FormatMetadata.php

Go to the documentation of this file.

<?php
namespace MediaWiki\Media;
 
use InvalidArgumentException;
use MediaWiki\Api\ApiResult;
use MediaWiki\Context\ContextSource;
use MediaWiki\Context\IContextSource;
use MediaWiki\FileRepo\File\File;
use MediaWiki\FileRepo\File\ForeignAPIFile;
use MediaWiki\HookContainer\ProtectedHookAccessorTrait;
use MediaWiki\Html\Html;
use MediaWiki\Logger\LoggerFactory;
use MediaWiki\MediaWikiServices;
use Wikimedia\Timestamp\TimestampFormat as TS;
 
class FormatMetadata extends ContextSource {
    use ProtectedHookAccessorTrait;
 
    protected $singleLang = false;
 
    public function setSingleLanguage( $val ) {
        $this->singleLang = $val;
    }
 
    public static function getFormattedData( $tags, $context = false ) {
        $obj = new self;
        if ( $context ) {
            $obj->setContext( $context );
        }
 
        return $obj->makeFormattedData( $tags );
    }
 
    public function makeFormattedData( $tags ) {
        $resolutionunit = !isset( $tags['ResolutionUnit'] ) || $tags['ResolutionUnit'] == 2 ? 2 : 3;
        unset( $tags['ResolutionUnit'] );
 
        // Ignore these complex values
        unset( $tags['HasExtendedXMP'] );
        unset( $tags['AuthorsPosition'] );
        unset( $tags['LocationCreated'] );
        unset( $tags['LocationShown'] );
        unset( $tags['GPSAltitudeRef'] );
 
        foreach ( $tags as $tag => &$vals ) {
            // This seems ugly to wrap non-array's in an array just to unwrap again,
            // especially when most of the time it is not an array
            if ( !is_array( $vals ) ) {
                $vals = [ $vals ];
            }
 
            // _type is a special value to say what array type
            if ( isset( $vals['_type'] ) ) {
                $type = $vals['_type'];
                unset( $vals['_type'] );
            } else {
                $type = 'ul'; // default unordered list.
            }
 
            // _formatted is a special value to indicate the subclass
            // already handled & formatted this tag as wikitext
            if ( isset( $tags[$tag]['_formatted'] ) ) {
                $tags[$tag] = $this->flattenArrayReal(
                    $tags[$tag]['_formatted'], $type
                );
                continue;
            }
 
            // This is done differently as the tag is an array.
            if ( $tag === 'GPSTimeStamp' && count( $vals ) === 3 ) {
                // hour min sec array
 
                $h = explode( '/', $vals[0], 2 );
                $m = explode( '/', $vals[1], 2 );
                $s = explode( '/', $vals[2], 2 );
 
                // this should already be validated
                // when loaded from file, but it could
                // come from a foreign repo, so be
                // paranoid.
                if ( !isset( $h[1] )
                    || !isset( $m[1] )
                    || !isset( $s[1] )
                    || $h[1] == 0
                    || $m[1] == 0
                    || $s[1] == 0
                ) {
                    continue;
                }
                $vals = str_pad( (string)( (int)$h[0] / (int)$h[1] ), 2, '0', STR_PAD_LEFT )
                    . ':' . str_pad( (string)( (int)$m[0] / (int)$m[1] ), 2, '0', STR_PAD_LEFT )
                    . ':' . str_pad( (string)( (int)$s[0] / (int)$s[1] ), 2, '0', STR_PAD_LEFT );
 
                $time = wfTimestamp( TS::MW, '1971:01:01 ' . $vals );
                // the 1971:01:01 is just a placeholder, and not shown to user.
                if ( $time && (int)$time > 0 ) {
                    $vals = $this->getLanguage()->time( $time );
                }
                continue;
            }
 
            // The contact info is a multi-valued field
            // instead of the other props which are single
            // valued (mostly) so handle as a special case.
            if ( $tag === 'Contact' || $tag === 'CreatorContactInfo' ) {
                $vals = $this->collapseContactInfo( $vals );
                continue;
            }
 
            foreach ( $vals as &$val ) {
                switch ( $tag ) {
                    case 'Compression':
                        $val = match ( $val ) {
                            1, 2, 3, 4, 5, 6, 7, 8,
                            32773,
                            32946,
                            34712 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'PhotometricInterpretation':
                        $val = match ( $val ) {
                            0, 1, 2, 3, 4, 5, 6, 8, 9, 10,
                            32803,
                            34892 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'Orientation':
                        $val = match ( $val ) {
                            1, 2, 3, 4, 5, 6, 7, 8 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'PlanarConfiguration':
                    // TODO: YCbCrSubSampling
                    case 'YCbCrPositioning':
                        $val = match ( $val ) {
                            1, 2 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'XResolution':
                    case 'YResolution':
                        $val = match ( $resolutionunit ) {
                            2 => $this->exifMsg( 'XYResolution', 'i', $this->formatNum( $val ) ),
                            3 => $this->exifMsg( 'XYResolution', 'c', $this->formatNum( $val ) ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    // TODO: YCbCrCoefficients  #p27 (see annex E)
                    case 'ExifVersion':
                    // PHP likes to be the odd one out with casing of FlashPixVersion;
                    // https://www.exif.org/Exif2-2.PDF#page=32 and
                    // https://www.digitalgalen.net/Documents/External/XMP/XMPSpecificationPart2.pdf#page=51
                    // both use FlashpixVersion. However, since at least 2002, PHP has used FlashPixVersion at
                    // https://github.com/php/php-src/blame/master/ext/exif/exif.c#L725
                    case 'FlashPixVersion':
                    // But we can still get the correct casing from
                    // Wikimedia\XMPReader on PDFs
                    case 'FlashpixVersion':
                        $val = $this->literal( (int)$val / 100 );
                        break;
 
                    case 'ColorSpace':
                        $val = match ( $val ) {
                            1, 65535 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'ComponentsConfiguration':
                        $val = match ( $val ) {
                            0, 1, 2, 3, 4, 5, 6 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'DateTime':
                    case 'DateTimeOriginal':
                    case 'DateTimeDigitized':
                    case 'DateTimeReleased':
                    case 'DateTimeExpires':
                    case 'GPSDateStamp':
                    case 'dc-date':
                    case 'DateTimeMetadata':
                    case 'FirstPhotoDate':
                    case 'LastPhotoDate':
                        if ( $val === null ) {
                            // T384879 - we don't need to call literal to turn this into a string, but
                            // we might as well call it for consistency and future proofing of the default value
                            $val = $this->literal( $val );
                            break;
                        }
 
