#!/usr/bin/ruby
#
# youtube-scanner.rb xuzo@cuarentaydos.com
# 
# v 0.0.1 Initial version.
#	Gaim, Gajim, Firefox and ClawsMail scanners.
#   Youtube URL detection.
#	Cache and error cache support.
#
# vim: ts=2
#

require 'sqlite3'  # GajimScanner
require 'net/http' # HTTP library

#
# Cache class for store results
#
class Cache
	def initialize(file)
		@data = {}
		@file = file
		
		# Read cache
		if FileTest.exists?(@file) and
			FileTest.file?(@file)
			f = File.new(@file,"r")
			f.each { |line|
				url, title = line.chomp.split(/\s+/, 2)
				@data[url] = title
			}
			f.close()
		end
	end

	def add(url,title = nil)
		if title != nil
			title.chomp!
		end
		@data[url] = title
	end

	def del(url)
		@data.delete([url])
	end

	def get(url)
		return @data[url]
	end

	def flush
		f = File.new(@file, 'w+')
		@data.each do |url,title|
			if not url.nil? and not url.empty?
				f.write("#{url} #{title}\n")
			end
		end
		f.close()
	end
end

class Url
	attr_accessor :url, :provider, :data

	def initialize(url, provider = nil, data = {})
		@url = url
		@provider = provider
		@data = data
	end

	def to_s
		return @url
	end

	def <=>(other_url)
		return @url <=> other_url.url
	end

	def ==(other_url)
		return @url == other_url.url
	end

	def ===(other_url)
		return @url == other_url.url
	end
end

#
# Base class for backends
#
class Scanner
	def scan(location)
		ret = []
		lineno = 1

		File.new(location, "r").each do |line|
			if (url = self.is_youtube?(line)) != nil
				ret.push(Url.new(url,'Scanner', {:file => location, :line => lineno}))
			end
			lineno = lineno + 1
		end

		return ret.uniq
	end

	def is_youtube?(string)
			start_url = string.index('youtube.com/watch?')
			if start_url.nil?
				return nil
			end
			string = string[start_url..-1]

			# Cortar por espacio
			end_url = string.index(' ', start_url)
			if not end_url.nil?
					string[0..end_url-1]
			end

			# Buscar el parametro v en la url
			v_param = string.index(/v=[A-Za-z0-9_\-\=]{11}/)
			if not v_param.nil?
				return "http://youtube.com/watch?" + string[v_param..v_param+12]
			end
	end

	def ls_R(location)
		ret = { :directories => [], :files => [] }

		if not FileTest.exists?(location)
			return []
		end

		if FileTest.file?(location)
			ret[:files].push(location)

		elsif FileTest.directory?(location)
			entries = Dir.entries(location)
			entries.slice!(0, 2) # Skip '.' and '..'

			entries.each do |e|
				fullpath = location + '/' + e

				if FileTest.directory?(fullpath)
					ret[:directories].push(fullpath)
					subs = self.ls_R(fullpath)
					ret[:directories] += subs[:directories]
					ret[:files]       += subs[:files]

				elsif FileTest.file?(fullpath)
					ret[:files].push(fullpath)
				end
			end

			return ret
		end
	end

end

#
# Backends
#
class GaimScanner < Scanner
	def scan(location = ENV['HOME'] + '/.gaim')
		location += '/logs'

		if not FileTest.directory?(location)
			return []
		end
		ret = []
		entries = self.ls_R(location)
		entries[:files].delete_if do |e| e.index(".txt", -4) == nil end
		entries[:files].each do |file| ret += super(file) end

		ret.each { |url| url.provider = self.class }

		return ret
	end
end

class GajimScanner < Scanner
	def scan(location = ENV['HOME'] + '/.gajim')
		location += '/logs.db'

		if not FileTest.file?(location)
			return []
		end

		ret = []
		db = SQLite3::Database.new(location)
		res = db.query("SELECT log_line_id,message FROM logs WHERE message LIKE '%youtube.com/watch%'")
		res.each do |row|
			if (url = self.is_youtube?(row[1])) != nil
				ret.push(Url.new(url, 'GajimScanner', { :file => location, :id => row[0] }))
			end
		end

		return ret
	end
end

class ClawsMailScanner < Scanner
	def scan(location = ENV['HOME'] + '/.sylpheed-claws' )
		location += '/imapcache'

		ret = []
		paths = self.ls_R(location)
		paths[:files].each do |path|
			ret += super(path)
		end

		ret.each do |url| url.provider = self.class end

		return ret
	end
end

class FirefoxScanner < Scanner
	def scan(location = ENV['HOME'] + '/.mozilla/firefox')
		ret = []
		
		Dir.foreach(location) do |profile|
			if /^[A-Za-z0-9]{8}\.default$/.match(profile) \
			and FileTest.exists?(location + '/' + profile + '/history.dat') 
				ret += super(location + '/' + profile + '/history.dat')
			end
		end

		ret.each do |url| url.provider = self.class end
	end
end

#
# HACK: Uniq over urls
#
def custom_uniq(urls)
	tmp = {}
	urls.each do |url| tmp[url.url] = url end

	ret = []
	tmp.each_value do |obj| ret.push(obj) end

	return ret
end

if __FILE__ == $0
	urls = []
  urls += GajimScanner.new.scan()
	urls += ClawsMailScanner.new.scan()
	urls += FirefoxScanner.new.scan()
	urls += GaimScanner.new.scan()

	cache = Cache.new(ENV['HOME'] + '/.video-cache')
	error_cache = Cache.new(ENV['HOME'] + '/.video-cache.error')
	conn = Net::HTTP.new('www.youtube.com', 80)
	urls = custom_uniq(urls)

	urls.each do |url|
		in_cache = true

		title = cache.get(url.url)
		if title.nil?

			if error_cache.get(url.url)
				$stderr.puts "[E] URL is blacklisted in error cache #{url}"
				next
			end

			path_idx = url.url.index('/watch?')
			if path_idx.nil?
				$stderr.puts "[E] Cannot find /watch? in [#{url}]"
				# puts "[D] #{url.inspect}"
				next
			end
			path = url.url[path_idx..-1]

			resp, data = conn.get(path)
			if resp.code != "200"
				$stderr.puts "[E] Server sais: #{resp.code}: #{resp.message} for #{url}"
				error_cache.add(url.url, nil)
				# puts "[D] #{url.inspect}"
				next
			end

			start_pos = data.index('<title>YouTube - ')
			end_pos = data.index('</title>')
			if start_pos.nil? or end_pos.nil?
				$stderr.puts "[E] Cannot find title in [#{data}]"
				next
			end

			title = data[start_pos+'<title>YouTube - '.size..end_pos-1]
			cache.add(url.url, title)
			in_cache = false
		end

		puts "[#{in_cache ? '*' : ' '}] [#{url}] #{title}"
	end

	cache.flush()
	error_cache.flush()
end