Chris@439: /* -*- c-basic-offset: 4 indent-tabs-mode: nil -*-  vi:set ts=8 sts=4 sw=4: */
Chris@439: 
Chris@439: /*
Chris@439:     Sonic Visualiser
Chris@439:     An audio file viewer and annotation editor.
Chris@439:     Centre for Digital Music, Queen Mary, University of London.
Chris@727:     This file copyright 2008-2012 QMUL.
Chris@439:    
Chris@439:     This program is free software; you can redistribute it and/or
Chris@439:     modify it under the terms of the GNU General Public License as
Chris@439:     published by the Free Software Foundation; either version 2 of the
Chris@439:     License, or (at your option) any later version.  See the file
Chris@439:     COPYING included with this distribution for more information.
Chris@439: */
Chris@439: 
Chris@439: #include "RDFImporter.h"
Chris@439: 
Chris@439: #include <map>
Chris@439: #include <vector>
Chris@439: 
Chris@439: #include <iostream>
Chris@439: #include <cmath>
Chris@439: 
Chris@439: #include "base/ProgressReporter.h"
Chris@439: #include "base/RealTime.h"
Chris@439: 
Chris@439: #include "data/model/SparseOneDimensionalModel.h"
Chris@439: #include "data/model/SparseTimeValueModel.h"
Chris@439: #include "data/model/EditableDenseThreeDimensionalModel.h"
Chris@449: #include "data/model/NoteModel.h"
Chris@510: #include "data/model/TextModel.h"
Chris@449: #include "data/model/RegionModel.h"
Chris@1122: #include "data/model/ReadOnlyWaveFileModel.h"
Chris@499: 
Chris@499: #include "data/fileio/FileSource.h"
Chris@520: #include "data/fileio/CachedFile.h"
Chris@581: #include "data/fileio/FileFinder.h"
Chris@522: 
Chris@726: #include <dataquay/BasicStore.h>
Chris@726: #include <dataquay/PropertyObject.h>
Chris@726: 
Chris@726: using Dataquay::Uri;
Chris@726: using Dataquay::Node;
Chris@726: using Dataquay::Nodes;
Chris@726: using Dataquay::Triple;
Chris@726: using Dataquay::Triples;
Chris@726: using Dataquay::BasicStore;
Chris@726: using Dataquay::PropertyObject;
Chris@726: 
Chris@439: class RDFImporterImpl
Chris@439: {
Chris@439: public:
Chris@1040:     RDFImporterImpl(QString url, sv_samplerate_t sampleRate);
Chris@439:     virtual ~RDFImporterImpl();
Chris@490: 
Chris@1040:     void setSampleRate(sv_samplerate_t sampleRate) { m_sampleRate = sampleRate; }
Chris@439:     
Chris@439:     bool isOK();
Chris@439:     QString getErrorString() const;
Chris@439: 
Chris@1752:     std::vector<ModelId> getDataModels(ProgressReporter *);
Chris@439: 
Chris@439: protected:
Chris@726:     BasicStore *m_store;
Chris@730:     Uri expand(QString s) { return m_store->expand(s); }
Chris@726: 
Chris@439:     QString m_uristring;
Chris@439:     QString m_errorString;
Chris@1752:     std::map<QString, ModelId> m_audioModelMap;
Chris@1040:     sv_samplerate_t m_sampleRate;
Chris@439: 
Chris@1752:     std::map<ModelId, std::map<QString, float> > m_labelValueMap;
Chris@617: 
Chris@1752:     void getDataModelsAudio(std::vector<ModelId> &, ProgressReporter *);
Chris@1752:     void getDataModelsSparse(std::vector<ModelId> &, ProgressReporter *);
Chris@1752:     void getDataModelsDense(std::vector<ModelId> &, ProgressReporter *);
Chris@440: 
Chris@1752:     QString getDenseModelTitle(QString featureUri, QString featureTypeUri);
Chris@493: 
Chris@440:     void getDenseFeatureProperties(QString featureUri,
Chris@1040:                                    sv_samplerate_t &sampleRate, int &windowLength,
Chris@440:                                    int &hopSize, int &width, int &height);
Chris@440: 
Chris@1752:     void fillModel(ModelId, sv_frame_t, sv_frame_t,
Chris@1039:                    bool, std::vector<float> &, QString);
Chris@439: };
Chris@439: 
Chris@439: QString
Chris@439: RDFImporter::getKnownExtensions()
Chris@439: {
Chris@439:     return "*.rdf *.n3 *.ttl";
Chris@439: }
Chris@439: 
Chris@1040: RDFImporter::RDFImporter(QString url, sv_samplerate_t sampleRate) :
Chris@439:     m_d(new RDFImporterImpl(url, sampleRate)) 
Chris@439: {
Chris@439: }
Chris@439: 
Chris@439: RDFImporter::~RDFImporter()
Chris@439: {
Chris@439:     delete m_d;
Chris@439: }
Chris@439: 
Chris@490: void
Chris@1040: RDFImporter::setSampleRate(sv_samplerate_t sampleRate)
Chris@490: {
Chris@490:     m_d->setSampleRate(sampleRate);
Chris@490: }
Chris@490: 
Chris@439: bool
Chris@439: RDFImporter::isOK()
Chris@439: {
Chris@439:     return m_d->isOK();
Chris@439: }
Chris@439: 
Chris@439: QString
Chris@439: RDFImporter::getErrorString() const
Chris@439: {
Chris@439:     return m_d->getErrorString();
Chris@439: }
Chris@439: 
Chris@1752: std::vector<ModelId>
