Started on cms page import

* created rake tasks
* still needed: cleanups, more tests and docs

lib/tasks/wordpress.rake

@@ -5,7 +5,8 @@ namespace :wordpress do
   task :reset_blog do
     Rake::Task["environment"].invoke
 
-    %w(taggings tags blog_comments blog_categories blog_categories_blog_posts blog_posts).each do |table_name|
+    %w(taggings tags blog_comments blog_categories blog_categories_blog_posts 
+       blog_posts).each do |table_name|
       p "Truncating #{table_name} ..."
       ActiveRecord::Base.connection.execute "DELETE FROM #{table_name}"
     end
@@ -22,12 +23,13 @@ namespace :wordpress do
     only_published = ENV['ONLY_PUBLISHED'] == 'true' ? true : false
     dump.posts(only_published).each(&:to_refinery)
 
+    Refinery::WordPress::Post.create_blog_page_if_necessary
+
     ENV["MODEL"] = 'BlogPost'
     Rake::Task["friendly_id:redo_slugs"].invoke
     ENV.delete("MODEL")
   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|
     Rake::Task["environment"].invoke
@@ -35,4 +37,45 @@ namespace :wordpress do
     Rake::Task["wordpress:import_blog"].invoke(params[:file_name])
   end
 
+
+  desc "Reset the cms relevant tables for a clean import"
+  task :reset_pages do
+    Rake::Task["environment"].invoke
+
+    %w(page_part_translations page_translations page_parts pages).each do |table_name|
+      p "Truncating #{table_name} ..."
+      ActiveRecord::Base.connection.execute "DELETE FROM #{table_name}"
+    end
+  end
+
+  desc "import cms data from a Refinery::WordPress XML dump"
+  task :import_pages, :file_name do |task, params|
+    Rake::Task["environment"].invoke
+    dump = Refinery::WordPress::Dump.new(params[:file_name])
+
+    only_published = ENV['ONLY_PUBLISHED'] == 'true' ? true : false
+    dump.pages(only_published).each(&:to_refinery)
+
+    # After all pages are persisted we can now create the parent - child
+    # relationships. This is necessary, as WordPress doesn't dump the pages in
+    # a correct order. 
+    dump.pages(only_published).each do |dump_page|
+      page = ::Page.find(dump_page.post_id)
+      page.parent_id = dump_page.parent_id
+      page.save!
+    end
+
+    Refinery::WordPress::Post.create_blog_page_if_necessary
+        
+    ENV["MODEL"] = 'Page'
+    Rake::Task["friendly_id:redo_slugs"].invoke
+    ENV.delete("MODEL")
+  end
+  
+  desc "reset cms tables and then import cms data from a Refinery::WordPress XML dump"
+  task :reset_and_import_pages, :file_name do |task, params|
+    Rake::Task["environment"].invoke
+    Rake::Task["wordpress:reset_pages"].invoke
+    Rake::Task["wordpress:import_pages"].invoke(params[:file_name])
+  end
 end

lib/wordpress/dump.rb

@@ -19,10 +19,13 @@ module Refinery
         end
       end
 
-      def pages
+      def pages(only_published=false)
-        doc.xpath("//item[wp:post_type = 'page']").collect do |page|
+        pages = doc.xpath("//item[wp:post_type = 'page']").collect do |page|
           Page.new(page)
         end
+
+        pages = pages.select(&:published?) if only_published
+        pages
       end
 
       def posts(only_published=false)

lib/wordpress/page.rb

@@ -36,7 +36,6 @@ module Refinery
         # [ruby]p "Hello World"[/ruby] 
         # -> <pre class="brush: ruby">p "Hello world"</pre> 
         formatted.gsub!(/\[(\w+)\](.+?)\[\/\1\]/m, '<pre class="brush: \1">\2</pre>')
-        #formatted.gsub!(/\[\/\w+\]/, '</pre>')
 
         # remove all tags inside <pre> that simple_format created
         # TODO: replace simple_format with a method, that ignores pre-tags
@@ -60,7 +59,8 @@ module Refinery
       end
 
       def parent_id
-        node.xpath("wp:post_parent").text.to_i
+        dump_id = node.xpath("wp:post_parent").text.to_i
+        dump_id == 0 ? nil : dump_id
       end
 
       def status
@@ -80,8 +80,8 @@ module Refinery
       end
 
       def to_refinery
-        page = ::Page.create!(:title => title, :created_at => post_date, 
+        page = ::Page.create!(:id => post_id, :title => title, 
-                              :draft => draft?, :parent_id => parent_id)
+          :created_at => post_date, :draft => draft?)
 
         page.parts.create(:title => 'Body', :body => content_formatted)
         page
@@ -95,7 +95,6 @@ module Refinery
         
         text.gsub!(/\r\n?/, "\n")                    # \r\n and \r -> \n
         text.gsub!(/\n\n+/, "</p>\n\n#{start_tag}")  # 2+ newline  -> paragraph
-        #text.gsub!(/([^\n]\n)(?=[^\n])/, '\1<br />') # 1 newline   -> br
         text.insert 0, start_tag
 
         text.html_safe.safe_concat("</p>")

lib/wordpress/post.rb

@@ -61,6 +61,25 @@ module Refinery
 
         post
       end
+
+      def self.create_blog_page_if_necessary
+        # refinerycms wants a page at /blog, so let's make sure there is one
+        # taken from the original db seeds from refinery-blog
+        unless ::Page.where("link_url = ?", '/blog').exists?
+          page = ::Page.create(
+            :title => "Blog",
+            :link_url => "/blog",
+            :deletable => false,
+            :position => ((::Page.maximum(:position, :conditions => {:parent_id => nil}) || -1)+1),
+            :menu_match => "^/blogs?(\/|\/.+?|)$"
+          )
+
+          ::Page.default_parts.each do |default_page_part|
+            page.parts.create(:title => default_page_part, :body => nil)
+          end
+        end
+      end
+
     end
   end
 end