lzip/plzip
1
0
Fork 0
Code Issues Activity

Merging upstream version 1.8.

Browse source 
Signed-off-by: Daniel Baumann <daniel@debian.org>
This commit is contained in:
Daniel Baumann 2025-02-24 04:16:09 +01:00
parent 95e76700ee
commit 3ab3342c4f
Signed by: daniel
GPG key ID: FBB4F0E80A80222F
21 changed files with 729 additions and 460 deletions
Download patch file Download diff file Expand all files Collapse all files

99
decompress.cc
View file

@ -1,6 +1,6 @@
/* Plzip - Parallel compressor compatible with lzip
/* Plzip - Massively parallel implementation of lzip
Copyright (C) 2009 Laszlo Ersek.
Copyright (C) 2009-2018 Antonio Diaz Diaz.
Copyright (C) 2009-2019 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
@ -34,7 +34,46 @@
#include <lzlib.h>
#include "lzip.h"
#include "file_index.h"
#include "lzip_index.h"
// This code is based on a patch by Hannes Domani, ssbssa@yahoo.de
// to be able to compile plzip under MS Windows (with MINGW compiler).
#if defined(__MSVCRT__) && defined(WITH_MINGW)
#include <windows.h>
#warning "Parallel I/O is not guaranteed to work on Windows."
ssize_t pread( int fd, void *buf, size_t count, uint64_t offset )
{
OVERLAPPED o = {0,0,0,0,0};
HANDLE fh = (HANDLE)_get_osfhandle(fd);
DWORD bytes;
BOOL ret;
if( fh == INVALID_HANDLE_VALUE ) { errno = EBADF; return -1; }
o.Offset = offset & 0xffffffff;
o.OffsetHigh = (offset >> 32) & 0xffffffff;
ret = ReadFile( fh, buf, (DWORD)count, &bytes, &o );
if( !ret ) { errno = EIO; return -1; }
return (ssize_t)bytes;
}
ssize_t pwrite( int fd, const void *buf, size_t count, uint64_t offset )
{
OVERLAPPED o = {0,0,0,0,0};
HANDLE fh = (HANDLE)_get_osfhandle(fd);
DWORD bytes;
BOOL ret;
if( fh == INVALID_HANDLE_VALUE ) { errno = EBADF; return -1; }
o.Offset = offset & 0xffffffff;
o.OffsetHigh = (offset >> 32) & 0xffffffff;
ret = WriteFile(fh, buf, (DWORD)count, &bytes, &o);
if( !ret ) { errno = EIO; return -1; }
return (ssize_t)bytes;
}
#endif // __MSVCRT__
// Returns the number of bytes really read.
@ -95,7 +134,7 @@ namespace {
struct Worker_arg
{
const File_index * file_index;
const Lzip_index * lzip_index;
const Pretty_print * pp;
int worker_id;
int num_workers;
@ -108,8 +147,8 @@ struct Worker_arg
// write the produced data to file.
extern "C" void * dworker( void * arg )
{
const Worker_arg & tmp = *(Worker_arg *)arg;
const File_index & file_index = *tmp.file_index;
const Worker_arg & tmp = *(const Worker_arg *)arg;
const Lzip_index & lzip_index = *tmp.lzip_index;
const Pretty_print & pp = *tmp.pp;
const int worker_id = tmp.worker_id;
const int num_workers = tmp.num_workers;
@ -124,12 +163,12 @@ extern "C" void * dworker( void * arg )
LZ_decompress_errno( decoder ) != LZ_ok )
{ pp( "Not enough memory." ); cleanup_and_fail(); }
for( long i = worker_id; i < file_index.members(); i += num_workers )
for( long i = worker_id; i < lzip_index.members(); i += num_workers )
{
long long data_pos = file_index.dblock( i ).pos();
long long data_rest = file_index.dblock( i ).size();
long long member_pos = file_index.mblock( i ).pos();
long long member_rest = file_index.mblock( i ).size();
long long data_pos = lzip_index.dblock( i ).pos();
long long data_rest = lzip_index.dblock( i ).size();
long long member_pos = lzip_index.mblock( i ).pos();
