includes/IEContentAnalyzer.php

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