includes/media/SVGMetadataExtractor.php

   1 <?php
   2 /**
   3  * Extraction of SVG image metadata.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Media
  22  * @author "Derk-Jan Hartman <hartman _at_ videolan d0t org>"
  23  * @author Brion Vibber
  24  * @copyright Copyright © 2010-2010 Brion Vibber, Derk-Jan Hartman
  25  * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License
  26  */
  27
  28 /**
  29  * @ingroup Media
  30  */
  31 class SVGMetadataExtractor {
  32         static function getMetadata( $filename ) {
  33                 $svg = new SVGReader( $filename );
  34
  35                 return $svg->getMetadata();
  36         }
  37 }
  38
  39 /**
  40  * @ingroup Media
  41  */
  42 class SVGReader {
  43         const DEFAULT_WIDTH = 512;
  44         const DEFAULT_HEIGHT = 512;
  45         const NS_SVG = 'http://www.w3.org/2000/svg';
  46         const LANG_PREFIX_MATCH = 1;
  47         const LANG_FULL_MATCH = 2;
  48
  49         /** @var null|XMLReader */
  50         private $reader = null;
  51
  52         /** @var bool */
  53         private $mDebug = false;
  54
  55         /** @var array */
  56         private $metadata = array();
  57         private $languages = array();
  58         private $languagePrefixes = array();
  59
  60         /**
  61          * Constructor
  62          *
  63          * Creates an SVGReader drawing from the source provided
  64          * @param string $source URI from which to read
  65          * @throws MWException|Exception
  66          */
  67         function __construct( $source ) {
  68                 global $wgSVGMetadataCutoff;
  69                 $this->reader = new XMLReader();
  70
  71                 // Don't use $file->getSize() since file object passed to SVGHandler::getMetadata is bogus.
  72                 $size = filesize( $source );
  73                 if ( $size === false ) {
  74                         throw new MWException( "Error getting filesize of SVG." );
  75                 }
  76
  77                 if ( $size > $wgSVGMetadataCutoff ) {
  78                         $this->debug( "SVG is $size bytes, which is bigger than $wgSVGMetadataCutoff. Truncating." );
  79                         $contents = file_get_contents( $source, false, null, -1, $wgSVGMetadataCutoff );
  80                         if ( $contents === false ) {
  81                                 throw new MWException( 'Error reading SVG file.' );
  82                         }
  83                         $this->reader->XML( $contents, null, LIBXML_NOERROR | LIBXML_NOWARNING );
  84                 } else {
  85                         $this->reader->open( $source, null, LIBXML_NOERROR | LIBXML_NOWARNING );
  86                 }
  87
  88                 // Expand entities, since Adobe Illustrator uses them for xmlns
  89                 // attributes (bug 31719). Note that libxml2 has some protection
  90                 // against large recursive entity expansions so this is not as
  91                 // insecure as it might appear to be. However, it is still extremely
  92                 // insecure. It's necessary to wrap any read() calls with
  93                 // libxml_disable_entity_loader() to avoid arbitrary local file
  94                 // inclusion, or even arbitrary code execution if the expect
  95                 // extension is installed (bug 46859).
  96                 $oldDisable = libxml_disable_entity_loader( true );
  97                 $this->reader->setParserProperty( XMLReader::SUBST_ENTITIES, true );
  98
  99                 $this->metadata['width'] = self::DEFAULT_WIDTH;
 100                 $this->metadata['height'] = self::DEFAULT_HEIGHT;
 101
 102                 // The size in the units specified by the SVG file
 103                 // (for the metadata box)
 104                 // Per the SVG spec, if unspecified, default to '100%'
 105                 $this->metadata['originalWidth'] = '100%';
 106                 $this->metadata['originalHeight'] = '100%';
 107
 108                 // Because we cut off the end of the svg making an invalid one. Complicated
 109                 // try catch thing to make sure warnings get restored. Seems like there should
 110                 // be a better way.
 111                 MediaWiki\suppressWarnings();
 112                 try {
 113                         $this->read();
 114                 } catch ( Exception $e ) {
 115                         // Note, if this happens, the width/height will be taken to be 0x0.
 116                         // Should we consider it the default 512x512 instead?
