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@439:     This file copyright 2008 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 "SimpleSPARQLQuery.h"
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@499: #include "data/model/WaveFileModel.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@439: using std::cerr;
Chris@439: using std::endl;
Chris@439: 
Chris@439: class RDFImporterImpl
Chris@439: {
Chris@439: public:
Chris@439:     RDFImporterImpl(QString url, int sampleRate);
Chris@439:     virtual ~RDFImporterImpl();
Chris@490: 
Chris@490:     void setSampleRate(int sampleRate) { m_sampleRate = sampleRate; }
Chris@439:     
Chris@439:     bool isOK();
Chris@439:     QString getErrorString() const;
Chris@439: 
Chris@439:     std::vector<Model *> getDataModels(ProgressReporter *);
Chris@439: 
Chris@439: protected:
Chris@439:     QString m_uristring;
Chris@439:     QString m_errorString;
Chris@499:     std::map<QString, Model *> m_audioModelMap;
Chris@439:     int m_sampleRate;
Chris@439: 
Chris@617:     std::map<Model *, std::map<QString, float> > m_labelValueMap;
Chris@617: 
Chris@520:     static bool m_prefixesLoaded;
Chris@520:     static void loadPrefixes(ProgressReporter *reporter);
Chris@520: 
Chris@499:     void getDataModelsAudio(std::vector<Model *> &, ProgressReporter *);
Chris@440:     void getDataModelsSparse(std::vector<Model *> &, ProgressReporter *);
Chris@440:     void getDataModelsDense(std::vector<Model *> &, ProgressReporter *);
Chris@440: 
Chris@493:     void getDenseModelTitle(Model *, QString, QString);
Chris@493: 
Chris@440:     void getDenseFeatureProperties(QString featureUri,
Chris@440:                                    int &sampleRate, int &windowLength,
Chris@440:                                    int &hopSize, int &width, int &height);
Chris@440: 
Chris@449:     void fillModel(Model *, long, long, bool, std::vector<float> &, QString);
Chris@439: };
Chris@439: 
Chris@520: bool RDFImporterImpl::m_prefixesLoaded = false;
Chris@439: 
Chris@439: QString
Chris@439: RDFImporter::getKnownExtensions()
Chris@439: {
Chris@439:     return "*.rdf *.n3 *.ttl";
Chris@439: }
Chris@439: 
Chris@439: RDFImporter::RDFImporter(QString url, int 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@490: RDFImporter::setSampleRate(int 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@439: std::vector<Model *>
Chris@439: RDFImporter::getDataModels(ProgressReporter *r)
Chris@439: {
Chris@439:     return m_d->getDataModels(r);
Chris@439: }
Chris@439: 
Chris@439: RDFImporterImpl::RDFImporterImpl(QString uri, int sampleRate) :
Chris@439:     m_uristring(uri),
Chris@439:     m_sampleRate(sampleRate)
Chris@439: {
Chris@439: }
Chris@439: 
Chris@439: RDFImporterImpl::~RDFImporterImpl()
Chris@439: {
Chris@492:     SimpleSPARQLQuery::closeSingleSource(m_uristring);
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@439: std::vector<Model *>
Chris@439: RDFImporterImpl::getDataModels(ProgressReporter *reporter)
Chris@439: {
Chris@520:     loadPrefixes(reporter);
Chris@520: 
Chris@439:     std::vector<Model *> 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@616:         std::cerr << m_errorString.toStdString() << std::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@499: RDFImporterImpl::getDataModelsAudio(std::vector<Model *> &models,
Chris@499:                                     ProgressReporter *reporter)
Chris@499: {
Chris@589:     SimpleSPARQLQuery query
Chris@499:         (SimpleSPARQLQuery::QueryFromSingleSource,
Chris@499:          QString
Chris@499:          (
Chris@499:              " PREFIX mo: <http://purl.org/ontology/mo/> "
Chris@499:              " SELECT ?signal ?source FROM <%1> "
Chris@589:              " WHERE { ?source a mo:AudioFile . "
Chris@589:              "         ?signal a mo:Signal . "
Chris@589:              "         ?source mo:encodes ?signal } "
Chris@499:              )
Chris@499:          .arg(m_uristring));
Chris@499: 
Chris@499:     SimpleSPARQLQuery::ResultList results = query.execute();
Chris@499: 
Chris@589:     if (results.empty()) {
Chris@589: 
Chris@589:         SimpleSPARQLQuery query2
Chris@589:             (SimpleSPARQLQuery::QueryFromSingleSource,
Chris@589:              QString
Chris@589:              (
Chris@589:                  " PREFIX mo: <http://purl.org/ontology/mo/> "
Chris@589:                  " SELECT ?signal ?source FROM <%1> "
Chris@589:                  " WHERE { ?signal a mo:Signal ; mo:available_as ?source } "
Chris@589:                  )
Chris@589:              .arg(m_uristring));
Chris@589:         
Chris@589:         results = query.execute();
Chris@589:     }
Chris@589: 
Chris@499:     for (int i = 0; i < results.size(); ++i) {
Chris@499: 
Chris@499:         QString signal = results[i]["signal"].value;
Chris@499:         QString source = results[i]["source"].value;
Chris@499: 
Chris@616:         std::cerr << "NOTE: Seeking signal source \"" << source.toStdString()
Chris@616:                   << "\"..." << std::endl;
Chris@616: 
Chris@522:         FileSource *fs = new FileSource(source, reporter);
Chris@616:         if (fs->isAvailable()) {
Chris@616:             std::cerr << "NOTE: Source is available: Local filename is \""
