/**
* Decode file charset.
*
* @param f File to process.
* @return File charset.
* @throws IOException in case of error.
*/
public static Charset decode(File f) throws IOException {
SortedMap<String, Charset> charsets = Charset.availableCharsets();
String[] firstCharsets = {
Charset.defaultCharset().name(), "US-ASCII", "UTF-8", "UTF-16BE", "UTF-16LE"
};
Collection<Charset> orderedCharsets = U.newLinkedHashSet(charsets.size());
for (String c : firstCharsets)
if (charsets.containsKey(c)) orderedCharsets.add(charsets.get(c));
orderedCharsets.addAll(charsets.values());
try (RandomAccessFile raf = new RandomAccessFile(f, "r")) {
FileChannel ch = raf.getChannel();
ByteBuffer buf = ByteBuffer.allocate(4096);
ch.read(buf);
buf.flip();
for (Charset charset : orderedCharsets) {
CharsetDecoder decoder = charset.newDecoder();
decoder.reset();
try {
decoder.decode(buf);
return charset;
} catch (CharacterCodingException ignored) {
}
}
}
return Charset.defaultCharset();
}
private void parsestories() {
try {
List<String> lns =
Files.readAllLines(Paths.get("datasets/" + name + ".tsv"), Charset.defaultCharset());
for (String ln : lns) stories.add(Story.fromtext(ln));
} catch (IOException e) {
System.out.println("Error reading dataset.");
System.exit(1);
}
}
/**
* Returns the connection charset. Defaults to {@link Charset} defaultCharset if not set
*
* @return charset
*/
public String getCharset()
{
return charset == null ? Charset.defaultCharset().name() : charset;
}