 117                         MediaWiki\restoreWarnings();
 118                         libxml_disable_entity_loader( $oldDisable );
 119                         throw $e;
 120                 }
 121                 MediaWiki\restoreWarnings();
 122                 libxml_disable_entity_loader( $oldDisable );
 123         }
 124
 125         /**
 126          * @return array Array with the known metadata
 127          */
 128         public function getMetadata() {
 129                 return $this->metadata;
 130         }
 131
 132         /**
 133          * Read the SVG
 134          * @throws MWException
 135          * @return bool
 136          */
 137         protected function read() {
 138                 $keepReading = $this->reader->read();
 139
 140                 /* Skip until first element */
 141                 while ( $keepReading && $this->reader->nodeType != XMLReader::ELEMENT ) {
 142                         $keepReading = $this->reader->read();
 143                 }
 144
 145                 if ( $this->reader->localName != 'svg' || $this->reader->namespaceURI != self::NS_SVG ) {
 146                         throw new MWException( "Expected <svg> tag, got " .
 147                                 $this->reader->localName . " in NS " . $this->reader->namespaceURI );
 148                 }
 149                 $this->debug( "<svg> tag is correct." );
 150                 $this->handleSVGAttribs();
 151
 152                 $exitDepth = $this->reader->depth;
 153                 $keepReading = $this->reader->read();
 154                 while ( $keepReading ) {
 155                         $tag = $this->reader->localName;
 156                         $type = $this->reader->nodeType;
 157                         $isSVG = ( $this->reader->namespaceURI == self::NS_SVG );
 158
 159                         $this->debug( "$tag" );
 160
 161                         if ( $isSVG && $tag == 'svg' && $type == XMLReader::END_ELEMENT
 162                                 && $this->reader->depth <= $exitDepth
 163                         ) {
 164                                 break;
 165                         } elseif ( $isSVG && $tag == 'title' ) {
 166                                 $this->readField( $tag, 'title' );
 167                         } elseif ( $isSVG && $tag == 'desc' ) {
 168                                 $this->readField( $tag, 'description' );
 169                         } elseif ( $isSVG && $tag == 'metadata' && $type == XMLReader::ELEMENT ) {
 170                                 $this->readXml( $tag, 'metadata' );
 171                         } elseif ( $isSVG && $tag == 'script' ) {
 172                                 // We normally do not allow scripted svgs.
 173                                 // However its possible to configure MW to let them
 174                                 // in, and such files should be considered animated.
 175                                 $this->metadata['animated'] = true;
 176                         } elseif ( $tag !== '#text' ) {
 177                                 $this->debug( "Unhandled top-level XML tag $tag" );
 178
 179                                 // Recurse into children of current tag, looking for animation and languages.
 180                                 $this->animateFilterAndLang( $tag );
 181                         }
 182
 183                         // Goto next element, which is sibling of current (Skip children).
 184                         $keepReading = $this->reader->next();
 185                 }
 186
 187                 $this->reader->close();
 188
 189                 $this->metadata['translations'] = $this->languages + $this->languagePrefixes;
 190
 191                 return true;
 192         }
 193
 194         /**
 195          * Read a textelement from an element
 196          *
 197          * @param string $name Name of the element that we are reading from
 198          * @param string $metafield Field that we will fill with the result
 199          */
 200         private function readField( $name, $metafield = null ) {
 201                 $this->debug( "Read field $metafield" );
 202                 if ( !$metafield || $this->reader->nodeType != XMLReader::ELEMENT ) {
 203                         return;
 204                 }
 205                 $keepReading = $this->reader->read();
 206                 while ( $keepReading ) {
 207                         if ( $this->reader->localName == $name
 208                                 && $this->reader->namespaceURI == self::NS_SVG
 209                                 && $this->reader->nodeType == XMLReader::END_ELEMENT
 210                         ) {
 211                                 break;
 212                         } elseif ( $this->reader->nodeType == XMLReader::TEXT ) {
 213                                 $this->metadata[$metafield] = trim( $this->reader->value );
 214                         }
 215                         $keepReading = $this->reader->read();
 216                 }
 217         }
 218
 219         /**
 220          * Read an XML snippet from an element
 221          *
 222          * @param string $metafield Field that we will fill with the result
 223          * @throws MWException
 224          */
 225         private function readXml( $metafield = null ) {
 226                 $this->debug( "Read top level metadata" );
 227                 if ( !$metafield || $this->reader->nodeType != XMLReader::ELEMENT ) {
 228                         return;
 229                 }
 230                 // @todo Find and store type of xml snippet. metadata['metadataType'] = "rdf"
 231                 if ( method_exists( $this->reader, 'readInnerXML' ) ) {
 232                         $this->metadata[$metafield] = trim( $this->reader->readInnerXML() );
 233                 } else {
 234                         throw new MWException( "The PHP XMLReader extension does not come " .
