index.cpp

来自「奇趣公司比较新的qt/emd版本」· C++ 代码 · 共 573 行 · 第 1/2 页

        dict.insert( key, new Entry( docs ) );
    }
    f.close();
    readDocumentList();
}

void Index::readDocumentList()
{
    QFile f( docListFile );
    if ( !f.open(QFile::ReadOnly ) )
        return;
    QDataStream s( &f );
    s >> docList;
}

QStringList Index::query( const QStringList &terms, const QStringList &termSeq, const QStringList &seqWords )
{
    QList<Term> termList;
    for (QStringList::ConstIterator it = terms.begin(); it != terms.end(); ++it ) {
        Entry *e = 0;
        if ( (*it).contains(QLatin1Char('*')) ) {
            QVector<Document> wcts = setupDummyTerm( getWildcardTerms( *it ) );
            termList.append( Term(QLatin1String("dummy"), wcts.count(), wcts ) );
        } else if ( dict[ *it ] ) {
            e = dict[ *it ];
            termList.append( Term( *it, e->documents.count(), e->documents ) );
        } else {
            return QStringList();
        }
    }
    if ( !termList.count() )
        return QStringList();
    qSort(termList);

    QVector<Document> minDocs = termList.takeFirst().documents;
    for(QList<Term>::Iterator it = termList.begin(); it != termList.end(); ++it) {
        Term *t = &(*it);
        QVector<Document> docs = t->documents;
        for(QVector<Document>::Iterator minDoc_it = minDocs.begin(); minDoc_it != minDocs.end(); ) {
            bool found = false;
            for (QVector<Document>::ConstIterator doc_it = docs.constBegin(); doc_it != docs.constEnd(); ++doc_it ) {
                if ( (*minDoc_it).docNumber == (*doc_it).docNumber ) {
                    (*minDoc_it).frequency += (*doc_it).frequency;
                    found = true;
                    break;
                }
            }
            if ( !found )
                minDoc_it = minDocs.erase( minDoc_it );
            else
                ++minDoc_it;
        }
    }

    QStringList results;
    qSort( minDocs );
    if ( termSeq.isEmpty() ) {
        for(QVector<Document>::Iterator it = minDocs.begin(); it != minDocs.end(); ++it)
            results << docList.at((int)(*it).docNumber);
        return results;
    }

    QString fileName;
    for(QVector<Document>::Iterator it = minDocs.begin(); it != minDocs.end(); ++it) {
        fileName =  docList[ (int)(*it).docNumber ];
        if ( searchForPattern( termSeq, seqWords, fileName ) )
            results << fileName;
    }
    return results;
}

QString Index::getDocumentTitle( const QString &fullFileName )
{
    QUrl url(fullFileName);
    QString fileName = url.toLocalFile();

    if (documentTitleCache.contains(fileName))
        return documentTitleCache.value(fileName);

    QFile file( fileName );
    if ( !file.open( QFile::ReadOnly ) ) {
        qWarning( "cannot open file %s", qPrintable(fileName) );
        return fileName;
    }
    QTextStream s( &file );
    QString text = s.readAll();

    int start = text.indexOf(QLatin1String("<title>"), 0, Qt::CaseInsensitive) + 7;
    int end = text.indexOf(QLatin1String("</title>"), 0, Qt::CaseInsensitive);

    QString title = tr("Untitled");
    if (end - start > 0) {
        title = text.mid(start, end - start);
        if (Qt::mightBeRichText(title)) {
            QTextDocument doc;
            doc.setHtml(title);
            title = doc.toPlainText();
        }
    }
    documentTitleCache.insert(fileName, title);
    return title;
}

QStringList Index::getWildcardTerms( const QString &term )
{
    QStringList lst;
    QStringList terms = split( term );
    QStringList::Iterator iter;

    for(QHash<QString, Entry*>::Iterator it = dict.begin(); it != dict.end(); ++it) {
        int index = 0;
        bool found = false;
        QString text( it.key() );
        for ( iter = terms.begin(); iter != terms.end(); ++iter ) {
            if ( *iter == QLatin1String("*") ) {
                found = true;
                continue;
            }
            if ( iter == terms.begin() && (*iter)[0] != text[0] ) {
                found = false;
                break;
            }
            index = text.indexOf( *iter, index );
            if ( *iter == terms.last() && index != (int)text.length()-1 ) {
                index = text.lastIndexOf( *iter );
                if ( index != (int)text.length() - (int)(*iter).length() ) {
                    found = false;
                    break;
                }
            }
            if ( index != -1 ) {
                found = true;
                index += (*iter).length();
                continue;
            } else {
                found = false;
                break;
            }
        }
        if ( found )
            lst << text;
    }

