DiffHistoryBlob.php

Go to the documentation of this file.

<?php
class DiffHistoryBlob implements HistoryBlob {
    public $mItems = [];
 
    public $mSize = 0;
 
    public $mDiffs;
 
    public $mDiffMap;
 
    public $mDefaultKey;
 
    public $mCompressed;
 
    public $mFrozen = false;
 
    public $mMaxSize = 10000000;
 
    public $mMaxCount = 100;
 
    private const XDL_BDOP_INS = 1;
    private const XDL_BDOP_CPY = 2;
    private const XDL_BDOP_INSB = 3;
 
    public function __construct() {
        if ( !function_exists( 'gzdeflate' ) ) {
            throw new MWException( "Need zlib support to read or write DiffHistoryBlob\n" );
        }
    }
 
    public function addItem( $text ) {
        if ( $this->mFrozen ) {
            throw new MWException( __METHOD__ . ": Cannot add more items after sleep/wakeup" );
        }
 
        $this->mItems[] = $text;
        $this->mSize += strlen( $text );
        $this->mDiffs = null; // later
        return count( $this->mItems ) - 1;
    }
 
    public function getItem( $key ) {
        return $this->mItems[$key];
    }
 
    public function setText( $text ) {
        $this->mDefaultKey = $this->addItem( $text );
    }
 
    public function getText() {
        return $this->getItem( $this->mDefaultKey );
    }
 
    private function compress() {
        if ( !function_exists( 'xdiff_string_rabdiff' ) ) {
            throw new MWException( "Need xdiff 1.5+ support to write DiffHistoryBlob\n" );
        }
        if ( isset( $this->mDiffs ) ) {
            // Already compressed
            return;
        }
        if ( $this->mItems === [] ) {
            return;
        }
 
        // Create two diff sequences: one for main text and one for small text
        $sequences = [
            'small' => [
                'tail' => '',
                'diffs' => [],
                'map' => [],
            ],
            'main' => [
                'tail' => '',
                'diffs' => [],
                'map' => [],
            ],
        ];
        $smallFactor = 0.5;
 
        $mItemsCount = count( $this->mItems );
        for ( $i = 0; $i < $mItemsCount; $i++ ) {
            $text = $this->mItems[$i];
            if ( $i == 0 ) {
                $seqName = 'main';
            } else {
                $mainTail = $sequences['main']['tail'];
                if ( strlen( $text ) < strlen( $mainTail ) * $smallFactor ) {
                    $seqName = 'small';
                } else {
                    $seqName = 'main';
                }
            }
 
            $tail = $sequences[$seqName]['tail'];
            $diff = $this->diff( $tail, $text );
            $sequences[$seqName]['diffs'][] = $diff;
            $sequences[$seqName]['map'][] = $i;
            $sequences[$seqName]['tail'] = $text;
        }
 
        // Knit the sequences together
        $tail = '';
        $this->mDiffs = [];
        $this->mDiffMap = [];
        foreach ( $sequences as $seq ) {
            if ( $seq['diffs'] === [] ) {
                continue;
            }
            if ( $tail === '' ) {
                $this->mDiffs[] = $seq['diffs'][0];
            } else {
                $head = $this->patch( '', $seq['diffs'][0] );
                $this->mDiffs[] = $this->diff( $tail, $head );
            }
            $this->mDiffMap[] = $seq['map'][0];
            $diffsCount = count( $seq['diffs'] );
            for ( $i = 1; $i < $diffsCount; $i++ ) {
                $this->mDiffs[] = $seq['diffs'][$i];
                $this->mDiffMap[] = $seq['map'][$i];
            }
            $tail = $seq['tail'];
        }
    }
 
    private function diff( $t1, $t2 ) {
        # Need to do a null concatenation with warnings off, due to bugs in the current version of xdiff
        # "String is not zero-terminated"
        Wikimedia\suppressWarnings();
        $diff = xdiff_string_rabdiff( $t1, $t2 ) . '';
        Wikimedia\restoreWarnings();
        return $diff;
    }
 
    private function patch( $base, $diff ) {
        if ( function_exists( 'xdiff_string_bpatch' ) ) {
            Wikimedia\suppressWarnings();
            $text = xdiff_string_bpatch( $base, $diff ) . '';
            Wikimedia\restoreWarnings();
            return $text;
        }
 
        # Pure PHP implementation
 
        $header = unpack( 'Vofp/Vcsize', substr( $diff, 0, 8 ) );
 
