Correctly handle unicode in anchors.

URL-encode all unicode characters.

Fixes #397

lib/jsduck/guide_anchors.rb

@@ -1,3 +1,5 @@
+require 'cgi'
+
 module JsDuck
 
   # Transforms in-page links so they won't break docs app #!-navigation.
@@ -9,7 +11,7 @@ class GuideAnchors
 
     def self.transform(html, guide_name)
       html.gsub(/(<a\s+(?:[^<>]*\s+)?href=['"]#)([^!\/].*?)(['"])/i) do |m|
-        "#{$1}!/guide/#{guide_name}-section-#{$2}#{$3}"
+        $1 + "!/guide/" + transform_id($2, guide_name) + $3
 
       end.gsub(/(<a\s+(?:[^<>]*\s+)?name=['"])(.*?)(['"])/i) do |m|
         $1 + transform_id($2, guide_name) + $3
@@ -23,6 +25,11 @@ def self.transform_id(id, guide_name)
       if id =~ /^#{guide_name}-section-/
         id
       else
+        # Escape the ID if it's not already escaped.  This check is
+        # needed to avoid re-escaping anchor-links created with
+        # Markdown - these get auto-escaped by RDiscount.
+        id = (id =~ /%[0-9A-F]{2}/) ? id : CGI::escape(id)
+
         "#{guide_name}-section-#{id}"
       end
     end

lib/jsduck/guide_toc.rb

@@ -41,7 +41,7 @@ def self.inject(html, guide_name)
     end
 
     def self.title_to_id(title)
-      title.downcase.gsub(/[^\w]+/, "-")
+      CGI::escape(title.downcase.gsub(/ /, "-"))
     end
 
   end

spec/guide_anchors_spec.rb

@@ -1,3 +1,4 @@
+# -*- coding: utf-8 -*-
 require "jsduck/guide_anchors"
 
 describe JsDuck::GuideAnchors do
@@ -21,6 +22,11 @@ def transform(html)
       "Some\nlong\ntext\nhere...\n\n <a href='#!/guide/myguide-section-blah'>label</a>"
   end
 
+  it "URL-encodes unicode anchors links" do
+    transform("<a href='#fäg'>label</a>").should ==
+      "<a href='#!/guide/myguide-section-f%C3%A4g'>label</a>"
+  end
+
   it "doesn't transform normal links" do
     transform("<a href='http://example.com'>label</a>").should ==
       "<a href='http://example.com'>label</a>"
@@ -41,6 +47,11 @@ def transform(html)
       "<a name='myguide-section-blah'>target</a>"
   end
 
+  it "URL-encodes unicode in anchors" do
+    transform("<a name='fäg'>target</a>").should ==
+      "<a name='myguide-section-f%C3%A4g'>target</a>"
+  end
+
   it "doesn't transform anchors already in target format" do
     transform("<a name='myguide-section-blah'>target</a>").should ==
       "<a name='myguide-section-blah'>target</a>"
@@ -51,6 +62,11 @@ def transform(html)
       "<h1 id='myguide-section-blah'>target</h1>"
   end
 
+  it "URL-encodes unicode in ID-s" do
+    transform("<h1 id='fäg'>target</h1>").should ==
+      "<h1 id='myguide-section-f%C3%A4g'>target</h1>"
+  end
+
   it "doesn't transform ID-s already in target format" do
     transform("<h1 id='myguide-section-blah'>target</h1>").should ==
       "<h1 id='myguide-section-blah'>target</h1>"

spec/guide_toc_spec.rb

@@ -1,3 +1,4 @@
+# -*- coding: utf-8 -*-
 require "jsduck/guide_toc"
 
 describe JsDuck::GuideToc do
@@ -30,13 +31,26 @@ def inject(html)
     EOHTML
   end
 
+  it "URL-encodes unicode in heading ID-s" do
+    inject(<<-EOHTML).should =~ /<h2 id='myguide-section-my-f%C3%A4pter'>My Fäpter/
+      <h2>My Fäpter</h2>
+    EOHTML
+  end
+
   it "links to headings from TOC" do
     inject(<<-EOHTML).should =~ /<a href='#!\/guide\/myguide-section-my-chapter'>/
       <h2>My Chapter</h2>
       <h2>Another Chapter</h2>
     EOHTML
   end
 
+  it "links to unicode headings from TOC" do
+    inject(<<-EOHTML).should =~ /<a href='#!\/guide\/myguide-section-my-f%C3%A4pter'>/
+      <h2>My Fäpter</h2>
+      <h2>Another Fäpter</h2>
+    EOHTML
+  end
+
   it "adds ID-s to H2 headings even when no TOC" do
     inject(<<-EOHTML).should =~ /<h2 id='myguide-section-my-chapter'>My Chapter/
       <h2>My Chapter</h2>