/**
* Divide the given record into sub-records.
*
* @return the sub-records or null if the record cannot be divided by this strategy.
*/
public List<Record> divide(Record record) {
List<Record> result = null;
//
// algorithm:
// - find repeating text with repeating paths using the view.
// - group the most (equally) frequent paths into records.
// - distribute same paths across records.
//
final PathHistogram pathHistogram = new PathHistogram(record);
final PathHistogram.FieldGroupCollector fieldGroupCollector =
new PathHistogram.FieldGroupCollector();
for (Iterator<PathHistogram.PathGroup> pgIter = pathHistogram.iterator(); pgIter.hasNext(); ) {
final PathHistogram.PathGroup pathGroup = pgIter.next();
if (!accept(pathGroup)) continue;
final boolean complete = fieldGroupCollector.addAll(pathGroup);
if (accept(fieldGroupCollector)) {
result = fieldGroupCollector.getRecords(false /*includeIncomplete*/);
if (accept(result)) break;
} else {
// zero out for another try
fieldGroupCollector.clear();
}
}
return result;
}
protected boolean accept(PathHistogram.FieldGroupCollector fieldGroupCollector) {
boolean result = false;
// accept if there are more than 2 records and none of them are null.
final List<Record> records = fieldGroupCollector.getRecords(false /*includeIncomplete*/);
if (records.size() > 2) {
result = true;
for (Record record : records) {
if (record == null) {
result = false;
break;
}
}
}
return result;
}