                        if ( $val === '0000:00:00 00:00:00' || $val === '    :  :     :  :  ' ) {
                            $val = $this->msg( 'exif-unknowndate' )->text();
                            break;
                        }
                        if ( preg_match(
                            '/^(?:\d{4}):(?:\d\d):(?:\d\d) (?:\d\d):(?:\d\d):(?:\d\d)$/D',
                            $val
                        ) ) {
                            // Full date.
                            $time = wfTimestamp( TS::MW, $val );
                            if ( $time && (int)$time > 0 ) {
                                $val = $this->getLanguage()->timeanddate( $time );
                                break;
                            }
                        } elseif ( preg_match( '/^(?:\d{4}):(?:\d\d):(?:\d\d) (?:\d\d):(?:\d\d)$/D', $val ) ) {
                            // No second field. Still format the same
                            // since timeanddate doesn't include seconds anyways,
                            // but second still available in api
                            $time = wfTimestamp( TS::MW, $val . ':00' );
                            if ( $time && (int)$time > 0 ) {
                                $val = $this->getLanguage()->timeanddate( $time );
                                break;
                            }
                        } elseif ( preg_match( '/^(?:\d{4}):(?:\d\d):(?:\d\d)$/D', $val ) ) {
                            // If only the date but not the time is filled in.
                            $time = wfTimestamp( TS::MW, substr( $val, 0, 4 )
                                . substr( $val, 5, 2 )
                                . substr( $val, 8, 2 )
                                . '000000' );
                            if ( $time && (int)$time > 0 ) {
                                $val = $this->getLanguage()->date( $time );
                                break;
                            }
                        }
                        // else it will just output $val without formatting it.
                        $val = $this->literal( $val );
                        break;
 
                    case 'ExposureProgram':
                        $val = match ( $val ) {
                            0, 1, 2, 3, 4, 5, 6, 7, 8 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'SubjectDistance':
                        $val = $this->exifMsg( $tag, '', $this->formatNum( $val ) );
                        break;
 
                    case 'MeteringMode':
                        $val = match ( $val ) {
                            0, 1, 2, 3, 4, 5, 6, 7, 255 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'LightSource':
                        $val = match ( $val ) {
                            0, 1, 2, 3, 4, 9, 10, 11, 12, 13, 14, 15, 17, 18, 19, 20, 21, 22, 23, 24,
                            255 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'Flash':
                        if ( !is_int( $val ) ) {
                            $val = 0;
                        }
                        $flashDecode = [
                            'fired' => $val & 0b00000001,
                            'return' => ( $val & 0b00000110 ) >> 1,
                            'mode' => ( $val & 0b00011000 ) >> 3,
                            'function' => ( $val & 0b00100000 ) >> 5,
                            'redeye' => ( $val & 0b01000000 ) >> 6,
                            // 'reserved' => ( $val & 0b10000000 ) >> 7,
                        ];
                        $flashMsgs = [];
                        # We do not need to handle unknown values since all are used.
                        foreach ( $flashDecode as $subTag => $subValue ) {
                            # We do not need any message for zeroed values.
                            if ( $subTag !== 'fired' && $subValue === 0 ) {
                                continue;
                            }
                            $fullTag = $tag . '-' . $subTag;
                            $flashMsgs[] = $this->exifMsg( $fullTag, $subValue );
                        }
                        $val = $this->getLanguage()->commaList( $flashMsgs );
                        break;
 
                    case 'FocalPlaneResolutionUnit':
                        $val = match ( $val ) {
                            2 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'SensingMethod':
                        $val = match ( $val ) {
                            1, 2, 3, 4, 5, 7, 8 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'FileSource':
                        $val = match ( $val ) {
                            3 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'SceneType':
                        $val = match ( $val ) {
                            1 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'CustomRendered':
                        $val = match ( $val ) {
                            0, /* normal */
                            1, /* custom */
                            /* The following are unofficial Apple additions */
                            2, /* HDR (no original saved) */
                            3, /* HDR (original saved) */
                            4, /* Original (for HDR) */
                            6, /* Panorama */
                            7, /* Portrait HDR */
                            8 /* Portrait */ => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'ExposureMode':
                    case 'Contrast':
                    case 'Saturation':
                    case 'Sharpness':
                        $val = match ( $val ) {
                            0, 1, 2 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'WhiteBalance':
                        $val = match ( $val ) {
                            0, 1 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'SceneCaptureType':
                    case 'SubjectDistanceRange':
                        $val = match ( $val ) {
                            0, 1, 2, 3 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'GainControl':
                        $val = match ( $val ) {
                            0, 1, 2, 3, 4 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    // The GPS...Ref values are kept for compatibility, probably won't be reached.
                    case 'GPSLatitudeRef':
                    case 'GPSDestLatitudeRef':
                        $val = match ( $val ) {
                            'N', 'S' => $this->exifMsg( 'GPSLatitude', $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'GPSLongitudeRef':
                    case 'GPSDestLongitudeRef':
                        $val = match ( $val ) {
                            'E', 'W' => $this->exifMsg( 'GPSLongitude', $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'GPSAltitude':
                        if ( $val < 0 ) {
                            $val = $this->exifMsg( 'GPSAltitude', 'below-sealevel', $this->formatNum( -$val, 3 ) );
                        } else {
                            $val = $this->exifMsg( 'GPSAltitude', 'above-sealevel', $this->formatNum( $val, 3 ) );
                        }
                        break;
 
