Convert a wiki from MediaWiki to Gollum.

README.md

Note that while this will convert a wiki from MediaWiki to Gollum, and links will be preserved, content formatting for the most part will be broken until you set the proper content type for Gollum.

1. Install dependancies:

   gem install hpricot
   gem install gollum
   gem install gollum-lib
   gem install wikicloth

2. Perform a Special:Export

3. Place the XML in a Gollum repo directory, along with the .rb

4. Edit the .rb for your particular needs (.xml path, commit credentials)

5. Run the .rb

   ruby mw-to-gollum.rb

mw-to-gollum.rb

#!/usr/bin/env ruby
require 'rubygems'
require 'hpricot'
require 'gollum'
require 'gollum-lib'

wiki = Gollum::Wiki.new(".")

file = File.open("G.R.A.C.E.-20130904175032.xml", "r")
 
doc = Hpricot(file)

doc.search('/mediawiki/page').each do |el|
  title = el.at('title').inner_text
  content = el.at('text').inner_text
  commit = { :message => "Import MediaWiki page #{title} into Gollum",
             :name => 'Tim Honeywell',
             :email => 'tfhoneywell@gmail.com' }
  begin
    puts "Writing page #{title}"
    wiki.write_page(title, :mediawiki, content, commit)
  rescue Gollum::DuplicatePageError => e
    p "Duplicate #{title}"
  end
end

file.close

For anyone else getting encoding errors: gollum/gollum#843

If you get encoding errors: the problem is indeed the one listed by @davidar (tl;dr: creating a file with accented characters is a problem with Git), but the solution is hard to put in place, and didn't work in my case.

I simply used Ruby's native i18n to remove diacritics. This UTF-supporting (with updated install instructions) version can convert international MediaWiki to Gollum / GitHub wiki pages.