XMPValidate.php

Go to the documentation of this file.

<?php
use Psr\Log\LoggerInterface;
use Psr\Log\LoggerAwareInterface;
use Wikimedia\Timestamp\ConvertibleTimestamp;
 
class XMPValidate implements LoggerAwareInterface {
 
    private $logger;
 
    public function __construct( LoggerInterface $logger ) {
        $this->setLogger( $logger );
    }
 
    public function setLogger( LoggerInterface $logger ) {
        $this->logger = $logger;
    }
    public function validateBoolean( $info, &$val, $standalone ) {
        if ( !$standalone ) {
            // this only validates standalone properties, not arrays, etc
            return;
        }
        if ( $val !== 'True' && $val !== 'False' ) {
            $this->logger->info( __METHOD__ . " Expected True or False but got $val" );
            $val = null;
        }
    }
 
    public function validateRational( $info, &$val, $standalone ) {
        if ( !$standalone ) {
            // this only validates standalone properties, not arrays, etc
            return;
        }
        if ( !preg_match( '/^(?:-?\d+)\/(?:\d+[1-9]|[1-9]\d*)$/D', $val ) ) {
            $this->logger->info( __METHOD__ . " Expected rational but got $val" );
            $val = null;
        }
    }
 
    public function validateRating( $info, &$val, $standalone ) {
        if ( !$standalone ) {
            // this only validates standalone properties, not arrays, etc
            return;
        }
        if ( !preg_match( '/^[-+]?\d*(?:\.?\d*)$/D', $val )
            || !is_numeric( $val )
        ) {
            $this->logger->info( __METHOD__ . " Expected rating but got $val" );
            $val = null;
 
            return;
        } else {
            $nVal = (float)$val;
            if ( $nVal < 0 ) {
                // We do < 0 here instead of < -1 here, since
                // the values between 0 and -1 are also illegal
                // as -1 is meant as a special reject rating.
                $this->logger->info( __METHOD__ . " Rating too low, setting to -1 (Rejected)" );
                $val = '-1';
 
                return;
            }
            if ( $nVal > 5 ) {
                $this->logger->info( __METHOD__ . " Rating too high, setting to 5" );
                $val = '5';
 
                return;
            }
        }
    }
 
    public function validateInteger( $info, &$val, $standalone ) {
        if ( !$standalone ) {
            // this only validates standalone properties, not arrays, etc
            return;
        }
        if ( !preg_match( '/^[-+]?\d+$/D', $val ) ) {
            $this->logger->info( __METHOD__ . " Expected integer but got $val" );
            $val = null;
        }
    }
 
    public function validateClosed( $info, &$val, $standalone ) {
        if ( !$standalone ) {
            // this only validates standalone properties, not arrays, etc
            return;
        }
 
        // check if its in a numeric range
        $inRange = false;
        if ( isset( $info['rangeLow'] )
            && isset( $info['rangeHigh'] )
            && is_numeric( $val )
            && ( intval( $val ) <= $info['rangeHigh'] )
            && ( intval( $val ) >= $info['rangeLow'] )
        ) {
            $inRange = true;
        }
 
        if ( !isset( $info['choices'][$val] ) && !$inRange ) {
            $this->logger->info( __METHOD__ . " Expected closed choice, but got $val" );
            $val = null;
        }
    }
 
    public function validateFlash( $info, &$val, $standalone ) {
        if ( $standalone ) {
            // this only validates flash structs, not individual properties
            return;
        }
        if ( !( isset( $val['Fired'] )
            && isset( $val['Function'] )
            && isset( $val['Mode'] )
            && isset( $val['RedEyeMode'] )
            && isset( $val['Return'] )
        ) ) {
            $this->logger->info( __METHOD__ . " Flash structure did not have all the required components" );
            $val = null;
        } else {
            $val = ( 0 | ( $val['Fired'] === 'True' )
                | ( intval( $val['Return'] ) << 1 )
                | ( intval( $val['Mode'] ) << 3 )
                | ( ( $val['Function'] === 'True' ) << 5 )
                | ( ( $val['RedEyeMode'] === 'True' ) << 6 ) );
        }
    }
 
    public function validateLangCode( $info, &$val, $standalone ) {
        if ( !$standalone ) {
            // this only validates standalone properties, not arrays, etc
            return;
        }
        if ( !preg_match( '/^[-A-Za-z0-9]{2,}$/D', $val ) ) {
            // this is a rather naive check.
            $this->logger->info( __METHOD__ . " Expected Lang code but got $val" );
            $val = null;
        }
    }
 
