annotate rdf/PluginRDFIndexer.cpp @ 467:c9b055f84326

* make use of CachedFile (untested)
author Chris Cannam
date Mon, 27 Oct 2008 18:15:20 +0000
parents 2019d89ebcf9
children 0dc158c1a6c9
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@467 20 //!!!#include "data/fileio/FileSource.h"
Chris@467 21 #include "data/fileio/CachedFile.h"
Chris@461 22 #include "data/fileio/PlaylistFileReader.h"
Chris@439 23 #include "plugin/PluginIdentifier.h"
Chris@439 24
Chris@457 25 #include "base/Profiler.h"
Chris@457 26
Chris@439 27 #include <vamp-sdk/PluginHostAdapter.h>
Chris@439 28
Chris@439 29 #include <QFileInfo>
Chris@439 30 #include <QDir>
Chris@439 31 #include <QUrl>
Chris@461 32 #include <QDateTime>
Chris@461 33 #include <QSettings>
Chris@461 34 #include <QFile>
Chris@439 35
Chris@439 36 #include <iostream>
Chris@439 37 using std::cerr;
Chris@439 38 using std::endl;
Chris@439 39 using std::vector;
Chris@439 40 using std::string;
Chris@439 41 using Vamp::PluginHostAdapter;
Chris@439 42
Chris@439 43 PluginRDFIndexer *
Chris@439 44 PluginRDFIndexer::m_instance = 0;
Chris@439 45
Chris@439 46 PluginRDFIndexer *
Chris@439 47 PluginRDFIndexer::getInstance()
Chris@439 48 {
Chris@439 49 if (!m_instance) m_instance = new PluginRDFIndexer();
Chris@439 50 return m_instance;
Chris@439 51 }
Chris@439 52
Chris@439 53 PluginRDFIndexer::PluginRDFIndexer()
Chris@439 54 {
Chris@439 55 vector<string> paths = PluginHostAdapter::getPluginPath();
Chris@439 56
Chris@439 57 QStringList filters;
Chris@439 58 filters << "*.n3";
Chris@439 59 filters << "*.N3";
Chris@439 60 filters << "*.rdf";
Chris@439 61 filters << "*.RDF";
Chris@439 62
Chris@439 63 // Search each Vamp plugin path for a .rdf file that either has
Chris@439 64 // name "soname", "soname:label" or "soname/label" plus RDF
Chris@439 65 // extension. Use that order of preference, and prefer n3 over
Chris@439 66 // rdf extension.
Chris@439 67
Chris@439 68 for (vector<string>::const_iterator i = paths.begin(); i != paths.end(); ++i) {
Chris@439 69
Chris@439 70 QDir dir(i->c_str());
Chris@439 71 if (!dir.exists()) continue;
Chris@439 72
Chris@439 73 QStringList entries = dir.entryList
Chris@439 74 (filters, QDir::Files | QDir::Readable);
Chris@439 75
Chris@439 76 for (QStringList::const_iterator j = entries.begin();
Chris@439 77 j != entries.end(); ++j) {
Chris@439 78 QFileInfo fi(dir.filePath(*j));
Chris@439 79 indexFile(fi.absoluteFilePath());
Chris@439 80 }
Chris@439 81
Chris@439 82 QStringList subdirs = dir.entryList
Chris@439 83 (QDir::AllDirs | QDir::NoDotAndDotDot | QDir::Readable);
Chris@439 84
Chris@439 85 for (QStringList::const_iterator j = subdirs.begin();
Chris@439 86 j != subdirs.end(); ++j) {
Chris@439 87 QDir subdir(dir.filePath(*j));
Chris@439 88 if (subdir.exists()) {
Chris@439 89 entries = subdir.entryList
Chris@439 90 (filters, QDir::Files | QDir::Readable);
Chris@439 91 for (QStringList::const_iterator k = entries.begin();
Chris@439 92 k != entries.end(); ++k) {
Chris@439 93 QFileInfo fi(subdir.filePath(*k));
Chris@439 94 indexFile(fi.absoluteFilePath());
Chris@439 95 }
Chris@439 96 }
Chris@439 97 }
Chris@439 98 }
Chris@439 99 }
Chris@439 100
Chris@439 101 PluginRDFIndexer::~PluginRDFIndexer()
Chris@439 102 {
Chris@461 103 QMutexLocker locker(&m_mutex);
Chris@467 104 /*!!!
