Peter's Blog

Rake is the ruby equivalent of make except it is much better in that you are scripting in ruby and not the nasty primeval make language. Rails uses rake a lot as a simple place to gather together a load of administration tasks such as running tests, running database migrations etc.

When I moved all my drupal blog posts from drupal over to PetersBlogger I decided to use a nice organised rake script rather than do my usual practise of having a directory full of utility scripts.

In my rails installation I created a file called lib/tasks/jobs.rake and in it I put my import code:

   1  task (:import_drupal => :environment) do
   2    print "Importing Drupal\n"
   3    oTagTable = {}
   4    #
   5    # Import all the articles.
   6    #
   7    Drupal::Article.find(:all, :conditions => "type <> 'image'").each do |oArticle|
   8      print "#{oArticle.title}\n"
   9      oTime = Time.at( oArticle.created)
  10      oPost = Post.create( :nid => oArticle.nid,
  11                          :title => oArticle.title,
  12        :body => oArticle.body,
  13        :created => oTime)
  14  
  15      #
  16      # import the tags related to the article.
  17      #
  18      oArticle.tags.each do |oTag|
  19        if oTagTable[oTag.tag]
  20          oNewTag = oTagTable[oTag.tag]
  21        else
  22          oNewTag = Tag.new( :tag=>oTag.tag)
  23          oTagTable[oTag.tag] = oNewTag
  24        end
  25  
  26        oPost.tags << oNewTag
  27      end
  28  
  29      #
  30      # Import the comments for the article.
  31      #
  32      oArticle.comments.each do |oComment|
  33        strWho = oComment.name
  34        strComment = oComment.comment
  35        if strWho == nil or strWho == ""
  36          strWho = "Peter"
  37        end
  38        if strComment == nil or strComment.strip == ""
  39          strComment = "{nothing to say}"
  40        end
  41        oPost.comments << Comment.create( :who => strWho,
  42            :body => strComment,
  43            :created => Time.at( oComment.timestamp), :published => 1)
  44        if not oPost.comments[-1].valid?
  45          p oPost.comments[-1]
  46          p oPost.comments[-1].errors
  47        end
  48      end
  49  
  50      oPost.save!
  51    end
  52  end

This creates a rake task called 'import_drupal' which I can invoke simply by running

rake import_drupal

The task runs with the appropriate rails environment so all my models are in place. I had to create a model to access the records in the drupal database thus:

   1  module Drupal
   2    class Comment < ActiveRecord::Base
   3      establish_connection configurations['drupal']
   4      set_table_name 'comments'
   5    end
   6    class Tag < ActiveRecord::Base
   7      establish_connection configurations['drupal']
   8      set_primary_key 'tid'
   9      has_and_belongs_to_many :articles,
  10        :class_name => 'Drupal::Article',
  11        :join_table => 'awtags_node',
  12        :foreign_key => 'tid',
  13        :association_foreign_key => 'nid'
  14      set_table_name 'awtags'
  15    end
  16    class Article < ActiveRecord::Base
  17      establish_connection configurations['drupal']
  18      set_primary_key 'nid'
  19      self.inheritance_column = 'poopy'
  20      has_many :comments, :table_name => 'comments', :foreign_key => 'nid', :class_name => 'Drupal::Comment'
  21      has_and_belongs_to_many :tags,
  22        :class_name => 'Drupal::Tag',
  23        :join_table => 'awtags_node',
  24        :foreign_key => 'nid',
  25        :association_foreign_key => 'tid'
  26      set_table_name 'node'
  27    end
  28  end

This model handles all the relationships between the records, including the has_and_belongs_to_many relationship between posts and tags (provided by awtags).

I needed an entry in config/database.yml to define the connection to the drupal database:

drupal:
  adapter: mysql
  database: petersblog
  username: secret
  password: doyouthinkidpostithere

and it all works nicely. If I invoke it thusly:

rake import_drupal RAILS_ENV=production

then it will import the posts into the rails production database.

I'm giving away a few of my schema secrets here which pedantic rails developers might leap on, specifically I have used my own fields called 'nid' and 'created' to perform roles that rails already provides 'id' and 'created_on' for. The reason for the duplication was that during development I was having problems setting these fields to the values from the drupal database, rails was insisting on giving them it's own values. By having my own fields I had total control. There are no doubt ways around this but I didn't have time to find them (although I did find other people moaning about the same problem).

And that's how I managed to save 1500 posts from the hell that is php.