List of usage examples for org.apache.hadoop.mapreduce.lib.output LazyOutputFormat setOutputFormatClass
@SuppressWarnings("unchecked") public static void setOutputFormatClass(Job job, Class<? extends OutputFormat> theClass)
From source file:de.tudarmstadt.ukp.dkpro.c4corpus.hadoop.full.Phase3Step1ExtractNearDupInfo.java
License:Apache License
@Override public int run(String[] args) throws Exception { Job job = Job.getInstance(getConf()); job.setJarByClass(Phase3Step1ExtractNearDupInfo.class); job.setJobName(Phase3Step1ExtractNearDupInfo.class.getName()); // mapper/*from w ww .j a v a 2 s. co m*/ job.setMapperClass(MapperClass.class); job.setMapOutputKeyClass(Text.class); job.setMapOutputValueClass(DocumentInfo.class); // reducer job.setReducerClass(DeDuplicationTextOutputReducer.class); job.setOutputKeyClass(NullWritable.class); job.setOutputValueClass(List.class); job.setInputFormatClass(WARCInputFormat.class); LazyOutputFormat.setOutputFormatClass(job, DocumentInfoOutputFormat.class); // paths String commaSeparatedInputFiles = args[0]; String outputPath = args[1]; FileInputFormat.addInputPaths(job, commaSeparatedInputFiles); FileOutputFormat.setOutputPath(job, new Path(outputPath)); return job.waitForCompletion(true) ? 0 : 1; }
From source file:de.tudarmstadt.ukp.dkpro.c4corpus.hadoop.full.Phase3Step2DistinctDataJob.java
License:Apache License
@Override public int run(String[] args) throws Exception { Job job = Job.getInstance(getConf()); job.setJarByClass(Phase3Step2DistinctDataJob.class); job.setJobName(Phase3Step2DistinctDataJob.class.getName()); //mapper/* ww w.j ava2 s.com*/ job.setMapperClass(RemoveRedundantDataMapper.class); job.setMapOutputKeyClass(Text.class); job.setMapOutputValueClass(NullWritable.class); //reducer job.setReducerClass(RemoveRedundantDataReducer.class); job.setOutputKeyClass(Text.class); job.setOutputValueClass(NullWritable.class); //paths String commaSeparatedInputFiles = args[0]; String outputPath = args[1]; job.setInputFormatClass(TextInputFormat.class); LazyOutputFormat.setOutputFormatClass(job, TextOutputFormat.class); //i/o paths FileInputFormat.addInputPaths(job, commaSeparatedInputFiles); FileOutputFormat.setOutputPath(job, new Path(outputPath)); return job.waitForCompletion(true) ? 0 : 1; }
From source file:de.tudarmstadt.ukp.dkpro.c4corpus.hadoop.full.Phase3Step3NearDupTuplesCreation.java
License:Apache License
@Override public int run(String[] args) throws Exception { Job job = Job.getInstance(getConf()); job.setJarByClass(Phase3Step3NearDupTuplesCreation.class); job.setJobName(Phase3Step3NearDupTuplesCreation.class.getName()); // mapper//from w w w . j a v a2s . c om job.setMapperClass(CreateTuplesMapper.class); job.setMapOutputKeyClass(NullWritable.class); job.setMapOutputValueClass(TreeSet.class); job.setInputFormatClass(TextInputFormat.class); LazyOutputFormat.setOutputFormatClass(job, TextOutputFormat.class); // paths String commaSeparatedInputFiles = args[0]; String outputPath = args[1]; FileInputFormat.addInputPaths(job, commaSeparatedInputFiles); FileOutputFormat.setOutputPath(job, new Path(outputPath)); job.setNumReduceTasks(0); //must be added or the mapper wont be called return job.waitForCompletion(true) ? 0 : 1; }
From source file:de.tudarmstadt.ukp.dkpro.c4corpus.hadoop.full.Phase3Step4LocalDeDuplication.java
License:Apache License
@Override public int run(String[] args) throws Exception { Job job = Job.getInstance(getConf()); job.setJarByClass(Phase3Step4LocalDeDuplication.class); job.setJobName(Phase3Step4LocalDeDuplication.class.getName()); // paths//from www . j av a 2 s. c o m String inputPath = args[0]; // text files of ids to be deleted String outputPath = args[1]; // input: reading max N lines for each mapper job.setInputFormatClass(NLineInputFormat.class); NLineInputFormat.addInputPath(job, new Path(inputPath)); job.getConfiguration().setInt("mapreduce.input.lineinputformat.linespermap", LINES); // mapper job.setMapperClass(LocalGreedyDeDuplicationMapper.class); LazyOutputFormat.setOutputFormatClass(job, TextOutputFormat.class); // reducer job.setReducerClass(IDCollectorReducer.class); job.setOutputKeyClass(Text.class); job.setOutputValueClass(NullWritable.class); FileOutputFormat.setOutputPath(job, new Path(outputPath)); return job.waitForCompletion(true) ? 0 : 1; }
From source file:edu.nyu.vida.data_polygamy.feature_identification.IndexCreation.java
License:BSD License
/** * @param args/* w ww.j a va 2s . c om*/ */ @SuppressWarnings({ "deprecation" }) public static void main(String[] args) throws IOException, InterruptedException, ClassNotFoundException { Options options = new Options(); Option forceOption = new Option("f", "force", false, "force the computation of the index and events " + "even if files already exist"); forceOption.setRequired(false); options.addOption(forceOption); Option thresholdOption = new Option("t", "use-custom-thresholds", false, "use custom thresholds for regular and rare events, defined in HDFS_HOME/" + FrameworkUtils.thresholdDir + " file"); thresholdOption.setRequired(false); options.addOption(thresholdOption); Option gOption = new Option("g", "group", true, "set group of datasets for which the indices and events" + " will be computed"); gOption.setRequired(true); gOption.setArgName("GROUP"); gOption.setArgs(Option.UNLIMITED_VALUES); options.addOption(gOption); Option machineOption = new Option("m", "machine", true, "machine identifier"); machineOption.setRequired(true); machineOption.setArgName("MACHINE"); machineOption.setArgs(1); options.addOption(machineOption); Option nodesOption = new Option("n", "nodes", true, "number of nodes"); nodesOption.setRequired(true); nodesOption.setArgName("NODES"); nodesOption.setArgs(1); options.addOption(nodesOption); Option s3Option = new Option("s3", "s3", false, "data on Amazon S3"); s3Option.setRequired(false); options.addOption(s3Option); Option awsAccessKeyIdOption = new Option("aws_id", "aws-id", true, "aws access key id; " + "this is required if the execution is on aws"); awsAccessKeyIdOption.setRequired(false); awsAccessKeyIdOption.setArgName("AWS-ACCESS-KEY-ID"); awsAccessKeyIdOption.setArgs(1); options.addOption(awsAccessKeyIdOption); Option awsSecretAccessKeyOption = new Option("aws_key", "aws-id", true, "aws secrect access key; " + "this is required if the execution is on aws"); awsSecretAccessKeyOption.setRequired(false); awsSecretAccessKeyOption.setArgName("AWS-SECRET-ACCESS-KEY"); awsSecretAccessKeyOption.setArgs(1); options.addOption(awsSecretAccessKeyOption); Option bucketOption = new Option("b", "s3-bucket", true, "bucket on s3; " + "this is required if the execution is on aws"); bucketOption.setRequired(false); bucketOption.setArgName("S3-BUCKET"); bucketOption.setArgs(1); options.addOption(bucketOption); Option helpOption = new Option("h", "help", false, "display this message"); helpOption.setRequired(false); options.addOption(helpOption); HelpFormatter formatter = new HelpFormatter(); CommandLineParser parser = new PosixParser(); CommandLine cmd = null; try { cmd = parser.parse(options, args); } catch (ParseException e) { formatter.printHelp("hadoop jar data-polygamy.jar " + "edu.nyu.vida.data_polygamy.feature_identification.IndexCreation", options, true); System.exit(0); } if (cmd.hasOption("h")) { formatter.printHelp("hadoop jar data-polygamy.jar " + "edu.nyu.vida.data_polygamy.feature_identification.IndexCreation", options, true); System.exit(0); } boolean s3 = cmd.hasOption("s3"); String s3bucket = ""; String awsAccessKeyId = ""; String awsSecretAccessKey = ""; if (s3) { if ((!cmd.hasOption("aws_id")) || (!cmd.hasOption("aws_key")) || (!cmd.hasOption("b"))) { System.out.println( "Arguments 'aws_id', 'aws_key', and 'b'" + " are mandatory if execution is on AWS."); formatter.printHelp("hadoop jar data-polygamy.jar " + "edu.nyu.vida.data_polygamy.feature_identification.IndexCreation", options, true); System.exit(0); } s3bucket = cmd.getOptionValue("b"); awsAccessKeyId = cmd.getOptionValue("aws_id"); awsSecretAccessKey = cmd.getOptionValue("aws_key"); } boolean snappyCompression = false; boolean bzip2Compression = false; String machine = cmd.getOptionValue("m"); int nbNodes = Integer.parseInt(cmd.getOptionValue("n")); Configuration s3conf = new Configuration(); if (s3) { s3conf.set("fs.s3.awsAccessKeyId", awsAccessKeyId); s3conf.set("fs.s3.awsSecretAccessKey", awsSecretAccessKey); s3conf.set("bucket", s3bucket); } String datasetNames = ""; String datasetIds = ""; ArrayList<String> shortDataset = new ArrayList<String>(); ArrayList<String> shortDatasetIndex = new ArrayList<String>(); HashMap<String, String> datasetAgg = new HashMap<String, String>(); HashMap<String, String> datasetId = new HashMap<String, String>(); HashMap<String, HashMap<Integer, Double>> datasetRegThreshold = new HashMap<String, HashMap<Integer, Double>>(); HashMap<String, HashMap<Integer, Double>> datasetRareThreshold = new HashMap<String, HashMap<Integer, Double>>(); Path path = null; FileSystem fs = FileSystem.get(new Configuration()); BufferedReader br; boolean removeExistingFiles = cmd.hasOption("f"); boolean isThresholdUserDefined = cmd.hasOption("t"); for (String dataset : cmd.getOptionValues("g")) { // getting aggregates String[] aggregate = FrameworkUtils.searchAggregates(dataset, s3conf, s3); if (aggregate.length == 0) { System.out.println("No aggregates found for " + dataset + "."); continue; } // getting aggregates header String aggregatesHeaderFileName = FrameworkUtils.searchAggregatesHeader(dataset, s3conf, s3); if (aggregatesHeaderFileName == null) { System.out.println("No aggregate header for " + dataset); continue; } String aggregatesHeader = s3bucket + FrameworkUtils.preProcessingDir + "/" + aggregatesHeaderFileName; shortDataset.add(dataset); datasetId.put(dataset, null); if (s3) { path = new Path(aggregatesHeader); fs = FileSystem.get(path.toUri(), s3conf); } else { path = new Path(fs.getHomeDirectory() + "/" + aggregatesHeader); } br = new BufferedReader(new InputStreamReader(fs.open(path))); datasetAgg.put(dataset, br.readLine().split("\t")[1]); br.close(); if (s3) fs.close(); } if (shortDataset.size() == 0) { System.out.println("No datasets to process."); System.exit(0); } // getting dataset id if (s3) { path = new Path(s3bucket + FrameworkUtils.datasetsIndexDir); fs = FileSystem.get(path.toUri(), s3conf); } else { path = new Path(fs.getHomeDirectory() + "/" + FrameworkUtils.datasetsIndexDir); } br = new BufferedReader(new InputStreamReader(fs.open(path))); String line = br.readLine(); while (line != null) { String[] dt = line.split("\t"); if (datasetId.containsKey(dt[0])) { datasetId.put(dt[0], dt[1]); datasetNames += dt[0] + ","; datasetIds += dt[1] + ","; } line = br.readLine(); } br.close(); datasetNames = datasetNames.substring(0, datasetNames.length() - 1); datasetIds = datasetIds.substring(0, datasetIds.length() - 1); Iterator<String> it = shortDataset.iterator(); while (it.hasNext()) { String dataset = it.next(); if (datasetId.get(dataset) == null) { System.out.println("No dataset id for " + dataset); System.exit(0); } } // getting user defined thresholds if (isThresholdUserDefined) { if (s3) { path = new Path(s3bucket + FrameworkUtils.thresholdDir); fs = FileSystem.get(path.toUri(), s3conf); } else { path = new Path(fs.getHomeDirectory() + "/" + FrameworkUtils.thresholdDir); } br = new BufferedReader(new InputStreamReader(fs.open(path))); line = br.readLine(); while (line != null) { // getting dataset name String dataset = line.trim(); HashMap<Integer, Double> regThresholds = new HashMap<Integer, Double>(); HashMap<Integer, Double> rareThresholds = new HashMap<Integer, Double>(); line = br.readLine(); while ((line != null) && (line.split("\t").length > 1)) { // getting attribute ids and thresholds String[] keyVals = line.trim().split("\t"); int att = Integer.parseInt(keyVals[0].trim()); regThresholds.put(att, Double.parseDouble(keyVals[1].trim())); rareThresholds.put(att, Double.parseDouble(keyVals[2].trim())); line = br.readLine(); } datasetRegThreshold.put(dataset, regThresholds); datasetRareThreshold.put(dataset, rareThresholds); } br.close(); } if (s3) fs.close(); // datasets that will use existing merge tree ArrayList<String> useMergeTree = new ArrayList<String>(); // creating index for each spatio-temporal resolution FrameworkUtils.createDir(s3bucket + FrameworkUtils.indexDir, s3conf, s3); HashSet<String> input = new HashSet<String>(); for (String dataset : shortDataset) { String indexCreationOutputFileName = s3bucket + FrameworkUtils.indexDir + "/" + dataset + "/"; String mergeTreeFileName = s3bucket + FrameworkUtils.mergeTreeDir + "/" + dataset + "/"; if (removeExistingFiles) { FrameworkUtils.removeFile(indexCreationOutputFileName, s3conf, s3); FrameworkUtils.removeFile(mergeTreeFileName, s3conf, s3); FrameworkUtils.createDir(mergeTreeFileName, s3conf, s3); } else if (datasetRegThreshold.containsKey(dataset)) { FrameworkUtils.removeFile(indexCreationOutputFileName, s3conf, s3); if (FrameworkUtils.fileExists(mergeTreeFileName, s3conf, s3)) { useMergeTree.add(dataset); } } if (!FrameworkUtils.fileExists(indexCreationOutputFileName, s3conf, s3)) { input.add(s3bucket + FrameworkUtils.aggregatesDir + "/" + dataset); shortDatasetIndex.add(dataset); } } if (input.isEmpty()) { System.out.println("All the input datasets have indices."); System.out.println("Use -f in the beginning of the command line to force the computation."); System.exit(0); } String aggregateDatasets = ""; it = input.iterator(); while (it.hasNext()) { aggregateDatasets += it.next() + ","; } Job icJob = null; Configuration icConf = new Configuration(); Machine machineConf = new Machine(machine, nbNodes); String jobName = "index"; String indexOutputDir = s3bucket + FrameworkUtils.indexDir + "/tmp/"; FrameworkUtils.removeFile(indexOutputDir, s3conf, s3); icConf.set("dataset-name", datasetNames); icConf.set("dataset-id", datasetIds); if (!useMergeTree.isEmpty()) { String useMergeTreeStr = ""; for (String dt : useMergeTree) { useMergeTreeStr += dt + ","; } icConf.set("use-merge-tree", useMergeTreeStr.substring(0, useMergeTreeStr.length() - 1)); } for (int i = 0; i < shortDataset.size(); i++) { String dataset = shortDataset.get(i); String id = datasetId.get(dataset); icConf.set("dataset-" + id + "-aggregates", datasetAgg.get(dataset)); if (datasetRegThreshold.containsKey(dataset)) { HashMap<Integer, Double> regThresholds = datasetRegThreshold.get(dataset); String thresholds = ""; for (int att : regThresholds.keySet()) { thresholds += String.valueOf(att) + "-" + String.valueOf(regThresholds.get(att)) + ","; } icConf.set("regular-" + id, thresholds.substring(0, thresholds.length() - 1)); } if (datasetRareThreshold.containsKey(dataset)) { HashMap<Integer, Double> rareThresholds = datasetRareThreshold.get(dataset); String thresholds = ""; for (int att : rareThresholds.keySet()) { thresholds += String.valueOf(att) + "-" + String.valueOf(rareThresholds.get(att)) + ","; } icConf.set("rare-" + id, thresholds.substring(0, thresholds.length() - 