From 802c527fcdef77879cd66a547e8b0a7f3f1a21f1 Mon Sep 17 00:00:00 2001
From: Samuel Vasko <samvasko@gmail.com>
Date: Sat, 25 Jan 2014 11:36:53 +0100
Subject: [PATCH] Small formatting fixes

---
 lib/convert.rb | 21 ++++++++++++++++++++-
 lib/fetch.rb   | 16 ++++++++++++++--
 2 files changed, 34 insertions(+), 3 deletions(-)

diff --git a/lib/convert.rb b/lib/convert.rb
index 4cad284..24033ed 100644
--- a/lib/convert.rb
+++ b/lib/convert.rb
@@ -1 +1,20 @@
-require 'eeepub'
\ No newline at end of file
+require 'eeepub'
+
+epub = EeePub.make do
+  title       'Vala Tutorial'
+  creator     'Maciej Piechotka'
+  publisher   'Samuel Vasko'
+  date        '2014-01-25'
+  identifier  'http://example.com/book/foo', :scheme => 'URL'
+  uid         'http://example.com/book/foo'
+
+
+  files ['/path/to/foo.html', '/path/to/bar.html'] # or files [{'/path/to/foo.html' => 'dest/dir'}, {'/path/to/bar.html' => 'dest/dir'}]
+  nav [
+    {:label => '1. foo', :content => 'foo.html', :nav => [
+      {:label => '1.1 foo-1', :content => 'foo.html#foo-1'}
+    ]},
+    {:label => '1. bar', :content => 'bar.html'}
+  ]
+end
+epub.save('sample.epub')
\ No newline at end of file
diff --git a/lib/fetch.rb b/lib/fetch.rb
index 3865a77..36fcb0e 100644
--- a/lib/fetch.rb
+++ b/lib/fetch.rb
@@ -13,10 +13,15 @@ def self.run
         self.transform_code content
         self.absolute_links content
         self.cleanup_paragraphs content
+        self.cleanup_headings content
+
+        ['id', 'lang', 'dir'].each { |e| content[0].remove_attribute(e) }
 
         return content
     end
 
+    private
+
     # Get the goods, sorry for the global
     def self.get_content
         $doc = Nokogiri::HTML(self.download('https://wiki.gnome.org/Projects/Vala/Tutorial'))
@@ -27,6 +32,12 @@ def self.parse_toc content
         content.css('.table-of-contents > ol > li > ol')
     end
 
+    def self.cleanup_headings content
+        content.css('h1, h2, h3, h4, h5, h6').each do |h|
+            h.set_attribute('id', h.attr('id').gsub('?', ''))
+        end
+    end
+
     def self.absolute_links content
         content.css('a').each do |a|
             if a.attr('href').starts_with '/'
@@ -50,13 +61,14 @@ def self.remove_elements content
     def self.transform_code content
         content.css('.highlight').each do |el|
             pre = Nokogiri::XML::Node.new 'pre', $doc
-            pre.content = el.text
+            # cleaning non breaking spaces
+            pre.content = el.text.gsub("\u00A0", ' ')
             el.replace(pre)
         end
 
         content.css('tt.backtick').each do |el|
             code = Nokogiri::XML::Node.new 'code', $doc
-            code.content = el.text
+            code.content = el.text.gsub("\u00A0", ' ')
             el.replace(code)
         end
     end