Chris@616:                       << fs->getLocalFilename().toStdString()
Chris@616:                       << "\"..." << std::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@616:             std::cerr << "NOTE: Signal source \"" << source.toStdString()
Chris@616:                       << "\" is not available, using file finder..." << std::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@617:                     std::cerr << "File finder returns: \"" << path.toStdString()
Chris@617:                               << "\"" << std::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@522:         WaveFileModel *newModel = new WaveFileModel(*fs, m_sampleRate);
Chris@522:         if (newModel->isOK()) {
Chris@522:             std::cerr << "Successfully created wave file model from source at \"" << source.toStdString() << "\"" << std::endl;
Chris@522:             models.push_back(newModel);
Chris@522:             m_audioModelMap[signal] = newModel;
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@522:             delete newModel;
Chris@499:         }
Chris@522:         delete fs;
Chris@499:     }
Chris@499: }
Chris@499: 
Chris@499: void
Chris@440: RDFImporterImpl::getDataModelsDense(std::vector<Model *> &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@589:     SimpleSPARQLQuery query
Chris@489:         (SimpleSPARQLQuery::QueryFromSingleSource,
Chris@480:          QString
Chris@440:          (
Chris@440:              " PREFIX mo: <http://purl.org/ontology/mo/>"
Chris@440:              " PREFIX af: <http://purl.org/ontology/af/>"
Chris@440:              
Chris@493:              " SELECT ?feature ?feature_signal_type ?value "
Chris@440:              " FROM <%1> "
Chris@440:              
Chris@440:              " WHERE { "
Chris@440:              
Chris@493:              "   ?signal af:signal_feature ?feature . "
Chris@440:              
Chris@440:              "   ?feature a ?feature_signal_type ; "
Chris@440:              "            af:value ?value . "
Chris@440:     
Chris@440:              " } "
Chris@440:              )
Chris@440:          .arg(m_uristring));
Chris@440: 
Chris@440:     SimpleSPARQLQuery::ResultList results = query.execute();
Chris@440: 
Chris@440:     if (!query.isOK()) {
Chris@440:         m_errorString = query.getErrorString();
Chris@440:         return;
Chris@440:     }
Chris@440: 
Chris@440:     if (query.wasCancelled()) {
Chris@440:         m_errorString = "Query cancelled";
Chris@440:         return;
Chris@440:     }        
Chris@440: 
Chris@440:     for (int i = 0; i < results.size(); ++i) {
Chris@440: 
Chris@440:         QString feature = results[i]["feature"].value;
Chris@440:         QString type = results[i]["feature_signal_type"].value;
Chris@440:         QString value = results[i]["value"].value;
Chris@440: 
Chris@440:         int 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@440:             SparseTimeValueModel *m = new 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@440:                 SparseTimeValueModel::Point point(j * hopSize, f, "");
Chris@440:                 m->addPoint(point);
Chris@440:             }
Chris@493: 
Chris@493:             getDenseModelTitle(m, feature, type);
Chris@440:         
Chris@558:             m->setRDFTypeURI(type);
Chris@558: 
Chris@440:             models.push_back(m);
Chris@440: 
Chris@440:         } else {
Chris@440: 
Chris@440:             EditableDenseThreeDimensionalModel *m =
Chris@535:                 new 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@493:             getDenseModelTitle(m, feature, type);
Chris@493:         
Chris@558:             m->setRDFTypeURI(type);
Chris@558: 
Chris@440:             models.push_back(m);
Chris@440:         }
Chris@440:     }
Chris@440: }
Chris@440: 
Chris@440: void
Chris@493: RDFImporterImpl::getDenseModelTitle(Model *m,
Chris@493:                                     QString featureUri,
Chris@493:                                     QString featureTypeUri)
Chris@493: {
Chris@493:     QString titleQuery = QString
Chris@493:         (
Chris@493:             " PREFIX dc: <http://purl.org/dc/elements/1.1/> "
Chris@493:             " SELECT ?title "
Chris@493:             " FROM <%1> " 
Chris@493:             " WHERE { "
Chris@493:             "   <%2> dc:title ?title . "
Chris@493:             " } "
Chris@493:             ).arg(m_uristring);
Chris@493:     
Chris@493:     SimpleSPARQLQuery::Value v;
Chris@493: 
Chris@493:     v = SimpleSPARQLQuery::singleResultQuery
Chris@493:         (SimpleSPARQLQuery::QueryFromSingleSource,
Chris@493:          titleQuery.arg(featureUri),
Chris@493:          "title");
Chris@493: 
Chris@493:     if (v.value != "") {
Chris@493:         std::cerr << "RDFImporterImpl::getDenseModelTitle: Title (from signal) \"" << v.value.toStdString() << "\"" << std::endl;
Chris@493:         m->setObjectName(v.value);
Chris@493:         return;
Chris@493:     }
Chris@493: 
Chris@493:     v = SimpleSPARQLQuery::singleResultQuery
Chris@493:         (SimpleSPARQLQuery::QueryFromSingleSource,
Chris@493:          titleQuery.arg(featureTypeUri),
Chris@493:          "title");
Chris@493:     
Chris@493:     if (v.value != "") {
