Examples with WordCount org.apache.gora.examples.mapreduce.WordCount used on opensource projects

Example 1 with WordCount

use of org.apache.gora.examples.mapreduce.WordCount in project gora by apache.

the class MapReduceTestUtils method testWordCount.

public static void testWordCount(Configuration conf, DataStore<String, WebPage> inStore, DataStore<String, TokenDatum> outStore) throws Exception {
    //Datastore now has to be a Hadoop based datastore
    ((DataStoreBase<String, WebPage>) inStore).setConf(conf);
    ((DataStoreBase<String, TokenDatum>) outStore).setConf(conf);
    //create input
    WebPageDataCreator.createWebPageData(inStore);
    //run the job
    WordCount wordCount = new WordCount(conf);
    wordCount.wordCount(inStore, outStore);
    //assert results
    HashMap<String, Integer> actualCounts = new HashMap<>();
    for (String content : WebPageDataCreator.CONTENTS) {
        if (content != null) {
            for (String token : content.split(" ")) {
                Integer count = actualCounts.get(token);
                if (count == null)
                    count = 0;
                actualCounts.put(token, ++count);
            }
        }
    }
    for (Map.Entry<String, Integer> entry : actualCounts.entrySet()) {
        assertTokenCount(outStore, entry.getKey(), entry.getValue());
    }
}