UnicodeDecodeError: 'ascii' codec can't decode byte 0xd1 in position 2: ordinal not in range(128)
Question:
I am attempting to work with a very large dataset that has some non-standard characters in it. I need to use unicode, as per the job specs, but I am baffled. (And quite possibly doing it all wrong.)
I open the CSV using:
15 ncesReader = csv.reader(open('geocoded_output.csv', 'rb'), delimiter='t', quotechar='"')
Then, I attempt to encode it with:
name=school_name.encode('utf-8'), street=row[9].encode('utf-8'), city=row[10].encode('utf-8'), state=row[11].encode('utf-8'), zip5=row[12], zip4=row[13],county=row[25].encode('utf-8'), lat=row[22], lng=row[23])
I’m encoding everything except the lat and lng because those need to be sent out to an API. When I run the program to parse the dataset into what I can use, I get the following Traceback.
Traceback (most recent call last):
File "push_into_db.py", line 80, in <module>
main()
File "push_into_db.py", line 74, in main
district_map = buildDistrictSchoolMap()
File "push_into_db.py", line 32, in buildDistrictSchoolMap
county=row[25].encode('utf-8'), lat=row[22], lng=row[23])
UnicodeDecodeError: 'ascii' codec can't decode byte 0xd1 in position 2: ordinal not in range(128)
I think I should tell you that I’m using python 2.7.2, and this is part of an app build on django 1.4. I’ve read several posts on this topic, but none of them seem to directly apply. Any help will be greatly appreciated.
You might also want to know that some of the non-standard characters causing the issue are Ñ and possibly É.
Answers:
Unicode is not equal to UTF-8. The latter is just an encoding for the former.
You are doing it the wrong way around. You are reading UTF-8-encoded data, so you have to decode the UTF-8-encoded String into a unicode string.
So just replace .encode with .decode, and it should work (if your .csv is UTF-8-encoded).
Nothing to be ashamed of, though. I bet 3 in 5 programmers had trouble at first understanding this, if not more 😉
Update:
If your input data is not UTF-8 encoded, then you have to .decode() with the appropriate encoding, of course. If nothing is given, python assumes ASCII, which obviously fails on non-ASCII-characters.
Just add this lines to your code:
1.Python2
import sys
reload(sys)
sys.setdefaultencoding('utf-8')
2.Python3
import sys
from importlib import reload
reload(sys)
sys.setdefaultencoding('utf-8')
The main reason for the error is that the default encoding assumed by python is ASCII.
Hence, if the string data to be encoded by encode('utf8') contains character that is outside of ASCII range e.g. for a string like ‘hgvcj터파크387’, python would throw error because the string is not in the expected encoding format.
If you are using python version earlier than version 3.5, a reliable fix would be to set the default encoding assumed by python to utf8:
import sys
reload(sys)
sys.setdefaultencoding('utf8')
name = school_name.encode('utf8')
This way python would be able to anticipate characters within a string that fall outside of ASCII range.
However, if you are using python version 3.5 or above, reload() function is not available, so you would have to fix it using decode e.g.
name = school_name.decode('utf8').encode('utf8')
for Python 3 users. you can do
with open(csv_name_here, 'r', encoding="utf-8") as f:
#some codes
it works with flask too 🙂
For Python 3 users:
changing the encoding from ‘ascii’ to ‘latin1’ works.
Also, you can try finding the encoding automatically by reading the top 10000 bytes using the below snippet:
import chardet
with open("dataset_path", 'rb') as rawdata:
result = chardet.detect(rawdata.read(10000))
print(result)
open with encoding UTF 16 because of lat and long.
with open(csv_name_here, 'r', encoding="utf-16") as f:
if you get this issue while running certbot while creating or renewing certificate, Please use the following method
grep -r -P '[^x00-x7f]' /etc/apache2 /etc/letsencrypt /etc/nginx
That command found the offending character “´” in one .conf file in the comment. After removing it (you can edit comments as you wish) and reloading nginx, everything worked again.
Check which locale you’re using with the locale command. If it’s not en_US.UTF-8, change it like this:
sudo apt install locales
sudo locale-gen en_US en_US.UTF-8
sudo dpkg-reconfigure locales
If you don’t have permission to do that you can run all your Python code like this:
PYTHONIOENCODING="UTF-8" python3 ./path/to/your/script.py
or run this command before running your Python code
export PYTHONIOENCODING="UTF-8"
to set it in the shell you run that in.