                    case 'GPSStatus':
                        $val = match ( $val ) {
                            'A', 'V' => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'GPSMeasureMode':
                        $val = match ( $val ) {
                            2, 3 => $this->exifMsg( $tag, $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'GPSTrackRef':
                    case 'GPSImgDirectionRef':
                    case 'GPSDestBearingRef':
                        $val = match ( $val ) {
                            'T', 'M' => $this->exifMsg( 'GPSDirection', $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'GPSLatitude':
                    case 'GPSDestLatitude':
                        $val = $this->formatCoords( $val, 'latitude' );
                        break;
                    case 'GPSLongitude':
                    case 'GPSDestLongitude':
                        $val = $this->formatCoords( $val, 'longitude' );
                        break;
 
                    case 'GPSSpeedRef':
                        $val = match ( $val ) {
                            'K', 'M', 'N' => $this->exifMsg( 'GPSSpeed', $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'GPSDestDistanceRef':
                        $val = match ( $val ) {
                            'K', 'M', 'N' => $this->exifMsg( 'GPSDestDistance', $val ),
                            /* If not recognized, display as is. */
                            default => $this->literal( $val )
                        };
                        break;
 
                    case 'GPSDOP':
                        // See https://en.wikipedia.org/wiki/Dilution_of_precision_(GPS)
                        if ( $val <= 2 ) {
                            $val = $this->exifMsg( $tag, 'excellent', $this->formatNum( $val ) );
                        } elseif ( $val <= 5 ) {
                            $val = $this->exifMsg( $tag, 'good', $this->formatNum( $val ) );
                        } elseif ( $val <= 10 ) {
                            $val = $this->exifMsg( $tag, 'moderate', $this->formatNum( $val ) );
                        } elseif ( $val <= 20 ) {
                            $val = $this->exifMsg( $tag, 'fair', $this->formatNum( $val ) );
                        } else {
                            $val = $this->exifMsg( $tag, 'poor', $this->formatNum( $val ) );
                        }
                        break;
 
                    // This is not in the Exif standard, just a special
                    // case for our purposes which enables wikis to wikify
                    // the make, model and software name to link to their articles.
                    case 'Make':
                    case 'Model':
                        $val = $this->exifMsg( $tag, '', $this->literal( $val ) );
                        break;
 
                    case 'Software':
                        if ( is_array( $val ) ) {
                            if ( count( $val ) > 1 ) {
                                // if its a software, version array.
                                $val = $this->msg(
                                    'exif-software-version-value',
                                    $this->literal( $val[0] ),
                                    $this->literal( $val[1] )
                                )->text();
                            } else {
                                // https://phabricator.wikimedia.org/T178130
                                $val = $this->exifMsg( $tag, '', $this->literal( $val[0] ) );
                            }
                        } else {
                            $val = $this->exifMsg( $tag, '', $this->literal( $val ) );
                        }
                        break;
 
                    case 'ExposureTime':
                        // Show the pretty fraction as well as decimal version
                        $val = $this->msg( 'exif-exposuretime-format',
                            $this->formatFraction( $val ), $this->formatNum( $val ) )->text();
                        break;
                    case 'ISOSpeedRatings':
                        // If it's 65535 that means it's at the
                        // limit of the size of Exif::short and
                        // is really higher.
                        if ( $val === '65535' ) {
                            $val = $this->exifMsg( $tag, 'overflow' );
                        } else {
                            $val = $this->formatNum( $val );
                        }
                        break;
                    case 'FNumber':
                        $val = $this->msg( 'exif-fnumber-format',
                            $this->formatNum( $val ) )->text();
                        break;
 
                    case 'FocalLength':
                    case 'FocalLengthIn35mmFilm':
                        $val = $this->msg( 'exif-focallength-format',
                            $this->formatNum( $val ) )->text();
                        break;
 
                    case 'MaxApertureValue':
                        if ( str_contains( $val, '/' ) ) {
                            // need to expand this earlier to calculate fNumber
                            [ $n, $d ] = explode( '/', $val, 2 );
                            if ( is_numeric( $n ) && is_numeric( $d ) ) {
                                $val = (int)$n / (int)$d;
                            }
                        }
                        if ( is_numeric( $val ) ) {
                            $fNumber = 2 ** ( $val / 2 );
                            if ( is_finite( $fNumber ) ) {
                                $val = $this->msg( 'exif-maxaperturevalue-value',
                                    $this->formatNum( $val ),
                                    $this->formatNum( $fNumber, 2 )
                                )->text();
                                break;
                            }
                        }
                        $val = $this->literal( $val );
                        break;
 
                    case 'iimCategory':
                        $val = match ( strtolower( $val ) ) {
                            // See pg 29 of IPTC photo
                            // metadata standard.
                            'ace', 'clj', 'dis', 'fin', 'edu', 'evn', 'hth', 'hum', 'lab', 'lif',
                            'pol', 'rel', 'sci', 'soi', 'spo', 'war',
                            'wea' => $this->exifMsg( 'iimcategory', $val ),
                            default => $this->literal( $val )
                        };
                        break;
                    case 'SubjectNewsCode':
                        // Essentially like iimCategory.
                        // 8 (numeric) digit hierarchical
                        // classification. We decode the
                        // first 2 digits, which provide
                        // a broad category.
                        $val = $this->convertNewsCode( $val );
                        break;
                    case 'Urgency':
                        // 1-8 with 1 being highest, 5 normal
                        // 0 is reserved, and 9 is 'user-defined'.
                        $urgency = '';
                        if ( $val === 0 || $val === 9 ) {
                            $urgency = 'other';
                        } elseif ( $val < 5 && $val > 1 ) {
                            $urgency = 'high';
                        } elseif ( $val === 5 ) {
                            $urgency = 'normal';
                        } elseif ( $val <= 8 && $val > 5 ) {
                            $urgency = 'low';
                        }
 
