includes/libs/mime/MSCompoundFileReader.php

   1 <?php
   2 /*
   3  * Copyright 2019 Wikimedia Foundation
   4  *
   5  * Licensed under the Apache License, Version 2.0 (the "License"); you may
   6  * not use this file except in compliance with the License.
   7  * You may obtain a copy of the License at
   8  *
   9  *     http://www.apache.org/licenses/LICENSE-2.0
  10  *
  11  * Unless required by applicable law or agreed to in writing, software distributed
  12  * under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS
  13  * OF ANY KIND, either express or implied. See the License for the
  14  * specific language governing permissions and limitations under the License.
  15  */
  16
  17 /**
  18  * Read the directory of a Microsoft Compound File Binary file, a.k.a. an OLE
  19  * file, and detect the MIME type.
  20  *
  21  * References:
  22  *  - MS-CFB https://msdn.microsoft.com/en-us/library/dd942138.aspx
  23  *  - MS-XLS https://msdn.microsoft.com/en-us/library/cc313154.aspx
  24  *  - MS-PPT https://msdn.microsoft.com/en-us/library/cc313106.aspx
  25  *  - MS-DOC https://msdn.microsoft.com/en-us/library/cc313153.aspx
  26  *  - Python olefile https://github.com/decalage2/olefile
  27  *  - OpenOffice.org's Documentation of the Microsoft Compound Document
  28  *    File Format https://www.openoffice.org/sc/compdocfileformat.pdf
  29  *
  30  * @since 1.33
  31  */
  32 class MSCompoundFileReader {
  33         private $file;
  34         private $header;
  35         private $mime;
  36         private $mimeFromClsid;
  37         private $error;
  38         private $errorCode;
  39         private $valid = false;
  40
  41         private $sectorLength;
  42         private $difat;
  43         private $fat = [];
  44         private $fileLength;
  45
  46         const TYPE_UNALLOCATED = 0;
  47         const TYPE_STORAGE = 1;
  48         const TYPE_STREAM = 2;
  49         const TYPE_ROOT = 5;
  50
  51         const ERROR_FILE_OPEN = 1;
  52         const ERROR_SEEK = 2;
  53         const ERROR_READ = 3;
  54         const ERROR_INVALID_SIGNATURE = 4;
  55         const ERROR_READ_PAST_END = 5;
  56         const ERROR_INVALID_FORMAT = 6;
  57
  58         private static $mimesByClsid = [
  59                 // From http://justsolve.archiveteam.org/wiki/Microsoft_Compound_File
  60                 '00020810-0000-0000-C000-000000000046' => 'application/vnd.ms-excel',
  61                 '00020820-0000-0000-C000-000000000046' => 'application/vnd.ms-excel',
  62                 '00020906-0000-0000-C000-000000000046' => 'application/msword',
  63                 '64818D10-4F9B-11CF-86EA-00AA00B929E8' => 'application/vnd.ms-powerpoint',
  64         ];
  65
  66         /**
  67          * Read a file by name
  68          *
  69          * @param string $fileName The full path to the file
  70          * @return array An associative array of information about the file:
  71          *    - valid: true if the file is valid, false otherwise
  72          *    - error: An error message in English, should be present if valid=false
  73          *    - errorCode: One of the self::ERROR_* constants
  74          *    - mime: The MIME type detected from the directory contents
  75          *    - mimeFromClsid: The MIME type detected from the CLSID on the root
  76          *      directory entry
  77          */
  78         public static function readFile( $fileName ) {
  79                 $handle = fopen( $fileName, 'r' );
  80                 if ( $handle === false ) {
  81                         return [
  82                                 'valid' => false,
  83                                 'error' => 'file does not exist',
  84                                 'errorCode' => self::ERROR_FILE_OPEN
  85                         ];
  86                 }
  87                 return self::readHandle( $handle );
  88         }
  89
  90         /**
  91          * Read from an open seekable handle
  92          *
  93          * @param resource $fileHandle The file handle
  94          * @return array An associative array of information about the file:
  95          *    - valid: true if the file is valid, false otherwise
  96          *    - error: An error message in English, should be present if valid=false
  97          *    - errorCode: One of the self::ERROR_* constants
  98          *    - mime: The MIME type detected from the directory contents
  99          *    - mimeFromClsid: The MIME type detected from the CLSID on the root
 100          *      directory entry
 101          */
 102         public static function readHandle( $fileHandle ) {
 103                 $reader = new self( $fileHandle );
 104                 $info = [
 105                         'valid' => $reader->valid,
 106                         'mime' => $reader->mime,
 107                         'mimeFromClsid' => $reader->mimeFromClsid
 108                 ];
 109                 if ( $reader->error ) {
 110                         $info['error'] = $reader->error;
 111                         $info['errorCode'] = $reader->errorCode;
 112                 }
 113                 return $info;
 114         }
 115
 116         private function __construct( $fileHandle ) {
 117                 $this->file = $fileHandle;
 118                 try {
 119                         $this->init();
 120                 } catch ( RuntimeException $e ) {
 121                         $this->valid = false;
 122                         $this->error = $e->getMessage();
