Mercurial > judyates

     1 <?php

     2 // +----------------------------------------------------------------------+

     3 // | PHP version 5                                                        |

     4 // +----------------------------------------------------------------------+

     5 // | Copyright (c) 2002-2006 James Heinrich, Allan Hansen                 |

     6 // +----------------------------------------------------------------------+

     7 // | This source file is subject to version 2 of the GPL license,         |

     8 // | that is bundled with this package in the file license.txt and is     |

     9 // | available through the world-wide-web at the following url:           |

    10 // | http://www.gnu.org/copyleft/gpl.html                                 |

    11 // +----------------------------------------------------------------------+

    12 // | getID3() - http://getid3.sourceforge.net or http://www.getid3.org    |

    13 // +----------------------------------------------------------------------+

    14 // | Authors: James Heinrich <infoØgetid3*org>                            |

    15 // |          Allan Hansen <ahØartemis*dk>                                |

    16 // +----------------------------------------------------------------------+

    17 // | module.archive.zip.php                                               |

    18 // | Module for analyzing pkZip files                                     |

    19 // | dependencies: NONE                                                   |

    20 // +----------------------------------------------------------------------+

    21 //

    22 // $Id: module.archive.zip.php,v 1.4 2006/11/02 10:48:00 ah Exp $

    23 

    24 

    25 

    26 class getid3_zip extends getid3_handler

    27 {

    28 

    29     public function Analyze() {

    30 

    31         $getid3 = $this->getid3;

    32         

    33         $getid3->info['zip'] = array ();

    34         $info_zip = &$getid3->info['zip'];

    35         

    36         $getid3->info['fileformat'] = 'zip';

    37         

    38         $info_zip['encoding'] = 'ISO-8859-1';

    39         $info_zip['files']    = array ();

    40         $info_zip['compressed_size'] = $info_zip['uncompressed_size'] = $info_zip['entries_count'] = 0;

    41 

    42         $eocd_search_data    = '';

    43         $eocd_search_counter = 0;

    44         while ($eocd_search_counter++ < 512) {

    45 

    46             fseek($getid3->fp, -128 * $eocd_search_counter, SEEK_END);

    47             $eocd_search_data = fread($getid3->fp, 128).$eocd_search_data;

    48 

    49             if (strstr($eocd_search_data, 'PK'."\x05\x06")) {

    50 

    51                 $eocd_position = strpos($eocd_search_data, 'PK'."\x05\x06");

    52                 fseek($getid3->fp, (-128 * $eocd_search_counter) + $eocd_position, SEEK_END);

    53                 $info_zip['end_central_directory'] = $this->ZIPparseEndOfCentralDirectory();

    54 

    55                 fseek($getid3->fp, $info_zip['end_central_directory']['directory_offset'], SEEK_SET);

    56                 $info_zip['entries_count'] = 0;

    57                 while ($central_directoryentry = $this->ZIPparseCentralDirectory($getid3->fp)) {

    58                     $info_zip['central_directory'][] = $central_directoryentry;

    59                     $info_zip['entries_count']++;

    60                     $info_zip['compressed_size']   += $central_directoryentry['compressed_size'];

    61                     $info_zip['uncompressed_size'] += $central_directoryentry['uncompressed_size'];

    62 

    63                     if ($central_directoryentry['uncompressed_size'] > 0) {

    64                         $info_zip['files'] = getid3_zip::array_merge_clobber($info_zip['files'], getid3_zip::CreateDeepArray($central_directoryentry['filename'], '/', $central_directoryentry['uncompressed_size']));

    65                     }

    66                 }

    67 

    68                 if ($info_zip['entries_count'] == 0) {

    69                     throw new getid3_exception('No Central Directory entries found (truncated file?)');

    70                 }

    71 

    72                 if (!empty($info_zip['end_central_directory']['comment'])) {

    73                     $info_zip['comments']['comment'][] = $info_zip['end_central_directory']['comment'];

    74                 }

    75 

    76                 if (isset($info_zip['central_directory'][0]['compression_method'])) {

