def run(args: Namespace):
"""
run function which is the start point of program
Args:
args: program arguments
"""
aligner = Aligner(args.rsc_src)
funmap = open(args.unmapped, 'w',
encoding='UTF-8') if args.unmapped else None
for sent in sent_iter(args):
if 0.0 < args.sample < 1.0 and random.random() >= args.sample:
continue
word_per_maps = []
for word in sent.words:
try:
maps = aligner.align(word)
except AlignError as algn_err:
if funmap:
algn_err.add_msg(str(word))
print(algn_err, file=funmap)
maps = []
word_per_maps.append(maps)
_print_sent(sent, word_per_maps)
_print_restore_dic(args)
aligner.print_middle_cnt()
def _set_align(aligner: Aligner, Word: type, entries: List[Entry]): # pylint: disable=invalid-name
"""
음절과 형태소 분석 결과를 정렬한다.
Args:
aligner: Aligner 객체
Word: Word 타입
entries: 엔트리 리스트
"""
bad_entries = []
for entry in entries:
if entry.is_sharp:
continue
word = Word.parse(
'\t'.join(['', entry.word,
Morph.to_str(entry.morphs)]), '', 0)
try:
entry.align = aligner.align(word)
except AlignError as map_exc:
entry.err_msg = 'fail to align'
logging.error(map_exc)
bad_entries.append(entry)
print_errors(bad_entries)