includes/libs/IEContentAnalyzer.php

   1 <?php
   2 /**
   3  * Simulation of Microsoft Internet Explorer's MIME type detection algorithm.
   4  *
   5  * @file
   6  * @todo Define the exact license of this file.
   7  */
   8
   9 /**
  10  * This class simulates Microsoft Internet Explorer's terribly broken and
  11  * insecure MIME type detection algorithm. It can be used to check web uploads
  12  * with an apparently safe type, to see if IE will reinterpret them to produce
  13  * something dangerous.
  14  *
  15  * It is full of bugs and strange design choices should not under any
  16  * circumstances be used to determine a MIME type to present to a user or
  17  * client. (Apple Safari developers, this means you too.)
  18  *
  19  * This class is based on a disassembly of IE 5.0, 6.0 and 7.0. Although I have
  20  * attempted to ensure that this code works in exactly the same way as Internet
  21  * Explorer, it does not share any source code, or creative choices such as
  22  * variable names, thus I (Tim Starling) claim copyright on it.
  23  *
  24  * It may be redistributed without restriction. To aid reuse, this class does
  25  * not depend on any MediaWiki module.
  26  */
  27 class IEContentAnalyzer {
  28         /**
  29          * Relevant data taken from the type table in IE 5
  30          */
  31         protected $baseTypeTable = array(
  32                 'ambiguous' /*1*/ => array(
  33                         'text/plain',
  34                         'application/octet-stream',
  35                         'application/x-netcdf', // [sic]
  36                 ),
  37                 'text' /*3*/ => array(
  38                         'text/richtext', 'image/x-bitmap', 'application/postscript', 'application/base64',
  39                         'application/macbinhex40', 'application/x-cdf', 'text/scriptlet'
  40                 ),
  41                 'binary' /*4*/ => array(
  42                         'application/pdf', 'audio/x-aiff', 'audio/basic', 'audio/wav', 'image/gif',
  43                         'image/pjpeg', 'image/jpeg', 'image/tiff', 'image/x-png', 'image/png', 'image/bmp',
  44                         'image/x-jg', 'image/x-art', 'image/x-emf', 'image/x-wmf', 'video/avi',
  45                         'video/x-msvideo', 'video/mpeg', 'application/x-compressed',
  46                         'application/x-zip-compressed', 'application/x-gzip-compressed', 'application/java',
  47                         'application/x-msdownload'
  48                 ),
  49                 'html' /*5*/ => array( 'text/html' ),
  50         );
  51
  52         /**
  53          * Changes to the type table in later versions of IE
  54          */
  55         protected $addedTypes = array(
  56                 'ie07' => array(
  57                         'text' => array( 'text/xml', 'application/xml' )
  58                 ),
  59         );
  60
  61         /**
  62          * An approximation of the "Content Type" values in HKEY_CLASSES_ROOT in a
  63          * typical Windows installation.
  64          *
  65          * Used for extension to MIME type mapping if detection fails.
  66          */
  67         protected $registry = array(
  68                 '.323' => 'text/h323',
  69                 '.3g2' => 'video/3gpp2',
  70                 '.3gp' => 'video/3gpp',
  71                 '.3gp2' => 'video/3gpp2',
  72                 '.3gpp' => 'video/3gpp',
  73                 '.aac' => 'audio/aac',
  74                 '.ac3' => 'audio/ac3',
  75                 '.accda' => 'application/msaccess',
  76                 '.accdb' => 'application/msaccess',
  77                 '.accdc' => 'application/msaccess',
  78                 '.accde' => 'application/msaccess',
  79                 '.accdr' => 'application/msaccess',
  80                 '.accdt' => 'application/msaccess',
  81                 '.ade' => 'application/msaccess',
  82                 '.adp' => 'application/msaccess',
  83                 '.adts' => 'audio/aac',
  84                 '.ai' => 'application/postscript',
  85                 '.aif' => 'audio/aiff',
  86                 '.aifc' => 'audio/aiff',
  87                 '.aiff' => 'audio/aiff',
  88                 '.amc' => 'application/x-mpeg',
  89                 '.application' => 'application/x-ms-application',
  90                 '.asf' => 'video/x-ms-asf',
  91                 '.asx' => 'video/x-ms-asf',
  92                 '.au' => 'audio/basic',
  93                 '.avi' => 'video/avi',
  94                 '.bmp' => 'image/bmp',
  95                 '.caf' => 'audio/x-caf',
  96                 '.cat' => 'application/vnd.ms-pki.seccat',
  97                 '.cbo' => 'application/sha',
  98                 '.cdda' => 'audio/aiff',
  99                 '.cer' => 'application/x-x509-ca-cert',
 100                 '.conf' => 'text/plain',
 101                 '.crl' => 'application/pkix-crl',
 102                 '.crt' => 'application/x-x509-ca-cert',
 103                 '.css' => 'text/css',
 104                 '.csv' => 'application/vnd.ms-excel',
 105                 '.der' => 'application/x-x509-ca-cert',
 106                 '.dib' => 'image/bmp',
 107                 '.dif' => 'video/x-dv',
