package ivory.lsh.eval;
import java.io.IOException;
import java.net.URI;
import java.util.Iterator;
import java.util.TreeSet;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.conf.Configured;
import org.apache.hadoop.filecache.DistributedCache;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.IntWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapred.FileInputFormat;
import org.apache.hadoop.mapred.FileOutputFormat;
import org.apache.hadoop.mapred.JobClient;
import org.apache.hadoop.mapred.JobConf;
import org.apache.hadoop.mapred.MapReduceBase;
import org.apache.hadoop.mapred.Mapper;
import org.apache.hadoop.mapred.OutputCollector;
import org.apache.hadoop.mapred.Reducer;
import org.apache.hadoop.mapred.Reporter;
import org.apache.hadoop.mapred.SequenceFileInputFormat;
import org.apache.hadoop.mapred.lib.IdentityReducer;
import org.apache.hadoop.util.LineReader;
import org.apache.hadoop.util.Tool;
import org.apache.hadoop.util.ToolRunner;
import org.apache.log4j.Level;
import org.apache.log4j.Logger;
import edu.umd.cloud9.io.map.HMapIIW;
import edu.umd.cloud9.io.pair.PairOfInts;
@SuppressWarnings("deprecation")
public class FilterResults extends Configured implements Tool {
public static final String[] RequiredParameters = {};
private static final Logger sLogger = Logger.getLogger(FilterResults.class);
static enum mapoutput {
count
};
private static int printUsage() {
System.out
.println("usage: [input-path] [output-path] [sample-docnos] [threshold] [num-results]");
return -1;
}
public FilterResults() {
super();
}
/**
* Filter results that are not from sample and/or have distance more than specified in option
* Ivory.MaxHammingDistance. Reducer selects closest N pairs for each sample foreign-language
* document.
*
* @author ferhanture
*
*/
public static class MyMapperTopN extends MapReduceBase implements
Mapper<PairOfInts, IntWritable, IntWritable, PairOfInts> {
static Path[] localFiles;
HMapIIW samplesMap = null;
int maxDist;
public void configure(JobConf job) {
sLogger.setLevel(Level.INFO);
maxDist = job.getInt("Ivory.MaxHammingDistance", -1);
// read doc ids of sample into vectors
try {
localFiles = DistributedCache.getLocalCacheFiles(job);
} catch (Exception e) {
throw new RuntimeException("Error reading doc vectors!");
}
if (localFiles != null && localFiles.length > 0) {
samplesMap = new HMapIIW();
try {
LineReader reader = new LineReader(FileSystem.getLocal(job).open(localFiles[0]));
Text t = new Text();
while (reader.readLine(t) != 0) {
int docno = Integer.parseInt(t.toString());
sLogger.info(docno + " --> sample");
samplesMap.put(docno, 1);
}
reader.close();
} catch (IOException e1) {
}
sLogger.info(samplesMap.size() + " sampled");
} else {
sLogger.info("samples file not specified in local cache");
}
}
public void map(PairOfInts key, IntWritable value,
OutputCollector<IntWritable, PairOfInts> output, Reporter reporter) throws IOException {
int leftKey = key.getLeftElement(); // english docno
int rightKey = key.getRightElement(); // german docno
sLogger.debug(rightKey);
if (samplesMap == null || samplesMap.containsKey(rightKey)) {
if (maxDist == -1 || value.get() <= maxDist) {
output.collect(new IntWritable(rightKey), new PairOfInts(value.get(), leftKey));
// symmetric implementation. change when not desired.
// output.collect(new IntWritable(leftKey), new PairOfInts(value.get(),rightKey));
}
}
}
}
public static class MyReducerTopN extends MapReduceBase implements
Reducer<IntWritable, PairOfInts, IntWritable, PairOfInts> {
int numResults;
TreeSet<PairOfInts> list = new TreeSet<PairOfInts>();
public void configure(JobConf conf) {
numResults = conf.getInt("Ivory.NumResults", -1);
sLogger.info("numResults");
}
public void reduce(IntWritable key, Iterator<PairOfInts> values,
OutputCollector<IntWritable, PairOfInts> output, Reporter reporter) throws IOException {
list.clear();
while (values.hasNext()) {
PairOfInts p = values.next();
list.add(new PairOfInts(p.getLeftElement(), p.getRightElement()));
reporter.incrCounter(mapoutput.count, 1);
}
int cntr = 0;
while (!list.isEmpty() && cntr < numResults) {
output.collect(key, list.pollFirst());
cntr++;
}
}
}
public static class MyMapper extends MapReduceBase implements
Mapper<PairOfInts, IntWritable, PairOfInts, IntWritable> {
static Path[] localFiles;
HMapIIW samplesMap = null;
int maxDist;
IntWritable outValue = new IntWritable();
PairOfInts outKey = new PairOfInts();
public void configure(JobConf job) {
sLogger.setLevel(Level.INFO);
maxDist = job.getInt("Ivory.MaxHammingDistance", -1);
// read doc ids of sample into vectors
try {
localFiles = DistributedCache.getLocalCacheFiles(job);
} catch (Exception e) {
throw new RuntimeException("Error reading doc vectors!");
}
if (localFiles != null && localFiles.length > 0) {
samplesMap = new HMapIIW();
try {
LineReader reader = new LineReader(FileSystem.getLocal(job).open(localFiles[0]));
Text t = new Text();
while (reader.readLine(t) != 0) {
int docno = Integer.parseInt(t.toString());
sLogger.info(docno + " --> sample");
samplesMap.put(docno, 1);
}
reader.close();
} catch (IOException e1) {
}
sLogger.info(samplesMap.size() + " sampled");
} else {
sLogger.info("samples file not specified in option SampleDocnosFile");
}
}
public void map(PairOfInts key, IntWritable value,
OutputCollector<PairOfInts, IntWritable> output, Reporter reporter) throws IOException {
int leftKey = key.getLeftElement(); // english docno
int rightKey = key.getRightElement(); // german docno
sLogger.debug(rightKey);
if (samplesMap == null || samplesMap.containsKey(rightKey)) {
if (maxDist == -1 || value.get() <= maxDist) {
outKey.set(leftKey, rightKey);
outValue.set(value.get());
output.collect(outKey, outValue);
}
}
}
}
// @author: ferhanture
// I wrote this to be used on a text dataset. needs some fixing.
