import chardet
import argparse
import os
import codecs
def get_encoding(file_name):
f = open(file_name, 'rb')
encoding = chardet.detect(f.read())['encoding']
return encoding
def write(file_name, content):
with codecs.open(file_name, 'w', 'utf-8') as f:
f.write(content)
def read(file_name, encoding):
with codecs.open(file_name, 'r', encoding) as f:
return f.read()
def build(input_file_name):
dirname = os.path.dirname(input_file_name)
base = os.path.splitext(os.path.basename(input_file_name))
return os.path.join(dirname, ''.join(base[0] + r' utf-8'+base[-1]))
def init_parser():
parser = argparse.ArgumentParser()
parser.add_argument(r'input')
parser.add_argument(r'-e', '--encoding')
parser.add_argument(r'-o', '--output')
return parser
if __name__ == '__main__':
parser = init_parser()
args = parser.parse_args()
input_file_name = args.input
encoding_file = args.encoding or get_encoding(input_file_name)
output = args.output or build(input_file_name)
content = None
try:
content = read(input_file_name, encoding_file)
except OSError:
print("something wrong")
finally:
write(output, content)
print("Done")
时隔多年，我答我自己
关注
