bigpicvotes

#!/usr/bin/env ruby
# Start Date: Wednesday November 19, 2008
# Current Version: 0.9
# Author: Joseph Pecoraro
# Contact: joepeck02@gmail.com
# Decription: Analyze the user comments for top images
#   because the commenters often put the numbers for
#   the images they really liked.

require 'rubygems'
require 'hpricot'
require 'net/http'
require 'uri'


# Wrapper around the functionality
class BigPictureAnalyzer
  
  # Takes in a valid Hpricot doc (assumes validation
  # of the URL it was taken from occured elsewhere)
  def initialize(url, body)
		@url = url
		@doc = Hpricot(body)
		@votes = nil
  end


	# Analyze the doc
	def analyze(n=0)
		
		# The title
		@title = (@doc/'h2')[0].innerText
		
		# Find the Number of images to make a vote array
		max = (@doc/'div.bpBody a')[-1].innerText.match(/\d+/).to_s.to_i
		@votes = Array.new(max+1, 0)
		
		# Search Comments
		(@doc/'.commentBodyText').each do |com|
			com.to_s.gsub(/https?:.*?(\s|[<>])/i, 'zz').scan(/\d+/) do |num|
				num = num.to_i
				@votes[num] = @votes[num] + 1 if num <= max && num > 0
			end
		end

		# Ugly Sort Trick
		# [1, 5, 3] => ["1:0", "5:1", "3:2"] => sort => ["5:1", "3:2", "1:0"]
		@votes.each_with_index { |e, i| @votes[i] = "#{@votes[i]}:#{i}" }
		@votes = @votes.sort.sort do |a,b|
			numa = a.split(/:/)[0].to_i
			numb = b.split(/:/)[0].to_i
			numb - numa
		end
		
		# Possibly print the results
		print_top_n(n) if n > 0
		
	end


	# Print the results
	def print_top_n(n=10)
		
		# Error handling
		n = 10 if n <= 0
		n = @votes.length-1 if n > @votes.length
			
		# Print header and top n
		puts
		puts "----------------------------"
		puts @title
		puts "Top Images Based on Comments"
		puts "----------------------------\n\n"
		0.upto(n-1) do |i|
			cnt, img = @votes[i].split(':')
			puts "  %2d: Image #%-2d - %d votes" % [i+1, img, cnt]
		end
		puts
		
	end

end



# When run as as script
if $0 == __FILE__
	
	# Expect at least one command line arg (1 to 2 args)
	unless (1..2) === ARGV.size
		program_name = $0.split(/\//).last
		puts "usage: #{program_name} url [top#]"
		exit 1
	end
	
	# Command line args
	url = ARGV[0]
	n = ARGV[1].to_i || 10
	n = 10 if n <= 0
	
	# Turn the url into a 'comments_' url if needed
	uri = URI.parse(url)
	parts = uri.path.split(/\//)
	last = parts.last
	unless last =~ /^comments_/
		parts[ parts.length-1 ] = 'comments_' + last
		uri.path = parts.join('/')
	end

	# Validate the the URI exists, and parse it if it does
	begin
		res = Net::HTTP.get_response uri
		BigPictureAnalyzer.new(url, res.body).analyze(n)
	rescue
		puts "Page did not exist. Please correct the URL"
		exit 2
	end
	
end