String pwsimFile = job.get("PwsimPairs");
for (Path localFile : localFiles) {
if (localFile.toString().contains(getFilename(pwsimFile))) {
SequenceFile.Reader reader = new SequenceFile.Reader(FileSystem.getLocal(job), localFile, job);
PairOfInts key = (PairOfInts) reader.getKeyClass().newInstance();
IntWritable value = (IntWritable) reader.getValueClass().newInstance();
int cnt = 0;
while (reader.next(key, value)) {
int fDocno = key.getRightElement();
// fDocno -= 1000000000;
int eDocno = key.getLeftElement();
if(langID == CLIRUtils.E){
if(!pwsimMapping.containsKey(eDocno)){
pwsimMapping.put(eDocno, new ArrayListOfIntsWritable());
}
pwsimMapping.get(eDocno).add(fDocno); // we add 1000000000 to foreign docnos to distinguish them during pwsim algo