HTTPFileStreamer.php

Go to the documentation of this file.

<?php
namespace Wikimedia\FileBackend;
 
use Wikimedia\Http\HttpStatus;
use Wikimedia\Timestamp\ConvertibleTimestamp;
use Wikimedia\Timestamp\TimestampFormat as TS;
 
class HTTPFileStreamer {
    protected $path;
    protected $obResetFunc;
    protected $streamMimeFunc;
    protected $headerFunc;
 
    // Do not send any HTTP headers (i.e. body only)
    public const STREAM_HEADLESS = 1;
    // Do not try to tear down any PHP output buffers
    public const STREAM_ALLOW_OB = 2;
 
    public static function preprocessHeaders( $headers ) {
        $rawHeaders = [];
        $optHeaders = [];
        foreach ( $headers as $name => $header ) {
            $nameLower = strtolower( $name );
            if ( in_array( $nameLower, [ 'range', 'if-modified-since' ], true ) ) {
                $optHeaders[$nameLower] = $header;
            } else {
                $rawHeaders[] = "$name: $header";
            }
        }
        return [ $rawHeaders, $optHeaders ];
    }
 
    public function __construct( $path, array $params = [] ) {
        $this->path = $path;
        $this->obResetFunc = $params['obResetFunc'] ?? self::resetOutputBuffers( ... );
        $this->streamMimeFunc = $params['streamMimeFunc'] ?? self::contentTypeFromPath( ... );
        $this->headerFunc = $params['headerFunc'] ?? header( ... );
    }
 
    public function stream(
        $headers = [], $sendErrors = true, $optHeaders = [], $flags = 0
    ) {
        $headless = ( $flags & self::STREAM_HEADLESS );
 
        // Don't stream it out as text/html if there was a PHP error
        if ( $headers && headers_sent() ) {
            echo "Headers already sent, terminating.\n";
            return false;
        }
 
        $headerFunc = $headless
            ? static function ( $header ) {
                // no-op
            }
            : $this->header( ... );
 
        // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged
        $info = @stat( $this->path );
 
        if ( !is_array( $info ) ) {
            if ( $sendErrors ) {
                self::send404Message( $this->path, $flags );
            }
            return false;
        }
 
        // Send Last-Modified HTTP header for client-side caching
        $mtimeCT = new ConvertibleTimestamp( $info['mtime'] );
        $headerFunc( 'Last-Modified: ' . $mtimeCT->getTimestamp( TS::RFC2822 ) );
 
        if ( ( $flags & self::STREAM_ALLOW_OB ) == 0 ) {
            ( $this->obResetFunc )();
        }
 
        $type = ( $this->streamMimeFunc )( $this->path );
        if ( $type && $type != 'unknown/unknown' ) {
            $headerFunc( "Content-type: $type" );
        } else {
            // Send a content type which is not known to Internet Explorer, to
            // avoid triggering IE's content type detection. Sending a standard
            // unknown content type here essentially gives IE license to apply
            // whatever content type it likes.
            $headerFunc( 'Content-type: application/x-wiki' );
        }
 
        // Don't send if client has up to date cache
        if ( isset( $optHeaders['if-modified-since'] ) ) {
            $modsince = preg_replace( '/;.*$/', '', $optHeaders['if-modified-since'] );
            if ( $mtimeCT->getTimestamp( TS::UNIX ) <= strtotime( $modsince ) ) {
                ini_set( 'zlib.output_compression', 0 );
                $headerFunc( 304 );
                return true; // ok
            }
        }
 
