OMRAS2 Classical Browser

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

#include "Objects.h"

#include "TypeRegistrar.h"

#include <dataquay/BasicStore.h>

#include <dataquay/TransactionalStore.h>

#include <dataquay/RDFException.h>

#include <dataquay/objectmapper/ObjectLoader.h>

#include <dataquay/objectmapper/ObjectStorer.h>

#include <dataquay/objectmapper/ObjectMapper.h>

#include <dataquay/objectmapper/TypeMapping.h>

#include <dataquay/Debug.h>

#include "data/fileio/AudioFileReaderFactory.h"

#include "data/fileio/AudioFileReader.h"

#include "base/TempDirectory.h"

#include "Matcher.h"

#include <vamp-hostsdk/PluginLoader.h>

#include <QMultiMap>

#include <QFileInfo>

#include <iostream>

using namespace Dataquay;

using namespace ClassicalData;

using namespace std;

using namespace Vamp;

using namespace Vamp::HostExt;

ostream &operator<<(ostream &target, const QString &str)

{

    return target << str.toLocal8Bit().data();

}

ostream &operator<<(ostream &target, const QUrl &u)

{

    return target << "<" << u.toString() << ">";

}

bool

load(BasicStore *store, QString fileName)

{

    QUrl url = QUrl::fromLocalFile(fileName);

    cerr << "Importing from URL " << url << " ...";

    try {

        store->import(url, BasicStore::ImportPermitDuplicates);

    } catch (RDFException e) {

        cerr << " retrying with explicit ntriples type...";

        try {

            store->import(url, BasicStore::ImportPermitDuplicates, "ntriples");

        } catch (RDFException e) {

            cerr << "failed" << endl;

            cerr << "Import failed: " << e.what() << endl;

            return false;

        }

    }

    cerr << " done" << endl;

    return true;

}

void

usage(char *name)

{

    int s = 0;

    for (int i = 0; name[i]; ++i) if (name[i] == '/') s = i + 1;

    name = name + s;

    cerr << "Usage:" << endl;

    cerr << "  " << name << " <input-rdf-file> guess <track.wav> [<track.wav> ...]" << endl;

    exit(-1);

}

static QList<Composer *> allComposers;

static QHash<QString, Composer *> composerAliases;

static QHash<Uri, Composer *> composerUris;

static QMap<Composer *, QList<Work *> > worksMap;

static QList<Work *> allWorks;

void

show(Composer *c)

{

    cout << c->property("uri").value<Uri>() << endl;

    cout << c->getSortName(true);

    QString d = c->getDisplayDates();

    if (d != "") cout << " (" << d << ")";

    if (!c->nationality().empty() || c->period() != "") {

        cout << " [";

        bool first = true;

        foreach (QString n, c->nationality()) {

            if (!first) cout << "/";

            cout << n;

            first = false;

        }

        if (c->period() != "") {

            if (!first) cout << ", ";

            cout << c->period();

        }

        cout << "]";

    }

    if (c->gender() != "") {

        cout << " *" << c->gender();

    }

    if (!worksMap[c].empty()) {

        cout << " [" << worksMap[c].size() << " work(s)]";

    }

    cout << endl;

    foreach (QString a, c->aliases()) {

        cout << " - " << a << endl;

    }

    if (c->remarks() != "") {

        cout << " " << c->remarks() << endl;

    }

    foreach (Document *d, c->pages()) {

        cout << d->siteName() << " -> " << d->uri() << endl;

    }

    foreach (Uri u, c->otherURIs()) {

        cout << "Same as " << u << endl;

    }

}

void

showBrief(Composer *c)

{

    cout << c->property("uri").value<Uri>() << endl;

    cout << c->getSortName(false);

    QString d = c->getDisplayDates();

    if (d != "") cout << " (" << d << ")";

    if (!c->nationality().empty() || c->period() != "") {

        cout << " [";

        bool first = true;

        foreach (QString n, c->nationality()) {

            if (!first) cout << "/";

            cout << n;

            first = false;

        }

        if (c->period() != "") {

            if (!first) cout << " ";

            cout << c->period();

