#305 Support for transliteration in console output
Since some days/weeks ago, I can't get non-latin characters to be displayed in the Windows console. Usually, they would be transliterated and in yellow, but now I only see either question marks or blocks (if I use an Unicode font). There is an exception for Greek characters which are displayed correctly if I use an Unicode font. As of now, I can't confirm an interwiki link in Cyrilic simply because I can't see what text is in [[ru:????????]].
I use the cp850 encoding. Using cp65001 is not supported by Pywikipedia.
It would be great to get the transliteration feature back working. For instance, r9974 worked fine.
Comment on config.py:
############## USER INTERFACE SETTINGS ##############
# The encoding that's used in the user's console, i.e. how strings are encoded
# when they are read by raw_input(). On Windows systems' DOS box, this should
# be 'cp850' ('cp437' for older versions). Linux users might try 'iso-8859-1'
# or 'utf-8'.
# This default code should work fine, so you don't have to think about it.
# TODO: consider getting rid of this config variable.
Python 2.7.2 (default, Jun 12 2011, 14:24:46) [MSC v.1500 64 bit (AMD64)]
config-settings:
use_api = True
use_api_login = True
unicode test: ok
Sorry, that was me. I forgot to login.
Set your console character set to Consolas or Lucida instead of raster fonts at the properties, and you will find it better than earlier. :-)
Not really. Lucida Console doesn't show Cyrilic or Georgian or Arabic. AFAICT, it only shows Latin and Greek.
It got better with DejaVu Sans Mono [check http://stackoverflow.com/questions/878972/windows-cmd-encoding-change-causes-python-crash/3259271#3259271\] but I still see some unsupported characters showing up sometimes.
Is there a complete mono font out there (free)?
Try reverting r10048 of terminal_interface_base.py to the previous r10047. I found it helps and gives the right transliterations. Could you verify?
I'm moving this to feature requests with a lower priority. In the current setup, we output the correct characters, but your font is unable to correctly render them. However, copying does work, and I think this is more useful than transliterated characters.
Checking if an interwiki link is correct is not possible using transliterations, but is possible by copy-pasting the unicode characters.
Implemented in r10187. Please see http://www.mediawiki.org/wiki/Manual:Pywikipediabot/Windows on how to set user-config.py.
Please note that the blocks you see are correct -- you can copy them to the browser and visit that page, which is impossible with transliterated characters!