                        if ( $urgency !== '' ) {
                            $val = $this->exifMsg( 'urgency',
                                $urgency, $this->literal( $val )
                            );
                        } else {
                            $val = $this->literal( $val );
                        }
                        break;
                    case 'DigitalSourceType':
                        // Should be a url starting with
                        // http://cv.iptc.org/newscodes/digitalsourcetype/
                        if ( str_starts_with( $val, 'http://cv.iptc.org/newscodes/digitalsourcetype/' ) ) {
                            $code = substr( $val, 47 );
                            $msg = $this->msg( 'exif-digitalsourcetype-' . strtolower( $code ) );
                            if ( !$msg->isDisabled() ) {
                                $val = $msg->text();
                                break;
                            }
                        }
                        $val = $this->literal( $val );
                        break;
                    // Things that have a unit of pixels.
                    case 'OriginalImageHeight':
                    case 'OriginalImageWidth':
                    case 'PixelXDimension':
                    case 'PixelYDimension':
                    case 'ImageWidth':
                    case 'ImageLength':
                        $val = $this->formatNum( $val ) . ' ' . $this->msg( 'unit-pixel' )->text();
                        break;
 
                    // Do not transform fields with pure text.
                    // For some languages the formatNum()
                    // conversion results to wrong output like
                    // foo,bar@example,com or foo٫bar@example٫com.
                    // Also some 'numeric' things like Scene codes
                    // are included here as we really don't want
                    // commas inserted.
                    case 'ImageDescription':
                    case 'UserComment':
                    case 'Artist':
                    case 'Copyright':
                    case 'RelatedSoundFile':
                    case 'ImageUniqueID':
                    case 'SpectralSensitivity':
                    case 'GPSSatellites':
                    case 'GPSVersionID':
                    case 'GPSMapDatum':
                    case 'Keywords':
                    case 'WorldRegionDest':
                    case 'CountryDest':
                    case 'CountryCodeDest':
                    case 'ProvinceOrStateDest':
                    case 'CityDest':
                    case 'SublocationDest':
                    case 'WorldRegionCreated':
                    case 'CountryCreated':
                    case 'CountryCodeCreated':
                    case 'ProvinceOrStateCreated':
                    case 'CityCreated':
                    case 'SublocationCreated':
                    case 'ObjectName':
                    case 'SpecialInstructions':
                    case 'Headline':
                    case 'Credit':
                    case 'Source':
                    case 'EditStatus':
                    case 'FixtureIdentifier':
                    case 'LocationDest':
                    case 'LocationDestCode':
                    case 'Writer':
                    case 'JPEGFileComment':
                    case 'iimSupplementalCategory':
                    case 'OriginalTransmissionRef':
                    case 'Identifier':
                    case 'dc-contributor':
                    case 'dc-coverage':
                    case 'dc-publisher':
                    case 'dc-relation':
                    case 'dc-rights':
                    case 'dc-source':
                    case 'dc-type':
                    case 'Lens':
                    case 'SerialNumber':
                    case 'CameraOwnerName':
                    case 'Label':
                    case 'Nickname':
                    case 'RightsCertificate':
                    case 'CopyrightOwner':
                    case 'UsageTerms':
                    case 'WebStatement':
                    case 'OriginalDocumentID':
                    case 'LicenseUrl':
                    case 'MorePermissionsUrl':
                    case 'AttributionUrl':
                    case 'PreferredAttributionName':
                    case 'PNGFileComment':
                    case 'Disclaimer':
                    case 'ContentWarning':
                    case 'GIFFileComment':
                    case 'SceneCode':
                    case 'IntellectualGenre':
                    case 'Event':
                    case 'OrganisationInImage':
                    case 'PersonInImage':
                    case 'CaptureSoftware':
                    case 'GPSAreaInformation':
                    case 'GPSProcessingMethod':
                    case 'StitchingSoftware':
                    case 'SubSecTime':
                    case 'SubSecTimeOriginal':
                    case 'SubSecTimeDigitized':
                        $val = $this->literal( $val );
                        break;
 
                    case 'ProjectionType':
                        $val = match ( $val ) {
                            'equirectangular' => $this->exifMsg( $tag, $val ),
                            default => $this->literal( $val )
                        };
                        break;
                    case 'ObjectCycle':
                        $val = match ( $val ) {
                            'a', 'p', 'b' => $this->exifMsg( $tag, $val ),
                            default => $this->literal( $val )
                        };
                        break;
                    case 'Copyrighted':
                    case 'UsePanoramaViewer':
                    case 'ExposureLockUsed':
                        $val = match ( $val ) {
                            'True', 'False' => $this->exifMsg( $tag, $val ),
                            default => $this->literal( $val )
                        };
                        break;
                    case 'Rating':
                        if ( $val === '-1' ) {
                            $val = $this->exifMsg( $tag, 'rejected' );
                        } else {
                            $val = $this->formatNum( $val );
                        }
                        break;
 
                    case 'LanguageCode':
                        $lang = MediaWikiServices::getInstance()
                            ->getLanguageNameUtils()
                            ->getLanguageName( strtolower( $val ), $this->getLanguage()->getCode() );
                        $val = $this->literal( $lang ?: $val );
                        break;
 
