From da8e72bf808af9d21ecd9b96f7daa5eaf9afb4f2 Mon Sep 17 00:00:00 2001
From: Jesse Luehrs <doy@tozt.net>
Date: Sat, 2 May 2009 15:10:08 -0500
Subject: update it for things i don't remember at the moment

---
 lib/IkiWiki/Plugin/highlight.pm | 71 ++++++++++++++++++++++++++++-------------
 1 file changed, 49 insertions(+), 22 deletions(-)

diff --git a/lib/IkiWiki/Plugin/highlight.pm b/lib/IkiWiki/Plugin/highlight.pm
index f2de12d..860b9c3 100644
--- a/lib/IkiWiki/Plugin/highlight.pm
+++ b/lib/IkiWiki/Plugin/highlight.pm
@@ -7,7 +7,12 @@ use Syntax::Highlight::Engine::Kate;
 
 # variables {{{
 my %hl_args = (
-    substitutions => { },
+    substitutions => {
+        '<' => '&lt;',
+        '>' => '&gt;',
+        '&' => '&amp;',
+        '[' => '&#91;', # wow is this a huge hack
+    },
     format_table  => {
         Alert        => ["<span class='synAlert'>",        "</span>"],
         BaseN        => ["<span class='synBaseN'>",        "</span>"],
@@ -31,35 +36,47 @@ my %hl_args = (
         Warning      => ["<span class='synWarning'>",      "</span>"],
     },
 );
-my %syntaxes;
+my $syntaxes;
 # }}}
 
 sub import { # {{{
-    hook(type => 'sanitize', id => __PACKAGE__, call => \&sanitize, first => 1);
+    my $hl = Syntax::Highlight::Engine::Kate->new(%hl_args);
+    $syntaxes = { map { +lc($_) => $hl->syntaxes->{$_} }
+                      keys %{ $hl->syntaxes } };
+    $syntaxes->{c}     = 'ANSI C89';
+    $syntaxes->{cpp}   = 'C++';
+    $syntaxes->{ocaml} = 'Objective Caml';
+    hook(type => 'filter', id => __PACKAGE__, call => \&filter);
 } # }}}
 
-sub sanitize { # {{{
+sub filter { # {{{
     my %args = @_;
     my $content = '';
+    my $code_block = '';
     my $in_code = 0;
+    my $code_re = qr/^(?: {4}|\t)/;
+    my $empty_re = qr/^\s*$/;
     for my $line (split /^/m, $args{content}) {
         if ($in_code) {
-            if ($line =~ s{(</code></pre>.*)}{}s) {
-                my $rest_line = $1;
+            if ($line =~ $code_re || $line =~ $empty_re) {
+                $line =~ s/^(?: ? ? ? ?|\t)//;
                 $code_block .= $line;
-                $content .= highlight($code_block) . $rest_line;
-                $code_block = '';
-                $in_code = 0;
             }
             else {
-                $code_block .= $line;
+                $code_block =~ s/(\s*)$//;
+                my $ws = $1;
+                $content .= highlight($code_block);
+                $in_code = 0;
+                $code_block = '';
+                $content .= $ws;
+                $content .= $line;
             }
         }
         else {
-            if ($line =~ s{(.*<pre><code>)}{}s) {
-                $content .= $1;
-                $code_block .= $line;
+            if ($line =~ $code_re && $line !~ $empty_re) {
                 $in_code = 1;
+                $line =~ s/^(?: ? ? ? ?|\t)//;
+                $code_block .= $line;
             }
             else {
                 $content .= $line;
@@ -73,21 +90,31 @@ sub sanitize { # {{{
 sub highlight { # {{{
     my ($code_block) = @_;
     my $filetype = guess_filetype($code_block);
-    return unless $filetype;
-    my $hl = Syntax::Highlight::Engine::Kate->new(
-        language => $filetype, %hl_args
-    );
-    %syntaxes ||= map { lc($_) => $hl->syntaxes->{$_} } keys %$hl->syntaxes;
-    return $hl->highlightText($code_block);
+    return $code_block unless $filetype;
+    my $hl = Syntax::Highlight::Engine::Kate->new(%hl_args,
+                                                  language => $filetype);
+    $code_block =~ s/[^\n]*\n//s;
+    $filetype =~ s/\s/_/g;
+    return "<pre><code class='lang$filetype'>" .
+           $hl->highlightText($code_block) .
+           '</code></pre>';
 } # }}}
 
 sub guess_filetype { # {{{
     my ($code_block) = @_;
     my ($first_line) = split /\n/, $code_block;
-    if ($first_line =~ /^#!(\w+)/) {
-        return $syntaxes{$1};
+    if ($first_line =~ m{^#!\s*(?:.+[\s/]+)*(\w+)}) {
+        return $syntaxes->{$1};
+    }
+    elsif ($first_line =~ m{^[/(]\*\s*(\w+)\s*\*[/)]}) {
+        return $syntaxes->{$1};
+    }
+    elsif ($first_line =~ m{^[-/]{2}\s*(\w+)}) {
+        return $syntaxes->{$1};
+    }
+    elsif ($first_line =~ m{^<!--\s*(\w+)\s*-->}) {
+        return $syntaxes->{$1};
     }
-    return;
 } # }}}
 
 1;
-- 
cgit v1.2.3