(* Copyright (C) 2007 Mauricio Fernandez <mfp@acm.org> http//eigenclass.org
 * See README.txt and LICENSE for the redistribution and modification terms *)

let prefix = Bigstring.BM_search.make "GET /ongoing/When/"
let re = "[0-9][0-9][0-9]x/\\([0-9][0-9][0-9][0-9]/[0-9][0-9]/[0-9][0-9]/[^ .]+\\) "
let re = Str.regexp re

let copy_until_space src off dst =
  let rec loop src i dst j =
    let c = Bigstring.String.unsafe_get src i in
      dst.[j] <- c;
      if c <> ' ' then loop src (i+1) dst (j+1)
  in loop (Bigstring.unsafe_string src) off dst 0

let process_block hash bs =
  try
    let buf = String.make 2048 ' ' in
    let off = ref 0 in
    let len = Bigstring.length bs in
      while true do
        off := Bigstring.BM_search.find_end prefix bs !off;
        if !off >= len then raise Not_found;
        copy_until_space bs !off buf;
        if Str.string_match re buf 0 then
          let k = Str.matched_group 1 buf in
            off := !off + 17; (* the match is at least this long *)
            try incr (Hashtbl.find hash k)
            with Not_found -> Hashtbl.add hash k (ref 1)
      done
  with Not_found -> ()

let sort_results hash =
  List.sort (fun (_,a) (_,b) -> b - a)
    (Hashtbl.fold (fun url hits l -> (url, !hits) :: l) hash [])

let rec print_top_n n = function
    [] -> ()
  | ((url, hits) :: tl) ->
      if n > 0 then (Printf.printf "%d: %s\n" hits url; print_top_n (n-1) tl)

let line_finder filename =
  let bs = Bigstring.map_file
             (Unix.openfile filename [Unix.O_RDONLY] 0o644) (-1) in
  let hash = Hashtbl.create 2000 in
    process_block hash bs;
    print_top_n 10 (sort_results hash)

let () = line_finder (Array.get Sys.argv 1)