open Printf

let http_scheme_RE = Pcre.regexp ~flags:[`CASELESS] "^http://"
let url_RE = Pcre.regexp "^([\\w.-]+)(:(\\d+))?(/.*)?$"

let tcp_bufsiz = 4096 (* for TCP I/O *)

let parse_url url =
  try
    let subs =
      Pcre.extract ~rex:url_RE (Pcre.replace ~rex:http_scheme_RE url)
    in
    (subs.(1),
    (if subs.(2) = "" then 80 else int_of_string subs.(3)),
    (if subs.(4) = "" then "/" else subs.(4)))
  with exc ->
    failwith
      (sprintf "Can't parse url: %s (exception: %s)"
        url (Printexc.to_string exc))

let init_socket addr port =
  let inet_addr = (Unix.gethostbyname addr).Unix.h_addr_list.(0) in
  let sockaddr = Unix.ADDR_INET (inet_addr, port) in
  let suck = Unix.socket Unix.PF_INET Unix.SOCK_STREAM 0 in
  Unix.connect suck sockaddr;
  let outchan = Unix.out_channel_of_descr suck in
  let inchan = Unix.in_channel_of_descr suck in
  (inchan, outchan)

let http_get_iter_buf ~callback url =
  let (address, port, path) = parse_url url in
  let buf = String.create tcp_bufsiz in
  let (inchan, outchan) = init_socket address port in
  output_string outchan (sprintf "GET %s HTTP/1.0\r\n\r\n" path);
  flush outchan;
  (try
    while true do
      match input inchan buf 0 tcp_bufsiz with
      | 0 -> raise End_of_file
      | bytes when bytes = tcp_bufsiz ->  (* buffer full, no need to slice it *)
          callback buf
      | bytes when bytes < tcp_bufsiz ->  (* buffer not full, slice it *)
          callback (String.sub buf 0 bytes)
      | _ -> (* ( bytes < 0 ) || ( bytes > tcp_bufsiz ) *)
          assert false
    done
  with End_of_file -> ());
  close_in inchan (* close also outchan, same fd *)

let http_get url =
  let buf = Buffer.create 10240 in
  http_get_iter_buf ~callback:(Buffer.add_string buf) url;
  Buffer.contents buf