    public function validateDate( $info, &$val, $standalone ) {
        if ( !$standalone ) {
            // this only validates standalone properties, not arrays, etc
            return;
        }
        $res = [];
        if ( !preg_match(
            /* ahh! scary regex... */
            // phpcs:ignore Generic.Files.LineLength
            '/^([0-3]\d{3})(?:-([01]\d)(?:-([0-3]\d)(?:T([0-2]\d):([0-6]\d)(?::([0-6]\d)(?:\.\d+)?)?([-+]\d{2}:\d{2}|Z)?)?)?)?$/D',
            $val, $res )
        ) {
            $this->logger->info( __METHOD__ . " Expected date but got $val" );
            $val = null;
        } else {
            /*
             * $res is formatted as follows:
             * 0 -> full date.
             * 1 -> year, 2-> month, 3-> day, 4-> hour, 5-> minute, 6->second
             * 7-> Timezone specifier (Z or something like +12:30 )
             * many parts are optional, some aren't. For example if you specify
             * minute, you must specify hour, day, month, and year but not second or TZ.
             */
 
            /*
             * First of all, if year = 0000, Something is wrongish,
             * so don't extract. This seems to happen when
             * some programs convert between metadata formats.
             */
            if ( $res[1] === '0000' ) {
                $this->logger->info( __METHOD__ . " Invalid date (year 0): $val" );
                $val = null;
 
                return;
            }
 
            if ( !isset( $res[4] ) ) { // hour
                // just have the year month day (if that)
                $val = $res[1];
                if ( isset( $res[2] ) ) {
                    $val .= ':' . $res[2];
                }
                if ( isset( $res[3] ) ) {
                    $val .= ':' . $res[3];
                }
 
                return;
            }
 
            if ( !isset( $res[7] ) || $res[7] === 'Z' ) {
                // if hour is set, then minute must also be or regex above will fail.
                $val = $res[1] . ':' . $res[2] . ':' . $res[3]
                    . ' ' . $res[4] . ':' . $res[5];
                if ( isset( $res[6] ) && $res[6] !== '' ) {
                    $val .= ':' . $res[6];
                }
 
                return;
            }
 
            // Extra check for empty string necessary due to TZ but no second case.
            $stripSeconds = false;
            if ( !isset( $res[6] ) || $res[6] === '' ) {
                $res[6] = '00';
                $stripSeconds = true;
            }
 
            // Do timezone processing. We've already done the case that tz = Z.
 
            // We know that if we got to this step, year, month day hour and min must be set
            // by virtue of regex not failing.
 
            $unix = ConvertibleTimestamp::convert( TS_UNIX,
                $res[1] . $res[2] . $res[3] . $res[4] . $res[5] . $res[6]
            );
            $offset = intval( substr( $res[7], 1, 2 ) ) * 60 * 60;
            $offset += intval( substr( $res[7], 4, 2 ) ) * 60;
            if ( substr( $res[7], 0, 1 ) === '-' ) {
                $offset = -$offset;
            }
            $val = ConvertibleTimestamp::convert( TS_EXIF, $unix + $offset );
 
            if ( $stripSeconds ) {
                // If seconds weren't specified, remove the trailing ':00'.
                $val = substr( $val, 0, -3 );
            }
        }
    }
 
    public function validateGPS( $info, &$val, $standalone ) {
        if ( !$standalone ) {
            return;
        }
 
        $m = [];
        if ( preg_match(
            '/(\d{1,3}),(\d{1,2}),(\d{1,2})([NWSE])/D',
            $val, $m )
        ) {
            $coord = intval( $m[1] );
            $coord += intval( $m[2] ) * ( 1 / 60 );
            $coord += intval( $m[3] ) * ( 1 / 3600 );
            if ( $m[4] === 'S' || $m[4] === 'W' ) {
                $coord = -$coord;
            }
            $val = $coord;
 
            return;
        } elseif ( preg_match(
            '/(\d{1,3}),(\d{1,2}(?:.\d*)?)([NWSE])/D',
            $val, $m )
        ) {
            $coord = intval( $m[1] );
            $coord += floatval( $m[2] ) * ( 1 / 60 );
            if ( $m[3] === 'S' || $m[3] === 'W' ) {
                $coord = -$coord;
            }
            $val = $coord;
 
            return;
        } else {
            $this->logger->info( __METHOD__
                . " Expected GPSCoordinate, but got $val." );
            $val = null;
 
            return;
        }
    }
}