import sys sys.path.append('/home/pat/.pylib') from lexicon import database from copy import copy from languages import * from collections import defaultdict from sqlite3 import dbapi2 as sqlite lex = sqlite.connect('/home/pat/lang/de/lexicon.db') lang = sys.argv[1] e2d = lex.execute('select source_term, target_term from lexicon where source_code = ? and target_code = ?', (fromlang, tolang)) e2d = dict([(e,d) for e,d in e2d]) d2e = copy(e2d) e2d = [(v,k) for k,v in d2e.items()] e2d = dict([(v,k) for k,v in d2e.items()]) dewords = ' '.join(d2e.values()).split() def freq(seq): d = defaultdict(int) for e in seq: d[e] += 1 return d dewordfq = freq(dewords) byfq = sorted([(v,k) for k,v in dewordfq.items()]) for fq,word in byfq[-100:]: print word