it can recognize only French, English or German text.
accuracy depends how many characters are in text.

run:
./tri.py corpus/eng.txt >eng.tri.txt
./tri.py corpus/de.txt >de.tri.txt
./tri.py corpus/fr.txt >fr.tri.txt

and, for testing, run:

./test.py eng.tri.txt fr.tri.txt de.tri.txt justtext.txt

and Voilà!