Chris@439: RDFImporter::getDataModels(ProgressReporter *r)
Chris@439: {
Chris@439:     return m_d->getDataModels(r);
Chris@439: }
Chris@439: 
Chris@1040: RDFImporterImpl::RDFImporterImpl(QString uri, sv_samplerate_t sampleRate) :
Chris@726:     m_store(new BasicStore),
Chris@439:     m_uristring(uri),
Chris@439:     m_sampleRate(sampleRate)
Chris@439: {
Chris@726:     //!!! retrieve data if remote... then
Chris@726: 
Chris@726:     m_store->addPrefix("mo", Uri("http://purl.org/ontology/mo/"));
Chris@726:     m_store->addPrefix("af", Uri("http://purl.org/ontology/af/"));
Chris@726:     m_store->addPrefix("dc", Uri("http://purl.org/dc/elements/1.1/"));
Chris@726:     m_store->addPrefix("tl", Uri("http://purl.org/NET/c4dm/timeline.owl#"));
Chris@726:     m_store->addPrefix("event", Uri("http://purl.org/NET/c4dm/event.owl#"));
Chris@726:     m_store->addPrefix("rdfs", Uri("http://www.w3.org/2000/01/rdf-schema#"));
Chris@727: 
Chris@738:     try {
Chris@738:         QUrl url;
Chris@738:         if (uri.startsWith("file:")) {
Chris@738:             url = QUrl(uri);
Chris@738:         } else {
Chris@738:             url = QUrl::fromLocalFile(uri);
Chris@738:         }
Chris@738:         m_store->import(url, BasicStore::ImportIgnoreDuplicates);
Chris@738:     } catch (std::exception &e) {
Chris@738:         m_errorString = e.what();
Chris@736:     }
Chris@439: }
Chris@439: 
Chris@439: RDFImporterImpl::~RDFImporterImpl()
Chris@439: {
Chris@726:     delete m_store;
Chris@439: }
Chris@439: 
Chris@439: bool
Chris@439: RDFImporterImpl::isOK()
Chris@439: {
Chris@439:     return (m_errorString == "");
Chris@439: }
Chris@439: 
Chris@439: QString
Chris@439: RDFImporterImpl::getErrorString() const
Chris@439: {
Chris@439:     return m_errorString;
Chris@439: }
Chris@439: 
Chris@1752: std::vector<ModelId>
Chris@439: RDFImporterImpl::getDataModels(ProgressReporter *reporter)
Chris@439: {
Chris@1752:     std::vector<ModelId> models;
Chris@439: 
Chris@499:     getDataModelsAudio(models, reporter);
Chris@499: 
Chris@490:     if (m_sampleRate == 0) {
Chris@616:         m_errorString = QString("Invalid audio data model (is audio file format supported?)");
Chris@843:         cerr << m_errorString << endl;
Chris@490:         return models;
Chris@490:     }
Chris@490: 
Chris@508:     QString error;
Chris@508: 
Chris@522:     if (m_errorString != "") {
Chris@522:         error = m_errorString;
Chris@522:     }
Chris@508:     m_errorString = "";
Chris@508: 
Chris@440:     getDataModelsDense(models, reporter);
Chris@440: 
Chris@522:     if (m_errorString != "") {
Chris@522:         error = m_errorString;
Chris@522:     }
Chris@440:     m_errorString = "";
Chris@440: 
Chris@440:     getDataModelsSparse(models, reporter);
Chris@440: 
Chris@522:     if (m_errorString == "" && error != "") {
Chris@522:         m_errorString = error;
Chris@522:     }
Chris@440: 
Chris@440:     return models;
Chris@440: }
Chris@440: 
Chris@440: void
Chris@1752: RDFImporterImpl::getDataModelsAudio(std::vector<ModelId> &models,
Chris@499:                                     ProgressReporter *reporter)
Chris@499: {
Chris@726:     Nodes sigs = m_store->match
Chris@730:         (Triple(Node(), Uri("a"), expand("mo:Signal"))).subjects();
Chris@499: 
Chris@726:     foreach (Node sig, sigs) {
Chris@726:         
Chris@730:         Node file = m_store->complete(Triple(Node(), expand("mo:encodes"), sig));
Chris@726:         if (file == Node()) {
Chris@730:             file = m_store->complete(Triple(sig, expand("mo:available_as"), Node()));
Chris@726:         }
Chris@726:         if (file == Node()) {
Chris@843:             cerr << "RDFImporterImpl::getDataModelsAudio: ERROR: No source for signal " << sig << endl;
Chris@726:             continue;
Chris@726:         }
Chris@499: 
Chris@726:         QString signal = sig.value;
Chris@726:         QString source = file.value;
Chris@589: 
Chris@726:         SVDEBUG << "NOTE: Seeking signal source \"" << source
Chris@726:                 << "\"..." << endl;
Chris@616: 
Chris@522:         FileSource *fs = new FileSource(source, reporter);
Chris@616:         if (fs->isAvailable()) {
Chris@690:             SVDEBUG << "NOTE: Source is available: Local filename is \""
Chris@726:                     << fs->getLocalFilename()
Chris@726:                     << "\"..." << endl;
Chris@616:         }
Chris@616:             
Chris@522: #ifdef NO_SV_GUI
Chris@522:         if (!fs->isAvailable()) {
Chris@522:             m_errorString = QString("Signal source \"%1\" is not available").arg(source);
