Url conversion heroku app to help a content manager migration

ncea-convert.rb

require 'sinatra'
require 'haml'

class String
  def proper
    self.split(/\s+/).each{ |word| word.capitalize! }.join(' ')
  end
end


helpers do
  def parse(url)
    url.gsub!(/\/$/, '')
    url.gsub!('www.', '')
    url.gsub!('legacy.', '')
    ncea = {}
    if /alignment-standards\.php/i.match(url)
      ncea[:alignment_landing_page] = true
      ncea
    elsif /http:\/\/tki.org.nz\/e\/community\/ncea\/\w+/i.match(url)
      match = /community\/ncea\/((\w+-\w+-\w+|\w+-\w+|\w+))+/i.match(url)
      if match
        ncea[:subject] = match[1].proper
        ncea[:subject] = map_names(ncea[:subject])
      end
      ncea[:aligned] = true unless /community\/ncea\/\\w+-\w+|\w+(-int)/i.match(url)
      match = /-lvl(\d)/i.match(url)
      ncea[:level] = match[1] if match
      if ncea[:subject] == nil
        nil
      else
        ncea
      end
    elsif /http:\/\/tki.org.nz\/e\/community\/ncea$/i.match(url)
      ncea[:root] = true
      ncea
    else
      nil
    end
  end

  def map_names(name)
    name = name.gsub(/-lvl\d/, '').gsub('.php', '').gsub('-int', '')

    map = {
      'Health' => 'Health-education',
      'Agriculture' => 'Agriculture-and-horticultural-science',
      'Space-science' => 'Earth-and-space-science',
      'Space-sci' => 'Earth-and-space-science',
      'Mathematics' => 'Mathematics-and-statistics',
      'Maths' => 'Mathematics-and-statistics',
      'Media' => 'Media-studies',
      'Home-ec' => 'Home-economics',
      'Physed' => 'Physical-education',
      'Graphics' => 'DVC',
      'Socstud' => 'Social-studies',
      'Te-reo' => 'Te-reo-maori',
      'Tech' => 'Technology',
    }

    map.each do |k,v|
      if name == k
        name = v
      end
    end

    name
  end

  def format(ncea)
    status_prefix = {
      nil => '/Resources-for-non-aligned-standards',
      true => '/Resources-for-aligned-standards',
    }
    category = {
      'Art-history' => '/The-arts',
      'Dance' => '/The-arts',
      'Drama' => '/The-arts',
      'Music' => '/The-arts',
      'Visual-arts' => '/The-arts',
      'Chinese' => '/Learning-languages',
      'Cook-islands-maori' => '/Learning-languages',
      'French' => '/Learning-languages',
      'German' => '/Learning-languages',
      'Japanese' => '/Learning-languages',
      'Latin' => '/Learning-languages',
      'Samoan' => '/Learning-languages',
      'Spanish' => '/Learning-languages',
      'English' => '',
      'Te-reo-maori' => '',
      'Te-reo-rangatira' => '',
      'Health-education' => '/Health-and-physical-education',
      'Home-economics' => '/Health-and-physical-education',
      'Home-ec' => '/Health-and-physical-education',
      'Physical-education' => '/Health-and-physical-education',
      'Mathematics-and-statistics' => '',
      'Agriculture-and-horticultural-science' => '/Science',
      'Biology' => '/Science',
      'Chemistry' => '/Science',
      'Earth-and-space-science' => '/Science',
      'Physics' => '/Science',
      'Science' => '/Science',
      'Accounting' => '/Social-sciences',
      'Business-studies' => '/Social-sciences',
      'Classical-studies' => '/Social-sciences',
      'Economics' => '/Social-sciences',
      'Education-for-sustainability' => '/Social-sciences',
      'Geography' => '/Social-sciences',
      'History' => '/Social-sciences',
      'Media-studies' => '/Social-sciences',
      'Social-studies' => '/Social-sciences',
      'Technology' => '',
      'DVC' => 'Technology',
    }
    url = "http://ncea.tki.org.nz"
    if ncea[:root]
      url
    elsif ncea[:alignment_landing_page]
      url << '/Resources-for-aligned-standards'
      url
    else
      url << status_prefix[ncea[:aligned]]
      url << category[ncea[:subject]]
      url << "/#{ncea[:subject]}"
      url << "/Level-#{ncea[:level]}-#{ncea[:subject]}" if ncea[:level]
      url
    end
  end
end




get '/' do
  haml :index
end

get '/url/?' do
  ncea = parse params[:url]
  if ncea == nil
    @url = "Unable to parse url"
  else
    @url = format ncea
  end
  haml :index
end



__END__
@@layout
%html
    %head
        %link(rel="stylesheet" href="http://twitter.github.com/bootstrap/assets/css/bootstrap.css")
        :css
            body {padding:30px; padding-left:60px;}
            input {height:27px;margin-bottom:0px;}
            h1,h2,h3 {padding-bottom:20px;}
            .response {font-size:150%;}
    %body
        .row
            = yield

@@index
%h1
    Enter a legacy NCEA url:
%form(id='url-form' method='get' action='/url' class='well form-vertical')
    %input(name='url' type='text' required='required' class='input-xlarge span6')
    %button(type='submit' class='btn')
        Submit
- if @url
    = haml :response

@@response
%p.response
    %strong= @url