In Part One of this series, we constructed a "hello world" Rack app, so to speak. In this part, we'll dive right into using the ruby-freshbooks gem and a little metaprogramming to keep things DRY.

Once again, for the entire source code, head to https://github.com/mikepack/freshbooks_on_rack.

Working with the ruby-freshbooks gem

ruby-freshbooks maps API calls in a somewhat object-oriented fashion. You can call #create, #update, #list, etc on a good number of API entities (like time_entry, client and staff). Check the FreshBooks API docs for a full list of available methods. Generally, the methods are called like the following:

connection = FreshBooks::Client.new('youraccount.freshbooks.com', 'yourfreshbooksapitoken')
connection.client.list
connection.staff.get :staff_id => 1

You can authenticate with your API token (as shown above) or OAuth. For instruction on authenticating with OAuth, check the ruby-freshbooks docs.

Now, lets take a look at the full Rack app that simply prints out all the projects for N number of accounts, and totals the number of hours along with the total income.

Again, FBOnRack#call is invoked upon a request to our Rack app. This method is the heart and soul of our app.

fb_on_rack.rb

require 'ruby-freshbooks'

class FBOnRack
  @cachable_entities = ['staff', 'task']

  def initialize
    @connections = [FreshBooks::Client.new('account1.freshbooks.com', 'apitoken1'),
                    FreshBooks::Client.new('account2.freshbooks.com', 'apitoken2')]
  end

  def call(env)
    res = Rack::Response.new
    res.write "<title>FreshBooks on Rack</title>"
    
    @connections.each do |connection|
      connection.project.list['projects']['project'].each do |project|
        res.write "<h1>Project: #{project['name']}</h1>"
        total_income = 0.0
        total_hours = 0.0

        connection.time_entry.list(:project_id => project['project_id'])['time_entries']['time_entry'].each do |entry|
          rate = get_rate(connection, project, entry)
          total_hours += entry['hours'].to_f
          total_income += rate.to_f * entry['hours'].to_f
        end
        res.write "Total hours: #{total_hours}<br />"
        res.write "Total income: #{total_income}<br />"
      end
    end

    res.finish
  end

private

  @cachable_entities.each do |entity_name|
    cache_var = instance_variable_set("@#{entity_name}_cache", {})
    get_entity = lambda do |connection, entity_id|
      if cache_var.has_key?(entity_id) # Check if the entity is already cached
        cache_var[entity_id]
      else
        entity = connection.send(entity_name).get(("#{entity_name}_id").to_sym => entity_id)[entity_name] # Make the API call for whatever entity
        cache_var[entity_id] = entity # Cache the API call
      end
    end
    define_method(("get_#{entity_name}").to_sym, get_entity)
  end

  def get_rate(connection, project, entry)
    case project['bill_method']
    when 'project-rate'
      project['rate']
    when 'staff-rate'
      get_staff(connection, entry['staff_id'])['rate']
    when 'task-rate'
      get_task(connection, entry['task_id'])['rate']
    end
  end
end

Lets break this down a little.

The first thing we should tackle is that strange loop at the start of the private definitions:

@cachable_entities.each do |entity_name|
  cache_var = instance_variable_set("@#{entity_name}_cache", {})
  get_entity = lambda do |connection, entity_id|
    if cache_var.has_key?(entity_id) # Check if the entity is already cached
      cache_var[entity_id]
    else
      entity = connection.send(entity_name).get(("#{entity_name}_id").to_sym => entity_id)[entity_name] # Make the API call for whatever entity
      cache_var[entity_id] = entity # Cache the API call
    end
  end
  define_method(("get_#{entity_name}").to_sym, get_entity)
end

The gist of this is to define a caching mechanism so we're not slamming the FreshBooks API. If we fetch an entity by the entity's ID, cache the result for that ID. Let's break this chunk of code down once we're inside the loop:

cache_var = instance_variable_set("@#{entity_name}_cache", {})

This does what you would expect: it defines an instance variable. @cachable_entities contains two entities we want to cache, staff and task. So in the end we have two class instance variables which act as in-memory cache: @staff_cache = {} and @task_cache = {}.

get_entity = lambda do |connection, entity_id|
  if cache_var.has_key?(entity_id) # Check if the entity is already cached
    cache_var[entity_id]
  else
    entity = connection.send(entity_name).get(("#{entity_name}_id").to_sym => entity_id)[entity_name] # Make the API call for whatever entity
    cache_var[entity_id] = entity # Cache the API call
  end
end

Here we define a closure which will fetch our result either from the cache or make a call to the API. After retrieving our entity from the API, we cache it.

define_method(("get_#{entity_name}").to_sym, get_entity)

Here we define our methods (#get_staff(connection, staff_id) and #get_task(connection, task_id)) as private instance methods. The get_entity parameter here is our lambda closure we defined above.

#get_staff and #get_task are called within our #get_rate method (but could be used elsewhere). #get_rate returns the rate which should be used for a given time entry. Rates can be project-based, staff-based or task-based. We need to find the appropriate rate based on the project['bill_method'].

Modify this code to your needs, restart your Rack server, visit http://localhost:9292/ and you should see all your projects, the total time spent on each and the total income from each.

