We Keep Coding

Unicode with cp1251 and utf-8 on windows

I am playing around with unicode in python.
So there is a simple script:
# -*- coding: cp1251 -*-
print 'юникод'.decode('cp1251')
print unicode('юникод', 'cp1251')
print unicode('юникод', 'utf-8')
In cmd I've switched encoding to Active code page: 1251.
And there is the output:
СЋРЅРёРєРѕРґ
СЋРЅРёРєРѕРґ
юникод
I am a little bit confused.
Since I've specified encoding to cp1251 I expect that it would be decoded correctly.
But as result there is some trash code points were interpreted.
I am understand that 'юникод' is just a bytes like:
'\xd1\x8e\xd0\xbd\xd0\xb8\xd0\xba\xd0\xbe\xd0\xb4'.
But there is a way to get correct output in terminal with cp1251?
Should I build byte string manually?
Seems like I misunderstood something.

I think I can understand what happened to you. The last line gave me the hint, that your trash codepoints confirmed. You try to display cp1251 characters but your editor is configured to use utf8.
The # -*- coding: cp1251 -*- is only used by the Python interpretor to convert characters from source python files that are outside of the ASCII range. And anyway it it is only used for unicode litterals because bytes from original source give er... exactly same bytes in byte strings. Some text editors are kind enough to automagically use this line (IDLE editor is), but I'm little confident in that and allways switch manually to the proper encoding when I use gvim for example. Short story: # -*- coding: cp1251 -*- in unused in your code and can only mislead a reader since it it not the actual encoding.
If you want to be sure of what lies in your source, you'd better use explicit escapes. In code page 1251, this word юникод is composed by those characters: '\xfe\xed\xe8\xea\xee\xe4'
If you write this source:
txt = '\xfe\xed\xe8\xea\xee\xe4'
print txt
print txt.decode('cp1251')
print unicode(txt, 'cp1251')
print unicode(txt, 'utf-8')
and execute it in a console configured to use CP1251 charset, the first three lines will output юникод, and the last one will throw a UnicodeDecodeError exception because the input is no longer valid 'utf8'.
Alternatively, if you find comfortable with you current editor, you could write:
# -*- coding: utf8 -*-
txt = 'юникод'.decode('utf8').encode('cp1251') # or simply txt = u'юникод'.encode('cp1251')
print txt
print txt.decode('cp1251')
print unicode(txt, 'cp1251')
print unicode(txt, 'utf-8')
which should give same results - but now the declared source encoding should be the actual encoding of your python source.
BTW, a Python 3.5 IDLE that natively uses unicode confirmed that:
>>> 'СЋРЅРёРєРѕРґ'.encode('cp1251').decode('utf8')
'юникод'

Your issue is that the encoding declaration is wrong: your editor uses utf-8 character encoding to save the source code. Use # -*- coding: utf-8 -*- to fix it.
>>> u'юникод'
u'\u044e\u043d\u0438\u043a\u043e\u0434'
>>> u'юникод'.encode('utf-8')
'\xd1\x8e\xd0\xbd\xd0\xb8\xd0\xba\xd0\xbe\xd0\xb4'
>>> print _.decode('cp1251') # mojibake due to the wrong encoding
СЋРЅРёРєРѕРґ
>>> print u'юникод'
юникод
Do not use bytestrings ('' literals create bytes object on Python 2) to represent text; use Unicode strings (u'' literals -- unicode type) instead.
If your code uses Unicode strings then a code page that your Windows console uses doesn't matter as long as the chosen font can display the corresponding (non-BMP) characters. See Python, Unicode, and the Windows console
Here's complete code, for reference:
#!/usr/bin/env python
# -*- coding: utf-8 -*-
print(u'юникод')
Note: no .decode(), unicode(). If you are using a literal to create a string; you should use Unicode literals if the string contains text. It is the only option on Python 3 where you can't put non-ascii characters inside a bytes literal and it is a good practice (to use Unicode for text instead of bytestrings) on Python 2 too.
If you are given a bytestring as an input (not literal) by some API then its encoding has nothing to do with the encoding declaration. What specific encoding to use depends on the source of the data.

Just use the following, but ensure you save the source code in the declared encoding. It can be any encoding that supports the characters you want to print. The terminal can be in a different encoding, as long as it also supports the characters you want to print:
#coding:utf8
print u'юникод'
The advantage is that you don't need to know the terminal's encoding. Python will normally1 detect the terminal encoding and encode the print output correctly.
1Unless your terminal is misconfigured.

