rlm@3: <?php
rlm@3: // +----------------------------------------------------------------------+
rlm@3: // | PHP version 5                                                        |
rlm@3: // +----------------------------------------------------------------------+
rlm@3: // | Copyright (c) 2002-2006 James Heinrich, Allan Hansen                 |
rlm@3: // +----------------------------------------------------------------------+
rlm@3: // | This source file is subject to version 2 of the GPL license,         |
rlm@3: // | that is bundled with this package in the file license.txt and is     |
rlm@3: // | available through the world-wide-web at the following url:           |
rlm@3: // | http://www.gnu.org/copyleft/gpl.html                                 |
rlm@3: // +----------------------------------------------------------------------+
rlm@3: // | getID3() - http://getid3.sourceforge.net or http://www.getid3.org    |
rlm@3: // +----------------------------------------------------------------------+
rlm@3: // | Authors: James Heinrich <infoØgetid3*org>                            |
rlm@3: // |          Allan Hansen <ahØartemis*dk>                                |
rlm@3: // +----------------------------------------------------------------------+
rlm@3: // | module.archive.gzip.php                                              |
rlm@3: // | module for analyzing GZIP files                                      |
rlm@3: // | dependencies: PHP compiled with zlib support (optional)              |
rlm@3: // +----------------------------------------------------------------------+
rlm@3: // | Module originally written by Mike Mozolin <teddybearØmail*ru>        |
rlm@3: // +----------------------------------------------------------------------+
rlm@3: //
rlm@3: // $Id: module.archive.gzip.php,v 1.4 2006/12/04 16:00:35 ah Exp $
rlm@3: 
rlm@3:         
rlm@3:         
rlm@3: class getid3_gzip extends getid3_handler
rlm@3: {
rlm@3: 
rlm@3:     // public: Optional file list - disable for speed.
rlm@3:     public $option_gzip_parse_contents = true; // decode gzipped files, if possible, and parse recursively (.tar.gz for example)
rlm@3: 
rlm@3:     
rlm@3:     // Reads the gzip-file
rlm@3:     function Analyze() {
rlm@3: 
rlm@3:         $info = &$this->getid3->info;
rlm@3: 
rlm@3:         $info['fileformat'] = 'gzip';
rlm@3: 
rlm@3:         $start_length = 10;
rlm@3:         $unpack_header = 'a1id1/a1id2/a1cmethod/a1flags/a4mtime/a1xflags/a1os';
rlm@3:         
rlm@3:         //+---+---+---+---+---+---+---+---+---+---+
rlm@3:         //|ID1|ID2|CM |FLG|     MTIME     |XFL|OS |
rlm@3:         //+---+---+---+---+---+---+---+---+---+---+
rlm@3:         
rlm@3:         @fseek($this->getid3->fp, 0);
rlm@3:         $buffer = @fread($this->getid3->fp, $info['filesize']);
rlm@3: 
rlm@3:         $arr_members = explode("\x1F\x8B\x08", $buffer);
rlm@3:         
rlm@3:         while (true) {
rlm@3:             $is_wrong_members = false;
rlm@3:             $num_members = intval(count($arr_members));
rlm@3:             for ($i = 0; $i < $num_members; $i++) {
rlm@3:                 if (strlen($arr_members[$i]) == 0) {
rlm@3:                     continue;
rlm@3:                 }
rlm@3:                 $buf = "\x1F\x8B\x08".$arr_members[$i];
rlm@3: 
rlm@3:                 $attr = unpack($unpack_header, substr($buf, 0, $start_length));
rlm@3:                 if (!$this->get_os_type(ord($attr['os']))) {
rlm@3:         
