How do I resolve difficulties with decoding and printing Greek characters using Python?

I am creating a simple game designed to prompt the user for the Greek translation of an English word. For example:

cow: # here, the gamer would answer with *η αγελάδα* in order to score one point.

I use a helper function to read and decode from a txt file. I do so using the following code in said function:

# The variable filename refers to my helper function's sole parameter, it takes the 
# above mentioned txt file as an argument.
words_text = codecs.open(filename, 'r', 'utf-8')

This helper function then reads each line. The lines resemble something like this:

# In stack data, when I debug, it reads as u"\η αγελάδα - cow\r\n".
u"\u03b7 \u03b1\u03b3\u03b5\u03bb\u03ac\u03开发者_开发技巧b4\u03b1 - cow\r\n"

The first line of the file when read, however, has an unwanted prefix, ueff-:

# u"\ufeffη αγελάδα - cow\r\n"
u"\ufeff\u03b7 \u03b1\u03b3\u03b5\u03bb\u03ac\u03b4\u03b1 - cow\r\n"

Note: After reviewing Mark's answer, I found out that the prepended oject (ueff) was a BOM signature (it is used to distinguish UTF-8 from other encodings).

It's a minor issue and I am not sure how to remove it in the tidiest of manners. Anyways, my helper function then creates and returns a new dictionary which looks something like this:

{u'\u03b7 \u03b1\u03b3\u03b5\u03bb\u03ac\u03b4\u03b1': 'cow'}

Then, in my main function, I use the following in order to store the user's input:

# This is the code for the prompt I noted at the beginning.
# The variable gr_en_dict is the dictionary noted right above.
for key in gr_en_dict:
    user_reply = raw_input('%s: ' % (gr_en_dict[key])).decode(sys.stdout.encoding)

I then compare the value of the user's input with the appropriate key in the dictionary:

# I imported unicodedata as ud.
if ud.normalize('NFC', user_reply) == ud.normalize('NFC', key):
        score += 1

In a response to a question similar to mine, the user ΤΖΩΤΖΙΟΥ said to import the module unicodedata and to call the normalize method (which I did in the code above), but I suspect that might not be necessary. Unfortunately, this step of the program is of no concern just yet because I have a problem decoding the user's input. To demonstrate, when I print the canonical string representation of user_reply and that of the corresponding key in my dictionary [using the built-in repr()] I get the following result:

user's input (user_reply):

u'? \u03b1?\u03b5??\u03b4\u03b1'

If I print the user's input without the repr() function, it looks like this:

? α?ε??δα

key in my dictionary:

u'\u03b7 \u03b1\u03b3\u03b5\u03bb\u03ac\u03b4\u03b1'

If I print it without repr(), I get an error:

UnicodeEncodeError: 'charmap' codec can't encode character u'\u03b7' in position 0: character maps to <undefined>

Notice the question marks in the user's input and the error I get when I try to print the Greek word proper. This seems to be the crux of my problem.

So, what exactly do I need to do in order to decode the user's input and to display all Greek characters properly?

When using my native code page:

C:\>chcp
Active code page: 437

C:\>\python25\python
Python 2.5.4 (r254:67916, Dec 23 2008, 15:10:54) [MSC v.1310 32 bit (Intel)] on
win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import sys
>>> sys.stdout.encoding
'cp437'
>>> print '? α?ε??δα'
? α?ε??δα
>>>

When using the Greek code page: (strangely, it appears correctly only when I copy it to clipboard first and then paste it into a word type application. I would post an image of the what it actually prints in default console, but I lack the reputation to do so.)

C:\>chcp 869
Active code page: 869

C:\>\python25\python
Python 2.5.4 (r254:67916, Dec 23 2008, 15:10:54) [MSC v.1310 32 bit (Intel)] on
win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import sys
>>> sys.stdout.encoding
'cp869'
>>> print ' η αγελάδα'
 η αγελάδα
>>> print 'η αγελάδα'
η αγελάδα
>>>

UP: I had to change default console's font to Lucida Console. That solved my discrepancy.

For part of your question, use:

words_text = codecs.open(filename, 'r', 'utf-8-sig')

and it will handle processing the byte-order-mark of \ufeff.

Technically, this:

user_reply = raw_input('%s: ' % (gr_en_dict[key])).decode(sys.stdout.encoding)

should be:

user_reply = raw_input('%s: ' % (gr_en_dict[key])).decode(sys.stdin.encoding)

but in practice they should be the same encoding.

I believe the problem is the encoding in your default console does not support all Greek characters. When I change to a Greek code page, things begin to work better. Note that I can paste the correct characters into the print statement below, but cp437 doesn't actually support all the characters, so when printed the unsupported characters are replaced with a question mark:

C:\>chcp
Active code page: 437

C:\>python
Python 2.7.1 (r271:86832, Nov 27 2010, 18:30:46) [MSC v.1500 32 bit (Intel)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import sys
>>> sys.stdout.encoding
'cp437'
>>> print 'η αγελάδα - cow'
? α?ε??δα - cow

If I switch to a Greek code page (869 or 1253), it works:

C:\>chcp 869
Active code page: 869

C:\>python
Python 2.7.1 (r271:86832, Nov 27 2010, 18:30:46) [MSC v.1500 32 bit (Intel)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import sys
>>> sys.stdout.encoding
'cp869'
>>> print 'η αγελάδα - cow'
η αγελάδα - cow
>>>

The standard windows shell has issues with extended characters. I would suggest using something like Windows PowerShell.

For the '\ufeff' character, which is the byte order mark, you could perform the following check after reading in the file:

words_text = codecs.open(filename, 'r', 'utf-8')
words_text_lines = words_text.readlines()

if words_text_lines and words_text_lines[0][0]==unicode(codecs.BOM_UTF8, 'utf8'):
    words_text_lines[0] = words_text_lines[0][1:]

That way you're discarding it if it's there.