一个mapreduce得到需要计算单词概率的基础数据
第一步,先计算需要计算概率的词频,单词种类数,类别单词总数(类别均是按照文件夹名区分)(基础数据以及分词了,每个单词一行,以及预处理好)
![](https://images.cnblogs.com/OutliningIndicators/ContractedBlock.gif)
package org.lukey.hadoop.classifyBayes; import java.io.IOException; import java.net.URI; import java.util.ArrayList; import java.util.HashMap; import java.util.List; import java.util.Map; import org.apache.hadoop.conf.Configuration; import org.apache.hadoop.fs.FSDataOutputStream; import org.apache.hadoop.fs.FileStatus; import org.apache.hadoop.fs.FileSystem; import org.apache.hadoop.fs.Path; import org.apache.hadoop.io.IOUtils; import org.apache.hadoop.io.IntWritable; import org.apache.hadoop.io.LongWritable; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.Counter; import org.apache.hadoop.mapreduce.Counters; import org.apache.hadoop.mapreduce.Job; import org.apache.hadoop.mapreduce.Mapper; import org.apache.hadoop.mapreduce.Reducer; import org.apache.hadoop.mapreduce.lib.input.FileInputFormat; import org.apache.hadoop.mapreduce.lib.input.FileSplit; import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat; import org.apache.hadoop.mapreduce.lib.output.MultipleOutputs; /** * * 一次将需要的结果都统计到对应的文件夹中 AFRICA 484017newsML.txt afford 1 * * 按照这个格式输出给后面处理得到需要的: 1. AFRICA 484017newsML.txt AFRICA 487141newsML.txt * 类别中的文本数, ---> 计算先验概率(单独解决这个) 所有类别中的文本总数, ---> 可以由上面得到,计算先验概率 * * 2. AFRICA afford 1 AFRICA boy 3 每个类中的每个单词的个数,---> 计算各个类中单词的概率 * * 3. AFRICA 768 类中单词总数, ---> 将2中的第一个key相同的第三个数相加即可 * * 4. AllWORDS 12345 所有类别中单词种类数 ---> 将1中的第三个key归并,计算个数 * */ public class MyWordCount { private static MultipleOutputs<Text, IntWritable> mos; static String baseOutputPath = "/user/hadoop/test_out"; // 设计两个map分别计算每个类别的文本数//和每个类别的单词总数 private static Map<String, List<String>> fileCountMap = new HashMap<String, List<String>>(); private static Map<String, Integer> fileCount = new HashMap<String, Integer>(); // static Map<String, List<String>> wordsCountInClassMap = new // HashMap<String, List<String>>(); static enum WordsNature { CLSASS_NUMBER, CLASS_WORDS, TOTALWORDS } public static void main(String[] args) throws Exception { Configuration conf = new Configuration(); // 设置不同文件的路径 // 文本数路径 String priorProbality = "hdfs://192.168.190.128:9000/user/hadoop/output/priorP/priorProbality.txt"; conf.set("priorProbality", priorProbality); String[] otherArgs = { "/user/hadoop/input/NBCorpus/Country", "/user/hadoop/mid/wordsFre" }; Job job = new Job(conf, "file count"); job.setJarByClass(MyWordCount.class); // job.setInputFormatClass(CustomInputFormat.class); job.setMapperClass(First_Mapper.class); job.setReducerClass(First_Reducer.class); //过滤掉文本数少于10的类别 List<Path> inputPaths = getSecondDir(conf, otherArgs[0]); for (Path path : inputPaths) { FileInputFormat.addInputPath(job, path); } // 调用自己写的方法 // MyUtils.addInputPath(job, inputpath, conf); // CustomInputFormat.setInputPaths(job, inputpath); // FileInputFormat.addInputPath(job, inputpath); FileOutputFormat.setOutputPath(job, new Path(otherArgs[1])); job.setOutputKeyClass(Text.class); job.setOutputValueClass(IntWritable.class); int exitCode = job.waitForCompletion(true) ? 