@Override
public void map(
Writable key,
Indexable doc,
Mapper<Writable, Indexable, ContextPatternWritable, ContextPatternStatsWritable>.Context
context)
throws IOException, InterruptedException {
// set current document
mExtractor.setDocument(doc);
// extract example counts
while (mExtractor.getNextPair(mPair)) {
if (mPatternTarget) {
if (mExampleStats.containsKey(mPair.getPattern())) {
mExampleStats.increment(mPair.getPattern());
}
} else if (mContextTarget) {
if (mExampleStats.containsKey(mPair.getContext())) {
mExampleStats.increment(mPair.getContext());
}
}
// increment number of pairs
mTotalTerms++;
}
}
@Override
public void cleanup(
Mapper<Writable, Indexable, ContextPatternWritable, ContextPatternStatsWritable>.Context
context)
throws IOException, InterruptedException {
Configuration conf = context.getConfiguration();
String examplesPath = conf.get("Mavuno.ExtractGlobalStats.ExamplesPath", null);
mContext.clear();
mStats.clear();
if (mPatternTarget) {
mStats.globalPatternCount = mTotalTerms;
} else if (mContextTarget) {
mStats.globalContextCount = mTotalTerms;
}
context.write(mContext, mStats);
BufferedReader reader = MavunoUtils.getBufferedReader(conf, examplesPath);
long count = 0L;
String input;
while ((input = reader.readLine()) != null) {
String[] cols = input.split("\t");
mContext.setId(cols[ContextPatternWritable.ID_FIELD]);
mStats.weight = Double.parseDouble(cols[ContextPatternWritable.TOTAL_FIELDS]);
if (mPatternTarget) {
mContext.setPattern(cols[ContextPatternWritable.PATTERN_FIELD]);
mStats.globalPatternCount = mExampleStats.get(mContext.getPattern());
count = mStats.globalPatternCount;
} else if (mContextTarget) {
mContext.setContext(cols[ContextPatternWritable.CONTEXT_FIELD]);
mStats.globalContextCount = mExampleStats.get(mContext.getContext());
count = mStats.globalContextCount;
}
if (count > 0) {
context.write(mContext, mStats);
}
}
reader.close();
}
@Override
public void reduce(
ContextPatternWritable key,
Iterable<ContextPatternStatsWritable> values,
Reducer<
ContextPatternWritable,
ContextPatternStatsWritable,
ContextPatternWritable,
ContextPatternStatsWritable>
.Context
context)
throws IOException, InterruptedException {
Configuration conf = context.getConfiguration();
mResult.zero();
for (ContextPatternStatsWritable c : values) {
mResult.increment(c);
}
// write total terms
if (key.getId().equals(ContextPatternWritable.ASTERISK)) {
if (!MavunoUtils.pathExists(conf, mTotalTermsPath)) {
BufferedWriter writer = MavunoUtils.getBufferedWriter(conf, mTotalTermsPath);
if (mPatternTarget) {
writer.write(Long.toString(mResult.globalPatternCount));
} else if (mContextTarget) {
writer.write(Long.toString(mResult.globalContextCount));
}
writer.close();
}
return;
}
context.write(key, mResult);
}