1)); } } icConf.set("mapreduce.tasktracker.map.tasks.maximum", String.valueOf(machineConf.getMaximumTasks())); icConf.set("mapreduce.tasktracker.reduce.tasks.maximum", String.valueOf(machineConf.getMaximumTasks())); icConf.set("mapreduce.jobtracker.maxtasks.perjob", "-1"); icConf.set("mapreduce.reduce.shuffle.parallelcopies", "20"); icConf.set("mapreduce.input.fileinputformat.split.minsize", "0"); icConf.set("mapreduce.task.io.sort.mb", "200"); icConf.set("mapreduce.task.io.sort.factor", "100"); //icConf.set("mapreduce.task.timeout", "1800000"); machineConf.setMachineConfiguration(icConf); if (s3) { machineConf.setMachineConfiguration(icConf); icConf.set("fs.s3.awsAccessKeyId", awsAccessKeyId); icConf.set("fs.s3.awsSecretAccessKey", awsSecretAccessKey); icConf.set("bucket", s3bucket); } if (snappyCompression) { icConf.set("mapreduce.map.output.compress", "true"); icConf.set("mapreduce.map.output.compress.codec", "org.apache.hadoop.io.compress.SnappyCodec"); //icConf.set("mapreduce.output.fileoutputformat.compress.codec", "org.apache.hadoop.io.compress.SnappyCodec"); } if (bzip2Compression) { icConf.set("mapreduce.map.output.compress", "true"); icConf.set("mapreduce.map.output.compress.codec", "org.apache.hadoop.io.compress.BZip2Codec"); //icConf.set("mapreduce.output.fileoutputformat.compress.codec", "org.apache.hadoop.io.compress.BZip2Codec"); } icJob = new Job(icConf); icJob.setJobName(jobName); icJob.setMapOutputKeyClass(AttributeResolutionWritable.class); icJob.setMapOutputValueClass(SpatioTemporalFloatWritable.class); icJob.setOutputKeyClass(AttributeResolutionWritable.class); icJob.setOutputValueClass(TopologyTimeSeriesWritable.class); //icJob.setOutputKeyClass(Text.class); //icJob.setOutputValueClass(Text.class); icJob.setMapperClass(IndexCreationMapper.class); icJob.setReducerClass(IndexCreationReducer.class); icJob.setNumReduceTasks(machineConf.getNumberReduces()); icJob.setInputFormatClass(SequenceFileInputFormat.class); //icJob.setOutputFormatClass(SequenceFileOutputFormat.class); LazyOutputFormat.setOutputFormatClass(icJob, SequenceFileOutputFormat.class); //LazyOutputFormat.setOutputFormatClass(icJob, TextOutputFormat.class); SequenceFileOutputFormat.setCompressOutput(icJob, true); SequenceFileOutputFormat.setOutputCompressionType(icJob, CompressionType.BLOCK); FileInputFormat.setInputDirRecursive(icJob, true); FileInputFormat.setInputPaths(icJob, aggregateDatasets.substring(0, aggregateDatasets.length() - 1)); FileOutputFormat.setOutputPath(icJob, new Path(indexOutputDir)); icJob.setJarByClass(IndexCreation.class); long start = System.currentTimeMillis(); icJob.submit(); icJob.waitForCompletion(true); System.out.println(jobName + "\t" + (System.currentTimeMillis() - start)); // moving files to right place for (String dataset : shortDatasetIndex) { String from = s3bucket + FrameworkUtils.indexDir + "/tmp/" + dataset + "/"; String to = s3bucket + FrameworkUtils.indexDir + "/" + dataset + "/"; FrameworkUtils.renameFile(from, to, s3conf, s3); } }
From source file:edu.nyu.vida.data_polygamy.relationship_computation.Relationship.java
License:BSD License
/** * @param args/*from w w w. jav a 2 s .co m*/ * @throws ParseException */ @SuppressWarnings({ "deprecation" }) public static void main(String[] args) throws IOException, InterruptedException, ClassNotFoundException { Options options = new Options(); Option forceOption = new Option("f", "force", false, "force the computation of the relationship " + "even if files already exist"); forceOption.setRequired(false); options.addOption(forceOption); Option scoreOption = new Option("sc", "score", true, "set threhsold for relationship score"); scoreOption.setRequired(false); scoreOption.setArgName("SCORE THRESHOLD"); options.addOption(scoreOption); Option strengthOption = new Option("st", "strength", true, "set threhsold for relationship strength"); strengthOption.setRequired(false); strengthOption.setArgName("STRENGTH THRESHOLD"); options.addOption(strengthOption); Option completeRandomizationOption = new Option("c", "complete-randomization", false, "use complete randomization when performing significance tests"); completeRandomizationOption.setRequired(false); options.addOption(completeRandomizationOption); Option idOption = new Option("id", "ids", false, "output id instead of names for datasets and attributes"); idOption.setRequired(false); options.addOption(idOption); Option g1Option = new Option("g1", "first-group", true, "set first group of datasets"); g1Option.setRequired(true); g1Option.setArgName("FIRST GROUP"); g1Option.setArgs(Option.UNLIMITED_VALUES); options.addOption(g1Option); Option g2Option = new Option("g2", "second-group", true, "set second group of datasets"); g2Option.setRequired(false); g2Option.setArgName("SECOND GROUP"); g2Option.setArgs(Option.UNLIMITED_VALUES); options.addOption(g2Option); Option machineOption = new Option("m", "machine", true, "machine identifier"); machineOption.setRequired(true); machineOption.setArgName("MACHINE"); machineOption.setArgs(1); options.addOption(machineOption); Option nodesOption = new Option("n", "nodes", true, "number of nodes"); nodesOption.setRequired(true); nodesOption.setArgName("NODES"); nodesOption.setArgs(1); options.addOption(nodesOption); Option s3Option = new Option("s3", "s3", false, "data on Amazon S3"); s3Option.setRequired(false); options.addOption(s3Option); Option awsAccessKeyIdOption = new Option("aws_id", "aws-id", true, "aws access key id; " + "this is required if the execution is on aws"); awsAccessKeyIdOption.setRequired(false); awsAccessKeyIdOption.setArgName("AWS-ACCESS-KEY-ID"); awsAccessKeyIdOption.setArgs(1); options.addOption(awsAccessKeyIdOption); Option awsSecretAccessKeyOption = new Option("aws_key", "aws-id", true, "aws secrect access key; " + "this is required if the execution is on aws"); awsSecretAccessKeyOption.setRequired(false); awsSecretAccessKeyOption.setArgName("AWS-SECRET-ACCESS-KEY"); awsSecretAccessKeyOption.setArgs(1); options.addOption(awsSecretAccessKeyOption); Option bucketOption = new Option("b", "s3-bucket", true, "bucket on s3; " + "this is required if the execution is on aws"); bucketOption.setRequired(false); bucketOption.setArgName("S3-BUCKET"); bucketOption.setArgs(1); options.addOption(bucketOption); Option helpOption = new Option("h", "help", false, "display this message"); helpOption.setRequired(false); options.addOption(helpOption); Option removeOption = new Option("r", "remove-not-significant", false, "remove relationships that are not" + "significant from the final output"); removeOption.setRequired(false); options.addOption(removeOption); HelpFormatter formatter = new HelpFormatter(); CommandLineParser parser = new PosixParser(); CommandLine cmd = null; try { cmd = parser.parse(options, args); } catch (ParseException e) { formatter.printHelp("hadoop jar data-polygamy.jar " + "edu.nyu.vida.data_polygamy.relationship_computation.Relationship", options, true); System.exit(0); } if (cmd.hasOption("h")) { formatter.printHelp("hadoop jar data-polygamy.jar " + "edu.nyu.vida.data_polygamy.relationship_computation.Relationship", options, true); System.exit(0); } boolean s3 = cmd.hasOption("s3"); String s3bucket = ""; String awsAccessKeyId = ""; String awsSecretAccessKey = ""; if (s3) { if ((!cmd.hasOption("aws_id")) || (!cmd.hasOption("aws_key")) || (!cmd.hasOption("b"))) { System.out.println( "Arguments 'aws_id', 'aws_key', and 'b'" + " are mandatory if execution is on AWS."); formatter.printHelp( "hadoop jar data-polygamy.jar " + "edu.nyu.vida.data_polygamy.relationship_computation.Relationship", options, true); System.exit(0); } s3bucket = cmd.getOptionValue("b"); awsAccessKeyId = cmd.getOptionValue("aws_id"); awsSecretAccessKey = cmd.getOptionValue("aws_key"); } boolean snappyCompression = false; boolean bzip2Compression = false; String machine = cmd.getOptionValue("m"); int nbNodes = Integer.parseInt(cmd.getOptionValue("n")); Configuration s3conf = new Configuration(); if (s3) { s3conf.set("fs.s3.awsAccessKeyId", awsAccessKeyId); s3conf.set("fs.s3.awsSecretAccessKey", awsSecretAccessKey); s3conf.set("bucket", s3bucket); } Path path = null; FileSystem fs = FileSystem.get(new Configuration()); ArrayList<String> shortDataset = new ArrayList<String>(); ArrayList<String> firstGroup = new ArrayList<String>(); ArrayList<String> secondGroup = new ArrayList<String>(); HashMap<String, String> datasetAgg = new HashMap<String, String>(); boolean removeNotSignificant = cmd.hasOption("r"); boolean removeExistingFiles = cmd.hasOption("f"); boolean completeRandomization = cmd.hasOption("c"); boolean hasScoreThreshold = cmd.hasOption("sc"); boolean hasStrengthThreshold = cmd.hasOption("st"); boolean outputIds = cmd.hasOption("id"); String scoreThreshold = hasScoreThreshold ? cmd.getOptionValue("sc") : ""; String strengthThreshold = hasStrengthThreshold ? cmd.getOptionValue("st") : ""; // all datasets ArrayList<String> all_datasets = new ArrayList<String>(); if (s3) { path = new Path(s3bucket + FrameworkUtils.datasetsIndexDir); fs = FileSystem.get(path.toUri(), s3conf); } else { path = new Path(fs.getHomeDirectory() + "/" + FrameworkUtils.datasetsIndexDir); } BufferedReader br = new BufferedReader(new InputStreamReader(fs.open(path))); String line = br.readLine(); while (line != null) { all_datasets.add(line.split("\t")[0]); line = br.readLine(); } br.close(); if (s3) fs.close(); String[] all_datasets_array = new String[all_datasets.size()]; all_datasets.toArray(all_datasets_array); String[] firstGroupCmd = cmd.getOptionValues("g1"); String[] secondGroupCmd = cmd.hasOption("g2") ? cmd.getOptionValues("g2") : all_datasets_array; addDatasets(firstGroupCmd, firstGroup, shortDataset, datasetAgg, path, fs, s3conf, s3, s3bucket); addDatasets(secondGroupCmd, secondGroup, shortDataset, datasetAgg, path, fs, s3conf, s3, s3bucket); if (shortDataset.size() == 0) { System.out.println("No datasets to process."); System.exit(0); } if (firstGroup.isEmpty()) { System.out.println("No indices from datasets in G1."); System.exit(0); } if (secondGroup.isEmpty()) { System.out.println("No indices from datasets in G2."); System.exit(0); } // getting dataset ids String datasetNames = ""; String datasetIds = ""; HashMap<String, String> datasetId = new HashMap<String, String>(); Iterator<String> it = shortDataset.iterator(); while (it.hasNext()) { datasetId.put(it.next(), null); } if (s3) { path = new Path(s3bucket + FrameworkUtils.datasetsIndexDir); fs = FileSystem.get(path.toUri(), s3conf); } else { path = new Path(fs.getHomeDirectory() + "/" + FrameworkUtils.datasetsIndexDir); } br = new BufferedReader(new InputStreamReader(fs.open(path))); line = br.readLine(); while (line != null) { String[] dt = line.split("\t"); all_datasets.add(dt[0]); if (datasetId.containsKey(dt[0])) { datasetId.put(dt[0], dt[1]); datasetNames += dt[0] + ","; datasetIds += dt[1] + ","; } line = br.readLine(); } br.close(); if (s3) fs.close(); datasetNames = datasetNames.substring(0, datasetNames.length() - 1); datasetIds = datasetIds.substring(0, datasetIds.length() - 1); it = shortDataset.iterator(); while (it.hasNext()) { String dataset = it.next(); if (datasetId.get(dataset) == null) { System.out.println("No dataset id for " + dataset); System.exit(0); } } String firstGroupStr = ""; String secondGroupStr = ""; for (String dataset : firstGroup) { firstGroupStr += datasetId.get(dataset) + ","; } for (String dataset : secondGroup) { secondGroupStr += datasetId.get(dataset) + ","; } firstGroupStr = firstGroupStr.substring(0, firstGroupStr.length() - 1); secondGroupStr = secondGroupStr.substring(0, secondGroupStr.length() - 1); String relationshipsDir = ""; if (outputIds) { relationshipsDir = FrameworkUtils.relationshipsIdsDir; } else { relationshipsDir = FrameworkUtils.relationshipsDir; } FrameworkUtils.createDir(s3bucket + relationshipsDir, s3conf, s3); String random = completeRandomization ? "complete" : "restricted"; String indexInputDirs = ""; String noRelationship = ""; HashSet<String> dirs = new HashSet<String>(); String dataset1; String dataset2; String datasetId1; String datasetId2; for (int i = 0; i < firstGroup.size(); i++) { for (int j = 0; j < secondGroup.size(); j++) { if (Integer.parseInt(datasetId.get(firstGroup.get(i))) < Integer .parseInt(datasetId.get(secondGroup.get(j)))) { dataset1 = firstGroup.get(i); dataset2 = secondGroup.get(j); } else { dataset1 = secondGroup.get(j); dataset2 = firstGroup.get(i); } datasetId1 = datasetId.get(dataset1); datasetId2 = datasetId.get(dataset2); if (dataset1.equals(dataset2)) continue; String correlationOutputFileName = s3bucket + relationshipsDir + "/" + dataset1 + "-" + dataset2 + "/"; if (removeExistingFiles) { FrameworkUtils.removeFile(correlationOutputFileName, s3conf, s3); } if (!FrameworkUtils.fileExists(correlationOutputFileName, s3conf, s3)) { dirs.add(s3bucket + FrameworkUtils.indexDir + "/" + dataset1); dirs.add(s3bucket + FrameworkUtils.indexDir + "/" + dataset2); } else { noRelationship += datasetId1 + "-" + datasetId2 + ","; } } } if (dirs.isEmpty()) { System.out.println("All the relationships were already computed."); System.out.println("Use -f in the beginning of the command line to force the computation."); System.exit(0); } for (String dir : dirs) { indexInputDirs += dir + ","; } Configuration conf = new Configuration(); Machine machineConf = new Machine(machine, nbNodes); String jobName = "relationship" + "-" + random; String relationshipOutputDir = s3bucket + relationshipsDir + "/tmp/"; FrameworkUtils.removeFile(relationshipOutputDir, s3conf, s3); for (int i = 0; i < shortDataset.size(); i++) { conf.set("dataset-" + datasetId.get(shortDataset.get(i)) + "-agg", datasetAgg.get(shortDataset.get(i))); } for (int i = 0; i < shortDataset.size(); i++) { conf.set("dataset-" + datasetId.get(shortDataset.get(i)) + "-agg-size", Integer.toString(datasetAgg.get(shortDataset.get(i)).split(",").length)); } conf.set("dataset-keys", datasetIds); conf.set("dataset-names", datasetNames); conf.set("first-group", firstGroupStr); conf.set("second-group", secondGroupStr); conf.set("complete-random", String.valueOf(completeRandomization)); conf.set("output-ids", String.valueOf(outputIds)); conf.set("complete-random-str", random); conf.set("main-dataset-id", datasetId.get(shortDataset.get(0))); conf.set("remove-not-significant", String.valueOf(removeNotSignificant)); if (noRelationship.length() > 0) { conf.set("no-relationship", noRelationship.substring(0, noRelationship.length() - 1)); } if (hasScoreThreshold) { conf.set("score-threshold", scoreThreshold); } if (hasStrengthThreshold) { conf.set("strength-threshold", strengthThreshold); } conf.set("mapreduce.tasktracker.map.tasks.maximum", String.valueOf(machineConf.getMaximumTasks())); conf.set("mapreduce.tasktracker.reduce.tasks.maximum", String.valueOf(machineConf.getMaximumTasks())); conf.set("mapreduce.jobtracker.maxtasks.perjob", "-1"); conf.set("mapreduce.reduce.shuffle.parallelcopies", "20"); conf.set("mapreduce.input.fileinputformat.split.minsize", "0"); conf.set("mapreduce.task.io.sort.mb", "200"); conf.set("mapreduce.task.io.sort.factor", "100"); conf.set("mapreduce.task.timeout", "2400000"); if (s3) { machineConf.setMachineConfiguration(conf); conf.set("fs.s3.awsAccessKeyId", awsAccessKeyId); conf.set("fs.s3.awsSecretAccessKey", awsSecretAccessKey); conf.set("bucket", s3bucket); } if (snappyCompression) { conf.set("mapreduce.map.output.compress", "true"); conf.set("mapreduce.map.output.compress.codec", "org.apache.hadoop.io.compress.SnappyCodec"); //conf.set("mapreduce.output.fileoutputformat.compress.codec", "org.apache.hadoop.io.compress.SnappyCodec"); } if (bzip2Compression) { conf.set("mapreduce.map.output.compress", "true"); conf.set("mapreduce.map.output.compress.codec", "org.apache.hadoop.io.compress.BZip2Codec"); //conf.set("mapreduce.output.fileoutputformat.compress.codec", "org.apache.hadoop.io.compress.BZip2Codec"); } Job job = new Job(conf); job.setJobName(jobName); job.setMapOutputKeyClass(PairAttributeWritable.class); job.setMapOutputValueClass(TopologyTimeSeriesWritable.class); job.setOutputKeyClass(Text.class); job.setOutputValueClass(Text.class); job.setMapperClass(CorrelationMapper.class); job.setReducerClass(CorrelationReducer.class); job.setNumReduceTasks(machineConf.getNumberReduces()); job.setInputFormatClass(SequenceFileInputFormat.class); //job.setOutputFormatClass(TextOutputFormat.class); LazyOutputFormat.setOutputFormatClass(job, TextOutputFormat.class); FileInputFormat.setInputDirRecursive(job, true); FileInputFormat.setInputPaths(job, indexInputDirs.substring(0, indexInputDirs.length() - 1)); FileOutputFormat.setOutputPath(job, new Path(relationshipOutputDir)); job.setJarByClass(Relationship.class); long start = System.currentTimeMillis(); job.submit(); job.waitForCompletion(true); System.out.println(jobName + "\t" + (System.currentTimeMillis() - start)); // moving files to right place for (int i = 0; i < firstGroup.size(); i++) { for (int j = 0; j < secondGroup.size(); j++) { if (Integer.parseInt(datasetId.get(firstGroup.get(i))) < Integer .parseInt(datasetId.get(secondGroup.get(j)))) { dataset1 = firstGroup.get(i); dataset2 = secondGroup.get(j); } else { dataset1 = secondGroup.get(j); dataset2 = firstGroup.get(i); } if (dataset1.equals(dataset2)) continue; String from = s3bucket + relationshipsDir + "/tmp/" + dataset1 + "-" + dataset2 + "/"; String to = s3bucket + relationshipsDir + "/" + dataset1 + "-" + dataset2 + "/"; FrameworkUtils.renameFile(from, to, s3conf, s3); } } }
From source file:edu.nyu.vida.data_polygamy.scalar_function_computation.Aggregation.java
License:BSD License
/** * @param args//from w w w . j a v a 2 s . co m */ @SuppressWarnings({ "deprecation" }) public static void main(String[] args) throws IOException, InterruptedException, ClassNotFoundException { Options options = new Options(); Option forceOption = new Option("f", "force", false, "force the computation of the aggregate functions " + "even if files already exist"); forceOption.setRequired(false); options.addOption(forceOption); Option gOption = new Option("g", "group", true, "set group of datasets for which the aggregate functions" + " will be computed, followed by their temporal and spatial attribute indices"); gOption.setRequired(true); gOption.setArgName("GROUP"); gOption.setArgs(Option.UNLIMITED_VALUES); options.addOption(gOption); Option machineOption = new Option("m", "machine", true, "machine identifier"); machineOption.setRequired(true); machineOption.setArgName("MACHINE"); machineOption.setArgs(1); options.addOption(machineOption); Option nodesOption = new Option("n", "nodes", true, "number of nodes"); nodesOption.setRequired(true); nodesOption.setArgName("NODES"); nodesOption.setArgs(1); options.addOption(nodesOption); Option s3Option = new Option("s3", "s3", false, "data on Amazon S3"); s3Option.setRequired(false); options.addOption(s3Option); Option awsAccessKeyIdOption = new Option("aws_id", "aws-id", true, "aws access key id; " + "this is required if the execution is on aws"); awsAccessKeyIdOption.setRequired(false); awsAccessKeyIdOption.setArgName("AWS-ACCESS-KEY-ID"); awsAccessKeyIdOption.setArgs(1); options.addOption(awsAccessKeyIdOption); Option awsSecretAccessKeyOption = new Option("aws_key", "aws-id", true, "aws secrect access key; " + "this is required if the execution is on aws"); awsSecretAccessKeyOption.setRequired(false); awsSecretAccessKeyOption.setArgName("AWS-SECRET-ACCESS-KEY"); awsSecretAccessKeyOption.setArgs(1); options.addOption(awsSecretAccessKeyOption); Option bucketOption = new Option("b", "s3-bucket", true, "bucket on s3; " + "this is required if the execution is on aws"); bucketOption.setRequired(false); bucketOption.setArgName("S3-BUCKET"); bucketOption.setArgs(1); options.addOption(bucketOption); Option helpOption = new Option("h", "help", false, "display this message"); helpOption.setRequired(false); options.addOption(helpOption); HelpFormatter formatter = new HelpFormatter(); CommandLineParser parser = new PosixParser(); CommandLine cmd = null; try { cmd = parser.parse(options, args); } catch (ParseException e) { formatter.printHelp("hadoop jar data-polygamy.jar " + "edu.nyu.vida.data_polygamy.scalar_function_computation.Aggregation", options, true); System.exit(0); } if (cmd.hasOption("h")) { formatter.printHelp("hadoop jar data-polygamy.jar " + "edu.nyu.vida.data_polygamy.scalar_function_computation.Aggregation", options, true); System.exit(0); } boolean s3 = cmd.hasOption("s3"); String s3bucket = ""; String awsAccessKeyId = ""; String awsSecretAccessKey = ""; if (s3) { if ((!cmd.hasOption("aws_id")) || (!cmd.hasOption("aws_key")) || (!cmd.hasOption("b"))) { System.out.println( "Arguments 'aws_id', 'aws_key', and 'b'" + " are mandatory if execution is on AWS."); formatter.printHelp( "hadoop jar data-polygamy.jar " + "edu.nyu.vida.data_polygamy.scalar_function_computation.Aggregation", options, true); System.exit(0); } s3bucket = cmd.getOptionValue("b"); awsAccessKeyId = cmd.getOptionValue("aws_id"); awsSecretAccessKey = cmd.getOptionValue("aws_key"); } boolean snappyCompression = false; boolean bzip2Compression = false; String machine = cmd.getOptionValue("m"); int nbNodes = Integer.parseInt(cmd.getOptionValue("n")); Configuration s3conf = new Configuration(); if (s3) { s3conf.set("fs.s3.awsAccessKeyId", awsAccessKeyId); s3conf.set("fs.s3.awsSecretAccessKey", awsSecretAccessKey); s3conf.set("bucket", s3bucket); } String datasetNames = ""; String datasetIds = ""; String preProcessingDatasets = ""; ArrayList<String> shortDataset = new ArrayList<String>(); ArrayList<String> shortDatasetAggregation = new ArrayList<String>(); HashMap<String, String> datasetTempAtt = new HashMap<String, String>(); HashMap<String, String> datasetSpatialAtt = new HashMap<String, String>(); HashMap<String, String> preProcessingDataset = new HashMap<String, String>(); HashMap<String, String> datasetId = new HashMap<String, String>(); boolean removeExistingFiles = cmd.hasOption("f"); String[] datasetArgs = cmd.getOptionValues("g"); for (int i = 0; i < datasetArgs.length; i += 3) { String dataset = datasetArgs[i]; // getting pre-processing String tempPreProcessing = FrameworkUtils.searchPreProcessing(dataset, s3conf, s3); if (tempPreProcessing == null) { System.out.println("No pre-processing available for " + dataset); continue; } preProcessingDataset.put(dataset, tempPreProcessing); shortDataset.add(dataset); datasetTempAtt.put(dataset, ((datasetArgs[i + 1] == "null") ? null : datasetArgs[i + 1])); datasetSpatialAtt.put(dataset, ((datasetArgs[i + 2] == "null") ? null : datasetArgs[i + 2])); datasetId.put(dataset, null); } if (shortDataset.size() == 0) { System.out.println("No datasets to process."); System.exit(0); } // getting dataset id Path path = null; FileSystem fs = null; if (s3) { path = new Path(s3bucket + FrameworkUtils.datasetsIndexDir); fs = FileSystem.get(path.toUri(), s3conf); } else { fs = FileSystem.get(new Configuration()); path = new Path(fs.getHomeDirectory() + "/" + FrameworkUtils.datasetsIndexDir); } BufferedReader br = new BufferedReader(new InputStreamReader(fs.open(path))); String line = br.readLine(); while (line != null) { String[] dt = line.split("\t"); if (datasetId.containsKey(dt[0])) { datasetId.put(dt[0], dt[1]); datasetNames += dt[0] + ","; datasetIds += dt[1] + ","; } line = br.readLine(); } br.close(); if (s3) fs.close(); datasetNames = datasetNames.substring(0, datasetNames.length() - 1); datasetIds = datasetIds.substring(0, datasetIds.length() - 1); Iterator<String> it = shortDataset.iterator(); while (it.hasNext()) { String dataset = it.next(); if (datasetId.get(dataset) == null) { System.out.println("No dataset id for " + dataset); System.exit(0); } } FrameworkUtils.createDir(s3bucket + FrameworkUtils.aggregatesDir, s3conf, s3); // getting smallest resolution HashMap<String, String> tempResMap = new HashMap<String, String>(); HashMap<String, String> spatialResMap = new HashMap<String, String>(); HashMap<String, String> datasetTemporalStrMap = new HashMap<String, String>(); HashMap<String, String> datasetSpatialStrMap = new HashMap<String, String>(); HashSet<String> input = new HashSet<String>(); for (String dataset : shortDataset) { String[] datasetArray = preProcessingDataset.get(dataset).split("-"); String datasetTemporalStr = datasetArray[datasetArray.length - 2]; int datasetTemporal = utils.temporalResolution(datasetTemporalStr); String datasetSpatialStr = datasetArray[datasetArray.length - 1]; int datasetSpatial = utils.spatialResolution(datasetSpatialStr); // finding all possible resolutions String[] temporalResolutions = FrameworkUtils.getAggTempResolutions(datasetTemporal); String[] spatialResolutions = FrameworkUtils.getAggSpatialResolutions(datasetSpatial); String temporalResolution = ""; String spatialResolution = ""; String tempRes = ""; String spatialRes = ""; boolean dataAdded = false; for (int i = 0; i < temporalResolutions.length; i++) { for (int j = 0; j < spatialResolutions.length; j++) { temporalResolution = temporalResolutions[i]; spatialResolution = spatialResolutions[j]; String aggregatesOutputFileName = s3bucket + FrameworkUtils.aggregatesDir + "/" + dataset + "/"; if (removeExistingFiles) { FrameworkUtils.removeFile(aggregatesOutputFileName, s3conf, s3); } if (!FrameworkUtils.fileExists(aggregatesOutputFileName, s3conf, s3)) { dataAdded = true; tempRes += temporalResolution + "-"; spatialRes += spatialResolution + "-"; } } } if (dataAdded) { input.add(s3bucket + FrameworkUtils.preProcessingDir + "/" + preProcessingDataset.get(dataset)); shortDatasetAggregation.add(dataset); tempResMap.put(dataset, tempRes.substring(0, tempRes.length() - 1)); spatialResMap.put(dataset, spatialRes.substring(0, spatialRes.length() - 1)); datasetTemporalStrMap.put(dataset, datasetTemporalStr); datasetSpatialStrMap.put(dataset, datasetSpatialStr); } } if (input.isEmpty()) { System.out.println("All the input datasets have aggregates."); System.out.println("Use -f in the beginning of the command line to force the computation."); System.exit(0); } it = input.iterator(); while (it.hasNext()) { preProcessingDatasets += it.next() + ","; } Job aggJob = null; String aggregatesOutputDir = s3bucket + FrameworkUtils.aggregatesDir + "/tmp/"; String jobName = "aggregates"; FrameworkUtils.removeFile(aggregatesOutputDir, s3conf, s3); Configuration aggConf = new Configuration(); Machine machineConf = new Machine(machine, nbNodes); aggConf.set("dataset-name", datasetNames); aggConf.set("dataset-id", datasetIds); for (int i = 0; i < shortDatasetAggregation.size(); i++) { String dataset = shortDatasetAggregation.get(i); String id = datasetId.get(dataset); aggConf.set("dataset-" + id + "-temporal-resolutions", tempResMap.get(dataset)); aggConf.set("dataset-" + id + "-spatial-resolutions", spatialResMap.get(dataset)); aggConf.set("dataset-" + id + "-temporal-att", datasetTempAtt.get(dataset)); aggConf.set("dataset-" + id + "-spatial-att", datasetSpatialAtt.get(dataset)); aggConf.set("dataset-" + id + "-temporal", datasetTemporalStrMap.get(dataset)); aggConf.set("dataset-" + id + "-spatial", datasetSpatialStrMap.get(dataset)); if (s3) aggConf.set("dataset-" + id, s3bucket + FrameworkUtils.preProcessingDir + "/" + preProcessingDataset.get(dataset)); else aggConf.set("dataset-" + id, FileSystem.get(new Configuration()).getHomeDirectory() + "/" + FrameworkUtils.preProcessingDir + "/" + preProcessingDataset.get(dataset)); } aggConf.set("mapreduce.tasktracker.map.tasks.maximum", String.valueOf(machineConf.getMaximumTasks())); aggConf.set("mapreduce.tasktracker.reduce.tasks.maximum", String.valueOf(machineConf.getMaximumTasks())); aggConf.set("mapreduce.jobtracker.maxtasks.perjob", "-1"); aggConf.set("mapreduce.reduce.shuffle.parallelcopies", "20"); aggConf.set("mapreduce.input.fileinputformat.split.minsize", "0"); aggConf.set("mapreduce.task.io.sort.mb", "200"); aggConf.set("mapreduce.task.io.sort.factor", "100"); machineConf.setMachineConfiguration(aggConf); if (s3) { machineConf.setMachineConfiguration(aggConf); aggConf.set("fs.s3.awsAccessKeyId", awsAccessKeyId); aggConf.set("fs.s3.awsSecretAccessKey", awsSecretAccessKey); } if (snappyCompression) { aggConf.set("mapreduce.map.output.compress", "true"); aggConf.set("mapreduce.map.output.compress.codec", "org.apache.hadoop.io.compress.SnappyCodec"); //aggConf.set("mapreduce.output.fileoutputformat.compress.codec", "org.apache.hadoop.io.compress.SnappyCodec"); } if (bzip2Compression) { aggConf.set("mapreduce.map.output.compress", "true"); aggConf.set("mapreduce.map.output.compress.codec", "org.apache.hadoop.io.compress.BZip2Codec"); //aggConf.set("mapreduce.output.fileoutputformat.compress.codec", "org.apache.hadoop.io.compress.BZip2Codec"); } aggJob = new Job(aggConf); aggJob.setJobName(jobName); aggJob.setMapOutputKeyClass(SpatioTemporalWritable.class); aggJob.setMapOutputValueClass(AggregationArrayWritable.class); aggJob.setOutputKeyClass(SpatioTemporalWritable.class); aggJob.setOutputValueClass(FloatArrayWritable.class); //aggJob.setOutputKeyClass(Text.class); //aggJob.setOutputValueClass(Text.class); aggJob.setMapperClass(AggregationMapper.class); aggJob.setCombinerClass(AggregationCombiner.class); aggJob.setReducerClass(AggregationReducer.class); aggJob.setNumReduceTasks(machineConf.getNumberReduces()); aggJob.setInputFormatClass(SequenceFileInputFormat.class); //aggJob.setOutputFormatClass(SequenceFileOutputFormat.class); LazyOutputFormat.setOutputFormatClass(aggJob, SequenceFileOutputFormat.class); //LazyOutputFormat.setOutputFormatClass(aggJob, TextOutputFormat.class); SequenceFileOutputFormat.setCompressOutput(aggJob, true); SequenceFileOutputFormat.setOutputCompressionType(aggJob, CompressionType.BLOCK); FileInputFormat.setInputDirRecursive(aggJob, true); FileInputFormat.setInputPaths(aggJob, preProcessingDatasets.substring(0, preProcessingDatasets.length() - 1)); FileOutputFormat.setOutputPath(aggJob, new Path(aggregatesOutputDir)); aggJob.setJarByClass(Aggregation.class); long start = System.currentTimeMillis(); aggJob.submit(); aggJob.waitForCompletion(true); System.out.println(jobName + "\t" + (System.currentTimeMillis() - start)); // moving files to right place for (String dataset : shortDatasetAggregation) { String from = s3bucket + FrameworkUtils.aggregatesDir + "/tmp/" + dataset + "/"; String to = s3bucket + FrameworkUtils.aggregatesDir + "/" + dataset + "/"; FrameworkUtils.renameFile(from, to, s3conf, s3); } }
From source file:edu.nyu.vida.data_polygamy.standard_techniques.CorrelationTechniques.java
License:BSD License
