List of usage examples for org.apache.hadoop.mapreduce.lib.output SequenceFileOutputFormat setOutputCompressionType
public static void setOutputCompressionType(Job job, CompressionType style)
From source file:edu.isi.mavuno.app.nlp.TratzParse.java
License:Apache License
@SuppressWarnings({ "unchecked", "rawtypes" }) public int run() throws ClassNotFoundException, InterruptedException, IOException { Configuration conf = getConf(); String corpusPath = MavunoUtils.getRequiredParam("Mavuno.Parse.CorpusPath", conf); String corpusClass = MavunoUtils.getRequiredParam("Mavuno.Parse.CorpusClass", conf); String outputPath = MavunoUtils.getRequiredParam("Mavuno.Parse.OutputPath", conf); // optional parameter that allows the parsed documents to be output in text format String textOutput = MavunoUtils.getOptionalParam("Mavuno.Parse.TextOutputFormat", conf); boolean textOutputFormat = false; if (textOutput != null && Boolean.parseBoolean(textOutput)) { textOutputFormat = true;/*from www.j a v a 2 s . co m*/ } sLogger.info("Tool name: TratzParse"); sLogger.info(" - Corpus path: " + corpusPath); sLogger.info(" - Corpus class: " + corpusClass); sLogger.info(" - Output path: " + outputPath); Job job = new Job(conf); job.setJobName("TratzParse"); MavunoUtils.recursivelyAddInputPaths(job, corpusPath); FileOutputFormat.setOutputPath(job, new Path(outputPath)); job.setInputFormatClass((Class<? extends InputFormat>) Class.forName(corpusClass)); // output format -- either plain text or sequencefile (default) if (textOutputFormat) { job.setOutputFormatClass(TextOutputFormat.class); } else { job.setOutputFormatClass(SequenceFileOutputFormat.class); FileOutputFormat.setCompressOutput(job, true); SequenceFileOutputFormat.setOutputCompressionType(job, SequenceFile.CompressionType.BLOCK); } job.setMapOutputKeyClass(Text.class); job.setMapOutputValueClass(TratzParsedDocument.class); job.setOutputKeyClass(Text.class); job.setOutputValueClass(TratzParsedDocument.class); job.setMapperClass(MyMapper.class); job.setJarByClass(TratzParse.class); // no reducers needed job.setNumReduceTasks(0); // run job job.waitForCompletion(true); // print job statistics Counters counters = job.getCounters(); sLogger.info(" - Total documents: " + counters.findCounter(StatCounters.TOTAL_DOCUMENTS).getValue()); sLogger.info(" - Total sentences: " + counters.findCounter(StatCounters.TOTAL_SENTENCES).getValue()); sLogger.info(" - Total tokens: " + counters.findCounter(StatCounters.TOTAL_TOKENS).getValue()); sLogger.info(" - Total dropped sentences: " + counters.findCounter(StatCounters.TOTAL_DROPPED_SENTENCES).getValue()); sLogger.info( " - Total tokenization time (ms): " + counters.findCounter(StatCounters.TOKENIZE_TIME).getValue()); sLogger.info( " - Total POS tagging time (ms): " + counters.findCounter(StatCounters.POSTAG_TIME).getValue()); sLogger.info(" - Total chunking time (ms): " + counters.findCounter(StatCounters.CHUNK_TIME).getValue()); sLogger.info(" - Total named entity tagging time (ms): " + counters.findCounter(StatCounters.NETAG_TIME).getValue()); sLogger.info(" - Total parse time (ms): " + counters.findCounter(StatCounters.PARSE_TIME).getValue()); return 0; }
From source file:edu.isi.mavuno.extract.CombineGlobalStats.java
License:Apache License
