annotate rdf/PluginRDFIndexer.cpp @ 440:5746c559af15

* Merge revisions 1131 to 1201 from sv-rdf-import branch
author Chris Cannam
date Thu, 18 Sep 2008 12:33:30 +0000
parents beb2948baa77
children 64e64e304a12
rev   line source
Chris@439 1 /* -*- c-basic-offset: 4 indent-tabs-mode: nil -*- vi:set ts=8 sts=4 sw=4: */
Chris@439 2
Chris@439 3 /*
Chris@439 4 Sonic Visualiser
Chris@439 5 An audio file viewer and annotation editor.
Chris@439 6 Centre for Digital Music, Queen Mary, University of London.
Chris@439 7 This file copyright 2008 QMUL.
Chris@439 8
Chris@439 9 This program is free software; you can redistribute it and/or
Chris@439 10 modify it under the terms of the GNU General Public License as
Chris@439 11 published by the Free Software Foundation; either version 2 of the
Chris@439 12 License, or (at your option) any later version. See the file
Chris@439 13 COPYING included with this distribution for more information.
Chris@439 14 */
Chris@439 15
Chris@439 16 #include "PluginRDFIndexer.h"
Chris@439 17
Chris@439 18 #include "SimpleSPARQLQuery.h"
Chris@439 19
Chris@439 20 #include "data/fileio/FileSource.h"
Chris@439 21 #include "plugin/PluginIdentifier.h"
Chris@439 22
Chris@439 23 #include <vamp-sdk/PluginHostAdapter.h>
Chris@439 24
Chris@439 25 #include <QFileInfo>
Chris@439 26 #include <QDir>
Chris@439 27 #include <QUrl>
Chris@439 28
Chris@439 29 #include <iostream>
Chris@439 30 using std::cerr;
Chris@439 31 using std::endl;
Chris@439 32 using std::vector;
Chris@439 33 using std::string;
Chris@439 34 using Vamp::PluginHostAdapter;
Chris@439 35
Chris@439 36 PluginRDFIndexer *
Chris@439 37 PluginRDFIndexer::m_instance = 0;
Chris@439 38
Chris@439 39 PluginRDFIndexer *
Chris@439 40 PluginRDFIndexer::getInstance()
Chris@439 41 {
Chris@439 42 if (!m_instance) m_instance = new PluginRDFIndexer();
Chris@439 43 return m_instance;
Chris@439 44 }
Chris@439 45
Chris@439 46 PluginRDFIndexer::PluginRDFIndexer()
Chris@439 47 {
Chris@439 48 vector<string> paths = PluginHostAdapter::getPluginPath();
Chris@439 49
Chris@439 50 QStringList filters;
Chris@439 51 filters << "*.n3";
Chris@439 52 filters << "*.N3";
Chris@439 53 filters << "*.rdf";
Chris@439 54 filters << "*.RDF";
Chris@439 55
Chris@439 56 // Search each Vamp plugin path for a .rdf file that either has
Chris@439 57 // name "soname", "soname:label" or "soname/label" plus RDF
Chris@439 58 // extension. Use that order of preference, and prefer n3 over
Chris@439 59 // rdf extension.
Chris@439 60
Chris@439 61 for (vector<string>::const_iterator i = paths.begin(); i != paths.end(); ++i) {
Chris@439 62
Chris@439 63 QDir dir(i->c_str());
Chris@439 64 if (!dir.exists()) continue;
Chris@439 65
Chris@439 66 QStringList entries = dir.entryList
Chris@439 67 (filters, QDir::Files | QDir::Readable);
Chris@439 68
Chris@439 69 for (QStringList::const_iterator j = entries.begin();
Chris@439 70 j != entries.end(); ++j) {
Chris@439 71 QFileInfo fi(dir.filePath(*j));
Chris@439 72 indexFile(fi.absoluteFilePath());
Chris@439 73 }
Chris@439 74
Chris@439 75 QStringList subdirs = dir.entryList
Chris@439 76 (QDir::AllDirs | QDir::NoDotAndDotDot | QDir::Readable);
Chris@439 77
Chris@439 78 for (QStringList::const_iterator j = subdirs.begin();
Chris@439 79 j != subdirs.end(); ++j) {
Chris@439 80 QDir subdir(dir.filePath(*j));
Chris@439 81 if (subdir.exists()) {
Chris@439 82 entries = subdir.entryList
Chris@439 83 (filters, QDir::Files | QDir::Readable);
Chris@439 84 for (QStringList::const_iterator k = entries.begin();
Chris@439 85 k != entries.end(); ++k) {
Chris@439 86 QFileInfo fi(subdir.filePath(*k));
Chris@439 87 indexFile(fi.absoluteFilePath());
Chris@439 88 }
Chris@439 89 }
Chris@439 90 }
Chris@439 91 }
Chris@439 92 }
Chris@439 93
Chris@439 94 PluginRDFIndexer::~PluginRDFIndexer()
Chris@439 95 {
Chris@439 96 while (!m_cache.empty()) {
Chris@439 97 delete *m_cache.begin();
Chris@439 98 m_cache.erase(m_cache.begin());
Chris@439 99 }
Chris@439 100 }
Chris@439 101
Chris@439 102 QString
Chris@439 103 PluginRDFIndexer::getURIForPluginId(QString pluginId)
Chris@439 104 {
Chris@439 105 if (m_idToUriMap.find(pluginId) == m_idToUriMap.end()) return "";
Chris@439 106 return m_idToUriMap[pluginId];
Chris@439 107 }
Chris@439 108
Chris@439 109 QString
Chris@439 110 PluginRDFIndexer::getIdForPluginURI(QString uri)
Chris@439 111 {
Chris@439 112 if (m_uriToIdMap.find(uri) == m_uriToIdMap.end()) {
Chris@439 113
Chris@439 114 // Haven't found this uri referenced in any document on the
Chris@439 115 // local filesystem; try resolving the pre-fragment part of
Chris@439 116 // the uri as a document URL and reading that if possible.