/** * @param args//from w w w. j av a2 s . co m * @throws ParseException */ @SuppressWarnings({ "deprecation" }) public static void main(String[] args) throws IOException, InterruptedException, ClassNotFoundException { Options options = new Options(); Option forceOption = new Option("f", "force", false, "force the computation of the relationship " + "even if files already exist"); forceOption.setRequired(false); options.addOption(forceOption); Option g1Option = new Option("g1", "first-group", true, "set first group of datasets"); g1Option.setRequired(true); g1Option.setArgName("FIRST GROUP"); g1Option.setArgs(Option.UNLIMITED_VALUES); options.addOption(g1Option); Option g2Option = new Option("g2", "second-group", true, "set second group of datasets"); g2Option.setRequired(false); g2Option.setArgName("SECOND GROUP"); g2Option.setArgs(Option.UNLIMITED_VALUES); options.addOption(g2Option); Option machineOption = new Option("m", "machine", true, "machine identifier"); machineOption.setRequired(true); machineOption.setArgName("MACHINE"); machineOption.setArgs(1); options.addOption(machineOption); Option nodesOption = new Option("n", "nodes", true, "number of nodes"); nodesOption.setRequired(true); nodesOption.setArgName("NODES"); nodesOption.setArgs(1); options.addOption(nodesOption); Option s3Option = new Option("s3", "s3", false, "data on Amazon S3"); s3Option.setRequired(false); options.addOption(s3Option); Option awsAccessKeyIdOption = new Option("aws_id", "aws-id", true, "aws access key id; " + "this is required if the execution is on aws"); awsAccessKeyIdOption.setRequired(false); awsAccessKeyIdOption.setArgName("AWS-ACCESS-KEY-ID"); awsAccessKeyIdOption.setArgs(1); options.addOption(awsAccessKeyIdOption); Option awsSecretAccessKeyOption = new Option("aws_key", "aws-id", true, "aws secrect access key; " + "this is required if the execution is on aws"); awsSecretAccessKeyOption.setRequired(false); awsSecretAccessKeyOption.setArgName("AWS-SECRET-ACCESS-KEY"); awsSecretAccessKeyOption.setArgs(1); options.addOption(awsSecretAccessKeyOption); Option bucketOption = new Option("b", "s3-bucket", true, "bucket on s3; " + "this is required if the execution is on aws"); bucketOption.setRequired(false); bucketOption.setArgName("S3-BUCKET"); bucketOption.setArgs(1); options.addOption(bucketOption); Option helpOption = new Option("h", "help", false, "display this message"); helpOption.setRequired(false); options.addOption(helpOption); HelpFormatter formatter = new HelpFormatter(); CommandLineParser parser = new PosixParser(); CommandLine cmd = null; try { cmd = parser.parse(options, args); } catch (ParseException e) { formatter.printHelp( "hadoop jar data-polygamy.jar " + "edu.nyu.vida.data_polygamy.standard_techniques.CorrelationTechniques", options, true); System.exit(0); } if (cmd.hasOption("h")) { formatter.printHelp( "hadoop jar data-polygamy.jar " + "edu.nyu.vida.data_polygamy.standard_techniques.CorrelationTechniques", options, true); System.exit(0); } boolean s3 = cmd.hasOption("s3"); String s3bucket = ""; String awsAccessKeyId = ""; String awsSecretAccessKey = ""; if (s3) { if ((!cmd.hasOption("aws_id")) || (!cmd.hasOption("aws_key")) || (!cmd.hasOption("b"))) { System.out.println( "Arguments 'aws_id', 'aws_key', and 'b'" + " are mandatory if execution is on AWS."); formatter.printHelp( "hadoop jar data-polygamy.jar " + "edu.nyu.vida.data_polygamy.standard_techniques.CorrelationTechniques", options, true); System.exit(0); } s3bucket = cmd.getOptionValue("b"); awsAccessKeyId = cmd.getOptionValue("aws_id"); awsSecretAccessKey = cmd.getOptionValue("aws_key"); } boolean snappyCompression = false; boolean bzip2Compression = false; String machine = cmd.getOptionValue("m"); int nbNodes = Integer.parseInt(cmd.getOptionValue("n")); Configuration s3conf = new Configuration(); if (s3) { s3conf.set("fs.s3.awsAccessKeyId", awsAccessKeyId); s3conf.set("fs.s3.awsSecretAccessKey", awsSecretAccessKey); s3conf.set("bucket", s3bucket); } Path path = null; FileSystem fs = FileSystem.get(new Configuration()); ArrayList<String> shortDataset = new ArrayList<String>(); ArrayList<String> firstGroup = new ArrayList<String>(); ArrayList<String> secondGroup = new ArrayList<String>(); HashMap<String, String> datasetAgg = new HashMap<String, String>(); boolean removeExistingFiles = cmd.hasOption("f"); String[] firstGroupCmd = cmd.getOptionValues("g1"); String[] secondGroupCmd = cmd.hasOption("g2") ? cmd.getOptionValues("g2") : new String[0]; addDatasets(firstGroupCmd, firstGroup, shortDataset, datasetAgg, path, fs, s3conf, s3, s3bucket); addDatasets(secondGroupCmd, secondGroup, shortDataset, datasetAgg, path, fs, s3conf, s3, s3bucket); if (shortDataset.size() == 0) { System.out.println("No datasets to process."); System.exit(0); } if (firstGroup.isEmpty()) { System.out.println("First group of datasets (G1) is empty. " + "Doing G1 = G2."); firstGroup.addAll(secondGroup); } if (secondGroup.isEmpty()) { System.out.println("Second group of datasets (G2) is empty. " + "Doing G2 = G1."); secondGroup.addAll(firstGroup); } // getting dataset ids String datasetNames = ""; String datasetIds = ""; HashMap<String, String> datasetId = new HashMap<String, String>(); Iterator<String> it = shortDataset.iterator(); while (it.hasNext()) { datasetId.put(it.next(), null); } if (s3) { path = new Path(s3bucket + FrameworkUtils.datasetsIndexDir); fs = FileSystem.get(path.toUri(), s3conf); } else { path = new Path(fs.getHomeDirectory() + "/" + FrameworkUtils.datasetsIndexDir); } BufferedReader br = new BufferedReader(new InputStreamReader(fs.open(path))); String line = br.readLine(); while (line != null) { String[] dt = line.split("\t"); if (datasetId.containsKey(dt[0])) { datasetId.put(dt[0], dt[1]); datasetNames += dt[0] + ","; datasetIds += dt[1] + ","; } line = br.readLine(); } br.close(); if (s3) fs.close(); datasetNames = datasetNames.substring(0, datasetNames.length() - 1); datasetIds = datasetIds.substring(0, datasetIds.length() - 1); it = shortDataset.iterator(); while (it.hasNext()) { String dataset = it.next(); if (datasetId.get(dataset) == null) { System.out.println("No dataset id for " + dataset); System.exit(0); } } String firstGroupStr = ""; String secondGroupStr = ""; for (String dataset : firstGroup) { firstGroupStr += datasetId.get(dataset) + ","; } for (String dataset : secondGroup) { secondGroupStr += datasetId.get(dataset) + ","; } firstGroupStr = firstGroupStr.substring(0, firstGroupStr.length() - 1); secondGroupStr = secondGroupStr.substring(0, secondGroupStr.length() - 1); FrameworkUtils.createDir(s3bucket + FrameworkUtils.correlationTechniquesDir, s3conf, s3); String dataAttributesInputDirs = ""; String noRelationship = ""; HashSet<String> dirs = new HashSet<String>(); String dataset1; String dataset2; String datasetId1; String datasetId2; for (int i = 0; i < firstGroup.size(); i++) { for (int j = 0; j < secondGroup.size(); j++) { if (Integer.parseInt(datasetId.get(firstGroup.get(i))) < Integer .parseInt(datasetId.get(secondGroup.get(j)))) { dataset1 = firstGroup.get(i); dataset2 = secondGroup.get(j); } else { dataset1 = secondGroup.get(j); dataset2 = firstGroup.get(i); } datasetId1 = datasetId.get(dataset1); datasetId2 = datasetId.get(dataset2); if (dataset1.equals(dataset2)) continue; String correlationOutputFileName = s3bucket + FrameworkUtils.correlationTechniquesDir + "/" + dataset1 + "-" + dataset2 + "/"; if (removeExistingFiles) { FrameworkUtils.removeFile(correlationOutputFileName, s3conf, s3); } if (!FrameworkUtils.fileExists(correlationOutputFileName, s3conf, s3)) { dirs.add(s3bucket + FrameworkUtils.aggregatesDir + "/" + dataset1); dirs.add(s3bucket + FrameworkUtils.aggregatesDir + "/" + dataset2); } else { noRelationship += datasetId1 + "-" + datasetId2 + ","; } } } if (dirs.isEmpty()) { System.out.println("All the relationships were already computed."); System.out.println("Use -f in the beginning of the command line to force the computation."); System.exit(0); } for (String dir : dirs) { dataAttributesInputDirs += dir + ","; } Configuration conf = new Configuration(); Machine machineConf = new Machine(machine, nbNodes); String jobName = "correlation"; String correlationOutputDir = s3bucket + FrameworkUtils.correlationTechniquesDir + "/tmp/"; FrameworkUtils.removeFile(correlationOutputDir, s3conf, s3); for (int i = 0; i < shortDataset.size(); i++) { conf.set("dataset-" + datasetId.get(shortDataset.get(i)) + "-agg", datasetAgg.get(shortDataset.get(i))); } for (int i = 0; i < shortDataset.size(); i++) { conf.set("dataset-" + datasetId.get(shortDataset.get(i)) + "-agg-size", Integer.toString(datasetAgg.get(shortDataset.get(i)).split(",").length)); } conf.set("dataset-keys", datasetIds); conf.set("dataset-names", datasetNames); conf.set("first-group", firstGroupStr); conf.set("second-group", secondGroupStr); conf.set("main-dataset-id", datasetId.get(shortDataset.get(0))); if (noRelationship.length() > 0) { conf.set("no-relationship", noRelationship.substring(0, noRelationship.length() - 1)); } conf.set("mapreduce.tasktracker.map.tasks.maximum", String.valueOf(machineConf.getMaximumTasks())); conf.set("mapreduce.tasktracker.reduce.tasks.maximum", String.valueOf(machineConf.getMaximumTasks())); conf.set("mapreduce.jobtracker.maxtasks.perjob", "-1"); conf.set("mapreduce.reduce.shuffle.parallelcopies", "20"); conf.set("mapreduce.input.fileinputformat.split.minsize", "0"); conf.set("mapreduce.task.io.sort.mb", "200"); conf.set("mapreduce.task.io.sort.factor", "100"); conf.set("mapreduce.task.timeout", "2400000"); if (s3) { machineConf.setMachineConfiguration(conf); conf.set("fs.s3.awsAccessKeyId", awsAccessKeyId); conf.set("fs.s3.awsSecretAccessKey", awsSecretAccessKey); conf.set("bucket", s3bucket); } if (snappyCompression) { conf.set("mapreduce.map.output.compress", "true"); conf.set("mapreduce.map.output.compress.codec", "org.apache.hadoop.io.compress.SnappyCodec"); //conf.set("mapreduce.output.fileoutputformat.compress.codec", "org.apache.hadoop.io.compress.SnappyCodec"); } if (bzip2Compression) { conf.set("mapreduce.map.output.compress", "true"); conf.set("mapreduce.map.output.compress.codec", "org.apache.hadoop.io.compress.BZip2Codec"); //conf.set("mapreduce.output.fileoutputformat.compress.codec", "org.apache.hadoop.io.compress.BZip2Codec"); } Job job = new Job(conf); job.setJobName(jobName); job.setMapOutputKeyClass(PairAttributeWritable.class); job.setMapOutputValueClass(SpatioTemporalValueWritable.class); job.setOutputKeyClass(Text.class); job.setOutputValueClass(Text.class); job.setMapperClass(CorrelationTechniquesMapper.class); job.setReducerClass(CorrelationTechniquesReducer.class); job.setNumReduceTasks(machineConf.getNumberReduces()); job.setInputFormatClass(SequenceFileInputFormat.class); LazyOutputFormat.setOutputFormatClass(job, TextOutputFormat.class); FileInputFormat.setInputDirRecursive(job, true); FileInputFormat.setInputPaths(job, dataAttributesInputDirs.substring(0, dataAttributesInputDirs.length() - 1)); FileOutputFormat.setOutputPath(job, new Path(correlationOutputDir)); job.setJarByClass(CorrelationTechniques.class); long start = System.currentTimeMillis(); job.submit(); job.waitForCompletion(true); System.out.println(jobName + "\t" + (System.currentTimeMillis() - start)); // moving files to right place for (int i = 0; i < firstGroup.size(); i++) { for (int j = 0; j < secondGroup.size(); j++) { if (Integer.parseInt(datasetId.get(firstGroup.get(i))) < Integer .parseInt(datasetId.get(secondGroup.get(j)))) { dataset1 = firstGroup.get(i); dataset2 = secondGroup.get(j); } else { dataset1 = secondGroup.get(j); dataset2 = firstGroup.get(i); } if (dataset1.equals(dataset2)) continue; String from = s3bucket + FrameworkUtils.correlationTechniquesDir + "/tmp/" + dataset1 + "-" + dataset2 + "/"; String to = s3bucket + FrameworkUtils.correlationTechniquesDir + "/" + dataset1 + "-" + dataset2 + "/"; FrameworkUtils.renameFile(from, to, s3conf, s3); } } }
From source file:eu.scape_project.spacip.Spacip.java
License:Apache License
/** * Start Hadoop job//from www . j a v a 2 s . co m * * @param conf Hadoop job configuration */ public static void startHadoopJob(Configuration conf) { try { Job job = new Job(conf, "spacip_" + conf.getInt("num_items_per_task", 0)); // local debugging (pseudo-distributed) // job.getConfiguration().set("mapred.job.tracker", "local"); // job.getConfiguration().set("fs.default.name", "file:///"); job.setJarByClass(Spacip.class); job.setMapperClass(Spacip.ContainerProcessingMapper.class); // No reducer needed job.setNumReduceTasks(0); job.setInputFormatClass(TextInputFormat.class); MultipleOutputs.addNamedOutput(job, "keyfilmapping", TextOutputFormat.class, Text.class, Text.class); MultipleOutputs.addNamedOutput(job, "tomarinput", TextOutputFormat.class, Text.class, Text.class); MultipleOutputs.addNamedOutput(job, "error", TextOutputFormat.class, Text.class, Text.class); job.setMapOutputKeyClass(Text.class); job.setMapOutputValueClass(ObjectWritable.class); job.setOutputKeyClass(Text.class); job.setOutputValueClass(ObjectWritable.class); TextInputFormat.addInputPath(job, new Path(config.getDirStr())); String outpath = StringUtils.normdir(conf.get("joboutput_hdfs_path", "spacip_joboutput")) + System.currentTimeMillis(); FileOutputFormat.setOutputPath(job, new Path(outpath)); LazyOutputFormat.setOutputFormatClass(job, TextOutputFormat.class); job.waitForCompletion(true); // print output path (taverna integration) System.out.print(outpath); System.exit(0); } catch (Exception e) { logger.error("I/O error", e); } }
From source file:eu.scape_project.tpid.TomarPrepareInputdata.java
License:Apache License
/** * Start Hadoop job// w w w. jav a 2 s. com * * @param conf Hadoop job configuration */ public static void startHadoopJob(Configuration conf) { try { Job job = new Job(conf, "tpid_" + conf.getInt("num_items_per_task", 0)); if (conf.getBoolean("pseudo_distributed", false)) { job.getConfiguration().set("mapred.job.tracker", "local"); job.getConfiguration().set("fs.default.name", "file:///"); } job.setJarByClass(TomarPrepareInputdata.class); job.setMapperClass(TomarPrepareInputdata.ContainerProcessingMapper.class); // No reducer needed job.setNumReduceTasks(0); job.setInputFormatClass(TextInputFormat.class); MultipleOutputs.addNamedOutput(job, "keyfilmapping", TextOutputFormat.class, Text.class, Text.class); MultipleOutputs.addNamedOutput(job, "tomarinput", TextOutputFormat.class, Text.class, Text.class); MultipleOutputs.addNamedOutput(job, "error", TextOutputFormat.class, Text.class, Text.class); job.setMapOutputKeyClass(Text.class); job.setMapOutputValueClass(ObjectWritable.class); job.setOutputKeyClass(Text.class); job.setOutputValueClass(ObjectWritable.class); TextInputFormat.addInputPath(job, new Path(config.getInputStr())); String outpath = StringUtils.normdir(conf.get("joboutput_hdfs_path", "tpid_joboutput")) + System.currentTimeMillis(); FileOutputFormat.setOutputPath(job, new Path(outpath)); LazyOutputFormat.setOutputFormatClass(job, TextOutputFormat.class); job.waitForCompletion(true); // print output path (taverna integration) System.out.print(outpath); System.exit(0); } catch (Exception e) { LOG.error("I/O error", e); } }