Revision 46:c8b777862198 common/FeatureFileIndex.cpp
| common/FeatureFileIndex.cpp | ||
|---|---|---|
| 89 | 89 |
} |
| 90 | 90 |
updateIndex(); |
| 91 | 91 |
|
| 92 |
QSet<Uri> fileUris; |
|
| 93 |
|
|
| 94 |
// The same file may be referred to with more than one URI; we |
|
| 95 |
// want to load any or all of: the URI in our file object; encoded |
|
| 96 |
// version of same; and any other file that is recorded as having |
|
| 97 |
// the same hash (i.e. it is the same file). |
|
| 98 |
|
|
| 99 |
fileUris.insert(tf->uri()); |
|
| 100 |
|
|
| 101 |
// and again with encoded version of file URI |
|
| 102 |
QByteArray enc = QUrl(tf->uri().toString()).toEncoded(); |
|
| 103 |
fileUris.insert(Uri(QString::fromUtf8(enc))); |
|
| 104 |
|
|
| 105 |
// and again with anything else having the same hash |
|
| 106 |
if (tf->hash() != "") {
|
|
| 107 |
Triple pattern(Node(), "foaf:sha1", Node(Node::Literal, tf->hash())); |
|
| 108 |
Triples results = m_index->match(pattern); |
|
| 109 |
std::cerr << "FeatureFileIndex::loadFor: " << results.size() << " audio file(s) found with hash " << tf->hash().toStdString() << std::endl; |
|
| 110 |
foreach (Triple t, results) {
|
|
| 111 |
fileUris.insert(Uri(t.a.value)); |
|
| 112 |
} |
|
| 113 |
} |
|
| 114 |
|
|
| 115 |
foreach (Uri u, fileUris) {
|
|
| 116 |
loadFor(tf->uri(), u, tf->hash(), store); |
|
| 117 |
} |
|
| 118 |
} |
|
| 119 |
|
|
| 120 |
bool |
|
| 121 |
FeatureFileIndex::loadFor(Uri canonicalUri, Uri afuri, |
|
| 122 |
QString hash, Store *store) |
|
| 123 |
{
|
|
| 92 | 124 |
// The AudioFile object has a URI and a hash. Feature files |
| 93 | 125 |
// generated for this AudioFile should ideally have a matching |
| 94 | 126 |
// hash; if they have no hash, then the URI should match. If the |
| 95 | 127 |
// hash is present in the feature file but does not match, then it |
| 96 | 128 |
// cannot be the right track even if the URI matches. |
| 97 | 129 |
|
| 98 |
Triple t(Node(), "foaf:primaryTopic", tf->uri());
|
|
| 99 |
Triples results = m_index->match(t);
|
|
| 100 |
std::cerr << "FeatureFileIndex::loadFor: " << results.size() << " feature file(s) for audio file " << tf->uri() << std::endl;
|
|
| 130 |
Triple pattern(Node(), "foaf:primaryTopic", afuri);
|
|
| 131 |
Triples results = m_index->match(pattern);
|
|
| 132 |
std::cerr << "FeatureFileIndex::loadFor: " << results.size() << " feature file(s) for audio file " << afuri << std::endl;
|
|
| 101 | 133 |
|
| 102 |
t = Triple(Node(), "foaf:primaryTopic", |
|
| 103 |
Uri(QString::fromUtf8(QUrl(tf->uri().toString()).toEncoded()))); |
|
| 104 |
Triples moreResults = m_index->match(t); |
|
| 105 |
std::cerr << "FeatureFileIndex::loadFor: " << moreResults.size() << " feature file(s) for audio file " << t.c << std::endl; |
|
| 134 |
bool loadedSomething = false; |
|
| 106 | 135 |
|
| 107 |
//!!! what's the right approach here? |
|
| 108 |
|
|
| 109 |
if (results.empty() && moreResults.empty()) {
|
|
| 110 |
return; |
|
| 136 |
foreach (Triple t, results) {
|
|
| 137 |
try {
|
|
| 138 |
BasicStore *b = BasicStore::load(QUrl(t.a.value)); |
|
| 139 |
Triples ts = b->match |
|
| 140 |
(Triple(afuri, "a", m_index->expand("mo:AudioFile")));
|
|
| 141 |
std::cerr << "FeatureFileIndex::loadFor: feature file " |
|
| 142 |
<< t.a << " has " << ts.size() << " type node(s) for this audio file" << std::endl; |
|
| 143 |
