public TrigramLanguageGuesser(URL[] urls) throws IOException {
tg = new TrigramGenerator();
tg.addTrigramListener(test);
for (int i = 0; i < urls.length; i++) {
URL url = urls[i];
String path = url.getPath();
int fileNameStart = path.lastIndexOf('/') + 1;
int fileNameEnd = path.lastIndexOf('.');
String code = path.substring(fileNameStart, fileNameEnd);
addFromStream(code, url.openStream());
}
}
/*
* Construct a LanguageGuesser
* fileLocation is a directory containing
* xx.tri files where xx is the ISO-639 Language Code
* see http://ftp.ics.uci.edu/pub/ietf/http/related/iso639.txt
*/
public TrigramLanguageGuesser(String fileLocation) throws IOException {
tg = new TrigramGenerator();
tg.addTrigramListener(test);
File f = new File(fileLocation);
if (f.isDirectory()) {
String[] files = f.list();
for (int i = 0; i < files.length; i++) {
if (files[i].endsWith("tri")) {
addFile(f.getAbsolutePath() + File.separator + files[i]);
}
}
}
if (trigramsmap.size() == 0)
throw new RuntimeException(
"Location [ " + f.getAbsolutePath() + " ] doesn't contain any .tri file");
}