(******************************************************************************)
(*  SXSI : XPath evaluator                                                    *)
(*  Kim Nguyen (Kim.Nguyen@nicta.com.au)                                      *)
(*  Copyright NICTA 2008                                                      *)
(*  Distributed under the terms of the LGPL (see LICENCE)                     *)
(******************************************************************************)
INCLUDE "utils.ml"

open Ata

let () = init_timer();;


let default_gc = Gc.get()
let tuned_gc = { default_gc with
  Gc.minor_heap_size = 32*1024*1024;
  Gc.major_heap_increment = 8*1024*1024;
  Gc.max_overhead = 1000000;
  Gc.space_overhead = 100;
}

let mk_runtime run auto doc arg count print outfile =
  fun () ->
    if !Options.do_perf then start_perf ();
    let r = time ~count:!Options.repeat ~msg:"Execution time" (run auto doc) arg in
    if !Options.do_perf then stop_perf ();
    Logger.verbose Format.err_formatter "Number of results: %i@\n" (count r);
    match outfile with
	None -> ()
      | Some file ->
	time ~count:1 ~msg:"Serialization time" (print file !Options.no_wrap_results doc) r
;;

let main v query_string output =
  Tag.init (Tree.tag_operations v);
  if !Options.docstats then Tree.stats v;
  let query =
    time ~msg:"Parsing query" XPath.parse query_string
  in
  Logger.verbose Format.err_formatter "Parsed query:%a@\n"
    XPath.Ast.print query;
  let auto, bu_info =
    time ~msg:"Compiling query" Compile.compile query
  in
  if !Options.verbose then Ata.print Format.err_formatter auto;
  Gc.full_major();
  Gc.compact();
  Gc.set (tuned_gc);
  let runtime =
    match !Options.bottom_up, bu_info with

    | true, Some [ (query, pattern) ] ->
      if !Options.count_only then
        let module R = ResJIT.Count in
        let module M = Runtime.Make(R) in
        mk_runtime M.bottom_up_run auto v (query, pattern) R.NS.length R.NS.serialize !Options.output_file
      else
        let module R = ResJIT.Mat in
        let module M = Runtime.Make(R) in
        mk_runtime M.bottom_up_run auto v (query, pattern) R.NS.length R.NS.serialize !Options.output_file

    | _ ->
      (* run the query top_down *)

      if !Options.bottom_up then
        Logger.verbose Format.err_formatter "Cannot run the query in bottom-up mode, using top-down evaluator@\n@?";
      if !Options.count_only then
        let module R = ResJIT.Count in
        let module M = Runtime.Make(R) in
        if !Options.twopass then
          mk_runtime M.twopass_top_down_run auto v Tree.root R.NS.length R.NS.serialize None
        else
          mk_runtime M.top_down_run auto v Tree.root R.NS.length R.NS.serialize !Options.output_file
      else
        let module R = ResJIT.Mat in
        let module M = Runtime.Make(R) in
        mk_runtime M.top_down_run auto v Tree.root R.NS.length R.NS.serialize !Options.output_file
  in
  runtime ()
;;

let () = Options.parse_cmdline()
;;
let _ =
  try
    Printexc.record_backtrace true;
    let document =
      if Filename.check_suffix !Options.input_file ".srx"
      then
	time
	  ~msg:"Loading file"
	  (Tree.load
	     ~sample:!Options.sample_factor
	     ~load_text:(not !Options.disable_text_collection))
	  !Options.input_file
      else
	let v =
	  time
	    ~msg:"Parsing document"
	    (Tree.parse_xml_uri)
	    !Options.input_file
	in
        let () =
          if !Options.save_file <> ""
          then
	    time
	      ~msg:"Writing file to disk"
	      (Tree.save v)
	      !Options.save_file;
	in
	v
    in
    main document !Options.query !Options.output_file;
    Logger.verbose Format.err_formatter "Maximum resident set size: %s @\n" (read_procmem());
    Gc.full_major();
    Profile.summary Format.err_formatter
  with
  | Ulexer.Loc.Exc_located ((x,y),e) ->
    Logger.print Format.err_formatter "character %i-%i %s@\n" x y (Printexc.to_string e);
    exit 1

  | e ->
    Logger.print Format.err_formatter "BACKTRACE: %s@\n@?" (Printexc.get_backtrace());
    Logger.print Format.err_formatter "FATAL ERROR: %s@\n@?" (Printexc.to_string e);
    exit 2