 235                                 "with readInnerXML() method. Your libxml is probably out of " .
 236                                 "date (need 2.6.20 or later)." );
 237                 }
 238                 $this->reader->next();
 239         }
 240
 241         /**
 242          * Filter all children, looking for animated elements.
 243          * Also get a list of languages that can be targeted.
 244          *
 245          * @param string $name Name of the element that we are reading from
 246          */
 247         private function animateFilterAndLang( $name ) {
 248                 $this->debug( "animate filter for tag $name" );
 249                 if ( $this->reader->nodeType != XMLReader::ELEMENT ) {
 250                         return;
 251                 }
 252                 if ( $this->reader->isEmptyElement ) {
 253                         return;
 254                 }
 255                 $exitDepth = $this->reader->depth;
 256                 $keepReading = $this->reader->read();
 257                 while ( $keepReading ) {
 258                         if ( $this->reader->localName == $name && $this->reader->depth <= $exitDepth
 259                                 && $this->reader->nodeType == XMLReader::END_ELEMENT
 260                         ) {
 261                                 break;
 262                         } elseif ( $this->reader->namespaceURI == self::NS_SVG
 263                                 && $this->reader->nodeType == XMLReader::ELEMENT
 264                         ) {
 265                                 $sysLang = $this->reader->getAttribute( 'systemLanguage' );
 266                                 if ( !is_null( $sysLang ) && $sysLang !== '' ) {
 267                                         // See http://www.w3.org/TR/SVG/struct.html#SystemLanguageAttribute
 268                                         $langList = explode( ',', $sysLang );
 269                                         foreach ( $langList as $langItem ) {
 270                                                 $langItem = trim( $langItem );
 271                                                 if ( Language::isWellFormedLanguageTag( $langItem ) ) {
 272                                                         $this->languages[$langItem] = self::LANG_FULL_MATCH;
 273                                                 }
 274                                                 // Note, the standard says that any prefix should work,
 275                                                 // here we do only the initial prefix, since that will catch
 276                                                 // 99% of cases, and we are going to compare against fallbacks.
 277                                                 // This differs mildly from how the spec says languages should be
 278                                                 // handled, however it matches better how the MediaWiki language
 279                                                 // preference is generally handled.
 280                                                 $dash = strpos( $langItem, '-' );
 281                                                 // Intentionally checking both !false and > 0 at the same time.
 282                                                 if ( $dash ) {
 283                                                         $itemPrefix = substr( $langItem, 0, $dash );
 284                                                         if ( Language::isWellFormedLanguageTag( $itemPrefix ) ) {
 285                                                                 $this->languagePrefixes[$itemPrefix] = self::LANG_PREFIX_MATCH;
 286                                                         }
 287                                                 }
 288                                         }
 289                                 }
 290                                 switch ( $this->reader->localName ) {
 291                                         case 'script':
 292                                                 // Normally we disallow files with
 293                                                 // <script>, but its possible
 294                                                 // to configure MW to disable
 295                                                 // such checks.
 296                                         case 'animate':
 297                                         case 'set':
 298                                         case 'animateMotion':
 299                                         case 'animateColor':
 300                                         case 'animateTransform':
 301                                                 $this->debug( "HOUSTON WE HAVE ANIMATION" );
 302                                                 $this->metadata['animated'] = true;
 303                                                 break;
 304                                 }
 305                         }
 306                         $keepReading = $this->reader->read();
 307                 }
 308         }
 309
 310         // @todo FIXME: Unused, remove?
