// The #bytes required to serialize the count map.
// Here let us assume to use 4 bytes for each of the int items. Normally it will consume lesser
// bytes as we will use vints.
// TODO Do we need to consider 5 as the number of bytes for each of the int field? Else there is
// a chance of ArrayIndexOutOfBoundsException when all the int fields are having very large
// values. Will that ever occur?
private int countMapSerializationSize() {
int size = Bytes.SIZEOF_INT; // Write the number of entries in the Map
for (ImmutableBytesPtr key : this.valueVsCount.keySet()) {
// Add up the key and key's lengths (Int) and the value
size += key.getLength() + Bytes.SIZEOF_INT + Bytes.SIZEOF_INT;
}
return size;
}
// The heap size which will be taken by the count map.
private int countMapHeapSize() {
int size = 0;
if (this.valueVsCount.size() > 0) {
for (ImmutableBytesPtr key : this.valueVsCount.keySet()) {
size +=
SizedUtil.MAP_ENTRY_SIZE
+ // entry
Bytes.SIZEOF_INT
+ // key size
key.getLength()
+ SizedUtil.ARRAY_SIZE; // value size
}
} else {
// Initially when the getSize() is called, we dont have any entries in the map so as to
// tell the exact heap need. Let us approximate the #entries
SizedUtil.sizeOfMap(
DEFAULT_ESTIMATED_DISTINCT_VALUES, SizedUtil.IMMUTABLE_BYTES_PTR_SIZE, Bytes.SIZEOF_INT);
}
return size;
}
@Override
public boolean evaluate(Tuple tuple, ImmutableBytesWritable ptr) {
// This serializes the Map. The format is as follows
// Map size(VInt ie. 1 to 5 bytes) +
// ( key length [VInt ie. 1 to 5 bytes] + key bytes + value [VInt ie. 1 to 5 bytes] )*
buffer = new byte[countMapSerializationSize()];
int offset = 0;
offset += ByteUtil.vintToBytes(buffer, offset, this.valueVsCount.size());
for (Entry<ImmutableBytesPtr, Integer> entry : this.valueVsCount.entrySet()) {
ImmutableBytesPtr key = entry.getKey();
offset += ByteUtil.vintToBytes(buffer, offset, key.getLength());
System.arraycopy(key.get(), key.getOffset(), buffer, offset, key.getLength());
offset += key.getLength();
offset += ByteUtil.vintToBytes(buffer, offset, entry.getValue().intValue());
}
ptr.set(buffer, 0, offset);
return true;
}