Encoding Decoding Python

I have text "confrères" in a text file with encoded format "ISO-8859-2". I want to encode this value in "UTF-8" in python.
I used following code in python(2.7) to convert it but the converted value ["confrčres"] is different from original value ["confrères"].
# -*- coding: utf-8 -*-
import chardet
import codecs
a1=codecs.open('.../test.txt', 'r')
a=a1.read()
b = a.decode(chardet.detect(a)['encoding']).encode('utf8')
a1=codecs.open('.../test_out.txt', 'w').write(b)
Any idea how to get actual value but in UTF8 encoded format in the output file.
Thanks

If you know the codec used, don't use chardet. Character detection is never foolproof, the library guessed wrong for your file.
Note that ISO-8859-2 is the wrong codec, as that codec cannot even encode the letter è. You have ISO-8859-1 (Latin-1) or Windows codepage 1252 data instead; è in 8859-1 and cp1252 is encoded to 0xE8, and 0xE8 in 8859-2 is č:
>>> print u'confrčres'.encode('iso-8859-2').decode('iso-8859-1')
confrères
Was 8859-2 perhaps the guess chardet made?
You can use the io library to handle decoding and encoding on the fly; it is the same codebase that handles all I/O in Python 3 and has fewer issues than codecs:
from shutil import copyfileobj
with open('test.txt', 'r', encoding='iso-8859-1') as inf:
with open('test_out.txt', 'w', encoding='utf8') as outf:
copyfileobj(inf, outf)
I used shutil.copyfileobj() to handle the copying across of data.

How to print symbols like ● to files in Python

I'm trying to write the symbol ● to a text file in python. I think it has something to do with the encoding (utf-8). Here is the code:
# -*- coding: utf-8 -*-
outFile = open('./myFile.txt', 'wb')
outFile.write("●")
outFile.close()
Instead of the black "●" I get "â—". How can I fix this?

Open the file using the io package for this to work with both python2 and python3 with encoding set to utf8 for this to work. When printing, When writing, write as a unicode string.
import io
outFile = io.open('./myFile.txt', 'w', encoding='utf8')
outFile.write(u'●')
outFile.close()
Tested on Python 2.7.8 and Python 3.4.2

If you are using Python 2, use codecs.open instead of open and unicode instead of str:
# -*- coding: utf-8 -*-
import codecs
outFile = codecs.open('./myFile.txt', 'wb', 'utf-8')
outFile.write(u"●")
outFile.close()
In Python 3, pass the encoding keyword argument to open:
# -*- coding: utf-8 -*-
outFile = open('./myFile.txt', 'w', encoding='utf-8')
outFile.write("●")
outFile.close()

>>> ec = u'\u25cf' # unicode("●", "UTF-8")
>>> open("/tmp/file.txt", "w").write(ec.encode('UTF-8'))

This should do the trick
# -*- coding: utf-8 -*-
outFile = open('./myFile.txt', 'wb')
outFile.write(u"\u25CF".encode('utf-8'))
outFile.close()
have a look at this

What your program does is to produce an output file in the same encoding as your program editor (the coding at the top does not matter, unless your program editor uses it for saving the file). Thus, if you open myFile.txt with a program that uses the same encoding as your program editor, everything looks fine.
This does not mean that your program works for everybody.
For this, you must do two things. You must first indicate the encoding used for text files on your machine. This is a little hard to detect, but the following should often work:
# coding=utf-8 # Put your editor's encoding here
import codecs
import locale
import sys
# Selection of the first non-None, reasonable encoding:
out_encoding = (locale.getlocale()[1]
or locale.getpreferredencoding()
or sys.stdin.encoding or sys.stdout.encoding
# Default:
or "UTF8")
outFile = codecs.open('./myFile.txt', 'w', out_encoding)
Note that it is very important to specify the right coding on top of the file: this must be your program editor's encoding.
If you know the encoding you want for your output file, you can directly put it in open(). Otherwise, the more general and portable out_encoding expression above should work for most users on most computers (i.e., whatever their encoding of choice is, they should be able to read "●" in the resulting file—assuming their computer's encoding can represent it).
Then you must print a string, not bytes:
outFile.write(u"●")
(note the leading u, meaning "unicode string").
For a deeper understanding of the issues at hand, one of my previous answers should be very helpful: UnicodeDecodeError when redirecting to file.

