Just a few thoughts on this:
You could assume that any characters marking a quotation will be non
alphanumeric (makes some things simpler).
But this brings up other issues:
1) how to recognize a single line being quoted (ie the original
message has only one line). I would assume that you would have to have
2+ lines to make a valid assumption that the character(s) you're
looking at are a quotation marker, and not actually part of the text
2) what happens if the original message has some kind of list:
eg - item 1
- item 2
- item 3
This could easily be interpreted as a quotation marker. How to
distinguish?
Definitely not a simple problem....it might come to the point of
a) not doing anything at all....leave things as-is
b) make certain assumptions, causing some mis-quoted messages
c) create a list of standard quotation markers and work from that list
On Thu, 2001-11-15 at 14:36, Brett Johnson wrote:
> On Thu, 2001-11-15 at 11:50, NotZed wrote:
> >
> > Or, we could just use '> ' like most everyone else, and the problem
> > isn't there.
>
> ?? I don't see how this makes the problem go away. Just because
> Evolution decides to use '> ' for quoting doesn't mean that every other
> mail client in the world will suddenly fall into line. So, we still
> need a way to recognize quoted text that doesn't conform to the '> '
> convention.
>
>
> Jim Meyer wrote:
> > > It appears that an poorly worded query on my part spawned an interesting
> > > topic: how can we recognize arbitrarily quoted messages for purposes of
> > > colorization, requoting, and rewrapping?
>
> Personally, I'm partial to the regexp solution (this shouldn't be a
> surprise ;o). Simply use a regular expression to describe the various
> ways that a string can be quoted. Every line that matches said regular
> expression is considered to be an attribution. This works very well in
> my experience (using Balsa and VM under XEmacs). It's even a fairly
> simple process to recognize different levels of quotation (simply remove
> the first part of the line that matched the regexp, then try the match
> again - recurse until it doesn't match anymore).
>
> > > It feels like a potentially worthwhile approach is to spend some time
> > > trying to recognize attribution strings; if we could learn to do that,
> > > we can infer that the next line is quoted somehow, and start to unravel
> > > which quoting strings relate to which attributions.
>
> Wow, that's ambitious. Seems like it could easily be broken as well. Take
> this message for example: NotZed's reply to your message didn't contain an
> attribution at all, so I added one. How would you distinguish between
> NotZed's comments and your own, if I hadn't added the attribution? For
> that matter, how do you distinguish between them now?
>
> > > Ideally, we could completely unroll any number of arbitrarily quoted
> > > messages, which would allow us to then requote or rewrap tidily.
> > >
> > > However, this is no more easy than the first problem, and perhaps
> > > considerably harder. I'm thinking about this a bit in my spare moments,
> > > but to paraphrase ESR, many eyes make light work. ;]
>
> Cheers!
> --
> Brett Johnson <[EMAIL PROTECTED]>
> - i n v e n t -
>
>
> _______________________________________________
> evolution maillist - [EMAIL PROTECTED]
> http://lists.ximian.com/mailman/listinfo/evolution
_______________________________________________
evolution maillist - [EMAIL PROTECTED]
http://lists.ximian.com/mailman/listinfo/evolution
