PATCH: Give filters access to scope objects.

[Bob Aman]

Index: lib/haml/engine.rb
===================================================================
--- lib/haml/engine.rb (revision 606)
+++ lib/haml/engine.rb (working copy)
@@ -590,12 +590,22 @@
raise HamlError.new("Filter \"#{filter}\" is not defined!")
end
else
- filtered = filter.new(@filter_buffer).render
+ if filter.instance_method(:render).arity == 0
+ filtered = filter.new(@filter_buffer).render

- unless filter == Haml::Filters::Preserve
- push_text(filtered.rstrip.gsub("\n", "\n#{' ' *
@output_tabs}"))
+ unless filter == Haml::Filters::Preserve
+ push_text(filtered.rstrip.gsub("\n", "\n#{' ' *
@output_tabs}"))
+ else
+ push_silent("_hamlout.buffer << #{filtered.dump} << \"\\n
\"\n")
+ end
else
- push_silent("_hamlout.buffer << #{filtered.dump} << \"\\n
\"\n")
+ push_silent(
+ "haml_temp = #{filter.name}" +
+ ".new(#{@filter_buffer.inspect})" +
+ ".render(self).rstrip" +
+ ".gsub(\"\\n\", \"\\n#{' ' * @output_tabs}\")\n", true)
+ @precompiled <<
+ "haml_temp = _hamlout.push_script(haml_temp, false, nil)
\n"
end
end

[Bob Aman]

By the way... apologies in advance, since this probably constitutes
"code fisting" (as defined by Zed).

[Mislav Marohnić]

Bob,

It's pretty hard to copy-paste patches from force-wrapped e-mails and apply them. You should have used Pastie.

Other thing: what do half of your patches do?? You could be nice enough to share this little useful info with us.

[Tom Stuart]

On 22 Sep 2007, at 12:52, Mislav Marohnić wrote:
> It's pretty hard to copy-paste patches from force-wrapped e-mails
> and apply them. You should have used Pastie.

I'm guessing that uploading them to http://groups.google.com/group/
haml/files is even better.

[Mislav Marohnić]

On 9/22/07, Tom Stuart <t...@experthuman.com> wrote:

I'm guessing that uploading them to http://groups.google.com/group/
haml/files is even better.

Pastie is better because you can apply patches directly from command-line.

curl pastie.caboo.se/12345.txt | patch -p0

[Brett Rogers]

Or even better:

sake pastie:patch PASTE=12345

Regards,
Brett

[Nathan Weizenbaum]

However you want to upload the patch, Bob, Mislav's right: it's a pain
to apply from one of these emails. Thanks for the patches, though; I
definitely want to include at least the first three. I'm having a little
trouble figuring out the usage of the oneline filter, though; what do
you have in mind for it?

- Nathan

[Bob Aman]

> However you want to upload the patch, Bob, Mislav's right: it's a pain
> to apply from one of these emails. Thanks for the patches, though; I
> definitely want to include at least the first three. I'm having a little
> trouble figuring out the usage of the oneline filter, though; what do
> you have in mind for it?
>
> - Nathan

Yeah, my bad. I didn't notice they all got line-wrapped. I'll go
throw them up on Pastie instead.

Regarding the :oneline filter, the intended use is this scenario:

Most likely:

:oneline
%pre
%code
= preserve("some multi-line code here")

But in general:

:oneline
%div
%span
= "Something that might normally cause line-wrapping in a place
where you don't actually want line-wrapping to occur for whatever
reason."

[Nathan Weizenbaum]

This is an interesting way to solve that problem. Why not go with a
helper function?

- online do
whatever

- Nathan

[Bob Aman]

http://pastie.caboo.se/100417

Also, I updated the patch to allow the ERB filter to take advantage of
this code.

[Bob Aman]

> This is an interesting way to solve that problem. Why not go with a
> helper function?
>
> - online do
> whatever
>
> - Nathan

Because I wanted the stuff within the "oneline" construct to be
specified as Haml. It seemed inelegant to me to have to drop into
Ruby when the content wasn't even dynamic to overcome what I felt was
a minor limitation of the templating system. Seemed to make more
sense to just go ahead and remove the limitation.

For example, what if I had:

:oneline
%pre
%code
:preserve
# This is some static ruby code
# which I want to display for the
# world to see.
puts eval(2 + 2)

Surely you don't disagree that that is a much nicer way of outputting
"<pre><code># This is some static ruby code&#x000A;# which I want to
display for the&#x000A;# world to see.&#x000A;puts eval(2 +
2)&#x000A;</code></pre>\n" than it would be if I tried to use a helper
method to generate that.

[Nathan Weizenbaum]

The thing is, filters exist expressly so that non-Haml syntaxes can be
used from within Haml. Using them to do processing on Haml code is
wrong; not only is it much easier with a helper method, that's what
helper methods are /for/.

Also, could you work up unit tests for the rest of your patches? Thanks.

- Nathan