HTML-Formatting

* added simple_format for content content
* added support for SyntaxHighlighter JS
* finished rake tasks for blog

lib/tasks/wordpress.rake

@@ -1,4 +1,3 @@
-require 'nokogiri'
 require 'wordpress'
 
 namespace :wordpress do
@@ -10,8 +9,8 @@ namespace :wordpress do
       p "Truncating #{table_name} ..."
       ActiveRecord::Base.connection.execute "DELETE FROM #{table_name}"
     end
-  end
 
+  end
 
   desc "import blog data from a Refinery::WordPress XML dump"
   task :import_blog, :file_name do |task, params|
@@ -20,47 +19,18 @@ namespace :wordpress do
 
     dump.authors.each(&:to_refinery)
     dump.posts.each(&:to_refinery)
+
+    ENV["MODEL"] = 'BlogPost'
+    Rake::Task["friendly_id:redo_slugs"].invoke
+    ENV["MODEL"] = nil
   end
 
 
-
+  desc "reset blog tables and then import blog data from a Refinery::WordPress XML dump"
-
+  task :reset_and_import_blog, :file_name do |task, params|
-
-
-
-
-
-  desc "Import data from a Refinery::WordPress XML dump"
-  task :import, :file_name do |task, params|
     Rake::Task["environment"].invoke
+    Rake::Task["wordpress:reset_blog"].invoke
+    Rake::Task["wordpress:import_blog"].invoke(params[:file_name])
   end
 
-  desc "New import (testing)"
-  task :new_import, :file_name do |task, params|
-    Rake::Task["environment"].invoke
-
-    file_name = File.absolute_path(params[:file_name])
-    unless File.file?(file_name) && File.readable?(file_name)
-      raise "Given file '#{file_name}' no file or not readable."
-    end
-
-    dump = Refinery::WordPress::Dump.new(file_name)
-    p dump.authors
-    p dump.pages
-    dump.posts.each do |post|
-      p post.title
-      p post.categories
-      p post.tags
-      p post.creator
-      #p post.content
-    end
-  end
-
-  
-  desc "Import data from a Refinery::WordPress XML dump into a clean database (reset first)"
-  task :import_clean, :file_name do |task, params|
-    Rake::Task["wordpress:reset"].invoke
-    Rake::Task["wordpress:import"].invoke(params[:file_name])
-
-  end
 end

lib/wordpress/page.rb

@@ -1,6 +1,9 @@
 module Refinery
   module WordPress
     class Page
+      include ::ActionView::Helpers::TagHelper
+      include ::ActionView::Helpers::TextHelper
+
       attr_reader :node
 
       def initialize(node)
@@ -19,6 +22,31 @@ module Refinery
         node.xpath("content:encoded").text
       end
 
+      def content_formatted
+        # WordPress doesn't export <p>-Tags, so let's run a simple_format over
+        # the content
+        formatted = simple_format(content)
+
+        # Support for SyntaxHighlighter:
+        # In WordPress you can (via a plugin) enclose code in [lang][/lang]
+        # blocks, which are converted to a <pre>-tag with a class corresponding
+        # to the language.
+        # 
+        # Example:
+        # [ruby]p "Hello World"[/ruby] 
+        # -> <pre class="brush: ruby">p "Hello world"</pre> 
+        formatted.gsub!(/\[(\w+)\]/, '<pre class="brush: \1">')
+        formatted.gsub!(/\[\/\w+\]/, '</pre>')
+
+        # remove all tags inside <pre> that simple_format created
+        # TODO: replace simple_format with a method, that ignores pre-tags
+        formatted.gsub!(/(<pre.*?>)(.+?)(<\/pre>)/m) do |match| 
+          "#{$1}#{strip_tags($2)}#{$3}"
+        end
+          
+        formatted
+      end
+
       def creator
         node.xpath("dc:creator").text
       end
@@ -51,7 +79,7 @@ module Refinery
         page = ::Page.create!(:title => title, :created_at => post_date, 
                               :draft => draft?, :parent_id => parent_id)
 
-        page.parts.create(:title => 'Body', :body => content)
+        page.parts.create(:title => 'Body', :body => content_formatted)
         page
       end
     end

lib/wordpress/post.rb

@@ -31,7 +31,7 @@ module Refinery
         post = BlogPost.new
         
 
-        post = ::BlogPost.create! :title => title, :body => content, :draft => draft?, 
+        post = ::BlogPost.create! :title => title, :body => content_formatted, :draft => draft?, 
           :published_at => post_date, :created_at => post_date, :author => user,
           :tag_list => tag_list
 

spec/lib/wordpress/dump_spec.rb

@@ -94,6 +94,9 @@ describe Refinery::WordPress::Dump, :type => :model do
       it { page.should_not == dump.pages.first }
 
       describe "#to_refinery" do
+        include ::ActionView::Helpers::TagHelper
+        include ::ActionView::Helpers::TextHelper
+
         before do
           # "About me" has a parent page with id 8 in the XML  dump, 
           # would otherwise fails creation
@@ -111,7 +114,7 @@ describe Refinery::WordPress::Dump, :type => :model do
           @page.title.should == page.title
           @page.draft.should == page.draft?
           @page.created_at.should == page.post_date
-          @page.parts.first.body.should == "<p>#{page.content}</p>"
+          @page.parts.first.body.should == "#{simple_format(page.content)}"
         end
       end
     end