        }

        cout << "]";

    }

    if (c->gender() != "") {

        cout << " *" << c->gender();

    }

    if (!worksMap[c].empty()) {

        cout << " [" << worksMap[c].size() << " work(s)]";

    }

    cout << endl;

}

void

listBrief(QList<Composer *> composers)

{

    QMultiMap<QString, Composer *> sorted;

    foreach (Composer *c, composers) {

        sorted.insert(c->getSortName(false), c);

    }

    foreach (Composer *c, sorted) {

        showBrief(c);

    }

}

void

listUris(QList<Composer *> composers)

{

    QMultiMap<Uri, Composer *> sorted;

    foreach (Composer *c, composers) {

        sorted.insert(c->property("uri").value<Uri>(), c);

    }

    foreach (Uri uri, sorted.keys()) {

        cout << uri << endl;

    }

}

void

showSearchResults(QMultiMap<float, Composer *> matches, int count)

{

    int n = 0;

    for (QMultiMap<float, Composer *>::const_iterator i = matches.end();

         i != matches.begin(); ) {

        --i;

        if (i.key() <= 0) continue;

        cout << endl;

        if (n == 0) {

            cout << "Best match:" << endl;

        } else if (n == 1) {

            cout << "Other candidate(s):" << endl;

        }

        cout << "[" << i.key() << "] ";

        if (n == 0) show(i.value());

        else showBrief(i.value());

        if (++n > count) break;

    }

    if (n == 0) cout << "No matches" << endl;

    cout << endl;

}

void

getTrackData(FileSource source, QString &fingerprint, QString &puid,

             QString &title, QString &maker, AudioFileReader::TagMap &tags)

{

    AudioFileReader *reader = AudioFileReaderFactory::createReader(source);

//    AudioFileReader *reader = AudioFileReaderFactory::createThreadingReader(source);

    if (!reader || !reader->isOK()) {

        cerr << "Failed to open audio file" << endl;

        return;

    }

    title = reader->getTitle();

    maker = reader->getMaker();

//    cout << "File tag title: " << reader->getTitle() << endl;

//    cout << "File tag maker: " << reader->getMaker() << endl;

    cout << "All tags:" << endl;

    tags = reader->getTags();

    for (AudioFileReader::TagMap::const_iterator i = tags.begin();

         i != tags.end(); ++i) {

        cout << i->first << " " << i->second << endl;

    }

    PluginLoader *pl = PluginLoader::getInstance();

    Plugin *plugin = pl->loadPlugin

        ("ofa-vamp-plugin:ofa_puid_and_fingerprint", reader->getSampleRate(), PluginLoader::ADAPT_ALL);

    if (!plugin) {

        cerr << "Failed to load OFA Vamp plugin" << endl;

        return;

    }

    // 135 seconds... well, ok, let's have 136

    int secs = 136;

    int want = int(secs * reader->getSampleRate());

    int ch = reader->getChannelCount();

    std::vector<SampleBlock> samples;

    reader->getDeInterleavedFrames(0, want, samples);

    int have = samples[0].size();

    if (!plugin->initialise(ch, have, have)) {

        cerr << "Failed to initialise(" << ch << "," << have << "," << have << ") plugin" << endl;

        return;

    }

    float **input = new float *[ch];

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

        input[i] = &samples[i][0];

    }

    Plugin::FeatureSet features = plugin->process(input, RealTime::zeroTime);

    if (!features[0].empty()) {

        fingerprint = QString::fromStdString(features[0][0].label);

    }

    if (!features[1].empty()) {

        puid = QString::fromStdString(features[1][0].label);

    }

    features = plugin->getRemainingFeatures();

    if (fingerprint == "" && !features[0].empty()) {

        fingerprint = QString::fromStdString(features[0][0].label);

    }

    if (puid == "" && !features[1].empty()) {

        puid = QString::fromStdString(features[1][0].label);

    }

    delete[] input;

    delete plugin;

    delete reader;

}

float

bonusFactor(NamedEntity *e)

