SVGMetadataExtractor.php

Go to the documentation of this file.

<?php
class SVGMetadataExtractor {
    static function getMetadata( $filename ) {
        $svg = new SVGReader( $filename );

        return $svg->getMetadata();
    }
}

class SVGReader {
    const DEFAULT_WIDTH = 512;
    const DEFAULT_HEIGHT = 512;
    const NS_SVG = 'http://www.w3.org/2000/svg';
    const LANG_PREFIX_MATCH = 1;
    const LANG_FULL_MATCH = 2;

    private $reader = null;

    private $mDebug = false;

    private $metadata = [];
    private $languages = [];
    private $languagePrefixes = [];

    function __construct( $source ) {
        global $wgSVGMetadataCutoff;
        $this->reader = new XMLReader();

        // Don't use $file->getSize() since file object passed to SVGHandler::getMetadata is bogus.
        $size = filesize( $source );
        if ( $size === false ) {
            throw new MWException( "Error getting filesize of SVG." );
        }

        if ( $size > $wgSVGMetadataCutoff ) {
            $this->debug( "SVG is $size bytes, which is bigger than $wgSVGMetadataCutoff. Truncating." );
            $contents = file_get_contents( $source, false, null, -1, $wgSVGMetadataCutoff );
            if ( $contents === false ) {
                throw new MWException( 'Error reading SVG file.' );
            }
            $this->reader->XML( $contents, null, LIBXML_NOERROR | LIBXML_NOWARNING );
        } else {
            $this->reader->open( $source, null, LIBXML_NOERROR | LIBXML_NOWARNING );
        }

        // Expand entities, since Adobe Illustrator uses them for xmlns
        // attributes (bug 31719). Note that libxml2 has some protection
        // against large recursive entity expansions so this is not as
        // insecure as it might appear to be. However, it is still extremely
        // insecure. It's necessary to wrap any read() calls with
        // libxml_disable_entity_loader() to avoid arbitrary local file
        // inclusion, or even arbitrary code execution if the expect
        // extension is installed (bug 46859).
        $oldDisable = libxml_disable_entity_loader( true );
        $this->reader->setParserProperty( XMLReader::SUBST_ENTITIES, true );

        $this->metadata['width'] = self::DEFAULT_WIDTH;
        $this->metadata['height'] = self::DEFAULT_HEIGHT;

        // The size in the units specified by the SVG file
        // (for the metadata box)
        // Per the SVG spec, if unspecified, default to '100%'
        $this->metadata['originalWidth'] = '100%';
        $this->metadata['originalHeight'] = '100%';

        // Because we cut off the end of the svg making an invalid one. Complicated
        // try catch thing to make sure warnings get restored. Seems like there should
        // be a better way.
        MediaWiki\suppressWarnings();
        try {
            $this->read();
        } catch ( Exception $e ) {
            // Note, if this happens, the width/height will be taken to be 0x0.
            // Should we consider it the default 512x512 instead?
            MediaWiki\restoreWarnings();
            libxml_disable_entity_loader( $oldDisable );
            throw $e;
        }
        MediaWiki\restoreWarnings();
        libxml_disable_entity_loader( $oldDisable );
    }

    public function getMetadata() {
        return $this->metadata;
    }

    protected function read() {
        $keepReading = $this->reader->read();

        /* Skip until first element */
        while ( $keepReading && $this->reader->nodeType != XMLReader::ELEMENT ) {
            $keepReading = $this->reader->read();
        }

        if ( $this->reader->localName != 'svg' || $this->reader->namespaceURI != self::NS_SVG ) {
            throw new MWException( "Expected <svg> tag, got " .
                $this->reader->localName . " in NS " . $this->reader->namespaceURI );
        }
        $this->debug( "<svg> tag is correct." );
        $this->handleSVGAttribs();

        $exitDepth = $this->reader->depth;
        $keepReading = $this->reader->read();
        while ( $keepReading ) {
            $tag = $this->reader->localName;
            $type = $this->reader->nodeType;
            $isSVG = ( $this->reader->namespaceURI == self::NS_SVG );

