Resource import working

* added rake tasks

lib/tasks/wordpress.rake

@@ -19,7 +19,6 @@ namespace :wordpress do
     dump = Refinery::WordPress::Dump.new(params[:file_name])
 
     dump.authors.each(&:to_refinery)
-    attachments = dump.attachments.each(&:to_refinery)
     
     only_published = ENV['ONLY_PUBLISHED'] == 'true' ? true : false
     dump.posts(only_published).each(&:to_refinery)
@@ -29,12 +28,6 @@ namespace :wordpress do
     ENV["MODEL"] = 'BlogPost'
     Rake::Task["friendly_id:redo_slugs"].invoke
     ENV.delete("MODEL")
-
-    # parse all created BlogPosts bodys and replace the old wordpress image uls 
-    # with the newly created ones
-    attachments.each do |attachment|
-      attachment.replace_image_url_in_blog_posts
-    end
   end
 
   desc "reset blog tables and then import blog data from a Refinery::WordPress XML dump"
@@ -55,7 +48,7 @@ namespace :wordpress do
     end
   end
 
-  desc "import cms data from a Refinery::WordPress XML dump"
+  desc "import cms data from a WordPress XML dump"
   task :import_pages, :file_name do |task, params|
     Rake::Task["environment"].invoke
     dump = Refinery::WordPress::Dump.new(params[:file_name])
@@ -79,10 +72,50 @@ namespace :wordpress do
     ENV.delete("MODEL")
   end
   
-  desc "reset cms tables and then import cms data from a Refinery::WordPress XML dump"
+  desc "reset cms tables and then import cms data from a 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
+
+
+  desc "Reset the media relevant tables for a clean import"
+  task :reset_media do
+    Rake::Task["environment"].invoke
+
+    %w(images resources).each do |table_name|
+      p "Truncating #{table_name} ..."
+      ActiveRecord::Base.connection.execute "DELETE FROM #{table_name}"
+    end
+  end
+
+  desc "import media data (images and files) from a WordPress XML dump and replace target URLs in pages and posts"
+  task :import_and_replace_media, :file_name do |task, params|
+    Rake::Task["environment"].invoke
+    dump = Refinery::WordPress::Dump.new(params[:file_name])
+    
+    attachments = dump.attachments.each(&:to_refinery)
+    
+    # parse all created BlogPost and Page bodys and replace the old wordpress media uls 
+    # with the newly created ones
+    attachments.each do |attachment|
+      attachment.replace_url
+    end
+  end
+
+  desc "reset media tables and then import media data from a WordPress XML dump"
+  task :reset_import_and_replace_media, :file_name do |task, params|
+    Rake::Task["environment"].invoke
+    Rake::Task["wordpress:reset_media"].invoke
+    Rake::Task["wordpress:import_and_replace_media"].invoke(params[:file_name])
+  end
+
+  desc "reset and import all data (see the other tasks)"
+  task :full_import, :file_name do |task, params|
+    Rake::Task["environment"].invoke
+    Rake::Task["wordpress:reset_and_import_blog"].invoke(params[:file_name])
+    Rake::Task["wordpress:reset_and_import_pages"].invoke(params[:file_name])
+    Rake::Task["wordpress:reset_import_and_replace_media"].invoke(params[:file_name])
+  end
 end

lib/wordpress/attachment.rb

@@ -3,6 +3,7 @@ module Refinery
     class Attachment
       attr_reader :node
       attr_reader :refinery_image
+      attr_reader :refinery_resource
 
       def initialize(node)
         @node = node
@@ -28,6 +29,14 @@ module Refinery
         node.xpath("wp:attachment_url").text
       end
 
+      def url_pattern
+        url_parts = url.split('.')
+        extension = url_parts.pop
+        url_without_extension = url_parts.join('.')
+
+        /#{url_without_extension}(-\d+x\d+)?\.#{extension}/
+      end
+
       def image?
         url.match /\.(png|jpg|jpeg|gif)$/ 
       end
@@ -36,21 +45,15 @@ module Refinery
         if image?
           to_image
         else
-          to_file
+          to_resource
         end
       end
 
-      def replace_image_url_in_blog_posts
+      def replace_url
-        ::BlogPost.all.each do |post|
+        if image?
-          if post.body.include? url
+          replace_image_url
-            url_parts = url.split('.')
+        else
-            extension = url_parts.pop
+          replace_resource_url
-            url_without_extension = url_parts.join('.')
-            pattern = /#{url_without_extension}(-\d+x\d+)?\.#{extension}/
-
-            post.body = post.body.gsub(pattern, refinery_image.image.url)
-            post.save!
-          end
         end
       end
 