Chris@439 117
Chris@439 118 // Because we may want to refer to this document again, we
Chris@439 119 // cache it locally if it turns out to exist.
Chris@439 120
Chris@439 121 cerr << "PluginRDFIndexer::getIdForPluginURI: NOTE: Failed to find a local RDF document describing plugin <" << uri.toStdString() << ">: attempting to retrieve one remotely by guesswork" << endl;
Chris@439 122
Chris@439 123 QString baseUrl = QUrl(uri).toString(QUrl::RemoveFragment);
Chris@439 124
Chris@439 125 FileSource source(baseUrl);
Chris@439 126 if (source.isAvailable()) {
Chris@439 127 source.waitForData();
Chris@439 128 if (indexFile(source.getLocalFilename())) {
Chris@439 129 m_cache.insert(new FileSource(source));
Chris@439 130 }
Chris@439 131 }
Chris@439 132
Chris@439 133 if (m_uriToIdMap.find(uri) == m_uriToIdMap.end()) {
Chris@439 134 m_uriToIdMap[uri] = "";
Chris@439 135 }
Chris@439 136 }
Chris@439 137
Chris@439 138 return m_uriToIdMap[uri];
Chris@439 139 }
Chris@439 140
Chris@439 141 QString
Chris@439 142 PluginRDFIndexer::getDescriptionURLForPluginId(QString pluginId)
Chris@439 143 {
Chris@439 144 if (m_idToDescriptionMap.find(pluginId) == m_idToDescriptionMap.end()) return "";
Chris@439 145 return m_idToDescriptionMap[pluginId];
Chris@439 146 }
Chris@439 147
Chris@439 148 QString
Chris@439 149 PluginRDFIndexer::getDescriptionURLForPluginURI(QString uri)
Chris@439 150 {
Chris@439 151 QString id = getIdForPluginURI(uri);
Chris@439 152 if (id == "") return "";
Chris@439 153 return getDescriptionURLForPluginId(id);
Chris@439 154 }
Chris@439 155
Chris@439 156 bool
Chris@439 157 PluginRDFIndexer::indexFile(QString filepath)
Chris@439 158 {
Chris@439 159 QUrl url = QUrl::fromLocalFile(filepath);
Chris@439 160 QString urlString = url.toString();
Chris@439 161 return indexURL(urlString);
Chris@439 162 }
Chris@439 163
Chris@439 164 bool
Chris@439 165 PluginRDFIndexer::indexURL(QString urlString)
Chris@439 166 {
Chris@439 167 // cerr << "PluginRDFIndexer::indexURL: url = <" << urlString.toStdString() << ">" << endl;
Chris@439 168
Chris@439 169 SimpleSPARQLQuery query
Chris@439 170 (QString
Chris@439 171 (
Chris@439 172 " PREFIX vamp: <http://purl.org/ontology/vamp/> "
Chris@439 173
Chris@439 174 " SELECT ?plugin ?library_id ?plugin_id "
Chris@439 175 " FROM <%1> "
Chris@439 176
Chris@439 177 " WHERE { "
Chris@439 178 " ?plugin a vamp:Plugin . "
Chris@439 179
Chris@439 180 // Make the identifier and library parts optional, so
Chris@439 181 // that we can check and report helpfully if one or both
Chris@439 182 // is absent instead of just getting no results
Chris@439 183
Chris@440 184 //!!! No -- because of rasqal's inability to correctly
Chris@440 185 // handle more than one OPTIONAL graph in a query, let's
Chris@440 186 // make identifier compulsory after all
Chris@440 187 //" OPTIONAL { ?plugin vamp:identifier ?plugin_id } . "
Chris@440 188
Chris@440 189 " ?plugin vamp:identifier ?plugin_id . "
Chris@439 190
Chris@439 191 " OPTIONAL { "
Chris@439 192 " ?library a vamp:PluginLibrary ; "
Chris@439 193 " vamp:available_plugin ?plugin ; "
Chris@439 194 " vamp:identifier ?library_id "
Chris@439 195 " } "
Chris@439 196 " } "
Chris@439 197 )
Chris@439 198 .arg(urlString));
Chris@439 199
Chris@439 200 SimpleSPARQLQuery::ResultList results = query.execute();
Chris@439 201
Chris@439 202 if (!query.isOK()) {
Chris@439 203 cerr << "ERROR: PluginRDFIndexer::indexURL: ERROR: Failed to index document at <"
Chris@439 204 << urlString.toStdString() << ">: "
Chris@439 205 << query.getErrorString().toStdString() << endl;
Chris@439 206 return false;
Chris@439 207 }
Chris@439 208
Chris@439 209 if (results.empty()) {