rlm@3:                     // Merge member with previous if wrong OS type
rlm@3:                     $arr_members[$i - 1] .= $buf;
rlm@3:                     $arr_members[$i] = '';
rlm@3:                     $is_wrong_members = true;
rlm@3:                     continue;
rlm@3:                 }
rlm@3:             }
rlm@3:             if (!$is_wrong_members) {
rlm@3:                 break;
rlm@3:             }
rlm@3:         }
rlm@3: 
rlm@3:         $fpointer = 0;
rlm@3:         $idx = 0;
rlm@3:         for ($i = 0; $i < $num_members; $i++) {
rlm@3:             if (strlen($arr_members[$i]) == 0) {
rlm@3:                 continue;
rlm@3:             }
rlm@3:             $info_gzip_member_header_idx = &$info['gzip']['member_header'][++$idx];
rlm@3: 
rlm@3:             $buff = "\x1F\x8B\x08".$arr_members[$i];
rlm@3: 
rlm@3:             $attr = unpack($unpack_header, substr($buff, 0, $start_length));
rlm@3:             $info_gzip_member_header_idx['filemtime']      = getid3_lib::LittleEndian2Int($attr['mtime']);
rlm@3:             $info_gzip_member_header_idx['raw']['id1']     = ord($attr['cmethod']);
rlm@3:             $info_gzip_member_header_idx['raw']['id2']     = ord($attr['cmethod']);
rlm@3:             $info_gzip_member_header_idx['raw']['cmethod'] = ord($attr['cmethod']);
rlm@3:             $info_gzip_member_header_idx['raw']['os']      = ord($attr['os']);
rlm@3:             $info_gzip_member_header_idx['raw']['xflags']  = ord($attr['xflags']);
rlm@3:             $info_gzip_member_header_idx['raw']['flags']   = ord($attr['flags']);
rlm@3: 
rlm@3:             $info_gzip_member_header_idx['flags']['crc16']    = (bool) ($info_gzip_member_header_idx['raw']['flags'] & 0x02);
rlm@3:             $info_gzip_member_header_idx['flags']['extra']    = (bool) ($info_gzip_member_header_idx['raw']['flags'] & 0x04);
rlm@3:             $info_gzip_member_header_idx['flags']['filename'] = (bool) ($info_gzip_member_header_idx['raw']['flags'] & 0x08);
rlm@3:             $info_gzip_member_header_idx['flags']['comment']  = (bool) ($info_gzip_member_header_idx['raw']['flags'] & 0x10);
rlm@3: 
rlm@3:             $info_gzip_member_header_idx['compression'] = $this->get_xflag_type($info_gzip_member_header_idx['raw']['xflags']);
rlm@3: 
rlm@3:             $info_gzip_member_header_idx['os'] = $this->get_os_type($info_gzip_member_header_idx['raw']['os']);
rlm@3:             if (!$info_gzip_member_header_idx['os']) {
rlm@3:                 $info['error'][] = 'Read error on gzip file';
rlm@3:                 return false;
rlm@3:             }
rlm@3: 
rlm@3:             $fpointer = 10;
rlm@3:             $arr_xsubfield = array ();
rlm@3:             
rlm@3:             // bit 2 - FLG.FEXTRA
rlm@3:             //+---+---+=================================+
rlm@3:             //| XLEN  |...XLEN bytes of "extra field"...|
rlm@3:             //+---+---+=================================+
rlm@3:             
rlm@3:             if ($info_gzip_member_header_idx['flags']['extra']) {
rlm@3:                 $w_xlen = substr($buff, $fpointer, 2);
rlm@3:                 $xlen = getid3_lib::LittleEndian2Int($w_xlen);
rlm@3:                 $fpointer += 2;
rlm@3: 
rlm@3:                 $info_gzip_member_header_idx['raw']['xfield'] = substr($buff, $fpointer, $xlen);
rlm@3:             
rlm@3:                 // Extra SubFields
rlm@3:                 //+---+---+---+---+==================================+