Chris@522:             delete fs;
Chris@522:             continue;
Chris@522:         }
Chris@522: #else
Chris@522:         if (!fs->isAvailable()) {
Chris@726:             SVDEBUG << "NOTE: Signal source \"" << source
Chris@726:                     << "\" is not available, using file finder..." << endl;
Chris@522:             FileFinder *ff = FileFinder::getInstance();
Chris@581:             if (ff) {
Chris@581:                 QString path = ff->find(FileFinder::AudioFile,
Chris@581:                                         fs->getLocation(),
Chris@581:                                         m_uristring);
Chris@581:                 if (path != "") {
Chris@844:                     cerr << "File finder returns: \"" << path
Chris@843:                               << "\"" << endl;
Chris@522:                     delete fs;
Chris@581:                     fs = new FileSource(path, reporter);
Chris@581:                     if (!fs->isAvailable()) {
Chris@581:                         delete fs;
Chris@581:                         m_errorString = QString("Signal source \"%1\" is not available").arg(source);
Chris@581:                         continue;
Chris@581:                     }
Chris@522:                 }
Chris@499:             }
Chris@522:         }
Chris@522: #endif
Chris@522: 
Chris@522:         if (reporter) {
Chris@522:             reporter->setMessage(RDFImporter::tr("Importing audio referenced in RDF..."));
Chris@522:         }
Chris@522:         fs->waitForData();
Chris@1752:         auto newModel = std::make_shared<ReadOnlyWaveFileModel>
Chris@1752:             (*fs, m_sampleRate);
Chris@522:         if (newModel->isOK()) {
Chris@843:             cerr << "Successfully created wave file model from source at \"" << source << "\"" << endl;
Chris@1752:             auto modelId = ModelById::add(newModel);
Chris@1752:             models.push_back(modelId);
Chris@1752:             m_audioModelMap[signal] = modelId;
Chris@522:             if (m_sampleRate == 0) {
Chris@522:                 m_sampleRate = newModel->getSampleRate();
Chris@499:             }
Chris@508:         } else {
Chris@522:             m_errorString = QString("Failed to create wave file model from source at \"%1\"").arg(source);
Chris@499:         }
Chris@522:         delete fs;
Chris@499:     }
Chris@499: }
Chris@499: 
Chris@499: void
Chris@1752: RDFImporterImpl::getDataModelsDense(std::vector<ModelId> &models,
Chris@440:                                     ProgressReporter *reporter)
Chris@440: {
Chris@499:     if (reporter) {
Chris@499:         reporter->setMessage(RDFImporter::tr("Importing dense signal data from RDF..."));
Chris@499:     }
Chris@499: 
Chris@726:     Nodes sigFeatures = m_store->match
Chris@730:         (Triple(Node(), expand("af:signal_feature"), Node())).objects();
Chris@440: 
Chris@726:     foreach (Node sf, sigFeatures) {
Chris@440: 
Chris@726:         if (sf.type != Node::URI && sf.type != Node::Blank) continue;
Chris@726:         
Chris@730:         Node t = m_store->complete(Triple(sf, expand("a"), Node()));
Chris@730:         Node v = m_store->complete(Triple(sf, expand("af:value"), Node()));
Chris@440: 
Chris@726:         QString feature = sf.value;
Chris@726:         QString type = t.value;
Chris@726:         QString value = v.value;
Chris@726:         
Chris@726:         if (type == "" || value == "") continue;
Chris@440: 
Chris@1040:         sv_samplerate_t sampleRate = 0;
Chris@440:         int windowLength = 0;
Chris@440:         int hopSize = 0;
Chris@440:         int width = 0;
Chris@440:         int height = 0;
Chris@440:         getDenseFeatureProperties
Chris@440:             (feature, sampleRate, windowLength, hopSize, width, height);
Chris@440: 
Chris@440:         if (sampleRate != 0 && sampleRate != m_sampleRate) {
Chris@440:             cerr << "WARNING: Sample rate in dense feature description does not match our underlying rate -- using rate from feature description" << endl;
Chris@440:         }
Chris@440:         if (sampleRate == 0) sampleRate = m_sampleRate;
Chris@440: 
Chris@440:         if (hopSize == 0) {
Chris@440:             cerr << "WARNING: Dense feature description does not specify a hop size -- assuming 1" << endl;
Chris@440:             hopSize = 1;
Chris@440:         }
Chris@440: 
Chris@440:         if (height == 0) {
Chris@440:             cerr << "WARNING: Dense feature description does not specify feature signal dimensions -- assuming one-dimensional (height = 1)" << endl;
Chris@440:             height = 1;
