def replaceJapanese(msg): if not translator2.mecab_initialized: translator2.initialize() msg = translator2.japanese_braille_separate(msg, _logwrite)[0] global kanadic if kanadic is None: kanadic = load_kanadic() for p in kanadic: try: msg = re.sub(p[0], p[1], msg) except: pass return msg
def pass2(verboseMode=False):
global output
outfile = '__h2output.txt'
with open(outfile, 'w') as f:
output = cStringIO.StringIO()
translator2.initialize(__print, jtalk_dir, dic_dir, user_dics)
log = output.getvalue()
output.close()
f.write(log)
f.write("\n")
count = 0
for t in tests:
if not t.has_key('input'):
continue
nabcc = False
if t.has_key('mode') and t['mode'] == 'NABCC':
nabcc = True
if t.has_key('text'):
output = cStringIO.StringIO()
result, pat, inpos1, inpos2 = translator2.translateWithInPos2(
t['text'], logwrite=__print, nabcc=nabcc)
log = output.getvalue()
output.close()
# inpos2
if t.has_key('inpos2'):
correct_inpos2 = ','.join(['%d' % n for n in t['inpos2'] ])
else:
correct_inpos2 = None
# inpos1
if t.has_key('inpos1'):
correct_inpos1 = ','.join(['%d' % n for n in t['inpos1'] ])
else:
correct_inpos1 = None
# merged inpos
inpos, outpos_ = translator2.mergePositionMap(
inpos1, inpos2, len(pat), len(t['text']))
# outpos
outpos = translator2.makeOutPos(inpos, len(t['text']), len(pat))
if t.has_key('inpos'):
correct_inpos = ','.join(['%d' % n for n in t['inpos'] ])
else:
correct_inpos = None
if t.has_key('outpos'):
correct_outpos = ','.join(['%d' % n for n in t['outpos'] ])
else:
correct_outpos = None
# result
result_inpos2 = ','.join(['%d' % n for n in inpos2])
result_inpos1 = ','.join(['%d' % n for n in inpos1])
result_inpos = ','.join(['%d' % n for n in inpos])
result_outpos = ','.join(['%d' % n for n in outpos])
# output
isError = False
if result != t['input'] or \
(correct_inpos2 and result_inpos2 != correct_inpos2) or \
(correct_inpos and result_inpos != correct_inpos) or \
(correct_outpos and result_outpos != correct_outpos):
isError = True
count+=1
if isError or verboseMode:
f.write("text : " + t['text'].encode('utf-8') + "\n")
f.write("correct: " + t['input'].encode('utf-8') + "\n")
f.write("result : " + result.encode('utf-8') + "\n")
f.write("pat : " + pat.encode('utf-8') + "\n")
if correct_inpos2:
f.write("cor_in2: " + correct_inpos2 + "\n")
if correct_inpos1:
f.write("cor_in1: " + correct_inpos1 + "\n")
if correct_inpos:
f.write("cor_in : " + correct_inpos + "\n")
if correct_outpos:
f.write("cor_out: " + correct_outpos + "\n")
f.write("res_in2: " + result_inpos2 + "\n")
f.write("res_in1: " + result_inpos1 + "\n")
f.write("res_in : " + result_inpos + "\n")
f.write("res_out: " + result_outpos + "\n")
if 'comment' in t and t['comment']:
f.write("comment: " + t['comment'].encode('utf-8') + "\n")
f.write("\n")
f.write(log)
f.write("\n")
print 'h2: %d error(s). see %s' % (count, outfile)
return (count, outfile)