                    default:
                        $val = $this->formatNum( $val, false, $tag );
                        break;
                }
            }
            // End formatting values, start flattening arrays.
            $vals = $this->flattenArrayReal( $vals, $type );
        }
 
        return $tags;
    }
 
    public function flattenArrayReal( $vals, $type = 'ul', $noHtml = false ) {
        if ( !is_array( $vals ) ) {
            return $vals; // do nothing if not an array;
        }
 
        if ( isset( $vals['_type'] ) ) {
            $type = $vals['_type'];
            unset( $vals['_type'] );
        }
 
        if ( count( $vals ) === 1 && $type !== 'lang' && isset( $vals[0] ) ) {
            return $vals[0];
        }
        if ( count( $vals ) === 0 ) {
            wfDebug( __METHOD__ . " metadata array with 0 elements!" );
 
            return ""; // paranoia. This should never happen
        }
        // Check if $vals contains nested arrays
        $containsNestedArrays = in_array( true, array_map( 'is_array', $vals ), true );
        if ( $containsNestedArrays ) {
            wfLogWarning( __METHOD__ . ': Invalid $vals, contains nested arrays: ' . json_encode( $vals ) );
        }
 
        /* @todo FIXME: This should hide some of the list entries if there are
         * say more than four. Especially if a field is translated into 20
         * languages, we don't want to show them all by default
         */
        switch ( $type ) {
            case 'lang':
                // Display default, followed by ContentLanguage,
                // followed by the rest in no particular order.
 
                // Todo: hide some items if really long list.
 
                $content = '';
 
                $priorityLanguages = $this->getPriorityLanguages();
                $defaultItem = false;
                $defaultLang = false;
 
                // If default is set, save it for later,
                // as we don't know if it's equal to one of the lang codes.
                // (In xmp you specify the language for a default property by having
                // both a default prop, and one in the language that are identical)
                if ( isset( $vals['x-default'] ) ) {
                    $defaultItem = $vals['x-default'];
                    unset( $vals['x-default'] );
                }
                foreach ( $priorityLanguages as $pLang ) {
                    if ( isset( $vals[$pLang] ) ) {
                        $isDefault = false;
                        if ( $vals[$pLang] === $defaultItem ) {
                            $defaultItem = false;
                            $isDefault = true;
                        }
                        $content .= $this->langItem( $vals[$pLang], $pLang, $isDefault, $noHtml );
 
                        unset( $vals[$pLang] );
 
                        if ( $this->singleLang ) {
                            return Html::rawElement( 'span', [ 'lang' => $pLang ], $vals[$pLang] );
                        }
                    }
                }
 
                // Now do the rest.
                foreach ( $vals as $lang => $item ) {
                    if ( $item === $defaultItem ) {
                        $defaultLang = $lang;
                        continue;
                    }
                    $content .= $this->langItem( $item, $lang, false, $noHtml );
                    if ( $this->singleLang ) {
                        return Html::rawElement( 'span', [ 'lang' => $lang ], $item );
                    }
                }
                if ( $defaultItem !== false ) {
                    $content = $this->langItem( $defaultItem, $defaultLang, true, $noHtml ) . $content;
                    if ( $this->singleLang ) {
                        return $defaultItem;
                    }
                }
                if ( $noHtml ) {
                    return $content;
                }
 
                return '<ul class="metadata-langlist">' . $content . '</ul>';
            case 'ol':
                if ( $noHtml ) {
                    return "\n#" . implode( "\n#", $vals );
                }
 
                return "<ol><li>" . implode( "</li>\n<li>", $vals ) . '</li></ol>';
            case 'ul':
            default:
                if ( $noHtml ) {
                    return "\n*" . implode( "\n*", $vals );
                }
 
                return "<ul><li>" . implode( "</li>\n<li>", $vals ) . '</li></ul>';
        }
    }
 
    private function langItem( $value, $lang, $default = false, $noHtml = false ) {
        if ( $lang === false && $default === false ) {
            throw new InvalidArgumentException( '$lang and $default cannot both be false.' );
        }
 
        if ( $noHtml ) {
            $wrappedValue = $this->literal( $value );
        } else {
            $wrappedValue = '<span class="mw-metadata-lang-value">' . $this->literal( $value ) . '</span>';
        }
 
        if ( $lang === false ) {
            $msg = $this->msg( 'metadata-langitem-default', $wrappedValue );
            if ( $noHtml ) {
                return $msg->text() . "\n\n";
            } /* else */
 
            return '<li class="mw-metadata-lang-default">' . $msg->text() . "</li>\n";
        }
 
        $lowLang = strtolower( $lang );
        $languageNameUtils = MediaWikiServices::getInstance()->getLanguageNameUtils();
        $langName = $languageNameUtils->getLanguageName( $lowLang );
        if ( $langName === '' ) {
            // try just the base language name. (aka en-US -> en ).
            $langPrefix = explode( '-', $lowLang, 2 )[0];
            $langName = $languageNameUtils->getLanguageName( $langPrefix );
            if ( $langName === '' ) {
                // give up.
                $langName = $lang;
            }
        }
        // else we have a language specified
 
        $msg = $this->msg( 'metadata-langitem', $wrappedValue, $langName, $lang );
        if ( $noHtml ) {
            return '*' . $msg->text();
        } /* else: */
 
