https://beam.apache.org/get-started/wordcount-example/
来自官网的：
The WordCount examples demonstrate how to set up a processing pipeline that can read text, tokenize the text lines into individual words, and perform a frequency count on each of those words. The Beam SDKs contain a series of these four successively more detailed WordCount examples that build on each other. The input text for all the examples is a set of Shakespeare’s texts.
Each WordCount example introduces different concepts in the Beam programming model. Begin by understanding Minimal WordCount, the simplest of the examples. Once you feel comfortable with the basic principles in building a pipeline, continue on to learn more concepts in the other examples.
- Minimal WordCount demonstrates the basic principles involved in building a pipeline.
- WordCount introduces some of the more common best practices in creating re-usable and maintainable pipelines.
- Debugging WordCount introduces logging and debugging practices.
- Windowed WordCount demonstrates how you can use Beam’s programming model to handle both bounded and unbounded datasets.
我这里仅以Minimal WordCount为例。
首先说明一下，为了简单起见，我直接在代码中显式配置指定PipelineRunner，示例代码片段如下所示：
PipelineOptions options = PipelineOptionsFactory.create(); options.setRunner(DirectRunner.class);
如果要部署到服务器上，可以通过命令行的方式指定PipelineRunner，比如要在Spark集群上运行，类似如下所示命令行：
spark-submit --class org.shirdrn.beam.examples.MinimalWordCountBasedSparkRunner 2017-01-18 --master spark://myserver:7077 target/my-beam-apps-0.0.1-SNAPSHOT-shaded.jar --runner=SparkRunner
下面，我们从几个典型的例子来看（基于Apache Beam软件包的examples有所改动），Apache Beam如何构建Pipeline并运行在指定的PipelineRunner上：
- WordCount（Count/Source/Sink）
我们根据Apache Beam的MinimalWordCount示例代码开始，看如何构建一个Pipeline，并最终执行它。 MinimalWordCount的实现，代码如下所示：
package org.shirdrn.beam.examples; import org.apache.beam.runners.direct.DirectRunner; import org.apache.beam.sdk.Pipeline; import org.apache.beam.sdk.io.TextIO; import org.apache.beam.sdk.options.PipelineOptions; import org.apache.beam.sdk.options.PipelineOptionsFactory; import org.apache.beam.sdk.transforms.Count; import org.apache.beam.sdk.transforms.DoFn; import org.apache.beam.sdk.transforms.MapElements; import org.apache.beam.sdk.transforms.ParDo; import org.apache.beam.sdk.transforms.SimpleFunction; import org.apache.beam.sdk.values.KV; public class MinimalWordCount { @SuppressWarnings("serial") public static void main(String[] args) { PipelineOptions options = PipelineOptionsFactory.create(); options.setRunner(DirectRunner.class); // 显式指定PipelineRunner：DirectRunner（Local模式） Pipeline pipeline = Pipeline.create(options); pipeline.apply(TextIO.Read.from("/tmp/dataset/apache_beam.txt")) // 读取本地文件，构建第一个PTransform .apply("ExtractWords", ParDo.of(new DoFn<String, String>() { // 对文件中每一行进行处理（实际上Split） @ProcessElement public void processElement(ProcessContext c) { for (String word : c.element().split("[\\s:\\,\\.\\-]+")) { if (!word.isEmpty()) { c.output(word); } } } })) .apply(Count.<String> perElement()) // 统计每一个Word的Count .apply("ConcatResultKVs", MapElements.via( // 拼接最后的格式化输出（Key为Word，Value为Count） new SimpleFunction<KV<String, Long>, String>() { @Override public String apply(KV<String, Long> input) { return input.getKey() + ": " + input.getValue(); } })) .apply(TextIO.Write.to("wordcount")); // 输出结果 pipeline.run().waitUntilFinish(); } }
Pipeline的具体含义，可以看上面代码的注释信息。下面，我们考虑以HDFS数据源作为Source，如何构建第一个PTransform，代码片段如下所示：
