Differences
This shows you the differences between two versions of the page.
Both sides previous revision Previous revision Next revision | Previous revision Next revision Both sides next revision | ||
courses:mapreduce-tutorial:step-29 [2012/01/29 17:42] straka |
courses:mapreduce-tutorial:step-29 [2012/02/05 18:49] straka |
||
---|---|---|---|
Line 1: | Line 1: | ||
- | ====== MapReduce Tutorial : Custom | + | ====== MapReduce Tutorial : Custom |
- | Every custom format reading keys of type '' | + | ====== Fast sorting comparator ====== |
- | ===== FileAsPathInputFormat ===== | + | The keys are sorted before processed by a reducer, using a |
+ | [[http:// | ||
- | We start by creating '' | + | <code java> |
+ | public static class IntPair implements WritableComparable< | ||
+ | private int first = 0; | ||
+ | private int second = 0; | ||
- | When implementing new input format, we must | + | public void set(int left, int right) { first = left; second = right; } |
- | | + | |
- | | + | |
- | When | + | |
- | <code java> | + | |
- | public static class FileAsPathInputFormat extends FileInputFormat< | + | |
- | public static class FileAsPathRecordReader extends RecordReader< | + | |
- | private Path file; | + | |
- | long start, length; | + | |
- | private Text key, value; | + | |
- | + | ||
- | | + | |
- | | + | |
- | file = split.getPath(); | + | |
- | start = split.getStart(); | + | |
- | length = split.getLength(); | + | |
- | key = null; | + | |
- | value = null; | + | |
- | | + | |
- | public boolean nextKeyValue() throws IOException { | + | |
- | if (key != null) return false; | + | |
- | + | ||
- | key = new Text(file.toString()); | + | |
- | value = new Text(String.format(" | + | |
- | + | ||
- | return true; | + | |
- | } | + | |
- | + | ||
- | public Text getCurrentKey() { return key; } | + | |
- | public Text getCurrentValue() { return value; } | + | |
- | public float getProgress() { return (key == null) ? 0 : 1; } | + | |
- | public synchronized void close() throws IOException {} | + | |
} | } | ||
- | + | | |
- | | + | |
- | | + | out.writeInt(second); |
- | } | + | } |
- | + | ||
- | | + | |
- | | + | |
- | return | + | |
} | } | ||
} | } | ||
</ | </ | ||
- | ===== WholeFileInputFormat ===== | + | If we would like in a Hadoop job to sort the '' |
- | We start by creating '' | ||
- | The main functionality lays in '' | ||
- | <code java> | + | ====== Grouping comparator ====== |
- | public class WholeFileInputFormat extends FileInputFormat< | + | |
- | // Helper class, which does the actual work -- reads the (path, content) input pair. | + | |
- | public static class WholeFileRecordReader extends RecordReader< | + | |
- | private Path file; | + | |
- | int length; | + | |
- | private boolean value_read; | + | |
- | private Text key; | + | |
- | private BytesWritable value; | + | |
- | DataInputStream in; | + | |
- | public void initialize(InputSplit genericSplit, TaskAttemptContext context) throws IOException { | + | In a reduce, it is guaranteed that keys are processed in ascending order. Sometimes it would be useful if the //values associated with one key// could also be processed in ascending order. |
- | FileSplit split = (FileSplit) genericSplit; | + | |
- | file = split.getPath(); | + | |
- | length = (int) split.getLength(); | + | |
- | | + | |
- | value = null; | + | |
- | value_read = false; | + | |
- | FileSystem fs = file.getFileSystem(context.getConfiguration()); | + | ---- |
- | in = fs.open(split.getPath()); | + | |
- | CompressionCodecFactory compressionCodecs = new CompressionCodecFactory(context.getConfiguration()); | + | < |
- | | + | <table style=" |
- | if (codec != null) | + | <tr> |
- | | + | <td style=" |
- | } | + | <td style=" |
- | + | <td style=" | |
- | public boolean nextKeyValue() throws IOException { | + | </tr> |
- | if (value_read) return false; | + | </table> |
- | + | </html> | |
- | byte[] data = new byte[length]; | + | |
- | in.readFully(data); | + | |
- | + | ||
- | key = new Text(file.toString()); | + | |
- | | + | |
- | value_read = true; | + | |
- | + | ||
- | return true; | + | |
- | } | + | |
- | + | ||
- | public Text getCurrentKey() { return key; } | + | |
- | public BytesWritable getCurrentValue() { return value; } | + | |
- | public float getProgress() { return value_read ? 0 : 1; } | + | |
- | | + | |
- | } | + | |
- | + | ||
- | | + | |
- | public RecordReader<Text, BytesWritable> createRecordReader(InputSplit split, TaskAttemptContext context) { | + | |
- | return new WholeFileRecordReader(); | + | |
- | } | + | |
- | + | ||
- | // Do not allow splitting. | + | |
- | protected boolean isSplittable(JobContext context, Path filename) { | + | |
- | return false; | + | |
- | } | + | |
- | } | + | |
- | + | ||
- | </code> | + | |