        $item = '<li class="mw-metadata-lang-code-' . $lang;
        if ( $default ) {
            $item .= ' mw-metadata-lang-default';
        }
        $item .= '" lang="' . $lang . '">';
        $item .= $msg->text();
        $item .= "</li>\n";
 
        return $item;
    }
 
    protected function literal( $val ): string {
        if ( $val === null ) {
            return '';
        }
        // T266707: historically this has used htmlspecialchars to protect
        // the string contents, but it should probably be changed to use
        // wfEscapeWikitext() instead -- however, "we still want to auto-link
        // urls" so wfEscapeWikitext isn't *quite* right...
        return htmlspecialchars( $val );
    }
 
    private function exifMsg( $tag, $val, $arg = null, $arg2 = null ) {
        if ( $val === '' ) {
            $val = 'value';
        }
 
        return $this->msg(
            MediaWikiServices::getInstance()->getContentLanguage()->lc( "exif-$tag-$val" ),
            $arg,
            $arg2
        )->text();
    }
 
    private function formatNum( $num, $round = false, $tagName = null ) {
        $m = [];
        if ( is_array( $num ) ) {
            $out = [];
            foreach ( $num as $number ) {
                $out[] = $this->formatNum( $number, $round, $tagName );
            }
 
            return $this->getLanguage()->commaList( $out );
        }
        if ( is_numeric( $num ) ) {
            if ( $round !== false ) {
                $num = round( $num, $round );
            }
            return $this->getLanguage()->formatNum( $num );
        }
        $num ??= '';
        if ( preg_match( '/^(-?\d+)\/(\d+)$/', $num, $m ) ) {
            if ( $m[2] !== 0 ) {
                $newNum = (int)$m[1] / (int)$m[2];
                if ( $round !== false ) {
                    $newNum = round( $newNum, $round );
                }
            } else {
                $newNum = $num;
            }
 
            return $this->getLanguage()->formatNum( $newNum );
        }
        # T267370: there are a lot of strange EXIF tags floating around.
        LoggerFactory::getInstance( 'formatnum' )->warning(
            'FormatMetadata::formatNum with non-numeric value',
            [
                'tag' => $tagName,
                'value' => $num,
            ]
        );
        return $this->literal( $num );
    }
 
    private function formatFraction( $num ) {
        $m = [];
        $num ??= '';
        if ( preg_match( '/^(-?\d+)\/(\d+)$/', $num, $m ) ) {
            $numerator = (int)$m[1];
            $denominator = (int)$m[2];
            $gcd = $this->gcd( abs( $numerator ), $denominator );
            if ( $gcd !== 0 ) {
                // 0 shouldn't happen! ;)
                return $this->formatNum( $numerator / $gcd ) . '/' . $this->formatNum( $denominator / $gcd );
            }
        }
 
        return $this->formatNum( $num );
    }
 
    private function gcd( $a, $b ) {
        /*
            // https://en.wikipedia.org/wiki/Euclidean_algorithm
            // Recursive form would be:
            if ( $b == 0 )
                return $a;
            else
                return gcd( $b, $a % $b );
        */
        while ( $b != 0 ) {
            $remainder = $a % $b;
 
            // tail recursion...
            $a = $b;
            $b = $remainder;
        }
 
        return $a;
    }
 
    private function convertNewsCode( $val ) {
        if ( !preg_match( '/^\d{8}$/D', $val ) ) {
            // Not a valid news code.
            return $val;
        }
        $cat = match ( substr( $val, 0, 2 ) ) {
            '01' => 'ace',
            '02' => 'clj',
            '03' => 'dis',
            '04' => 'fin',
            '05' => 'edu',
            '06' => 'evn',
            '07' => 'hth',
            '08' => 'hum',
            '09' => 'lab',
            '10' => 'lif',
            '11' => 'pol',
            '12' => 'rel',
            '13' => 'sci',
            '14' => 'soi',
            '15' => 'spo',
            '16' => 'war',
            '17' => 'wea',
            default => '',
        };
        if ( $cat !== '' ) {
            $catMsg = $this->exifMsg( 'iimcategory', $cat );
            $val = $this->exifMsg( 'subjectnewscode', '', $this->literal( $val ), $catMsg );
        }
 
        return $val;
    }
 
    private function formatCoords( $coord, string $type ) {
        if ( !is_numeric( $coord ) ) {
            wfDebugLog( 'exif', __METHOD__ . ": \"$coord\" is not a number" );
            return $this->literal( (string)$coord );
        }
 
        $ref = '';
        if ( $coord < 0 ) {
            $nCoord = -$coord;
            if ( $type === 'latitude' ) {
                $ref = 'S';
            } elseif ( $type === 'longitude' ) {
                $ref = 'W';
            }
        } else {
            $nCoord = (float)$coord;
            if ( $type === 'latitude' ) {
                $ref = 'N';
            } elseif ( $type === 'longitude' ) {
                $ref = 'E';
            }
        }
 
        $deg = floor( $nCoord );
        $min = floor( ( $nCoord - $deg ) * 60 );
        $sec = round( ( ( $nCoord - $deg ) * 60 - $min ) * 60, 2 );
 
        $deg = $this->formatNum( $deg );
        $min = $this->formatNum( $min );
        $sec = $this->formatNum( $sec );
 
        // Note the default message "$1° $2′ $3″ $4" ignores the 5th parameter
        return $this->msg( 'exif-coordinate-format', $deg, $min, $sec, $ref, $this->literal( $coord ) )->text();
    }
 
    public function collapseContactInfo( array $vals ) {
        if ( !( isset( $vals['CiAdrExtadr'] )
            || isset( $vals['CiAdrCity'] )
            || isset( $vals['CiAdrCtry'] )
            || isset( $vals['CiEmailWork'] )
            || isset( $vals['CiTelWork'] )
            || isset( $vals['CiAdrPcode'] )
            || isset( $vals['CiAdrRegion'] )
            || isset( $vals['CiUrlWork'] )
        ) ) {
            // We don't have any sub-properties
            // This could happen if its using old
            // iptc that just had this as a free-form
            // text value.
            // Note: people often insert >, etc into
            // the metadata which should not be interpreted
            // but we still want to auto-link urls.
            foreach ( $vals as &$val ) {
                $val = $this->literal( $val );
            }
 
            return $this->flattenArrayReal( $vals );
        }
 
        // We have a real ContactInfo field.
        // Its unclear if all these fields have to be
        // set, so assume they do not.
        $url = $tel = $street = $city = $country = '';
        $email = $postal = $region = '';
 