Chris@457 105 while (!m_sources.empty()) {
Chris@457 106 delete *m_sources.begin();
Chris@457 107 m_sources.erase(m_sources.begin());
Chris@439 108 }
Chris@467 109 */
Chris@439 110 }
Chris@439 111
Chris@461 112 bool
Chris@461 113 PluginRDFIndexer::indexConfiguredURLs()
Chris@461 114 {
Chris@461 115 std::cerr << "PluginRDFIndexer::indexConfiguredURLs" << std::endl;
Chris@461 116
Chris@461 117 QSettings settings;
Chris@461 118 settings.beginGroup("RDF");
Chris@461 119
Chris@461 120 QString indexKey("rdf-indices");
Chris@461 121 QStringList indices = settings.value(indexKey).toStringList();
Chris@461 122
Chris@461 123 for (int i = 0; i < indices.size(); ++i) {
Chris@461 124
Chris@461 125 QString index = indices[i];
Chris@461 126
Chris@461 127 std::cerr << "PluginRDFIndexer::indexConfiguredURLs: index url is "
Chris@461 128 << index.toStdString() << std::endl;
Chris@461 129
Chris@467 130 /*!!!
Chris@461 131 expireCacheMaybe(index);
Chris@461 132
Chris@461 133 FileSource indexSource(index, 0, FileSource::PersistentCache);
Chris@461 134 if (!indexSource.isAvailable()) continue;
Chris@461 135 indexSource.waitForData();
Chris@467 136 */
Chris@467 137
Chris@467 138 CachedFile cf(index);
Chris@467 139 if (!cf.isOK()) continue;
Chris@467 140
Chris@467 141 FileSource indexSource(cf.getLocalFilename());
Chris@461 142
Chris@461 143 PlaylistFileReader reader(indexSource);
Chris@461 144 if (!reader.isOK()) continue;
Chris@461 145
Chris@461 146 PlaylistFileReader::Playlist list = reader.load();
Chris@461 147 for (PlaylistFileReader::Playlist::const_iterator j = list.begin();
Chris@461 148 j != list.end(); ++j) {
Chris@461 149 std::cerr << "PluginRDFIndexer::indexConfiguredURLs: url is "
Chris@461 150 << j->toStdString() << std::endl;
Chris@461 151 indexURL(*j);
Chris@461 152 }
Chris@461 153 }
Chris@461 154
Chris@461 155 QString urlListKey("rdf-urls");
Chris@461 156 QStringList urls = settings.value(urlListKey).toStringList();
Chris@461 157
Chris@461 158 for (int i = 0; i < urls.size(); ++i) {
Chris@461 159 indexURL(urls[i]);
Chris@461 160 }
Chris@461 161
Chris@461 162 settings.endGroup();
Chris@461 163 return true;
Chris@461 164 }
Chris@461 165
Chris@439 166 QString
Chris@439 167 PluginRDFIndexer::getURIForPluginId(QString pluginId)
Chris@439 168 {
Chris@461 169 QMutexLocker locker(&m_mutex);
Chris@461 170
Chris@439 171 if (m_idToUriMap.find(pluginId) == m_idToUriMap.end()) return "";
Chris@439 172 return m_idToUriMap[pluginId];
Chris@439 173 }
Chris@439 174
Chris@439 175 QString
Chris@439 176 PluginRDFIndexer::getIdForPluginURI(QString uri)
Chris@439 177 {
Chris@461 178 QMutexLocker locker(&m_mutex);
Chris@461 179
Chris@439 180 if (m_uriToIdMap.find(uri) == m_uriToIdMap.end()) {
Chris@439 181
Chris@439 182 // Haven't found this uri referenced in any document on the
Chris@439 183 // local filesystem; try resolving the pre-fragment part of
Chris@439 184 // the uri as a document URL and reading that if possible.
Chris@439 185
Chris@439 186 // Because we may want to refer to this document again, we
Chris@439 187 // cache it locally if it turns out to exist.