0 : 1; // 调用计数器 Counters counters = job.getCounters(); Counter c1 = counters.findCounter(WordsNature.TOTALWORDS); System.out.println("-------------->>>>: " + c1.getDisplayName() + ":" + c1.getName() + ": " + c1.getValue()); // 将单词种类数写入文件中 Path totalWordsPath = new Path("/user/hadoop/output/totalwords.txt"); FileSystem fs = FileSystem.get(conf); FSDataOutputStream outputStream = fs.create(totalWordsPath); outputStream.writeBytes(c1.getDisplayName() + ":" + c1.getValue()); IOUtils.closeStream(outputStream); // 下次求概率是尝试单词总种类数写到configuration中 // // conf.set("TOTALWORDS", totalWords.toString()); System.exit(exitCode); } // Mapper static class First_Mapper extends Mapper<LongWritable, Text, Text, IntWritable> { private final static IntWritable one = new IntWritable(1); private final static IntWritable zero = new IntWritable(0); private Text className = new Text(); private Text countryName = new Text(); @Override protected void cleanup(Mapper<LongWritable, Text, Text, IntWritable>.Context context) throws IOException, InterruptedException { Configuration conf = context.getConfiguration(); String file = conf.get("priorProbality"); FileSystem fs = FileSystem.get(URI.create(file), conf); Path priorPath = new Path(file); FSDataOutputStream priorStream = fs.create(priorPath); for (Map.Entry<String, List<String>> entry : fileCountMap.entrySet()) { fileCount.put(entry.getKey(), entry.getValue().size()); priorStream.writeBytes(entry.getKey() + "\t" + entry.getValue().size()); } // 求文本总数 int fileSum = 0; for (Integer num : fileCount.values()) { fileSum += num; } System.out.println("fileSum = " + fileSum); // 计算每个类的先验概率并写入文件 for (Map.Entry<String, Integer> entry : fileCount.entrySet()) { double p = (double) entry.getValue() / fileSum; priorStream.writeBytes(entry.getKey() + ":" + p); } IOUtils.closeStream(priorStream); } @Override protected void map(LongWritable key, Text value, Mapper<LongWritable, Text, Text, IntWritable>.Context context) throws IOException, InterruptedException { // TODO Auto-generated method stub FileSplit fileSplit = (FileSplit) context.getInputSplit(); // 文件名 String fileName = fileSplit.getPath().getName(); // 文件夹名(即类别名) String dirName = fileSplit.getPath().getParent().getName(); className.set(dirName + "\t" + value.toString()); countryName.set(dirName + "\t" + fileName + "\t" + value.toString()); // 将文件名添加到map中用于统计文本个数(单独跑了一个程序计算主要还是为了筛选文本数太少的类别) if (fileCountMap.containsKey(dirName)) { if (!fileCountMap.get(dirName).contains(fileName)) { fileCountMap.get(dirName).add(fileName); } } else { List<String> oneList = new ArrayList<String>(); oneList.add(fileName); fileCountMap.put(dirName, oneList); } context.write(className, one); // 每个类别的每个单词数 // ABDBI hello 1 context.write(new Text(dirName), one);// 统计每个类中的单词总数 //ABDBI 1 context.write(value, zero); // 用于统计所有类中单词个数 } } // Reducer static class First_Reducer extends Reducer<Text, IntWritable, Text, IntWritable> { // result 表示每个类别中每个单词的个数 IntWritable result = new IntWritable(); Map<String, List<String>> classMap = new HashMap<String, List<String>>(); Map<String, List<String>> fileMap = new HashMap<String, List<String>>(); @Override protected void reduce(Text key, Iterable<IntWritable> values, Reducer<Text, IntWritable, Text, IntWritable>.Context context) throws IOException, InterruptedException { int sum = 0; for (IntWritable value : values) { sum += value.get(); } // sum为0,总得单词数加1,统计所有单词的种类 if (sum == 0) { context.getCounter(WordsNature.TOTALWORDS).increment(1); } else {// sum不为0时,通过key的长度来判断, String[] temp = key.toString().split("\t"); if (temp.length == 2) { // 用tab分隔类别和单词 result.set(sum); context.write(key, result); // mos.write(new Text(temp[1]), result, temp[0]); } else { // 类别中单词总数 result.set(sum); mos.write(key, result, "wordsInClass"); } } } @Override protected void cleanup(Reducer<Text, IntWritable, Text, IntWritable>.Context context) throws