解析GenericOptionsParser

hadoop源代码分析(4)-org.apache.hadoop.util包-GenericOptionsParser类【原创】

 

一  准备

  hadoop版本:1.0.3,GenericOptionsParser所在的包:org.apache.hadoop.util

  学习方法:理解GenericOptionsParser这个解析器的使用地方,从构造函数入手,理解GenericOptionsParser整个类的使用情况。

  时间:2013-02-21

二  GenericOptionsParser功能描述

   GenericOptionsParser是hadoop框架中解析命令行参数的基本类。它能够辨别一些标准的命令行参数,能够使应用程序轻易地指定namenode,jobtracker,以及其他额外的配置资源。一般它使用方法如下:

 

复制代码
 1     out.println("Generic options supported are");
 2     out.println("-conf <configuration file>     specify an application configuration file");
 3     out.println("-D <property=value>            use value for given property");
 4     out.println("-fs <local|namenode:port>      specify a namenode");
 5     out.println("-jt <local|jobtracker:port>    specify a job tracker");
 6     out.println("-files <comma separated list of files>    " +
 7       "specify comma separated files to be copied to the map reduce cluster");
 8     out.println("-libjars <comma separated list of jars>    " +
 9       "specify comma separated jar files to include in the classpath.");
10     out.println("-archives <comma separated list of archives>    " +
11                 "specify comma separated archives to be unarchived" +
12                 " on the compute machines.\n");
13     out.println("The general command line syntax is");
14     out.println("bin/hadoop command [genericOptions] [commandOptions]\n");
复制代码

 

  使用示例:

 

复制代码
 1  * $ bin/hadoop dfs -fs darwin:8020 -ls /data
 2  * list /data directory in dfs with namenode darwin:8020
 3  *
 4  * $ bin/hadoop dfs -D fs.default.name=darwin:8020 -ls /data
 5  * list /data directory in dfs with namenode darwin:8020
 6  *     
 7  * $ bin/hadoop dfs -conf hadoop-site.xml -ls /data
 8  * list /data directory in dfs with conf specified in hadoop-site.xml
 9  *     
10  * $ bin/hadoop job -D mapred.job.tracker=darwin:50020 -submit job.xml
11  * submit a job to job tracker darwin:50020
12  *     
13  * $ bin/hadoop job -jt darwin:50020 -submit job.xml
14  * submit a job to job tracker darwin:50020
15  *     
16  * $ bin/hadoop job -jt local -submit job.xml
17  * submit a job to local runner
18  *
19  * $ bin/hadoop jar -libjars testlib.jar
20  * -archives test.tgz -files file.txt inputjar args
21  * job submission with libjars, files and archives
复制代码

 

四  GenericOptionsParser主要方法、属性分析

  GenericOptionsParser这个类是从构造函数开始的,它有多个构造函数,真正的处理是在parseGeneralOptions(options, conf, args)这个函数中。

复制代码
 1   /** 
 2    * 构造GenericOptionsParser来解析给定的选项以及基本的hadoop选项
 3    * 命令行对象可以通过getCommandLine()函数获得
 4    * @param conf the configuration to modify  
 5    * @param options options built by the caller 
 6    * @param args User-specified arguments
 7    * @throws IOException 
 8    */
 9   public GenericOptionsParser(Configuration conf,
10       Options options, String[] args) throws IOException {
11     parseGeneralOptions(options, conf, args);
12     this.conf = conf;
13   }
复制代码

  parseGeneralOptions(options, conf, args)这个函数解析用户指定的参数,获取基本选项以及根据需要修改配置。它首先指定每个通用选项的属性,然后解析选项,参数,把它转化为命令行对象(CommandLine),紧接着把设定好的命令行参数写入系统配置,源代码如下:

