From 7550dfa16407e098966e8e060b7b764b512dbe6c Mon Sep 17 00:00:00 2001
From: Marc Remolt <marc.remolt@googlemail.com>
Date: Sun, 5 Jun 2011 19:14:52 +0200
Subject: [PATCH] First shot at attachment import

* only images for now
* only blog posts for now
* thumbnailed images are not yet replaced in blog posts body
---
 lib/tasks/wordpress.rake              |  7 +++++
 lib/wordpress/attachment.rb           | 43 +++++++++++++++++++++++++++
 spec/lib/wordpress/attachment_spec.rb | 33 ++++++++++++++++++++
 3 files changed, 83 insertions(+)

diff --git a/lib/tasks/wordpress.rake b/lib/tasks/wordpress.rake
index f31f120..44c9bbe 100644
--- a/lib/tasks/wordpress.rake
+++ b/lib/tasks/wordpress.rake
@@ -19,6 +19,7 @@ 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)
@@ -28,6 +29,12 @@ 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"
diff --git a/lib/wordpress/attachment.rb b/lib/wordpress/attachment.rb
index ee5a6b2..f4f898d 100644
--- a/lib/wordpress/attachment.rb
+++ b/lib/wordpress/attachment.rb
@@ -2,6 +2,7 @@ module Refinery
   module WordPress
     class Attachment
       attr_reader :node
+      attr_reader :refinery_image
 
       def initialize(node)
         @node = node
@@ -15,6 +16,10 @@ module Refinery
         node.xpath("description").text
       end
 
+      def file_name
+        url.split('/').last
+      end
+
       def post_date
         DateTime.parse node.xpath("wp:post_date").text
       end
@@ -22,6 +27,44 @@ module Refinery
       def url
         node.xpath("wp:attachment_url").text
       end
+
+      def image?
+        url.match /\.(png|jpg|jpeg|gif)$/ 
+      end
+
+      def to_refinery
+        if image?
+          to_image
+        else
+          to_file
+        end
+      end
+
+      def replace_image_url_in_blog_posts
+        ::BlogPost.all.each do |post|
+          if post.body.include? url
+            post.body = post.body.gsub(url, refinery_image.image.url)
+            post.save!
+          end
+        end
+      end
+
+      private
+
+      def to_image
+        image = ::Image.new
+        image.created_at = post_date
+        image.image_url = url
+        image.save!
+
+        @refinery_image = image
+        image
+      end
+
+      def to_file
+        raise "to_file is not implemented yet, sorry!"
+      end
+
     end
   end
 end
diff --git a/spec/lib/wordpress/attachment_spec.rb b/spec/lib/wordpress/attachment_spec.rb
index 002a39b..8221a0b 100644
--- a/spec/lib/wordpress/attachment_spec.rb
+++ b/spec/lib/wordpress/attachment_spec.rb
@@ -6,5 +6,38 @@ describe Refinery::WordPress::Attachment, :type => :model do
   specify { attachment.title.should == '200px-Tux.svg' }
   specify { attachment.description.should == 'Tux, the Linux mascot' }
   specify { attachment.url.should == 'http://localhost/wordpress/wp-content/uploads/2011/05/200px-Tux.svg_.png' }
+  specify { attachment.file_name.should == '200px-Tux.svg_.png' }
   specify { attachment.post_date.should == DateTime.new(2011, 6, 5, 15, 26, 51) }
+  specify { attachment.should be_an_image }
+
+  describe "#to_refinery" do
+    before do
+      @image = attachment.to_refinery
+    end
+
+    it "should create an Image from the Attachment" do
+      @image.should be_a(Image)
+    end
+
+    it "should copy the attributes from Attachment" do
+      @image.created_at.should == attachment.post_date
+      @image.image.url.end_with?(attachment.file_name).should be_true
+    end
+  end
+
+  describe "#replace_image_url" do
+    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
+    end
+
+    it "should replace attachment urls in the generated BlogPosts" do
+      BlogPost.first.body.should_not include(attachment.url)
+      BlogPost.first.body.should include(@image.image.url)
+    end
+
+  end
 end