IOException, InterruptedException { // TODO Auto-generated method stub mos.close(); } @Override protected void setup(Reducer<Text, IntWritable, Text, IntWritable>.Context context) throws IOException, InterruptedException { // TODO Auto-generated method stub mos = new MultipleOutputs<Text, IntWritable>(context); } } // 获取文件夹下面二级文件夹路径的方法 static List<Path> getSecondDir(Configuration conf, String folder) throws Exception { FileSystem fs = FileSystem.get(conf); Path path = new Path(folder); FileStatus[] stats = fs.listStatus(path); List<Path> folderPath = new ArrayList<Path>(); for (FileStatus stat : stats) { if (stat.isDir()) { if (fs.listStatus(stat.getPath()).length > 10) { //筛选出文件数大于10个的类别作为 输入路径 folderPath.add(stat.getPath()); } } } return folderPath; } }
第二步,计算每个类别单词的概率,需提前读取每个类别单词总数,以及总得单词种类数(都可以通过configuration.set)也可以在setup里面先于map处理前读取数据。
![](https://images.cnblogs.com/OutliningIndicators/ContractedBlock.gif)
package org.lukey.hadoop.classifyBayes; import java.io.BufferedReader; import java.io.IOException; import java.io.InputStreamReader; import java.net.URI; import java.util.HashMap; import java.util.Map; import org.apache.commons.logging.Log; import org.apache.commons.logging.LogFactory; import org.apache.hadoop.conf.Configuration; import org.apache.hadoop.fs.FSDataInputStream; import org.apache.hadoop.fs.FileSystem; import org.apache.hadoop.fs.Path; import org.apache.hadoop.io.DoubleWritable; import org.apache.hadoop.io.LongWritable; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.Job; import org.apache.hadoop.mapreduce.Mapper; import org.apache.hadoop.mapreduce.Reducer; import org.apache.hadoop.mapreduce.lib.input.FileInputFormat; import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat; import org.apache.hadoop.mapreduce.lib.output.MultipleOutputs; public class Probability { private static final Log LOG = LogFactory.getLog(FileInputFormat.class); public static int total = 0; private static MultipleOutputs<Text, DoubleWritable> mos; // Client public static void main(String[] args) throws Exception { Configuration conf = new Configuration(); conf.set("mapred.job.tracker", "192.168.190.128:9001"); conf.set("mapred.jar", "probability.jar"); // 读取单词总数,设置到congfiguration中 String totalWordsPath = "hdfs://192.168.190.128:9000/user/hadoop/output/totalwords.txt"; String wordsInClassPath = "hdfs://192.168.190.128:9000/user/hadoop/mid/wordsFrequence/wordsInClass-r-00000"; conf.set("wordsInClassPath", wordsInClassPath); // Map<String, Integer> wordsInClassMap = new HashMap<String, // Integer>();//保存每个类别的单词总数 // 先读取单词总类别数 FileSystem fs = FileSystem.get(URI.create(totalWordsPath), conf); FSDataInputStream inputStream = fs.open(new Path(totalWordsPath)); BufferedReader buffer = new BufferedReader(new InputStreamReader(inputStream)); String strLine = buffer.readLine(); String[] temp = strLine.split(":"); if (temp.length == 2) { // temp[0] = TOTALWORDS conf.set(temp[0], temp[1]);// 设置两个String } total = Integer.parseInt(conf.get("TOTALWORDS")); LOG.info("------>total = " + total); System.out.println("total ==== " + total); /* * String[] otherArgs = new GenericOptionsParser(conf, * args).getRemainingArgs(); * * if (otherArgs.length != 2) { System.out.println("Usage <in> <out>"); * System.exit(-1); } */ Job job = new Job(conf, "file count"); job.setJarByClass(Probability.class); job.setMapperClass(WordsOfClassCountMapper.class); job.setReducerClass(WordsOfClassCountReducer.class); String input = "hdfs://192.168.190.128:9000/user/hadoop/mid/wordsFrequence"; String output = "hdfs://192.168.190.128:9000/user/hadoop/output/probability/"; FileInputFormat.addInputPath(job, new Path(input)); FileOutputFormat.setOutputPath(job, new Path(output)); job.setOutputKeyClass(Text.class); job.setOutputValueClass(DoubleWritable.class); System.exit(job.waitForCompletion(true) ? 