public static DepContextCounts importFromReader(BufferedReader in) throws IOException {
Helper.report("[ContextCounts] Importing context word counts...");
DepContextCounts dmc = new DepContextCounts();
String line;
while ((line = in.readLine()) != null) {
if (line.startsWith("<contextcounts")) {
Matcher matcher = contextCountsPattern.matcher(line);
if (matcher.find()) { // ignore first entry: corpus name
Corpus.setName(matcher.group(1));
}
} else if (line.startsWith("<deprelation")) {
Matcher matcher = depRelationPattern.matcher(line);
if (matcher.find()) { // ignore first entry: corpus name
String depRelationString = matcher.group(1);
importDepRelationCounts(in, dmc, depRelationString);
}
} else if (line.equals("</contextcounts>")) {
break;
}
}
Helper.report("[ContextCounts] ...Finished importing context word counts.");
return dmc;
}
// marginalise over all corpus files using threads
public synchronized void count() {
Helper.report("[ContextCounter] Counting over all corpus files...");
File corpusFolder = new File(DepNeighbourhoodSpace.getProjectFolder(), Corpus.getFolderName());
String[] corpusFilenames = corpusFolder.list();
Arrays.sort(corpusFilenames);
// run each dep marginaliser thread
for (String corpusFilename : corpusFilenames) {
DepContextCounterThread ccThread =
new DepContextCounterThread(
this, corpusFilename, new File(corpusFolder, corpusFilename), amountOfSentences);
threads.add(ccThread);
(new Thread(ccThread)).start();
}
// wait for all threads to finish
try {
while (!threads.isEmpty()) {
wait();
}
} catch (InterruptedException e) {
}
Helper.report("[ContextCounter] ...Finished counting over all corpus files...");
}
public void saveToFile(File contextCountsFile) {
try {
Helper.report("[ContextCounts] Saving counts to \"" + contextCountsFile + "\"...");
BufferedWriter out = Helper.getFileWriter(contextCountsFile);
saveToWriter(out);
out.close();
Helper.report(
"[ContextCounts] ...Finished saving counts to \"" + contextCountsFile + "\"...");
} catch (IOException e) {
e.printStackTrace();
}
}
public static DepContextCounts importFromFile(File file) {
DepContextCounts dcc = null;
try {
BufferedReader in = Helper.getFileReader(file);
dcc = importFromReader(in);
in.close();
} catch (IOException e) {
e.printStackTrace();
}
return dcc;
}
public void saveToWriter(BufferedWriter out) throws IOException {
out.write("<contextcounts corpus=\"" + DepNeighbourhoodSpace.getName() + "\">\n");
for (String depRelation : depRelationWordCountMap.keySet()) {
Helper.report(
"[ContextCounts] Writing counts for dep relation \"" + depRelation + "\" to file");
out.write("<deprelation name=\"" + depRelation + "\">\n");
TreeSet<Entry<String, Long>> sortedWordCountMap = getSortedMap(depRelation);
while (!sortedWordCountMap.isEmpty()) {
Entry<String, Long> topWordCountPair = sortedWordCountMap.pollLast();
out.write(topWordCountPair.getValue() + "\t" + topWordCountPair.getKey() + "\n");
}
out.write("</deprelation>\n");
}
out.write("</contextcounts>\n");
}
public static void importDepRelationCounts(
BufferedReader in, DepContextCounts dmc, String depRelationString) throws IOException {
Helper.report(
"[ContextCounts] Importing context word counts for dep relation \""
+ depRelationString
+ "\"...");
String line;
while ((line = in.readLine()) != null) {
if (line.equals("</deprelation>")) {
break;
} else {
String[] entries = line.split("\t");
long count = Long.parseLong(entries[0]);
String contextWord = entries[1];
dmc.setCount(depRelationString, contextWord, count);
}
}
}
