MediaWiki  REL1_22
SVGMetadataExtractor.php
Go to the documentation of this file.
00001 <?php
00031 class SVGMetadataExtractor {
00032     static function getMetadata( $filename ) {
00033         $svg = new SVGReader( $filename );
00034         return $svg->getMetadata();
00035     }
00036 }
00037 
00041 class SVGReader {
00042     const DEFAULT_WIDTH = 512;
00043     const DEFAULT_HEIGHT = 512;
00044     const NS_SVG = 'http://www.w3.org/2000/svg';
00045 
00046     private $reader = null;
00047     private $mDebug = false;
00048     private $metadata = Array();
00049 
00057     function __construct( $source ) {
00058         global $wgSVGMetadataCutoff;
00059         $this->reader = new XMLReader();
00060 
00061         // Don't use $file->getSize() since file object passed to SVGHandler::getMetadata is bogus.
00062         $size = filesize( $source );
00063         if ( $size === false ) {
00064             throw new MWException( "Error getting filesize of SVG." );
00065         }
00066 
00067         if ( $size > $wgSVGMetadataCutoff ) {
00068             $this->debug( "SVG is $size bytes, which is bigger than $wgSVGMetadataCutoff. Truncating." );
00069             $contents = file_get_contents( $source, false, null, -1, $wgSVGMetadataCutoff );
00070             if ( $contents === false ) {
00071                 throw new MWException( 'Error reading SVG file.' );
00072             }
00073             $this->reader->XML( $contents, null, LIBXML_NOERROR | LIBXML_NOWARNING );
00074         } else {
00075             $this->reader->open( $source, null, LIBXML_NOERROR | LIBXML_NOWARNING );
00076         }
00077 
00078         // Expand entities, since Adobe Illustrator uses them for xmlns
00079         // attributes (bug 31719). Note that libxml2 has some protection
00080         // against large recursive entity expansions so this is not as
00081         // insecure as it might appear to be. However, it is still extremely
00082         // insecure. It's necessary to wrap any read() calls with
00083         // libxml_disable_entity_loader() to avoid arbitrary local file
00084         // inclusion, or even arbitrary code execution if the expect
00085         // extension is installed (bug 46859).
00086         $oldDisable = libxml_disable_entity_loader( true );
00087         $this->reader->setParserProperty( XMLReader::SUBST_ENTITIES, true );
00088 
00089         $this->metadata['width'] = self::DEFAULT_WIDTH;
00090         $this->metadata['height'] = self::DEFAULT_HEIGHT;
00091 
00092         // The size in the units specified by the SVG file
00093         // (for the metadata box)
00094         // Per the SVG spec, if unspecified, default to '100%'
00095         $this->metadata['originalWidth'] = '100%';
00096         $this->metadata['originalHeight'] = '100%';
00097 
00098         // Because we cut off the end of the svg making an invalid one. Complicated
00099         // try catch thing to make sure warnings get restored. Seems like there should
00100         // be a better way.
00101         wfSuppressWarnings();
00102         try {
00103             $this->read();
00104         } catch ( Exception $e ) {
00105             // Note, if this happens, the width/height will be taken to be 0x0.
00106             // Should we consider it the default 512x512 instead?
00107             wfRestoreWarnings();
00108             libxml_disable_entity_loader( $oldDisable );
00109             throw $e;
00110         }
00111         wfRestoreWarnings();
00112         libxml_disable_entity_loader( $oldDisable );
00113     }
00114 
00118     public function getMetadata() {
00119         return $this->metadata;
00120     }
00121 
00127     protected function read() {
00128         $keepReading = $this->reader->read();
00129 
00130         /* Skip until first element */
00131         while ( $keepReading && $this->reader->nodeType != XmlReader::ELEMENT ) {
00132             $keepReading = $this->reader->read();
00133         }
00134 
00135         if ( $this->reader->localName != 'svg' || $this->reader->namespaceURI != self::NS_SVG ) {
00136             throw new MWException( "Expected <svg> tag, got " .