Chris@440:         }
Chris@440: 
Chris@440:         QStringList values = value.split(' ', QString::SkipEmptyParts);
Chris@440: 
Chris@440:         if (values.empty()) {
Chris@440:             cerr << "WARNING: Dense feature description does not specify any values!" << endl;
Chris@440:             continue;
Chris@440:         }
Chris@440: 
Chris@440:         if (height == 1) {
Chris@440: 
Chris@1752:             auto m = std::make_shared<SparseTimeValueModel>
Chris@440:                 (sampleRate, hopSize, false);
Chris@440: 
Chris@440:             for (int j = 0; j < values.size(); ++j) {
Chris@440:                 float f = values[j].toFloat();
Chris@1651:                 Event e(j * hopSize, f, "");
Chris@1651:                 m->add(e);
Chris@440:             }
Chris@493: 
Chris@1752:             m->setObjectName(getDenseModelTitle(feature, type));
Chris@558:             m->setRDFTypeURI(type);
Chris@1752:             models.push_back(ModelById::add(m));
Chris@440: 
Chris@440:         } else {
Chris@440: 
Chris@1752:             auto m = std::make_shared<EditableDenseThreeDimensionalModel>
Chris@535:                 (sampleRate, hopSize, height, 
Chris@535:                  EditableDenseThreeDimensionalModel::NoCompression, false);
Chris@440:             
Chris@440:             EditableDenseThreeDimensionalModel::Column column;
Chris@440: 
Chris@440:             int x = 0;
Chris@440: 
Chris@440:             for (int j = 0; j < values.size(); ++j) {
Chris@440:                 if (j % height == 0 && !column.empty()) {
Chris@440:                     m->setColumn(x++, column);
Chris@440:                     column.clear();
Chris@440:                 }
Chris@440:                 column.push_back(values[j].toFloat());
Chris@440:             }
Chris@440: 
Chris@440:             if (!column.empty()) {
Chris@440:                 m->setColumn(x++, column);
Chris@440:             }
Chris@440: 
Chris@1752:             m->setObjectName(getDenseModelTitle(feature, type));
Chris@558:             m->setRDFTypeURI(type);
Chris@1752:             models.push_back(ModelById::add(m));
Chris@440:         }
Chris@440:     }
Chris@440: }
Chris@440: 
Chris@1752: QString
Chris@1752: RDFImporterImpl::getDenseModelTitle(QString featureUri,
Chris@493:                                     QString featureTypeUri)
Chris@493: {
Chris@730:     Node n = m_store->complete
Chris@730:         (Triple(Uri(featureUri), expand("dc:title"), Node()));
Chris@493: 
Chris@726:     if (n.type == Node::Literal && n.value != "") {
Chris@726:         SVDEBUG << "RDFImporterImpl::getDenseModelTitle: Title (from signal) \"" << n.value << "\"" << endl;
Chris@1752:         return n.value;
Chris@493:     }
Chris@493: 
Chris@730:     n = m_store->complete
Chris@730:         (Triple(Uri(featureTypeUri), expand("dc:title"), Node()));
Chris@726: 
Chris@726:     if (n.type == Node::Literal && n.value != "") {
Chris@726:         SVDEBUG << "RDFImporterImpl::getDenseModelTitle: Title (from signal type) \"" << n.value << "\"" << endl;
Chris@1752:         return n.value;
Chris@493:     }
Chris@493: 
Chris@690:     SVDEBUG << "RDFImporterImpl::getDenseModelTitle: No title available for feature <" << featureUri << ">" << endl;
Chris@1752:     return {};
Chris@493: }
Chris@493: 
Chris@493: void
Chris@440: RDFImporterImpl::getDenseFeatureProperties(QString featureUri,
Chris@1040:                                            sv_samplerate_t &sampleRate, int &windowLength,
Chris@440:                                            int &hopSize, int &width, int &height)
Chris@440: {
Chris@730:     Node dim = m_store->complete
Chris@730:         (Triple(Uri(featureUri), expand("af:dimensions"), Node()));
Chris@489: 
Chris@726:     cerr << "Dimensions = \"" << dim.value << "\"" << endl;
Chris@440: 
Chris@726:     if (dim.type == Node::Literal && dim.value != "") {
Chris@726:         QStringList dl = dim.value.split(" ");
Chris@726:         if (dl.empty()) dl.push_back(dim.value);
Chris@440:         if (dl.size() > 0) height = dl[0].toInt();
Chris@440:         if (dl.size() > 1) width = dl[1].toInt();
Chris@440:     }
Chris@726:     
Chris@726:     // Looking for rate, hop, window from:
Chris@726:     //
Chris@726:     // ?feature mo:time ?time .
Chris@726:     // ?time a tl:Interval .
Chris@726:     // ?time tl:onTimeLine ?timeline .
Chris@726:     // ?map tl:rangeTimeLine ?timeline .
Chris@726:     // ?map tl:sampleRate ?rate .
Chris@726:     // ?map tl:hopSize ?hop .
Chris@726:     // ?map tl:windowLength ?window .