bool someGood = false; |
|
| 144 |
foreach (Triple t, ts) {
|
|
| 145 |
bool good = true; |
|
| 146 |
if (hash != "") {
|
|
| 147 |
Triples hashts = b->match |
|
| 148 |
(Triple(afuri, m_index->expand("foaf:sha1"), Node()));
|
|
| 149 |
std::cerr << "FeatureFileIndex::loadFor: feature file " |
|
| 150 |
<< t.a << " has " << hashts.size() << " hashes for this file" << std::endl; |
|
| 151 |
if (!hashts.empty()) {
|
|
| 152 |
good = false; |
|
| 153 |
foreach (Triple hasht, hashts) {
|
|
| 154 |
if (hasht.c.value == hash) {
|
|
| 155 |
std::cerr << "Hash " << hasht.c << " matches our hash " << hash.toStdString() << std::endl; |
|
| 156 |
good = true; |
|
| 157 |
break; |
|
| 158 |
} |
|
| 159 |
} |
|
| 160 |
if (!good) {
|
|
| 161 |
std::cerr << "(no hash matches, eliminating file)" << std::endl; |
|
| 162 |
} |
|
| 163 |
} else {
|
|
| 164 |
std::cerr << "(so cannot eliminate file via hash)" << std::endl; |
|
| 165 |
} |
|
| 166 |
} |
|
| 167 |
if (good) {
|
|
| 168 |
std::cerr << "...going to import this one" << std::endl; |
|
| 169 |
someGood = true; |
|
| 170 |
} |
|
| 171 |
} |
|
| 172 |
if (someGood) {
|
|
| 173 |
Triples all = b->match(Triple()); |
|
| 174 |
std::cerr << "Importing " << all.size() << " triple(s) into store" << std::endl; |
|
| 175 |
// Replace instances of the audio file URI with our |
|
| 176 |
// canonical URI (we want to make sure we're |
|
| 177 |
// associating these facts with our own URI for this |
|
| 178 |
// file, even if they originated from a different URI |
|
| 179 |
// with the same hash) |
|
| 180 |
Node from = Node(Node::URI, afuri.toString()); |
|
| 181 |
Node to = Node(Node::URI, canonicalUri.toString()); |
|
| 182 |
foreach (Triple t, all) {
|
|
| 183 |
if (t.a == from) t.a = to; |
|
| 184 |
if (t.c == from) t.c = to; |
|
| 185 |
store->add(t); |
|
| 186 |
} |
|
| 187 |
loadedSomething = true; |
|
| 188 |
} |
|
| 189 |
} catch (...) { }
|
|
| 111 | 190 |
} |
| 112 | 191 |
|
| 113 |
|
|
| 192 |
return loadedSomething; |
|
| 114 | 193 |
} |
| 115 | 194 |
|
| 116 | 195 |
void |
| ... | ... | |
| 125 | 204 |
QMutexLocker locker(&m_mutex); |
| 126 | 205 |
if (!m_index) return; |
| 127 | 206 |
|
| 207 |
std::cerr << "Generating index..." << std::endl; |
|
| 208 |
|
|
| 128 | 209 |
QDir featureDir; |
| 129 | 210 |
try {
|
| 130 | 211 |
QString s = getFeatureDirectoryName(); |
| ... | ... | |
| 147 | 228 |
|
| 148 | 229 |
std::cerr << "Saving index to " << m_indexFileName.toStdString() << std::endl; |
| 149 | 230 |
m_bs->save(m_indexFileName); |
| 231 |
|
|
| 232 |
std::cerr << "Done" << std::endl; |
|
| 150 | 233 |
} |
| 151 | 234 |
|
| 152 | 235 |
void |
| ... | ... | |
| 167 | 250 |
(Triple(Node(), "a", m_index->expand("mo:AudioFile")));
|
| 168 | 251 |
foreach (Triple t, ts) {
|
| 169 | 252 |
tx->add(Triple(Uri(fileUrl), "foaf:primaryTopic", t.a));; |
| 253 |
Triples hashts = b->match |
|
| 254 |
(Triple(t.a, m_index->expand("foaf:sha1"), Node()));
|
|
| 255 |
foreach (Triple hasht, hashts) {
|
|
| 256 |
tx->add(hasht); |
|
| 257 |
} |
|
| 170 | 258 |
} |
| 171 |
} catch (...) { }
|
|
| 259 |
} catch (std::exception &e) {
|
|
| 260 |
std::cerr << "Caught exception: \"" << e.what() << "\" while indexing " |
|
| 261 |
<< Uri(fileUrl) << ", skipping" << std::endl; |
|
| 262 |
} |
|
| 172 | 263 |
|
| 173 | 264 |
delete tx; |
| 174 | 265 |
} |
Also available in: Unified diff