Chris@493:         std::cerr << "RDFImporterImpl::getDenseModelTitle: Title (from signal type) \"" << v.value.toStdString() << "\"" << std::endl;
Chris@493:         m->setObjectName(v.value);
Chris@493:         return;
Chris@493:     }
Chris@493: 
Chris@493:     std::cerr << "RDFImporterImpl::getDenseModelTitle: No title available for feature <" << featureUri.toStdString() << ">" << std::endl;
Chris@493: }
Chris@493: 
Chris@493: void
Chris@440: RDFImporterImpl::getDenseFeatureProperties(QString featureUri,
Chris@440:                                            int &sampleRate, int &windowLength,
Chris@440:                                            int &hopSize, int &width, int &height)
Chris@440: {
Chris@489:     SimpleSPARQLQuery::QueryType s = SimpleSPARQLQuery::QueryFromSingleSource;
Chris@489: 
Chris@440:     QString dimensionsQuery 
Chris@440:         (
Chris@440:             " PREFIX mo: <http://purl.org/ontology/mo/>"
Chris@440:             " PREFIX af: <http://purl.org/ontology/af/>"
Chris@440:             
Chris@440:             " SELECT ?dimensions "
Chris@440:             " FROM <%1> "
Chris@440: 
Chris@440:             " WHERE { "
Chris@440: 
Chris@440:             "   <%2> af:dimensions ?dimensions . "
Chris@440:             
Chris@440:             " } "
Chris@440:             );
Chris@440: 
Chris@440:     SimpleSPARQLQuery::Value dimensionsValue =
Chris@489:         SimpleSPARQLQuery::singleResultQuery
Chris@489:         (s, dimensionsQuery.arg(m_uristring).arg(featureUri), "dimensions");
Chris@440: 
Chris@440:     cerr << "Dimensions = \"" << dimensionsValue.value.toStdString() << "\""
Chris@440:          << endl;
Chris@440: 
Chris@440:     if (dimensionsValue.value != "") {
Chris@440:         QStringList dl = dimensionsValue.value.split(" ");
Chris@440:         if (dl.empty()) dl.push_back(dimensionsValue.value);
Chris@440:         if (dl.size() > 0) height = dl[0].toInt();
Chris@440:         if (dl.size() > 1) width = dl[1].toInt();
Chris@440:     }
Chris@440: 
Chris@440:     QString queryTemplate
Chris@440:         (
Chris@440:             " PREFIX mo: <http://purl.org/ontology/mo/>"
Chris@440:             " PREFIX af: <http://purl.org/ontology/af/>"
Chris@440:             " PREFIX tl: <http://purl.org/NET/c4dm/timeline.owl#>"
Chris@440: 
Chris@440:             " SELECT ?%3 "
Chris@440:             " FROM <%1> "
Chris@440:             
Chris@440:             " WHERE { "
Chris@440:             
Chris@440:             "   <%2> mo:time ?time . "
Chris@440:             
Chris@440:             "   ?time a tl:Interval ; "
Chris@440:             "         tl:onTimeLine ?timeline . "
Chris@440: 
Chris@440:             "   ?map tl:rangeTimeLine ?timeline . "
Chris@440: 
Chris@440:             "   ?map tl:%3 ?%3 . "
Chris@440:             
Chris@440:             " } "
Chris@440:             );
Chris@440: 
Chris@440:     // Another laborious workaround for rasqal's failure to handle
Chris@440:     // multiple optionals properly
Chris@440: 
Chris@440:     SimpleSPARQLQuery::Value srValue = 
Chris@489:         SimpleSPARQLQuery::singleResultQuery(s,
Chris@480:                                              queryTemplate
Chris@440:                                              .arg(m_uristring).arg(featureUri)
Chris@440:                                              .arg("sampleRate"),
Chris@440:                                              "sampleRate");
Chris@440:     if (srValue.value != "") {
Chris@440:         sampleRate = srValue.value.toInt();
Chris@440:     }
Chris@440: 
Chris@440:     SimpleSPARQLQuery::Value hopValue = 
Chris@489:         SimpleSPARQLQuery::singleResultQuery(s,
Chris@480:                                              queryTemplate
Chris@440:                                              .arg(m_uristring).arg(featureUri)
Chris@440:                                              .arg("hopSize"),
Chris@440:                                              "hopSize");
Chris@440:     if (srValue.value != "") {
Chris@440:         hopSize = hopValue.value.toInt();
Chris@440:     }
Chris@440: 
Chris@440:     SimpleSPARQLQuery::Value winValue = 
Chris@489:         SimpleSPARQLQuery::singleResultQuery(s,
Chris@480:                                              queryTemplate
Chris@440:                                              .arg(m_uristring).arg(featureUri)
Chris@440:                                              .arg("windowLength"),
Chris@440:                                              "windowLength");
Chris@440:     if (winValue.value != "") {
Chris@440:         windowLength = winValue.value.toInt();
Chris@440:     }
Chris@440: 
Chris@440:     cerr << "sr = " << sampleRate << ", hop = " << hopSize << ", win = " << windowLength << endl;
Chris@440: }
Chris@440: 
Chris@440: void
Chris@440: RDFImporterImpl::getDataModelsSparse(std::vector<Model *> &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@489:     SimpleSPARQLQuery::QueryType s = SimpleSPARQLQuery::QueryFromSingleSource;
Chris@489: 
Chris@439:     // Our query is intended to retrieve every thing that has a time,
Chris@439:     // and every feature type and value associated with a thing that
Chris@439:     // has a time.
