Merge pull request #73 from jakedouglas/fix_ruby_1_8_anchors

TocFilter: keep using the old punctuation stripping on 1.8.7

Author: jch

lib/html/pipeline/toc_filter.rb

@@ -7,11 +7,13 @@ class Pipeline
     # eventually generating the Table of Contents itself, with links
     # to each header
     class TableOfContentsFilter < Filter
+      PUNCTUATION_REGEXP = RUBY_VERSION > "1.9" ? /[^\p{Word}\- ]/u : /[^\w\- ]/
+
       def call
         headers = Hash.new(0)
         doc.css('h1, h2, h3, h4, h5, h6').each do |node|
           name = node.text.downcase
-          name.gsub!(/[^\p{Word}\- ]/u, '') # remove punctuation
+          name.gsub!(PUNCTUATION_REGEXP, '') # remove punctuation
           name.gsub!(' ', '-') # replace spaces with dash
 
           uniq = (headers[name] > 0) ? "-#{headers[name]}" : ''

test/html/pipeline/toc_filter_test.rb

@@ -54,5 +54,5 @@ def test_anchors_with_utf8_characters
                  rendered_h1s[0]
     assert_equal "<h1>\n<a name=\"%D0%A0%D1%83%D1%81%D1%81%D0%BA%D0%B8%D0%B9\" class=\"anchor\" href=\"#%D0%A0%D1%83%D1%81%D1%81%D0%BA%D0%B8%D0%B9\"><span class=\"octicon octicon-link\"></span></a>Русский</h1>",
                  rendered_h1s[1]
-  end
+  end if RUBY_VERSION > "1.9" # not sure how to make this work on 1.8.7
 end