def _set_align(rsc_src: Tuple[Aligner, dict, Dict[str, int]],
entries: List[Entry]):
"""
음절과 형태소 분석 결과를 정렬한다.
Args:
rsc_src: (Aligner, restore dic, vocab out) resource triple
Word: Word 타입
entries: 엔트리 리스트
"""
bad_entries = []
for entry in entries:
if entry.is_sharp:
continue
entry.left_align = align_patch(rsc_src, entry.raw,
Morph.to_str(entry.left))
if not entry.left_align:
entry.err_msg = 'fail to align left'
bad_entries.append(entry)
continue
entry.right_align = align_patch(rsc_src, entry.raw,
Morph.to_str(entry.right))
if not entry.right_align:
entry.err_msg = 'fail to align right'
bad_entries.append(entry)
continue
assert len(entry.left_align) == len(entry.right_align)
print_errors(bad_entries)
def _load_entries(args: Namespace) -> List[Entry]:
"""
패치 엔트리를 파일로부터 로드한다.
Args:
args: program arguments
Returns:
엔트리 리스트
"""
good_entries = []
bad_entries = []
for file_path in glob.glob('{}/{}.errpatch.*'.format(
args.rsc_src, args.model_size)):
file_name = os.path.basename(file_path)
logging.info(file_name)
for line_num, line in enumerate(open(file_path, 'r', encoding='UTF-8'),
start=1):
line = line.rstrip('\r\n')
if not line:
continue
entry = Entry(file_path, line_num, line)
if entry.err_msg:
bad_entries.append(entry)
else:
good_entries.append(entry)
print_errors(bad_entries)
return good_entries
def _check_dup(entries: List[Entry]):
"""
중복된 엔트리가 없는 지 확인한다.
Args:
entries: 엔트리 리스트
"""
bad_entries = []
key_dic = {}
for entry in entries:
if entry.key_str() in key_dic:
dup_entry = key_dic[entry.key_str()]
entry.err_msg = '[DUPLICATED] with "{}"'.format(dup_entry)
bad_entries.append(entry)
else:
key_dic[entry.key_str()] = entry
print_errors(bad_entries)