In my case, I was using POSIX, the default Ubuntu locale instead of en_US.UTF-8, so I saw this output:
$ locale
LANG=
LANGUAGE=
LC_CTYPE="POSIX"
LC_NUMERIC="POSIX"
LC_TIME="POSIX"
LC_COLLATE="POSIX"
LC_MONETARY="POSIX"
LC_MESSAGES="POSIX"
LC_PAPER="POSIX"
LC_NAME="POSIX"
LC_ADDRESS="POSIX"
LC_TELEPHONE="POSIX"
LC_MEASUREMENT="POSIX"
LC_IDENTIFICATION="POSIX"
LC_ALL=
which caused Python to open files as ASCII instead of UTF-8.
You can check which locale Python is using like this:
>>> import locale
>>> locale.getpreferredencoding(False)
'ANSI_X3.4-1968'
locale.getpreferredencoding(False) is the function called by open() when you don’t provide an encoding. The output should be 'UTF-8', but in my case it was 'ANSI_X3.4-1968', some variant of ASCII.
Or when you deal with text in Python if it is a Unicode text, make a note it is Unicode.
Set text=u'unicode text' instead just text='unicode text'.
This worked in my case.
It does work by just taking the argument ‘rb’ read binary instead of ‘r’ read
Dealing with this issue inside of a Docker container.
It might be the case (as it was for me) that you only need to generate the locale and do nothing more:
sudo locale-gen en_US en_US.UTF-8
In some case that was sufficient for me because locales was already installed and configured. If you have to install locales and configure it, add the following part to your Dockerfile:
RUN apt update && apt install locales &&
sed -i -e 's/# en_US.UTF-8 UTF-8/en_US.UTF-8 UTF-8/' /etc/locale.gen &&
echo 'LANG="en_US.UTF-8"'>/etc/default/locale &&
dpkg-reconfigure --frontend=noninteractive locales &&
update-locale LANG=en_US.UTF-8
ENV LANG en_US.UTF-8
ENV LANGUAGE en_US.UTF-8
ENV LC_ALL en_US.UTF-8
I tested it like this:
cat <<EOF > /tmp/test.txt
++*=|@#|¼üöäàéàè!´]]¬|¢|¢¬|{ł|¼½{}}
EOF
python3
import pathlib; pathlib.Path("/tmp/test.txt").read_text()
I faced this issue while using Pickle for unloading.
Try,
data = pickle.load(f,encoding='latin1')
I am attempting to work with a very large dataset that has some non-standard characters in it. I need to use unicode, as per the job specs, but I am baffled. (And quite possibly doing it all wrong.)
I open the CSV using:
15 ncesReader = csv.reader(open('geocoded_output.csv', 'rb'), delimiter='t', quotechar='"')
Then, I attempt to encode it with:
name=school_name.encode('utf-8'), street=row[9].encode('utf-8'), city=row[10].encode('utf-8'), state=row[11].encode('utf-8'), zip5=row[12], zip4=row[13],county=row[25].encode('utf-8'), lat=row[22], lng=row[23])
I’m encoding everything except the lat and lng because those need to be sent out to an API. When I run the program to parse the dataset into what I can use, I get the following Traceback.
Traceback (most recent call last):
File "push_into_db.py", line 80, in <module>
main()
File "push_into_db.py", line 74, in main
district_map = buildDistrictSchoolMap()
File "push_into_db.py", line 32, in buildDistrictSchoolMap
county=row[25].encode('utf-8'), lat=row[22], lng=row[23])
UnicodeDecodeError: 'ascii' codec can't decode byte 0xd1 in position 2: ordinal not in range(128)
I think I should tell you that I’m using python 2.7.2, and this is part of an app build on django 1.4. I’ve read several posts on this topic, but none of them seem to directly apply. Any help will be greatly appreciated.
You might also want to know that some of the non-standard characters causing the issue are Ñ and possibly É.
Unicode is not equal to UTF-8. The latter is just an encoding for the former.
You are doing it the wrong way around. You are reading UTF-8-encoded data, so you have to decode the UTF-8-encoded String into a unicode string.
So just replace .encode with .decode, and it should work (if your .csv is UTF-8-encoded).
Nothing to be ashamed of, though. I bet 3 in 5 programmers had trouble at first understanding this, if not more 😉
Update:
If your input data is not UTF-8 encoded, then you have to .decode() with the appropriate encoding, of course. If nothing is given, python assumes ASCII, which obviously fails on non-ASCII-characters.
Just add this lines to your code:
1.Python2
import sys
reload(sys)
sys.setdefaultencoding('utf-8')
2.Python3
import sys
from importlib import reload
reload(sys)
sys.setdefaultencoding('utf-8')
The main reason for the error is that the default encoding assumed by python is ASCII.