{

    // tiny nudge to prefer composers we actually have works for

    Composer *c = qobject_cast<Composer *>(e);

    float f = 1.f;

    int sz = 0;

    if (c && worksMap.contains(c)) {

        sz = worksMap[c].size();

        while (sz > 0) {

            f += 0.01;

            sz = sz / 10;

        }

    }

    return f;

}

void

integrateGuesses(GuessSet &guesses, GuessSet newGuesses)

{

    QHash<NamedEntity *, float> ecmap;

    foreach (Guess g, guesses) {

        ecmap[g.entity()] += g.confidence() * bonusFactor(g.entity());

    }

    foreach (Guess g, newGuesses) {

        if (ecmap.contains(g.entity())) {

            ecmap[g.entity()] += g.confidence() / 2;

        } else {

            ecmap[g.entity()] = g.confidence();

        }

    }

    guesses.clear();

    foreach (NamedEntity *e, ecmap.keys()) {

        guesses.insert(Guess(ecmap[e], e));

    }

}

void

guessFromMaker(QString maker, float scale, GuessSet &guesses)

{

    if (maker == "") return;

//    cerr << "guessFromMaker: " << maker << endl;

    GuessSet myGuesses;

    if (composerAliases.contains(maker)) {

        QList<Composer *> matching = composerAliases.values(maker);

        foreach (Composer *c, matching) {

            myGuesses.insert(Guess(10 * scale, c));

        }

    } else {

        ComposerFullTextMatcher matcher(allComposers);

        GuessList gl(matcher.match(maker, 5, 0.5));

        if (!gl.empty()) {

            foreach (Guess guess, gl) {

                myGuesses.insert(Guess(guess.confidence() * scale, guess.entity()));

            }

        }

    }

    integrateGuesses(guesses, myGuesses);

}

void

guessFromMakerTag(AudioFileReader::TagMap tags, QString tag, float scale, GuessSet &guesses)

{

    if (tags.find(tag) != tags.end()) {

        guessFromMaker(tags[tag], scale, guesses);

    }

}

void

guessFromTitle(QString title, float scale, GuessSet &guesses)

{

    QStringList bits = title.split(QRegExp("[:,_-]"),

                                   QString::SkipEmptyParts);

    if (bits.size() > 1) {

        guessFromMaker(bits.first(), scale, guesses);

    }

}    

void

guessFromTitleTag(AudioFileReader::TagMap tags, QString tag, float scale, GuessSet &guesses)

{

    if (tags.find(tag) != tags.end()) {

        guessFromTitle(tags[tag], scale, guesses);

    }

}

void

guessFromFilename(QString filename, float scale, GuessSet &guesses)

{

    cerr << "guessFromFilename: " << filename << endl;

    QString dirpart = QFileInfo(filename).path();

    QStringList dirbits = dirpart.split("/", QString::SkipEmptyParts);

    dirbits = dirbits.last()

        .replace(QRegExp("^\\d+"), "")

        .split(QRegExp("[^\\w]"), QString::SkipEmptyParts);

    if (!dirbits.empty()) {

        guessFromMaker(dirbits.first(), scale, guesses);

    }

    QString filepart = QFileInfo(filename).fileName().replace(QRegExp("\\d+"), "");

    QStringList filebits = filepart.split(QRegExp("[^\\w]"),

                                          QString::SkipEmptyParts);

    if (!filebits.empty()) {

        guessFromMaker(filebits.first(), scale, guesses);

    }

}

void

guessWorkFromTitleByCatalogue(QString title, float scale,

                              Composer *composer, GuessSet &guesses)

{

    if (title == "") return;

    WorkCatalogueMatcher matcher(composer ? worksMap.value(composer) : allWorks);

    GuessList gl(matcher.match(title, 0));

    if (!gl.empty()) {

        foreach (Guess guess, gl) {

            guesses.insert(Guess(guess.confidence() * scale, guess.entity()));

        }

    }

}

void

guessWorkFromTitle(QString title, float scale,

                   Composer *composer, GuessSet &guesses)

{

    if (title == "") return;

    WorkTitleMatcher matcher(composer ? worksMap.value(composer) : allWorks);

