import re, json

FNAMES  = {
    "English": 
    {"Spanish":  
     {"file": "en_es.txt", "regex": "([a-z]+)\s+([a-z]+) \("}, 
     "French": 
     {"file": "en_fr.txt", "regex": "([a-z]+)\s+([a-z]+) \(" }
     }, 
    "German": 
    {"Italian": 
     {"file": "de_it.txt", "regex": "([a-z]+)\s+([a-z]+)\s*$"}
     }
    }

output = {}
r = re.compile("([a-z]+)\s+([a-z]+) \(")
for lang_from in FNAMES: 
    output[lang_from] = {}
    for lang_to in FNAMES[lang_from]: 
        params = FNAMES[lang_from][lang_to]
        f = open(params["file"])
        d = {}
        r = re.compile(params["regex"])
        n = 0
        total = 0
        for line in f: 
            total+=1
            m =  r.match(line)
            if m is not None: 
                n+=1
                d[m.group(1)] = m.group(2)
        output[lang_from][lang_to] =  d
        print "%s -> %s: keeping  %s words out of %s" %(lang_from, lang_to, n,total)
        f.close()
f_out = open("dicts.json", "w")
f_out.write("var dicts="+json.dumps(output))
f_out.close();