Re: completely rewind an HTML::TokeParser (PullParser)

D'oh, sorry, just realized that proposed sub change is useless. So, 
please ignore that part. Still looking for ideas to make the rewind 
work though.

On Wednesday, December 21, 2005, at 11:23  AM, Ashley Pond V wrote:

> I want to parse out some HTML to count the text characters and then 
> rewind to the beginning of the parser to step through tags and excerpt 
> to an argument supplied character or percentage point in the *text* 
> chars and then, keeping track of what tags are open, automatically add 
> close tags.
>
> To be used in a TT2 filter like "truncate" but for HTML instead of 
> plain text. So that...
>
>  [% html = "<p><b>this <i>is something to truncate</i></b></p>" %]
>  [% html | truncate_html(10) %]
>
> Would output
>  <p><b>this <i>is so...</i></b></p>
>  ^^^^^^12345^^^67890
>
> Anyway, the trouble I'm trying to address is rewinding the parser for 
> the second walkthrough once I've counted text characters.
>
> I want to do this: 1 while $p->unget_token()
>
> But it doesn't work because HTML::PullParser->unget_token returns the 
> parser object itself.
>
> To make a long story short, and it's not too late for that -- would 
> changing the return be reasonable or would it break code?
>
>  sub unget_token
>  {
>      my $self = shift;
>      unshift @{$self->{pullparser_accum}}, @_;
>  #    $self; <-- change, don't return $self anymore
>  }
>
> If not, does anyone have a smart idea for how to rewind it while 
> respecting the interface (ie, not testing $self->{pullparser_accum})?
>
> Thanks!
> -Ashley