基于文件系统（及MySQL）使用Java实现MapReduce

实现这个代码的原因是：

• 我会MapReduce，但是之前都是在AWS EMR上，自己搭过伪分布式的，但是感觉运维起来比较困难；
• 我就MySQL会一点（本来想用mongoDB的但是不太会啊）
• 数据量不是很大，至少对我来说。
• 希望不要出很么问题，这方面文件系统还是可以信任的。

设计思路如下：

• init阶段：将所需的文件添加到一个列表文件input_file_list.txt中。
• Map阶段：读取input_file_list.txt中的每一个文件的每一行，并将其映射成一个key-value对。
  
  考虑到key可能包含特殊字符，所以这里使用MySQL存储一个id到key的对应关系的数据。
• Reduce阶段：针对每一个key，读取对应的文件，最终生成一个name-value列表，该name-value列表对应一个json对象，如：{ "name": "zifeiy", "age": 88 }，将所有的json对象存储到一个结果文件reduceResult.txt中。
• 处理结果阶段，将reduceResult.txt文件进行解析，最终生成结果的CSV文件或者Excel文件。

主要代码：

package com.zifeiy.snowflake.tools.mapreduce.v1;

import java.io.BufferedReader;
import java.io.File;
import java.io.FileInputStream;
import java.io.FileOutputStream;
import java.io.IOException;
import java.io.InputStreamReader;
import java.io.OutputStreamWriter;
import java.sql.Connection;
import java.sql.DriverManager;
import java.sql.PreparedStatement;
import java.sql.ResultSet;
import java.sql.Statement;
import java.util.ArrayList;
import java.util.HashMap;
import java.util.List;
import java.util.Map;

import com.google.gson.Gson;
import com.zifeiy.snowflake.assist.CsvOneLineParser;
import com.zifeiy.snowflake.assist.FileHelper;

import jxl.Workbook;
import jxl.write.Label;
import jxl.write.WritableSheet;
import jxl.write.WritableWorkbook;

public abstract class MapReduceBaseVersion1 {

	private static final String APPENDED_DB_INFO = "?useUnicode=true&characterEncoding=UTF8"
										            + "&rewriteBatchedStatements=true"
										            + "&useLegacyDatetimeCode=false"
										            + "&serverTimezone=Asia/Shanghai"
										            + "&useSSL=false";
	private static final String classname		= "com.mysql.cj.jdbc.Driver";
	private static final String url				= "jdbc:mysql://localhost:3306/snowflake" + APPENDED_DB_INFO;
	private static final String username			= "root";
	private static final String password    = "password";

	public static final String taskRootPath = "D:\\snowflake\\task";

	private Connection connection = null;
	private File inputListFile = null;
	private File reduceResultFile = null;
	private File resultFile = null;
	private int taskId;

	public void addInputPath(File file) throws IOException {
		FileHelper.appendFile(inputListFile, file.getAbsolutePath() + "\r\n");
	}

	public void setKeyValuePair(String key, String value) throws Exception {
		int id = -1;
		Statement statement = connection.createStatement();
		ResultSet resultSet = statement.executeQuery(String.format("select id from tmp" + taskId + " where kname='%s'", key.replaceAll("'", "''")));
		if (resultSet.next()) {
			id = resultSet.getInt(1);
		}
		else {
			statement.execute(String.format("insert into tmp" + taskId + " (kname) values ('%s')", key.replaceAll("'", key.replaceAll("'", "''"))));
			resultSet = statement.executeQuery(String.format("select id from tmp" + taskId + " where kname='%s'", key.replaceAll("'", "''")));
			if (resultSet.next()) {
				id = resultSet.getInt(1);
			}
		}
        if (id == -1) throw new Exception("set key value pair failed: key = " + key + ", value = " + value);
        File tmpFile = new File(taskRootPath + File.separator + taskId + File.separator + "tmp" + File.separator + id + ".txt");
        if (tmpFile.exists() == false) {
        	tmpFile.createNewFile();
        }
        FileHelper.appendFile(tmpFile, value + "\r\n");
	}

