public static void main(String[] args) {
try {
byte[] b = FileUtils.read(new File(args[0]));
torrentParser parser = new torrentParser();
Document[] d =
parser.parse(
new DigestURL("http://localhost/test.torrent"),
null,
StandardCharsets.UTF_8.name(),
new VocabularyScraper(),
0,
new ByteArrayInputStream(b));
Condenser c = new Condenser(d[0], null, true, true, LibraryProvider.dymLib, false, false, 0);
Map<String, Word> w = c.words();
for (Map.Entry<String, Word> e : w.entrySet())
System.out.println("Word: " + e.getKey() + " - " + e.getValue().posInText);
} catch (final IOException e) {
e.printStackTrace();
} catch (final Parser.Failure e) {
e.printStackTrace();
} catch (final InterruptedException e) {
e.printStackTrace();
}
}
@Override
public Document[] parse(
final DigestURL location,
final String mimeType,
final String charset,
final VocabularyScraper scraper,
final int timezoneOffset,
final InputStream source)
throws Parser.Failure, InterruptedException {
byte[] b = null;
try {
b = FileUtils.read(source);
} catch (final IOException e1) {
throw new Parser.Failure(e1.toString(), location);
}
final BDecoder bd = new BDecoder(b);
final BObject bo = bd.parse();
if (bo == null) throw new Parser.Failure("BDecoder.parse returned null", location);
if (bo.getType() != BType.dictionary)
throw new Parser.Failure("BDecoder object is not a dictionary", location);
final Map<String, BObject> map = bo.getMap();
final BObject commento = map.get("comment");
final String comment = (commento == null) ? "" : UTF8.String(commento.getString());
// Date creation = new Date(map.get("creation date").getInteger());
final BObject infoo = map.get("info");
final StringBuilder filenames = new StringBuilder(80);
String title = "";
if (infoo != null) {
final Map<String, BObject> info = infoo.getMap();
final BObject fileso = info.get("files");
if (fileso != null) {
final List<BObject> filelist = fileso.getList();
for (final BObject fo : filelist) {
final BObject patho = fo.getMap().get("path");
if (patho != null) {
final List<BObject> l = patho.getList(); // one file may have several names
for (final BObject fl : l) {
filenames.append(fl.toString()).append(" ");
}
}
}
}
final BObject nameo = info.get("name");
if (nameo != null) title = UTF8.String(nameo.getString());
}
if (title == null || title.isEmpty()) title = MultiProtocolURL.unescape(location.getFileName());
return new Document[] {
new Document(
location,
mimeType,
charset,
this,
null,
null,
singleList(title), // title
comment, // author
location.getHost(),
null,
null,
0.0d,
0.0d,
filenames.toString(),
null,
null,
null,
false,
new Date())
};
}