    77                     $info_zip['compression_method'] = $info_zip['central_directory'][0]['compression_method'];

    78                 }

    79                 if (isset($info_zip['central_directory'][0]['flags']['compression_speed'])) {

    80                     $info_zip['compression_speed']  = $info_zip['central_directory'][0]['flags']['compression_speed'];

    81                 }

    82                 if (isset($info_zip['compression_method']) && ($info_zip['compression_method'] == 'store') && !isset($info_zip['compression_speed'])) {

    83                     $info_zip['compression_speed']  = 'store';

    84                 }

    85 

    86                 return true;

    87             }

    88         }

    89 

    90         if ($this->getZIPentriesFilepointer()) {

    91 

    92             // central directory couldn't be found and/or parsed

    93             // scan through actual file data entries, recover as much as possible from probable trucated file

    94             if (@$info_zip['compressed_size'] > ($getid3->info['filesize'] - 46 - 22)) {

    95                 throw new getid3_exception('Warning: Truncated file! - Total compressed file sizes ('.$info_zip['compressed_size'].' bytes) is greater than filesize minus Central Directory and End Of Central Directory structures ('.($getid3->info['filesize'] - 46 - 22).' bytes)');

    96             }

    97             throw new getid3_exception('Cannot find End Of Central Directory - returned list of files in [zip][entries] array may not be complete');

    98         }

    99         

   100         //throw new getid3_exception('Cannot find End Of Central Directory (truncated file?)');

   101     }

   102 

   103 

   104 

   105     private function getZIPHeaderFilepointerTopDown() {

   106         

   107         // shortcut

   108         $getid3 = $this->getid3;

   109         

   110         $getid3->info['fileformat'] = 'zip';

   111         

   112         $getid3->info['zip'] = array ();

   113         $info_zip['compressed_size'] = $info_zip['uncompressed_size'] = $info_zip['entries_count'] = 0;

   114         

   115         rewind($getid3->fp);

   116         while ($fileentry = $this->ZIPparseLocalFileHeader()) {

   117             $info_zip['entries'][] = $fileentry;

   118             $info_zip['entries_count']++;

   119         }

   120         if ($info_zip['entries_count'] == 0) {

   121             throw new getid3_exception('No Local File Header entries found');

   122         }

   123 

   124         $info_zip['entries_count']  = 0;

   125         while ($central_directoryentry = $this->ZIPparseCentralDirectory($getid3->fp)) {

   126             $info_zip['central_directory'][] = $central_directoryentry;

   127             $info_zip['entries_count']++;

   128             $info_zip['compressed_size']   += $central_directoryentry['compressed_size'];

   129             $info_zip['uncompressed_size'] += $central_directoryentry['uncompressed_size'];

   130         }

   131         if ($info_zip['entries_count'] == 0) {

   132             throw new getid3_exception('No Central Directory entries found (truncated file?)');

   133         }

   134 

   135         if ($eocd = $this->ZIPparseEndOfCentralDirectory()) {

   136             $info_zip['end_central_directory'] = $eocd;

   137         } else {

   138             throw new getid3_exception('No End Of Central Directory entry found (truncated file?)');

   139         }

   140 

   141         if (!@$info_zip['end_central_directory']['comment']) {

   142             $info_zip['comments']['comment'][] = $info_zip['end_central_directory']['comment'];

   143         }

   144 

   145         return true;

   146     }

   147 

   148 

   149 

   150     private function getZIPentriesFilepointer() {

   151         

   152         // shortcut

   153         $getid3 = $this->getid3;

   154         

   155         $getid3->info['zip'] = array ();

   156         $info_zip['compressed_size'] = $info_zip['uncompressed_size'] = $info_zip['entries_count'] = 0;

   157 

   158         rewind($getid3->fp);

   159         while ($fileentry = $this->ZIPparseLocalFileHeader($getid3->fp)) {

   160             $info_zip['entries'][] = $fileentry;

   161             $info_zip['entries_count']++;

   162             $info_zip['compressed_size']   += $fileentry['compressed_size'];

