diff options
Diffstat (limited to 'delete_lz.cc')
| -rw-r--r-- | delete_lz.cc | 172 |
1 files changed, 32 insertions, 140 deletions
diff --git a/delete_lz.cc b/delete_lz.cc index 961645f..c566c00 100644 --- a/delete_lz.cc +++ b/delete_lz.cc @@ -1,5 +1,5 @@ /* Tarlz - Archiver with multimember lzip compression - Copyright (C) 2013-2020 Antonio Diaz Diaz. + Copyright (C) 2013-2021 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 @@ -25,7 +25,7 @@ #include <cstring> #include <string> #include <vector> -#include <pthread.h> +#include <pthread.h> // for tarlz.h #include <stdint.h> #include <unistd.h> #include <lzlib.h> @@ -33,150 +33,52 @@ #include "arg_parser.h" #include "tarlz.h" #include "lzip_index.h" - - -/* Read 'size' decompressed bytes from the archive. - Return value: 0 = OK, 1 = damaged member, 2 = fatal error. */ -int archive_read_lz( LZ_Decoder * const decoder, const int infd, - long long & file_pos, const long long member_end, - const long long cdata_size, uint8_t * const buf, - const int size, const char ** msg ) - { - int sz = 0; - - while( sz < size ) - { - const int rd = LZ_decompress_read( decoder, buf + sz, size - sz ); - if( rd < 0 ) - { *msg = LZ_strerror( LZ_decompress_errno( decoder ) ); return 1; } - if( rd == 0 && LZ_decompress_finished( decoder ) == 1 ) - { *msg = end_msg; return 2; } - sz += rd; - if( sz < size && LZ_decompress_write_size( decoder ) > 0 ) - { - const long long ibuf_size = 16384; - uint8_t ibuf[ibuf_size]; - const long long rest = ( file_pos < member_end ) ? - member_end - file_pos : cdata_size - file_pos; - const int rsize = std::min( LZ_decompress_write_size( decoder ), - (int)std::min( ibuf_size, rest ) ); - if( rsize <= 0 ) LZ_decompress_finish( decoder ); - else - { - const int rd = preadblock( infd, ibuf, rsize, file_pos ); - if( LZ_decompress_write( decoder, ibuf, rd ) != rd ) - internal_error( "library error (LZ_decompress_write)." ); - file_pos += rd; - if( rd < rsize ) - { - LZ_decompress_finish( decoder ); - if( errno ) { *msg = "Error reading archive"; return 2; } - } - } - } - } - return 0; - } - - -int parse_records_lz( LZ_Decoder * const decoder, const int infd, - long long & file_pos, const long long member_end, - const long long cdata_size, long long & data_pos, - Extended & extended, const Tar_header header, - Resizable_buffer & rbuf, const char ** msg, - const bool permissive ) - { - const long long edsize = parse_octal( header + size_o, size_l ); - const long long bufsize = round_up( edsize ); - if( edsize <= 0 || edsize >= 1LL << 33 || bufsize >= INT_MAX ) - return 1; // overflow or no extended data - if( !rbuf.resize( bufsize ) ) return 1; // extended records buffer - int retval = archive_read_lz( decoder, infd, file_pos, member_end, - cdata_size, (uint8_t *)rbuf(), bufsize, msg ); - if( retval == 0 ) - { if( extended.parse( rbuf(), edsize, permissive ) ) data_pos += bufsize; - else retval = 2; } - return retval; - } - - -int skip_member_lz( LZ_Decoder * const decoder, const int infd, - long long & file_pos, const long long member_end, - const long long cdata_size, long long & data_pos, - long long rest, const char ** msg ) - { - const int bufsize = 32 * header_size; - uint8_t buf[bufsize]; - while( rest > 0 ) // skip tar member - { - const int rsize = ( rest >= bufsize ) ? bufsize : rest; - const int ret = archive_read_lz( decoder, infd, file_pos, member_end, - cdata_size, buf, rsize, msg ); - if( ret != 0 ) return ret; - data_pos += rsize; - rest -= rsize; - } - return 0; - } +#include "archive_reader.h" /* Deleting from a corrupt archive must not worsen the corruption. Stop and tail-copy as soon as corruption is found. */ int delete_members_lz( const Cl_options & cl_opts, - const char * const archive_namep, + const Archive_descriptor & ad, std::vector< char > & name_pending, - const Lzip_index & lzip_index, - const int infd, const int outfd ) + const int outfd ) { Resizable_buffer rbuf; - LZ_Decoder * const decoder = LZ_decompress_open(); - if( !rbuf.size() || !decoder || LZ_decompress_errno( decoder ) != LZ_ok ) - { show_error( mem_msg ); return 1; } + Archive_reader_i ar( ad ); // indexed reader + if( !rbuf.size() || ar.fatal() ) { show_error( mem_msg ); return 1; } long long istream_pos = 0; // source of next data move - const long long cdata_size = lzip_index.cdata_size(); int retval = 0, retval2 = 0; - for( long i = 0; i < lzip_index.members(); ++i ) + for( long i = 0; i < ad.lzip_index.members(); ++i ) { - const long long mdata_pos = lzip_index.dblock( i ).pos(); - long long data_pos = mdata_pos; - const long long mdata_end = lzip_index.dblock( i ).end(); - if( data_pos >= mdata_end ) continue; // empty lzip member - const long long member_pos = lzip_index.mblock( i ).pos(); - long long file_pos = member_pos; - const long long member_end = lzip_index.mblock( i ).end(); - + if( ad.lzip_index.dblock( i ).size() == 0 ) continue; // empty lzip member