public int run() throws ClassNotFoundException, InterruptedException, IOException { Configuration conf = getConf(); String inputPath = MavunoUtils.getRequiredParam("Mavuno.CombineGlobalStats.InputPath", conf); String outputPath = MavunoUtils.getRequiredParam("Mavuno.CombineGlobalStats.OutputPath", conf); int numSplits = conf.getInt("Mavuno.CombineGlobalStats.TotalSplits", 1); sLogger.info("Tool name: CombineGlobalStats"); sLogger.info(" - Input path: " + inputPath); sLogger.info(" - Output path: " + outputPath); sLogger.info(" - Number of splits: " + numSplits); Job job = new Job(conf); job.setJobName("CombineGlobalStats"); for (int split = 0; split < numSplits; split++) { FileInputFormat.addInputPath(job, new Path(inputPath + "/" + split)); }//from w w w . ja va 2 s . c o m FileOutputFormat.setOutputPath(job, new Path(outputPath)); job.setInputFormatClass(SequenceFileInputFormat.class); job.setOutputFormatClass(SequenceFileOutputFormat.class); FileOutputFormat.setCompressOutput(job, true); SequenceFileOutputFormat.setOutputCompressionType(job, SequenceFile.CompressionType.BLOCK); job.setMapOutputKeyClass(ContextPatternWritable.class); job.setSortComparatorClass(ContextPatternWritable.Comparator.class); job.setMapOutputValueClass(ContextPatternStatsWritable.class); job.setOutputKeyClass(ContextPatternWritable.class); job.setOutputValueClass(ContextPatternStatsWritable.class); job.setMapperClass(MyMapper.class); job.setCombinerClass(MyReducer.class); job.setReducerClass(MyReducer.class); job.waitForCompletion(true); return 0; }
From source file:edu.isi.mavuno.extract.CombineSplits.java
License:Apache License
public int run() throws ClassNotFoundException, InterruptedException, IOException { Configuration conf = getConf(); String examplesPath = MavunoUtils.getRequiredParam("Mavuno.CombineSplits.ExamplesPath", conf); String exampleStatsPath = MavunoUtils.getRequiredParam("Mavuno.CombineSplits.ExampleStatsPath", conf); String splitKey = MavunoUtils.getRequiredParam("Mavuno.CombineSplits.SplitKey", conf).toLowerCase(); int numSplits = conf.getInt("Mavuno.CombineSplits.TotalSplits", 1); String outputPath = MavunoUtils.getRequiredParam("Mavuno.CombineSplits.OutputPath", conf); sLogger.info("Tool name: CombineSplits"); sLogger.info(" - Examples path: " + examplesPath); sLogger.info(" - Example stats path: " + exampleStatsPath); sLogger.info(" - Split key: " + splitKey); sLogger.info(" - Total splits: " + numSplits); sLogger.info(" - Output path: " + outputPath); Job job = new Job(conf); job.setJobName("CombineSplits"); for (int split = 0; split < numSplits; split++) { FileInputFormat.addInputPath(job, new Path(examplesPath + "/" + split)); }// w ww. j a v a2 s . c o m if (MavunoUtils.pathExists(conf, exampleStatsPath)) { FileInputFormat.addInputPath(job, new Path(exampleStatsPath)); } FileOutputFormat.setOutputPath(job, new Path(outputPath)); job.setInputFormatClass(SequenceFileInputFormat.class); job.setOutputFormatClass(SequenceFileOutputFormat.class); FileOutputFormat.setCompressOutput(job, true); SequenceFileOutputFormat.setOutputCompressionType(job, SequenceFile.CompressionType.BLOCK); job.setMapOutputKeyClass(ContextPatternWritable.class); if ("pattern".equals(splitKey)) { job.setSortComparatorClass(ContextPatternWritable.Comparator.class); } else if ("context".equals(splitKey)) { job.setSortComparatorClass(ContextPatternWritable.IdPatternComparator.class); } else if ("pattern+context".equals(splitKey)) { job.setSortComparatorClass(ContextPatternWritable.Comparator.class); } else { throw new RuntimeException("Invalid SplitKey in CombineSplits! -- " + splitKey); } job.setMapOutputValueClass(ContextPatternStatsWritable.class); job.setOutputKeyClass(ContextPatternWritable.class); job.setOutputValueClass(ContextPatternStatsWritable.class); job.setMapperClass(MyMapper.class); job.setReducerClass(MyReducer.class); job.waitForCompletion(true); return 0; }
