lynx-dev
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: lynx-dev Re: external table filtering (was: TRST : the next step)


From: Klaus Weide
Subject: Re: lynx-dev Re: external table filtering (was: TRST : the next step)
Date: Tue, 23 Nov 1999 04:39:07 -0600 (CST)

On Sat, 20 Nov 1999, Hataguchi Takeshi wrote:

> On Fri, 19 Nov 1999, Klaus Weide wrote:
> > "something that preprocesses
> > 
> >  <TR><TD>Blah<br>more blah</TD><TD>Other blah<br>more other blah</TD></TR>
> > 
> > into
> > 
> >  <TR><TD>Blah             </TD><TD>Other blah                   </TD></TR>
> >  <TR><TD>        more blah</TD><TD>               more otherblah</TD></TR>"?
> 
> Yes and I've tried. Please try this.
> 
>     http://www.t3.rim.or.jp/~patakuti/tmp/t2thtml
> 
> I added two new options: "usepre" and "usetable".
> "usepre" means convert tables to preformatted text as old one does.
> "usetable" means convert tables to tables as you wrote
> (which is fitted for TRST support Lynx)
> 
> I tried this way. I think the output is GREAT!

Again, thank you very much.

Unfortunately, I cannot really agree that the output is that great. :(
In some simple cases yes, but in more general cases it can get much worse
than without *any* table handling.

I have made some more improvements (I hope).  Again I put the updated
version of your script in <http://enteract.com/~kweide/lynx/lxtab/>.
It can now work with nested tables, at least in some cases.

But still, the "usetable" approach isn't quite safe to use, in general.

Well, having played more with t2thtml in various forms now, I must say
I am somewhat less enthusiastic overall.  There are just so many things
that can (and do) go wrong, starting with FORM fields (not that I really
understand what the script tries to do with them, or with IMGs and some
other tags).

To get everything right, the script would have to grow into something
that includes a full (more or less) HTML parser.  That doesn't look
very practicable.  But otherwise there'll always be some constructs that
get mishandled.  (for example, SGML comments with arbitrary contents...
or ">" in attribute values.)

But probably our evolving t2thtml can be improved by upgrading it to
the level of Tom's latest t2tgrnest.  I haven't tried it, or checked
in detail how similar t2thtml and t2tgrnest are but they seem to have
a common base.  So (some of) the t2tgrnest should be backportable to
t2thtml.


   Klaus



reply via email to

[Prev in Thread] Current Thread [Next in Thread]