        // Send additional headers
        foreach ( $headers as $header ) {
            $headerFunc( $header );
        }
 
        if ( isset( $optHeaders['range'] ) ) {
            $range = self::parseRange( $optHeaders['range'], $info['size'] );
            if ( is_array( $range ) ) {
                $headerFunc( 206 );
                $headerFunc( 'Content-Length: ' . $range[2] );
                $headerFunc( "Content-Range: bytes {$range[0]}-{$range[1]}/{$info['size']}" );
            } elseif ( $range === 'invalid' ) {
                if ( $sendErrors ) {
                    $headerFunc( 416 );
                    $headerFunc( 'Cache-Control: no-cache' );
                    $headerFunc( 'Content-Type: text/html; charset=utf-8' );
                    $headerFunc( 'Content-Range: bytes */' . $info['size'] );
                }
                return false;
            } else { // unsupported Range request (e.g. multiple ranges)
                $range = null;
                $headerFunc( 'Content-Length: ' . $info['size'] );
            }
        } else {
            $range = null;
            $headerFunc( 'Content-Length: ' . $info['size'] );
        }
 
        if ( is_array( $range ) ) {
            $handle = fopen( $this->path, 'rb' );
            if ( $handle ) {
                $ok = true;
                fseek( $handle, $range[0] );
                $remaining = $range[2];
                while ( $remaining > 0 && $ok ) {
                    $bytes = min( $remaining, 8 * 1024 );
                    $data = fread( $handle, $bytes );
                    $remaining -= $bytes;
                    $ok = ( $data !== false );
                    print $data;
                }
            } else {
                return false;
            }
        } else {
            return readfile( $this->path ) !== false; // faster
        }
 
        return true;
    }
 
    public static function send404Message( $fname, $flags = 0 ) {
        if ( ( $flags & self::STREAM_HEADLESS ) == 0 ) {
            HttpStatus::header( 404 );
            header( 'Cache-Control: no-cache' );
            header( 'Content-Type: text/html; charset=utf-8' );
        }
        $encFile = htmlspecialchars( $fname );
        $encScript = htmlspecialchars( $_SERVER['SCRIPT_NAME'] );
        echo "<!DOCTYPE html><html><body>
            <h1>File not found</h1>
            <p>Although this PHP script ($encScript) exists, the file requested for output
            ($encFile) does not.</p>
            </body></html>
            ";
    }
 
    public static function parseRange( $range, $size ) {
        $m = [];
        if ( preg_match( '#^bytes=(\d*)-(\d*)$#', $range, $m ) ) {
            [ , $start, $end ] = $m;
            if ( $start === '' && $end === '' ) {
                $absRange = [ 0, $size - 1 ];
            } elseif ( $start === '' ) {
                $absRange = [ $size - (int)$end, $size - 1 ];
            } elseif ( $end === '' ) {
                $absRange = [ (int)$start, $size - 1 ];
            } else {
                $absRange = [ (int)$start, (int)$end ];
            }
            if ( $absRange[0] >= 0 && $absRange[1] >= $absRange[0] ) {
                if ( $absRange[0] < $size ) {
                    $absRange[1] = min( $absRange[1], $size - 1 ); // stop at EOF
                    $absRange[2] = $absRange[1] - $absRange[0] + 1;
                    return $absRange;
                } elseif ( $absRange[0] == 0 && $size == 0 ) {
                    return 'unrecognized'; // the whole file should just be sent
                }
            }
            return 'invalid';
        }
        return 'unrecognized';
    }
 
    protected static function resetOutputBuffers() {
        while ( ob_get_status() ) {
            if ( !ob_end_clean() ) {
                // Could not remove output buffer handler; abort now
                // to avoid getting in some kind of infinite loop.
                break;
            }
        }
    }
 
    protected static function contentTypeFromPath( string $filename ): string {
        $ext = strtolower( pathinfo( $filename, PATHINFO_EXTENSION ) );
        return match ( $ext ) {
            'gif' => 'image/gif',
            'png' => 'image/png',
            'jpg',
            'jpeg' => 'image/jpeg',
            'webp' => 'image/webp',
            default => 'unknown/unknown',
        };
    }
 
    private function header( $header ) {
        if ( is_int( $header ) ) {
            $header = HttpStatus::getHeader( $header );
        }
 
        ( $this->headerFunc )( $header );
    }
}
 
class_alias( HTTPFileStreamer::class, 'HTTPFileStreamer' );