    GuessList gl(matcher.match(title, 0));

    if (!gl.empty()) {

        foreach (Guess guess, gl) {

            guesses.insert(Guess(guess.confidence() * scale, guess.entity()));

        }

    }

}    

void

guessWorkFromTitleTag(AudioFileReader::TagMap tags, QString tag, float scale,

                      Composer *composer, GuessSet &guesses)

{

    cerr << "guessWorkFromTitleTag: " << tag << endl;

    if (tags.find(tag) != tags.end()) {

        cerr << "guessWorkFromTitleTag: tag is " << tags[tag] << endl;

        GuessSet myGuesses;

        guessWorkFromTitle(tags[tag], scale, composer, myGuesses);

        integrateGuesses(guesses, myGuesses);

        myGuesses.clear();

        guessWorkFromTitle(tags[tag], scale, composer, myGuesses);

        integrateGuesses(guesses, myGuesses);

    }

}

void

guessWorkFromFilenameByCatalogue(QString filename, float scale,

                      Composer *composer, GuessSet &guesses)

{

    cerr << "guessWorkFromFilename: " << filename << endl;

    QString dirpart = QFileInfo(filename).path();

    QStringList dirbits = dirpart.split("/", QString::SkipEmptyParts);

    if (!dirbits.empty()) {

        guessWorkFromTitleByCatalogue(dirbits.last(), scale * 0.7, composer, guesses);

    }

    QString filepart = QFileInfo(filename).fileName().replace

        (QRegExp("\\.[^\\.]*"), "").replace(QRegExp("^\\d+[^\\w]+"), "");

    guessWorkFromTitleByCatalogue(filepart, scale, composer, guesses);

}

void

guessWorkFromFilenameByTitle(QString filename, float scale,

                             Composer *composer, GuessSet &guesses)

{

    cerr << "guessWorkFromFilename: " << filename << endl;

    QString dirpart = QFileInfo(filename).path();

    QStringList dirbits = dirpart.split("/", QString::SkipEmptyParts);

    if (!dirbits.empty()) {

        guessWorkFromTitle(dirbits.last(), scale * 0.7, composer, guesses);

    }

    QString filepart = QFileInfo(filename).fileName().replace

        (QRegExp("\\.[^\\.]*"), "").replace(QRegExp("^\\d+[^\\w]+"), "");

    guessWorkFromTitle(filepart, scale, composer, guesses);

}

Signal *

guess(QString track)

{

    cout << endl;

    cout << "Guessing composer for: " << track << endl;

//    cerr << "Creating Signal...";

    FileSource fs(track);

    Signal *tf = new Signal;

    tf->addAvailableAs(new AudioFile(fs));

//    cerr << "done" << endl;

//    cerr << "hash = " << tf->hash() << endl;

    QString fingerprint, puid, maker, title;

    AudioFileReader::TagMap tags;

    //!!! bad api!:

    getTrackData(fs, fingerprint, puid, title, maker, tags);

    cout << "fingerprint: " << fingerprint.toStdString() << ", puid: "

         << puid.toStdString() << endl;

    GuessSet guesses;

    guessFromMakerTag(tags, "TCOM", 1.0, guesses);

    guessFromMakerTag(tags, "COMPOSER", 1.0, guesses);

    if (guesses.empty() || guesses.begin()->confidence() < 0.4) {

        guessFromMakerTag(tags, "TOPE", 0.8, guesses);

        guessFromMakerTag(tags, "TPE1", 0.8, guesses);

        guessFromMakerTag(tags, "ARTIST", 0.9, guesses);

        guessFromMakerTag(tags, "PERFORMER", 0.8, guesses);

        guessFromTitleTag(tags, "TIT1", 0.4, guesses);

        guessFromTitleTag(tags, "TIT2", 0.5, guesses);

        guessFromTitleTag(tags, "TALB", 0.5, guesses);

        guessFromTitleTag(tags, "TIT3", 0.3, guesses);

        guessFromTitleTag(tags, "TITLE", 0.5, guesses);

        guessFromTitleTag(tags, "ALBUM", 0.5, guesses);