From source file:edu.isi.mavuno.extract.Extract.java
License:Apache License
@SuppressWarnings({ "unchecked", "rawtypes" }) public int run() throws ClassNotFoundException, InterruptedException, IOException { Configuration conf = getConf(); String inputPath = MavunoUtils.getRequiredParam("Mavuno.Extract.InputPath", conf); String corpusPath = MavunoUtils.getRequiredParam("Mavuno.Extract.CorpusPath", conf); String corpusClass = MavunoUtils.getRequiredParam("Mavuno.Extract.CorpusClass", conf); String extractorClass = MavunoUtils.getRequiredParam("Mavuno.Extract.ExtractorClass", conf); String extractorArgs = MavunoUtils.getRequiredParam("Mavuno.Extract.ExtractorArgs", conf); String extractorTarget = MavunoUtils.getRequiredParam("Mavuno.Extract.ExtractorTarget", conf).toLowerCase(); int minContextMatches = Integer.parseInt(MavunoUtils.getRequiredParam("Mavuno.Extract.MinMatches", conf)); String outputPath = MavunoUtils.getRequiredParam("Mavuno.Extract.OutputPath", conf); sLogger.info("Tool name: Extract"); sLogger.info(" - Input path: " + inputPath); sLogger.info(" - Corpus path: " + corpusPath); sLogger.info(" - Corpus class: " + corpusClass); sLogger.info(" - Extractor class: " + extractorClass); sLogger.info(" - Extractor arguments: " + extractorArgs); sLogger.info(" - Extractor target: " + extractorTarget); sLogger.info(" - Min context matches: " + minContextMatches); sLogger.info(" - Output path: " + outputPath); Job job = new Job(conf); job.setJobName("Extract"); MavunoUtils.recursivelyAddInputPaths(job, corpusPath); FileOutputFormat.setOutputPath(job, new Path(outputPath)); job.setInputFormatClass((Class<? extends InputFormat>) Class.forName(corpusClass)); job.setOutputFormatClass(SequenceFileOutputFormat.class); FileOutputFormat.setCompressOutput(job, true); SequenceFileOutputFormat.setOutputCompressionType(job, SequenceFile.CompressionType.BLOCK); if ("pattern".equals(extractorTarget)) { job.setSortComparatorClass(ContextPatternWritable.Comparator.class); job.setPartitionerClass(ContextPatternWritable.IdContextPartitioner.class); } else if ("context".equals(extractorTarget)) { job.setSortComparatorClass(ContextPatternWritable.IdPatternComparator.class); job.setPartitionerClass(ContextPatternWritable.IdPatternPartitioner.class); } else {//w w w . j a v a2 s .c o m throw new RuntimeException("Invalid extractor target in Extract -- " + extractorTarget); } job.setMapOutputKeyClass(ContextPatternWritable.class); job.setMapOutputValueClass(ContextPatternStatsWritable.class); job.setOutputKeyClass(ContextPatternWritable.class); job.setOutputValueClass(ContextPatternStatsWritable.class); job.setMapperClass(MyMapper.class); job.setCombinerClass(MyCombiner.class); job.setReducerClass(MyReducer.class); job.waitForCompletion(true); return 0; }
From source file:edu.isi.mavuno.extract.ExtractGlobalStats.java
License:Apache License
@SuppressWarnings({ "unchecked", "rawtypes" }) public int run() throws ClassNotFoundException, InterruptedException, IOException { Configuration conf = getConf(); String inputPath = MavunoUtils.getRequiredParam("Mavuno.ExtractGlobalStats.InputPath", conf); String corpusPath = MavunoUtils.getRequiredParam("Mavuno.ExtractGlobalStats.CorpusPath", conf); String corpusClass = MavunoUtils.getRequiredParam("Mavuno.ExtractGlobalStats.CorpusClass", conf); String extractorClass = MavunoUtils.getRequiredParam("Mavuno.ExtractGlobalStats.ExtractorClass", conf); String