rlm@3:                 //|SI1|SI2|  LEN  |... LEN bytes of subfield data ...|
rlm@3:                 //+---+---+---+---+==================================+
rlm@3:             
rlm@3:                 $idx = 0;
rlm@3:                 while (true) {
rlm@3:                     if ($idx >= $xlen) {
rlm@3:                         break;
rlm@3:                     }
rlm@3:                     $si1 = ord(substr($buff, $fpointer + $idx++, 1));
rlm@3:                     $si2 = ord(substr($buff, $fpointer + $idx++, 1));
rlm@3:                     if (($si1 == 0x41) && ($si2 == 0x70)) {
rlm@3:                         $w_xsublen = substr($buff, $fpointer+$idx, 2);
rlm@3:                         $xsublen = getid3_lib::LittleEndian2Int($w_xsublen);
rlm@3:                         $idx += 2;
rlm@3:                         $arr_xsubfield[] = substr($buff, $fpointer+$idx, $xsublen);
rlm@3:                         $idx += $xsublen;
rlm@3:                     } else {
rlm@3:                         break;
rlm@3:                     }
rlm@3:                 }
rlm@3:                 $fpointer += $xlen;
rlm@3:             }
rlm@3:             
rlm@3:             // bit 3 - FLG.FNAME
rlm@3:             //+=========================================+
rlm@3:             //|...original file name, zero-terminated...|
rlm@3:             //+=========================================+
rlm@3:             // GZIP files may have only one file, with no filename, so assume original filename is current filename without .gz
rlm@3:             
rlm@3:             $info_gzip_member_header_idx['filename'] = eregi_replace('.gz$', '', @$info['filename']);
rlm@3:             if ($info_gzip_member_header_idx['flags']['filename']) {
rlm@3:                 while (true) {
rlm@3:                     if (ord($buff[$fpointer]) == 0) {
rlm@3:                         $fpointer++;
rlm@3:                         break;
rlm@3:                     }
rlm@3:                     $info_gzip_member_header_idx['filename'] .= $buff[$fpointer];
rlm@3:                     $fpointer++;
rlm@3:                 }
rlm@3:             }
rlm@3:             
rlm@3:             // bit 4 - FLG.FCOMMENT
rlm@3:             //+===================================+
rlm@3:             //|...file comment, zero-terminated...|
rlm@3:             //+===================================+
rlm@3:             
rlm@3:             if ($info_gzip_member_header_idx['flags']['comment']) {
rlm@3:                 while (true) {
rlm@3:                     if (ord($buff[$fpointer]) == 0) {
rlm@3:                         $fpointer++;
rlm@3:                         break;
rlm@3:                     }
rlm@3:                     $info_gzip_member_header_idx['comment'] .= $buff[$fpointer];
rlm@3:                     $fpointer++;
rlm@3:                 }
rlm@3:             }
rlm@3:             
rlm@3:             // bit 1 - FLG.FHCRC
rlm@3:             //+---+---+
rlm@3:             //| CRC16 |
rlm@3:             //+---+---+
rlm@3:             
rlm@3:             if ($info_gzip_member_header_idx['flags']['crc16']) {
rlm@3:                 $w_crc = substr($buff, $fpointer, 2);
rlm@3:                 $info_gzip_member_header_idx['crc16'] = getid3_lib::LittleEndian2Int($w_crc);
rlm@3:                 $fpointer += 2;
rlm@3:             }
rlm@3:             
rlm@3:             // bit 0 - FLG.FTEXT
rlm@3:             //if ($info_gzip_member_header_idx['raw']['flags'] & 0x01) {
rlm@3:             //  Ignored...
