#!/usr/bin/env ruby
 
$:.unshift "#{File.dirname(__FILE__)}/lib"
 
require 'environment'
require 'mechanize'
require 'open-uri'
require 'name'
require 'people'
require 'hpricot'
require 'configuration'
require 'json'
 
conf = Configuration.new
 
# Not using caching proxy since we will be running this script once a day and we
# always want to get the new data
agent = WWW::Mechanize.new
 
puts "Reading member data..."
people = PeopleCSVReader.read_members
 
puts "Twitter information (from tweetmp.org.au)..."
 
xml = File.open("#{conf.members_xml_path}/twitter.xml", 'w')
x = Builder::XmlMarkup.new(:target => xml, :indent => 1)
x.instruct!
x.peopleinfo do
  JSON.parse(agent.get("http://tweetmp.org.au/api/mps.json").body).each do |person|
    aph_id = person["GovernmentId"].upcase
    twitter = person["TwitterScreenName"]
    # Lookup the person based on their government id
    p = people.find_person_by_aph_id(aph_id)
    # Temporary workaround until we figure out what's going on with the aph_id's that start with '00'
    if p.nil?
      p = people.find_person_by_aph_id("00" + aph_id)
      puts "WARNING: Couldn't find person with aph id: #{aph_id}" if p.nil?
    end
    if twitter
      x.personinfo(:id => p.id, :mp_twitter_screen_name => twitter)
    else
      # Give the URL for inviting this person to Twitter using tweetmp.org.au
      x.personinfo(:id => p.id, :mp_twitter_invite_tweetmp => "http://tweetmp.org.au/mps/invite/#{person["Id"]}")
    end
  end
end
xml.close
 
puts "Personal home page & Contact Details (Gov website)..."
 
def extract_links(name, people, agent, link, x)
  person = people.find_person_by_name_current_on_date(name, Date.today)
  if person
    sub_page = agent.click(link)
    home_page_tag = sub_page.links.find{|l| l.text =~ /personal home page/i}
    
    params = {:id => person.id, :mp_contactdetails => sub_page.uri}
    params[:mp_website] = home_page_tag.uri if home_page_tag
    x.personinfo(params)
  else
    puts "WARNING: Could not find person with name #{name.full_name}"
  end
end
 
xml = File.open("#{conf.members_xml_path}/websites.xml", 'w')
x = Builder::XmlMarkup.new(:target => xml, :indent => 1)
x.instruct!
x.peopleinfo do
  agent.get(conf.alternative_current_house_members_url).links.each do |link|
    if link.to_s =~ /Member for/
      name = Name.last_title_first(link.text.split(',')[0..1].join(','))
      extract_links(name, people, agent, link, x)
    end
  end
  agent.get(conf.alternative_current_senate_members_url).links.each do |link|
    if link.to_s =~ /Senator/
      name = Name.last_title_first(link.to_s.split('-')[0..-2].join('-'))
      extract_links(name, people, agent, link, x)
    end
  end
end
xml.close
 
puts "Election results (from the abc.net.au) ..."
 
abc_root = "http://www.abc.net.au"
xml = File.open("#{conf.members_xml_path}/links-abc-election.xml", 'w')
x = Builder::XmlMarkup.new(:target => xml, :indent => 1)
x.instruct!
 
x.consinfos do
  # Representatives
  url = "#{conf.election_web_root}/results/electorateindex.htm"
  doc = Hpricot(open(url))
  (doc/"td.electorate").each do |td|
    href = td.at("a")['href']
    href = "#{abc_root}#{href}"
    name = td.at("a").inner_text
    name = name.gsub(/\*/,'').strip
    x.consinfo(:canonical => name, :abc_election_results_2007 => href)
  end
  # Senate
  url = "#{conf.election_web_root}/results/senate/"
  doc = Hpricot(open(url))
  (doc/:a).each do |a|
    if /results\/senate\/(\w+)\.htm/.match(a['href'])
      href = abc_root + a['href']
      name = a.inner_text
      x.consinfo(:canonical => name, :abc_election_results_2007 => href)
    end
  end
end
xml.close
 
puts "Q&A Links..."
 
data = {}
 
# First get mapping between constituency name and web page
page = agent.get(conf.qanda_electorate_url)
map = {}
 
page.links[35..184].each do |link|
  map[link.text.downcase] = (page.uri + link.uri).to_s
end
# Hack to deal with "Flynn" constituency incorrectly spelled as "Flyn"
map["flynn"] = "http://www.abc.net.au/tv/qanda/mp-profiles/flyn.htm"
 
bad_divisions = []
# Check that the links point to valid pages
map.each_pair do |division, url|
  begin
    agent.get(url)
  rescue WWW::Mechanize::ResponseCodeError
    bad_divisions << division
    puts "ERROR: Invalid url #{url} for division #{division}"
  end
end
# Clear out bad divisions
bad_divisions.each { |division| map.delete(division) }
 
people.find_current_members(House.representatives).each do |member|
  short_division = member.division.downcase[0..3]
  link = map[member.division.downcase]
  data[member.person.id] = link
  puts "ERROR: Couldn't lookup division #{member.division}" if link.nil?
end
 
page = agent.get(conf.qanda_all_senators_url)
page.links.each do |link|
  if link.uri.to_s =~ /^\/tv\/qanda\/senators\//
    # HACK to handle Unicode in Kerry O'Brien's name on Q&A site
    if link.to_s == "Kerry O\222Brien"
      name_text = "Kerry O'Brien"
    else
      name_text = link.to_s
    end
    member = people.find_member_by_name_current_on_date(Name.title_first_last(name_text), Date.today, House.senate)
    if member.nil?
      puts "WARNING: Can't find Senator #{link}"
    else
      data[member.person.id] = page.uri + link.uri
    end
  end
end
 
xml = File.open("#{conf.members_xml_path}/links-abc-qanda.xml", 'w')
x = Builder::XmlMarkup.new(:target => xml, :indent => 1)
x.instruct!
x.peopleinfo do
  data.each do |id, link|
    x.personinfo(:id => id, :mp_biography_qanda => link)
  end
end
xml.close
 
system(conf.web_root + "/twfy/scripts/mpinfoin.pl links")