extractorArgs = MavunoUtils.getRequiredParam("Mavuno.ExtractGlobalStats.ExtractorArgs", conf); String extractorTarget = MavunoUtils.getRequiredParam("Mavuno.ExtractGlobalStats.ExtractorTarget", conf) .toLowerCase();//from w w w . ja v a2 s. co m String outputPath = MavunoUtils.getRequiredParam("Mavuno.ExtractGlobalStats.OutputPath", conf); // split examples conf.set("Mavuno.Split.InputPath", inputPath); conf.set("Mavuno.Split.OutputPath", outputPath + "/../split"); conf.set("Mavuno.Split.SplitKey", extractorTarget); new Split(conf).run(); // get splits FileStatus[] files = MavunoUtils.getDirectoryListing(conf, outputPath + "/../split"); int split = 0; for (FileStatus file : files) { if (!file.getPath().getName().endsWith(".examples")) { continue; } conf.set("Mavuno.ExtractGlobalStats.ExamplesPath", file.getPath().toString()); sLogger.info("Tool name: ExtractGlobalStats"); sLogger.info(" - Input path: " + inputPath); sLogger.info(" - Examples path: " + file.getPath()); sLogger.info(" - Example split: " + split); sLogger.info(" - Corpus path: " + corpusPath); sLogger.info(" - Corpus class: " + corpusClass); sLogger.info(" - Extractor class: " + extractorClass); sLogger.info(" - Extractor class: " + extractorArgs); sLogger.info(" - Extractor target: " + extractorTarget); sLogger.info(" - Output path: " + outputPath); Job job = new Job(conf); job.setJobName("ExtractGlobalStats"); MavunoUtils.recursivelyAddInputPaths(job, corpusPath); FileOutputFormat.setOutputPath(job, new Path(outputPath + "/../split/" + split)); job.setInputFormatClass((Class<? extends InputFormat>) Class.forName(corpusClass)); job.setOutputFormatClass(SequenceFileOutputFormat.class); FileOutputFormat.setCompressOutput(job, true); SequenceFileOutputFormat.setOutputCompressionType(job, SequenceFile.CompressionType.BLOCK); job.setMapOutputKeyClass(ContextPatternWritable.class); job.setSortComparatorClass(ContextPatternWritable.Comparator.class); job.setPartitionerClass(ContextPatternWritable.FullPartitioner.class); job.setMapOutputValueClass(ContextPatternStatsWritable.class); job.setOutputKeyClass(ContextPatternWritable.class); job.setOutputValueClass(ContextPatternStatsWritable.class); job.setMapperClass(MyMapper.class); job.setReducerClass(MyReducer.class); job.waitForCompletion(true); split++; } // combine splits conf.setInt("Mavuno.CombineGlobalStats.TotalSplits", split); conf.set("Mavuno.CombineGlobalStats.InputPath", outputPath + "/../split/"); conf.set("Mavuno.CombineGlobalStats.OutputPath", outputPath); new CombineGlobalStats(conf).run(); MavunoUtils.removeDirectory(conf, outputPath + "/../split"); return 0; }
From source file:edu.isi.mavuno.score.ScoreContexts.java
License:Apache License
public int run() throws ClassNotFoundException, InterruptedException, IOException { Configuration conf = getConf(); String inputPath = MavunoUtils.getRequiredParam("Mavuno.ScoreContexts.InputPath", conf); String outputPath = MavunoUtils.getRequiredParam("Mavuno.ScoreContexts.OutputPath", conf); String contextScorerClass = MavunoUtils.getRequiredParam("Mavuno.Scorer.Class", conf); String contextScorerArgs = MavunoUtils.getRequiredParam("Mavuno.Scorer.Args", conf); sLogger.info("Tool name: ScoreContexts"); sLogger.info(" - Input path: " + inputPath); sLogger.info(" - Output path: " + outputPath); sLogger.info(" - Context scorer class: " + contextScorerClass); sLogger.info(" - Context scorer args: " + contextScorerArgs); Job job = new Job(conf); job.setJobName("ScoreContexts"); FileInputFormat.addInputPath(job, new