Chris@440: 
Chris@730:     Node interval = m_store->complete(Triple(Uri(featureUri), expand("mo:time"), Node()));
Chris@440: 
Chris@730:     if (!m_store->contains(Triple(interval, expand("a"), expand("tl:Interval")))) {
Chris@726:         cerr << "RDFImporterImpl::getDenseFeatureProperties: Feature time node "
Chris@726:              << interval << " is not a tl:Interval" << endl;
Chris@726:         return;
Chris@440:     }
Chris@440: 
Chris@730:     Node tl = m_store->complete(Triple(interval, expand("tl:onTimeLine"), Node()));
Chris@726:     
Chris@726:     if (tl == Node()) {
Chris@726:         cerr << "RDFImporterImpl::getDenseFeatureProperties: Interval node "
Chris@726:              << interval << " lacks tl:onTimeLine property" << endl;
Chris@726:         return;
Chris@440:     }
Chris@440: 
Chris@730:     Node map = m_store->complete(Triple(Node(), expand("tl:rangeTimeLine"), tl));
Chris@726:     
Chris@726:     if (map == Node()) {
Chris@726:         cerr << "RDFImporterImpl::getDenseFeatureProperties: No map for "
Chris@726:              << "timeline node " << tl << endl;
Chris@726:     }
Chris@726: 
Chris@726:     PropertyObject po(m_store, "tl:", map);
Chris@726: 
Chris@726:     if (po.hasProperty("sampleRate")) {
Chris@1040:         sampleRate = po.getProperty("sampleRate").toDouble();
Chris@726:     }
Chris@726:     if (po.hasProperty("hopSize")) {
Chris@726:         hopSize = po.getProperty("hopSize").toInt();
Chris@726:     }
Chris@726:     if (po.hasProperty("windowLength")) {
Chris@726:         windowLength = po.getProperty("windowLength").toInt();
Chris@440:     }
Chris@440: 
Chris@440:     cerr << "sr = " << sampleRate << ", hop = " << hopSize << ", win = " << windowLength << endl;
Chris@440: }
Chris@440: 
Chris@440: void
Chris@1752: RDFImporterImpl::getDataModelsSparse(std::vector<ModelId> &models,
Chris@440:                                      ProgressReporter *reporter)
Chris@440: {
Chris@499:     if (reporter) {
Chris@499:         reporter->setMessage(RDFImporter::tr("Importing event data from RDF..."));
Chris@499:     }
Chris@499: 
Chris@726:     /*
Chris@726:       This function is only used for sparse data (for dense data we
Chris@726:       would be in getDataModelsDense instead).
Chris@489: 
Chris@726:       Our query is intended to retrieve every thing that has a time,
Chris@726:       and every feature type and value associated with a thing that
Chris@726:       has a time.
Chris@439: 
Chris@726:       We will then need to refine this big bag of results into a set
Chris@726:       of data models.
Chris@439: 
Chris@726:       Results that have different source signals should go into
Chris@726:       different models.
Chris@439: 
Chris@726:       Results that have different feature types should go into
Chris@726:       different models.
Chris@726:     */
Chris@439: 
Chris@726:     Nodes sigs = m_store->match
Chris@730:         (Triple(Node(), expand("a"), expand("mo:Signal"))).subjects();
Chris@449: 
Chris@616:     // Map from timeline uri to event type to dimensionality to
Chris@1752:     // presence of duration to model id.  Whee!
Chris@1752:     std::map<QString, std::map<QString, std::map<int, std::map<bool, ModelId> > > >
Chris@449:         modelMap;
Chris@449: 
Chris@726:     foreach (Node sig, sigs) {
Chris@726:         
Chris@730:         Node interval = m_store->complete(Triple(sig, expand("mo:time"), Node()));
Chris@726:         if (interval == Node()) continue;
Chris@439: 
Chris@730:         Node tl = m_store->complete(Triple(interval, expand("tl:onTimeLine"), Node()));
Chris@726:         if (tl == Node()) continue;
Chris@499: 
Chris@730:         Nodes times = m_store->match(Triple(Node(), expand("tl:onTimeLine"), tl)).subjects();
Chris@449:         
Chris@726:         foreach (Node tn, times) {
Chris@726:             
Chris@730:             Nodes timedThings = m_store->match(Triple(Node(), expand("event:time"), tn)).subjects();
Chris@439: 
Chris@726:             foreach (Node thing, timedThings) {
Chris@726:                 
Chris@730:                 Node typ = m_store->complete(Triple(thing, expand("a"), Node()));
Chris@726:                 if (typ == Node()) continue;
Chris@439: 
Chris@730:                 Node valu = m_store->complete(Triple(thing, expand("af:feature"), Node()));
Chris@510: 
Chris@726:                 QString source = sig.value;
Chris@726:                 QString timeline = tl.value;
Chris@726:                 QString type = typ.value;
Chris@726:                 QString thinguri = thing.value;
Chris@510: 
Chris@726:                 /*
Chris@726:                   For sparse data, the determining factors in deciding
Chris@726:                   what model to use are: Do the features have values?
Chris@726:                   and Do the features have duration?
Chris@449: 
Chris@726:                   We can run through the results and check off whether
Chris@726:                   we find values and duration for each of the
Chris@726:                   source+type keys, and then run through the
Chris@726:                   source+type keys pushing each of the results into a
Chris@726:                   suitable model.
Chris@439: 
Chris@726:                   Unfortunately, at this point we do not yet have any
Chris@726:                   actual timing data (time/duration) -- just the time
Chris@726:                   URI.