 108                 '.dll' => 'application/x-msdownload',
 109                 '.doc' => 'application/msword',
 110                 '.docm' => 'application/vnd.ms-word.document.macroEnabled.12',
 111                 '.docx' => 'application/vnd.openxmlformats-officedocument.wordprocessingml.document',
 112                 '.dot' => 'application/msword',
 113                 '.dotm' => 'application/vnd.ms-word.template.macroEnabled.12',
 114                 '.dotx' => 'application/vnd.openxmlformats-officedocument.wordprocessingml.template',
 115                 '.dv' => 'video/x-dv',
 116                 '.dwfx' => 'model/vnd.dwfx+xps',
 117                 '.edn' => 'application/vnd.adobe.edn',
 118                 '.eml' => 'message/rfc822',
 119                 '.eps' => 'application/postscript',
 120                 '.etd' => 'application/x-ebx',
 121                 '.exe' => 'application/x-msdownload',
 122                 '.fdf' => 'application/vnd.fdf',
 123                 '.fif' => 'application/fractals',
 124                 '.gif' => 'image/gif',
 125                 '.gsm' => 'audio/x-gsm',
 126                 '.hqx' => 'application/mac-binhex40',
 127                 '.hta' => 'application/hta',
 128                 '.htc' => 'text/x-component',
 129                 '.htm' => 'text/html',
 130                 '.html' => 'text/html',
 131                 '.htt' => 'text/webviewhtml',
 132                 '.hxa' => 'application/xml',
 133                 '.hxc' => 'application/xml',
 134                 '.hxd' => 'application/octet-stream',
 135                 '.hxe' => 'application/xml',
 136                 '.hxf' => 'application/xml',
 137                 '.hxh' => 'application/octet-stream',
 138                 '.hxi' => 'application/octet-stream',
 139                 '.hxk' => 'application/xml',
 140                 '.hxq' => 'application/octet-stream',
 141                 '.hxr' => 'application/octet-stream',
 142                 '.hxs' => 'application/octet-stream',
 143                 '.hxt' => 'application/xml',
 144                 '.hxv' => 'application/xml',
 145                 '.hxw' => 'application/octet-stream',
 146                 '.ico' => 'image/x-icon',
 147                 '.iii' => 'application/x-iphone',
 148                 '.ins' => 'application/x-internet-signup',
 149                 '.iqy' => 'text/x-ms-iqy',
 150                 '.isp' => 'application/x-internet-signup',
 151                 '.jfif' => 'image/jpeg',
 152                 '.jnlp' => 'application/x-java-jnlp-file',
 153                 '.jpe' => 'image/jpeg',
 154                 '.jpeg' => 'image/jpeg',
 155                 '.jpg' => 'image/jpeg',
 156                 '.jtx' => 'application/x-jtx+xps',
 157                 '.latex' => 'application/x-latex',
 158                 '.log' => 'text/plain',
 159                 '.m1v' => 'video/mpeg',
 160                 '.m2v' => 'video/mpeg',
 161                 '.m3u' => 'audio/x-mpegurl',
 162                 '.mac' => 'image/x-macpaint',
 163                 '.man' => 'application/x-troff-man',
 164                 '.mda' => 'application/msaccess',
 165                 '.mdb' => 'application/msaccess',
 166                 '.mde' => 'application/msaccess',
 167                 '.mfp' => 'application/x-shockwave-flash',
 168                 '.mht' => 'message/rfc822',
 169                 '.mhtml' => 'message/rfc822',
 170                 '.mid' => 'audio/mid',
 171                 '.midi' => 'audio/mid',
 172                 '.mod' => 'video/mpeg',
 173                 '.mov' => 'video/quicktime',
 174                 '.mp2' => 'video/mpeg',
 175                 '.mp2v' => 'video/mpeg',
 176                 '.mp3' => 'audio/mpeg',
 177                 '.mp4' => 'video/mp4',
 178                 '.mpa' => 'video/mpeg',
 179                 '.mpe' => 'video/mpeg',
 180                 '.mpeg' => 'video/mpeg',
 181                 '.mpf' => 'application/vnd.ms-mediapackage',
 182                 '.mpg' => 'video/mpeg',
 183                 '.mpv2' => 'video/mpeg',
 184                 '.mqv' => 'video/quicktime',
 185                 '.NMW' => 'application/nmwb',
 186                 '.nws' => 'message/rfc822',
 187                 '.odc' => 'text/x-ms-odc',
 188                 '.ols' => 'application/vnd.ms-publisher',
 189                 '.p10' => 'application/pkcs10',
 190                 '.p12' => 'application/x-pkcs12',
 191                 '.p7b' => 'application/x-pkcs7-certificates',
 192                 '.p7c' => 'application/pkcs7-mime',
 193                 '.p7m' => 'application/pkcs7-mime',
 194                 '.p7r' => 'application/x-pkcs7-certreqresp',
 195                 '.p7s' => 'application/pkcs7-signature',
 196                 '.pct' => 'image/pict',
 197                 '.pdf' => 'application/pdf',
 198                 '.pdx' => 'application/vnd.adobe.pdx',
 199                 '.pfx' => 'application/x-pkcs12',
 200                 '.pic' => 'image/pict',
 201                 '.pict' => 'image/pict',
 202                 '.pinstall' => 'application/x-picasa-detect',
 203                 '.pko' => 'application/vnd.ms-pki.pko',
 204                 '.png' => 'image/png',
 205                 '.pnt' => 'image/x-macpaint',