复制代码
 1   /**
 2    * 解析用户指定的参数,获取基本选项以及根据需要修改配置
 3    * Parse the user-specified options, get the generic options, and modify
 4    * configuration accordingly
 5    * @param conf Configuration to be modified
 6    * @param args User-specified arguments
 7    * @return Command-specific arguments
 8    */
 9   private String[] parseGeneralOptions(Options opts, Configuration conf, 
10       String[] args) throws IOException {
11       // 指定每个通用选项的属性
12     opts = buildGeneralOptions(opts);
13     CommandLineParser parser = new GnuParser();
14     try {
15         // 解析选项,参数,获取命令行
16       commandLine = parser.parse(opts, args, true);
17       // 根据用户指定的参数(commandLine)修改系统的配置
18       processGeneralOptions(conf, commandLine);
19       return commandLine.getArgs();
20     } catch(ParseException e) {
21       LOG.warn("options parsing failed: "+e.getMessage());
22 
23       HelpFormatter formatter = new HelpFormatter();
24       formatter.printHelp("general options are: ", opts);
25     }
26     return args;
27   }
复制代码

  processGeneralOptions函数作用是修改配置,利用CommandLine对象的相关方法,这个类包含处理选项以及选项描述,选项值的方法,源代码如下:

复制代码
 1  /**
 2    * 根据用户指定的参数修改配置
 3    * Modify configuration according user-specified generic options
 4    * @param conf Configuration to be modified
 5    * @param line User-specified generic options
 6    */
 7   private void processGeneralOptions(Configuration conf,
 8       CommandLine line) throws IOException {
 9     if (line.hasOption("fs")) {
10         // 设置NAMENODE的ip
11       FileSystem.setDefaultUri(conf, line.getOptionValue("fs"));
12     }
13 
14     if (line.hasOption("jt")) {
15       conf.set("mapred.job.tracker", line.getOptionValue("jt"));
16     }
17     if (line.hasOption("conf")) {
18       String[] values = line.getOptionValues("conf");
19       for(String value : values) {
20         // 新增配置文件,除非是final属性,不然新配置文件会覆盖旧的配置文件
21         conf.addResource(new Path(value));
22       }
23     }
24     if (line.hasOption("libjars")) {
25       conf.set("tmpjars", 
26                validateFiles(line.getOptionValue("libjars"), conf));
27       //setting libjars in client classpath
28       URL[] libjars = getLibJars(conf);
29       if(libjars!=null && libjars.length>0) {
30         conf.setClassLoader(new URLClassLoader(libjars, conf.getClassLoader()));
31         Thread.currentThread().setContextClassLoader(
32             new URLClassLoader(libjars, 
33                 Thread.currentThread().getContextClassLoader()));
34       }
35     }
36     if (line.hasOption("files")) {
37       conf.set("tmpfiles", 
38                validateFiles(line.getOptionValue("files"), conf));
39     }
40     if (line.hasOption("archives")) {
41       conf.set("tmparchives", 
42                 validateFiles(line.getOptionValue("archives"), conf));
43     }
44     if (line.hasOption('D')) {
45       String[] property = line.getOptionValues('D');
46       for(String prop : property) {
47         String[] keyval = prop.split("=", 2);
48         if (keyval.length == 2) {
49           conf.set(keyval[0], keyval[1]);
50         }
51       }
52     }
53     conf.setBoolean("mapred.used.genericoptionsparser", true);
54     
55     // tokensFile
56     if(line.hasOption("tokenCacheFile")) {
57       String fileName = line.getOptionValue("tokenCacheFile");
58       // check if the local file exists
59       try 
60       {
61         FileSystem localFs = FileSystem.getLocal(conf);
62         Path p = new Path(fileName);
63         if (!localFs.exists(p)) {
64           throw new FileNotFoundException("File "+fileName+" does not exist.");
65         }
66 
67         LOG.debug("setting conf tokensFile: " + fileName);
68         conf.set("mapreduce.job.credentials.json", 
69                  localFs.makeQualified(p).toString());
70       } catch (IOException e) {
71         throw new RuntimeException(e);
72       }
73     }
74   }
复制代码

 

五  GenericOptionsParser相关类、接口简述

  跟这个类相关的类是:Options,Option,ComandLine,FileSystem。

六  结语

posted @ 2016-09-20 16:57  Evil_XJZ  阅读(1173)  评论(0编辑  收藏  举报