00137                 $this->reader->localName . " in NS " . $this->reader->namespaceURI );
00138         }
00139         $this->debug( "<svg> tag is correct." );
00140         $this->handleSVGAttribs();
00141 
00142         $exitDepth = $this->reader->depth;
00143         $keepReading = $this->reader->read();
00144         while ( $keepReading ) {
00145             $tag = $this->reader->localName;
00146             $type = $this->reader->nodeType;
00147             $isSVG = ( $this->reader->namespaceURI == self::NS_SVG );
00148 
00149             $this->debug( "$tag" );
00150 
00151             if ( $isSVG && $tag == 'svg' && $type == XmlReader::END_ELEMENT && $this->reader->depth <= $exitDepth ) {
00152                 break;
00153             } elseif ( $isSVG && $tag == 'title' ) {
00154                 $this->readField( $tag, 'title' );
00155             } elseif ( $isSVG && $tag == 'desc' ) {
00156                 $this->readField( $tag, 'description' );
00157             } elseif ( $isSVG && $tag == 'metadata' && $type == XmlReader::ELEMENT ) {
00158                 $this->readXml( $tag, 'metadata' );
00159             } elseif ( $isSVG && $tag == 'script' ) {
00160                 // We normally do not allow scripted svgs.
00161                 // However its possible to configure MW to let them
00162                 // in, and such files should be considered animated.
00163                 $this->metadata['animated'] = true;
00164             } elseif ( $tag !== '#text' ) {
00165                 $this->debug( "Unhandled top-level XML tag $tag" );
00166 
00167                 if ( !isset( $this->metadata['animated'] ) ) {
00168                     // Recurse into children of current tag, looking for animation.
00169                     $this->animateFilter( $tag );
00170                 }
00171             }
00172 
00173             // Goto next element, which is sibling of current (Skip children).
00174             $keepReading = $this->reader->next();
00175         }
00176 
00177         $this->reader->close();
00178 
00179         return true;
00180     }
00181 
00188     private function readField( $name, $metafield = null ) {
00189         $this->debug( "Read field $metafield" );
00190         if ( !$metafield || $this->reader->nodeType != XmlReader::ELEMENT ) {
00191             return;
00192         }
00193         $keepReading = $this->reader->read();
00194         while ( $keepReading ) {
00195             if ( $this->reader->localName == $name && $this->reader->namespaceURI == self::NS_SVG && $this->reader->nodeType == XmlReader::END_ELEMENT ) {
00196                 break;
00197             } elseif ( $this->reader->nodeType == XmlReader::TEXT ) {
00198                 $this->metadata[$metafield] = trim( $this->reader->value );
00199             }
00200             $keepReading = $this->reader->read();
00201         }
00202     }
00203 
00210     private function readXml( $metafield = null ) {
00211         $this->debug( "Read top level metadata" );
00212         if ( !$metafield || $this->reader->nodeType != XmlReader::ELEMENT ) {
00213             return;
00214         }
00215         // TODO: find and store type of xml snippet. metadata['metadataType'] = "rdf"
00216         if ( method_exists( $this->reader, 'readInnerXML' ) ) {
00217             $this->metadata[$metafield] = trim( $this->reader->readInnerXML() );
00218         } else {
00219             throw new MWException( "The PHP XMLReader extension does not come with readInnerXML() method. Your libxml is probably out of date (need 2.6.20 or later)." );
00220         }
00221         $this->reader->next();
00222     }
00223 
00229     private function animateFilter( $name ) {
00230         $this->debug( "animate filter for tag $name" );
00231         if ( $this->reader->nodeType != XmlReader::ELEMENT ) {
00232             return;
00233         }
00234         if ( $this->reader->isEmptyElement ) {
00235             return;
00236         }
00237         $exitDepth = $this->reader->depth;
00238         $keepReading = $this->reader->read();
00239         while ( $keepReading ) {
00240             if ( $this->reader->localName == $name && $this->reader->depth <= $exitDepth
00241                 && $this->reader->nodeType == XmlReader::END_ELEMENT ) {
00242                 break;
00243             } elseif ( $this->reader->namespaceURI == self::NS_SVG && $this->reader->nodeType == XmlReader::ELEMENT ) {
00244                 switch ( $this->reader->localName ) {
00245                     case 'script':
00246                         // Normally we disallow files with
00247                         // <script>, but its possible
00248                         // to configure MW to disable
00249                         // such checks.