 206                 '.pntg' => 'image/x-macpaint',
 207                 '.pot' => 'application/vnd.ms-powerpoint',
 208                 '.potm' => 'application/vnd.ms-powerpoint.template.macroEnabled.12',
 209                 '.potx' => 'application/vnd.openxmlformats-officedocument.presentationml.template',
 210                 '.ppa' => 'application/vnd.ms-powerpoint',
 211                 '.ppam' => 'application/vnd.ms-powerpoint.addin.macroEnabled.12',
 212                 '.pps' => 'application/vnd.ms-powerpoint',
 213                 '.ppsm' => 'application/vnd.ms-powerpoint.slideshow.macroEnabled.12',
 214                 '.ppsx' => 'application/vnd.openxmlformats-officedocument.presentationml.slideshow',
 215                 '.ppt' => 'application/vnd.ms-powerpoint',
 216                 '.pptm' => 'application/vnd.ms-powerpoint.presentation.macroEnabled.12',
 217                 '.pptx' => 'application/vnd.openxmlformats-officedocument.presentationml.presentation',
 218                 '.prf' => 'application/pics-rules',
 219                 '.ps' => 'application/postscript',
 220                 '.pub' => 'application/vnd.ms-publisher',
 221                 '.pwz' => 'application/vnd.ms-powerpoint',
 222                 '.py' => 'text/plain',
 223                 '.pyw' => 'text/plain',
 224                 '.qht' => 'text/x-html-insertion',
 225                 '.qhtm' => 'text/x-html-insertion',
 226                 '.qt' => 'video/quicktime',
 227                 '.qti' => 'image/x-quicktime',
 228                 '.qtif' => 'image/x-quicktime',
 229                 '.qtl' => 'application/x-quicktimeplayer',
 230                 '.rat' => 'application/rat-file',
 231                 '.rmf' => 'application/vnd.adobe.rmf',
 232                 '.rmi' => 'audio/mid',
 233                 '.rqy' => 'text/x-ms-rqy',
 234                 '.rtf' => 'application/msword',
 235                 '.sct' => 'text/scriptlet',
 236                 '.sd2' => 'audio/x-sd2',
 237                 '.sdp' => 'application/sdp',
 238                 '.shtml' => 'text/html',
 239                 '.sit' => 'application/x-stuffit',
 240                 '.sldm' => 'application/vnd.ms-powerpoint.slide.macroEnabled.12',
 241                 '.sldx' => 'application/vnd.openxmlformats-officedocument.presentationml.slide',
 242                 '.slk' => 'application/vnd.ms-excel',
 243                 '.snd' => 'audio/basic',
 244                 '.so' => 'application/x-apachemodule',
 245                 '.sol' => 'text/plain',
 246                 '.sor' => 'text/plain',
 247                 '.spc' => 'application/x-pkcs7-certificates',
 248                 '.spl' => 'application/futuresplash',
 249                 '.sst' => 'application/vnd.ms-pki.certstore',
 250                 '.stl' => 'application/vnd.ms-pki.stl',
 251                 '.swf' => 'application/x-shockwave-flash',
 252                 '.thmx' => 'application/vnd.ms-officetheme',
 253                 '.tif' => 'image/tiff',
 254                 '.tiff' => 'image/tiff',
 255                 '.txt' => 'text/plain',
 256                 '.uls' => 'text/iuls',
 257                 '.vcf' => 'text/x-vcard',
 258                 '.vdx' => 'application/vnd.ms-visio.viewer',
 259                 '.vsd' => 'application/vnd.ms-visio.viewer',
 260                 '.vss' => 'application/vnd.ms-visio.viewer',
 261                 '.vst' => 'application/vnd.ms-visio.viewer',
 262                 '.vsx' => 'application/vnd.ms-visio.viewer',
 263                 '.vtx' => 'application/vnd.ms-visio.viewer',
 264                 '.wav' => 'audio/wav',
 265                 '.wax' => 'audio/x-ms-wax',
 266                 '.wbk' => 'application/msword',
 267                 '.wdp' => 'image/vnd.ms-photo',
 268                 '.wiz' => 'application/msword',
 269                 '.wm' => 'video/x-ms-wm',
 270                 '.wma' => 'audio/x-ms-wma',
 271                 '.wmd' => 'application/x-ms-wmd',
 272                 '.wmv' => 'video/x-ms-wmv',
 273                 '.wmx' => 'video/x-ms-wmx',
 274                 '.wmz' => 'application/x-ms-wmz',
 275                 '.wpl' => 'application/vnd.ms-wpl',
 276                 '.wsc' => 'text/scriptlet',
 277                 '.wvx' => 'video/x-ms-wvx',
 278                 '.xaml' => 'application/xaml+xml',
 279                 '.xbap' => 'application/x-ms-xbap',
 280                 '.xdp' => 'application/vnd.adobe.xdp+xml',
 281                 '.xfdf' => 'application/vnd.adobe.xfdf',
 282                 '.xht' => 'application/xhtml+xml',
 283                 '.xhtml' => 'application/xhtml+xml',
 284                 '.xla' => 'application/vnd.ms-excel',
 285                 '.xlam' => 'application/vnd.ms-excel.addin.macroEnabled.12',
 286                 '.xlk' => 'application/vnd.ms-excel',
 287                 '.xll' => 'application/vnd.ms-excel',
 288                 '.xlm' => 'application/vnd.ms-excel',
 289                 '.xls' => 'application/vnd.ms-excel',
 290                 '.xlsb' => 'application/vnd.ms-excel.sheet.binary.macroEnabled.12',
 291                 '.xlsm' => 'application/vnd.ms-excel.sheet.macroEnabled.12',
 292                 '.xlsx' => 'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet',