Chris@439: 
Chris@439:     // We will then need to refine this big bag of results into a set
Chris@439:     // of data models.
Chris@439: 
Chris@439:     // Results that have different source signals should go into
Chris@439:     // different models.
Chris@439: 
Chris@439:     // Results that have different feature types should go into
Chris@439:     // different models.
Chris@439: 
Chris@439:     // Results that are sparse should go into different models from
Chris@439:     // those that are dense (we need to examine the timestamps to
Chris@439:     // establish this -- if the timestamps are regular, the results
Chris@439:     // are dense -- so we can't do it as we go along, only after
Chris@439:     // collecting all results).
Chris@439: 
Chris@439:     // Timed things that have features associated with them should not
Chris@439:     // appear directly in any model -- their features should appear
Chris@439:     // instead -- and these should be different models from those used
Chris@439:     // for timed things that do not have features.
Chris@439: 
Chris@439:     // As we load the results, we'll push them into a partially
Chris@439:     // structured container that maps from source signal (URI as
Chris@439:     // string) -> feature type (likewise) -> time -> list of values.
Chris@439:     // If the source signal or feature type is unavailable, the empty
Chris@439:     // string will do.
Chris@439: 
Chris@449:     QString prefixes = QString(
Chris@439:         " PREFIX event: <http://purl.org/NET/c4dm/event.owl#>"
Chris@449:         " PREFIX tl: <http://purl.org/NET/c4dm/timeline.owl#>"
Chris@439:         " PREFIX mo: <http://purl.org/ontology/mo/>"
Chris@439:         " PREFIX af: <http://purl.org/ontology/af/>"
Chris@449:         " PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>"
Chris@449:         );
Chris@439: 
Chris@449:     QString queryString = prefixes + QString(
Chris@449: 
Chris@616:         " SELECT ?signal ?timed_thing ?timeline ?event_type ?value"
Chris@439:         " FROM <%1>"
Chris@439: 
Chris@439:         " WHERE {"
Chris@440: 
Chris@440:         "   ?signal a mo:Signal ."
Chris@440: 
Chris@439:         "   ?signal mo:time ?interval ."
Chris@616:         "   ?interval tl:onTimeLine ?timeline ."
Chris@616:         "   ?time tl:onTimeLine ?timeline ."
Chris@449:         "   ?timed_thing event:time ?time ."
Chris@440:         "   ?timed_thing a ?event_type ."