    }

    if (tags.find("MUSICBRAINZ_ARTISTID") != tags.end()) {

        QString id = tags["MUSICBRAINZ_ARTISTID"];

        Uri mbzUri = Uri("http://dbtune.org/musicbrainz/resource/artist/" + id);

        cout << "MBZ id found: " << id << endl;

        if (composerUris.contains(mbzUri)) {

            guesses.insert(Guess(2.0, composerUris[mbzUri]));

        }

    }

    cerr << "Composer guesses:" << endl;

    foreach (Guess g, guesses) {

        cerr << "[" << g.confidence() << "] " << g.entity()->uri() << endl;

    }

    float bc = 0.f;

    QString best;

    if (!guesses.empty()) {

        Guess bg = *guesses.begin();

        best = bg.entity()->name();

        bc = bg.confidence();

    }

    guessFromFilename(track, 0.5, guesses);

    float bc2 = 0.f;

    QString best2;

    if (!guesses.empty()) {

        Guess bg = *guesses.begin();

        best2 = bg.entity()->name();

        bc2 = bg.confidence();

    }

    // If we have only one confident composer guess, consider only

    // works from that composer (really this should permit considering

    // works from all confident composers)

    Composer *confidentComposer = 0;

    if (bc2 > 0.5) {

        confidentComposer = qobject_cast<Composer *>(guesses.begin()->entity());

    }

    QString bestTitle;

    GuessSet workGuesses;

    if (tags["TIT2"] != "") {

        bestTitle = tags["TIT2"];

        guessWorkFromTitleTag(tags, "TIT2", 0.5, confidentComposer, workGuesses);

    }

    if (tags["TITLE"] != "") {

        bestTitle = tags["TITLE"];

        guessWorkFromTitleTag(tags, "TITLE", 0.5, confidentComposer, workGuesses);

    }

    if (workGuesses.empty()) {

        guessWorkFromTitleTag(tags, "TIT1", 0.2, confidentComposer, workGuesses);

        guessWorkFromTitleTag(tags, "TALB", 0.2, confidentComposer, workGuesses);

        guessWorkFromTitleTag(tags, "TIT3", 0.1, confidentComposer, workGuesses);

        guessWorkFromTitleTag(tags, "ALBUM", 0.4, confidentComposer, workGuesses);

    }

    if (workGuesses.empty() || workGuesses.begin()->confidence() < 0.3) {

        guessWorkFromFilenameByCatalogue(track, 0.4, confidentComposer, workGuesses);

    }

    if (workGuesses.empty()) {

        guessWorkFromFilenameByTitle(track, 0.3, confidentComposer, workGuesses);

    }

    cerr << "Work guesses:" << endl;

    foreach (Guess g, workGuesses) {

        cerr << "[" << g.confidence() << "] " << g.entity()->uri() << endl;

    }

    GuessSet consistentComposers;

    GuessSet consistentWorks;

    foreach (Guess wg, workGuesses) {

        Work *w = qobject_cast<Work *>(wg.entity());

        if (!w || !w->getComposer()) continue;

        Composer *wc = w->getComposer();

        foreach (Guess g, guesses) {

            if (g.entity() == wc) {

                consistentComposers.insert(g);

                consistentWorks.insert(wg);

            }

        }

    }

    cerr << "Consistent composer guesses:" << endl;

    foreach (Guess g, consistentComposers) {

        cerr << "[" << g.confidence() << "] " << g.entity()->uri() << endl;

    }

    cerr << "Consistent work guesses:" << endl;

    foreach (Guess g, consistentWorks) {

        cerr << "[" << g.confidence() << "] " << g.entity()->uri() << endl;

    }

    float bc3 = bc2;

    QString best3 = best2;

    QString work;

    Work *bestWork = 0;

    if (!consistentWorks.empty()) {

        Guess bg = *consistentWorks.begin();

        bestWork = qobject_cast<Work *>(bg.entity());

        if (bestWork) {

            bc3 = bg.confidence();

            best3 = bestWork->getComposerName();

            work = bestWork->getDisplayName();