 293                 '.xlt' => 'application/vnd.ms-excel',
 294                 '.xltm' => 'application/vnd.ms-excel.template.macroEnabled.12',
 295                 '.xltx' => 'application/vnd.openxmlformats-officedocument.spreadsheetml.template',
 296                 '.xlw' => 'application/vnd.ms-excel',
 297                 '.xml' => 'text/xml',
 298                 '.xps' => 'application/vnd.ms-xpsdocument',
 299                 '.xsl' => 'text/xml',
 300         );
 301
 302         /**
 303          * IE versions which have been analysed to bring you this class, and for
 304          * which some substantive difference exists. These will appear as keys
 305          * in the return value of getRealMimesFromData(). The names are chosen to sort correctly.
 306          */
 307         protected $versions = array( 'ie05', 'ie06', 'ie07', 'ie07.strict', 'ie07.nohtml' );
 308
 309         /**
 310          * Type table with versions expanded
 311          */
 312         protected $typeTable = array();
 313
 314         /** constructor */
 315         function __construct() {
 316                 // Construct versioned type arrays from the base type array plus additions
 317                 $types = $this->baseTypeTable;
 318                 foreach ( $this->versions as $version ) {
 319                         if ( isset( $this->addedTypes[$version] ) ) {
 320                                 foreach ( $this->addedTypes[$version] as $format => $addedTypes ) {
 321                                         $types[$format] = array_merge( $types[$format], $addedTypes );
 322                                 }
 323                         }
 324                         $this->typeTable[$version] = $types;
 325                 }
 326         }
 327
 328         /**
 329          * Get the MIME types from getMimesFromData(), but convert the result from IE's
 330          * idiosyncratic private types into something other apps will understand.
 331          *
 332          * @param string $fileName the file name (unused at present)
 333          * @param string $chunk the first 256 bytes of the file
 334          * @param string $proposed the MIME type proposed by the server
 335          *
 336          * @return Array: map of IE version to detected mime type
 337          */
 338         public function getRealMimesFromData( $fileName, $chunk, $proposed ) {
 339                 $types = $this->getMimesFromData( $fileName, $chunk, $proposed );
 340                 $types = array_map( array( $this, 'translateMimeType' ), $types );
 341                 return $types;
 342         }
 343
 344         /**
 345          * Translate a MIME type from IE's idiosyncratic private types into
 346          * more commonly understood type strings
 347          * @param $type
 348          * @return string
 349          */
 350         public function translateMimeType( $type ) {
 351                 static $table = array(
 352                         'image/pjpeg' => 'image/jpeg',
 353                         'image/x-png' => 'image/png',
 354                         'image/x-wmf' => 'application/x-msmetafile',
 355                         'image/bmp' => 'image/x-bmp',
 356                         'application/x-zip-compressed' => 'application/zip',
 357                         'application/x-compressed' => 'application/x-compress',
 358                         'application/x-gzip-compressed' => 'application/x-gzip',
 359                         'audio/mid' => 'audio/midi',
 360                 );
 361                 if ( isset( $table[$type] ) ) {
 362                         $type = $table[$type];
 363                 }
 364                 return $type;
 365         }
 366
 367         /**
 368          * Get the untranslated MIME types for all known versions
 369          *
 370          * @param string $fileName the file name (unused at present)
 371          * @param string $chunk the first 256 bytes of the file
 372          * @param string $proposed the MIME type proposed by the server
 373          *
 374          * @return Array: map of IE version to detected mime type
 375          */
 376         public function getMimesFromData( $fileName, $chunk, $proposed ) {
 377                 $types = array();
 378                 foreach ( $this->versions as $version ) {
 379                         $types[$version] = $this->getMimeTypeForVersion( $version, $fileName, $chunk, $proposed );
 380                 }
 381                 return $types;
 382         }
 383
 384         /**
 385          * Get the MIME type for a given named version
 386          * @param $version
 387          * @param $fileName
 388          * @param $chunk
 389          * @param $proposed
 390          * @return bool|string
 391          */
 392         protected function getMimeTypeForVersion( $version, $fileName, $chunk, $proposed ) {
 393                 // Strip text after a semicolon
 394                 $semiPos = strpos( $proposed, ';' );
 395                 if ( $semiPos !== false ) {
 396                         $proposed = substr( $proposed, 0, $semiPos );
 397                 }
 398
 399                 $proposedFormat = $this->getDataFormat( $version, $proposed );
 400                 if ( $proposedFormat == 'unknown'