Path(inputPath)); FileOutputFormat.setOutputPath(job, new Path(outputPath)); job.setInputFormatClass(SequenceFileInputFormat.class); job.setSortComparatorClass(ContextPatternWritable.Comparator.class); job.setPartitionerClass(ContextPatternWritable.IdPartitioner.class); job.setOutputFormatClass(SequenceFileOutputFormat.class); FileOutputFormat.setCompressOutput(job, true); SequenceFileOutputFormat.setOutputCompressionType(job, SequenceFile.CompressionType.BLOCK); job.setMapOutputKeyClass(ContextPatternWritable.class); job.setMapOutputValueClass(ScoreWritable.class); job.setOutputKeyClass(ContextPatternWritable.class); job.setOutputValueClass(DoubleWritable.class); job.setMapperClass(MyMapper.class); job.setReducerClass(MyReducer.class); job.waitForCompletion(true);/*from w ww . j a v a 2 s . c om*/ return 0; }
From source file:edu.isi.mavuno.score.ScorePatterns.java
License:Apache License
public int run() throws ClassNotFoundException, InterruptedException, IOException { Configuration conf = getConf(); String inputPath = MavunoUtils.getRequiredParam("Mavuno.ScorePatterns.InputPath", conf); String outputPath = MavunoUtils.getRequiredParam("Mavuno.ScorePatterns.OutputPath", conf); String patternScorerClass = MavunoUtils.getRequiredParam("Mavuno.Scorer.Class", conf); String patternScorerArgs = MavunoUtils.getRequiredParam("Mavuno.Scorer.Args", conf); sLogger.info("Tool name: ScorePatterns"); sLogger.info(" - Input path: " + inputPath); sLogger.info(" - Output path: " + outputPath); sLogger.info(" - Pattern scorer class: " + patternScorerClass); sLogger.info(" - Pattern scorer args: " + patternScorerArgs); Job job = new Job(conf); job.setJobName("ScorePatterns"); FileInputFormat.addInputPath(job, new Path(inputPath)); FileOutputFormat.setOutputPath(job, new Path(outputPath)); job.setInputFormatClass(SequenceFileInputFormat.class); job.setSortComparatorClass(ContextPatternWritable.IdPatternComparator.class); job.setPartitionerClass(ContextPatternWritable.IdPartitioner.class); job.setOutputFormatClass(SequenceFileOutputFormat.class); FileOutputFormat.setCompressOutput(job, true); SequenceFileOutputFormat.setOutputCompressionType(job, SequenceFile.CompressionType.BLOCK); job.setMapOutputKeyClass(ContextPatternWritable.class); job.setMapOutputValueClass(ScoreWritable.class); job.setOutputKeyClass(ContextPatternWritable.class); job.setOutputValueClass(DoubleWritable.class); job.setMapperClass(MyMapper.class); job.setReducerClass(MyReducer.class); job.waitForCompletion(true);/* w w w.j av a2 s .c om*/ return 0; }
From source file:edu.nyu.vida.data_polygamy.feature_identification.IndexCreation.java
License:BSD License
/** * @param args/*ww w .j ava2 s . c o 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 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.pre_processing.PreProcessing.java
License:BSD License
/** * @param args/*ww w. java2 s . c om*/ * @throws IOException * @throws ClassNotFoundException * @throws InterruptedException */ @SuppressWarnings("deprecation") public static void main(String[] args) throws IOException, InterruptedException, ClassNotFoundException { Options options = new Options(); Option nameOption = new Option("dn", "name", true, "the name of the dataset"); nameOption.setRequired(true); nameOption.setArgName("DATASET NAME"); options.addOption(nameOption); Option headerOption = new Option("dh", "header", true, "the file that contains the header of the dataset"); headerOption.setRequired(true); headerOption.setArgName("DATASET