PCollection<KV<LongWritable, Text>> resultCollection = pipeline.apply(HDFSFileSource.readFrom( "hdfs://myserver:8020/data/ds/beam.txt", TextInputFormat.class, LongWritable.class, Text.class))
可以看到，返回的是具有键值分别为LongWritable、Text类型的KV对象集合，后续处理和上面处理逻辑类似。如果使用Maven构建Project，需要加上如下依赖（这里beam.version的值可以为最新Release版本0.4.0）：
<dependency>
<groupId>org.apache.beam</groupId>
<artifactId>beam-sdks-java-io-hdfs</artifactId>
<version>${beam.version}</version>
</dependency>
- 去重（Distinct）
去重也是对数据集比较常见的操作，使用Apache Beam来实现，示例代码如下所示：
package org.shirdrn.beam.examples; import org.apache.beam.runners.direct.DirectRunner; import org.apache.beam.sdk.Pipeline; import org.apache.beam.sdk.io.TextIO; import org.apache.beam.sdk.options.PipelineOptions; import org.apache.beam.sdk.options.PipelineOptionsFactory; import org.apache.beam.sdk.transforms.Distinct; public class DistinctExample { public static void main(String[] args) throws Exception { PipelineOptions options = PipelineOptionsFactory.create(); options.setRunner(DirectRunner.class); // 显式指定PipelineRunner：DirectRunner（Local模式） Pipeline pipeline = Pipeline.create(options); pipeline.apply(TextIO.Read.from("/tmp/dataset/MY_ID_FILE.txt")) .apply(Distinct.<String> create()) // 创建一个处理String类型的PTransform：Distinct .apply(TextIO.Write.to("deduped.txt")); // 输出结果 pipeline.run().waitUntilFinish(); } }
- 分组（GroupByKey）
对数据进行分组操作也非常普遍，我们拿一个最基础的PTransform实现GroupByKey来实现一个例子，代码如下所示：
package org.shirdrn.beam.examples; import org.apache.beam.runners.direct.DirectRunner; import org.apache.beam.runners.direct.repackaged.com.google.common.base.Joiner; import org.apache.beam.sdk.Pipeline; import org.apache.beam.sdk.io.TextIO; import org.apache.beam.sdk.options.PipelineOptions; import org.apache.beam.sdk.options.PipelineOptionsFactory; import org.apache.beam.sdk.transforms.DoFn; import org.apache.beam.sdk.transforms.GroupByKey; import org.apache.beam.sdk.transforms.MapElements; import org.apache.beam.sdk.transforms.ParDo; import org.apache.beam.sdk.transforms.SimpleFunction; import org.apache.beam.sdk.values.KV; public class GroupByKeyExample { @SuppressWarnings("serial") public static void main(String[] args) { PipelineOptions options = PipelineOptionsFactory.create(); options.setRunner(DirectRunner.class); // 显式指定PipelineRunner：DirectRunner（Local模式） Pipeline pipeline = Pipeline.create(options); pipeline.apply(TextIO.Read.from("/tmp/dataset/MY_INFO_FILE.txt")) .apply("ExtractFields", ParDo.of(new DoFn<String, KV<String, String>>() { @ProcessElement public void processElement(ProcessContext c) { // file format example: 35451605324179 3G CMCC String[] values = c.element().split("\t"); if(values.length == 3) { c.output(KV.of(values[1], values[0])); } } })) .apply("GroupByKey", GroupByKey.<String, String>create()) // 创建一个GroupByKey实例的PTransform .apply("ConcatResults", MapElements.via( new SimpleFunction<KV<String, Iterable<String>>, String>() { @Override public String apply(KV<String, Iterable<String>> input) { return new StringBuffer() .append(input.getKey()).append("\t") .append(Joiner.on(",").join(input.getValue())) .toString(); } })) .apply(TextIO.Write.to("grouppedResults")); pipeline.run().waitUntilFinish(); } }
使用DirectRunner运行，输出文件名称类似于grouppedResults-00000-of-00002、grouppedResults-00001-of-00002等等。
- 连接（Join）
最后，我们通过实现一个Join的例子，其中，用户的基本信息包含ID和名称，对应文件格式如下所示：