Chris@440: 
Chris@439:         "   OPTIONAL {"
Chris@440:         "     ?timed_thing af:feature ?value"
Chris@439:         "   }"
Chris@439:         " }"
Chris@439: 
Chris@439:         ).arg(m_uristring);
Chris@439: 
Chris@616:     //!!! NB we're using rather old terminology for these things, apparently:
Chris@616:     // beginsAt -> start
Chris@616:     // onTimeLine -> timeline
Chris@616: 
Chris@449:     QString timeQueryString = prefixes + QString(
Chris@449:         
Chris@449:         " SELECT ?time FROM <%1> "
Chris@449:         " WHERE { "        
Chris@449:         "   <%2> event:time ?t . "
Chris@449:         "   ?t tl:at ?time . "
Chris@449:         " } "
Chris@449: 
Chris@449:         ).arg(m_uristring);
Chris@449: 
Chris@449:     QString rangeQueryString = prefixes + QString(
Chris@449:         
Chris@449:         " SELECT ?time ?duration FROM <%1> "
Chris@449:         " WHERE { "
Chris@449:         "   <%2> event:time ?t . "
Chris@449:         "   ?t tl:beginsAt ?time . "
Chris@449:         "   ?t tl:duration ?duration . "
Chris@449:         " } "
Chris@449: 
Chris@449:         ).arg(m_uristring);
Chris@449: 
Chris@449:     QString labelQueryString = prefixes + QString(
Chris@449:         
Chris@449:         " SELECT ?label FROM <%1> "
Chris@449:         " WHERE { "
Chris@449:         "   <%2> rdfs:label ?label . "
Chris@449:         " } "
Chris@449: 
Chris@449:         ).arg(m_uristring);
Chris@449: 
Chris@510:     QString textQueryString = prefixes + QString(
Chris@510:         
Chris@510:         " SELECT ?label FROM <%1> "
Chris@510:         " WHERE { "
Chris@510:         "   <%2> af:text ?label . "
Chris@510:         " } "
Chris@510: 
Chris@510:         ).arg(m_uristring);
Chris@510: 
Chris@489:     SimpleSPARQLQuery query(s, queryString);
Chris@439:     query.setProgressReporter(reporter);
Chris@439: 
Chris@500: //    cerr << "Query will be: " << queryString.toStdString() << endl;
Chris@439: 
Chris@439:     SimpleSPARQLQuery::ResultList results = query.execute();
Chris@439: 
Chris@439:     if (!query.isOK()) {
Chris@439:         m_errorString = query.getErrorString();
Chris@440:         return;
Chris@439:     }
Chris@439: 
Chris@439:     if (query.wasCancelled()) {
Chris@439:         m_errorString = "Query cancelled";
Chris@440:         return;
Chris@439:     }        
Chris@439: 
Chris@449:     /*
Chris@449:       This function is now only used for sparse data (for dense data
Chris@449:       we would be in getDataModelsDense instead).
Chris@449: 
Chris@449:       For sparse data, the determining factors in deciding what model
Chris@449:       to use are: Do the features have values? and Do the features
Chris@449:       have duration?
Chris@449: 
Chris@449:       We can run through the results and check off whether we find
Chris@449:       values and duration for each of the source+type keys, and then
Chris@449:       run through the source+type keys pushing each of the results
Chris@449:       into a suitable model.
Chris@449: 
Chris@449:       Unfortunately, at this point we do not yet have any actual
Chris@449:       timing data (time/duration) -- just the time URI.
Chris@449: 
Chris@449:       What we _could_ do is to create one of each type of model at the
Chris@449:       start, for each of the source+type keys, and then push each
Chris@449:       feature into the relevant model depending on what we find out
Chris@449:       about it.  Then return only non-empty models.
Chris@449:     */
Chris@449: 
Chris@616:     // Map from timeline uri to event type to dimensionality to
Chris@449:     // presence of duration to model ptr.  Whee!
Chris@449:     std::map<QString, std::map<QString, std::map<int, std::map<bool, Model *> > > >
Chris@449:         modelMap;
Chris@449: 
Chris@439:     for (int i = 0; i < results.size(); ++i) {
Chris@439: 
Chris@499:         if (i % 4 == 0) {
Chris@499:             if (reporter) reporter->setProgress(i/4);
Chris@499:         }
Chris@499: 
Chris@499:         QString source = results[i]["signal"].value;
Chris@616:         QString timeline = results[i]["timeline"].value;
Chris@449:         QString type = results[i]["event_type"].value;
Chris@449:         QString thinguri = results[i]["timed_thing"].value;
Chris@449:         
Chris@449:         RealTime time;
Chris@449:         RealTime duration;
Chris@439: 
Chris@449:         bool haveTime = false;
Chris@449:         bool haveDuration = false;
Chris@439: 
Chris@510:         QString label = "";
Chris@510:         bool text = (type.contains("Text") || type.contains("text")); // Ha, ha
Chris@510: 
Chris@510:         if (text) {
Chris@510:             label = SimpleSPARQLQuery::singleResultQuery
Chris@510:                 (s, textQueryString.arg(thinguri), "label").value;
Chris@510:         }
Chris@510: 
Chris@510:         if (label == "") {
Chris@510:             label = SimpleSPARQLQuery::singleResultQuery
Chris@510:                 (s, labelQueryString.arg(thinguri), "label").value;
Chris@510:         }
Chris@449: 