        # Check the checksum if hash extension is available
        $ofp = $this->xdiffAdler32( $base );
        if ( $ofp !== false && $ofp !== substr( $diff, 0, 4 ) ) {
            wfDebug( __METHOD__ . ": incorrect base checksum" );
            return false;
        }
        if ( $header['csize'] != strlen( $base ) ) {
            wfDebug( __METHOD__ . ": incorrect base length" );
            return false;
        }
 
        $p = 8;
        $out = '';
        while ( $p < strlen( $diff ) ) {
            $x = unpack( 'Cop', substr( $diff, $p, 1 ) );
            $op = $x['op'];
            ++$p;
            switch ( $op ) {
                case self::XDL_BDOP_INS:
                    $x = unpack( 'Csize', substr( $diff, $p, 1 ) );
                    $p++;
                    $out .= substr( $diff, $p, $x['size'] );
                    $p += $x['size'];
                    break;
                case self::XDL_BDOP_INSB:
                    $x = unpack( 'Vcsize', substr( $diff, $p, 4 ) );
                    $p += 4;
                    $out .= substr( $diff, $p, $x['csize'] );
                    $p += $x['csize'];
                    break;
                case self::XDL_BDOP_CPY:
                    $x = unpack( 'Voff/Vcsize', substr( $diff, $p, 8 ) );
                    $p += 8;
                    $out .= substr( $base, $x['off'], $x['csize'] );
                    break;
                default:
                    wfDebug( __METHOD__ . ": invalid op" );
                    return false;
            }
        }
        return $out;
    }
 
    public function xdiffAdler32( $s ) {
        if ( !function_exists( 'hash' ) ) {
            return false;
        }
 
        static $init;
        if ( $init === null ) {
            $init = str_repeat( "\xf0", 205 ) . "\xee" . str_repeat( "\xf0", 67 ) . "\x02";
        }
 
        // The real Adler-32 checksum of $init is zero, so it initialises the
        // state to zero, as it is at the start of LibXDiff's checksum
        // algorithm. Appending the subject string then simulates LibXDiff.
        return strrev( hash( 'adler32', $init . $s, true ) );
    }
 
    private function uncompress() {
        if ( !$this->mDiffs ) {
            return;
        }
        $tail = '';
        $mDiffsCount = count( $this->mDiffs );
        for ( $diffKey = 0; $diffKey < $mDiffsCount; $diffKey++ ) {
            $textKey = $this->mDiffMap[$diffKey];
            $text = $this->patch( $tail, $this->mDiffs[$diffKey] );
            $this->mItems[$textKey] = $text;
            $tail = $text;
        }
    }
 
    public function __sleep() {
        $this->compress();
        if ( $this->mItems === [] ) {
            $info = false;
        } else {
            // Take forward differences to improve the compression ratio for sequences
            $map = '';
            $prev = 0;
            foreach ( $this->mDiffMap as $i ) {
                if ( $map !== '' ) {
                    $map .= ',';
                }
                $map .= $i - $prev;
                $prev = $i;
            }
            $info = [
                'diffs' => $this->mDiffs,
                'map' => $map
            ];
        }
        if ( isset( $this->mDefaultKey ) ) {
            $info['default'] = $this->mDefaultKey;
        }
        $this->mCompressed = gzdeflate( serialize( $info ) );
        return [ 'mCompressed' ];
    }
 
    public function __wakeup() {
        // addItem() doesn't work if mItems is partially filled from mDiffs
        $this->mFrozen = true;
        $info = unserialize( gzinflate( $this->mCompressed ) );
        $this->mCompressed = null;
 
        if ( !$info ) {
            // Empty object
            return;
        }
 
        if ( isset( $info['default'] ) ) {
            $this->mDefaultKey = $info['default'];
        }
        $this->mDiffs = $info['diffs'];
        if ( isset( $info['base'] ) ) {
            // Old format
            $this->mDiffMap = range( 0, count( $this->mDiffs ) - 1 );
            array_unshift( $this->mDiffs,
                pack( 'VVCV', 0, 0, self::XDL_BDOP_INSB, strlen( $info['base'] ) ) .
                $info['base'] );
        } else {
            // New format
            $map = explode( ',', $info['map'] );
            $cur = 0;
            $this->mDiffMap = [];
            foreach ( $map as $i ) {
                $cur += $i;
                $this->mDiffMap[] = $cur;
            }
        }
        $this->uncompress();
    }
 
    public function isHappy() {
        return $this->mSize < $this->mMaxSize
            && count( $this->mItems ) < $this->mMaxCount;
    }
 
}