public void map(
LongWritable key,
TrecDocument doc,
OutputCollector<Text, IntWritable> output,
Reporter reporter)
throws IOException {
reporter.incrCounter(Count.DOCS, 1);
docid.set(doc.getDocid());
one.set(docMapping.getDocno(doc.getDocid()));
output.collect(docid, one);
}
public void configure(JobConf job) {
try {
Path[] localFiles = DistributedCache.getLocalCacheFiles(job);
// Instead of hard-coding the actual concrete DocnoMapping class, have the name of the
// class passed in as a property; this makes the mapper more general.
docMapping = (DocnoMapping) Class.forName(job.get("DocnoMappingClass")).newInstance();
// Simply assume that the mappings file is the only file in the distributed cache.
docMapping.loadMapping(localFiles[0], FileSystem.getLocal(job));
} catch (Exception e) {
e.printStackTrace();
throw new RuntimeException("Error initializing DocnoMapping!");
}
}