 123                         $this->errorCode = $e->getCode();
 124                 }
 125         }
 126
 127         private function init() {
 128                 $this->header = $this->unpackOffset( 0, [
 129                         'header_signature' => 8,
 130                         'header_clsid' => 16,
 131                         'minor_version' => 2,
 132                         'major_version' => 2,
 133                         'byte_order' => 2,
 134                         'sector_shift' => 2,
 135                         'mini_sector_shift' => 2,
 136                         'reserved' => 6,
 137                         'num_dir_sectors' => 4,
 138                         'num_fat_sectors' => 4,
 139                         'first_dir_sector' => 4,
 140                         'transaction_signature_number' => 4,
 141                         'mini_stream_cutoff_size' => 4,
 142                         'first_mini_fat_sector' => 4,
 143                         'num_mini_fat_sectors' => 4,
 144                         'first_difat_sector' => 4,
 145                         'num_difat_sectors' => 4,
 146                         'difat' => 436,
 147                 ] );
 148                 if ( $this->header['header_signature'] !== "\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" ) {
 149                         $this->error( 'invalid signature: ' . bin2hex( $this->header['header_signature'] ),
 150                                 self::ERROR_INVALID_SIGNATURE );
 151                 }
 152                 // @phan-suppress-next-line PhanTypeInvalidRightOperandOfIntegerOp
 153                 $this->sectorLength = 1 << $this->header['sector_shift'];
 154                 $this->readDifat();
 155                 $this->readDirectory();
 156
 157                 $this->valid = true;
 158         }
 159
 160         private function sectorOffset( $sectorId ) {
 161                 return $this->sectorLength * ( $sectorId + 1 );
 162         }
 163
 164         private function decodeClsid( $binaryClsid ) {
 165                 $parts = unpack( 'Va/vb/vc/C8d', $binaryClsid );
 166                 return sprintf( "%08X-%04X-%04X-%02X%02X-%02X%02X%02X%02X%02X%02X",
 167                         $parts['a'],
 168                         $parts['b'],
 169                         $parts['c'],
 170                         $parts['d1'],
 171                         $parts['d2'],
 172                         $parts['d3'],
 173                         $parts['d4'],
 174                         $parts['d5'],
 175                         $parts['d6'],
 176                         $parts['d7'],
 177                         $parts['d8']
 178                 );
 179         }
 180
 181         private function unpackOffset( $offset, $struct ) {
 182                 $block = $this->readOffset( $offset, array_sum( $struct ) );
 183                 return $this->unpack( $block, 0, $struct );
 184         }
 185
 186         private function unpack( $block, $offset, $struct ) {
 187                 $data = [];
 188                 foreach ( $struct as $key => $length ) {
 189                         if ( $length > 4 ) {
 190                                 $data[$key] = substr( $block, $offset, $length );
 191                         } else {
 192                                 $data[$key] = $this->bin2dec( $block, $offset, $length );
 193                         }
 194                         $offset += $length;
 195                 }
 196                 return $data;
 197         }
 198
 199         private function bin2dec( $str, $offset, $length ) {
 200                 $value = 0;
 201                 for ( $i = $length - 1; $i >= 0; $i-- ) {
 202                         $value *= 256;
 203                         $value += ord( $str[$offset + $i] );
 204                 }
 205                 return $value;
 206         }
 207
 208         private function readOffset( $offset, $length ) {
 209                 $this->fseek( $offset );
 210                 Wikimedia\suppressWarnings();
 211                 $block = fread( $this->file, $length );
 212                 Wikimedia\restoreWarnings();
 213                 if ( $block === false ) {
 214                         $this->error( 'error reading from file', self::ERROR_READ );
 215                 }
 216                 if ( strlen( $block ) !== $length ) {
 217                         $this->error( 'unable to read the required number of bytes from the file',
 218                                 self::ERROR_READ_PAST_END );
 219                 }
 220                 return $block;
 221         }
 222
 223         private function readSector( $sectorId ) {
 224                 // @phan-suppress-next-line PhanTypeInvalidRightOperandOfIntegerOp
 225                 return $this->readOffset( $this->sectorOffset( $sectorId ), 1 << $this->header['sector_shift'] );
 226         }
 227
 228         private function error( $message, $code ) {
 229                 throw new RuntimeException( $message, $code );
 230         }
 231
 232         private function fseek( $offset ) {
 233                 Wikimedia\suppressWarnings();
 234                 $result = fseek( $this->file, $offset );
 235                 Wikimedia\restoreWarnings();
 236                 if ( $result !== 0 ) {
 237                         $this->error( "unable to seek to offset $offset", self::ERROR_SEEK );
 238                 }
 239         }
 240
 241         private function readDifat() {
 242                 $binaryDifat = $this->header['difat'];