If you've made it this far, give yourself a pat on the rear because this part in the series is definitely the hardest. Let me know if you have any issues understanding the FBOnRack class above. In Part Three of this series, we'll finish off by deploying to Heroku and baking a cake.

New in Ruby 1.9 is the ability to name capture groups so you don't have to use $1, $2...$n. First a demonstration:

Ruby Pre-1.9

regex = /(\w+),(\w+),(\w+)/
"Mike,Pack,Ruby".match regex
"First Name: #$1"
"Last Name: #$2"
"Favorite Language: #$3"

Ruby 1.9.2

regex = /(?<first_name>\w+),(?<last_name>\w+),(?<favorite_language>\w+)/
m = "Mike,Pack,Ruby".match regex
"First Name: #{m[:first_name]}"
"Last Name: #{m[:last_name]}"
"Favorite Language: #{m[:favorite_language]}"

Note: If you use named groups, Ruby won't process unnammed groups. So the following won't work:

regex = /(?<first_name>\w+),(?<last_name>\w+),(?<favorite_language>\w+),(\w+)/
m = "Mike,Pack,Ruby,Colorado".match regex
"First Name: #{m[:first_name]}"
"Last Name: #{m[:last_name]}"
"Favorite Language: #{m[:favorite_language]}"
"Location: #$4"

Note: Even though Ruby won't populate $4, it will still populate $1, $2 and $3.

This Tuesday's Trick

Perl had named regex groups, now Ruby has them. Naming your regex groups can be extremely helpful, especially when the regex becomes complex. Use 'em.

I love FreshBooks. It makes my time tracking incredibly easy and my invoicing hassle free. That's not all; their website is extremely powerful but feels lightweight and friendly to use. As a software engineer, I really appreciate and expect my invoicing tool to be Web 2.0 and fun.

For it's free account, FreshBooks only allows you to have 3 clients. You can have any number of projects under those 3 clients but they set a cap in hopes you'll pay for their service. Well, I have more than 3 clients and I love freemium. FreshBooks allows you to have any number of freemium accounts. While it's a pain in the ass to have to switch between accounts for invoicing, it's worth the $20/month I'm saving...for now.

Another annoying thing about working with numerous freemium accounts is you can't quickly calculate numbers based on all projects you have. For instance, the total income from all projects or the total projected income for a single month. To remedy this, I wanted to create a lightweight Heroku app which would poll all my FreshBook freemium accounts and calculate some numbers, specifically the total hours spent and the total income for each project. FreshBooks has an API which allows me to do just that. +1 FreshBooks!

In this three-part series we'll build the app from the ground up, starting with Rack and then deploying to Heroku. Lets get started.

For the source of the entire project, head to https://github.com/mikepack/freshbooks_on_rack.

Rails to Rack

One consideration for this little project is the framework to use or if a framework is appropriate at all. I sought the following:

• Rack based...It needs to run on Heroku
• Lightweight...I don't have a model so I don't need MVC

I love frameworks. They make most arduous tasks simple. I'm a Rails programmer but I realize for a project this small, Rails is way overkill. Other potential overkill frameworks include Merb, Sinatra and even Camping. I stumbled upon Waves, a resource oriented, ultra-lightweight framework. Unfortunately, Waves has been dead since 2009. So I decided to ditch the framework and go straight to Rack.

Rack Setup

If you don't have Rack yet, install the gem:

gem install rack

Create a directory for your project:

mkdir freshbooks_on_rack
cd freshbooks_on_rack

Rack expects a configuration file to end in .ru:

config.ru

require 'rack'

You should now be able to run your Rack application from the command line:

rackup

Visit http://localhost:9292/ to see your app in action. It won't do anything yet (it'll throw an error), but the Rack basics are there.

FreshBooks Setup

FreshBooks has a great API. There's also a great corresponding gem which ruby-fies the API responses. The ruby-freshbooks gem is an isomorphic, flexible library with very little dependencies. freshbooks.rb is another FreshBooks API wrapper gem but has one major quip: it uses a global connection so you can't (naturally) pull in information from different FreshBooks accounts.

Heroku uses Bundler so create a Gemfile (you don't need the rack gem in your Gemfile):

Gemfile

source 'http://rubygems.org'
gem 'ruby-freshbooks'

Install the gem with Bundler:

bundle

Creating the Rack App

 Let's create a class which will handle our Rack implementation:

fb_on_rack.rb

require 'ruby-freshbooks'

class FBOnRack
  def call(env)
    res = Rack::Response.new
    res.write '<title>FreshBooks on Rack</title>'
    res.finish
  end
end

All our class does so far is respond to a Rack call with a title tag. I use Rack::Response here to make things a little easier when it comes to the expected return value of our #call method. Take a look at this example for a reference on how to respond without using Rack::Response.

Now lets update our config.ru file to call our FBOnRack class:

config.ru

require 'rack'
require 'fb_on_rack'

run FBOnRack.new

FBOnRack#call will now be invoked when a request is made to our Rack app. Restart your Rack app (you'll need to do this on every code change), visit http://localhost:9292/ and you should see a blank page with a title of "FreshBooks on Rack".

Tada! You've just created a sweet Rack app, you should feel proud. In Part Two of this series, we'll take a look at how to work with the ruby-freshbooks gem. We'll generate simple, yet useful data and get our hands dirty in some metaprogramming.