Hence, if the string data to be encoded by encode('utf8') contains character that is outside of ASCII range e.g. for a string like ‘hgvcj터파크387’, python would throw error because the string is not in the expected encoding format.
If you are using python version earlier than version 3.5, a reliable fix would be to set the default encoding assumed by python to utf8:
import sys
reload(sys)
sys.setdefaultencoding('utf8')
name = school_name.encode('utf8')
This way python would be able to anticipate characters within a string that fall outside of ASCII range.
However, if you are using python version 3.5 or above, reload() function is not available, so you would have to fix it using decode e.g.
name = school_name.decode('utf8').encode('utf8')
for Python 3 users. you can do
with open(csv_name_here, 'r', encoding="utf-8") as f:
#some codes
it works with flask too 🙂
For Python 3 users:
changing the encoding from ‘ascii’ to ‘latin1’ works.
Also, you can try finding the encoding automatically by reading the top 10000 bytes using the below snippet:
import chardet
with open("dataset_path", 'rb') as rawdata:
result = chardet.detect(rawdata.read(10000))
print(result)
open with encoding UTF 16 because of lat and long.
with open(csv_name_here, 'r', encoding="utf-16") as f:
if you get this issue while running certbot while creating or renewing certificate, Please use the following method
grep -r -P '[^x00-x7f]' /etc/apache2 /etc/letsencrypt /etc/nginx
That command found the offending character “´” in one .conf file in the comment. After removing it (you can edit comments as you wish) and reloading nginx, everything worked again.
Check which locale you’re using with the locale command. If it’s not en_US.UTF-8, change it like this:
sudo apt install locales
sudo locale-gen en_US en_US.UTF-8
sudo dpkg-reconfigure locales
If you don’t have permission to do that you can run all your Python code like this:
PYTHONIOENCODING="UTF-8" python3 ./path/to/your/script.py
or run this command before running your Python code
export PYTHONIOENCODING="UTF-8"
to set it in the shell you run that in.
In my case, I was using POSIX, the default Ubuntu locale instead of en_US.UTF-8, so I saw this output:
$ locale
LANG=
LANGUAGE=
LC_CTYPE="POSIX"
LC_NUMERIC="POSIX"
LC_TIME="POSIX"
LC_COLLATE="POSIX"
LC_MONETARY="POSIX"
LC_MESSAGES="POSIX"
LC_PAPER="POSIX"
LC_NAME="POSIX"
LC_ADDRESS="POSIX"
LC_TELEPHONE="POSIX"
LC_MEASUREMENT="POSIX"
LC_IDENTIFICATION="POSIX"
LC_ALL=
which caused Python to open files as ASCII instead of UTF-8.
You can check which locale Python is using like this:
>>> import locale
>>> locale.getpreferredencoding(False)
'ANSI_X3.4-1968'
locale.getpreferredencoding(False) is the function called by open() when you don’t provide an encoding. The output should be 'UTF-8', but in my case it was 'ANSI_X3.4-1968', some variant of ASCII.
Or when you deal with text in Python if it is a Unicode text, make a note it is Unicode.
Set text=u'unicode text' instead just text='unicode text'.
This worked in my case.
It does work by just taking the argument ‘rb’ read binary instead of ‘r’ read
Dealing with this issue inside of a Docker container.
It might be the case (as it was for me) that you only need to generate the locale and do nothing more:
sudo locale-gen en_US en_US.UTF-8
In some case that was sufficient for me because locales was already installed and configured. If you have to install locales and configure it, add the following part to your Dockerfile:
RUN apt update && apt install locales &&
sed -i -e 's/# en_US.UTF-8 UTF-8/en_US.UTF-8 UTF-8/' /etc/locale.gen &&
echo 'LANG="en_US.UTF-8"'>/etc/default/locale &&
dpkg-reconfigure --frontend=noninteractive locales &&
update-locale LANG=en_US.UTF-8
ENV LANG en_US.UTF-8
ENV LANGUAGE en_US.UTF-8
ENV LC_ALL en_US.UTF-8
I tested it like this:
cat <<EOF > /tmp/test.txt
++*=|@#|¼üöäàéàè!´]]¬|¢|¢¬|{ł|¼½{}}
EOF
python3
import pathlib; pathlib.Path("/tmp/test.txt").read_text()
I faced this issue while using Pickle for unloading.
Try,
data = pickle.load(f,encoding='latin1')