 401                         && $proposed != 'multipart/mixed'
 402                         && $proposed != 'multipart/x-mixed-replace' )
 403                 {
 404                         return $proposed;
 405                 }
 406                 if ( strval( $chunk ) === '' ) {
 407                         return $proposed;
 408                 }
 409
 410                 // Truncate chunk at 255 bytes
 411                 $chunk = substr( $chunk, 0, 255 );
 412
 413                 // IE does the Check*Headers() calls last, and instead does the following image
 414                 // type checks by directly looking for the magic numbers. What I do here should
 415                 // have the same effect since the magic number checks are identical in both cases.
 416                 $result = $this->sampleData( $version, $chunk );
 417                 $sampleFound = $result['found'];
 418                 $counters = $result['counters'];
 419                 $binaryType = $this->checkBinaryHeaders( $version, $chunk );
 420                 $textType = $this->checkTextHeaders( $version, $chunk );
 421
 422                 if ( $proposed == 'text/html' && isset( $sampleFound['html'] ) ) {
 423                         return 'text/html';
 424                 }
 425                 if ( $proposed == 'image/gif' && $binaryType == 'image/gif' ) {
 426                         return 'image/gif';
 427                 }
 428                 if ( ( $proposed == 'image/pjpeg' || $proposed == 'image/jpeg' )
 429                         && $binaryType == 'image/pjpeg' )
 430                 {
 431                         return $proposed;
 432                 }
 433                 // PNG check added in IE 7
 434                 if ( $version >= 'ie07'
 435                         && ( $proposed == 'image/x-png' || $proposed == 'image/png' )
 436                         && $binaryType == 'image/x-png' )
 437                 {
 438                         return $proposed;
 439                 }
 440
 441                 // CDF was removed in IE 7 so it won't be in $sampleFound for later versions
 442                 if ( isset( $sampleFound['cdf'] ) ) {
 443                         return 'application/x-cdf';
 444                 }
 445
 446                 // RSS and Atom were added in IE 7 so they won't be in $sampleFound for
 447                 // previous versions
 448                 if ( isset( $sampleFound['rss'] ) ) {
 449                         return 'application/rss+xml';
 450                 }
 451                 if ( isset( $sampleFound['rdf-tag'] )
 452                         && isset( $sampleFound['rdf-url'] )
 453                         && isset( $sampleFound['rdf-purl'] ) )
 454                 {
 455                         return 'application/rss+xml';
 456                 }
 457                 if ( isset( $sampleFound['atom'] ) ) {
 458                         return 'application/atom+xml';
 459                 }
 460
 461                 if ( isset( $sampleFound['xml'] ) ) {
 462                         // TODO: I'm not sure under what circumstances this flag is enabled
 463                         if ( strpos( $version, 'strict' ) !== false ) {
 464                                 if ( $proposed == 'text/html' || $proposed == 'text/xml' ) {
 465                                         return 'text/xml';
 466                                 }
 467                         } else {
 468                                 return 'text/xml';
 469                         }
 470                 }
 471                 if ( isset( $sampleFound['html'] ) ) {
 472                         // TODO: I'm not sure under what circumstances this flag is enabled
 473                         if ( strpos( $version, 'nohtml' ) !== false ) {
 474                                 if ( $proposed == 'text/plain' ) {
 475                                         return 'text/html';
 476                                 }
 477                         } else {
 478                                 return 'text/html';
 479                         }
 480                 }
 481                 if ( isset( $sampleFound['xbm'] ) ) {
 482                         return 'image/x-bitmap';
 483                 }
 484                 if ( isset( $sampleFound['binhex'] ) ) {
 485                         return 'application/macbinhex40';
 486                 }
 487                 if ( isset( $sampleFound['scriptlet'] ) ) {
 488                         if ( strpos( $version, 'strict' ) !== false ) {
 489                                 if ( $proposed == 'text/plain' || $proposed == 'text/scriptlet' ) {
 490                                         return 'text/scriptlet';
 491                                 }
 492                         } else {
 493                                 return 'text/scriptlet';
 494                         }
 495                 }
 496
 497                 // Freaky heuristics to determine if the data is text or binary
 498                 // The heuristic is of course broken for non-ASCII text
 499                 if ( $counters['ctrl'] != 0 && ( $counters['ff'] + $counters['low'] )
 500                         < ( $counters['ctrl'] + $counters['high'] ) * 16 )
 501                 {