   163             $info_zip['uncompressed_size'] += $fileentry['uncompressed_size'];

   164         }

   165         if ($info_zip['entries_count'] == 0) {

   166             throw new getid3_exception('No Local File Header entries found');

   167         }

   168 

   169         return true;

   170     }

   171 

   172 

   173 

   174     private function ZIPparseLocalFileHeader() {

   175         

   176         // shortcut

   177         $getid3 = $this->getid3;

   178 

   179         $local_file_header['offset'] = ftell($getid3->fp);

   180         

   181         $zip_local_file_header = fread($getid3->fp, 30);

   182 

   183         $local_file_header['raw']['signature'] = getid3_lib::LittleEndian2Int(substr($zip_local_file_header,  0, 4));

   184         

   185         // Invalid Local File Header Signature

   186         if ($local_file_header['raw']['signature'] != 0x04034B50) {

   187             fseek($getid3->fp, $local_file_header['offset'], SEEK_SET); // seek back to where filepointer originally was so it can be handled properly

   188             return false;

   189         }

   190         

   191         getid3_lib::ReadSequence('LittleEndian2Int', $local_file_header['raw'], $zip_local_file_header,  4, 

   192             array (

   193                 'extract_version'    => 2, 

   194                 'general_flags'      => 2, 

   195                 'compression_method' => 2, 

   196                 'last_mod_file_time' => 2, 

   197                 'last_mod_file_date' => 2, 

   198                 'crc_32'             => 2, 

   199                 'compressed_size'    => 2, 

   200                 'uncompressed_size'  => 2, 

   201                 'filename_length'    => 2, 

   202                 'extra_field_length' => 2

   203             )

   204         );        

   205 

   206         $local_file_header['extract_version']         = sprintf('%1.1f', $local_file_header['raw']['extract_version'] / 10);

   207         $local_file_header['host_os']                 = $this->ZIPversionOSLookup(($local_file_header['raw']['extract_version'] & 0xFF00) >> 8);

   208         $local_file_header['compression_method']      = $this->ZIPcompressionMethodLookup($local_file_header['raw']['compression_method']);

   209         $local_file_header['compressed_size']         = $local_file_header['raw']['compressed_size'];

   210         $local_file_header['uncompressed_size']       = $local_file_header['raw']['uncompressed_size'];

   211         $local_file_header['flags']                   = $this->ZIPparseGeneralPurposeFlags($local_file_header['raw']['general_flags'], $local_file_header['raw']['compression_method']);

   212         $local_file_header['last_modified_timestamp'] = $this->DOStime2UNIXtime($local_file_header['raw']['last_mod_file_date'], $local_file_header['raw']['last_mod_file_time']);

   213 

   214         $filename_extra_field_length = $local_file_header['raw']['filename_length'] + $local_file_header['raw']['extra_field_length'];

   215         if ($filename_extra_field_length > 0) {

   216             $zip_local_file_header .= fread($getid3->fp, $filename_extra_field_length);

   217 

   218             if ($local_file_header['raw']['filename_length'] > 0) {

   219                 $local_file_header['filename'] = substr($zip_local_file_header, 30, $local_file_header['raw']['filename_length']);

   220             }

   221             if ($local_file_header['raw']['extra_field_length'] > 0) {

   222                 $local_file_header['raw']['extra_field_data'] = substr($zip_local_file_header, 30 + $local_file_header['raw']['filename_length'], $local_file_header['raw']['extra_field_length']);

   223             }

   224         }

   225 

   226         $local_file_header['data_offset'] = ftell($getid3->fp);

   227         fseek($getid3->fp, $local_file_header['raw']['compressed_size'], SEEK_CUR);

   228 

   229         if ($local_file_header['flags']['data_descriptor_used']) {

   230             $data_descriptor = fread($getid3->fp, 12);

   231             

   232             getid3_lib::ReadSequence('LittleEndian2Int', $local_file_header['data_descriptor'], $data_descriptor, 0, 

   233                 array (

   234                 'crc_32'            => 4,

   235                 'compressed_size'   => 4,

   236                 'uncompressed_size' => 4 

   237                 )