	public void addParamList(List<Map<String, String>> paramList) throws Exception {
		String content = "";
		Gson gson = new Gson();
		for (Map<String, String> params : paramList) {
			String jsonString = gson.toJson(params);
			content += jsonString + "\r\n";
		}
		FileHelper.appendFile(reduceResultFile, content);
	}

	public void generateFile(String[] columns, String[] nameColumns) throws Exception {
		if (reduceResultFile == null || reduceResultFile.exists() == false) {
			throw new Exception("[mapreduce.v1] in generateFile function: reduceResultFile do not exist!");
		}
//		if (false) {	// test
		if (reduceResultFile.length() > 1 * 1024 * 1024) {	// 如果文件大小超过1MB，导出成csv
			resultFile = new File(taskRootPath + File.separator + taskId + File.separator + "result.csv");

			Gson gson = new Gson();

			BufferedReader br = new BufferedReader(new InputStreamReader(new FileInputStream(reduceResultFile), "UTF-8"));
			FileOutputStream fos = new FileOutputStream(resultFile);
	        OutputStreamWriter osw = new OutputStreamWriter(fos, "UTF-8");

	        String content = "";
	        for (int i = 0; i < nameColumns.length; i ++) {
	        	if (i > 0)
	        		content += ",";
	        	content += '"' + nameColumns[i] + '"';
	        }
	        osw.write(content + "\r\n");

	        String line = null;
	        while ((line = br.readLine()) != null) {
	        	content = "";
	        	Map<String, String> map = gson.fromJson(line, Map.class);
	        	if (map == null) { throw new Exception("map is null by parsing line: " + line); }
	        	for (int i = 0; i < columns.length; i ++) {
	        		if (i > 0) content += ",";
	        		String c = columns[i];
	        		String v = map.get(c);
	        		if (v != null) {
	        			content += '"' + v + '"';
	        		}
	        	}
	        	osw.write(content + "\r\n");
	        }
	        br.close();
	        osw.write(content);
	        osw.flush();
	        osw.close();
		} else {	// 如果文件大小小于1MB，导出成Excel文件
			resultFile = new File(taskRootPath + File.separator + taskId + File.separator + "result.xls");

			WritableWorkbook workbook = Workbook.createWorkbook(resultFile);
			WritableSheet sheet = workbook.createSheet("sheet1", 0);

			BufferedReader br = new BufferedReader(new InputStreamReader(new FileInputStream(reduceResultFile), "UTF-8"));
	        String line = null;

	        for (int i = 0; i < nameColumns.length; i ++) {
	        	sheet.addCell(new Label(i, 0, nameColumns[i]));
	        }

	        int rowId = 1;
	        while ((line = br.readLine()) != null) {
	        	Gson gson = new Gson();
	            List<String> rowList = new ArrayList<String>();

	            Map<String, String> map = gson.fromJson(line, Map.class);
	        	if (map == null) { throw new Exception("map is null by parsing line: " + line); }
	        	for (int i = 0; i < columns.length; i ++) {
	        		String c = columns[i];
	        		String v = map.get(c);
	        		String innerContent = "";
	        		if (v != null) {
	        			innerContent = v;
	        		}
	        		sheet.addCell(new Label(i, rowId, innerContent));
	        	}
            	rowId ++;

	        }
	        br.close();

	        workbook.write();
	        workbook.close();

		}
	}

	public abstract void init() throws Exception;

	public abstract void map(String line) throws Exception;

	public abstract void reduce(String key, ReduceReader reduceReader) throws Exception;

	public abstract void generate() throws Exception;

	public String mapreduce() {
		try {
			Class.forName(classname);
			connection = DriverManager.getConnection(url, username, password);