        // Also note, some of the class names this uses
        // are similar to those used by hCard. This is
        // mostly because they're sensible names. This
        // does not (and does not attempt to) output
        // stuff in the hCard microformat. However it
        // might output in the adr microformat.
 
        if ( isset( $vals['CiAdrExtadr'] ) ) {
            // Todo: This can potentially be multi-line.
            // Need to check how that works in XMP.
            $street = '<span class="extended-address">'
                . $this->literal(
                    $vals['CiAdrExtadr'] )
                . '</span>';
        }
        if ( isset( $vals['CiAdrCity'] ) ) {
            $city = '<span class="locality">'
                . $this->literal( $vals['CiAdrCity'] )
                . '</span>';
        }
        if ( isset( $vals['CiAdrCtry'] ) ) {
            $country = '<span class="country-name">'
                . $this->literal( $vals['CiAdrCtry'] )
                . '</span>';
        }
        if ( isset( $vals['CiEmailWork'] ) ) {
            $emails = [];
            // Have to split multiple emails at commas/new lines.
            $splitEmails = explode( "\n", $vals['CiEmailWork'] );
            foreach ( $splitEmails as $e1 ) {
                // Also split on comma
                foreach ( explode( ',', $e1 ) as $e2 ) {
                    $finalEmail = trim( $e2 );
                    if ( $finalEmail === ',' || $finalEmail === '' ) {
                        continue;
                    }
                    if ( str_contains( $finalEmail, '<' ) ) {
                        // Don't do fancy formatting to
                        // "My name" <foo@bar.com> style stuff
                        $emails[] = $this->literal( $finalEmail );
                    } else {
                        $emails[] = '[mailto:'
                            . $finalEmail
                            . ' <span class="email">'
                            . $this->literal( $finalEmail )
                            . '</span>]';
                    }
                }
            }
            $email = implode( ', ', $emails );
        }
        if ( isset( $vals['CiTelWork'] ) ) {
            $tel = '<span class="tel">'
                . $this->literal( $vals['CiTelWork'] )
                . '</span>';
        }
        if ( isset( $vals['CiAdrPcode'] ) ) {
            $postal = '<span class="postal-code">'
                . $this->literal( $vals['CiAdrPcode'] )
                . '</span>';
        }
        if ( isset( $vals['CiAdrRegion'] ) ) {
            // Note this is province/state.
            $region = '<span class="region">'
                . $this->literal( $vals['CiAdrRegion'] )
                . '</span>';
        }
        if ( isset( $vals['CiUrlWork'] ) ) {
            $url = '<span class="url">'
                . $this->literal( $vals['CiUrlWork'] )
                . '</span>';
        }
 
        return $this->msg( 'exif-contact-value', $email, $url,
            $street, $city, $region, $postal, $country, $tel )->text();
    }
 
    public static function getVisibleFields() {
        $fields = [];
        $lines = explode( "\n", wfMessage( 'metadata-fields' )->inContentLanguage()->text() );
        foreach ( $lines as $line ) {
            $matches = [];
            if ( preg_match( '/^\\*\s*(.*?)\s*$/', $line, $matches ) ) {
                $fields[] = $matches[1];
            }
        }
        $fields = array_map( 'strtolower', $fields );
 
        return $fields;
    }
 
    public function fetchExtendedMetadata( File $file ) {
        $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
 
        // If revision deleted, exit immediately
        if ( $file->isDeleted( File::DELETED_FILE ) ) {
            return [];
        }
 
        $cacheKey = $cache->makeKey(
            'getExtendedMetadata',
            $this->getLanguage()->getCode(),
            (int)$this->singleLang,
            $file->getSha1()
        );
        $maxCacheTime = ( $file instanceof ForeignAPIFile ) ? 60 * 60 * 12 : 60 * 60 * 24 * 30;
 
        $cachedValue = $cache->getWithSetCallback(
            $cacheKey,
            $maxCacheTime,
            function () use ( $file ) {
                $fileMetadata = $this->getExtendedMetadataFromFile( $file );
                $extendedMetadata = $this->getExtendedMetadataFromHook( $file, $fileMetadata, $maxCacheTime );
                if ( $this->singleLang ) {
                    $this->resolveMultilangMetadata( $extendedMetadata );
                }
                $this->discardMultipleValues( $extendedMetadata );
                // Make sure the metadata won't break the API when an XML format is used.
                // This is an API-specific function so it would be cleaner to call it from
                // outside fetchExtendedMetadata, but this way we don't need to redo the
                // computation on a cache hit.
                $this->sanitizeArrayForAPI( $extendedMetadata );
 
                return [ 'data' => $extendedMetadata, 'timestamp' => wfTimestampNow() ];
            },
            [
                'touchedCallback' => function ( $value ) use ( $file ) {
                    if (
                        !$this->getHookRunner()->onValidateExtendedMetadataCache( $value['timestamp'], $file )
                    ) {
                        // Reject cache and regenerate
                        return time();
                    }
                    return null;
                }
            ]
        );
 
        return $cachedValue['data'];
    }
 
    protected function getExtendedMetadataFromFile( File $file ) {
        // If this is a remote file accessed via an API request, we already
        // have remote metadata so we just ignore any local one
        if ( $file instanceof ForeignAPIFile ) {
            // In case of error we pretend no metadata - this will get cached.
            // Might or might not be a good idea.
            return $file->getExtendedMetadata() ?: [];
        }
 
        $uploadDate = wfTimestamp( TS::ISO_8601, $file->getTimestamp() );
 