Chris@449: 
Chris@726:                   What we _could_ do is to create one of each type of
Chris@726:                   model at the start, for each of the source+type
Chris@726:                   keys, and then push each feature into the relevant
Chris@726:                   model depending on what we find out about it.  Then
Chris@726:                   return only non-empty models.
Chris@726:                 */
Chris@439: 
Chris@726:                 QString label = "";
Chris@726:                 bool text = (type.contains("Text") || type.contains("text")); // Ha, ha
Chris@726:                 bool note = (type.contains("Note") || type.contains("note")); // Guffaw
Chris@449: 
Chris@726:                 if (text) {
Chris@730:                     label = m_store->complete(Triple(thing, expand("af:text"), Node())).value;
Chris@726:                 }
Chris@726:                 
Chris@726:                 if (label == "") {
Chris@730:                     label = m_store->complete(Triple(thing, expand("rdfs:label"), Node())).value;
Chris@726:                 }
Chris@449: 
Chris@726:                 RealTime time;
Chris@726:                 RealTime duration;
Chris@726: 
Chris@930: //                bool haveTime = false;
Chris@726:                 bool haveDuration = false;
Chris@726: 
Chris@730:                 Node at = m_store->complete(Triple(tn, expand("tl:at"), Node()));
Chris@726: 
Chris@726:                 if (at != Node()) {
Chris@726:                     time = RealTime::fromXsdDuration(at.value.toStdString());
Chris@930: //                    haveTime = true;
Chris@726:                 } else {
Chris@726:     //!!! NB we're using rather old terminology for these things, apparently:
Chris@726:     // beginsAt -> start
Chris@726:     // onTimeLine -> timeline
Chris@726: 
Chris@730:                     Node start = m_store->complete(Triple(tn, expand("tl:beginsAt"), Node()));
Chris@730:                     Node dur = m_store->complete(Triple(tn, expand("tl:duration"), Node()));
Chris@726:                     if (start != Node() && dur != Node()) {
Chris@726:                         time = RealTime::fromXsdDuration
Chris@726:                             (start.value.toStdString());
Chris@726:                         duration = RealTime::fromXsdDuration
Chris@726:                             (dur.value.toStdString());
Chris@930: //                        haveTime = haveDuration = true;
Chris@726:                     }
Chris@726:                 }
Chris@726: 
Chris@726:                 QString valuestring = valu.value;
Chris@726:                 std::vector<float> values;
Chris@726: 
Chris@726:                 if (valuestring != "") {
Chris@726:                     QStringList vsl = valuestring.split(" ", QString::SkipEmptyParts);
Chris@726:                     for (int j = 0; j < vsl.size(); ++j) {
Chris@726:                         bool success = false;
Chris@726:                         float v = vsl[j].toFloat(&success);
Chris@726:                         if (success) values.push_back(v);
Chris@726:                     }
Chris@726:                 }
Chris@726:                 
Chris@726:                 int dimensions = 1;
Chris@726:                 if (values.size() == 1) dimensions = 2;
Chris@726:                 else if (values.size() > 1) dimensions = 3;
Chris@726: 
Chris@1752:                 ModelId modelId;
Chris@726: 
Chris@726:                 if (modelMap[timeline][type][dimensions].find(haveDuration) ==
Chris@726:                     modelMap[timeline][type][dimensions].end()) {
Chris@449: 
Chris@449: /*
Chris@690:             SVDEBUG << "Creating new model: source = " << source                      << ", type = " << type << ", dimensions = "
Chris@449:                       << dimensions << ", haveDuration = " << haveDuration
Chris@449:                       << ", time = " << time << ", duration = " << duration
Chris@687:                       << endl;
Chris@449: */
Chris@1752: 
Chris@1752:                     Model *model = nullptr;
Chris@1752:                     
Chris@726:                     if (!haveDuration) {
Chris@449: 
Chris@726:                         if (dimensions == 1) {
Chris@726:                             if (text) {
Chris@726:                                 model = new TextModel(m_sampleRate, 1, false);
Chris@726:                             } else {
Chris@726:                                 model = new SparseOneDimensionalModel(m_sampleRate, 1, false);
Chris@726:                             }
Chris@726:                         } else if (dimensions == 2) {
Chris@726:                             if (text) {
Chris@726:                                 model = new TextModel(m_sampleRate, 1, false);
Chris@726:                             } else {
Chris@726:                                 model = new SparseTimeValueModel(m_sampleRate, 1, false);
Chris@726:                             }
Chris@726:                         } else {
Chris@726:                             // We don't have a three-dimensional sparse model,
Chris@726:                             // so use a note model.  We do have some logic (in
Chris@726:                             // extractStructure below) for guessing whether
Chris@726:                             // this should after all have been a dense model,
