#!/usr/bin/env ruby

require 'fuserofs'
require 'socket'
require 'openssl'
require 'timeout'

# 
# A ruby HTTP filesystem for fuse
#
# rhttpfs supports proxies, basic authentication, ssl
# read requests uses http Range to get only required bytes, with a configurable temporary cache to allow seamless streaming
#
# Usage:
# ruby [-v] [-d] rhttpfs.rb httpurl mountpoint
#
# httpurl exemples:
#
#  http://www.site.com/
#
#  http://log:pass@www.site.com:8080/mydir/
#  
#  http://virtualhost.undef@10.0.0.1/mydir/
#
#  http-proxy://l:p@proxy.com:2121/https://www.blabla.com:8000/
#
# on the mounted filesystem, use :
#  * echo 1   > [mountdir]/.rhttpfs-ctrl/umount to unmount
#  * echo 1   > [mountdir]/.rhttpfs-ctrl/flush  to empty the directory listing / file size cache
#  * echo <n> > [mountdir]/.rhttpfs-ctrl/minreadsize to change the minimal size of http requests for file content
#
# Author::    Yoann Guillot (at ofjj.net)
# Copyright:: Copyright (c) 2006 Yoann Guillot
# Licence::   Redistributes under the terms of the GPL version 2, with specific permission to be linked to OpenSSL without invoking GPL clause 2 (b)
#
#
# Changelog:
#  before:	silent updates
#  05/06/07:	lol @ date
#		adopt great patch from Arnaud Cornet (allow HTTP/1.0 replies)
#		add entities decoder in dir listing
#  25/12/11:    add chunked transer-encoding support (dir listing)


class HttpFS < FuseROFS
	def http_get(path, range=nil, method='GET')
		path = (@rootdir + path).gsub(/\/+/, '/')

		path = path.gsub(/[^a-zA-Z0-9_.\/-]/) { |c| '%' << c.unpack('H2').first }

		off, offend = range.split('-').map { |i| i.to_i } if range

		range = "Range: bytes=#{range}\r\n" if range
		retried = false
		begin
			info "#{method} #{path}" + (range ? " [#{off}, #{offend-off+1}]" : '')
			@socket.write 	"#{method} #{"http://#@host/" if @proxyh and not @use_ssl}#{path} HTTP/1.1\r\n" <<
				"Host: #{@vhost}\r\n" <<
				"#{@proxylp}"        <<
				"#{@loginpass}"      <<
				"#{range}"           <<
				"\r\n"

			buf = nil
			begin
				Timeout::timeout(20) {
					raise 'eof' if not buf = @socket.readpartial(4096)
				}
			rescue Timeout::Error
				return ''
			end

			raise "wtf #{buf.to_s[0, 64].inspect}" if buf !~ /^HTTP\/1.[01] (\d+) /
			
			case $1.to_i
			when 200, 206
				# ok
			when 302
				info '302 redirect to ' << buf[/^Location: (.*?)\r?$/, 1].inspect
			when 416
				# bad range : beyond end of file
				@socket.close
				return ''
			else
				info 'unhandled http response:', buf
				raise 'unhandled http response:'+ buf
			end

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

			head = buf[0, headlen-4]

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

			debug head

			if method == 'HEAD'
				# return file size
				len or 2**20
			else
				# return body
				if len
					data = buf[headlen..-1]
					data << @socket.read(len - data.length) if len > data.length
				elsif head =~ /^Transfer-encoding: Chunked/i
					raw = buf[headlen..-1]
					data = ''
					loop do
						if not raw.empty?
							if not ll = raw.index("\r\n")
								raw << @socket.read(4096)
								next
							end
							line = raw[0, ll]
							raw[0, ll+2] = ''
						else
							line = @socket.gets
						end
						chunksize = line.to_i(16)
						break if chunksize == 0
						if raw.length > 0
							if raw.length < chunksize
								data << raw
								data << @socket.read(chunksize-raw.length)
								raw = ''
								@socket.read 2
							else
								data << raw[0, chunksize]
								raw[0, chunksize] = ''
							end
						else
							data << @socket.read(chunksize)
							@socket.read 2
						end
					end
				else
					data = buf[headlen..-1]
					data << @socket.read if buf.length >= 4096
				end
				data
			end

		rescue
			info "exception: #{$!.class} #{$!.message}"
			raise if retried
			retried = true
			connect_socket
			retry
		end
	end

	def connect_socket
		debug 'connecting'
		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 initialize(url)
		super()

		raise 'Unparsed url' unless md = %r{(?:http-proxy://(\w+:\w+@)?([\w.-]+)(:\d+)?/)?http(s)?://(\w+:\w+@)?([\w.-]+@)?([\w.-]+)(:\d+)?(/.*)}.match(url)

		@control_dir = '.rhttpfs-ctrl/'

		proxylp, @proxyh, proxyp, @use_ssl, loginpass, vhost, @host, port, @rootdir = 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

		@socket = nil
		@cache_timeout = 2*3600

		@minreadsize = 64*1024
		file_accessor_int :minreadsize, :cache_timeout

		file_accessor :rootdir, :host, :port, :vhost, :proxyh, :proxyp

		connect_socket

		debug 'connected'
	end

	def get_file_content(path, off)
		http_get(path, "#{off}-#{off+@minreadsize-1}")
	end

	def directory_listing(path)
		connect_socket if path == '/'

		list = []
		p = http_get(path)
		debug "dirlist #{path}"
		debug p
		p.scan(/<a\s+[^>]*href=(["'])(.*?)\1/i) {
			match = $2
			match.gsub!(/%([a-fA-F0-9]{2})/) { $1.hex.chr }
			match.gsub!(/&(.*?);/) { |o| {'amp' => '&', 'quot' => '"', 'apos' => '\'', 'lt' => '<', 'gt' => '>'}.fetch($1, o) }
			next if match == '../' or match == '/' or match[0] == ?? or match[0..-2].index '/'
			list << match
		}
		list
	end

	def get_file_size(path)
		http_get(path, nil, 'HEAD')
	end
end

if $0 == __FILE__
	begin
		Timeout::timeout(30, RuntimeError) { HttpFS.new(ARGV.shift) }.mount_under(ARGV.shift)
	rescue
		puts $!
		puts "Usage : #$0 [http-proxy://[login:pass@]host[:port]/]http[s]://[login:pass@][virtualhost@]host[:port]/[rootdir/] <mountpoint>"
	end
end