[IPython-dev] Re: A few remaining coloring bugs?

[Fernando Perez]

Hi everyone,

Gary Bishop has been helping me track down a problem with syntax-highlighted 
source under Windows.  A simple way to see the problem is to do

import code
code??

You should see nicely syntax-highlighted source:

http://ipython.scipy.org/screenshots/syntax_highlight.png

Instead, win32 users get strangely broken lines with incorrect colors and 
visible escape sequences.  Here is some of what Gary has found:

Gary Bishop wrote:
> A little more debugging.
> 
>>From outside my code, ipython calls "write" on my file object. The CR 
> characters are already in the string at that time. So it looks to me 
> like the CR's are coming from your code. So, it looks like they are 
> already in there. This is output from od -c
> 
> 0012720   m  \r  \r  \n 033   [   1   ;   3   2   m   f   r   o   m 033
> 0012740   [   0   m     033   [   1   ;   3   7   m   c   o   d   e   o
> 0012760   p 033   [   0   m     033   [   1   ;   3   2   m   i   m   p
> 0013000   o   r   t 033   [   0   m     033   [   1   ;   3   7   m   C
> 0013020   o   m   m   a   n   d   C   o   m   p   i   l   e   r 033   [
> 0013040   0   m 033   [  \r  \r  \n   1   ;   3   3   m   , 033   [   0
> 0013060   m     033   [   1   ;   3   7   m   c   o   m   p   i   l   e
> 0013100   _   c   o   m   m   a   n   d 033   [   0   m  \r  \r  \n  \r
> 0013120  \r  \n 033   [   1   ;   3   7   m   _   _   a   l   l   _   _
> 0013140 033   [   0   m     033   [   1   ;   3   3   m   = 033   [   0
> 0013160   m     033   [   1   ;   3   3   m   [ 033   [   0   m 033   [

[...]

> No, I think the key problem is the \r's are embedded inside the escape 
> sequence. I have an RE that is scanning for the escape sequences. It 
> doesn't allow embedded \r\n characters. I guess it could. But I'm 
> amazed that escape sequences allow embedded control characters.

Unfortunately, my debugging efforts came to a screeching halt today, as the 
old win32 laptop I was trying to use seems to have finally died.  It has a bad 
screen, and today I just couldn't turn it on anymore.  Since I don't have an 
external screen at home I can plug into it, for now I'm stuck.

But looking at the code, I think I have  a good idea of where the problem 
might be.  I was wondering if I ask anyone on the lists with win32 access to 
help me track this down, by running a simple test for me.

Please replace PyColorize.py in the site-packages/IPython directory with the 
one I'm sending here.  I've basically replaced all explicit '\n' references in 
the tokenizer call to os.linesep.  This might prevent it from inserting things 
in the wrong places, though I'm not 100% sure it can fix it.

If this works, I think we'll finally have an ipython which under win32 would 
really provide _all_ the functionality of the linux version.

Thanks in advance for any help.

Best,

f
-------------- next part --------------
A non-text attachment was scrubbed...
Name: PyColorize.py
Type: text/x-python
Size: 9293 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/ipython-dev/attachments/20050127/93176f20/attachment.py>