Chris@489:         SimpleSPARQLQuery rangeQuery(s, rangeQueryString.arg(thinguri));
Chris@450:         SimpleSPARQLQuery::ResultList rangeResults = rangeQuery.execute();
Chris@450:         if (!rangeResults.empty()) {
Chris@450: //                std::cerr << rangeResults.size() << " range results" << std::endl;
Chris@450:             time = RealTime::fromXsdDuration
Chris@450:                 (rangeResults[0]["time"].value.toStdString());
Chris@450:             duration = RealTime::fromXsdDuration
Chris@450:                 (rangeResults[0]["duration"].value.toStdString());
Chris@450: //                std::cerr << "duration string " << rangeResults[0]["duration"].value.toStdString() << std::endl;
Chris@449:             haveTime = true;
Chris@450:             haveDuration = true;
Chris@449:         } else {
Chris@450:             QString timestring = SimpleSPARQLQuery::singleResultQuery
Chris@489:                 (s, timeQueryString.arg(thinguri), "time").value;
Chris@520: //            std::cerr << "timestring = " << timestring.toStdString() << std::endl;
Chris@450:             if (timestring != "") {
Chris@450:                 time = RealTime::fromXsdDuration(timestring.toStdString());
Chris@449:                 haveTime = true;
Chris@449:             }
Chris@449:         }
Chris@439: 
Chris@439:         QString valuestring = results[i]["value"].value;
Chris@449:         std::vector<float> values;
Chris@449: 
Chris@439:         if (valuestring != "") {
Chris@449:             QStringList vsl = valuestring.split(" ", QString::SkipEmptyParts);
Chris@449:             for (int j = 0; j < vsl.size(); ++j) {
Chris@449:                 bool success = false;
Chris@449:                 float v = vsl[j].toFloat(&success);
Chris@449:                 if (success) values.push_back(v);
Chris@449:             }
Chris@439:         }
Chris@439: 
Chris@449:         int dimensions = 1;
Chris@449:         if (values.size() == 1) dimensions = 2;
Chris@449:         else if (values.size() > 1) dimensions = 3;
Chris@449: 
Chris@449:         Model *model = 0;
Chris@449: 
Chris@616:         if (modelMap[timeline][type][dimensions].find(haveDuration) ==
Chris@616:             modelMap[timeline][type][dimensions].end()) {
Chris@449: 
Chris@449: /*
Chris@449:             std::cerr << "Creating new model: source = " << source.toStdString()
Chris@449:                       << ", type = " << type.toStdString() << ", dimensions = "
Chris@449:                       << dimensions << ", haveDuration = " << haveDuration
Chris@449:                       << ", time = " << time << ", duration = " << duration
Chris@449:                       << std::endl;
Chris@449: */
Chris@449:             
Chris@449:             if (!haveDuration) {
Chris@449: 
Chris@449:                 if (dimensions == 1) {
Chris@449: 
Chris@510:                     if (text) {
Chris@510:                         
Chris@510:                         model = new TextModel(m_sampleRate, 1, false);
Chris@510: 
Chris@510:                     } else {
Chris@510: 
Chris@510:                         model = new SparseOneDimensionalModel(m_sampleRate, 1, false);
Chris@510:                     }
Chris@449: 
Chris@449:                 } else if (dimensions == 2) {
Chris@449: 
Chris@510:                     if (text) {
Chris@510: 
Chris@510:                         model = new TextModel(m_sampleRate, 1, false);
Chris@510: 
Chris@510:                     } else {
Chris@510: 
Chris@510:                         model = new SparseTimeValueModel(m_sampleRate, 1, false);
Chris@510:                     }
Chris@449: 
Chris@449:                 } else {
Chris@449: 
Chris@449:                     // We don't have a three-dimensional sparse model,
Chris@449:                     // so use a note model.  We do have some logic (in
Chris@449:                     // extractStructure below) for guessing whether
Chris@449:                     // this should after all have been a dense model,
Chris@449:                     // but it's hard to apply it because we don't have
Chris@449:                     // all the necessary timing data yet... hmm
Chris@449: 
Chris@449:                     model = new NoteModel(m_sampleRate, 1, false);
Chris@449:                 }
Chris@449: 
Chris@449:             } else { // haveDuration
Chris@449: 
Chris@449:                 if (dimensions == 1 || dimensions == 2) {
Chris@449: 
Chris@449:                     // If our units are frequency or midi pitch, we
Chris@449:                     // should be using a note model... hm
Chris@449:                     
Chris@449:                     model = new RegionModel(m_sampleRate, 1, false);
Chris@449: 
Chris@449:                 } else {
Chris@449: 
Chris@449:                     // We don't have a three-dimensional sparse model,
Chris@449:                     // so use a note model.  We do have some logic (in
Chris@449:                     // extractStructure below) for guessing whether
Chris@449:                     // this should after all have been a dense model,