// public static class MyReducerAltOutput extends MapReduceBase implements
// Reducer<IntWritable, PairOfInts, Text, Text> {
// int numResults;
// TreeSet<PairOfInts> list = new TreeSet<PairOfInts>();
// private DocnoMapping mDocMapping;
//
// public void configure(JobConf conf){
// numResults = conf.getInt("Ivory.NumResults", -1);
// sLogger.info("numResults");
// mDocMapping = new TextDocnoMapping();
// try {
// mDocMapping.loadMapping(new Path("/user/fture/doug/docno-mapping.dat"), FileSystem.get(conf));
// } catch (IOException e) {
// e.printStackTrace();
// }
// }
//
// public void reduce(IntWritable key, Iterator<PairOfInts> values,
// OutputCollector<Text, Text> output, Reporter reporter)
// throws IOException {
// list.clear();
// while(values.hasNext()){
// PairOfInts p = values.next();
// list.add(new PairOfInts(p.getLeftElement(),p.getRightElement()));
// reporter.incrCounter(mapoutput.count, 1);
// }
// int cntr = 0;
// while(!list.isEmpty() && cntr<numResults){
// PairOfInts nextClosest = list.pollFirst();
// String keyDocid = mDocMapping.getDocid(key.get());
// String valueDocid = mDocMapping.getDocid(nextClosest.getRightElement());
// int dist= nextClosest.getLeftElement();
// output.collect(new Text(keyDocid), new Text(valueDocid+"\t"+dist));
// cntr++;
// }
// }
//
// }
public String[] getRequiredParameters() {
return RequiredParameters;
}
public int run(String[] args) throws Exception {
if (args.length != 5) {
printUsage();
return -1;
}
JobConf job = new JobConf(getConf(), FilterResults.class);
String samplesFile = args[2];
int maxHammingDistance = Integer.parseInt(args[3]);
job.setInt("Ivory.MaxHammingDistance", maxHammingDistance);
int numResults = Integer.parseInt(args[4]);
job.setInt("Ivory.NumResults", numResults);
job.setJobName("FilterResults_" + maxHammingDistance + "_" + numResults);
FileSystem fs2 = FileSystem.get(job);
String inputPath2 = args[0];// job2.get("Ivory.PWSimOutputPath");
String outputPath2 = args[1];// job2.get("FilteredPWSimFile");
int numMappers2 = 300;
int numReducers2 = 1;
if (fs2.exists(new Path(outputPath2))) {
sLogger.info("FilteredPwsim output already exists! Quitting...");
return 0;
}
FileInputFormat.setInputPaths(job, new Path(inputPath2));
FileOutputFormat.setOutputPath(job, new Path(outputPath2));
FileOutputFormat.setCompressOutput(job, false);
job.set("mapred.child.java.opts", "-Xmx2048m");
job.setInt("mapred.map.max.attempts", 10);
job.setInt("mapred.reduce.max.attempts", 10);
job.setInt("mapred.task.timeout", 6000000);
sLogger.info("Running job " + job.getJobName());
sLogger.info("Input directory: " + inputPath2);
sLogger.info("Output directory: " + outputPath2);
sLogger.info("Samples file: " + samplesFile);
if (!samplesFile.equals("none")) {
DistributedCache.addCacheFile(new URI(samplesFile), job); // sample doc vectors in file
}
if (numResults > 0) {
sLogger.info("Number of results = " + numResults);
job.setMapperClass(MyMapperTopN.class);
job.setReducerClass(MyReducerTopN.class);
job.setMapOutputKeyClass(IntWritable.class);
job.setMapOutputValueClass(PairOfInts.class);
} else {
sLogger.info("Number of results = all");
job.setMapperClass(MyMapper.class);
job.setReducerClass(IdentityReducer.class);
job.setMapOutputKeyClass(PairOfInts.class);
job.setMapOutputValueClass(IntWritable.class);
}
job.setJarByClass(FilterResults.class);
job.setNumMapTasks(numMappers2);
job.setNumReduceTasks(numReducers2);
job.setInputFormat(SequenceFileInputFormat.class);
JobClient.runJob(job);
return 0;
}
public static void main(String[] args) throws Exception {
ToolRunner.run(new Configuration(), new FilterResults(), args);
return;
}
}