long long member_begin = 0; // first pos of current tar member Extended extended; // metadata from extended records bool prev_extended = false; // prev header was extended - LZ_decompress_reset( decoder ); // prepare for new member - if( !safe_seek( infd, member_pos ) ) { retval = 1; break; } + ar.set_member( i ); // prepare for new member while( true ) // process one tar header per iteration { - if( data_pos >= mdata_end ) + if( ar.data_pos() >= ar.mdata_end() ) { - if( data_pos == mdata_end && !prev_extended ) break; + if( ar.at_member_end() && !prev_extended ) break; // member end exceeded or ends in extended - show_file_error( archive_namep, "Member misalignment found." ); + show_file_error( ad.namep, "Member misalignment found." ); retval = 2; goto done; } - if( !prev_extended ) member_begin = data_pos; + if( !prev_extended ) member_begin = ar.data_pos(); Tar_header header; - const char * msg = 0; - retval = archive_read_lz( decoder, infd, file_pos, member_end, - cdata_size, header, header_size, &msg ); - if( retval != 0 ) { show_file_error( archive_namep, msg ); goto done; } - data_pos += header_size; + retval = ar.read( header, header_size ); + if( retval != 0 ) { show_file_error( ad.namep, ar.e_msg() ); goto done; } if( !verify_ustar_chksum( header ) ) { if( block_is_zero( header, header_size ) ) // EOF { if( prev_extended && !cl_opts.permissive ) - { show_file_error( archive_namep, fv_msg1 ); retval = 2; } + { show_file_error( ad.namep, fv_msg1 ); retval = 2; } goto done; } - show_file_error( archive_namep, ( data_pos > header_size ) ? + show_file_error( ad.namep, ( ar.data_pos() > header_size ) ? bad_hdr_msg : posix_lz_msg ); retval = 2; goto done; @@ -186,67 +88,57 @@ int delete_members_lz( const Cl_options & cl_opts, if( typeflag == tf_global ) { if( prev_extended && !cl_opts.permissive ) - { show_file_error( archive_namep, fv_msg2 ); retval = 2; goto done; } + { show_file_error( ad.namep, fv_msg2 ); retval = 2; goto done; } Extended dummy; // global headers are parsed and ignored - retval = parse_records_lz( decoder, infd, file_pos, member_end, - cdata_size, data_pos, dummy, header, - rbuf, &msg, true ); + retval = ar.parse_records( dummy, header, rbuf, true ); if( retval == 0 ) continue; - show_file_error( archive_namep, gblrec_msg ); + show_file_error( ad.namep, gblrec_msg ); goto done; } if( typeflag == tf_extended ) { + const char * msg = 0; if( prev_extended && !cl_opts.permissive ) { msg = fv_msg3; retval = 2; } - else retval = parse_records_lz( decoder, infd, file_pos, member_end, - cdata_size, data_pos, extended, header, - rbuf, &msg, cl_opts.permissive ); + else retval = ar.parse_records( extended, header, rbuf, + cl_opts.permissive ); if( retval == 0 && !extended.crc_present() && cl_opts.missing_crc ) { msg = mcrc_msg; retval = 2; } if( retval == 0 ) { prev_extended = true; continue; } - show_file_error( archive_namep, extrec_msg ); + show_file_error( ad.namep, msg ? msg : extrec_msg ); goto done; } prev_extended = false; extended.fill_from_ustar( header ); // copy metadata from header - long long rest = round_up( extended.file_size() ); // size + padding - if( data_pos + rest >= mdata_end ) data_pos += rest; - else // skip tar member - if( ( retval = skip_member_lz( decoder, infd, file_pos, member_end, - cdata_size, data_pos, rest, &msg ) ) != 0 ) - goto done; + if( ( retval = ar.skip_member( extended ) ) != 0 ) goto done; // delete tar member if( !check_skip_filename( cl_opts, name_pending, extended.path().c_str() ) ) { // verify that members match - if( member_begin != mdata_pos || data_pos != mdata_end ) + if( member_begin != ad.lzip_index.dblock( i ).pos() || !ar.at_member_end() ) { show_file_error( extended.path().c_str(), "Can't delete: not compressed individually." ); retval2 = 2; extended.reset(); continue; } if( !show_member_name( extended, header, 1, rbuf ) ) { retval = 1; goto done; } - const long long size = member_pos - istream_pos; + const long long size = ad.lzip_index.mblock( i ).pos() - istream_pos; if( size > 0 ) // move pending data each time a member is deleted { if( istream_pos == 0 ) { if( !safe_seek( outfd, size ) ) { retval = 1; goto done; } } - else if( !safe_seek( infd, istream_pos ) || - !copy_file( infd, outfd, size ) ) { retval = 1; goto done; } + else if( !safe_seek( ad.infd, istream_pos ) || + !copy_file( ad.infd, outfd, size ) ) { retval = 1; goto done; } } - istream_pos = member_end; + istream_pos = ad.lzip_index.mblock( i ).end(); // member end } extended.reset(); } } done: if( retval < retval2 ) retval = retval2; - if( LZ_decompress_close( decoder ) < 0 && !retval ) - { show_error( "LZ_decompress_close failed." ); retval = 1; } // tail copy keeps trailing data - return tail_copy( archive_namep, cl_opts.parser, name_pending, lzip_index, - istream_pos, infd, outfd, retval ); + return tail_copy( cl_opts.parser, ad, name_pending, istream_pos, outfd, retval ); } |