public void fingerprintFiles(List<String> paths, Polynomial p)
throws FileNotFoundException, IOException {
final RabinFingerprintLong rabin = new RabinFingerprintLong(p);
for (String path : paths) {
File file = new File(path);
if (file.exists()) {
rabin.reset();
rabin.pushBytes(ByteStreams.toByteArray(new FileInputStream(file)));
System.out.println(
String.format("%X %s", rabin.getFingerprintLong(), file.getAbsolutePath()));
System.out.flush();
} else {
System.err.print(String.format("Could not find file %s", path));
System.err.flush();
}
}
}
/**
* Fingerprint the file into chunks called "Fingers". The chunk boundaries are determined using a
* windowed fingerprinter {@link RabinFingerprintLongWindowed}.
*
* <p>The chunk detector is position independent. Therefore, even if a file is rearranged or
* partially corrupted, the untouched chunks can be efficiently discovered.
*/
public void getChunkFingerprints(InputStream is, ChunkVisitor visitor) throws IOException {
// windowing fingerprinter for finding chunk boundaries. this is only
// reset at the beginning of the file
final RabinFingerprintLong window = newWindowedFingerprint();
// fingerprinter for chunks. this is reset after each chunk
final RabinFingerprintLong finger = newFingerprint();
// counters
long chunkStart = 0;
long chunkEnd = 0;
/*
* fingerprint one byte at a time. we have to use this granularity to
* ensure that, for example, a one byte offset at the beginning of the
* file won't effect the chunk boundaries
*/
for (byte b : ByteStreams.toByteArray(is)) {
// push byte into fingerprints
window.pushByte(b);
finger.pushByte(b);
chunkEnd++;
/*
* if we've reached a boundary (which we will at some probability
* based on the boundary pattern and the size of the fingerprint
* window), we store the current chunk fingerprint and reset the
* chunk fingerprinter.
*/
if (boundaryDetector.isBoundary(window)) {
visitor.visit(finger.getFingerprintLong(), chunkStart, chunkEnd);
finger.reset();
// store last chunk offset
chunkStart = chunkEnd;
}
}
// final chunk
visitor.visit(finger.getFingerprintLong(), chunkStart, chunkEnd);
}
/** Rapidly fingerprint an entire stream's contents. */
public long getFullFingerprint(InputStream is) throws IOException {
final RabinFingerprintLong finger = newFingerprint();
finger.pushBytes(ByteStreams.toByteArray(is));
return finger.getFingerprintLong();
}
public void fingerprintStdin(Polynomial p) throws IOException {
final RabinFingerprintLong rabin = new RabinFingerprintLong(p);
rabin.pushBytes(ByteStreams.toByteArray(System.in));
System.out.println(String.format("%X", rabin.getFingerprintLong()));
}
