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

let prefix = 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 process_block hash buf len =
  try
    let off = ref 0 in
      while true do
        off := BM_search.find_end prefix buf !off;
        if !off >= len then raise Not_found;
        if Str.string_match re buf !off then
          let k = Str.matched_group 1 buf in
            Hashtbl.replace hash k (try Hashtbl.find hash k + 1 with Not_found -> 1);
            off := !off + 17 (* the match is at least this long *)
      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 input_bytes ic buf off n =
  let rec aux ic buf bytes off =
    if (off < bytes) then
      let br = input ic buf off (bytes-off) in
        if (br > 0) then aux ic buf bytes (off + br) else off
    else off
  in aux ic buf (n + off) off

let line_finder filename =
  let ic = open_in filename in
  let hash = Hashtbl.create 2000 in
  let buf = String.create Sys.max_string_length in
  let off = ref 0 in
    try
      while true do
        let bytes = input_bytes ic buf !off (Sys.max_string_length - 2048) in
          process_block hash buf bytes;
          let last_line_start = String.rindex_from buf bytes '\n' in
            off := bytes - last_line_start;
            String.blit buf !off buf 0 !off;
      done
    with End_of_file | Not_found -> print_top_n 10 (sort_results hash)

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