@Override
public void stage(MultiStagedJob mjob) {
mjob.removeIntermediate(true);
SequenceFileStage<Text, BytesWritable, Text, LongWritable, LongWritable, Text> collateWords =
new SequenceFileStage<Text, BytesWritable, Text, LongWritable, LongWritable, Text>() {
@Override
public void setup(Job job) {
job.getConfiguration().setInt(WORDCOUNT_THRESH, wordCountThreshold);
job.getConfiguration().setInt(WORDCOUNT_TIMETHRESH, wordTimeThreshold);
job.setNumReduceTasks(1);
}
@Override
public Class<? extends Mapper<Text, BytesWritable, Text, LongWritable>> mapper() {
return WordIndex.Map.class;
}
@Override
public Class<? extends Reducer<Text, LongWritable, LongWritable, Text>> reducer() {
return WordIndex.Reduce.class;
}
@Override
public String outname() {
return "words-collated";
}
};
SequenceFileTextStage<LongWritable, Text, LongWritable, Text, NullWritable, Text> sortedWords =
new SequenceFileTextStage<LongWritable, Text, LongWritable, Text, NullWritable, Text>() {
@Override
public void setup(Job job) {
job.getConfiguration().setInt(WORDCOUNT_TOPN, topNWords);
job.setSortComparatorClass(LongWritable.DecreasingComparator.class);
job.setNumReduceTasks(1);
}
@Override
public Class<? extends Reducer<LongWritable, Text, NullWritable, Text>> reducer() {
return WordIndexSort.Reduce.class;
}
@Override
public String outname() {
return "words";
}
};
mjob.queueStage(collateWords);
mjob.queueStage(sortedWords);
}
@Override
public void write(HadoopTwitterTokenToolOptions opts, TwitterTokenMode completedMode)
throws Exception {
MultiStagedJob stages;
stages =
new MultiStagedJob(
HadoopToolsUtil.getInputPaths(
completedMode.finalOutput(opts), CountWordsAcrossTimeperiod.WORDCOUNT_DIR),
HadoopToolsUtil.getOutputPath(outputPath),
opts.getArgs());
stages.queueStage(new TimeWordJacardIndex());
stages.runAll();
}
@Override
public void write(HadoopTwitterTokenToolOptions opts, TwitterTokenMode completedMode)
throws Exception {
this.stages =
new MultiStagedJob(
HadoopToolsUtil.getInputPaths(
completedMode.finalOutput(opts), CountWordsAcrossTimeperiod.WORDCOUNT_DIR),
HadoopToolsUtil.getOutputPath(outputPath),
opts.getArgs());
// Three stage process
// 1a. Write all the words (word per line)
new WordIndex().stage(stages);
final Path wordIndex = stages.runAll();
HashMap<String, IndependentPair<Long, Long>> wordCountLines =
WordIndex.readWordCountLines(wordIndex.toString(), "");
StatsWordMatch matches = new StatsWordMatch();
for (Entry<String, IndependentPair<Long, Long>> entry : wordCountLines.entrySet()) {
String word = entry.getKey();
IndependentPair<Long, Long> countLine = entry.getValue();
Long count = countLine.firstObject();
matches.updateStats(word, count);
}
System.out.println(matches);
}
