def write_file_eval(self, out_fname, trans, data_prefix, alns=None):
if alns is not None:
fout_aln = open('{}.aln'.format(out_fname), 'w') # valids/trans
fout_aln.writelines(alns)
fout_aln.close()
fout = open(out_fname, 'w') # valids/trans
fout.writelines(trans)
fout.close()
ref_fpaths = []
# *.ref
ref_fpath = '{}{}.{}'.format(wargs.val_tst_dir, data_prefix,
wargs.val_ref_suffix)
if os.path.exists(ref_fpath): ref_fpaths.append(ref_fpath)
for idx in range(wargs.ref_cnt):
# *.ref0, *.ref1, ...
ref_fpath = '{}{}.{}{}'.format(wargs.val_tst_dir, data_prefix,
wargs.val_ref_suffix, idx)
if not os.path.exists(ref_fpath): continue
ref_fpaths.append(ref_fpath)
if wargs.with_bpe is True:
os.system('cp {} {}.bpe'.format(out_fname, out_fname))
wlog('cp {} {}.bpe'.format(out_fname, out_fname))
os.system("sed -r 's/(@@ )|(@@ ?$)//g' {}.bpe > {}".format(
out_fname, out_fname))
wlog("sed -r 's/(@@ )|(@@ ?$)//g' {}.bpe > {}".format(
out_fname, out_fname))
if wargs.with_postproc is True:
opost_name = '{}.opost'.format(out_fname)
os.system('cp {} {}'.format(out_fname, opost_name))
wlog('cp {} {}'.format(out_fname, opost_name))
os.system("sh postproc.sh {} {}".format(opost_name, out_fname))
wlog("sh postproc.sh {} {}".format(opost_name, out_fname))
mteval_bleu_opost = bleu_file(opost_name, ref_fpaths)
os.rename(opost_name, "{}_{}.txt".format(opost_name,
mteval_bleu_opost))
'''
os.system('cp {} {}.bpe'.format(out_fname, out_fname))
wlog('cp {} {}.bpe'.format(out_fname, out_fname))
os.system("sed -r 's/(@@ )|(@@ ?$)//g' {}.bpe > {}".format(out_fname, out_fname))
wlog("sed -r 's/(@@ )|(@@ ?$)//g' {}.bpe > {}".format(out_fname, out_fname))
src_sgm = '{}/{}.src.sgm'.format(wargs.val_tst_dir, data_prefix)
os.system('./scripts/wrap_xml.pl zh {} {} < {} > {}.sgm'.format(src_sgm, data_prefix, out_fname, out_fname))
wlog('./scripts/wrap_xml.pl zh {} {} < {} > {}.sgm'.format(src_sgm, data_prefix, out_fname, out_fname))
os.system('./scripts/chi_char_segment.pl -t xml < {}.sgm > {}.seg.sgm'.format(out_fname, out_fname))
wlog('./scripts/chi_char_segment.pl -t xml < {}.sgm > {}.seg.sgm'.format(out_fname, out_fname))
os.system('./scripts/de-xml.pl < {}.seg.sgm > {}.seg.plain'.format(out_fname, out_fname))
wlog('./scripts/de-xml.pl < {}.seg.sgm > {}.seg.plain'.format(out_fname, out_fname))
'''
mteval_bleu = bleu_file(out_fname, ref_fpaths)
#mteval_bleu = bleu_file(out_fname + '.seg.plain', ref_fpaths)
os.rename(out_fname, "{}_{}.txt".format(out_fname, mteval_bleu))
return mteval_bleu_opost if wargs.with_postproc is True else mteval_bleu
def write_file_eval(self,
out_fname,
trans,
data_prefix,
alns=None,
subw=None):
if alns is not None:
fout_aln = open('{}.aln'.format(out_fname), 'w') # valids/trans
fout_aln.writelines(alns)
fout_aln.close()
if subw is not None:
fout_subw = open('{}.subword'.format(out_fname),
'w') # valids/trans
fout_subw.writelines(subw)
fout_subw.close()
fout = open(out_fname, 'w') # valids/trans
fout.writelines(trans)
fout.close()
ref_fpaths = []
# *.ref
ref_fpath = '{}{}.{}'.format(wargs.val_tst_dir, data_prefix,
wargs.val_ref_suffix)
if os.path.exists(ref_fpath): ref_fpaths.append(ref_fpath)
for idx in range(wargs.ref_cnt):
# *.ref0, *.ref1, ...
ref_fpath = '{}{}.{}{}'.format(wargs.val_tst_dir, data_prefix,
wargs.val_ref_suffix, idx)
if not os.path.exists(ref_fpath): continue
ref_fpaths.append(ref_fpath)
assert os.path.exists(out_fname), 'translation do not exist ...'
if wargs.with_bpe is True:
bpe_fname = '{}.bpe'.format(out_fname)
wlog('copy {} to {} ... '.format(out_fname, bpe_fname), 0)
#os.system('cp {} {}.bpe'.format(out_fname, out_fname))
copyfile(out_fname, bpe_fname)
assert os.path.exists(bpe_fname), 'bpe file do not exist ...'