Chris@439 188
Chris@439 189 cerr << "PluginRDFIndexer::getIdForPluginURI: NOTE: Failed to find a local RDF document describing plugin <" << uri.toStdString() << ">: attempting to retrieve one remotely by guesswork" << endl;
Chris@439 190
Chris@439 191 QString baseUrl = QUrl(uri).toString(QUrl::RemoveFragment);
Chris@439 192
Chris@457 193 indexURL(baseUrl);
Chris@439 194
Chris@439 195 if (m_uriToIdMap.find(uri) == m_uriToIdMap.end()) {
Chris@439 196 m_uriToIdMap[uri] = "";
Chris@439 197 }
Chris@439 198 }
Chris@439 199
Chris@439 200 return m_uriToIdMap[uri];
Chris@439 201 }
Chris@439 202
Chris@439 203 QString
Chris@439 204 PluginRDFIndexer::getDescriptionURLForPluginId(QString pluginId)
Chris@439 205 {
Chris@461 206 QMutexLocker locker(&m_mutex);
Chris@461 207
Chris@439 208 if (m_idToDescriptionMap.find(pluginId) == m_idToDescriptionMap.end()) return "";
Chris@439 209 return m_idToDescriptionMap[pluginId];
Chris@439 210 }
Chris@439 211
Chris@439 212 QString
Chris@439 213 PluginRDFIndexer::getDescriptionURLForPluginURI(QString uri)
Chris@439 214 {
Chris@461 215 QMutexLocker locker(&m_mutex);
Chris@461 216
Chris@439 217 QString id = getIdForPluginURI(uri);
Chris@439 218 if (id == "") return "";
Chris@439 219 return getDescriptionURLForPluginId(id);
Chris@439 220 }
Chris@439 221
Chris@456 222 QStringList
Chris@456 223 PluginRDFIndexer::getIndexedPluginIds()
Chris@456 224 {
Chris@461 225 QMutexLocker locker(&m_mutex);
Chris@461 226
Chris@456 227 QStringList ids;
Chris@456 228 for (StringMap::const_iterator i = m_idToDescriptionMap.begin();
Chris@456 229 i != m_idToDescriptionMap.end(); ++i) {
Chris@456 230 ids.push_back(i->first);
Chris@456 231 }
Chris@456 232 return ids;
Chris@456 233 }
Chris@456 234
Chris@439 235 bool
Chris@439 236 PluginRDFIndexer::indexFile(QString filepath)
Chris@439 237 {
Chris@439 238 QUrl url = QUrl::fromLocalFile(filepath);
Chris@439 239 QString urlString = url.toString();
Chris@439 240 return indexURL(urlString);
Chris@439 241 }
Chris@467 242 /*!!!
Chris@461 243 void
Chris@461 244 PluginRDFIndexer::expireCacheMaybe(QString urlString)
Chris@461 245 {
Chris@461 246 QString cacheFile = FileSource::getPersistentCacheFilePath(urlString);
Chris@461 247
Chris@461 248 QSettings settings;
Chris@461 249 settings.beginGroup("RDF");
Chris@461 250
Chris@461 251 QString key("rdf-expiry-times");
Chris@461 252
Chris@461 253 QMap<QString, QVariant> expiryMap = settings.value(key).toMap();
Chris@461 254 QDateTime lastExpiry = expiryMap[urlString].toDateTime();
Chris@461 255
Chris@461 256 if (!QFileInfo(cacheFile).exists()) {
Chris@461 257 expiryMap[urlString] = QDateTime::currentDateTime();
Chris@461 258 settings.setValue(key, expiryMap);
Chris@461 259 settings.endGroup();
Chris@461 260 return;
Chris@461 261 }
Chris@461 262
Chris@461 263 if (!lastExpiry.isValid() ||
Chris@461 264 (lastExpiry.addDays(2) < QDateTime::currentDateTime())) {
Chris@461 265
Chris@461 266 std::cerr << "Expiring old cache file " << cacheFile.toStdString()
Chris@461 267 << std::endl;
Chris@461 268
Chris@461 269 if (QFile(cacheFile).remove()) {
Chris@461 270
Chris@461 271 expiryMap[urlString] = QDateTime::currentDateTime();
Chris@461 272 settings.setValue(key, expiryMap);
Chris@461 273 }
Chris@461 274 }
Chris@461 275
Chris@461 276 settings.endGroup();
Chris@461 277 }
Chris@467 278 */
Chris@439 279 bool
Chris@439 280 PluginRDFIndexer::indexURL(QString urlString)
Chris@439 281 {
Chris@457 282 Profiler profiler("PluginRDFIndexer::indexURL");
Chris@457 283
Chris@461 284 std::cerr << "PluginRDFIndexer::indexURL(" << urlString.toStdString() << ")" << std::endl;
Chris@461 285
Chris@461 286 QMutexLocker locker(&m_mutex);
Chris@461 287
Chris@457 288 QString localString = urlString;
Chris@457 289
Chris@457 290 if (FileSource::isRemote(urlString) &&
Chris@457 291 FileSource::canHandleScheme(urlString)) {
Chris@457 292
Chris@461 293 //!!! how do we avoid hammering the server if it doesn't have
Chris@461 294 //!!! the file, and/or the network if it can't get through?