Chris@726:                             // but it's hard to apply it because we don't have
Chris@726:                             // all the necessary timing data yet... hmm
Chris@726:                             model = new NoteModel(m_sampleRate, 1, false);
Chris@726:                         }
Chris@449: 
Chris@726:                     } else { // haveDuration
Chris@510: 
Chris@726:                         if (note || (dimensions > 2)) {
Chris@726:                             model = new NoteModel(m_sampleRate, 1, false);
Chris@726:                         } else {
Chris@726:                             // If our units are frequency or midi pitch, we
Chris@726:                             // should be using a note model... hm
Chris@726:                             model = new RegionModel(m_sampleRate, 1, false);
Chris@726:                         }
Chris@510:                     }
Chris@449: 
Chris@726:                     model->setRDFTypeURI(type);
Chris@449: 
Chris@726:                     if (m_audioModelMap.find(source) != m_audioModelMap.end()) {
Chris@843:                         cerr << "source model for " << model << " is " << m_audioModelMap[source] << endl;
Chris@1752:                         model->setSourceModel(m_audioModelMap[source]);
Chris@510:                     }
Chris@449: 
Chris@730:                     QString title = m_store->complete
Chris@730:                         (Triple(typ, expand("dc:title"), Node())).value;
Chris@726:                     if (title == "") {
Chris@726:                         // take it from the end of the event type
Chris@726:                         title = type;
Chris@726:                         title.replace(QRegExp("^.*[/#]"), "");
Chris@726:                     }
Chris@726:                     model->setObjectName(title);
Chris@449: 
Chris@1752:                     modelId = ModelById::add(std::shared_ptr<Model>(model));
Chris@1752:                     modelMap[timeline][type][dimensions][haveDuration] = modelId;
Chris@1752:                     models.push_back(modelId);
Chris@449:                 }
Chris@449: 
Chris@1752:                 modelId = modelMap[timeline][type][dimensions][haveDuration];
Chris@449: 
Chris@1752:                 if (!modelId.isNone()) {
Chris@1752:                     sv_frame_t ftime =
Chris@1752:                         RealTime::realTime2Frame(time, m_sampleRate);
Chris@1752:                     sv_frame_t fduration =
Chris@1752:                         RealTime::realTime2Frame(duration, m_sampleRate);
Chris@1752:                     fillModel(modelId, ftime, fduration,
Chris@1752:                               haveDuration, values, label);
Chris@449:                 }
Chris@449:             }
Chris@439:         }
Chris@439:     }
Chris@439: }
Chris@439: 
Chris@439: void
Chris@1752: RDFImporterImpl::fillModel(ModelId modelId,
Chris@1039:                            sv_frame_t ftime,
Chris@1039:                            sv_frame_t fduration,
Chris@449:                            bool haveDuration,
Chris@449:                            std::vector<float> &values,
Chris@449:                            QString label)
Chris@449: {
Chris@690: //    SVDEBUG << "RDFImporterImpl::fillModel: adding point at frame " << ftime << endl;
Chris@492: 
Chris@1752:     if (auto sodm = ModelById::getAs<SparseOneDimensionalModel>(modelId)) {
Chris@1658:         Event point(ftime, label);
Chris@1658:         sodm->add(point);
Chris@449:         return;
Chris@449:     }
Chris@449: 
Chris@1752:     if (auto tm = ModelById::getAs<TextModel>(modelId)) {
Chris@1661:         Event e
Chris@510:             (ftime,
Chris@510:              values.empty() ? 0.5f : values[0] < 0.f ? 0.f : values[0] > 1.f ? 1.f : values[0], // I was young and feckless once too
Chris@510:              label);
Chris@1661:         tm->add(e);
Chris@510:         return;
Chris@510:     }
Chris@510: 
Chris@1752:     if (auto stvm = ModelById::getAs<SparseTimeValueModel>(modelId)) {
Chris@1651:         Event e(ftime, values.empty() ? 0.f : values[0], label);
Chris@1651:         stvm->add(e);
Chris@449:         return;
Chris@449:     }
Chris@449: 
Chris@1752:     if (auto nm = ModelById::getAs<NoteModel>(modelId)) {
Chris@449:         if (haveDuration) {
Chris@449:             float value = 0.f, level = 1.f;
Chris@449:             if (!values.empty()) {
Chris@449:                 value = values[0];
Chris@449:                 if (values.size() > 1) {
Chris@449:                     level = values[1];
Chris@449:                 }
Chris@449:             }
Chris@1644:             Event e(ftime, value, fduration, level, label);
Chris@1644:             nm->add(e);
Chris@449:         } else {
Chris@449:             float value = 0.f, duration = 1.f, level = 1.f;
Chris@449:             if (!values.empty()) {
Chris@449:                 value = values[0];
Chris@449:                 if (values.size() > 1) {
Chris@449:                     duration = values[1];
Chris@449:                     if (values.size() > 2) {
Chris@449:                         level = values[2];
Chris@449:                     }
Chris@449:                 }
Chris@449:             }
Chris@1644:             Event e(ftime, value, sv_frame_t(lrintf(duration)),
Chris@1643:                         level, label);
Chris@1644:             nm->add(e);
