public void seek(TermEnum terms) throws IOException {
original.seek(terms);
docFreq = terms.docFreq();
pointer = -1;
if (docFreq > postingMaps.length) { // grow postingsMap
PostingMap[] newMap = new PostingMap[docFreq];
System.arraycopy(postingMaps, 0, newMap, 0, postingMaps.length);
for (int i = postingMaps.length; i < docFreq; i++) {
newMap[i] = new PostingMap();
}
postingMaps = newMap;
}
out.reset();
int i = 0;
while (original.next()) {
PostingMap map = postingMaps[i++];
map.newDoc = oldToNew[original.doc()]; // remap the newDoc id
map.offset = out.getFilePointer(); // save pointer to buffer
final int tf = original.freq(); // buffer tf & positions
out.writeVInt(tf);
int prevPosition = 0;
for (int j = tf; j > 0; j--) { // delta encode positions
int p = original.nextPosition();
out.writeVInt(p - prevPosition);
prevPosition = p;
}
}
out.flush();
docFreq = i; // allow for deletions
Arrays.sort(postingMaps, 0, docFreq); // resort by mapped doc ids
// HeapSorter.sort(postingMaps,docFreq); // TODO MC - due to the lack of space
// NOTE: this might be substantially faster if RAMInputStream were public
// and supported a reset() operation.
in = tempDir.openInput(TEMP_FILE);
}
private static class SortedTermPositions implements TermPositions {
private TermPositions original;
private int[] oldToNew;
private int docFreq;
private PostingMap[] postingMaps = new PostingMap[0];
private int pointer;
private int freq;
private int position;
private static final String TEMP_FILE = "temp";
private final RAMDirectory tempDir = new RAMDirectory();
private final RAMOutputStream out = (RAMOutputStream) tempDir.createOutput(TEMP_FILE);
private IndexInput in;
public SortedTermPositions(TermPositions original, int[] oldToNew) {
this.original = original;
this.oldToNew = oldToNew;
}
public void seek(Term term) throws IOException {
throw new UnsupportedOperationException();
}
public void seek(TermEnum terms) throws IOException {
original.seek(terms);
docFreq = terms.docFreq();
pointer = -1;
if (docFreq > postingMaps.length) { // grow postingsMap
PostingMap[] newMap = new PostingMap[docFreq];
System.arraycopy(postingMaps, 0, newMap, 0, postingMaps.length);
for (int i = postingMaps.length; i < docFreq; i++) {
newMap[i] = new PostingMap();
}
postingMaps = newMap;
}
out.reset();
int i = 0;
while (original.next()) {
PostingMap map = postingMaps[i++];
map.newDoc = oldToNew[original.doc()]; // remap the newDoc id
map.offset = out.getFilePointer(); // save pointer to buffer
final int tf = original.freq(); // buffer tf & positions
out.writeVInt(tf);
int prevPosition = 0;
for (int j = tf; j > 0; j--) { // delta encode positions
int p = original.nextPosition();
out.writeVInt(p - prevPosition);
prevPosition = p;
}
}
out.flush();
docFreq = i; // allow for deletions
Arrays.sort(postingMaps, 0, docFreq); // resort by mapped doc ids
// HeapSorter.sort(postingMaps,docFreq); // TODO MC - due to the lack of space
// NOTE: this might be substantially faster if RAMInputStream were public
// and supported a reset() operation.
in = tempDir.openInput(TEMP_FILE);
}
public boolean next() throws IOException {
pointer++;
if (pointer < docFreq) {
in.seek(postingMaps[pointer].offset);
freq = in.readVInt();
position = 0;
return true;
}
return false;
}
public int doc() {
return postingMaps[pointer].newDoc;
}
public int freq() {
return freq;
}
public int nextPosition() throws IOException {
int positionIncrement = in.readVInt();
position += positionIncrement;
return position;
}
public int read(int[] docs, int[] freqs) {
throw new UnsupportedOperationException();
}
public boolean skipTo(int target) {
throw new UnsupportedOperationException();
}
public void close() throws IOException {
original.close();
}
}