Chris@461 295
Chris@467 296 CachedFile cf(urlString);
Chris@467 297 if (!cf.isOK()) {
Chris@467 298 return false;
Chris@467 299 }
Chris@467 300
Chris@467 301 localString = cf.getLocalFilename();
Chris@467 302 /*!!!
Chris@461 303 expireCacheMaybe(urlString);
Chris@461 304
Chris@460 305 FileSource *source = new FileSource
Chris@460 306 (urlString, 0, FileSource::PersistentCache);
Chris@457 307 if (!source->isAvailable()) {
Chris@457 308 delete source;
Chris@457 309 return false;
Chris@457 310 }
Chris@457 311 source->waitForData();
Chris@457 312 localString = QUrl::fromLocalFile(source->getLocalFilename()).toString();
Chris@457 313 m_sources.insert(source);
Chris@467 314 */
Chris@457 315 }
Chris@457 316
Chris@439 317 // cerr << "PluginRDFIndexer::indexURL: url = <" << urlString.toStdString() << ">" << endl;
Chris@439 318
Chris@439 319 SimpleSPARQLQuery query
Chris@439 320 (QString
Chris@439 321 (
Chris@439 322 " PREFIX vamp: <http://purl.org/ontology/vamp/> "
Chris@439 323
Chris@439 324 " SELECT ?plugin ?library_id ?plugin_id "
Chris@439 325 " FROM <%1> "
Chris@439 326
Chris@439 327 " WHERE { "
Chris@439 328 " ?plugin a vamp:Plugin . "
Chris@439 329
Chris@439 330 // Make the identifier and library parts optional, so
Chris@439 331 // that we can check and report helpfully if one or both
Chris@439 332 // is absent instead of just getting no results
Chris@439 333
Chris@440 334 //!!! No -- because of rasqal's inability to correctly
Chris@440 335 // handle more than one OPTIONAL graph in a query, let's
Chris@440 336 // make identifier compulsory after all
Chris@440 337 //" OPTIONAL { ?plugin vamp:identifier ?plugin_id } . "
Chris@440 338
Chris@440 339 " ?plugin vamp:identifier ?plugin_id . "
Chris@439 340
Chris@439 341 " OPTIONAL { "
Chris@439 342 " ?library a vamp:PluginLibrary ; "
Chris@439 343 " vamp:available_plugin ?plugin ; "
Chris@439 344 " vamp:identifier ?library_id "
Chris@439 345 " } "
Chris@439 346 " } "
Chris@439 347 )
Chris@457 348 .arg(localString));
Chris@439 349
Chris@439 350 SimpleSPARQLQuery::ResultList results = query.execute();
Chris@439 351
Chris@439 352 if (!query.isOK()) {
Chris@439 353 cerr << "ERROR: PluginRDFIndexer::indexURL: ERROR: Failed to index document at <"
Chris@439 354 << urlString.toStdString() << ">: "
Chris@439 355 << query.getErrorString().toStdString() << endl;
Chris@439 356 return false;
Chris@439 357 }
Chris@439 358
Chris@439 359 if (results.empty()) {
Chris@439 360 cerr << "PluginRDFIndexer::indexURL: NOTE: Document at <"
Chris@439 361 << urlString.toStdString()
Chris@439 362 << "> does not describe any vamp:Plugin resources" << endl;
Chris@439 363 return false;
Chris@439 364 }
Chris@439 365
Chris@439 366 bool foundSomething = false;
Chris@439 367 bool addedSomething = false;
Chris@439 368
Chris@439 369 for (SimpleSPARQLQuery::ResultList::iterator i = results.begin();
Chris@439 370 i != results.end(); ++i) {
Chris@439 371
Chris@439 372 QString pluginUri = (*i)["plugin"].value;
Chris@439 373 QString soname = (*i)["library_id"].value;
