@Override
public void map(
PactNull pactNull, PactString line, Collector<YearMonthKey, PactInteger> collector) {
String[] fields = line.toString().split("\t");
short year = Short.parseShort(fields[0]);
short month = Short.parseShort(fields[1]);
int temperature = Integer.parseInt(fields[2]);
double quality = Double.parseDouble(fields[3]);
if (quality >= MINIMUM_QUALITY) {
collector.collect(new YearMonthKey(year, month), new PactInteger(temperature));
}
}
@Override
public void map(PactNull key, PactString value, Collector<PactString, PactString> out) {
String[] items = value.toString().split(Delimiter);
if (items.length == 2) // it's a docs-tuple
{
boolean ContainAllKeywords = true;
for (String keyword : keywords) {
if (!items[1].contains(keyword)) // check whether content contains keyword
{
ContainAllKeywords = false;
}
}
if (ContainAllKeywords) {
out.collect(new PactString(items[docs_KeyPosition]), value);
}
} else if (items.length == 3) // it's a ranks-tuple
{
if (Integer.parseInt(items[0]) > rank) // check whether the rank is bigger enough
{
out.collect(new PactString(items[ranks_KeyPosition]), value);
}
}
};
@Override
public void reduce(
PactString key, Iterator<PactString> tuples, Collector<PactNull, PactString> out) {
boolean rank_tuple_found = false;
boolean doc_tuple_found = false;
PactString rank_tuple = new PactString();
while (tuples.hasNext()) {
PactString tuple = tuples.next();
if (tuple.toString().split(Delimiter).length == 3) {
rank_tuple_found = true;
rank_tuple.setValue(tuple.getValue());
} else if (tuple.toString().split(Delimiter).length == 2) {
doc_tuple_found = true;
}
}
if (rank_tuple_found && doc_tuple_found) {
// emit the rank tuple
out.collect(PactNull.getInstance(), rank_tuple);
}
}