UnicodeDecodeError, invalid continuation byte

Question

Why is the below item failing? and why does it succeed with "latin-1" codec?

o = "a test of \xe9 char" #I want this to remain a string as this is what I am receiving
v = o.decode("utf-8")

results in:

 Traceback (most recent call last):  
 File "<stdin>", line 1, in <module>  
 File "C:\Python27\lib\encodings\utf_8.py",
 line 16, in decode
     return codecs.utf_8_decode(input, errors, True) UnicodeDecodeError:
 'utf8' codec can't decode byte 0xe9 in position 10: invalid continuation byte

Answer 1

If this error arises when manipulating a file that was just opened, check to see if you opened it in 'rb' mode

Answer 2

I had the same error when I tried to open a csv file by pandas read_csv method.

The solution was change the encoding to 'latin-1':

pd.read_csv('ml-100k/u.item', sep='|', names=m_cols , encoding='latin-1')

Answer 3

Because UTF-8 is multibyte and there is no char corresponding to your combination of \xe9 plus following space.

Why should it succeed in both utf-8 and latin-1?

Here how the same sentence should be in utf-8:

>>> o.decode('latin-1').encode("utf-8")
'a test of \xc3\xa9 char'

Answer 4

It is invalid UTF-8. That character is the e-acute character in ISO-Latin1, which is why it succeeds with that codeset.

If you don't know the codeset you're receiving strings in, you're in a bit of trouble. It would be best if a single codeset (hopefully UTF-8) would be chosen for your protocol/application and then you'd just reject ones that didn't decode.

If you can't do that, you'll need heuristics.

Answer 5

In binary, 0xE9 looks like 1110 1001. If you read about UTF-8 on Wikipedia, you’ll see that such a byte must be followed by two of the form 10xx xxxx. So, for example:

>>> b'\xe9\x80\x80'.decode('utf-8')
u'\u9000'

But that’s just the mechanical cause of the exception. In this case, you have a string that is almost certainly encoded in latin 1. You can see how UTF-8 and latin 1 look different:

>>> u'\xe9'.encode('utf-8')
b'\xc3\xa9'
>>> u'\xe9'.encode('latin-1')
b'\xe9'

(Note, I'm using a mix of Python 2 and 3 representation here. The input is valid in any version of Python, but your Python interpreter is unlikely to actually show both unicode and byte strings in this way.)