames[i].c_str() );
|
|
return 0;
|
|
}
|
|
}
|
|
return -1;
|
|
}
|
|
|
|
|
|
// merge block by block
|
|
bool try_merge_member( const long long mpos, const long long msize,
|
|
const std::vector< Block > & block_vector,
|
|
const std::vector< int > & color_vector,
|
|
const std::vector< int > & infd_vector,
|
|
const std::string & output_filename,
|
|
const int outfd, const int verbosity )
|
|
{
|
|
const int blocks = block_vector.size();
|
|
const int files = infd_vector.size();
|
|
const long variations = ipow( files, blocks );
|
|
if( variations >= LONG_MAX )
|
|
{
|
|
if( files > 2 )
|
|
show_error( "Too many damaged blocks. Try merging fewer files." );
|
|
else
|
|
show_error( "Too many damaged blocks. Merging is not possible." );
|
|
cleanup_and_fail( output_filename, outfd, 2 );
|
|
}
|
|
int bi = 0; // block index
|
|
std::vector< int > file_idx( blocks, 0 ); // file to read each block from
|
|
|
|
while( bi >= 0 )
|
|
{
|
|
if( verbosity >= 1 )
|
|
{
|
|
long var = 0;
|
|
for( int i = 0; i < blocks; ++i )
|
|
var = ( var * files ) + file_idx[i];
|
|
std::printf( "Trying variation %ld of %ld \r", var + 1, variations );
|
|
std::fflush( stdout );
|
|
}
|
|
while( bi < blocks )
|
|
{
|
|
const int infd = infd_vector[file_idx[bi]];
|
|
if( !safe_seek( infd, block_vector[bi].pos() ) ||
|
|
!safe_seek( outfd, block_vector[bi].pos() ) ||
|
|
!copy_file( infd, outfd, block_vector[bi].size() ) )
|
|
cleanup_and_fail( output_filename, outfd, 1 );
|
|
++bi;
|
|
}
|
|
if( !safe_seek( outfd, mpos ) )
|
|
cleanup_and_fail( output_filename, outfd, 1 );
|
|
long long failure_pos = 0;
|
|
if( try_decompress_member( outfd, msize, &failure_pos ) ) return true;
|
|
while( bi > 0 && mpos + failure_pos < block_vector[bi-1].pos() ) --bi;
|
|
while( --bi >= 0 )
|
|
{
|
|
while( ++file_idx[bi] < files )
|
|
{
|
|
const int color = color_vector[file_idx[bi]];
|
|
bool done = true;
|
|
for( int i = file_idx[bi] - 1; i >= 0; --i )
|
|
if( color_vector[i] == color ) { done = false; break; }
|
|
if( done ) break;
|
|
}
|
|
if( file_idx[bi] < files ) break;
|
|
file_idx[bi] = 0;
|
|
}
|
|
}
|
|
return false;
|
|
}
|
|
|
|
|
|
// merge a single block split at every possible position
|
|
bool try_merge_member1( const long long mpos, const long long msize,
|
|
const std::vector< Block > & block_vector,
|
|
const std::vector< int > & color_vector,
|
|
const std::vector< int > & infd_vector,
|
|
const std::string & output_filename,
|
|
const int outfd, const int verbosity )
|
|
{
|
|
if( block_vector.size() != 1 || block_vector[0].size() <= 1 ) return false;
|
|
const long long pos = block_vector[0].pos();
|
|
const long long size = block_vector[0].size();
|
|
const int files = infd_vector.size();
|
|
const int variations = files * ( files - 1 );
|
|
uint8_t byte;
|
|
|
|
for( int i1 = 0; i1 < files; ++i1 )
|
|
for( int i2 = 0; i2 < files; ++i2 )
|
|
{
|
|
if( i1 == i2 || color_vector[i1] == color_vector[i2] ) continue;
|
|
const int infd = infd_vector[i1];
|
|
if( !safe_seek( infd, pos ) ||
|
|
!safe_seek( infd_vector[i2], pos ) ||
|
|
!safe_seek( outfd, pos ) ||
|
|
!copy_file( infd_vector[i2], outfd, size ) )
|
|
cleanup_and_fail( output_filename, outfd, 1 );
|
|
const int var = ( i1 * ( files - 1 ) ) + i2 - ( i2 > i1 ) + 1;
|
|
for( long long i = 0; i < size; ++i )
|
|
{
|
|
if( verbosity >= 1 )
|
|
{
|
|
std::printf( "Trying variation %d of %d, position %lld \r",
|
|
var, variations, pos + i );
|
|
std::fflush( stdout );
|
|
}
|
|
if( !safe_seek( outfd, pos + i ) ||
|
|
readblock( infd, &byte, 1 ) != 1 ||
|
|
writeblock( outfd, &byte, 1 ) != 1 ||
|
|
!safe_seek( outfd, mpos ) )
|
|
cleanup_and_fail( output_filename, outfd, 1 );
|
|
long long failure_pos = 0;
|
|
if( try_decompress_member( outfd, msize, &failure_pos ) ) return true;
|
|
if( mpos + failure_pos <= pos + i ) break;
|
|
}
|
|
}
|
|
return false;
|
|
}
|
|
|
|
} // end namespace
|
|
|
|
|
|
void cleanup_and_fail( const std::string & output_filename,
|
|
const int outfd, const int retval )
|
|
{
|
|
if( outfd >= 0 ) close( outfd );
|
|
if( std::remove( output_filename.c_str() ) != 0 && errno != ENOENT )
|
|
show_error( "WARNING: deletion of output file (apparently) failed." );
|
|
std::exit( retval );
|
|
}
|
|
|
|
|
|
// max_size < 0 means no size limit.