35451605324179 Jack 35236905298306 Jim 35236905519469 John 35237005022314 Linda
另一个文件是用户使用手机的部分信息，文件格式如下所示：
35451605324179 3G 中国移动 35236905298306 2G 中国电信 35236905519469 4G 中国移动
我们希望通过Join操作后，能够知道用户使用的什么网络（用户名+网络），使用Apache Beam实现，具体实现代码如下所示：
package org.shirdrn.beam.examples; import org.apache.beam.runners.direct.DirectRunner; import org.apache.beam.sdk.Pipeline; import org.apache.beam.sdk.io.TextIO; import org.apache.beam.sdk.options.PipelineOptions; import org.apache.beam.sdk.options.PipelineOptionsFactory; import org.apache.beam.sdk.transforms.DoFn; import org.apache.beam.sdk.transforms.MapElements; import org.apache.beam.sdk.transforms.ParDo; import org.apache.beam.sdk.transforms.SimpleFunction; import org.apache.beam.sdk.transforms.join.CoGbkResult; import org.apache.beam.sdk.transforms.join.CoGroupByKey; import org.apache.beam.sdk.transforms.join.KeyedPCollectionTuple; import org.apache.beam.sdk.values.KV; import org.apache.beam.sdk.values.PCollection; import org.apache.beam.sdk.values.TupleTag; public class JoinExample { @SuppressWarnings("serial") public static void main(String[] args) { PipelineOptions options = PipelineOptionsFactory.create(); options.setRunner(DirectRunner.class); // 显式指定PipelineRunner：DirectRunner（Local模式） Pipeline pipeline = Pipeline.create(options); // create ID info collection final PCollection<KV<String, String>> idInfoCollection = pipeline .apply(TextIO.Read.from("/tmp/dataset/MY_ID_INFO_FILE.txt")) .apply("CreateUserIdInfoPairs", MapElements.via( new SimpleFunction<String, KV<String, String>>() { @Override public KV<String, String> apply(String input) { // line format example: 35451605324179 Jack String[] values = input.split("\t"); return KV.of(values[0], values[1]); } })); // create operation collection final PCollection<KV<String, String>> opCollection = pipeline .apply(TextIO.Read.from("/tmp/dataset/MY_ID_OP_INFO_FILE.txt")) .apply("CreateIdOperationPairs", MapElements.via( new SimpleFunction<String, KV<String, String>>() { @Override public KV<String, String> apply(String input) { // line format example: 35237005342309 3G CMCC String[] values = input.split("\t"); return KV.of(values[0], values[1]); } })); final TupleTag<String> idInfoTag = new TupleTag<String>(); final TupleTag<String> opInfoTag = new TupleTag<String>(); final PCollection<KV<String, CoGbkResult>> cogrouppedCollection = KeyedPCollectionTuple .of(idInfoTag, idInfoCollection) .and(opInfoTag, opCollection) .apply(CoGroupByKey.<String>create()); final PCollection<KV<String, String>> finalResultCollection = cogrouppedCollection .apply("CreateJoinedIdInfoPairs", ParDo.of(new DoFn<KV<String, CoGbkResult>, KV<String, String>>() { @ProcessElement public void processElement(ProcessContext c) { KV<String, CoGbkResult> e = c.element(); String id = e.getKey(); String name = e.getValue().getOnly(idInfoTag); for (String opInfo : c.element().getValue().getAll(opInfoTag)) { // Generate a string that combines information from both collection values c.output(KV.of(id, "\t" + name + "\t" + opInfo)); } } })); PCollection<String> formattedResults = finalResultCollection .apply("FormatFinalResults", ParDo.of(new DoFn<KV<String, String>, String>() { @ProcessElement public void processElement(ProcessContext c) { c.output(c.element().getKey() + "\t" + c.element().getValue()); } })); formattedResults.apply(TextIO.Write.to("joinedResults")); pipeline.run().waitUntilFinish(); }