Chris@449:         }
Chris@449:         return;
Chris@449:     }
Chris@449: 
Chris@1752:     if (auto rm = ModelById::getAs<RegionModel>(modelId)) {
Chris@617:         float value = 0.f;
Chris@617:         if (values.empty()) {
Chris@617:             // no values? map each unique label to a distinct value
Chris@1752:             if (m_labelValueMap[modelId].find(label) == m_labelValueMap[modelId].end()) {
Chris@1752:                 m_labelValueMap[modelId][label] = rm->getValueMaximum() + 1.f;
Chris@617:             }
Chris@1752:             value = m_labelValueMap[modelId][label];
Chris@617:         } else {
Chris@617:             value = values[0];
Chris@617:         }
Chris@449:         if (haveDuration) {
Chris@1649:             Event e(ftime, value, fduration, label);
Chris@1649:             rm->add(e);
Chris@449:         } else {
Chris@449:             // This won't actually happen -- we only create region models
Chris@449:             // if we do have duration -- but just for completeness
Chris@617:             float duration = 1.f;
Chris@449:             if (!values.empty()) {
Chris@449:                 value = values[0];
Chris@449:                 if (values.size() > 1) {
Chris@449:                     duration = values[1];
Chris@449:                 }
Chris@449:             }
Chris@1649:             Event e(ftime, value, sv_frame_t(lrintf(duration)), label);
Chris@1649:             rm->add(e);
Chris@449:         }
Chris@449:         return;
Chris@449:     }
Chris@449:             
Chris@843:     cerr << "WARNING: RDFImporterImpl::fillModel: Unknown or unexpected model type" << endl;
Chris@449:     return;
Chris@449: }
Chris@449: 
Chris@490: RDFImporter::RDFDocumentType
Chris@490: RDFImporter::identifyDocumentType(QString url)
Chris@490: {
Chris@490:     bool haveAudio = false;
Chris@490:     bool haveAnnotations = false;
Chris@726:     bool haveRDF = false;
Chris@449: 
Chris@1582:     BasicStore *store = nullptr;
Chris@726: 
Chris@726:     // This is not expected to return anything useful, but if it does
Chris@726:     // anything at all then we know we have RDF
Chris@726:     try {
Chris@738:         //!!! non-local document?
Chris@726:         store = BasicStore::load(QUrl(url));
Chris@730:         Triple t = store->matchOnce(Triple());
Chris@726:         if (t != Triple()) haveRDF = true;
Chris@1471:     } catch (std::exception &) {
Chris@738:         // nothing; haveRDF will be false so the next bit catches it
Chris@726:     }
Chris@726: 
Chris@726:     if (!haveRDF) {
Chris@726:         delete store;
Chris@499:         return NotRDF;
Chris@499:     }
Chris@499: 
Chris@726:     store->addPrefix("mo", Uri("http://purl.org/ontology/mo/"));
Chris@726:     store->addPrefix("event", Uri("http://purl.org/NET/c4dm/event.owl#"));
Chris@726:     store->addPrefix("af", Uri("http://purl.org/ontology/af/"));
Chris@726: 
Chris@588:     // "MO-conformant" structure for audio files
Chris@588: 
Chris@730:     Node n = store->complete(Triple(Node(), Uri("a"), store->expand("mo:AudioFile")));
Chris@726:     if (n != Node() && n.type == Node::URI) {
Chris@588: 
Chris@490:         haveAudio = true;
Chris@588: 
Chris@588:     } else {
Chris@588: 
Chris@588:         // Sonic Annotator v0.2 and below used to write this structure
Chris@588:         // (which is not properly in conformance with the Music
Chris@588:         // Ontology)
Chris@588: 
Chris@730:         Nodes sigs = store->match(Triple(Node(), Uri("a"), store->expand("mo:Signal"))).subjects();
Chris@726:         foreach (Node sig, sigs) {
Chris@730:             Node aa = store->complete(Triple(sig, store->expand("mo:available_as"), Node()));
Chris@726:             if (aa != Node()) {
Chris@726:                 haveAudio = true;
Chris@726:                 break;
Chris@726:             }
Chris@588:         }
Chris@490:     }
Chris@490: 
Chris@690:     SVDEBUG << "NOTE: RDFImporter::identifyDocumentType: haveAudio = "
Chris@687:               << haveAudio << endl;
Chris@616: 
Chris@736:     // can't call complete() with two Nothing nodes
Chris@736:     n = store->matchOnce(Triple(Node(), store->expand("event:time"), Node())).c;
Chris@726:     if (n != Node()) {
Chris@490:         haveAnnotations = true;
Chris@490:     }
Chris@490: 
Chris@490:     if (!haveAnnotations) {
Chris@736:         // can't call complete() with two Nothing nodes
Chris@736:         n = store->matchOnce(Triple(Node(), store->expand("af:signal_feature"), Node())).c;
Chris@726:         if (n != Node()) {
Chris@490:             haveAnnotations = true;
Chris@490:         }
Chris@490:     }
Chris@490: 
Chris@690:     SVDEBUG << "NOTE: RDFImporter::identifyDocumentType: haveAnnotations = "
Chris@687:               << haveAnnotations << endl;
Chris@616: 
Chris@726:     delete store;
Chris@542: 
Chris@490:     if (haveAudio) {
Chris@490:         if (haveAnnotations) {
Chris@490:             return AudioRefAndAnnotations;
Chris@490:         } else {
Chris@490:             return AudioRef;
Chris@490:         }
Chris@490:     } else {
Chris@490:         if (haveAnnotations) {
Chris@490:             return Annotations;
Chris@490:         } else {
Chris@499:             return OtherRDFDocument;
Chris@490:         }
Chris@490:     }
Chris@492: 
Chris@542:     return OtherRDFDocument;
Chris@490: }
Chris@490: