CODEWITHSUNDEEP
pythonfileencodingutf-8codec
Andrew_1510
Andrew_1510

Reputation: 13526

how to write utf-8 string to utf-8 file on utf-8 machine with python

Not the first time, by this confused me:

Open the file with codecs.open:

cfh = codecs.open('/tmp/ddfh', 'wb', 'utf-8')

Try to write the string, sa:

In [109]: sa
Out[109]: '\xe6\x96\xb0 \xe9\x97\xbb\xe3\x80\x80\xe7\xbd\x91 \xe9\xa1\xb5\xe3\x80\x80\xe8\xb4\xb4 \xe5\x90\xa7\xe3\x80\x80\xe7\x9f\xa5 \xe9\x81\x93\xe3\x80\x80\xe9\x9f\xb3 \xe4\xb9\x90\xe3\x80\x80\xe5\x9b\xbe \xe7\x89\x87\xe3\x80\x80\xe8\xa7\x86 \xe9\xa2\x91\xe3\x80\x80\xe5\x9c\xb0 \xe5\x9b\xbe'

In [110]: print sa
新 闻　网 页　贴 吧　知 道　音 乐　图 片　视 频　地 图

In [111]: sa.encode()
---------------------------------------------------------------------------
UnicodeDecodeError                        Traceback (most recent call last)
/home/za/tmp/<ipython-input-111-dea686030e89> in <module>()
----> 1 sa.encode()

UnicodeDecodeError: 'ascii' codec can't decode byte 0xe6 in position 0: ordinal not in range(128)

In [112]: sa.decode()
---------------------------------------------------------------------------
UnicodeDecodeError                        Traceback (most recent call last)
/home/za/tmp/<ipython-input-112-a79b22010b0e> in <module>()
----> 1 sa.decode()

UnicodeDecodeError: 'ascii' codec can't decode byte 0xe6 in position 0: ordinal not in range(128)

In [113]: sa.encode('utf-8')
---------------------------------------------------------------------------
UnicodeDecodeError                        Traceback (most recent call last)
/home/za/tmp/<ipython-input-113-ed97f8f61eb5> in <module>()
----> 1 sa.encode('utf-8')

UnicodeDecodeError: 'ascii' codec can't decode byte 0xe6 in position 0: ordinal not in range(128)

In [114]: sa.decode('utf-8')
Out[114]: u'\u65b0 \u95fb\u3000\u7f51 \u9875\u3000\u8d34 \u5427\u3000\u77e5 \u9053\u3000\u97f3 \u4e50\u3000\u56fe \u7247\u3000\u89c6 \u9891\u3000\u5730 \u56fe'

In [115]: cfh.write(sa.decode('utf-8'))

It works in the above, but FAILED with another machine, same Ubuntu machine, same $LANG env. I keep hitting "'ascii' codec can't ...."

Who can point me to a good doc? the official doc about module codecs is not good for me.

===

The problem comes from the codes:

# encoding=utf-8
# ......

def write_video_info_file(folder, filename, infos):
    # infos : a list of list, lines of text grouped by topic, results of language translations.

    absfn = os.path.join(folder, filename)
    with codecs.open(absfn, mode='wb', encoding='utf-8') as fh:
        for vinfo in infos:
            for v in vinfo:
                fh.write(v)
            fh.write("\n\n" + vi_delimit + "\n\n")

This was tested OK in my local machine, and deployed to a remote machine, then it get a lot: UnicodeDecodeError: 'ascii' codec can't.

After it, nearly all mode=, open without codecs tried.

$ echo $LANG  #  en_US.UTF-8

Python 2.7.3

Ubuntu 12.04

LANG=en_US.UTF-8

LANGUAGE=

LC_ALL=

===

I got the solution, use this to make sure all string are utf-8:

if isinstance(mystring, str):
    mystring = mystring.decode('utf-8')

Upvotes: 0

Views: 4034

Answers (1)

Martijn Pieters
Martijn Pieters

Reputation: 1121376

Your data is already encoded to UTF-8. Just open the file without codecs.open() and write the data directly:

with open('/tmp/ddfh', 'wb') as output:
    output.write(sa)

Unicode encoding / decoding errors usually occur because you are mixing byte strings and unicode strings; concatenation, comparisons, using str.join() when you needed to use unicode.join() instead, etc.

You may want to read up on Python and Unicode:

Upvotes: 3

Related Questions