#!/usr/bin/env ruby

require 'thread'
require 'socket'
begin ; require 'openssl' ; rescue LoadError ; end

# 
# A ruby http file downloader using multiple parallel connections
#
# supports proxies, basic authentication, ssl
#
# Usage:
# ruby httppget.rb -n 12 http://kik:oo@myhost.com/bigfile.lol
#
# Licence::   Redistributes under the terms of the GPL version 2, with specific permission to be linked to OpenSSL without invoking GPL clause 2 (b)
#

class Http
def initialize(url)
	raise 'Unparsed url' unless md = %r{(?:http-proxy://(\w+:\w+@)?([\w.-]+)(:\d+)?/)?(?:http(s)?://)?(\w+:\w+@)?([\w.-]+@)?([\w.-]+)(:\d+)?(/.*)}.match(url)
	@proxylp, @proxyh, @proxyp, @use_ssl, @loginpass, @vhost, @host, @port, @path = md.captures
	@proxyp = @proxyp ? @proxyp[1..-1].to_i : 3128
	@port = @port ? @port[1..-1].to_i : (@use_ssl ? 443 : 80)
	@proxylp = @proxylp ? "Proxy-Authorization: Basic #{[proxylp.chop].pack('m').chomp}\r\n" : ''
	@loginpass = @loginpass ? "Authorization: Basic #{[loginpass.chop].pack('m').chomp}\r\n" : ''
	@vhost = @vhost ? @vhost.chop : @host
	@path = @path.gsub(/[^a-zA-Z0-9_.\/-]/) { |c| '%' << c.unpack('H2').first }

	connect
end

def connect
	if @proxyh
		@socket = TCPSocket.new @proxyh, @proxyp
		if @use_ssl
			@socket.puts "CONNECT #@host:#@port HTTP/1.1\r\n" << @proxylp << "\r\n"
			buf = @socket.gets
			raise "non http answer #{buf[1..100].inspect}" if buf !~ /^HTTP\/1.. (\d+) /
			raise "CONNECT bad response: #{buf.inspect}" if $1.to_i != 200
			nil until @socket.gets.chomp.empty?
		end
	else
		@socket = TCPSocket.new @host, @port
	end
	if @use_ssl
		@socket = OpenSSL::SSL::SSLSocket.new(@socket, OpenSSL::SSL::SSLContext.new)
		@socket.sync_close = true
		@socket.connect
	end
end

def request(method='GET', range=nil)
	range = "Range: bytes=#{range}\r\n" if range
	request =  	"#{method} #{"http://#@host#{':'+@port if @port != 80}/" if @proxyh and not @use_ssl}#{@path} HTTP/1.1\r\n" <<
			"Host: #{@vhost}#{':'+@port if @port != 80}\r\n" <<
			"#{@proxylp}"        <<
			"#{@loginpass}"      <<
			"#{range}"           <<
			"\r\n"

	puts request if $DEBUG

	@socket.write request

	raise 'eof' if not buf = @socket.readpartial(4096)

	buf =~ /^HTTP\/1.1 (\d+) /
	case $1.to_i
	when 200, 206
		# ok
	when 301, 302
		puts '302 redirect to ' << buf[/^Location: (.*?)\r?$/, 1].inspect
		return ''
	when 416
		# bad range : beyond end of file
		return ''
	else
		raise 'unhandled http response:'+ buf
	end

	buf =~ /^Content-length: (\d+)/i
	len = $1.to_i if $1

	headlen = buf.index("\r\n\r\n") + 4

	puts buf[0, headlen+2] if $DEBUG

	return len if method == 'HEAD'

	data = buf[headlen..-1]
	if not len
		if buf.length < 4096
			data
		else
			data << @socket.read
		end
	else
		data << @socket.read(len - data.length) if len > data.length
	end
end

def self.download_file(urls, nthreads=nil, filename=nil, chunksize=1024*1024)
	nthreads ||= urls.length
	filename ||= urls.first[/[^\/]*$/]
	File.open(filename, 'w') {} if not File.exist? filename
	len = new(urls.first).request('HEAD')
	chunks = []
	(0..len).step(chunksize) { |i| chunks << i }
	m = Mutex.new
	tstart = Time.now
	threads = []
	idx = 0
	while not chunks.empty?
		$stderr.print "#{chunks.length} to go       \r" if $stderr.isatty
		if threads.length < nthreads
			threads << Thread.new {
				i = m.synchronize { chunks.shift }
				data = new(urls[idx = (idx + 1 % urls.length)]).request('GET', "#{i}-#{(i + chunksize) > len ? len : (i + chunksize)}")
				m.synchronize {
					File.open(filename, 'r+') { |fd|
						fd.pos = i
						fd.write data
					}
				}
			}
		end
		threads.delete_if { |t| not t.alive? }
		sleep 0.1
	end
	threads.each { |t| t.join }

	puts 'got %.02fM at %.02f ko/s' % [len/1024.0/1024, (len/(Time.now-tstart)/1024)]
end
end

if $0 == __FILE__
	require 'optparse'
	n = nil
	ofile = nil
	chunksize = 1024*1024
	OptionParser.new { |opts|
		opts.banner = "Usage: httppget.rb [options] url [mirror url ...]"
		opts.on('-n N', '--nthreads N', 'maximum number of parallel connections') { |o| n = o.to_i }
		opts.on('-O F', '--outfile F', 'file to write') { |o| ofile = o }
		opts.on('-c N', '--chunksize N', 'chunk size') { |o| chunksize = o.to_i }
		opts.on('-h', '--help', 'this message') { puts opts ; exit }
	}.parse!
	Http.download_file(ARGV, n, ofile, chunksize)
end