/**
* Filters for records of the visits relation where the year of visit is equal to a specified
* value. The URL of all visit records passing the filter is emitted.
*
* <p>Output Format: 0: URL
*/
@Override
public void map(PactRecord record, Collector<PactRecord> out) throws Exception {
// Parse date string with the format YYYY-MM-DD and extract the year
String dateString = record.getField(1, PactString.class).getValue();
int year = Integer.parseInt(dateString.substring(0, 4));
if (year == YEARFILTER) {
record.setNull(1);
out.collect(record);
}
}
/**
* Filters for documents that contain all of the given keywords and projects the records on the
* URL field.
*
* <p>Output Format: 0: URL
*/
@Override
public void map(PactRecord record, Collector<PactRecord> out) throws Exception {
// FILTER
// Only collect the document if all keywords are contained
String docText = record.getField(1, PactString.class).toString();
boolean allContained = true;
for (String kw : KEYWORDS) {
if (!docText.contains(kw)) {
allContained = false;
break;
}
}
if (allContained) {
record.setNull(1);
out.collect(record);
}
}