 243                 $nextDifatSector = $this->header['first_difat_sector'];
 244                 for ( $i = 0; $i < $this->header['num_difat_sectors']; $i++ ) {
 245                         $block = $this->readSector( $nextDifatSector );
 246                         $binaryDifat .= substr( $block, 0, $this->sectorLength - 4 );
 247                         $nextDifatSector = $this->bin2dec( $block, $this->sectorLength - 4, 4 );
 248                         if ( $nextDifatSector == 0xFFFFFFFE ) {
 249                                 break;
 250                         }
 251                 }
 252
 253                 $this->difat = [];
 254                 for ( $pos = 0; $pos < strlen( $binaryDifat ); $pos += 4 ) {
 255                         $fatSector = $this->bin2dec( $binaryDifat, $pos, 4 );
 256                         if ( $fatSector < 0xFFFFFFFC ) {
 257                                 $this->difat[] = $fatSector;
 258                         } else {
 259                                 break;
 260                         }
 261                 }
 262         }
 263
 264         private function getNextSectorIdFromFat( $sectorId ) {
 265                 $entriesPerSector = intdiv( $this->sectorLength, 4 );
 266                 $fatSectorId = intdiv( $sectorId, $entriesPerSector );
 267                 $fatSectorArray = $this->getFatSector( $fatSectorId );
 268                 return $fatSectorArray[$sectorId % $entriesPerSector];
 269         }
 270
 271         private function getFatSector( $fatSectorId ) {
 272                 if ( !isset( $this->fat[$fatSectorId] ) ) {
 273                         $fat = [];
 274                         if ( !isset( $this->difat[$fatSectorId] ) ) {
 275                                 $this->error( 'FAT sector requested beyond the end of the DIFAT', self::ERROR_INVALID_FORMAT );
 276                         }
 277                         $absoluteSectorId = $this->difat[$fatSectorId];
 278                         $block = $this->readSector( $absoluteSectorId );
 279                         for ( $pos = 0; $pos < strlen( $block ); $pos += 4 ) {
 280                                 $fat[] = $this->bin2dec( $block, $pos, 4 );
 281                         }
 282                         $this->fat[$fatSectorId] = $fat;
 283                 }
 284                 return $this->fat[$fatSectorId];
 285         }
 286
 287         private function readDirectory() {
 288                 $dirSectorId = $this->header['first_dir_sector'];
 289                 $binaryDir = '';
 290                 $seenSectorIds = [];
 291                 while ( $dirSectorId !== 0xFFFFFFFE ) {
 292                         if ( isset( $seenSectorIds[$dirSectorId] ) ) {
 293                                 $this->error( 'FAT loop detected', self::ERROR_INVALID_FORMAT );
 294                         }
 295                         $seenSectorIds[$dirSectorId] = true;
 296
 297                         $binaryDir .= $this->readSector( $dirSectorId );
 298                         $dirSectorId = $this->getNextSectorIdFromFat( $dirSectorId );
 299                 }
 300
 301                 $struct = [
 302                         'name_raw' => 64,
 303                         'name_length' => 2,
 304                         'object_type' => 1,
 305                         'color' => 1,
 306                         'sid_left' => 4,
 307                         'sid_right' => 4,
 308                         'sid_child' => 4,
 309                         'clsid' => 16,
 310                         'state_bits' => 4,
 311                         'create_time_low' => 4,
 312                         'create_time_high' => 4,
 313                         'modify_time_low' => 4,
 314                         'modify_time_high' => 4,
 315                         'first_sector' => 4,
 316                         'size_low' => 4,
 317                         'size_high' => 4,
 318                 ];
 319                 $entryLength = array_sum( $struct );
 320
 321                 for ( $pos = 0; $pos < strlen( $binaryDir ); $pos += $entryLength ) {
 322                         $entry = $this->unpack( $binaryDir, $pos, $struct );
 323
 324                         // According to [MS-CFB] size_high may contain garbage due to a
 325                         // bug in a writer, it's best to pretend it is zero
 326                         $entry['size_high'] = 0;
 327
 328                         $type = $entry['object_type'];
 329                         if ( $type == self::TYPE_UNALLOCATED ) {
 330                                 continue;
 331                         }
 332
 333                         $name = iconv( 'UTF-16LE', 'UTF-8', substr( $entry['name_raw'], 0, $entry['name_length'] - 2 ) );
 334
 335                         $clsid = $this->decodeClsid( $entry['clsid'] );
 336                         if ( $type == self::TYPE_ROOT && isset( self::$mimesByClsid[$clsid] ) ) {
 337                                 $this->mimeFromClsid = self::$mimesByClsid[$clsid];
 338                         }
 339
 340                         if ( $name === 'Workbook' ) {
 341                                 $this->mime = 'application/vnd.ms-excel';
 342                         } elseif ( $name === 'WordDocument' ) {
 343                                 $this->mime = 'application/msword';
 344                         } elseif ( $name === 'PowerPoint Document' ) {
 345                                 $this->mime = 'application/vnd.ms-powerpoint';
 346                         }
 347                 }
 348         }
 349 }