HEADER FILE"); options.addOption(headerOption); Option deafultsOption = new Option("dd", "defaults", true, "the file that contains the default values of the dataset"); deafultsOption.setRequired(true); deafultsOption.setArgName("DATASET DEFAULTS FILE"); options.addOption(deafultsOption); Option tempResOption = new Option("t", "temporal", true, "desired temporal resolution (hour, day, week, or month)"); tempResOption.setRequired(true); tempResOption.setArgName("TEMPORAL RESOLUTION"); options.addOption(tempResOption); Option spatialResOption = new Option("s", "spatial", true, "desired spatial resolution (points, nbhd, zip, grid, or city)"); spatialResOption.setRequired(true); spatialResOption.setArgName("SPATIAL RESOLUTION"); options.addOption(spatialResOption); Option currentSpatialResOption = new Option("cs", "current-spatial", true, "current spatial resolution (points, nbhd, zip, grid, or city)"); currentSpatialResOption.setRequired(true); currentSpatialResOption.setArgName("CURRENT SPATIAL RESOLUTION"); options.addOption(currentSpatialResOption); Option indexResOption = new Option("i", "index", true, "indexes of the temporal and spatial attributes"); indexResOption.setRequired(true); indexResOption.setArgName("INDEX OF SPATIO-TEMPORAL RESOLUTIONS"); indexResOption.setArgs(Option.UNLIMITED_VALUES); options.addOption(indexResOption); 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.pre_processing.PreProcessing", options, true); System.exit(0); } if (cmd.hasOption("h")) { formatter.printHelp( "hadoop jar data-polygamy.jar " + "edu.nyu.vida.data_polygamy.pre_processing.PreProcessing", 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.pre_processing.PreProcessing", 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); } Configuration conf = new Configuration(); Machine machineConf = new Machine(machine, nbNodes); String dataset = cmd.getOptionValue("dn"); String header = cmd.getOptionValue("dh"); String defaults = cmd.getOptionValue("dd"); String temporalResolution = cmd.getOptionValue("t"); String spatialResolution = cmd.getOptionValue("s"); String gridResolution = ""; String currentSpatialResolution = cmd.getOptionValue("cs"); if (spatialResolution.contains("grid")) { String[] res = spatialResolution.split("-"); spatialResolution = res[0]; gridResolution = res[1]; } conf.set("header", s3bucket + FrameworkUtils.dataDir + "/" + header); conf.set("defaults", s3bucket + FrameworkUtils.dataDir + "/" + defaults); conf.set("temporal-resolution", temporalResolution); conf.set("spatial-resolution", spatialResolution); conf.set("grid-resolution", gridResolution); conf.set("current-spatial-resolution", currentSpatialResolution); String[] indexes = cmd.getOptionValues("i"); String temporalPos = ""; Integer sizeSpatioTemp = 0; if (!(currentSpatialResolution.equals("points"))) { String spatialPos = ""; for (int i = 0; i < indexes.length; i++) { temporalPos += indexes[i] + ","; spatialPos += indexes[++i] + ","; sizeSpatioTemp++; } conf.set("spatial-pos", spatialPos); } else { String xPositions = "", yPositions = ""; for (int i = 0; i < indexes.length; i++) { temporalPos += indexes[i] + ","; xPositions += indexes[++i] + ","; yPositions += indexes[++i] + ","; sizeSpatioTemp++; } conf.set("xPositions", xPositions); conf.set("yPositions", yPositions); } conf.set("temporal-pos", temporalPos); conf.set("size-spatio-temporal", sizeSpatioTemp.toString()); // checking resolutions if (utils.spatialResolution(spatialResolution) < 0) { System.out.println("Invalid