0 : 1); } // Mapper static class WordsOfClassCountMapper extends Mapper<LongWritable, Text, Text, DoubleWritable> { private static DoubleWritable number = new DoubleWritable(); private static Text className = new Text(); // 保存类别中单词总数 private static Map<String, Integer> filemap = new HashMap<String, Integer>(); protected void map(LongWritable key, Text value, Mapper<LongWritable, Text, Text, DoubleWritable>.Context context) throws IOException, InterruptedException { Configuration conf = context.getConfiguration(); int tot = Integer.parseInt(conf.get("TOTALWORDS")); System.out.println("total = " + total); System.out.println("tot = " + tot); // 输入的格式如下: // ALB weekend 1 // ALB weeks 3 Map<String, Map<String, Integer>> baseMap = new HashMap<String, Map<String, Integer>>(); // 保存基础数据 // Map<String, Map<String, Double>> priorMap = new HashMap<String, // Map<String, Double>>(); // 保存每个单词出现的概率 String[] temp = value.toString().split("\t"); // 先将数据存到baseMap中 if (temp.length == 3) { // 文件夹名类别名 if (baseMap.containsKey(temp[0])) { baseMap.get(temp[0]).put(temp[1], Integer.parseInt(temp[2])); } else { Map<String, Integer> oneMap = new HashMap<String, Integer>(); oneMap.put(temp[1], Integer.parseInt(temp[2])); baseMap.put(temp[0], oneMap); } } // 读取数据完毕,全部保存在baseMap中 int allWordsInClass = 0; for (Map.Entry<String, Map<String, Integer>> entries : baseMap.entrySet()) { // 遍历类别 allWordsInClass = filemap.get(entries.getKey()); for (Map.Entry<String, Integer> entry : entries.getValue().entrySet()) { // 遍历类别中的单词词频求概率 double p = (entry.getValue() + 1.0) / (allWordsInClass + tot); className.set(entries.getKey() + "\t" + entry.getKey()); number.set(p); LOG.info("------>p = " + p); context.write(className, number); } } } protected void cleanup(Mapper<LongWritable, Text, Text, DoubleWritable>.Context context) throws IOException, InterruptedException { // TODO Auto-generated method stub mos.close(); } protected void setup(Mapper<LongWritable, Text, Text, DoubleWritable>.Context context) throws IOException, InterruptedException { // TODO Auto-generated method stub Configuration conf = context.getConfiguration(); mos = new MultipleOutputs<Text, DoubleWritable>(context); String filePath = conf.get("wordsInClassPath"); FileSystem fs = FileSystem.get(URI.create(filePath), conf); FSDataInputStream inputStream = fs.open(new Path(filePath)); BufferedReader buffer = new BufferedReader(new InputStreamReader(inputStream)); String strLine = null; while ((strLine = buffer.readLine()) != null) { String[] temp = strLine.split("\t"); filemap.put(temp[0], Integer.parseInt(temp[1])); } } } // Reducer static class WordsOfClassCountReducer extends Reducer<Text, DoubleWritable, Text, DoubleWritable> { // result 表示每个文件里面单词个数 DoubleWritable result = new DoubleWritable(); // Configuration conf = new Configuration(); // int total = conf.getInt("TOTALWORDS", 1); protected void reduce(Text key, Iterable<DoubleWritable> values, Reducer<Text, DoubleWritable, Text, DoubleWritable>.Context context) throws IOException, InterruptedException { double sum = 0L; for (DoubleWritable value : values) { sum += value.get(); } result.set(sum); context.write(key, result); } } }