rlm@3:             //}
rlm@3:             // bits 5, 6, 7 - reserved
rlm@3: 
rlm@3:             $info_gzip_member_header_idx['crc32']    = getid3_lib::LittleEndian2Int(substr($buff, strlen($buff) - 8, 4));
rlm@3:             $info_gzip_member_header_idx['filesize'] = getid3_lib::LittleEndian2Int(substr($buff, strlen($buff) - 4));
rlm@3: 
rlm@3:             if ($this->option_gzip_parse_contents) {
rlm@3: 
rlm@3:                 // Try to inflate GZip
rlm@3:                 
rlm@3:                 if (!function_exists('gzinflate')) {
rlm@3:                     $this->getid3->warning('PHP does not have zlib support - contents not parsed.');
rlm@3:                     return true;
rlm@3:                 }
rlm@3: 
rlm@3:                 $csize = 0;
rlm@3:                 $inflated = '';
rlm@3:                 $chkcrc32 = '';
rlm@3: 
rlm@3:                 $cdata = substr($buff, $fpointer);
rlm@3:                 $cdata = substr($cdata, 0, strlen($cdata) - 8);
rlm@3:                 $csize = strlen($cdata);
rlm@3:                 $inflated = gzinflate($cdata);
rlm@3: 
rlm@3:                 // Calculate CRC32 for inflated content
rlm@3:                 $info_gzip_member_header_idx['crc32_valid'] = (bool) (sprintf('%u', crc32($inflated)) == $info_gzip_member_header_idx['crc32']);
rlm@3: 
rlm@3:                 
rlm@3:                 //// Analyse contents
rlm@3:                 
rlm@3:                 // write content to temp file
rlm@3:                 if (($temp_file_name = tempnam('*', 'getID3'))  === false) {
rlm@3:                     throw new getid3_exception('Unable to create temporary file.');
rlm@3:                 }
rlm@3: 
rlm@3:                 if ($tmp = fopen($temp_file_name, 'wb')) {
rlm@3:                     fwrite($tmp, $inflated);
rlm@3:                     fclose($tmp);
rlm@3:                     
rlm@3:                     // clone getid3 - we want same settings
rlm@3:                     $clone = clone $this->getid3;
rlm@3:                     unset($clone->info);
rlm@3:                     try {
rlm@3:                         $clone->Analyze($temp_file_name);
rlm@3:                         $info_gzip_member_header_idx['parsed_content'] = $clone->info;
rlm@3:                     }
rlm@3:                     catch (getid3_exception $e) {
rlm@3:                         // unable to parse contents
rlm@3:                     }
rlm@3:                     
rlm@3:                     unlink($temp_file_name);
rlm@3:                 }
rlm@3:             
rlm@3:                 // Unknown/unhandled format 
rlm@3:                 else {
rlm@3:                                         
rlm@3:                 }
rlm@3:             }
rlm@3:         }
rlm@3:         return true;
rlm@3:     }
rlm@3: 
rlm@3: 
rlm@3:     // Converts the OS type
rlm@3:     public static function get_os_type($key) {
rlm@3:         static $os_type = array (
rlm@3:             '0'   => 'FAT filesystem (MS-DOS, OS/2, NT/Win32)',
rlm@3:             '1'   => 'Amiga',
rlm@3:             '2'   => 'VMS (or OpenVMS)',
rlm@3:             '3'   => 'Unix',
rlm@3:             '4'   => 'VM/CMS',
rlm@3:             '5'   => 'Atari TOS',
rlm@3:             '6'   => 'HPFS filesystem (OS/2, NT)',
rlm@3:             '7'   => 'Macintosh',
rlm@3:             '8'   => 'Z-System',
rlm@3:             '9'   => 'CP/M',
rlm@3:             '10'  => 'TOPS-20',
rlm@3:             '11'  => 'NTFS filesystem (NT)',
rlm@3:             '12'  => 'QDOS',
rlm@3:             '13'  => 'Acorn RISCOS',
rlm@3:             '255' => 'unknown'
rlm@3:         );
rlm@3:         return @$os_type[$key];
rlm@3:     }
rlm@3: 
rlm@3: 
rlm@3:     // Converts the eXtra FLags
rlm@3:     public static function get_xflag_type($key) {
rlm@3:         static $xflag_type = array (
rlm@3:             '0' => 'unknown',
rlm@3:             '2' => 'maximum compression',
rlm@3:             '4' => 'fastest algorithm'
rlm@3:         );
rlm@3:         return @$xflag_type[$key];
rlm@3:     }
rlm@3:     
rlm@3: }
rlm@3: 
rlm@3: ?>