00250                     case 'animate':
00251                     case 'set':
00252                     case 'animateMotion':
00253                     case 'animateColor':
00254                     case 'animateTransform':
00255                         $this->debug( "HOUSTON WE HAVE ANIMATION" );
00256                         $this->metadata['animated'] = true;
00257                         break;
00258                 }
00259             }
00260             $keepReading = $this->reader->read();
00261         }
00262     }
00263 
00264     private function throwXmlError( $err ) {
00265         $this->debug( "FAILURE: $err" );
00266         wfDebug( "SVGReader XML error: $err\n" );
00267     }
00268 
00269     private function debug( $data ) {
00270         if ( $this->mDebug ) {
00271             wfDebug( "SVGReader: $data\n" );
00272         }
00273     }
00274 
00275     private function warn( $data ) {
00276         wfDebug( "SVGReader: $data\n" );
00277     }
00278 
00279     private function notice( $data ) {
00280         wfDebug( "SVGReader WARN: $data\n" );
00281     }
00282 
00288     private function handleSVGAttribs() {
00289         $defaultWidth = self::DEFAULT_WIDTH;
00290         $defaultHeight = self::DEFAULT_HEIGHT;
00291         $aspect = 1.0;
00292         $width = null;
00293         $height = null;
00294 
00295         if ( $this->reader->getAttribute( 'viewBox' ) ) {
00296             // min-x min-y width height
00297             $viewBox = preg_split( '/\s+/', trim( $this->reader->getAttribute( 'viewBox' ) ) );
00298             if ( count( $viewBox ) == 4 ) {
00299                 $viewWidth = $this->scaleSVGUnit( $viewBox[2] );
00300                 $viewHeight = $this->scaleSVGUnit( $viewBox[3] );
00301                 if ( $viewWidth > 0 && $viewHeight > 0 ) {
00302                     $aspect = $viewWidth / $viewHeight;
00303                     $defaultHeight = $defaultWidth / $aspect;
00304                 }
00305             }
00306         }
00307         if ( $this->reader->getAttribute( 'width' ) ) {
00308             $width = $this->scaleSVGUnit( $this->reader->getAttribute( 'width' ), $defaultWidth );
00309             $this->metadata['originalWidth'] = $this->reader->getAttribute( 'width' );
00310         }
00311         if ( $this->reader->getAttribute( 'height' ) ) {
00312             $height = $this->scaleSVGUnit( $this->reader->getAttribute( 'height' ), $defaultHeight );
00313             $this->metadata['originalHeight'] = $this->reader->getAttribute( 'height' );
00314         }
00315 
00316         if ( !isset( $width ) && !isset( $height ) ) {
00317             $width = $defaultWidth;
00318             $height = $width / $aspect;
00319         } elseif ( isset( $width ) && !isset( $height ) ) {
00320             $height = $width / $aspect;
00321         } elseif ( isset( $height ) && !isset( $width ) ) {
00322             $width = $height * $aspect;
00323         }
00324 
00325         if ( $width > 0 && $height > 0 ) {
00326             $this->metadata['width'] = intval( round( $width ) );
00327             $this->metadata['height'] = intval( round( $height ) );
00328         }
00329     }
00330 
00339     static function scaleSVGUnit( $length, $viewportSize = 512 ) {
00340         static $unitLength = array(
00341             'px' => 1.0,
00342             'pt' => 1.25,
00343             'pc' => 15.0,
00344             'mm' => 3.543307,
00345             'cm' => 35.43307,
00346             'in' => 90.0,
00347             'em' => 16.0, // fake it?
00348             'ex' => 12.0, // fake it?
00349             '' => 1.0, // "User units" pixels by default
00350             );
00351         $matches = array();
00352         if ( preg_match( '/^\s*(\d+(?:\.\d+)?)(em|ex|px|pt|pc|cm|mm|in|%|)\s*$/', $length, $matches ) ) {
00353             $length = floatval( $matches[1] );
00354             $unit = $matches[2];
00355             if ( $unit == '%' ) {
00356                 return $length * 0.01 * $viewportSize;
00357             } else {
00358                 return $length * $unitLength[$unit];
00359             }
00360         } else {
00361             // Assume pixels
00362             return floatval( $length );
00363         }
00364     }
00365 }