			// generate taskId
			PreparedStatement preparedStatement = connection.prepareStatement("insert into task () values ()");
			preparedStatement.execute("insert into task () values ()", PreparedStatement.RETURN_GENERATED_KEYS);
            ResultSet resultSet = preparedStatement.getGeneratedKeys();
            if (resultSet.next()) {
            	taskId = resultSet.getInt(1);
            }
            else {
            	throw new Exception("[mapreduce.v1] Exception: can not generate taskId");
            }
            // generated task file path
            String taskPath = taskRootPath + File.separator + taskId;
            File taskPathDir = new File(taskPath);
            if (taskPathDir.exists() == true) {
            	throw new Exception("[mapreduce.v1] Exception: task directory already exists");
            }
            taskPathDir.mkdirs();
            String tmpDirPath = taskPath + File.separator + "tmp";
            File tmpDir = new File(tmpDirPath);
            tmpDir.mkdirs();
            this.inputListFile = new File(taskPath + File.separator + "input_file_list.txt");
            inputListFile.createNewFile();
            // period. 1: init()
            // during init period, we will use addInputPath function to add all the input files we need
            init();

            // begin to read each line of each file
            // peroid. 2: map(line)

            // db prepare
            Statement statement = connection.createStatement();
            statement.execute("create temporary table tmp" + taskId + " ( id int not null auto_increment primary key, kname varchar(200) )");

            // file content prepare

            BufferedReader br = new BufferedReader(new InputStreamReader(new FileInputStream(inputListFile), "UTF-8"));
            String inputFilename = null;
            while ((inputFilename = br.readLine()) != null) {
                File inputFile = new File(inputFilename);
                if (inputFile.exists() == false) {
                	throw new Exception("[mapreduce.v1] Exception: input file " + inputFilename + " do not exists!");
                }
                BufferedReader br2 = new BufferedReader(new InputStreamReader(new FileInputStream(inputFile), "GBK"));
                String line = null;
                while ((line = br2.readLine()) != null) {
                	map(line);
                }
            }
            br.close();

            // period. 3: reduce(key, valueList)
            reduceResultFile = new File(taskPath + File.separator + "reduce.txt");
            if (reduceResultFile.exists() == true) {
            	throw new Exception("[mapreduce.v1] reduce file already exists!");
            }
            reduceResultFile.createNewFile();

	        resultSet = statement.executeQuery("select * from tmp" + taskId);
	        while (resultSet.next()) {
	        	int id = resultSet.getInt(1);
	        	String key = resultSet.getString(2);
	        	File reduceFile = new File(tmpDirPath + File.separator + id + ".txt");
	        	if (reduceFile.exists() == false) {
	        		throw new Exception("[mapreduce.v1] Exception: reduce file " + reduceFile.getName() + " not exists!");
	        	}
	        	ReduceReader reduceReader = new ReduceReader(reduceFile);
	        	reduce(key, reduceReader);
	        }

	        // period. 4: generate
	        // generate the result file
	        generate();

	        connection.close();
		} catch (Exception e) {
			e.printStackTrace();
		}

		if (resultFile == null) return null;
		else return resultFile.getAbsolutePath();

	}

	// main for test
	public static void main(String[] args) {
		MapReduceBaseVersion1 mapReduceBaseVersion1 = new MapReduceBaseVersion1() {

			@Override
			public void reduce(String key, ReduceReader reduceReader) throws Exception {
				// TODO Auto-generated method stub
				List<Map<String, String>> paramList = new ArrayList<Map<String,String>>();

				String line;
				while ( (line = reduceReader.next()) != null ) {
					List<String> rowList = CsvOneLineParser.parseLine(line);
					Map<String, String> tmpMap = new HashMap<String, String>();
					int idx = 0;
					for (String s : rowList) {
						idx ++;
						tmpMap.put("" + idx, s);
					}
					paramList.add(tmpMap);
				}
				addParamList(paramList);
			}

			@Override
			public void map(String line) throws Exception {
				// TODO Auto-generated method stub
				setKeyValuePair(line.substring(1, 3), line);
			}

			@Override
			public void init() throws Exception {
				// TODO Auto-generated method stub
				addInputPath(new File("D:\\test\\test.del"));
			}

			@Override
			public void generate() throws Exception {
				// TODO Auto-generated method stub
				generateFile(new String[] { "1", "2", "3", "4", "5", "6" }, new String[] { "一", "二", "三", "四", "五", "六" });
			}
		};
		System.out.println(mapReduceBaseVersion1.mapreduce());
	}
}