Use Easy Batch to do some data mining on paris trees data #EasyBatch

Tree.java

package org.easybatch.flatfile;

public class Tree {

    private String adresse;

    private float hauteur;

    public Tree() {
    }

    public String getAdresse() {
        return adresse;
    }

    public void setAdresse(String adresse) {
        this.adresse = adresse;
    }

    public float getHauteur() {
        return hauteur;
    }

    public void setHauteur(float hauteur) {
        this.hauteur = hauteur;
    }
}

TreeParisTest.java

package org.easybatch.flatfile;

import org.easybatch.core.api.ComputationalRecordProcessor;
import org.easybatch.core.api.Report;
import org.easybatch.core.filter.HeaderRecordFilter;
import org.easybatch.core.impl.Engine;
import org.easybatch.core.record.StringRecord;
import org.easybatch.flatfile.dsv.DelimitedRecordMapper;
import org.junit.Test;

import java.io.File;

import static org.easybatch.core.impl.EngineBuilder.aNewEngine;

public class TreeParisTest {

    //data source : http://opendata.paris.fr/explore/dataset/arbresalignementparis2010/download/?format=csv

    @Test
    public void testWithRawData() throws Exception {

        File dataSource = new File(this.getClass().getResource("/arbresalignementparis2010.csv").toURI());

        Engine engine = aNewEngine()
                .reader(new FlatFileRecordReader(dataSource))
                .filter(new HeaderRecordFilter())
                .processor(new ComputationalRecordProcessor<StringRecord, StringRecord, Integer>() {

                    Integer count = 0;

                    @Override
                    public Integer getComputationResult() {
                        return count;
                    }

                    @Override
                    public StringRecord processRecord(StringRecord record) throws Exception {
                        if (Float.parseFloat(record.getPayload().split(";")[3]) > 0) {
                            count++;
                        }
                        return record;
                    }
                })
                .build();

        Report report = engine.call();
        Integer total = (Integer) report.getBatchResult();

        System.out.println("total = " + total + " calculated in " + report.getFormattedBatchDuration());

    }

    @Test
    public void testWithMappingDataToDomainObject() throws Exception {

        File dataSource = new File(this.getClass().getResource("/arbresalignementparis2010.csv").toURI());

        DelimitedRecordMapper<Tree> recordMapper = new DelimitedRecordMapper<Tree>(Tree.class, new Integer[]{2, 3}, new String[]{"adresse", "hauteur"});
        recordMapper.setDelimiter(";");

        Engine engine = aNewEngine()
                .reader(new FlatFileRecordReader(dataSource))
                .filter(new HeaderRecordFilter())
                .mapper(recordMapper)
                .processor(new ComputationalRecordProcessor<Tree, Tree, Integer>() {

                    Integer count = 0;

                    @Override
                    public Integer getComputationResult() {
                        return count;
                    }

                    @Override
                    public Tree processRecord(Tree tree) throws Exception {
                        if (tree.getHauteur() > 0) {
                            count++;
                        }
                        return tree;
                    }
                })
                .build();

        Report report = engine.call();
        Integer total = (Integer) report.getBatchResult();

        System.out.println("total = " + total + " calculated in " + report.getFormattedBatchDuration());

    }

}