 502                         $kindOfBinary = true;
 503                         $type = $binaryType ? $binaryType : $textType;
 504                         if ( $type === false ) {
 505                                 $type = 'application/octet-stream';
 506                         }
 507                 } else {
 508                         $kindOfBinary = false;
 509                         $type = $textType ? $textType : $binaryType;
 510                         if ( $type === false ) {
 511                                 $type = 'text/plain';
 512                         }
 513                 }
 514
 515                 // Check if the output format is ambiguous
 516                 // This generally means that detection failed, real types aren't ambiguous
 517                 $detectedFormat = $this->getDataFormat( $version, $type );
 518                 if ( $detectedFormat != 'ambiguous' ) {
 519                         return $type;
 520                 }
 521
 522                 if ( $proposedFormat != 'ambiguous' ) {
 523                         // FormatAgreesWithData()
 524                         if ( $proposedFormat == 'text' && !$kindOfBinary ) {
 525                                 return $proposed;
 526                         }
 527                         if ( $proposedFormat == 'binary' && $kindOfBinary ) {
 528                                 return $proposed;
 529                         }
 530                         if ( $proposedFormat == 'html' ) {
 531                                 return $proposed;
 532                         }
 533                 }
 534
 535                 // Find a MIME type by searching the registry for the file extension.
 536                 $dotPos = strrpos( $fileName, '.' );
 537                 if ( $dotPos === false ) {
 538                         return $type;
 539                 }
 540                 $ext = substr( $fileName, $dotPos );
 541                 if ( isset( $this->registry[$ext] ) ) {
 542                         return $this->registry[$ext];
 543                 }
 544
 545                 // TODO: If the extension has an application registered to it, IE will return
 546                 // application/octet-stream. We'll skip that, so we could erroneously
 547                 // return text/plain or application/x-netcdf where application/octet-stream
 548                 // would be correct.
 549
 550                 return $type;
 551         }
 552
 553         /**
 554          * Check for text headers at the start of the chunk
 555          * Confirmed same in 5 and 7.
 556          * @param $version
 557          * @param $chunk
 558          * @return bool|string
 559          */
 560         private function checkTextHeaders( $version, $chunk ) {
 561                 $chunk2 = substr( $chunk, 0, 2 );
 562                 $chunk4 = substr( $chunk, 0, 4 );
 563                 $chunk5 = substr( $chunk, 0, 5 );
 564                 if ( $chunk4 == '%PDF' ) {
 565                         return 'application/pdf';
 566                 }
 567                 if ( $chunk2 == '%!' ) {
 568                         return 'application/postscript';
 569                 }
 570                 if ( $chunk5 == '{\\rtf' ) {
 571                         return 'text/richtext';
 572                 }
 573                 if ( $chunk5 == 'begin' ) {
 574                         return 'application/base64';
 575                 }
 576                 return false;
 577         }
 578
 579         /**
 580          * Check for binary headers at the start of the chunk
 581          * Confirmed same in 5 and 7.
 582          * @param $version
 583          * @param $chunk
 584          * @return bool|string
 585          */
 586         private function checkBinaryHeaders( $version, $chunk ) {
 587                 $chunk2 = substr( $chunk, 0, 2 );
 588                 $chunk3 = substr( $chunk, 0, 3 );
 589                 $chunk4 = substr( $chunk, 0, 4 );
 590                 $chunk5 = substr( $chunk, 0, 5 );
 591                 $chunk5uc = strtoupper( $chunk5 );
 592                 $chunk8 = substr( $chunk, 0, 8 );
 593                 if ( $chunk5uc == 'GIF87' || $chunk5uc == 'GIF89' ) {
 594                         return 'image/gif';
 595                 }
 596                 if ( $chunk2 == "\xff\xd8" ) {
 597                         return 'image/pjpeg'; // actually plain JPEG but this is what IE returns
 598                 }
 599
 600                 if ( $chunk2 == 'BM'
 601                         && substr( $chunk, 6, 2 ) == "\000\000"
 602                         && substr( $chunk, 8, 2 ) == "\000\000" )
 603                 {
 604                         return 'image/bmp'; // another non-standard MIME
 605                 }
 606                 if ( $chunk4 == 'RIFF'
 607                         && substr( $chunk, 8, 4 ) == 'WAVE' )
 608                 {
 609                         return 'audio/wav';
 610                 }
 611                 // These were integer literals in IE
 612                 // Perhaps the author was not sure what the target endianness was
 613                 if ( $chunk4 == ".sd\000"
 614                         || $chunk4 == ".snd"
 615                         || $chunk4 == "\000ds."
