public static void main(String[] args)
throws IOException, IllegalAccessException, InstantiationException {
HashingIndexor indexor = new HashingIndexor();
BitSampling.readHashFunctions();
// BitSampling.readHashFunctions(new FileInputStream(BitSampling.hashFunctionsFileName));
// LocalitySensitiveHashing.readHashFunctions();
for (int i = 0; i < args.length; i++) {
String arg = args[i];
if (arg.startsWith("-i") || arg.startsWith("--input-file")) {
// infile ...
if ((i + 1) < args.length) indexor.addInputFile(new File(args[i + 1]));
else printHelp();
} else if (arg.startsWith("-l") || arg.startsWith("--index")) {
// index
if ((i + 1) < args.length) indexor.setIndexPath(args[i + 1]);
else printHelp();
} else if (arg.startsWith("-f") || arg.startsWith("--feature")) {
// index
if ((i + 1) < args.length)
try {
indexor.setFeatureClass(Class.forName(args[i + 1]));
} catch (ClassNotFoundException e) {
System.err.println("Could not find feature class named " + args[i + 1]);
printHelp();
}
else printHelp();
} else if (arg.startsWith("-h")) {
// help
printHelp();
} else if (arg.startsWith("-s")) {
// silent ...
verbose = false;
} else if (arg.startsWith("-c")) {
// list of input files within a file.
if ((i + 1) < args.length) {
BufferedReader br = new BufferedReader(new FileReader(new File(args[i + 1])));
String file;
while ((file = br.readLine()) != null) {
if (file.trim().length() > 2) {
File f = new File(file);
if (f.exists()) indexor.addInputFile(f);
else System.err.println("Did not find file " + f.getName());
}
}
} else printHelp();
}
}
// check if there is an infile, an outfile and some features to extract.
if (!indexor.isConfigured()) {
printHelp();
} else {
indexor.run();
}
}
protected void addToDocument(GlobalFeature feature, Document document, String featureFieldName) {
// This is for debugging the image features.
// try {
//// System.out.println(feature.getClass().getName() + " " +
// document.getValues(DocumentBuilder.FIELD_NAME_IDENTIFIER)[0]);
// LireFeature f1 = feature.getClass().newInstance();
// f1.extract(ImageIO.read(new
// File(document.getValues(DocumentBuilder.FIELD_NAME_IDENTIFIER)[0])));
// float distance = feature.getDistance(f1);
// if (distance != 0) {
// System.out.println("Extracted:" +
// java.util.Arrays.toString(f1.getFeatureVector()).replaceAll("\\.0,", "") + "\n" +
// "Data :" +
// java.util.Arrays.toString(feature.getFeatureVector()).replaceAll("\\.0,", "") + "\n" +
// "Problem with " + f1.getClass().getName() + " at file " +
// document.getValues(DocumentBuilder.FIELD_NAME_IDENTIFIER)[0] + ", distance=" + distance
// );
//// System.out.println("Problem with " + f1.getClass().getName() + " at file " +
// document.getValues(DocumentBuilder.FIELD_NAME_IDENTIFIER)[0] + ", distance=" + distance);
// }
// } catch (Exception e) {
// e.printStackTrace();
//
// }
if (feature.getClass().getCanonicalName().equals(featureClass.getCanonicalName())) {
// generate hashes here:
// int[] hashes =
// LocalitySensitiveHashing.generateHashes(feature.getFeatureVector());
int[] hashes = BitSampling.generateHashes(feature.getFeatureVector());
// System.out.println(Arrays.toString(hashes));
// store hashes in index as terms
document.add(
new TextField(
featureFieldName + "_hash",
SerializationUtils.arrayToString(hashes),
Field.Store.YES));
// add the specific feature
document.add(new StoredField(featureFieldName, feature.getByteArrayRepresentation()));
}
// add the specific feature
// document.add(new StoredField(featureFieldName, feature.getByteArrayRepresentation()));
}