            $this->debug( "$tag" );

            if ( $isSVG && $tag == 'svg' && $type == XMLReader::END_ELEMENT
                && $this->reader->depth <= $exitDepth
            ) {
                break;
            } elseif ( $isSVG && $tag == 'title' ) {
                $this->readField( $tag, 'title' );
            } elseif ( $isSVG && $tag == 'desc' ) {
                $this->readField( $tag, 'description' );
            } elseif ( $isSVG && $tag == 'metadata' && $type == XMLReader::ELEMENT ) {
                $this->readXml( $tag, 'metadata' );
            } elseif ( $isSVG && $tag == 'script' ) {
                // We normally do not allow scripted svgs.
                // However its possible to configure MW to let them
                // in, and such files should be considered animated.
                $this->metadata['animated'] = true;
            } elseif ( $tag !== '#text' ) {
                $this->debug( "Unhandled top-level XML tag $tag" );

                // Recurse into children of current tag, looking for animation and languages.
                $this->animateFilterAndLang( $tag );
            }

            // Goto next element, which is sibling of current (Skip children).
            $keepReading = $this->reader->next();
        }

        $this->reader->close();

        $this->metadata['translations'] = $this->languages + $this->languagePrefixes;

        return true;
    }

    private function readField( $name, $metafield = null ) {
        $this->debug( "Read field $metafield" );
        if ( !$metafield || $this->reader->nodeType != XMLReader::ELEMENT ) {
            return;
        }
        $keepReading = $this->reader->read();
        while ( $keepReading ) {
            if ( $this->reader->localName == $name
                && $this->reader->namespaceURI == self::NS_SVG
                && $this->reader->nodeType == XMLReader::END_ELEMENT
            ) {
                break;
            } elseif ( $this->reader->nodeType == XMLReader::TEXT ) {
                $this->metadata[$metafield] = trim( $this->reader->value );
            }
            $keepReading = $this->reader->read();
        }
    }

    private function readXml( $metafield = null ) {
        $this->debug( "Read top level metadata" );
        if ( !$metafield || $this->reader->nodeType != XMLReader::ELEMENT ) {
            return;
        }
        // @todo Find and store type of xml snippet. metadata['metadataType'] = "rdf"
        if ( method_exists( $this->reader, 'readInnerXML' ) ) {
            $this->metadata[$metafield] = trim( $this->reader->readInnerXml() );
        } else {
            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)." );
        }
        $this->reader->next();
    }

    private function animateFilterAndLang( $name ) {
        $this->debug( "animate filter for tag $name" );
        if ( $this->reader->nodeType != XMLReader::ELEMENT ) {
            return;
        }
        if ( $this->reader->isEmptyElement ) {
            return;
        }
        $exitDepth = $this->reader->depth;
        $keepReading = $this->reader->read();
        while ( $keepReading ) {
            if ( $this->reader->localName == $name && $this->reader->depth <= $exitDepth
                && $this->reader->nodeType == XMLReader::END_ELEMENT
            ) {
                break;
            } elseif ( $this->reader->namespaceURI == self::NS_SVG
                && $this->reader->nodeType == XMLReader::ELEMENT
            ) {
                $sysLang = $this->reader->getAttribute( 'systemLanguage' );
                if ( !is_null( $sysLang ) && $sysLang !== '' ) {
                    // See http://www.w3.org/TR/SVG/struct.html#SystemLanguageAttribute
                    $langList = explode( ',', $sysLang );
                    foreach ( $langList as $langItem ) {
                        $langItem = trim( $langItem );
                        if ( Language::isWellFormedLanguageTag( $langItem ) ) {
                            $this->languages[$langItem] = self::LANG_FULL_MATCH;
                        }
                        // Note, the standard says that any prefix should work,
                        // here we do only the initial prefix, since that will catch
                        // 99% of cases, and we are going to compare against fallbacks.
                        // This differs mildly from how the spec says languages should be
                        // handled, however it matches better how the MediaWiki language
                        // preference is generally handled.
                        $dash = strpos( $langItem, '-' );
                        // Intentionally checking both !false and > 0 at the same time.
                        if ( $dash ) {
                            $itemPrefix = substr( $langItem, 0, $dash );
                            if ( Language::isWellFormedLanguageTag( $itemPrefix ) ) {
                                $this->languagePrefixes[$itemPrefix] = self::LANG_PREFIX_MATCH;
                            }
                        }
                    }
                }
                switch ( $this->reader->localName ) {
                    case 'script':
                        // Normally we disallow files with
                        // <script>, but its possible
                        // to configure MW to disable
                        // such checks.
                    case 'animate':
                    case 'set':
                    case 'animateMotion':
                    case 'animateColor':
                    case 'animateTransform':
                        $this->debug( "HOUSTON WE HAVE ANIMATION" );
                        $this->metadata['animated'] = true;
                        break;
                }
            }
            $keepReading = $this->reader->read();
        }
    }