   238             );

   239         }

   240 

   241         return $local_file_header;

   242     }

   243 

   244 

   245 

   246     private function ZIPparseCentralDirectory() {

   247         

   248         // shortcut

   249         $getid3 = $this->getid3;

   250 

   251         $central_directory['offset'] = ftell($getid3->fp);

   252 

   253         $zip_central_directory = fread($getid3->fp, 46);

   254 

   255         $central_directory['raw']['signature']  = getid3_lib::LittleEndian2Int(substr($zip_central_directory,  0, 4));

   256         

   257         // invalid Central Directory Signature

   258         if ($central_directory['raw']['signature'] != 0x02014B50) {

   259             fseek($getid3->fp, $central_directory['offset'], SEEK_SET); // seek back to where filepointer originally was so it can be handled properly

   260             return false;

   261         }

   262         

   263         getid3_lib::ReadSequence('LittleEndian2Int', $central_directory['raw'], $zip_central_directory,  4, 

   264             array (

   265                 'create_version'       => 2,

   266                 'extract_version'      => 2,

   267                 'general_flags'        => 2,

   268                 'compression_method'   => 2,

   269                 'last_mod_file_time'   => 2,

   270                 'last_mod_file_date'   => 2,

   271                 'crc_32'               => 4,

   272                 'compressed_size'      => 4,

   273                 'uncompressed_size'    => 4,

   274                 'filename_length'      => 2,

   275                 'extra_field_length'   => 2,

   276                 'file_comment_length'  => 2,

   277                 'disk_number_start'    => 2,

   278                 'internal_file_attrib' => 2,

   279                 'external_file_attrib' => 4,

   280                 'local_header_offset'  => 4

   281             )

   282         );

   283         

   284         $central_directory['entry_offset']            = $central_directory['raw']['local_header_offset'];

   285         $central_directory['create_version']          = sprintf('%1.1f', $central_directory['raw']['create_version'] / 10);

   286         $central_directory['extract_version']         = sprintf('%1.1f', $central_directory['raw']['extract_version'] / 10);

   287         $central_directory['host_os']                 = $this->ZIPversionOSLookup(($central_directory['raw']['extract_version'] & 0xFF00) >> 8);

   288         $central_directory['compression_method']      = $this->ZIPcompressionMethodLookup($central_directory['raw']['compression_method']);

   289         $central_directory['compressed_size']         = $central_directory['raw']['compressed_size'];

   290         $central_directory['uncompressed_size']       = $central_directory['raw']['uncompressed_size'];

   291         $central_directory['flags']                   = $this->ZIPparseGeneralPurposeFlags($central_directory['raw']['general_flags'], $central_directory['raw']['compression_method']);

   292         $central_directory['last_modified_timestamp'] = $this->DOStime2UNIXtime($central_directory['raw']['last_mod_file_date'], $central_directory['raw']['last_mod_file_time']);

   293 

   294         $filename_extra_field_comment_length = $central_directory['raw']['filename_length'] + $central_directory['raw']['extra_field_length'] + $central_directory['raw']['file_comment_length'];

   295         if ($filename_extra_field_comment_length > 0) {

   296             $filename_extra_field_comment = fread($getid3->fp, $filename_extra_field_comment_length);

   297 

   298             if ($central_directory['raw']['filename_length'] > 0) {

   299                 $central_directory['filename']= substr($filename_extra_field_comment, 0, $central_directory['raw']['filename_length']);

   300             }

   301             if ($central_directory['raw']['extra_field_length'] > 0) {

   302                 $central_directory['raw']['extra_field_data'] = substr($filename_extra_field_comment, $central_directory['raw']['filename_length'], $central_directory['raw']['extra_field_length']);

   303             }

   304             if ($central_directory['raw']['file_comment_length'] > 0) {

   305                 $central_directory['file_comment'] = substr($filename_extra_field_comment, $central_directory['raw']['filename_length'] + $central_directory['raw']['extra_field_length'], $central_directory['raw']['file_comment_length']);

