(************************************************************
   get_links.ml		Created      : Tue Nov 26 00:39:00 2002
  			Last modified: Wed Nov 27 00:15:06 2002
  Compile: ocamlopt.opt str.cmxa get_links.ml -o get_links #
  FTP Directory: sources/ocaml #
************************************************************)
(**

  @author Takashi Masuyama <mamewo@dk9.so-net.ne.jp>

*)

let anchor_regexp = Str.regexp_case_fold "<a href=\"\\([^\"]+\\)\""
let get_all_links input =
  let rec read_input result =
    try
      read_input (result^(input_line input))
    with
      End_of_file -> result in
  let data = read_input "" in
  let rec get_group count result =
    try 
      let start_position = Str.search_forward anchor_regexp data count in
      if Str.string_match anchor_regexp data start_position then
	 get_group (Str.match_end ()) ((Str.matched_group 1 data)::result)
      else
	assert false
    with 
      Not_found -> result
  in
  get_group 0 []

let _ =
  let files = ref [] in
  let specs = [] in
  let help_head_string = "" in
  begin
    Arg.parse specs (fun x -> files := x::!files) help_head_string;
    List.iter (fun x ->
      let input = open_in x in
      begin
	List.iter print_endline (get_all_links input);
	close_in input;
      end) !files;
  end