 616                         || $chunk4 == "dns." )
 617                 {
 618                         return 'audio/basic';
 619                 }
 620                 if ( $chunk3 == "MM\000" ) {
 621                         return 'image/tiff';
 622                 }
 623                 if ( $chunk2 == 'MZ' ) {
 624                         return 'application/x-msdownload';
 625                 }
 626                 if ( $chunk8 == "\x89PNG\x0d\x0a\x1a\x0a" ) {
 627                         return 'image/x-png'; // [sic]
 628                 }
 629                 if ( strlen( $chunk ) >= 5 ) {
 630                         $byte2 = ord( $chunk[2] );
 631                         $byte4 = ord( $chunk[4] );
 632                         if ( $byte2 >= 3 && $byte2 <= 31 && $byte4 == 0 && $chunk2 == 'JG' ) {
 633                                 return 'image/x-jg';
 634                         }
 635                 }
 636                 // More endian confusion?
 637                 if ( $chunk4 == 'MROF' ) {
 638                         return 'audio/x-aiff';
 639                 }
 640                 $chunk4_8 = substr( $chunk, 8, 4 );
 641                 if ( $chunk4 == 'FORM' && ( $chunk4_8 == 'AIFF' || $chunk4_8 == 'AIFC' ) ) {
 642                         return 'audio/x-aiff';
 643                 }
 644                 if ( $chunk4 == 'RIFF' && $chunk4_8 == 'AVI ' ) {
 645                         return 'video/avi';
 646                 }
 647                 if ( $chunk4 == "\x00\x00\x01\xb3" || $chunk4 == "\x00\x00\x01\xba" ) {
 648                         return 'video/mpeg';
 649                 }
 650                 if ( $chunk4 == "\001\000\000\000"
 651                         && substr( $chunk, 40, 4 ) == ' EMF' )
 652                 {
 653                         return 'image/x-emf';
 654                 }
 655                 if ( $chunk4 == "\xd7\xcd\xc6\x9a" ) {
 656                         return 'image/x-wmf';
 657                 }
 658                 if ( $chunk4 == "\xca\xfe\xba\xbe" ) {
 659                         return 'application/java';
 660                 }
 661                 if ( $chunk2 == 'PK' ) {
 662                         return 'application/x-zip-compressed';
 663                 }
 664                 if ( $chunk2 == "\x1f\x9d" ) {
 665                         return 'application/x-compressed';
 666                 }
 667                 if ( $chunk2 == "\x1f\x8b" ) {
 668                         return 'application/x-gzip-compressed';
 669                 }
 670                 // Skip redundant check for ZIP
 671                 if ( $chunk5 == "MThd\000" ) {
 672                         return 'audio/mid';
 673                 }
 674                 if ( $chunk4 == '%PDF' ) {
 675                         return 'application/pdf';
 676                 }
 677                 return false;
 678         }
 679
 680         /**
 681          * Do heuristic checks on the bulk of the data sample.
 682          * Search for HTML tags.
 683          * @param $version
 684          * @param $chunk
 685          * @return array
 686          */
 687         protected function sampleData( $version, $chunk ) {
 688                 $found = array();
 689                 $counters = array(
 690                         'ctrl' => 0,
 691                         'high' => 0,
 692                         'low' => 0,
 693                         'lf' => 0,
 694                         'cr' => 0,
 695                         'ff' => 0
 696                 );
 697                 $htmlTags = array(
 698                         'html',
 699                         'head',
 700                         'title',
 701                         'body',
 702                         'script',
 703                         'a href',
 704                         'pre',
 705                         'img',
 706                         'plaintext',
 707                         'table'
 708                 );
 709                 $rdfUrl = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#';
 710                 $rdfPurl = 'http://purl.org/rss/1.0/';
 711                 $xbmMagic1 = '#define';
 712                 $xbmMagic2 = '_width';
 713                 $xbmMagic3 = '_bits';
 714                 $binhexMagic = 'converted with BinHex';
 715
 716                 for ( $offset = 0; $offset < strlen( $chunk ); $offset++ ) {
 717                         $curChar = $chunk[$offset];
 718                         if ( $curChar == "\x0a" ) {
 719                                 $counters['lf']++;
 720                                 continue;
 721                         } elseif ( $curChar == "\x0d" ) {
 722                                 $counters['cr']++;
 723                                 continue;
 724                         } elseif ( $curChar == "\x0c" ) {
 725                                 $counters['ff']++;
 726                                 continue;
 727                         } elseif ( $curChar == "\t" ) {
 728                                 $counters['low']++;
 729                                 continue;
 730                         } elseif ( ord( $curChar ) < 32 ) {
 731                                 $counters['ctrl']++;
 732                                 continue;
 733                         } elseif ( ord( $curChar ) >= 128 ) {
 734                                 $counters['high']++;
 735                                 continue;
 736                         }
 737
 738                         $counters['low']++;
 739                         if ( $curChar == '<' ) {
 740                                 // XML
 741                                 $remainder = substr( $chunk, $offset + 1 );
 742                                 if ( !strncasecmp( $remainder, '?XML', 4 ) ) {
 743                                         $nextChar = substr( $chunk, $offset + 5, 1 );
 744                                         if ( $nextChar == ':' || $nextChar == ' ' || $nextChar == "\t" ) {
 745                                                 $found['xml'] = true;
 746                                         }
 747                                 }
 748                                 // Scriptlet (JSP)
 749                                 if ( !strncasecmp( $remainder, 'SCRIPTLET', 9 ) ) {
 750                                         $found['scriptlet'] = true;
 751                                         break;
 752                                 }
 753                                 // HTML
 754                                 foreach ( $htmlTags as $tag ) {
 755                                         if ( !strncasecmp( $remainder, $tag, strlen( $tag ) ) ) {
 756                                                 $found['html'] = true;
 757                                         }
 758                                 }
 759                                 // Skip broken check for additional tags (HR etc.)
