file2.close() #Testing file pointers file3 = open(args.SLFT) file4 = open(args.TLFT) #Create new files where we could write our Pairs file5 = open(args.OUT, 'w') while True: line = preprocess(file3.readline()) line1 = preprocess(file4.readline()) if not line or not line1: break if len(line.split()) > max_len: continue for s,t in zip(src_sentences, tgt_sentences): fms = FMS(s, line) max_fms = fms.get_max_fms() #Get max possible FMS for the pair if max_fms >= min_fms: fms = fms.calculate_using_wanger_fischer() #Get actual FMS if fms >= min_fms and fms < 1.0: file5.write(s+"\n") file5.write(t+"\n") file5.write(line+"\n") file5.write(line1+"\n") #Close remaining files file3.close() file4.close() file5.close()
use_caching = True if cache_db_file else False
# Initiate and check Apertium
apertium = Apertium(lps[0], lps[1])
(out, err) = apertium.check_installations(lp_dir)
assertion(out, err)
tmxf = TMXFile(tmxfile, lps[0], lps[1])
tmunits = tmxf.getunits()
fmses = {}
for tmxu in tmunits:
src, tgt = preprocess(tmxu.getsource()), preprocess(tmxu.gettarget())
fms = FMS(s_sentence, src)
max_fms = fms.get_max_fms()
if max_fms >= min_fms:
fms = fms.calculate_using_wanger_fischer()
if fms >= min_fms:
fmses[(src, tgt)] = fms
assertion(fmses != {}, "No proper match with FMS > {0} could be found".format(min_fms))
sorted_fms = sorted(fmses, key=fmses.get)
(src, tgt) = sorted_fms[0] # Best match
patcher = Patcher(apertium, src, s_sentence, tgt, use_caching, cache_db_file)
patches = patcher.patch(min_len, max_len, grounded, lp_dir)
best_patch = patcher.get_best_patch()