|
|
bool copy_file( const int infd, const int outfd, const long long max_size )
|
|
{
|
|
const int buffer_size = 65536;
|
|
// remaining number of bytes to copy
|
|
long long rest = ( ( max_size >= 0 ) ? max_size : buffer_size );
|
|
uint8_t * const buffer = new uint8_t[buffer_size];
|
|
bool error = false;
|
|
|
|
while( rest > 0 )
|
|
{
|
|
const int size = std::min( (long long)buffer_size, rest );
|
|
if( max_size >= 0 ) rest -= size;
|
|
const int rd = readblock( infd, buffer, size );
|
|
if( rd != size && errno )
|
|
{ show_error( "Error reading input file", errno ); error = true; break; }
|
|
if( rd > 0 )
|
|
{
|
|
const int wr = writeblock( outfd, buffer, rd );
|
|
if( wr != rd )
|
|
{ show_error( "Error writing output file", errno );
|
|
error = true; break; }
|
|
}
|
|
if( rd < size ) break; // EOF
|
|
}
|
|
delete[] buffer;
|
|
return !error;
|
|
}
|
|
|
|
|
|
bool try_decompress_member( const int fd, const unsigned long long msize,
|
|
long long * failure_posp )
|
|
{
|
|
Range_decoder rdec( fd );
|
|
File_header header;
|
|
rdec.read_data( header.data, File_header::size );
|
|
if( !rdec.finished() && // End Of File
|
|
header.verify_magic() && header.verify_version() &&
|
|
header.dictionary_size() >= min_dictionary_size &&
|
|
header.dictionary_size() <= max_dictionary_size )
|
|
{
|
|
LZ_decoder decoder( header, rdec, -1 );
|
|
Pretty_print dummy( "", -1 );
|
|
|
|
if( decoder.decode_member( dummy ) == 0 &&
|
|
rdec.member_position() == msize ) return true;
|
|
if( failure_posp ) *failure_posp = rdec.member_position();
|
|
}
|
|
return false;
|
|
}
|
|
|
|
|
|
int merge_files( const std::vector< std::string > & filenames,
|
|
const std::string & output_filename, const int verbosity,
|
|
const bool force )
|
|
{
|
|
const int files = filenames.size();
|
|
std::vector< int > infd_vector( files );
|
|
File_index file_index;
|
|
const int retval =
|
|
open_input_files( filenames, infd_vector, file_index, verbosity );
|
|
if( retval >= 0 ) return retval;
|
|
if( !safe_seek( infd_vector[0], 0 ) ) return 1;
|
|
|
|
const int outfd = open_outstream_rw( output_filename, force );
|
|
if( outfd < 0 ) return 1;
|
|
if( !copy_file( infd_vector[0], outfd ) ) // copy whole file
|
|
cleanup_and_fail( output_filename, outfd, 1 );
|
|
|
|
for( long j = 0; j < file_index.members(); ++j )
|
|
{
|
|
const long long mpos = file_index.mblock( j ).pos();
|
|
const long long msize = file_index.mblock( j ).size();
|
|
// vector of data blocks differing among the copies of the current member
|
|
std::vector< Block > block_vector;
|
|
std::vector< int > color_vector( files, 0 );
|
|
if( !diff_member( mpos, msize, infd_vector, block_vector, color_vector ) ||
|
|
!safe_seek( outfd, mpos ) )
|
|
cleanup_and_fail( output_filename, outfd, 1 );
|
|
|
|
if( block_vector.empty() )
|
|
{
|
|
if( file_index.members() > 1 && try_decompress_member( outfd, msize ) )
|
|
continue;
|
|
show_error( "Input files are (partially) identical. Merging is not possible." );
|
|
cleanup_and_fail( output_filename, outfd, 2 );
|
|
}
|
|
|
|
if( verbosity >= 1 && file_index.members() > 1 )
|
|
{
|
|
std::printf( "Merging member %ld of %ld\n",
|
|
j + 1, file_index.members() );
|
|
std::fflush( stdout );
|
|
}
|
|
|
|
bool done = false;
|
|
if( file_index.members() > 1 || block_vector.size() > 1 )
|
|
{
|
|
done = try_merge_member( mpos, msize, block_vector, color_vector,
|
|
infd_vector, output_filename, outfd, verbosity );
|
|
if( !done && verbosity >= 1 ) std::fputc( '\n', stdout );
|
|
}
|
|
if( !done )
|
|
done = try_merge_member1( mpos, msize, block_vector, color_vector,
|
|
infd_vector, output_filename, outfd, verbosity );
|
|
if( verbosity >= 1 ) std::fputc( '\n', stdout );
|
|
if( !done )
|
|
{
|
|
if( verbosity >= 2 )
|
|
for( unsigned i = 0; i < block_vector.size(); ++i )
|
|
std::fprintf( stderr, "area %2d from position %6lld to %6lld\n", i + 1,
|
|
block_vector[i].pos(), block_vector[i].end() - 1 );
|
|
show_error( "Some error areas overlap. Can't recover input file." );
|
|
cleanup_and_fail( output_filename, outfd, 2 );
|
|
}
|
|
}
|
|
|
|
if( close( outfd ) != 0 )
|
|
{
|
|
show_error( "Error closing output file", errno );
|
|
cleanup_and_fail( output_filename, -1, 1 );
|
|
}
|
|
if( verbosity >= 1 )
|
|
std::fputs( "Input files merged successfully.\n", stdout );
|
|
return 0;
|
|
}
|