Chris@439 374 QString identifier = (*i)["plugin_id"].value;
Chris@439 375
Chris@439 376 if (identifier == "") {
Chris@439 377 cerr << "PluginRDFIndexer::indexURL: NOTE: Document at <"
Chris@439 378 << urlString.toStdString()
Chris@439 379 << "> fails to define any vamp:identifier for plugin <"
Chris@439 380 << pluginUri.toStdString() << ">"
Chris@439 381 << endl;
Chris@439 382 continue;
Chris@439 383 }
Chris@439 384 if (soname == "") {
Chris@439 385 cerr << "PluginRDFIndexer::indexURL: NOTE: Document at <"
Chris@439 386 << urlString.toStdString() << "> does not associate plugin <"
Chris@439 387 << pluginUri.toStdString() << "> with any implementation library"
Chris@439 388 << endl;
Chris@439 389 continue;
Chris@439 390 }
Chris@439 391 /*
Chris@439 392 cerr << "PluginRDFIndexer::indexURL: Document for plugin \""
Chris@439 393 << soname.toStdString() << ":" << identifier.toStdString()
Chris@439 394 << "\" (uri <" << pluginUri.toStdString() << ">) is at url <"
Chris@439 395 << urlString.toStdString() << ">" << endl;
Chris@439 396 */
Chris@439 397 QString pluginId = PluginIdentifier::createIdentifier
Chris@439 398 ("vamp", soname, identifier);
Chris@439 399
Chris@439 400 foundSomething = true;
Chris@439 401
Chris@439 402 if (m_idToDescriptionMap.find(pluginId) != m_idToDescriptionMap.end()) {
Chris@439 403 cerr << "PluginRDFIndexer::indexURL: NOTE: Plugin id \""
Chris@439 404 << pluginId.toStdString() << "\", described in document at <"
Chris@439 405 << urlString.toStdString()
Chris@439 406 << ">, has already been described in document <"
Chris@439 407 << m_idToDescriptionMap[pluginId].toStdString()
Chris@439 408 << ">: ignoring this new description" << endl;
Chris@439 409 continue;
Chris@439 410 }
Chris@439 411
Chris@439 412 m_idToDescriptionMap[pluginId] = urlString;
Chris@439 413 m_idToUriMap[pluginId] = pluginUri;
Chris@439 414
Chris@439 415 addedSomething = true;
Chris@439 416
Chris@439 417 if (pluginUri != "") {
Chris@439 418 if (m_uriToIdMap.find(pluginUri) != m_uriToIdMap.end()) {
Chris@439 419 cerr << "PluginRDFIndexer::indexURL: WARNING: Found multiple plugins with the same URI:" << endl;
Chris@439 420 cerr << " 1. Plugin id \"" << m_uriToIdMap[pluginUri].toStdString() << "\"" << endl;
Chris@439 421 cerr << " described in <" << m_idToDescriptionMap[m_uriToIdMap[pluginUri]].toStdString() << ">" << endl;
Chris@439 422 cerr << " 2. Plugin id \"" << pluginId.toStdString() << "\"" << endl;
Chris@439 423 cerr << " described in <" << urlString.toStdString() << ">" << endl;
Chris@439 424 cerr << "both claim URI <" << pluginUri.toStdString() << ">" << endl;
Chris@439 425 } else {
Chris@439 426 m_uriToIdMap[pluginUri] = pluginId;
Chris@439 427 }
Chris@439 428 }
Chris@439 429 }
Chris@439 430
Chris@439 431 if (!foundSomething) {
Chris@439 432 cerr << "PluginRDFIndexer::indexURL: NOTE: Document at <"
Chris@439 433 << urlString.toStdString()
Chris@439 434 << "> does not sufficiently describe any plugins" << endl;
Chris@439 435 }
Chris@439 436
Chris@439 437 return addedSomething;
Chris@439 438 }
Chris@439 439
Chris@439 440
Chris@439 441