        }

    }

    cout << track << "|" << best << "|" << bc << "|" << best2 << "|" << bc2 << "|" << best3 << "|" << bc3 << "|" << work << "|" << bestTitle << endl;

    tf->setOfaFingerprint(fingerprint);

    tf->setPuid(puid);

    tf->setComposer(confidentComposer);

    tf->setWork(bestWork);

    return tf;

}

int

main(int argc, char **argv)

{

    if (argc < 3) usage(argv[0]);

    QString inRDFFileName = argv[1];

    QString command = argv[2];

    QStringList args;

    for (int i = 3; i < argc; ++i) {

        args.push_back(argv[i]);

    }

    //!!! unit test!

    int c = Work::compareCatalogueNumberTexts("Op. 1 no 4", "Op. 3 no 2");

    std::cerr << c << std::endl;

    c = Work::compareCatalogueNumberTexts("1 no 4", "3 no 2");

    std::cerr << c << std::endl;

    c = Work::compareCatalogueNumberTexts("4 no 2", "3 no 2");

    std::cerr << c << std::endl;

    c = Work::compareCatalogueNumberTexts("Opus 4 no 2", "3 no 2");

    std::cerr << c << std::endl;

    c = Work::compareCatalogueNumberTexts("Op 141", "K. 21");

    std::cerr << c << std::endl;

    c = Work::compareCatalogueNumberTexts("Op 14", "K. 21");

    std::cerr << c << std::endl;

    c = Work::compareCatalogueNumberTexts("Op 6a", "Op 6");

    std::cerr << c << std::endl;

    c = Work::compareCatalogueNumberTexts("Op 6a", "Op 6b");

    std::cerr << c << std::endl;

    c = Work::compareCatalogueNumberTexts("Op 6a", "Op 7");

    std::cerr << c << std::endl;

    c = Work::compareCatalogueNumberTexts("Hob XXIIId:Es1", "Hob XXII:B04");

    std::cerr << c << std::endl;

    BasicStore *store = new BasicStore();

    store->setBaseUri(Uri("http://dbtune.org/classical/resource/"));

    ObjectLoader *loader = new ObjectLoader(store);

    TypeMapping tm;

    TypeRegistrar::registerTypes();

    TypeRegistrar::addMappings(store, &tm);

    loader->setTypeMapping(tm);

    if (!load(store, inRDFFileName)) {

        cerr << "Failed to load data source" << endl;

        return 1;

    }

    cerr << "Imported RDF data, mapping to objects...";

    QObjectList objects = loader->loadAll();

    cerr << " done" << endl;

    delete loader;

    foreach (QObject *o, objects) {

        Composer *c = qobject_cast<Composer *>(o);

        if (c) {

            allComposers.push_back(c);

            composerAliases.insert(c->name(), c);

            foreach (QString alias, c->aliases()) {

                composerAliases.insert(alias, c);

            }

            composerUris.insert(c->property("uri").value<Uri>(), c);

            foreach (Uri otherUri, c->otherURIs()) {

                composerUris.insert(otherUri, c);

            }

        }

    }

    QList<Work *> works;

    foreach (QObject *o, objects) {

        Work *w = qobject_cast<Work *>(o);

        if (w) works.push_back(w);

    }

    foreach (Work *w, works) {

        allWorks.push_back(w);

        Composition *c = w->composition();

        if (c) {

            Composer *cp = c->composer();

            if (cp) worksMap[cp].push_back(w);

        }

    }

    BasicStore localStore;

    TypeRegistrar::addMappings(&localStore, &tm);

    ObjectStorer *localStorer = new ObjectStorer(&localStore);

    localStorer->setTypeMapping(tm);

    localStorer->setFollowPolicy(ObjectStorer::FollowObjectProperties);

    localStorer->setPropertyStorePolicy(ObjectStorer::StoreIfChanged);

    if (command == "guess") {

        if (args.empty()) usage(argv[0]);

        foreach (QString track, args) {

            Signal *tf = guess(track);

            localStorer->store(tf);

        }

    } else {

        usage(argv[0]);

    }

    delete localStorer;

    localStore.save("local.ttl");

    delete store;

    TempDirectory::getInstance()->cleanup();

}