require 'net/http'
require 'time'
require 'uri'

class App

  def initialize argv
    @tsv = argv.first || 'pages.tsv'
    @srv = 'www.wmo.int'
    @port = 80
    @path = '/pages/prog/www/WIS/wiswiki/tiki-index.php'
    @tasks = []
  end

  def link title, page
    return '' if '-' == page
    ['"', (title || page).gsub(/"/, 'Q'), '":',
      URI::HTTP.build([nil, @srv, @port, @path, "page=#{page}", nil]).to_s
    ].join
  end

  def loadtab
    File.open(@tsv, 'r') {|fp|
      for line in fp
	lev, page, desc = line.chomp.split(/\t/, 3)
	lev = lev.to_i
	next if lev.zero?
	@tasks.push({:lev => lev, :page => page, :desc => desc})
      end
    }
  end

  def checkpage http, page
    path = [@path, '?page=', page].join
    resp = http.get(path)
    unless /\A2/ === resp.code
      return({:code => resp.code})
    end
    title = if /<title>([^<]*)<\/title/ === resp.body
      then $1.strip.sub(/ : WIS WIKI/, '')
      else ""
      end
    if title == page and /<div class="titlebar">([^<]*)<\/div/ === resp.body
      title = $1.strip
    end
    if title == 'This page has not yet been developed'
      title = "(#{page})"
    end
    title = "#{title[0,32]}..." if title.size > 32
    update = if /<p class="editdate">\s*Page last modified on ([^<]*)<\/p/ === resp.body
      then Time.parse($1.strip.sub(/ of/, '').sub(/CEST/, '-0100'))
      else nil
      end
    return({:code => resp.code, :title => title, :update => update,
      :page => page, :size => resp.body.size})
  end

  def checkweb
    Net::HTTP.start(@srv, @port) {|http|
      for task in @tasks
        if '-' == task[:page]
	  task[:title] = task[:desc]
	  task[:desc] = nil
	else
	  STDERR.puts "GET #{task[:page]}" if STDERR.tty?
	  r = checkpage(http, task[:page])
	  task[:title] = r[:title]
	  task[:update] = r[:update].strftime('%Y-%m-%d') if r[:update]
	  task[:update] = 'stub' if /\A2013-07/ === task[:update]
	  r = checkpage(http, '%2A' + task[:page])
	  task[:draft] = r[:title] ? true : false
	end
	report(task)
      end
    }
  end

  def report task
    puts ['', 
      [('~ ' * task[:lev]), task[:title]].join,
      link(task[:update], task[:page]),
      task[:draft] ? link('draft', '%2A' + task[:page]) : ' ',
      task[:desc],
    ''].join('|')
  end

  def run
    loadtab
    checkweb
  end

end


App.new(ARGV).run