 760
 761                                 // CHANNEL replaced by RSS, RDF and FEED in IE 7
 762                                 if ( $version < 'ie07' ) {
 763                                         if ( !strncasecmp( $remainder, 'CHANNEL', 7 ) ) {
 764                                                 $found['cdf'] = true;
 765                                         }
 766                                 } else {
 767                                         // RSS
 768                                         if ( !strncasecmp( $remainder, 'RSS', 3 ) ) {
 769                                                 $found['rss'] = true;
 770                                                 break; // return from SampleData
 771                                         }
 772                                         if ( !strncasecmp( $remainder, 'rdf:RDF', 7 ) ) {
 773                                                 $found['rdf-tag'] = true;
 774                                                 // no break
 775                                         }
 776                                         if ( !strncasecmp( $remainder, 'FEED', 4 ) ) {
 777                                                 $found['atom'] = true;
 778                                                 break;
 779                                         }
 780                                 }
 781                                 continue;
 782                         }
 783                         // Skip broken check for -->
 784
 785                         // RSS URL checks
 786                         // For some reason both URLs must appear before it is recognised
 787                         $remainder = substr( $chunk, $offset );
 788                         if ( !strncasecmp( $remainder, $rdfUrl, strlen( $rdfUrl ) ) ) {
 789                                 $found['rdf-url'] = true;
 790                                 if ( isset( $found['rdf-tag'] )
 791                                         && isset( $found['rdf-purl'] ) ) // [sic]
 792                                 {
 793                                         break;
 794                                 }
 795                                 continue;
 796                         }
 797
 798                         if ( !strncasecmp( $remainder, $rdfPurl, strlen( $rdfPurl ) ) ) {
 799                                 if ( isset( $found['rdf-tag'] )
 800                                         && isset( $found['rdf-url'] ) ) // [sic]
 801                                 {
 802                                         break;
 803                                 }
 804                                 continue;
 805                         }
 806
 807                         // XBM checks
 808                         if ( !strncasecmp( $remainder, $xbmMagic1, strlen( $xbmMagic1 ) ) ) {
 809                                 $found['xbm1'] = true;
 810                                 continue;
 811                         }
 812                         if ( $curChar == '_' ) {
 813                                 if ( isset( $found['xbm2'] ) ) {
 814                                         if ( !strncasecmp( $remainder, $xbmMagic3, strlen( $xbmMagic3 ) ) ) {
 815                                                 $found['xbm'] = true;
 816                                                 break;
 817                                         }
 818                                 } elseif ( isset( $found['xbm1'] ) ) {
 819                                         if ( !strncasecmp( $remainder, $xbmMagic2, strlen( $xbmMagic2 ) ) ) {
 820                                                 $found['xbm2'] = true;
 821                                         }
 822                                 }
 823                         }
 824
 825                         // BinHex
 826                         if ( !strncmp( $remainder, $binhexMagic, strlen( $binhexMagic ) ) ) {
 827                                 $found['binhex'] = true;
 828                         }
 829                 }
 830                 return array( 'found' => $found, 'counters' => $counters );
 831         }
 832
 833         /**
 834          * @param $version
 835          * @param $type
 836          * @return int|string
 837          */
 838         protected function getDataFormat( $version, $type ) {
 839                 $types = $this->typeTable[$version];
 840                 if ( $type == '(null)' || strval( $type ) === '' ) {
 841                         return 'ambiguous';
 842                 }
 843                 foreach ( $types as $format => $list ) {
 844                         if ( in_array( $type, $list ) ) {
 845                                 return $format;
 846                         }
 847                 }
 848                 return 'unknown';
 849         }
 850 }