Chris@439 210 cerr << "PluginRDFIndexer::indexURL: NOTE: Document at <"
Chris@439 211 << urlString.toStdString()
Chris@439 212 << "> does not describe any vamp:Plugin resources" << endl;
Chris@439 213 return false;
Chris@439 214 }
Chris@439 215
Chris@439 216 bool foundSomething = false;
Chris@439 217 bool addedSomething = false;
Chris@439 218
Chris@439 219 for (SimpleSPARQLQuery::ResultList::iterator i = results.begin();
Chris@439 220 i != results.end(); ++i) {
Chris@439 221
Chris@439 222 QString pluginUri = (*i)["plugin"].value;
Chris@439 223 QString soname = (*i)["library_id"].value;
Chris@439 224 QString identifier = (*i)["plugin_id"].value;
Chris@439 225
Chris@439 226 if (identifier == "") {
Chris@439 227 cerr << "PluginRDFIndexer::indexURL: NOTE: Document at <"
Chris@439 228 << urlString.toStdString()
Chris@439 229 << "> fails to define any vamp:identifier for plugin <"
Chris@439 230 << pluginUri.toStdString() << ">"
Chris@439 231 << endl;
Chris@439 232 continue;
Chris@439 233 }
Chris@439 234 if (soname == "") {
Chris@439 235 cerr << "PluginRDFIndexer::indexURL: NOTE: Document at <"
Chris@439 236 << urlString.toStdString() << "> does not associate plugin <"
Chris@439 237 << pluginUri.toStdString() << "> with any implementation library"
Chris@439 238 << endl;
Chris@439 239 continue;
Chris@439 240 }
Chris@439 241 /*
Chris@439 242 cerr << "PluginRDFIndexer::indexURL: Document for plugin \""
Chris@439 243 << soname.toStdString() << ":" << identifier.toStdString()
Chris@439 244 << "\" (uri <" << pluginUri.toStdString() << ">) is at url <"
Chris@439 245 << urlString.toStdString() << ">" << endl;
Chris@439 246 */
Chris@439 247 QString pluginId = PluginIdentifier::createIdentifier
Chris@439 248 ("vamp", soname, identifier);
Chris@439 249
Chris@439 250 foundSomething = true;
Chris@439 251
Chris@439 252 if (m_idToDescriptionMap.find(pluginId) != m_idToDescriptionMap.end()) {
Chris@439 253 cerr << "PluginRDFIndexer::indexURL: NOTE: Plugin id \""
Chris@439 254 << pluginId.toStdString() << "\", described in document at <"
Chris@439 255 << urlString.toStdString()
Chris@439 256 << ">, has already been described in document <"
Chris@439 257 << m_idToDescriptionMap[pluginId].toStdString()
Chris@439 258 << ">: ignoring this new description" << endl;
Chris@439 259 continue;
Chris@439 260 }
Chris@439 261
Chris@439 262 m_idToDescriptionMap[pluginId] = urlString;
Chris@439 263 m_idToUriMap[pluginId] = pluginUri;
Chris@439 264
Chris@439 265 addedSomething = true;
Chris@439 266
Chris@439 267 if (pluginUri != "") {
Chris@439 268 if (m_uriToIdMap.find(pluginUri) != m_uriToIdMap.end()) {
Chris@439 269 cerr << "PluginRDFIndexer::indexURL: WARNING: Found multiple plugins with the same URI:" << endl;
Chris@439 270 cerr << " 1. Plugin id \"" << m_uriToIdMap[pluginUri].toStdString() << "\"" << endl;
Chris@439 271 cerr << " described in <" << m_idToDescriptionMap[m_uriToIdMap[pluginUri]].toStdString() << ">" << endl;
Chris@439 272 cerr << " 2. Plugin id \"" << pluginId.toStdString() << "\"" << endl;
Chris@439 273 cerr << " described in <" << urlString.toStdString() << ">" << endl;
Chris@439 274 cerr << "both claim URI <" << pluginUri.toStdString() << ">" << endl;
Chris@439 275 } else {
Chris@439 276 m_uriToIdMap[pluginUri] = pluginId;
Chris@439 277 }
Chris@439 278 }
Chris@439 279 }
Chris@439 280
Chris@439 281 if (!foundSomething) {
Chris@439 282 cerr << "PluginRDFIndexer::indexURL: NOTE: Document at <"
Chris@439 283 << urlString.toStdString()
Chris@439 284 << "> does not sufficiently describe any plugins" << endl;
Chris@439 285 }
Chris@439 286
Chris@439 287 return addedSomething;
Chris@439 288 }
Chris@439 289
Chris@439 290
Chris@439 291