#!/usr/local/bin/ruby
=begin

= Overview
This script is for caching banner-images of others' sites.
Check abuse with referrer and file size.

== See also
<URL:http://www.pandora.nu/tempo-depot/notes/PC_side/Web_Browser/d050107x3.htm>

= Usage
 cache.rb?http://example.com/banner


= License
Creative Commons `by'
<URL:http://creativecommons.org/licenses/by/2.0/>

(c) 2005 cho45 <cho45@lowreal.net>
=end

@config = {
	# Cache directory.
	# Cache directory must be writable.
	"Cache-Directory" => ".",
	
	# Image when Error.
	"No-Banner" => "../nobanner.png",
	
	# Limit Cached File Size.
	# Raise <code>OverFilesizeError</code>
	# if caching file over this limit.
	"Filesize-Limit" => 40000, # (byte)
	
	# Recache after this days.
	# HEAD access -> GET if filesize is changed
	"Crawl-Interval" => 7, # (day)
	
	# Acceptable Content-types.
	# Raise <code>NotConfiguredContentTypeError</code> 
	# if you try caching file with not configured Content-Type.
	"Content-Type" => {
		"image/gif"  => "gif",
		"image/jpeg" => "jpg",
		"image/png"  => "png",
	},
}


require "net/http"
require "uri"

class CacheImg
	
	class CacheImgError < StandardError; end
	class RequestFailedError < CacheImgError; end
	class OverFilesizeError < CacheImgError; end
	class NotConfiguredContentTypeError < CacheImgError; end
	
	VERSION = "0.2"
	UA = "Mozilla/4.0 (MSIE 6) ImageCache/#{VERSION}"
	DEFAULT_CONTENT_TYPE_CONF = {
		"image/gif"  => "gif",
		"image/jpeg" => "jpg",
		"image/png"  => "png"
	}

	attr_accessor :dir, :content_type, :crawl_interval, :max_filesize, :content_type

	def initialize(dir, crawl_interval=7, max_filesize=40000, content_type=DEFAULT_CONTENT_TYPE_CONF)
		@dir = dir
		@crawl_interval = crawl_interval
		@max_filesize = max_filesize
		@content_type = content_type
	end

	def get_image(uri)
		filename = nil
		content_type = nil
		basename = "#{@dir}/#{uri2filename(uri)}"
		@content_type.each do |k, v|
			if File.exist?("#{basename}.#{v}")
				filename = "#{basename}.#{v}"
				content_type = k
			end
		end
		if File.exist?("#{basename}.err")
			filename = "#{basename}.err"
		end
		if filename
			if File.mtime(filename) + (@crawl_interval * 24 * 60 * 60) < Time.now
				uri = URI.parse(uri)
				res = nil
				Net::HTTP.start(uri.host, uri.port) do |http|
					header = {
						"User-Agent" => UA,
						"Accept" => @content_type.keys.join(",")
					}
					res = http.head(uri.request_uri, header)
				end
				if res["content-length"].to_i != File.size(filename)
					File.delete(filename)
					content_type, content = restore_img(uri)
				else
					t = Time.now
					File.utime(t, t, filename)
					content_type = @content_type.index(File.extname(filename)[1..-1])
					content = File.open(filename, "rb") {|f| f.read}
				end
			else
				content_type = @content_type.index(File.extname(filename)[1..-1])
				content = File.open(filename, "rb") {|f| f.read}
			end
		else
			content_type, content = restore_img(uri)
		end
		
		unless content_type
			raise CacheImg.const_get(content).new
		end
		
		[content_type, content]
	end

	private
	def restore_img(uri)
		uri = URI.parse(uri)
		content_type = nil
		ext = "err"
		content = ""
		Net::HTTP.start(uri.host, uri.port) do |http|
			header = {
				"User-Agent" => UA,
				"Accept" => @content_type.keys.join(",")
			}
			res = http.get(uri.request_uri, header)
			case res.code
			when "200"
				ext = @content_type[res["content-type"]]
				if ext
					if res.body.size < @max_filesize
						content_type = res["content-type"]
						content = res.body
					else
						ext = "err"
						content = "OverFilesizeError"
					end
				else
					ext = "err"
					content = "NotConfiguredContentTypeError"
				end
			else
				content = "RequestFailedError"
			end
		end
		File.open("#{@dir}/#{uri2filename(uri.to_s)}.#{ext}", "wb") do |f|
			f.print content
		end
		[content_type, content]
	end

	def uri2filename(uri)
		uri.gsub(/[\\\/\*\?\|"<>:,;% ']/) do |m|
			"%#{Regexp.last_match[0].unpack("H2")}"
		end
	end

	def filename2uri(filename)
		filename.gsub(/%([A-Za-z0-9]{2})/) do |m|
			Regexp.last_match[1].hex.chr
		end
	end
end


require "cgi"

@cgi = CGI.new
puts @cgi.header("Content-Type" => "text/plain") if $DEBUG

begin
	class InvalidReferrer < StandardError; end
	raise InvalidReferrer unless @cgi.referer =~ /^http:\/\/#{@cgi.host}.*$/
	
	uri = @cgi.query_string
	uri ||= ""
	if uri.empty?
		@cgi.out("Content-Type" => "image/png") do
			File.open(@config["No-Banner"], "rb") {|f| f.read}
		end unless $DEBUG
	else
		@ci = CacheImg.new(@config["Cache-Directory"], @config["Crawl-Interval"], @config["Filesize-Limit"], @config["Content-Type"])
		begin
			content_type, content = @ci.get_image(uri.gsub(/\s/, ''))
		rescue CacheImg::CacheImgError => e
			content_type = "image/png"
			content = File.open(@config["No-Banner"], "rb") {|f| f.read}
		end
		
		@cgi.out("Content-Type" => content_type) do
			content
		end unless $DEBUG
	end
rescue InvalidReferrer => e
	@cgi.out("Content-Type" => "text/plain") do
		"Invalid Referrer"
	end
	
rescue Exception => e
	if $DEBUG
		puts @cgi.header("Content-Type" => "text/plain")
		p e.message
		p e.backtrace
	else
		@cgi.out("Content-Type" => "image/png") do
			File.open(@config["No-Banner"], "rb") {|f| f.read}
		end
	end
end