OMRAS2 Classical Browser

/* -*- c-basic-offset: 4 indent-tabs-mode: nil -*-  vi:set ts=8 sts=4 sw=4: */

#include "ImportClassicalDotNet.h"

#include <dataquay/Debug.h>

#include <QFile>

#include <QFileInfo>

#include <QTextStream>

#include <QRegExp>

#include <QVariant>

#include <exception>

using namespace Dataquay;

namespace ClassicalData {

void

ClassicalDotNetImporter::setSource(QUrl source)

{

    DEBUG << "ClassicalDotNetImporter::setSource: " << source << endl;

    import(source);

}

void

parseNames(QString field, QStringList &names)

{

    field.replace("Ä", QChar(0x00C4)); // LATIN CAPITAL LETTER A WITH DIAERESIS

    field.replace("ł", QChar(0x0142)); // LATIN SMALL LETTER L WITH STROKE

    field.replace("Ř", QChar(0x0158)); // LATIN CAPITAL LETTER R WITH CARON

    field.replace("á", QChar(0x00E1));

    field.replace("Á", QChar(0x00C1));

    field.replace("ç", QChar(0x00E7));

    field.replace("é", QChar(0x00E9));

    field.replace("É", QChar(0x00C9));

    field.replace("È", QChar(0x00C8));

    field.replace("Ë", QChar(0x00CB));

    field.replace("í", QChar(0x00ED));

    field.replace("Ï", QChar(0x00CF));

    field.replace("Ñ", QChar(0x00D1));

    field.replace("Ó", QChar(0x00D3));

    field.replace("Ô", QChar(0x00D4));

    field.replace("ò", QChar(0x00F2));

    field.replace("ö", QChar(0x00F6));

    field.replace("Ÿ", QChar(0x0178));

    if (field.contains(QRegExp("&[^ ]+;"))) {

        DEBUG << "Failed to handle entity in " << field << endl;

    }

    // all-caps -> titlecase

    QRegExp re("[A-Z][^ ,]*[A-Z][^,]+");

    int mp = re.indexIn(field);

    if (mp >= 0) {

        int ml = re.matchedLength();

        bool initial = true;

        for (int i = 0; i < ml; ++i) {

            if (initial) {

                initial = false;

                continue;

            }

            if (field[mp + i].isUpper()) {

                field[mp + i] = field[mp + i].toLower();

            } else if (field[mp + i].isSpace()) {

                initial = true;

            }

        }

    }

    field = field.trimmed();

    names.push_back(field);

    // comma

    re = QRegExp("^([^,]+), ([^,]+)$");

    if ((mp = re.indexIn(field)) >= 0) {

        QString c(re.cap(1));

        QString d(re.cap(2));

        names.push_back(d + " " + c);

        return;

    }

}

void

ClassicalDotNetImporter::import(QUrl source)

{

    //!!! for now

    QString filename = source.toLocalFile();

    QFile file(filename);

    if (!file.open(QFile::ReadOnly | QFile::Text)) {

        throw std::exception();

    }

    QTextStream stream(&file);

    stream.setCodec("UTF-8");

    QString all = stream.readAll();

    all.replace(QRegExp("^.*<div id=\"center\">"), "");

    QRegExp matcher

        ("<li><a href=\"([^\"]+)\">([^<]+)</a></li>");

    int pos = 0, count = 0;

    while ((pos = matcher.indexIn(all, pos)) != -1) {

        pos += matcher.matchedLength();

        ++count;

        DEBUG << "Item " << count

              << ": page = " << matcher.cap(1)

              << ", name = " << matcher.cap(2);

        QString namefield = matcher.cap(2);

        QStringList names;

        parseNames(namefield, names);

        if (names.empty()) {

            DEBUG << "No name!" << endl;

            continue;

        }

        if (names[0].contains(" Collections")) {

            continue;

        }

        Composer *composer = new Composer();

        composer->setName(names[0]);

        for (int i = 1; i < names.size(); ++i) {

            composer->addAlias(names[i]);

        }

        if (matcher.cap(1) != "") {

            QString url = matcher.cap(1);

            url.replace(QRegExp("^\\.\\./"), "/music/");

            Document *d = new Document;

            d->setUri(Uri("http://www.classical.net" + url));

            d->setTopic(composer);

            d->setSiteName("Classical Net");

            composer->addPage(d);

        }

        m_objects.push_back(composer);

    }

    DEBUG << "Found " << count << " things" << endl;

}

}