gistfile1.rb

require 'rubygems'
require 'hpricot'
doc = open("initiative_shortlist.html") {|f| Hpricot(f) }
counter = 0
names = []
places = []
years = []
(doc/"#table1/tr").each do |tr|
  if counter > 0
    tds = tr.search("td")
    #names << tds[0].search("font").first.inner_html.gsub(/[^\w ]/,"")
    original_place = tds[1].search("font").first.inner_html.gsub(/[^\w]/,"").gsub("nbsp","").downcase.capitalize
    places <<  original_place if original_place.length > 0
    original_year =  tds[2].search("font").first.inner_html.gsub(/[^\w\/]/,"").gsub("nbsp","")
    year = original_year.split("/").last
    if !year
      #puts "Invalid year #{original_year}"
    else
      if year.length == 4
        years << year
      elsif year.length == 2 && year != "NA" && year.to_i < 10 # 20xx xx < 10
        years << "20" + year
      elsif year.length == 2 && year != "NA" && year.to_i > 30 # 19xx xx > 30
        years << "19" + year
      else 
        #puts "Invalid year #{original_year}"
      end
    end
  end
  counter = counter + 1
end


sorted_places = {}
counter = 0
place = places.first
places.sort.each do |p|
  if p == place
    counter = counter + 1
  else
    sorted_places[place] = counter
    counter = 1
  end
  place = p
end
sorted_places.sort{|a,b| a[1]<=>b[1]}.reverse.each { |elem|
    puts "#{elem[0]}, #{elem[1]}"
}

puts "***********"
sorted_years = {}
counter = 0
year = years.first
years.sort.each do |p|
  if p == year
    counter = counter + 1
  else
    sorted_years[year] = counter
    counter = 1
  end
  year = p
end
sorted_years.sort{|a,b| a[1]<=>b[1]}.reverse.each { |elem|
    puts "#{elem[0]}, #{elem[1]}"
}

puts "***********"
sorted_years.sort.each { |elem|
    puts "#{elem[0]}, #{elem[1]}"
}