Chris@449:                     // but it's hard to apply it because we don't have
Chris@449:                     // all the necessary timing data yet... hmm
Chris@449: 
Chris@449:                     model = new NoteModel(m_sampleRate, 1, false);
Chris@449:                 }
Chris@449:             }
Chris@449: 
Chris@558:             model->setRDFTypeURI(type);
Chris@558: 
Chris@499:             if (m_audioModelMap.find(source) != m_audioModelMap.end()) {
Chris@499:                 std::cerr << "source model for " << model << " is " << m_audioModelMap[source] << std::endl;
Chris@499:                 model->setSourceModel(m_audioModelMap[source]);
Chris@499:             }
Chris@499: 
Chris@493:             QString titleQuery = QString
Chris@493:                 (
Chris@493:                     " PREFIX dc: <http://purl.org/dc/elements/1.1/> "
Chris@493:                     " SELECT ?title "
Chris@493:                     " FROM <%1> " 
Chris@493:                     " WHERE { "
Chris@493:                     "   <%2> dc:title ?title . "
Chris@493:                     " } "
Chris@493:                     ).arg(m_uristring).arg(type);
Chris@493:             QString title = SimpleSPARQLQuery::singleResultQuery
Chris@493:                 (s, titleQuery, "title").value;
Chris@618:             if (title == "") {
Chris@618:                 // take it from the end of the event type
Chris@618:                 title = type;
Chris@618:                 title.replace(QRegExp("^.*[/#]"), "");
Chris@618:             }
Chris@618:             model->setObjectName(title);
Chris@493: 
Chris@616:             modelMap[timeline][type][dimensions][haveDuration] = model;
Chris@449:             models.push_back(model);
Chris@449:         }
Chris@449: 
Chris@616:         model = modelMap[timeline][type][dimensions][haveDuration];
Chris@449: 
Chris@449:         if (model) {
Chris@449:             long ftime = RealTime::realTime2Frame(time, m_sampleRate);
Chris@449:             long fduration = RealTime::realTime2Frame(duration, m_sampleRate);
Chris@449:             fillModel(model, ftime, fduration, haveDuration, values, label);
Chris@439:         }
Chris@439:     }
Chris@439: }
Chris@439: 
Chris@439: void
Chris@449: RDFImporterImpl::fillModel(Model *model,
Chris@449:                            long ftime,
Chris@449:                            long fduration,
Chris@449:                            bool haveDuration,
Chris@449:                            std::vector<float> &values,
Chris@449:                            QString label)
Chris@449: {
Chris@493: //    std::cerr << "RDFImporterImpl::fillModel: adding point at frame " << ftime << std::endl;
Chris@492: 
Chris@449:     SparseOneDimensionalModel *sodm =
Chris@449:         dynamic_cast<SparseOneDimensionalModel *>(model);
Chris@449:     if (sodm) {
Chris@449:         SparseOneDimensionalModel::Point point(ftime, label);
Chris@449:         sodm->addPoint(point);
Chris@449:         return;
Chris@449:     }
Chris@449: 
Chris@510:     TextModel *tm =
Chris@510:         dynamic_cast<TextModel *>(model);
Chris@510:     if (tm) {
Chris@510:         TextModel::Point point
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@510:         tm->addPoint(point);
Chris@510:         return;
Chris@510:     }
Chris@510: 
Chris@449:     SparseTimeValueModel *stvm =
Chris@449:         dynamic_cast<SparseTimeValueModel *>(model);
Chris@449:     if (stvm) {
Chris@449:         SparseTimeValueModel::Point point
Chris@449:             (ftime, values.empty() ? 0.f : values[0], label);
Chris@449:         stvm->addPoint(point);
Chris@449:         return;
Chris@449:     }
Chris@449: 
Chris@449:     NoteModel *nm =
Chris@449:         dynamic_cast<NoteModel *>(model);
Chris@449:     if (nm) {
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@449:             NoteModel::Point point(ftime, value, fduration, level, label);
Chris@449:             nm->addPoint(point);
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@449:             NoteModel::Point point(ftime, value, duration, level, label);
Chris@449:             nm->addPoint(point);
Chris@449:         }
Chris@449:         return;
Chris@449:     }
Chris@449: 
Chris@449:     RegionModel *rm = 
Chris@449:         dynamic_cast<RegionModel *>(model);
Chris@449:     if (rm) {
Chris@617:         float value = 0.f;
Chris@617:         if (values.empty()) {
Chris@617:             // no values? map each unique label to a distinct value
Chris@617:             if (m_labelValueMap[model].find(label) == m_labelValueMap[model].end()) {
Chris@617:                 m_labelValueMap[model][label] = rm->getValueMaximum() + 1.f;
Chris@617:             }
Chris@617:             value = m_labelValueMap[model][label];
Chris@617:         } else {
Chris@617:             value = values[0];
Chris@617:         }
Chris@449:         if (haveDuration) {
Chris@617:             RegionModel::Point point(ftime, value, fduration, label);
Chris@449:             rm->addPoint(point);
