UnicodeDecodeError: 'ascii' codec can't decode byte 0xe2 in position 13: ordinal not in range(128)

Question

I'm using NLTK to perform kmeans clustering on my text file in which each line is considered as a document. So for example, my text file is something like this:

belong finger death punch <br>
hasty <br>
mike hasty walls jericho <br>
jägermeister rules <br>
rules bands follow performing jägermeister stage <br>
approach 

Now the demo code I'm trying to run is this:

import sys

import numpy
from nltk.cluster import KMeansClusterer, GAAClusterer, euclidean_distance
import nltk.corpus
from nltk import decorators
import nltk.stem

stemmer_func = nltk.stem.EnglishStemmer().stem
stopwords = set(nltk.corpus.stopwords.words('english'))

@decorators.memoize
def normalize_word(word):
    return stemmer_func(word.lower())

def get_words(titles):
    words = set()
    for title in job_titles:
        for word in title.split():
            words.add(normalize_word(word))
    return list(words)

@decorators.memoize
def vectorspaced(title):
    title_components = [normalize_word(word) for word in title.split()]
    return numpy.array([
        word in title_components and not word in stopwords
        for word in words], numpy.short)

if __name__ == '__main__':

    filename = 'example.txt'
    if len(sys.argv) == 2:
        filename = sys.argv[1]

    with open(filename) as title_file:

        job_titles = [line.strip() for line in title_file.readlines()]

        words = get_words(job_titles)

        # cluster = KMeansClusterer(5, euclidean_distance)
        cluster = GAAClusterer(5)
        cluster.cluster([vectorspaced(title) for title in job_titles if title])

        # NOTE: This is inefficient, cluster.classify should really just be
        # called when you are classifying previously unseen examples!
        classified_examples = [
                cluster.classify(vectorspaced(title)) for title in job_titles
            ]

        for cluster_id, title in sorted(zip(classified_examples, job_titles)):
            print cluster_id, title

(which can also be found here)

The error I receive is this:

Traceback (most recent call last):
File "cluster_example.py", line 40, in
words = get_words(job_titles)
File "cluster_example.py", line 20, in get_words
words.add(normalize_word(word))
File "", line 1, in
File "/usr/local/lib/python2.7/dist-packages/nltk/decorators.py", line 183, in memoize
result = func(*args)
File "cluster_example.py", line 14, in normalize_word
return stemmer_func(word.lower())
File "/usr/local/lib/python2.7/dist-packages/nltk/stem/snowball.py", line 694, in stem
word = (word.replace(u"\u2019", u"\x27")
UnicodeDecodeError: 'ascii' codec can't decode byte 0xe2 in position 13: ordinal not in range(128)

What is happening here?

Staid

Answer 1

The file is being read as a bunch of strs, but it should be unicodes. Python tries to implicitly convert, but fails. Change:

job_titles = [line.strip() for line in title_file.readlines()]

to explicitly decode the strs to unicode (here assuming UTF-8):

job_titles = [line.decode('utf-8').strip() for line in title_file.readlines()]

It could also be solved by importing the codecs module and using codecs.open rather than the built-in open.

Lyublin

running this line.decode('utf-8').strip().lower().split() also gives me the same error. I have added the .deocode('utf-8') – Rhinology

@kathirraja: Can you provide a reference for that? As far as I know, even in Python 3, the decode method remains the preferred way to decode a byte string to a Unicode string. (Though, the types in my answer are not right for Python 3 -- for Python 3, we're trying to convert from bytes to str rather than from str to unicode.) – Lyublin

Answer 2

This works fine for me.

f = open(file_path, 'r+', encoding="utf-8")

You can add a third parameter encoding to ensure the encoding type is 'utf-8'

Note: this method works fine in Python3, I did not try it in Python2.7.

Moldavia

It doesn't work in Python 2.7.10: TypeError: 'encoding' is an invalid keyword argument for this function – Cheddar

It doesn't work in Python 2.7.10: TypeError: 'encoding' is an invalid keyword argument for this function This works fine: import io with io.open(file_path, 'r', encoding="utf-8") as f: for line in f: do_something(line) – Cheddar

Worked like a charm in python3.6 Thanks a lot! – Bennion

Answer 3

For me there was a problem with the terminal encoding. Adding UTF-8 to .bashrc solved the problem:

export LC_CTYPE=en_US.UTF-8

Don't forget to reload .bashrc afterwards:

source ~/.bashrc

Pashm

I had to use export LC_ALL=C.UTF-8 on Ubuntu 18.04.3 and Python 3.6.8. Otherwise this solved my problem, thanks. – Foursome