    // @todo FIXME: Unused, remove?
    private function throwXmlError( $err ) {
        $this->debug( "FAILURE: $err" );
        wfDebug( "SVGReader XML error: $err\n" );
    }

    private function debug( $data ) {
        if ( $this->mDebug ) {
            wfDebug( "SVGReader: $data\n" );
        }
    }

    private function handleSVGAttribs() {
        $defaultWidth = self::DEFAULT_WIDTH;
        $defaultHeight = self::DEFAULT_HEIGHT;
        $aspect = 1.0;
        $width = null;
        $height = null;

        if ( $this->reader->getAttribute( 'viewBox' ) ) {
            // min-x min-y width height
            $viewBox = preg_split( '/\s+/', trim( $this->reader->getAttribute( 'viewBox' ) ) );
            if ( count( $viewBox ) == 4 ) {
                $viewWidth = $this->scaleSVGUnit( $viewBox[2] );
                $viewHeight = $this->scaleSVGUnit( $viewBox[3] );
                if ( $viewWidth > 0 && $viewHeight > 0 ) {
                    $aspect = $viewWidth / $viewHeight;
                    $defaultHeight = $defaultWidth / $aspect;
                }
            }
        }
        if ( $this->reader->getAttribute( 'width' ) ) {
            $width = $this->scaleSVGUnit( $this->reader->getAttribute( 'width' ), $defaultWidth );
            $this->metadata['originalWidth'] = $this->reader->getAttribute( 'width' );
        }
        if ( $this->reader->getAttribute( 'height' ) ) {
            $height = $this->scaleSVGUnit( $this->reader->getAttribute( 'height' ), $defaultHeight );
            $this->metadata['originalHeight'] = $this->reader->getAttribute( 'height' );
        }

        if ( !isset( $width ) && !isset( $height ) ) {
            $width = $defaultWidth;
            $height = $width / $aspect;
        } elseif ( isset( $width ) && !isset( $height ) ) {
            $height = $width / $aspect;
        } elseif ( isset( $height ) && !isset( $width ) ) {
            $width = $height * $aspect;
        }

        if ( $width > 0 && $height > 0 ) {
            $this->metadata['width'] = intval( round( $width ) );
            $this->metadata['height'] = intval( round( $height ) );
        }
    }

    static function scaleSVGUnit( $length, $viewportSize = 512 ) {
        static $unitLength = [
            'px' => 1.0,
            'pt' => 1.25,
            'pc' => 15.0,
            'mm' => 3.543307,
            'cm' => 35.43307,
            'in' => 90.0,
            'em' => 16.0, // fake it?
            'ex' => 12.0, // fake it?
            '' => 1.0, // "User units" pixels by default
        ];
        $matches = [];
        if ( preg_match( '/^\s*(\d+(?:\.\d+)?)(em|ex|px|pt|pc|cm|mm|in|%|)\s*$/', $length, $matches ) ) {
            $length = floatval( $matches[1] );
            $unit = $matches[2];
            if ( $unit == '%' ) {
                return $length * 0.01 * $viewportSize;
            } else {
                return $length * $unitLength[$unit];
            }
        } else {
            // Assume pixels
            return floatval( $length );
        }
    }
}