        $fileMetadata = [
            // This is modification time, which is close to "upload" time.
            'DateTime' => [
                'value' => $uploadDate,
                'source' => 'mediawiki-metadata',
            ],
        ];
 
        $title = $file->getTitle();
        if ( $title ) {
            $text = $title->getText();
            $pos = strrpos( $text, '.' );
 
            if ( $pos ) {
                $name = substr( $text, 0, $pos );
            } else {
                $name = $text;
            }
 
            $fileMetadata['ObjectName'] = [
                'value' => $name,
                'source' => 'mediawiki-metadata',
            ];
        }
 
        return $fileMetadata;
    }
 
    protected function getExtendedMetadataFromHook( File $file, array $extendedMetadata,
        &$maxCacheTime
    ) {
        $this->getHookRunner()->onGetExtendedMetadata(
            $extendedMetadata,
            $file,
            $this->getContext(),
            $this->singleLang,
            $maxCacheTime
        );
 
        $visible = array_fill_keys( self::getVisibleFields(), true );
        foreach ( $extendedMetadata as $key => $value ) {
            if ( !isset( $visible[strtolower( $key )] ) ) {
                $extendedMetadata[$key]['hidden'] = '';
            }
        }
 
        return $extendedMetadata;
    }
 
    protected function resolveMultilangValue( $value ) {
        if (
            !is_array( $value )
            || !isset( $value['_type'] )
            || $value['_type'] !== 'lang'
        ) {
            return $value; // do nothing if not a multilang array
        }
 
        // choose the language best matching user or site settings
        $priorityLanguages = $this->getPriorityLanguages();
        foreach ( $priorityLanguages as $lang ) {
            if ( isset( $value[$lang] ) ) {
                return $value[$lang];
            }
        }
 
        // otherwise go with the default language, if set
        if ( isset( $value['x-default'] ) ) {
            return $value['x-default'];
        }
 
        // otherwise just return any one language
        unset( $value['_type'] );
        if ( $value ) {
            return reset( $value );
        }
 
        // this should not happen; signal error
        return null;
    }
 
    protected function resolveMultivalueValue( $value ) {
        if ( !is_array( $value ) ) {
            return $value;
        }
        if ( isset( $value['_type'] ) && $value['_type'] === 'lang' ) {
            // if this is a multilang array, process fields separately
            $newValue = [];
            foreach ( $value as $k => $v ) {
                $newValue[$k] = $this->resolveMultivalueValue( $v );
            }
            return $newValue;
        }
        // _type is 'ul' or 'ol' or missing in which case it defaults to 'ul'
        $v = reset( $value );
        if ( key( $value ) === '_type' ) {
            $v = next( $value );
        }
        return $v;
    }
 
    protected function resolveMultilangMetadata( &$metadata ) {
        if ( !is_array( $metadata ) ) {
            return;
        }
        foreach ( $metadata as &$field ) {
            if ( isset( $field['value'] ) ) {
                $field['value'] = $this->resolveMultilangValue( $field['value'] );
            }
        }
    }
 
    protected function discardMultipleValues( &$metadata ) {
        if ( !is_array( $metadata ) ) {
            return;
        }
        foreach ( $metadata as $key => &$field ) {
            if ( $key === 'Software' || $key === 'Contact' ) {
                // we skip some fields which have composite values. They are not particularly interesting
                // and you can get them via the metadata / commonmetadata APIs anyway.
                continue;
            }
            if ( isset( $field['value'] ) ) {
                $field['value'] = $this->resolveMultivalueValue( $field['value'] );
            }
        }
    }
 
    protected function sanitizeArrayForAPI( &$arr ) {
        if ( !is_array( $arr ) ) {
            return;
        }
 
        $counter = 1;
        foreach ( $arr as $key => &$value ) {
            $sanitizedKey = $this->sanitizeKeyForAPI( $key );
            if ( $sanitizedKey !== $key ) {
                if ( isset( $arr[$sanitizedKey] ) ) {
                    // Make the sanitized keys hopefully unique.
                    // To make it definitely unique would be too much effort, given that
                    // sanitizing is only needed for misformatted metadata anyway, but
                    // this at least covers the case when $arr is numeric.
                    $sanitizedKey .= $counter;
                    ++$counter;
                }
                $arr[$sanitizedKey] = $arr[$key];
                unset( $arr[$key] );
            }
            if ( is_array( $value ) ) {
                $this->sanitizeArrayForAPI( $value );
            }
        }
        unset( $value );
 
        // Handle API metadata keys (particularly "_type")
        $keys = array_filter( array_keys( $arr ), ApiResult::isMetadataKey( ... ) );
        if ( $keys ) {
            ApiResult::setPreserveKeysList( $arr, $keys );
        }
    }
 
    protected function sanitizeKeyForAPI( $key ) {
        // drop all characters which are not valid in an XML tag name
        // a bunch of non-ASCII letters would be valid but probably won't
        // be used so we take the easy way
        $key = preg_replace( '/[^a-zA-Z0-9_:.\-]/', '', $key );
        // drop characters which are invalid at the first position
        $key = preg_replace( '/^[\d\-.]+/', '', $key );
 
        if ( $key === '' ) {
            $key = '_';
        // special case for an internal keyword
        } elseif ( $key === '_element' ) {
            $key = 'element';
        }
 
        return $key;
    }
 
    protected function getPriorityLanguages() {
        $priorityLanguages = MediaWikiServices::getInstance()
            ->getLanguageFallback()
            ->getAllIncludingSiteLanguage( $this->getLanguage()->getCode() );
        $priorityLanguages = array_merge(
            (array)$this->getLanguage()->getCode(),
            $priorityLanguages[0],
            $priorityLanguages[1]
        );
 
        return $priorityLanguages;
    }
}
 
class_alias( FormatMetadata::class, 'FormatMetadata' );