For me it was solved with set -x LANG en_US.UTF-8 (fish) on macOS 10.15.7 and Python 3.6.7 – Tiemroth

In my case it was export LC_ALL= to revert a prior fix I made so that xclock would render fonts setting export LC_ALL=C. Then LC_TYPE went back to en_US.UTF-8. Of course we could also just remove the UTF-8 characters from the file and keep it to ASCII – Curtin

Answer 4

You can try this also:

import sys
reload(sys)
sys.setdefaultencoding('utf8')

Aciculum

What are the implications of this? It sounds like it's something global and not only applicable for this file. – Gynecocracy

Notice that the above is deprecated in Python 3. – Vermiculate

Answer 5

I got this error when trying to install a python package in a Docker container. For me, the issue was that the docker image did not have a locale configured. Adding the following code to the Dockerfile solved the problem for me.

# Avoid ascii errors when reading files in Python
RUN apt-get install -y locales && locale-gen en_US.UTF-8
ENV LANG='en_US.UTF-8' LANGUAGE='en_US:en' LC_ALL='en_US.UTF-8'

Fructuous

I had to use this: github.com/docker-library/python/issues/13 – Washerman

Answer 6

When on Ubuntu 18.04 using Python3.6 I have solved the problem doing both:

with open(filename, encoding="utf-8") as lines:

and if you are running the tool as command line:

export LC_ALL=C.UTF-8

Note that if you are in Python2.7 you have do to handle this differently. First you have to set the default encoding:

import sys
reload(sys)
sys.setdefaultencoding('utf-8')

and then to load the file you must use io.open to set the encoding:

import io
with io.open(filename, 'r', encoding='utf-8') as lines:

You still need to export the env

export LC_ALL=C.UTF-8

Aircrewman

Answer 7

To find ANY and ALL unicode error related... Using the following command:

grep -r -P '[^\x00-\x7f]' /etc/apache2 /etc/letsencrypt /etc/nginx

Found mine in

/etc/letsencrypt/options-ssl-nginx.conf:        # The following CSP directives don't use default-src as 

Using shed, I found the offending sequence. It turned out to be an editor mistake.

00008099:     C2  194 302 11000010
00008100:     A0  160 240 10100000
00008101:  d  64  100 144 01100100
00008102:  e  65  101 145 01100101
00008103:  f  66  102 146 01100110
00008104:  a  61  097 141 01100001
00008105:  u  75  117 165 01110101
00008106:  l  6C  108 154 01101100
00008107:  t  74  116 164 01110100
00008108:  -  2D  045 055 00101101
00008109:  s  73  115 163 01110011
00008110:  r  72  114 162 01110010
00008111:  c  63  099 143 01100011
00008112:     C2  194 302 11000010
00008113:     A0  160 240 10100000

Striated

Answer 8

Use open(fn, 'rb').read().decode('utf-8') instead of just open(fn).read()

Silber

Answer 9

You can try this before using job_titles string:

source = unicode(job_titles, 'utf-8')

Philana

For Python2 or Python3 as well? – Canfield

Answer 10

For python 3, the default encoding would be "utf-8". Following steps are suggested in the base documentation:https://docs.python.org/2/library/csv.html#csv-examples in case of any problem

1. Create a function

   def utf_8_encoder(unicode_csv_data):
       for line in unicode_csv_data:
           yield line.encode('utf-8')
   

2. Then use the function inside the reader, for e.g.

   csv_reader = csv.reader(utf_8_encoder(unicode_csv_data))
   

Folia

Answer 11

python3x or higher

1. load file in byte stream:

     body = ''
        for lines in open('website/index.html','rb'):
            decodedLine = lines.decode('utf-8')
            body = body+decodedLine.strip()
        return body

2. use global setting:

    import io
    import sys
    sys.stdout = io.TextIOWrapper(sys.stdout.buffer,encoding='utf-8')

Stromberg

Answer 12

It's better to use the encoding="utf-8" as the parameter while using open(). In order to make the Python (verified in 3.x) detect and use the utf-8 encoding, following steps may be followed (run them as root/sudo) if locale is not already set.

1. Run apt-update && apt install -y locales
2. Edit the file /etc/locale.gen and uncomment the line en_US.UTF-8 UTF-8
3. Run locale-gen && update-locale LANG=en_US.UTF-8 LC_ALL=en_US.UTF-8

Now try running the python script again and it should pickup the UTC-8 locale (unless some part of your code is explicitly overriding it).

NOTE: If you do not want en_US.UTF-8, you may skip step 2 and just run update-locale LANG=C.UTF-8 LC_ALL=C.UTF-8

Cavite