 311         private function throwXmlError( $err ) {
 312                 $this->debug( "FAILURE: $err" );
 313                 wfDebug( "SVGReader XML error: $err\n" );
 314         }
 315
 316         private function debug( $data ) {
 317                 if ( $this->mDebug ) {
 318                         wfDebug( "SVGReader: $data\n" );
 319                 }
 320         }
 321
 322         /**
 323          * Parse the attributes of an SVG element
 324          *
 325          * The parser has to be in the start element of "<svg>"
 326          */
 327         private function handleSVGAttribs() {
 328                 $defaultWidth = self::DEFAULT_WIDTH;
 329                 $defaultHeight = self::DEFAULT_HEIGHT;
 330                 $aspect = 1.0;
 331                 $width = null;
 332                 $height = null;
 333
 334                 if ( $this->reader->getAttribute( 'viewBox' ) ) {
 335                         // min-x min-y width height
 336                         $viewBox = preg_split( '/\s+/', trim( $this->reader->getAttribute( 'viewBox' ) ) );
 337                         if ( count( $viewBox ) == 4 ) {
 338                                 $viewWidth = $this->scaleSVGUnit( $viewBox[2] );
 339                                 $viewHeight = $this->scaleSVGUnit( $viewBox[3] );
 340                                 if ( $viewWidth > 0 && $viewHeight > 0 ) {
 341                                         $aspect = $viewWidth / $viewHeight;
 342                                         $defaultHeight = $defaultWidth / $aspect;
 343                                 }
 344                         }
 345                 }
 346                 if ( $this->reader->getAttribute( 'width' ) ) {
 347                         $width = $this->scaleSVGUnit( $this->reader->getAttribute( 'width' ), $defaultWidth );
 348                         $this->metadata['originalWidth'] = $this->reader->getAttribute( 'width' );
 349                 }
 350                 if ( $this->reader->getAttribute( 'height' ) ) {
 351                         $height = $this->scaleSVGUnit( $this->reader->getAttribute( 'height' ), $defaultHeight );
 352                         $this->metadata['originalHeight'] = $this->reader->getAttribute( 'height' );
 353                 }
 354
 355                 if ( !isset( $width ) && !isset( $height ) ) {
 356                         $width = $defaultWidth;
 357                         $height = $width / $aspect;
 358                 } elseif ( isset( $width ) && !isset( $height ) ) {
 359                         $height = $width / $aspect;
 360                 } elseif ( isset( $height ) && !isset( $width ) ) {
 361                         $width = $height * $aspect;
 362                 }
 363
 364                 if ( $width > 0 && $height > 0 ) {
 365                         $this->metadata['width'] = intval( round( $width ) );
 366                         $this->metadata['height'] = intval( round( $height ) );
 367                 }
 368         }
 369
 370         /**
 371          * Return a rounded pixel equivalent for a labeled CSS/SVG length.
 372          * http://www.w3.org/TR/SVG11/coords.html#UnitIdentifiers
 373          *
 374          * @param string $length CSS/SVG length.
 375          * @param float|int $viewportSize Optional scale for percentage units...
 376          * @return float Length in pixels
 377          */
 378         static function scaleSVGUnit( $length, $viewportSize = 512 ) {
 379                 static $unitLength = array(
 380                         'px' => 1.0,
 381                         'pt' => 1.25,
 382                         'pc' => 15.0,
 383                         'mm' => 3.543307,
 384                         'cm' => 35.43307,
 385                         'in' => 90.0,
 386                         'em' => 16.0, // fake it?
 387                         'ex' => 12.0, // fake it?
 388                         '' => 1.0, // "User units" pixels by default
 389                 );
 390                 $matches = array();
 391                 if ( preg_match( '/^\s*(\d+(?:\.\d+)?)(em|ex|px|pt|pc|cm|mm|in|%|)\s*$/', $length, $matches ) ) {
 392                         $length = floatval( $matches[1] );
 393                         $unit = $matches[2];
 394                         if ( $unit == '%' ) {
 395                                 return $length * 0.01 * $viewportSize;
 396                         } else {
 397                                 return $length * $unitLength[$unit];
 398                         }
 399                 } else {
 400                         // Assume pixels
 401                         return floatval( $length );
 402                 }
 403         }
 404 }