/** Called when we are done adding docs to this term */
@Override
public void finishTerm(TermStats stats) throws IOException {
if (DEBUG)
System.out.println(
"PW finishTerm docCount="
+ stats.docFreq
+ " pendingCount="
+ pendingCount
+ " pendingTerms.size()="
+ pendingTerms.size());
assert pendingCount > 0 || pendingCount == -1;
if (pendingCount == -1) {
wrappedPostingsWriter.finishTerm(stats);
// Must add null entry to record terms that our
// wrapped postings impl added
pendingTerms.add(null);
} else {
// There were few enough total occurrences for this
// term, so we fully inline our postings data into
// terms dict, now:
// TODO: it'd be better to share this encoding logic
// in some inner codec that knows how to write a
// single doc / single position, etc. This way if a
// given codec wants to store other interesting
// stuff, it could use this pulsing codec to do so
if (indexOptions.compareTo(IndexOptions.DOCS_AND_FREQS_AND_POSITIONS) >= 0) {
int lastDocID = 0;
int pendingIDX = 0;
int lastPayloadLength = -1;
int lastOffsetLength = -1;
while (pendingIDX < pendingCount) {
final Position doc = pending[pendingIDX];
final int delta = doc.docID - lastDocID;
lastDocID = doc.docID;
if (DEBUG) System.out.println(" write doc=" + doc.docID + " freq=" + doc.termFreq);
if (doc.termFreq == 1) {
buffer.writeVInt((delta << 1) | 1);
} else {
buffer.writeVInt(delta << 1);
buffer.writeVInt(doc.termFreq);
}
int lastPos = 0;
int lastOffset = 0;
for (int posIDX = 0; posIDX < doc.termFreq; posIDX++) {
final Position pos = pending[pendingIDX++];
assert pos.docID == doc.docID;
final int posDelta = pos.pos - lastPos;
lastPos = pos.pos;
if (DEBUG) System.out.println(" write pos=" + pos.pos);
final int payloadLength = pos.payload == null ? 0 : pos.payload.length;
if (storePayloads) {
if (payloadLength != lastPayloadLength) {
buffer.writeVInt((posDelta << 1) | 1);
buffer.writeVInt(payloadLength);
lastPayloadLength = payloadLength;
} else {
buffer.writeVInt(posDelta << 1);
}
} else {
buffer.writeVInt(posDelta);
}
if (indexOptions.compareTo(IndexOptions.DOCS_AND_FREQS_AND_POSITIONS_AND_OFFSETS)
>= 0) {
// System.out.println("write=" + pos.startOffset + "," + pos.endOffset);
int offsetDelta = pos.startOffset - lastOffset;
int offsetLength = pos.endOffset - pos.startOffset;
if (offsetLength != lastOffsetLength) {
buffer.writeVInt(offsetDelta << 1 | 1);
buffer.writeVInt(offsetLength);
} else {
buffer.writeVInt(offsetDelta << 1);
}
lastOffset = pos.startOffset;
lastOffsetLength = offsetLength;
}
if (payloadLength > 0) {
assert storePayloads;
buffer.writeBytes(pos.payload.bytes, 0, pos.payload.length);
}
}
}
} else if (indexOptions == IndexOptions.DOCS_AND_FREQS) {
int lastDocID = 0;
for (int posIDX = 0; posIDX < pendingCount; posIDX++) {
final Position doc = pending[posIDX];
final int delta = doc.docID - lastDocID;
assert doc.termFreq != 0;
if (doc.termFreq == 1) {
buffer.writeVInt((delta << 1) | 1);
} else {
buffer.writeVInt(delta << 1);
buffer.writeVInt(doc.termFreq);
}
lastDocID = doc.docID;
}
} else if (indexOptions == IndexOptions.DOCS_ONLY) {
int lastDocID = 0;
for (int posIDX = 0; posIDX < pendingCount; posIDX++) {
final Position doc = pending[posIDX];
buffer.writeVInt(doc.docID - lastDocID);
lastDocID = doc.docID;
}
}
final byte[] bytes = new byte[(int) buffer.getFilePointer()];
buffer.writeTo(bytes, 0);
pendingTerms.add(new PendingTerm(bytes));
buffer.reset();
}
pendingCount = 0;
}