@@ -66,8 +69,60 @@ module Refinery
         image
       end
 
-      def to_file
+      def to_resource
-        raise "to_file is not implemented yet, sorry!"
+        resource = ::Resource.new
+        resource.created_at = post_date
+        resource.file_url = url
+        resource.save!
+
+        @refinery_resource = resource
+        resource
+      end
+
+      def replace_image_url
+        replace_image_url_in_blog_posts
+        replace_image_url_in_pages
+      end
+
+      def replace_resource_url
+        replace_resource_url_in_blog_posts
+        replace_resource_url_in_pages
+      end
+
+      def replace_image_url_in_blog_posts
+        replace_url_in_blog_posts(refinery_image.image.url)
+      end
+      
+      def replace_image_url_in_pages
+        replace_url_in_pages(refinery_image.image.url)
+      end
+
+      def replace_resource_url_in_blog_posts
+        replace_url_in_blog_posts(refinery_resource.file.url)
+      end
+      
+      def replace_resource_url_in_pages
+        replace_url_in_pages(refinery_resource.file.url)
+      end
+
+      def replace_url_in_blog_posts(new_url)
+        ::BlogPost.all.each do |post|
+          if (! post.body.empty?) && post.body.include?(url)
+            post.body = post.body.gsub(url_pattern, new_url)
+            post.save!
+          end
+        end
+      end
+
+      def replace_url_in_pages(new_url)
+        ::Page.all.each do |page|
+          page.parts.each do |part|
+            if (! part.body.to_s.blank?) && part.body.include?(url)
+              part.body = part.body.gsub(url_pattern, new_url)
+              part.save!
+            end
+          end
+        end
       end
 
     end

spec/lib/wordpress/attachment_spec.rb

@@ -27,15 +27,16 @@ describe Refinery::WordPress::Attachment, :type => :model do
       end
     end
 
-    describe "#replace_image_url" do
+    describe "#replace_url" do
       let(:post) { BlogPost.first }
 
       before do
         test_dump.authors.each(&:to_refinery)
         test_dump.posts.each(&:to_refinery)
+
         @image = attachment.to_refinery
 
-        attachment.replace_image_url_in_blog_posts
+        attachment.replace_url
       end
 
       specify { post.body.should_not include attachment.url }
@@ -45,8 +46,6 @@ describe Refinery::WordPress::Attachment, :type => :model do
       it "should replace attachment urls in the generated BlogPosts" do
         post.body.should include(@image.image.url)
       end
-
-
     end
   end
 
@@ -60,8 +59,34 @@ describe Refinery::WordPress::Attachment, :type => :model do
     specify { attachment.should_not be_an_image }
 
     describe '#to_refinery' do
-      it "should raise an exception for now" do
+      before do 
-        lambda { attachment.to_refinery }.should raise_error
+        @resource = attachment.to_refinery
+      end
+
+      specify { Resource.should have(1).record }
+      specify { @resource.should be_a(Resource) }
+
+      it "should copy the attributes from Attachment" do
+        @resource.created_at.should == attachment.post_date
+        @resource.file.url.end_with?(attachment.file_name).should be_true
+      end
+
+    end
+
+    describe '#replace_resource_url' do
+      let(:page_part) { Page.last.parts.first }
+
+      before do
+        test_dump.pages.each(&:to_refinery)
+        @resource = attachment.to_refinery
+        attachment.replace_url
+      end
+
+      specify { page_part.body.should_not include attachment.url }
+      specify { page_part.body.should_not include 'wp-content' }
+
+      it "should replace attachment urls in the generated BlogPosts" do
+        page_part.body.should include(@resource.file.url)
       end
     end
   end

spec/spec_helper.rb

@@ -5,18 +5,6 @@ require File.expand_path("../dummy/config/environment.rb",  __FILE__)
 require "rails/test_help"
 require "rspec/rails"
 require "database_cleaner"
-require "fakeweb"
-
-FakeWeb.allow_net_connect = false
-
-# Simulating download of wordpress file attachments. The dump expects the files
-# to be at the given URLs
-FakeWeb.register_uri(:get, 
-  "http://localhost/wordpress/wp-content/uploads/2011/05/200px-Tux.svg_.png", 
-  :body => File.new('spec/fixtures/200px-Tux.svg_.png').read, 
-  :content_type => "image/png")
-
-FakeWeb.register_uri(:get, "http://localhost/wordpress/wp-content/uploads/2011/05/cv.txt", :body => "Hello World!", :content_type => "text/plain")
 
 ActionMailer::Base.delivery_method = :test
 ActionMailer::Base.perform_deliveries = true

spec/support/fakeweb.rb

@@ -0,0 +1,12 @@
+require "fakeweb"
+
+FakeWeb.allow_net_connect = false
+
+# Simulating download of wordpress file attachments. The dump expects the files
+# to be at the given URLs
+FakeWeb.register_uri(:get, 
+  "http://localhost/wordpress/wp-content/uploads/2011/05/200px-Tux.svg_.png", 
+  :body => File.new('spec/fixtures/200px-Tux.svg_.png').read, 
+  :content_type => "image/png")
+
+FakeWeb.register_uri(:get, "http://localhost/wordpress/wp-content/uploads/2011/05/cv.txt", :body => "Hello World!", :content_type => "text/plain")