【转】storm 开发系列一 第一个程序

原文： http://blog.csdn.net/csfreebird/article/details/49104777

-------------------------------------------------------------------------------------------------

本文将在本地开发环境创建一个storm程序，力求简单。

首先用mvn创建一个简单的工程hello_storm

[plain] view plain copy

print?

1. mvn archetype:generate -DgroupId=org.csfreebird -DartifactId=hello_storm -DarchetypeArtifactId=maven-archetype-quickstart -DinteractiveMode=false

编辑pom.xml，添加dependency

[html] view plain copy

print?

1. <project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
2. xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/maven-v4_0_0.xsd">
3. <modelVersion>4.0.0</modelVersion>
4. <groupId>org.csfreebird</groupId>
5. <artifactId>hello_storm</artifactId>
6. <version>0.9.5</version>
7. <packaging>jar</packaging>
8. <name>hello_storm</name>
9. <url>http://maven.apache.org</url>
10. <dependencies>
11. <dependency>
12. <groupId>org.apache.storm</groupId>
13. <artifactId>storm-core</artifactId>
14. <version>${project.version}</version>
15. <!-- keep storm out of the jar-with-dependencies -->
16. <scope>provided</scope>
17. </dependency>
18. </dependencies>
19. </project>

provided 表示storm-core的jar包只作为编译和测试时使用，在集群环境下运行时完全依赖集群环境的storm-core的jar包。

然后重命名App.Java为HelloTopology.java文件，开始编码。模仿之前的Example, 这里将所有的spout/bolt类都作为静态类定义，就放在HelloTopology.java文件。

功能如下

编写HelloTopology.java代码，spout代码来自于TestWordSpout,去掉了log的代码，改变了_引导的成员变量命名方法

[plain] view plain copy

print?

1. package org.csfreebird;
2. import backtype.storm.Config;
3. import backtype.storm.LocalCluster;
4. import backtype.storm.StormSubmitter;
5. import backtype.storm.task.OutputCollector;
6. import backtype.storm.task.TopologyContext;
7. import backtype.storm.testing.TestWordSpout;
8. import backtype.storm.topology.OutputFieldsDeclarer;
9. import backtype.storm.topology.TopologyBuilder;
10. import backtype.storm.topology.base.BaseRichBolt;
11. import backtype.storm.topology.base.BaseRichSpout;
12. import backtype.storm.tuple.Fields;
13. import backtype.storm.tuple.Tuple;
14. import backtype.storm.tuple.Values;
15. import backtype.storm.utils.Utils;
16. import backtype.storm.spout.SpoutOutputCollector;
17. import java.util.Map;
18. import java.util.TreeMap;
19. import java.util.Random;
20. public class HelloTopology {
21. public static class HelloSpout extends BaseRichSpout {
22. boolean isDistributed;
23. SpoutOutputCollector collector;
24. public HelloSpout() {
25. this(true);
26. }
27. public HelloSpout(boolean isDistributed) {
28. this.isDistributed = isDistributed;
29. }
30. public void open(Map conf, TopologyContext context, SpoutOutputCollector collector) {
31. this.collector = collector;
32. }
33. public void close() {
34. }
35. public void nextTuple() {
36. Utils.sleep(100);
37. final String[] words = new String[] {"china", "usa", "japan", "russia", "england"};
38. final Random rand = new Random();
39. final String word = words[rand.nextInt(words.length)];
40. this.collector.emit(new Values(word));
41. }
42. public void ack(Object msgId) {
43. }
44. public void fail(Object msgId) {
45. }
46. public void declareOutputFields(OutputFieldsDeclarer declarer) {
47. declarer.declare(new Fields("word"));
48. }
49. @Override
50. public Map<String, Object> getComponentConfiguration() {
51. if(!this.isDistributed) {
52. Map<String, Object> ret = new TreeMap<String, Object>();
53. ret.put(Config.TOPOLOGY_MAX_TASK_PARALLELISM, 1);
54. return ret;
55. } else {
56. return null;
57. }
58. }
59. }
60. public static class HelloBolt extends BaseRichBolt {
61. OutputCollector collector;
62. @Override
63. public void prepare(Map conf, TopologyContext context, OutputCollector collector) {
64. this.collector = collector;
65. }
66. @Override
67. public void execute(Tuple tuple) {
68. this.collector.emit(tuple, new Values("hello," + tuple.getString(0)));
69. this.collector.ack(tuple);
70. }
71. @Override
72. public void declareOutputFields(OutputFieldsDeclarer declarer) {
73. declarer.declare(new Fields("word"));
74. }
75. }
76. public static void main(String[] args) throws Exception {
77. TopologyBuilder builder = new TopologyBuilder();
78. builder.setSpout("a", new HelloSpout(), 10);
79. builder.setBolt("b", new HelloBolt(), 5).shuffleGrouping("a");
80. Config conf = new Config();
81. conf.setDebug(true);
82. if (args != null && args.length > 0) {
83. conf.setNumWorkers(3);
84. StormSubmitter.submitTopologyWithProgressBar(args[0], conf, builder.createTopology());
85. } else {
86. String test_id = "hello_test";
87. LocalCluster cluster = new LocalCluster();
88. cluster.submitTopology(test_id, conf, builder.createTopology());
89. Utils.sleep(10000);
90. cluster.killTopology(test_id);
91. cluster.shutdown();
92. }
93. }
94. }

编译成功

[plain] view plain copy

print?

1. mvn clean compile

为了能够在本地模式运行，需要在pom.xml中添加如下：

[html] view plain copy

print?

1. <build>
2. <plugins>
3. <plugin>
4. <groupId>org.codehaus.mojo</groupId>
5. <artifactId>exec-maven-plugin</artifactId>
6. <version>1.2.1</version>
7. <executions>
8. <execution>
9. <goals>
10. <goal>exec</goal>
11. </goals>
12. </execution>
13. </executions>
14. <configuration>
15. <executable>java</executable>
16. <includeProjectDependencies>true</includeProjectDependencies>
17. <includePluginDependencies>false</includePluginDependencies>
18. <classpathScope>compile</classpathScope>
19. <mainClass>${storm.topology}</mainClass>
20. </configuration>
21. </plugin>
22. </plugins>
23. </build>

然后运行命令

[plain] view plain copy

print?

1. mvn compile exec:java -Dstorm.topology=org.csfreebird.HelloTopology