Syntax higlighting improvements

[catagits/Gitalist.git] / lib / Gitalist / View / SyntaxHighlight.pm

diff --git a/lib/Gitalist/View/SyntaxHighlight.pm b/lib/Gitalist/View/SyntaxHighlight.pm
index 5e90ed8..de745ab 100644 (file)
--- a/lib/Gitalist/View/SyntaxHighlight.pm
+++ b/lib/Gitalist/View/SyntaxHighlight.pm
@@ -1,52 +1,26 @@
 package Gitalist::View::SyntaxHighlight;
 use Moose;
-use Gitalist; # ->path_to
 use namespace::autoclean;
 
 extends 'Catalyst::View';
 
 use Syntax::Highlight::Engine::Kate ();
-use Syntax::Highlight::Engine::Kate::Perl ();
 
 use HTML::Entities qw(encode_entities);
 
-=begin
-
-B<Notes>
-
-What should be done, but isn't currently:
-
- broquaint> Another Cat question - if I want to have arbitrary things highlighted is pushing things through a View at all costs terribly wrong?
- broquaint> e.g modifying this slightly to highlight anything (or arrays of anything) http://github.com/broquaint/Gitalist/blob/a7cc1ede5f9729465bb53da9c3a8b300a3aa8a0a/lib/Gitalist/View/SyntaxHighlight.pm
-       t0m> no, that's totally fine.. I'd tend to push the rendering logic into a model, so you end up doing something like: $c->model('SyntaxDriver')->highlight_all($stuff, $c->view('SyntaxHighlight'));
- broquaint> I'm thinking it's a bad idea because the Controller needs to munge data such that the View knows what to do
- broquaint> You just blew my mind ;)
-       t0m> ^^ That works _much_ better if you split up your view methods into process & render..
-       t0m> ala TT..
-       t0m> i.e. I'd have 'highlight this scalar' as the ->render method in the view..
-       t0m> And then the 'default' thing (i.e. process method) will do that and shove the output in the body..
-       t0m> but then you can write foreach my $thing (@things) { push(@highlighted_things, $c->view('SyntaxHighlight')->render($thing)); }
-       t0m> and then I'd move that ^^ loop down into a model which actually knows about / abstracts walking the data structures concerned..
-       t0m> But splitting render and process is the most important bit.. :) Otherwise you need to jump through hoops to render things that don't fit 'nicely' into the bits of stash / body that the view uses by 'default'
-       t0m> I wouldn't kill you for putting the structure walking code in the view given you're walking simple arrays / hashes.. It becomes more important if you have a more complex visitor..
-       t0m> (I use Visitor in the design patterns sense)
-       t0m> As the visitor is responsible for walking the structure, delegating to the ->render call in the view which is responsible for actually mangling the content..
-
-=cut
-
 sub process {
     my($self, $c) = @_;
 
-    for($c->stash->{blobs} ? @{$c->stash->{blobs}} : $c->stash->{blob}) {
-        $_ = $self->highlight($c->stash->{language} => $_);
-    }
-
-    $c->forward('View::Default');
+    $c->res->body($self->render($c, $c->res->body, $c->stash));
 }
 
-# XXX This takes for freakin' ever on big merges. A cache may be needed.
-sub highlight {
-    my($self, $lang, $blob) = @_;
+sub render {
+    my ($self, $c, $blob, $args) = @_;
+
+    # Don't bother with anything over 64kb, it'll be tragically slow.
+    return encode_entities $blob if length $blob > 65536;
+
+    my $lang = $args->{language};
 
     my $ret;
     if($lang) {
@@ -88,3 +62,29 @@ sub highlight {
 }
 
 __PACKAGE__->meta->make_immutable;
+
+__END__
+
+=head1 NAME
+
+Gitalist::View::SyntaxHighlight - Responsible for syntax highlighting code
+
+=head1 DESCRIPTION
+
+Catalyst View for Syntax highlighting.
+
+=head1 METHODS
+
+=head2 process
+
+=head2 highlight
+
+=head1 AUTHORS
+
+See L<Gitalist> for authors.
+
+=head1 LICENSE
+
+See L<Gitalist> for the license.
+
+=cut