   306             }

   307         }

   308 

   309         return $central_directory;

   310     }

   311 

   312     

   313     

   314     private function ZIPparseEndOfCentralDirectory() {

   315         

   316         // shortcut             

   317         $getid3 = $this->getid3;

   318     

   319         $end_of_central_directory['offset'] = ftell($getid3->fp);

   320 

   321         $zip_end_of_central_directory = fread($getid3->fp, 22);

   322 

   323         $end_of_central_directory['signature'] = getid3_lib::LittleEndian2Int(substr($zip_end_of_central_directory,  0, 4));

   324         

   325         // invalid End Of Central Directory Signature

   326         if ($end_of_central_directory['signature'] != 0x06054B50) {

   327             fseek($getid3->fp, $end_of_central_directory['offset'], SEEK_SET); // seek back to where filepointer originally was so it can be handled properly

   328             return false;

   329         }

   330         

   331         getid3_lib::ReadSequence('LittleEndian2Int', $end_of_central_directory, $zip_end_of_central_directory,  4, 

   332             array (

   333                 'disk_number_current'         => 2,

   334                 'disk_number_start_directory' => 2,

   335                 'directory_entries_this_disk' => 2,

   336                 'directory_entries_total'     => 2,

   337                 'directory_size'              => 4,

   338                 'directory_offset'            => 4,

   339                 'comment_length'              => 2

   340             )

   341         );

   342         

   343         if ($end_of_central_directory['comment_length'] > 0) {

   344             $end_of_central_directory['comment'] = fread($getid3->fp, $end_of_central_directory['comment_length']);

   345         }

   346 

   347         return $end_of_central_directory;

   348     }

   349     

   350 

   351 

   352     public static function ZIPparseGeneralPurposeFlags($flag_bytes, $compression_method) {

   353 

   354         $parsed_flags['encrypted'] = (bool)($flag_bytes & 0x0001);

   355 

   356         switch ($compression_method) {

   357             case 6:

   358                 $parsed_flags['dictionary_size']    = (($flag_bytes & 0x0002) ? 8192 : 4096);

   359                 $parsed_flags['shannon_fano_trees'] = (($flag_bytes & 0x0004) ? 3    : 2);

   360                 break;

   361 

   362             case 8:

   363             case 9:

   364                 switch (($flag_bytes & 0x0006) >> 1) {

   365                     case 0:

   366                         $parsed_flags['compression_speed'] = 'normal';

   367                         break;

   368                     case 1:

   369                         $parsed_flags['compression_speed'] = 'maximum';

   370                         break;

   371                     case 2:

   372                         $parsed_flags['compression_speed'] = 'fast';

   373                         break;

   374                     case 3:

   375                         $parsed_flags['compression_speed'] = 'superfast';

   376                         break;

   377                 }

   378                 break;

   379         }

   380         $parsed_flags['data_descriptor_used'] = (bool)($flag_bytes & 0x0008);

   381 

   382         return $parsed_flags;

   383     }

   384 

   385 

   386 

   387     public static function ZIPversionOSLookup($index) {

   388         

   389         static $lookup = array (

   390             0  => 'MS-DOS and OS/2 (FAT / VFAT / FAT32 file systems)',

   391             1  => 'Amiga',

   392             2  => 'OpenVMS',

   393             3  => 'Unix',

   394             4  => 'VM/CMS',

   395             5  => 'Atari ST',

   396             6  => 'OS/2 H.P.F.S.',

   397             7  => 'Macintosh',

   398             8  => 'Z-System',

   399             9  => 'CP/M',

   400             10 => 'Windows NTFS',

   401             11 => 'MVS',

   402             12 => 'VSE',

   403             13 => 'Acorn Risc',

   404             14 => 'VFAT',

   405             15 => 'Alternate MVS',

   406             16 => 'BeOS',

   407             17 => 'Tandem'

   408         );

   409         return (isset($lookup[$index]) ? $lookup[$index] : '[unknown]');