spatial resolution: " + spatialResolution); System.exit(-1); } if (utils.spatialResolution(spatialResolution) == FrameworkUtils.POINTS) { System.out.println("The data needs to be reduced at least to neighborhoods or grid."); System.exit(-1); } if (utils.spatialResolution(currentSpatialResolution) < 0) { System.out.println("Invalid spatial resolution: " + currentSpatialResolution); System.exit(-1); } if (utils.spatialResolution(currentSpatialResolution) > utils.spatialResolution(spatialResolution)) { System.out.println("The current spatial resolution is coarser than " + "the desired one. You can only navigate from a fine resolution" + " to a coarser one."); System.exit(-1); } if (utils.temporalResolution(temporalResolution) < 0) { System.out.println("Invalid temporal resolution: " + temporalResolution); System.exit(-1); } String fileName = s3bucket + FrameworkUtils.preProcessingDir + "/" + dataset + "-" + temporalResolution + "-" + spatialResolution + gridResolution; conf.set("aggregates", fileName + ".aggregates"); // making sure both files are removed, if they exist FrameworkUtils.removeFile(fileName, s3conf, s3); FrameworkUtils.removeFile(fileName + ".aggregates", s3conf, s3); /** * Hadoop Parameters * sources: http://www.slideshare.net/ImpetusInfo/ppt-on-advanced-hadoop-tuning-n-optimisation * https://cloudcelebrity.wordpress.com/2013/08/14/12-key-steps-to-keep-your-hadoop-cluster-running-strong-and-performing-optimum/ */ 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"); // using SnappyCodec for intermediate and output data ? // TODO: for now, using SnappyCodec -- what about LZO + Protocol Buffer serialization? // LZO - http://www.oberhumer.com/opensource/lzo/#download // Hadoop-LZO - https://github.com/twitter/hadoop-lzo // Protocol Buffer - https://github.com/twitter/elephant-bird // General Info - http://www.devx.com/Java/Article/47913 // Compression - http://comphadoop.weebly.com/index.html 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"); } // TODO: this is dangerous! if (s3) { conf.set("fs.s3.awsAccessKeyId", awsAccessKeyId); conf.set("fs.s3.awsSecretAccessKey", awsSecretAccessKey); } Job job = new Job(conf); job.setJobName(dataset + "-" + temporalResolution + "-" + spatialResolution); job.setMapOutputKeyClass(MultipleSpatioTemporalWritable.class); job.setMapOutputValueClass(AggregationArrayWritable.class); job.setOutputKeyClass(MultipleSpatioTemporalWritable.class); job.setOutputValueClass(AggregationArrayWritable.class); job.setMapperClass(PreProcessingMapper.class); job.setCombinerClass(PreProcessingCombiner.class); job.setReducerClass(PreProcessingReducer.class); job.setNumReduceTasks(machineConf.getNumberReduces()); //job.setNumReduceTasks(1); job.setInputFormatClass(TextInputFormat.class); job.setOutputFormatClass(SequenceFileOutputFormat.class); SequenceFileOutputFormat.setCompressOutput(job, true); SequenceFileOutputFormat.setOutputCompressionType(job, CompressionType.BLOCK); FileInputFormat.setInputPaths(job, new Path(s3bucket + FrameworkUtils.dataDir + "/" + dataset)); FileOutputFormat.setOutputPath(job, new Path(fileName)); job.setJarByClass(PreProcessing.class); long start = System.currentTimeMillis(); job.submit(); job.waitForCompletion(true); System.out.println(fileName + "\t" + (System.currentTimeMillis() - start)); }
From source file:edu.nyu.vida.data_polygamy.scalar_function_computation.Aggregation.java
License:BSD License
/** * @param args/*from www .j a va 2 s . c o 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); } }