    return lst;
}

QStringList Index::split( const QString &str )
{
    QStringList lst;
    int j = 0;
    int i = str.indexOf(QLatin1Char('*'), j );

    if (str.startsWith(QLatin1String("*")))
        lst << QLatin1String("*");

    while ( i != -1 ) {
        if ( i > j && i <= (int)str.length() ) {
            lst << str.mid( j, i - j );
            lst << QLatin1String("*");
        }
        j = i + 1;
        i = str.indexOf(QLatin1Char('*'), j );
    }

    int l = str.length() - 1;
    if ( str.mid( j, l - j + 1 ).length() > 0 )
        lst << str.mid( j, l - j + 1 );

    return lst;
}

QVector<Document> Index::setupDummyTerm( const QStringList &terms )
{
    QList<Term> termList;
    for (QStringList::ConstIterator it = terms.begin(); it != terms.end(); ++it) {
        Entry *e = 0;
        if ( dict[ *it ] ) {
            e = dict[ *it ];
            termList.append( Term( *it, e->documents.count(), e->documents ) );
        }
    }
    QVector<Document> maxList(0);
    if ( !termList.count() )
        return maxList;
    qSort(termList);

    maxList = termList.takeLast().documents;
    for(QList<Term>::Iterator it = termList.begin(); it != termList.end(); ++it) {
        Term *t = &(*it);
        QVector<Document> docs = t->documents;
        for (QVector<Document>::iterator docIt = docs.begin(); docIt != docs.end(); ++docIt ) {
            if ( maxList.indexOf( *docIt ) == -1 )
                maxList.append( *docIt );
        }
    }
    return maxList;
}

void Index::buildMiniDict( const QString &str )
{
    if ( miniDict[ str ] )
        miniDict[ str ]->positions.append( wordNum );
    ++wordNum;
}

bool Index::searchForPattern( const QStringList &patterns, const QStringList &words, const QString &fileName )
{
    QUrl url(fileName);
    QString fName = url.toLocalFile();
    QFile file( fName );
    if ( !file.open( QFile::ReadOnly ) ) {
        qWarning( "cannot open file %s", qPrintable(fName) );
        return false;
    }

    wordNum = 3;
    miniDict.clear();
    QStringList::ConstIterator cIt = words.begin();
    for ( ; cIt != words.end(); ++cIt )
        miniDict.insert( *cIt, new PosEntry( 0 ) );

    QTextStream s( &file );
    QString text = s.readAll();
    bool valid = true;
    const QChar *buf = text.unicode();
    QChar str[64];
    QChar c = buf[0];
    int j = 0;
    int i = 0;
    while ( j < text.length() ) {
        if ( c == QLatin1Char('<') || c == QLatin1Char('&') ) {
            valid = false;
            if ( i > 1 )
                buildMiniDict( QString(str,i) );
            i = 0;
            c = buf[++j];
            continue;
        }
        if ( ( c == QLatin1Char('>') || c == QLatin1Char(';') ) && !valid ) {
            valid = true;
            c = buf[++j];
            continue;
        }
        if ( !valid ) {
            c = buf[++j];
            continue;
        }
        if ( ( c.isLetterOrNumber() || c == QLatin1Char('_') ) && i < 63 ) {
            str[i] = c.toLower();
            ++i;
        } else {
            if ( i > 1 )
                buildMiniDict( QString(str,i) );
            i = 0;
        }
        c = buf[++j];
    }
    if ( i > 1 )
        buildMiniDict( QString(str,i) );
    file.close();

    QStringList::ConstIterator patIt = patterns.begin();
    QStringList wordLst;
    QList<uint> a, b;
    QList<uint>::iterator aIt;
    for ( ; patIt != patterns.end(); ++patIt ) {
        wordLst = (*patIt).split(QLatin1Char(' '));
        a = miniDict[ wordLst[0] ]->positions;
        for ( int j = 1; j < (int)wordLst.count(); ++j ) {
            b = miniDict[ wordLst[j] ]->positions;
            aIt = a.begin();
            while ( aIt != a.end() ) {
                if ( b.contains( *aIt + 1 )) {
                    (*aIt)++;
                    ++aIt;
                } else {
                    aIt = a.erase( aIt );
                }
            }
        }
    }
    if ( a.count() )
        return true;
    return false;
}