movie-date-search.rb

#!/usr/bin/env ruby
# Movie date search by filename

require 'rubygems'
require 'open-uri'
require 'cgi'
require 'hpricot'

files = Dir[File.join(ARGV[0], '*')]
unless files.any?
  puts "Call this script with the name of a directory full of movie files you want to add years to. For example:"
  puts "    movie-date-search /folder/with/movies/in/it"
end

trap(:INT) { exit }

files.each do |fname|
  title = File.basename(fname, File.extname(fname))

  unless title =~ /\(\d{4}\)/
    puts "Searching for #{title}"

    html = open("http://us.imdb.com/find?q=#{CGI.escape(title)}").read

    if html =~ /no matches/i
      puts "Couldn't find any matches on IMDB"
    else
      doc = Hpricot(html)
      # this is nth-child(1) instead of nth-child(3) because of a bug in hpricot
      # http://github.com/fizx/hpricot/commit/1208019ade1af02816de6217d3c34bd82054b9b7
      matches = doc.search("tr td:nth-child(1)").map do |td|
        td.inner_html.match(/">(.*?)<\/a> \((\d{4})/).to_a[1..-1]
      end.compact
      options = matches.map do |m|
        CGI.unescapeHTML(m.first.gsub(/<.*?>/, '') + " (#{m.last})")
      end

      options = [doc.at("//title").inner_html] unless options.any?

      options.reverse.each_with_index do |o, i|
        puts "#{options.size - i}. #{o}"
      end
      puts "Which one matches '#{title}'? "
      line = STDIN.gets
      next if !line || line =~ /^[\s\r\n]*$/
      choice = options[line.to_i - 1]

      command = %{mv "#{fname}" "#{File.dirname(fname)}/#{choice}#{File.extname(fname)}"}
      puts(command)
      system(command)
    end
  end
end