Fixing corrupt encoding (with Python)

Question

I have bunch of text files contains Korean characters with wrong encodings. Specifically, it seems the characters are encoded with EUC-KR, but the files themselves were saved with UTF8+BOM.

So far I managed to fix a file with the following:

1. Open a file with EditPlus (it shows the file's encoding is UTF8+BOM)
2. In EditPlus, save the file as ANSI

3. Lastly, in Python:

   with codecs.open(html, 'rb', encoding='euc-kr') as source_file:
       contents = source_file.read()
   
   with open(html, 'w+b') as dest_file:
       dest_file.write(contents.encode('utf-8'))
   

I want to automate this, but I have not been able to do so. I can open the original file in Python:

codecs.open(html, 'rb', encoding='utf-8-sig')

However, I haven't been able to figure out how to do the 2. part.

Answer 1

I am presuming here that you have text already encoded to EUC-KR, then encoded again to UTF-8. If so, encoding to Latin 1 (what Windows calls ANSI) is indeed the best way to get back to the original EUC-KR bytestring.

Open the file as UTF8 with BOM, encode to Latin1, decode as EUC-KR:

import io

with io.open(html, encoding='utf-8-sig') as infh:
    data = infh.read().encode('latin1').decode('euc-kr')

with io.open(html, 'w', encoding='utf8') as outfh:
    outfh.write(data)

I'm using the io.open() function here instead of codecs as the more robust method; io is the new Python 3 library also backported to Python 2.

Demo:

>>> broken = '\xef\xbb\xbf\xc2\xb9\xc3\x8c\xc2\xbc\xc3\xba'
>>> print broken.decode('utf-8-sig').encode('latin1').decode('euc-kr')
미술