wlog('done')
wlog(
"sed -r 's/(@@ )|(@@ ?$)//g' {} > {} ... ".format(
bpe_fname, out_fname), 0)
#os.system("sed -r 's/(@@ )|(@@ ?$)//g' {} > {}".format(bpe_fname, out_fname))
proc_bpe(bpe_fname, out_fname)
wlog('done')
# Luong: remove "rich-text format" --> rich ##AT##-##AT## text format
#os.system("sed -r -i 's/( ##AT##)|(##AT## )//g' {}".format(out_fname))
#wlog("sed -r -i 's/( ##AT##)|(##AT## )//g' {}".format(out_fname))
if wargs.with_postproc is True:
opost_name = '{}.opost'.format(out_fname)
wlog('copy {} to {} ... '.format(out_fname, opost_name), 0)
#os.system('cp {} {}'.format(out_fname, opost_name))
copyfile(out_fname, opost_name)
assert os.path.exists(opost_name), 'opost file do not exist ...'
wlog('done')
wlog("sh postproc.sh {} {}".format(opost_name, out_fname))
os.system("sh postproc.sh {} {}".format(opost_name, out_fname))
wlog('done')
mteval_bleu_opost = bleu_file(opost_name,
ref_fpaths,
cased=wargs.cased)
os.rename(opost_name, "{}_{}.txt".format(opost_name,
mteval_bleu_opost))
mteval_bleu = bleu_file(out_fname, ref_fpaths, cased=wargs.cased)
multi_bleu = print_multi_bleu(out_fname, ref_fpaths, cased=wargs.cased)
#mteval_bleu = bleu_file(out_fname + '.seg.plain', ref_fpaths)
if wargs.char is True:
c_mteval_bleu = bleu_file(out_fname,
ref_fpaths,
cased=wargs.cased,
char=True)
c_multi_bleu = print_multi_bleu(out_fname,
ref_fpaths,
cased=wargs.cased,
char=True)
os.rename(
out_fname,
"{}_{}_{}_c_{}_{}.txt".format(out_fname, mteval_bleu,
multi_bleu, c_mteval_bleu,
c_multi_bleu))
else:
os.rename(
out_fname, "{}_{}_{}.txt".format(out_fname, mteval_bleu,
multi_bleu))
if wargs.use_multi_bleu is False:
if wargs.char is False: final_bleu = mteval_bleu
else: final_bleu = c_mteval_bleu
else:
if wargs.char is False: final_bleu = multi_bleu
else: final_bleu = c_multi_bleu
return final_bleu
def write_file_eval(self, out_fname, trans, data_prefix, alns=None):
if alns is not None:
fout_aln = open('{}.aln'.format(out_fname), 'w') # valids/trans
fout_aln.writelines(alns)
fout_aln.close()
fout = open(out_fname, 'w') # valids/trans
fout.writelines(trans)
fout.close()
# *.ref
ref_fpath = '{}{}.{}'.format(wargs.val_tst_dir, data_prefix,
wargs.val_ref_suffix)
ref_fpaths = grab_all_trg_files(ref_fpath)
assert os.path.exists(out_fname), 'translation do not exist ...'
if wargs.with_bpe is True:
bpe_fname = '{}.bpe'.format(out_fname)
wlog('copy {} to {} ... '.format(out_fname, bpe_fname), 0)
#os.system('cp {} {}.bpe'.format(out_fname, out_fname))
copyfile(out_fname, bpe_fname)
assert os.path.exists(bpe_fname), 'bpe file do not exist ...'
wlog('done')
wlog(
"sed -r 's/(@@ )|(@@ ?$)//g' {} > {} ... ".format(
bpe_fname, out_fname), 0)
#os.system("sed -r 's/(@@ )|(@@ ?$)//g' {} > {}".format(bpe_fname, out_fname))
proc_bpe(bpe_fname, out_fname)
wlog('done')
if wargs.with_postproc is True:
opost_name = '{}.opost'.format(out_fname)
wlog('copy {} to {} ... '.format(out_fname, opost_name), 0)
#os.system('cp {} {}'.format(out_fname, opost_name))
copyfile(out_fname, opost_name)
assert os.path.exists(opost_name), 'opost file do not exist ...'
wlog('done')
wlog("sh postproc.sh {} {}".format(opost_name, out_fname))
os.system("sh postproc.sh {} {}".format(opost_name, out_fname))
wlog('done')
mteval_bleu_opost = bleu_file(opost_name,
ref_fpaths,
cased=wargs.cased)
os.rename(opost_name, "{}_{}.txt".format(opost_name,
mteval_bleu_opost))
mteval_bleu = bleu_file(out_fname,
ref_fpaths,
cased=wargs.cased,
char=wargs.char_bleu)
multi_bleu = print_multi_bleu(out_fname,
ref_fpaths,
cased=wargs.cased,
char=wargs.char_bleu)
#mteval_bleu = bleu_file(out_fname + '.seg.plain', ref_fpaths)
os.rename(
out_fname,
'{}{}_{}_{}.txt'.format(out_fname,
'_char' if wargs.char_bleu is True else '',
mteval_bleu, multi_bleu))
final_bleu = multi_bleu if wargs.use_multi_bleu is True else mteval_bleu
return final_bleu