long long member_rest = lzip_index.mblock( i ).size();
while( member_rest > 0 )
{
@ -180,7 +219,7 @@ extern "C" void * dworker( void * arg )
if( rd == 0 ) break;
}
}
show_progress( file_index.mblock( i ).size() );
show_progress( lzip_index.mblock( i ).size() );
}
delete[] obuffer; delete[] ibuffer;
@ -197,26 +236,27 @@ extern "C" void * dworker( void * arg )
// start the workers and wait for them to finish.
int decompress( const unsigned long long cfile_size, int num_workers,
const int infd, const int outfd, const Pretty_print & pp,
const int debug_level, const bool ignore_trailing,
const int debug_level, const int in_slots,
const int out_slots, const bool ignore_trailing,
const bool loose_trailing, const bool infd_isreg )
{
if( !infd_isreg )
return dec_stream( cfile_size, num_workers, infd, outfd, pp,
debug_level, ignore_trailing, loose_trailing );
return dec_stream( cfile_size, num_workers, infd, outfd, pp, debug_level,
in_slots, out_slots, ignore_trailing, loose_trailing );
const File_index file_index( infd, ignore_trailing, loose_trailing );
if( file_index.retval() == 1 )
const Lzip_index lzip_index( infd, ignore_trailing, loose_trailing );
if( lzip_index.retval() == 1 )
{
lseek( infd, 0, SEEK_SET );
return dec_stream( cfile_size, num_workers, infd, outfd, pp,
debug_level, ignore_trailing, loose_trailing );
return dec_stream( cfile_size, num_workers, infd, outfd, pp, debug_level,
in_slots, out_slots, ignore_trailing, loose_trailing );
}
if( file_index.retval() != 0 )
{ show_file_error( pp.name(), file_index.error().c_str() );
return file_index.retval(); }
if( lzip_index.retval() != 0 )
{ show_file_error( pp.name(), lzip_index.error().c_str() );
return lzip_index.retval(); }
if( num_workers > file_index.members() )
num_workers = file_index.members();
if( num_workers > lzip_index.members() )
num_workers = lzip_index.members();
if( verbosity >= 1 ) pp();
show_progress( 0, cfile_size, &pp ); // init
@ -225,7 +265,8 @@ int decompress( const unsigned long long cfile_size, int num_workers,
struct stat st;
if( fstat( outfd, &st ) != 0 || !S_ISREG( st.st_mode ) ||
lseek( outfd, 0, SEEK_CUR ) < 0 )
return dec_stdout( num_workers, infd, outfd, pp, debug_level, file_index );
return dec_stdout( num_workers, infd, outfd, pp, debug_level, out_slots,
lzip_index );
}
Worker_arg * worker_args = new( std::nothrow ) Worker_arg[num_workers];
@ -234,7 +275,7 @@ int decompress( const unsigned long long cfile_size, int num_workers,
{ pp( "Not enough memory." ); cleanup_and_fail(); }
for( int i = 0; i < num_workers; ++i )
{
worker_args[i].file_index = &file_index;
worker_args[i].lzip_index = &lzip_index;
worker_args[i].pp = &pp;
worker_args[i].worker_id = i;
worker_args[i].num_workers = num_workers;
@ -257,9 +298,9 @@ int decompress( const unsigned long long cfile_size, int num_workers,
if( verbosity >= 2 )
{
if( verbosity >= 4 ) show_header( file_index.dictionary_size( 0 ) );
const unsigned long long in_size = file_index.cdata_size();
const unsigned long long out_size = file_index.udata_size();
if( verbosity >= 4 ) show_header( lzip_index.dictionary_size( 0 ) );
const unsigned long long in_size = lzip_index.cdata_size();
const unsigned long long out_size = lzip_index.udata_size();
if( out_size == 0 || in_size == 0 )
std::fputs( "no data compressed. ", stderr );
else