I'm very sorry, but writing a symbol to a text file without saying what the encoding of the file should be is simply non-sense.
It may not be evident at first sight, but text files are indeed encoded and may be encoded in different ways. If you have only letters (upper and lower case, but not accented oned), digits and simple symbols (everything that has an ASCII code below 128), all should be fine, because ASCII 7 bits is now a standard and in fact those characters have same representation in major encodings.
But as soon as you get true symbols, or accented chars, their representation vary from one encoding to the other. For example, the symbol ● has a UTF-8 representation of (Python coding) : \xe2\x97\x8f. What is worse, it cannot be represented in latin1 (ISO-8859-1) encoding.
Another example is the french e accent aigu : é it is represented in UTF8 as \xc3\xa9 (note 2 bytes), but is represented in Latin1 as \x89 (one single byte)
So I tested your code in my Ubuntu box using a UTF8 encoding and the command
cat myFile.txt ... correctly showed the bullet !
sba#sba-ubuntu:~/stackoverflow$ cat myFile.txt
●sba#sba-ubuntu:~/stackoverflow$
(as you didn't add any newline after the bullet, the prompt immediately follows it)
In conclusion :
Your code correctly writes the bullet to the file in UTF8 encoding. If your system uses natively another encoding (ISO-8859-1 or its variant Windows-1252) you cannot natively convert it because this character simply does not exist in this encodings.
But you can always see it in a text editor that supports different encoding like the excellent vim that exists on all major systems.
Proof of above :
On a Windows 7 computer, I opened a vim window and instructed it to accept utf8 with :set encoding='utf8'. I then pasted original code from OP and saved it to a file foo.py.
I opened a cmd.exe window and executed python foo.py (using a Python 2.7) : it created a file myFile.txt containing the 3 bytes (hexa) : e2 97 8f that is the utf8 representation of the bullet ● (I could confirm it with vim Tools/Hexa convert).
I could even open myFile.txt in idle and actually saw the bullet. Even notepad.exe could show the bullet !
So even on a Windows 7 computer that does not natively accept utf-8, the code from OP correctly generates a text file that when opened with a text editor accepting UTF-8 contains the bullet ●.
Of course, if I try to open myFile.txt with vim in latin1 mode, I get : â—, on a cmd windows with codepage 850, type myFile.txt shows ÔùÅ, and with codepage 1252 (variant of latin1) : â—.
In conclusion original OP code creates a correct utf8 encoded file - it is up to the reading part to interpret correctly utf8.

Encoding error even using the right codec

I want to open files depending on the encoding format, therefore I do the following:
import magic
import csv
i_file = open(filename).read()
mag = magic.Magic(mime_encoding=True)
encoding = mag.from_buffer(i_file)
print "The encoding is ",encoding
Once I know the encoding format, I try to open the file using the right one:
with codecs.open(filename, "rb", encoding) as f_obj:
reader = csv.reader(f_obj)
for row in reader:
csvlist.append(row)
However, I get the next error:
UnicodeEncodeError: 'ascii' codec can't encode character u'\ufeff' in position 0: ordinal not in range(128)
trying to open a csv file which encoding is:
The encoding is utf-16le
The funny part comes here. If utf-16le is replaced by utf-16, the CSV utf-16le file is properly read. However, it is not well read when used in ascii csv files.
What am I doing wrong?

Python 2's csv module doesn't support Unicode. Is switching to Python 3 an option? If not, can you convert the input file to UTF-8 first?
From the docs linked above:
The csv module doesn’t directly support reading and writing Unicode,
but it is 8-bit-clean save (sic!) for some problems with ASCII NUL
characters. So you can write functions or classes that handle the
encoding and decoding for you as long as you avoid encodings like
UTF-16 that use NULs. UTF-8 is recommended.
Quick and dirty example:
with codecs.open(filename, "rb", encoding) as f_obj:
with codecs.open(filename+"u8", "wb", "utf-8") as utf8:
utf8.write(f_obj.read())
with codecs.open(filename+"u8", "rb", "utf-8") as f_obj:
reader = csv.reader(f_obj)
# etc.

This may be a bit useful to you.
Checkout python 2 documentation
https://docs.python.org/2/library/csv.html
Especially this section:
For all other encodings the following UnicodeReader and UnicodeWriter
classes can be used. They take an additional encoding parameter in
their constructor and make sure that the data passes the real reader
or writer encoded as UTF-8:
Look at the bottom of the page!!!!

Python unicode normalization issue

I am trying to convert a file that contains some unicode characters in it and replace it with normal characters. I am facing some problem with that and get the following error.
UnicodeDecodeError: 'utf8' codec can't decode bytes in position 1-3: invalid data
My file looks like below:
ecteDV
ecteBl
agnéto
the code to replace accents is shown below:
#!/usr/bin/python
# -*- coding: utf-8 -*-
import re, sys, unicodedata, codecs
f = codecs.open(sys.argv[1], 'r', 'utf-8')
for line in f:
name = line.lower().strip()
normal = unicodedata.normalize('NFKD', name).encode('ASCII', 'ignore')
print normal
f.close()
Is there a way I can replace all the accents and normalize the contents of the file?

Consider that your file is perhaps not using UTF-8 as the encoding.
You are reading the file with the UTF-8 codec but decoding fails. Check that your file encoding is really UTF-8.
Note that UTF-8 is an encoding out of many, it doesn't mean 'decode magically to Unicode'.
If you don't yet understand what encodings are (as opposed to what Unicode is, a related but separate concept), you need to do some reading:
The Absolute Minimum Every Software Developer Absolutely, Positively Must Know About Unicode and Character Sets (No Excuses!) by Joel Spolsky
The Python Unicode HOWTO
Pragmatic Unicode by Ned Batchelder

Try opening the file with the following code and replace sys.argv[1] with the filename.
import re, sys, unicodedata, codecs
with codecs.open("filename", 'r', 'utf-8') as f:
for line in f:
do something
f.close()