/**
* Creates the files containing all paths for all concepts up to a certain depth
*
* @param kb
* @param concepts
* @throws IOException
*/
public static void createIndexFiles(KnowledgeBase kb, Collection<String> concepts)
throws IOException {
Iterator<String> conceptIterator = kb.getConceptIterator();
String outputDir =
KnowledgeConfiguration.getInstance().getKnowledgeBasePathIndexDir(kb) + TXT_FILE_DIR;
if (!Files.isDirectoryEmpty(new File(outputDir))) {
log.warn(outputDir + " IS NOT EMPTY, BAILING OUT ...");
return;
}
int maxSearchDepth = getMaxSearchDepth();
DirectoryFileManager dirManager = new DirectoryFileManager(outputDir, 1000);
log.info("CREATING THE INDEXABLE PATH FILES FOR: " + kb.name() + " INTO " + outputDir);
int idx = 0;
List<String> goodConcepts = new ArrayList<String>();
while (conceptIterator.hasNext()) {
String concept = conceptIterator.next();
if (concepts.isEmpty() || concepts.contains(concept)) goodConcepts.add(concept);
idx++;
if ((idx % 100000) == 0) log.info("ITERATED THROUGH " + idx + " CONCEPTS SO FAR ...");
}
List<List<String>> splitConcepts =
Collections.split(goodConcepts, goodConcepts.size() / NSPLITS);
int nThreads = splitConcepts.size();
ExecutorService threadExecutor = Executors.newFixedThreadPool(nThreads);
CountDownLatch doneSignal = new CountDownLatch(nThreads);
for (int i = 0; i < nThreads; i++) {
ConceptIndexer worker =
new ConceptIndexer(splitConcepts.get(i), i, dirManager, kb, maxSearchDepth, doneSignal);
threadExecutor.execute(worker);
}
Timer timer = new Timer();
try {
doneSignal.await();
threadExecutor.shutdown();
} catch (InterruptedException ie) {
ie.printStackTrace();
}
timer.tick("TO CREATE INDEXABLE FILES");
}
public void getConcepts() throws IOException {
int counter = 0;
Timer timer = new Timer();
for (String concept : concepts) {
Set<List<String>> paths = kb.getAllPathsFrom(concept, maxSearchDepth);
FileWriter writerFrom = dirManager.getFileWriter(concept + ".txt");
// saves the paths starting from the concept
for (List<String> path : paths) writerFrom.write(Strings.join(path) + "\n");
writerFrom.close();
if ((counter % 100) == 0)
timer.tick(
"\t[WORKER] "
+ id
+ " TO CREATE INDEXABLE FILES FOR "
+ counter
+ " CONCEPTS SO FAR ... ");
counter++;
}
}