   410     }

   411 

   412 

   413 

   414     public static function ZIPcompressionMethodLookup($index) {

   415 

   416         static $lookup = array (

   417             0  => 'store',

   418             1  => 'shrink',

   419             2  => 'reduce-1',

   420             3  => 'reduce-2',

   421             4  => 'reduce-3',

   422             5  => 'reduce-4',

   423             6  => 'implode',

   424             7  => 'tokenize',

   425             8  => 'deflate',

   426             9  => 'deflate64',

   427             10 => 'PKWARE Date Compression Library Imploding'

   428         );

   429         return (isset($lookup[$index]) ? $lookup[$index] : '[unknown]');

   430     }

   431 

   432 

   433 

   434     public static function DOStime2UNIXtime($DOSdate, $DOStime) {

   435 

   436         /*

   437         // wFatDate

   438         // Specifies the MS-DOS date. The date is a packed 16-bit value with the following format:

   439         // Bits      Contents

   440         // 0-4    Day of the month (1-31)

   441         // 5-8    Month (1 = January, 2 = February, and so on)

   442         // 9-15   Year offset from 1980 (add 1980 to get actual year)

   443 

   444         $UNIXday    =  ($DOSdate & 0x001F);

   445         $UNIXmonth  = (($DOSdate & 0x01E0) >> 5);

   446         $UNIXyear   = (($DOSdate & 0xFE00) >> 9) + 1980;

   447 

   448         // wFatTime

   449         // Specifies the MS-DOS time. The time is a packed 16-bit value with the following format:

   450         // Bits   Contents

   451         // 0-4    Second divided by 2

   452         // 5-10   Minute (0-59)

   453         // 11-15  Hour (0-23 on a 24-hour clock)

   454 

   455         $UNIXsecond =  ($DOStime & 0x001F) * 2;

   456         $UNIXminute = (($DOStime & 0x07E0) >> 5);

   457         $UNIXhour   = (($DOStime & 0xF800) >> 11);

   458 

   459         return gmmktime($UNIXhour, $UNIXminute, $UNIXsecond, $UNIXmonth, $UNIXday, $UNIXyear);

   460         */

   461         return gmmktime(($DOStime & 0xF800) >> 11, ($DOStime & 0x07E0) >> 5, ($DOStime & 0x001F) * 2, ($DOSdate & 0x01E0) >> 5, $DOSdate & 0x001F, (($DOSdate & 0xFE00) >> 9) + 1980);

   462     }

   463     

   464     

   465     

   466     public static function array_merge_clobber($array1, $array2) {

   467 

   468         // written by kcØhireability*com

   469         // taken from http://www.php.net/manual/en/function.array-merge-recursive.php

   470         

   471         if (!is_array($array1) || !is_array($array2)) {

   472             return false;

   473         }

   474         

   475         $newarray = $array1;

   476         foreach ($array2 as $key => $val) {

   477             if (is_array($val) && isset($newarray[$key]) && is_array($newarray[$key])) {

   478                 $newarray[$key] = getid3_zip::array_merge_clobber($newarray[$key], $val);

   479             } else {

   480                 $newarray[$key] = $val;

   481             }

   482         }

   483         return $newarray;

   484     }

   485     

   486     

   487     

   488     public static function CreateDeepArray($array_path, $separator, $value) {

   489 

   490         // assigns $value to a nested array path:

   491         //   $foo = getid3_lib::CreateDeepArray('/path/to/my', '/', 'file.txt')

   492         // is the same as:

   493         //   $foo = array ('path'=>array('to'=>'array('my'=>array('file.txt'))));

   494         // or

   495         //   $foo['path']['to']['my'] = 'file.txt';

   496         

   497         while ($array_path{0} == $separator) {

   498             $array_path = substr($array_path, 1);

   499         }

   500         if (($pos = strpos($array_path, $separator)) !== false) {

   501             return array (substr($array_path, 0, $pos) => getid3_zip::CreateDeepArray(substr($array_path, $pos + 1), $separator, $value));

   502         }

   503         

   504         return array ($array_path => $value);

   505     }

   506 

   507 }

   508 

   509 

   510 ?>