Erik on Rails

I’m trying to set up a Scanty blog, using Christopher Swenson’s scanty_wordpress_import script to import my wordpress entries. I was getting a strange error, that Google didn’t recognize:

./lib/post.rb:7: undefined method `table_exists?' for Post:Class (NoMethodError)

I found the answer in a few Japanese language posts, but since it was a little difficult to decipher, I’m posting the deets here. The solution is simply to install the schema plugin in the Sequel model (Post, in this case):

class Post < Sequel::Model
  plugin :schema
 
  ...

If you’re getting an “ArgumentError: options :required are unknown” error when trying to use required columns in DataMapper (i.e. property :name, String, :required => true), you need to upgrade your DataMapper to version 0.10.2 or later.

The software I’m writing uses lots of dates, and I got tired of typing Date.parse over and over again, so I tried to write a little module that would let me create dates more simply:

module FancyDates
  (1..12).each do |mo| 
    t = Time.mktime(Date.today.year,mo)
    three_letter = t.strftime("%b")
    t.instance_eval do
      def [](d)
        t = Time.mktime(1999,month,d)
        t.instance_eval do
          def [](y)
            Time.mktime(y,month,day)
          end
        end
        t
      end
    end
 
    const_set(three_letter.to_sym, t)
  end
end

This lets me just type Jan[19][2010] or even Jan[19] and get a date back. It’s sort of messy looking, but I couldn’t think of a cleaner way to do it. If you have any suggestions for refactoring, submit them here.

Update: Pavel Gorbokon posted a much simpler way to do this on RefactorMyCode. Thanks Pavel!

Sometimes in rails I want to validate objects, but only sometimes. Particularly in my specs, sometimes I want to put an object in the database without validating because I know I’m not going to need some of the parameters, and it’d pollute my specs to include them.

Here’s how I do it:

class Outfit < ActiveRecord::Base
 
  def validate
    errors.add_on_empty :pants unless lazy_validation?
  end
 
  def lazy
    dup.instance_eval { @lazy_validation = true; self }
  end
 
  def lazy_validation?
    @lazy_validation == true
  end
end

That way if I use the object normally (Outfit.create!) I’ll get an error, but in my specs I can just do…

it "should be ok" do
  o = Outfit.new.lazy
  o.save!
  o.should be_valid
end

And everything works hunky-dory. You can also do it the other way, where you set the object to be strict, and only validate if you’ve done that.

This took a bit of excessive Googling to figure out, so I thought I’d post a wrap-up here. If you want to create a Rake task that has access to your Rails app’s ActiveModel objects and all the other good Railsy bits, you write it like this:

desc "Prints out the names of all users"
task :print_user_names => :environment do
  User.all.each do |user|
    puts "#{user.first} #{user.last}"
  end
end

If you put that in a .rake file in your RAILS_ROOT/lib/tasks folder, you can run it from your RAILS_ROOT by running “rake print_user_names”.

Easy enough, but the :task_name => :environment bit is a little tricky. But now what if you want to pass some parameters/arguments to your task? Say to print out the name of a specific user? That looks like this:

desc "Look up the name for a particular user id"
task :get_name, :id, :needs => :environment do |task, args|
  user = User.find(args[:id])
  puts "#{user.first} #{user.last}"
end

Then you can look up users by running “rake get_name[14]“, where 14 is the id of the user you’re looking up.

Simple enough but took me a little while to find.