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@449:             RegionModel::Point point(ftime, value, duration, label);
Chris@449:             rm->addPoint(point);
Chris@449:         }
Chris@449:         return;
Chris@449:     }
Chris@449:             
Chris@449:     std::cerr << "WARNING: RDFImporterImpl::fillModel: Unknown or unexpected model type" << std::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@449: 
Chris@499:     // This query is not expected to return any values, but if it
Chris@499:     // executes successfully (leaving no error in the error string)
Chris@499:     // then we know we have RDF
Chris@499:     SimpleSPARQLQuery q(SimpleSPARQLQuery::QueryFromSingleSource,
Chris@499:                         QString(" SELECT ?x FROM <%1> WHERE { ?x <y> <z> } ")
Chris@499:                         .arg(url));
Chris@499:     
Chris@499:     SimpleSPARQLQuery::ResultList r = q.execute();
Chris@499:     if (!q.isOK()) {
Chris@542:         SimpleSPARQLQuery::closeSingleSource(url);
Chris@499:         return NotRDF;
Chris@499:     }
Chris@499: 
Chris@588:     // "MO-conformant" structure for audio files
Chris@588: 
Chris@490:     SimpleSPARQLQuery::Value value =
Chris@490:         SimpleSPARQLQuery::singleResultQuery
Chris@490:         (SimpleSPARQLQuery::QueryFromSingleSource,
Chris@490:          QString
Chris@490:          (" PREFIX mo: <http://purl.org/ontology/mo/> "
Chris@490:           " SELECT ?url FROM <%1> "
Chris@588:           " WHERE { ?url a mo:AudioFile } "
Chris@490:              ).arg(url),
Chris@490:          "url");
Chris@490: 
Chris@490:     if (value.type == SimpleSPARQLQuery::URIValue) {
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@588:         value =
Chris@588:             SimpleSPARQLQuery::singleResultQuery
Chris@588:             (SimpleSPARQLQuery::QueryFromSingleSource,
Chris@588:              QString
Chris@588:              (" PREFIX mo: <http://purl.org/ontology/mo/> "
Chris@588:               " SELECT ?url FROM <%1> "
Chris@588:               " WHERE { ?signal a mo:Signal ; mo:available_as ?url } "
Chris@588:                  ).arg(url),
Chris@588:              "url");
Chris@588: 
Chris@588:         if (value.type == SimpleSPARQLQuery::URIValue) {
Chris@588:             haveAudio = true;
Chris@588:         }
Chris@490:     }
Chris@490: 
Chris@616:     std::cerr << "NOTE: RDFImporter::identifyDocumentType: haveAudio = "
Chris@616:               << haveAudio << std::endl;
Chris@616: 
Chris@490:     value =
Chris@490:         SimpleSPARQLQuery::singleResultQuery
Chris@490:         (SimpleSPARQLQuery::QueryFromSingleSource,
Chris@490:          QString
Chris@490:          (" PREFIX event: <http://purl.org/NET/c4dm/event.owl#> "
Chris@490:           " SELECT ?thing FROM <%1> "
Chris@490:           " WHERE { ?thing event:time ?time } "
Chris@490:              ).arg(url),
Chris@490:          "thing");
Chris@490: 
Chris@490:     if (value.type == SimpleSPARQLQuery::URIValue) {
Chris@490:         haveAnnotations = true;
Chris@490:     }
Chris@490: 
Chris@490:     if (!haveAnnotations) {
Chris@490:         
Chris@490:         value =
Chris@490:             SimpleSPARQLQuery::singleResultQuery
Chris@490:             (SimpleSPARQLQuery::QueryFromSingleSource,
Chris@490:              QString
Chris@490:              (" PREFIX af: <http://purl.org/ontology/af/> "
Chris@490:               " SELECT ?thing FROM <%1> "
Chris@490:               " WHERE { ?signal af:signal_feature ?thing } "
Chris@490:              ).arg(url),
Chris@490:              "thing");
Chris@490:         
Chris@490:         if (value.type == SimpleSPARQLQuery::URIValue) {
Chris@490:             haveAnnotations = true;
Chris@490:         }
Chris@490:     }
Chris@490: 
Chris@616:     std::cerr << "NOTE: RDFImporter::identifyDocumentType: haveAnnotations = "
Chris@616:               << haveAnnotations << std::endl;
Chris@616: 
Chris@542:     SimpleSPARQLQuery::closeSingleSource(url);
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: 
Chris@520: void
Chris@520: RDFImporterImpl::loadPrefixes(ProgressReporter *reporter)
Chris@520: {
Chris@520:     return;
Chris@520: //!!!
Chris@520:     if (m_prefixesLoaded) return;
Chris@520:     const char *prefixes[] = {
Chris@520:         "http://purl.org/NET/c4dm/event.owl",
Chris@520:         "http://purl.org/NET/c4dm/timeline.owl",
Chris@520:         "http://purl.org/ontology/mo/",
Chris@520:         "http://purl.org/ontology/af/",
Chris@520:         "http://www.w3.org/2000/01/rdf-schema",
Chris@520:         "http://purl.org/dc/elements/1.1/",
Chris@520:     };
Chris@520:     for (size_t i = 0; i < sizeof(prefixes)/sizeof(prefixes[0]); ++i) {
Chris@520:         CachedFile cf(prefixes[i], reporter, "application/rdf+xml");
Chris@520:         if (!cf.isOK()) continue;
Chris@520:         SimpleSPARQLQuery::addSourceToModel
Chris@520:             (QUrl::fromLocalFile(cf.getLocalFilename()).toString());
Chris@520:     }
Chris@520:     m_prefixesLoaded = true;
Chris@520: }