main.ml

   1 (******************************************************************************)
   2 (*  SXSI : XPath evaluator                                                    *)
   3 (*  Kim Nguyen (Kim.Nguyen@nicta.com.au)                                      *)
   4 (*  Copyright NICTA 2008                                                      *)
   5 (*  Distributed under the terms of the LGPL (see LICENCE)                     *)
   6 (******************************************************************************)
   7
   8 open Ata
   9 INCLUDE "utils.ml"
  10 let () = init_timer();;
  11
  12 let enabled_gc = Gc.get()
  13 let disabled_gc = { Gc.get() with
  14                       Gc.max_overhead = 1000000;
  15                       Gc.space_overhead = 100 }
  16
  17
  18
  19
  20
  21 let main v query_string output =
  22
  23     let _ = Tag.init (Tree.tag_pool v) in
  24       Printf.eprintf "Parsing query : ";
  25       let query = try
  26         time
  27           XPath.Parser.parse_string query_string
  28       with
  29           Ulexer.Loc.Exc_located ((x,y),e) -> Printf.eprintf "character %i-%i %s\n" x y (Printexc.to_string e);exit 1
  30       in
  31         XPath.Ast.print Format.err_formatter query;
  32         Format.fprintf Format.err_formatter "\n%!";
  33         Printf.eprintf "Compiling query : ";
  34         let auto,ltags,contains = time (XPath.Compile.compile ~querystring:query_string) query in
  35         let _ = Ata.dump Format.err_formatter auto in
  36         let _ = Printf.eprintf "%!" in
  37         let jump_to =
  38           match contains with
  39               None -> (max_int,`NOTHING)
  40             | Some s ->
  41                 let r = Tree.count v s
  42                 in
  43                   Printf.eprintf "%i documents in the TextCollection\n" (Tree.text_size v);
  44                   Printf.eprintf "Global count is %i, using " r;
  45                   if r < !Options.tc_threshold then begin
  46                     Printf.eprintf "TextCollection contains\nCalling global contains : ";
  47                     time (Tree.init_contains v) s;
  48                   end
  49                   else begin
  50                     Printf.eprintf "Naive contains\nCalling global contains : ";
  51                     time (Tree.init_naive_contains v) s
  52                   end;(r,`CONTAINS(s))
  53         in
  54         let test_list = jump_to in
  55         (*
  56         let test_list =
  57           if (!Options.backward) then begin
  58             Printf.eprintf "Finding min occurences : ";
  59             time
  60               ( List.fold_left (fun ((min_occ,kind)as acc)  (tag,_) ->
  61                               let numtags = Tree.subtree_tags v tag Tree.root in
  62                                 if  ((numtags < min_occ) && numtags >= 2)
  63                                 then (numtags,`TAG(tag))
  64                                 else acc) jump_to) ltags
  65           end
  66           else (max_int,`NOTHING)
  67         in*)
  68         let _ = if (snd test_list) != `NOTHING then
  69           let occ,s1,s2 = match test_list with
  70             | (x,`TAG (tag)) -> (x, "tag", (Tag.to_string tag))
  71             | (x,`CONTAINS(s)) -> (x, "contains()", s)
  72             | _ -> assert false
  73           in
  74             Printf.eprintf "Will jump to %s %s occuring %i time\n%!" s1 s2 occ
  75         in
  76           Printf.eprintf "Execution time %s : "
  77             (if !Options.count_only then "(counting only)" else if !Options.backward then "(bottomup)" else "");
  78           begin
  79             let _ = Gc.full_major();Gc.compact() in
  80             let _ = Printf.eprintf "%!" in
  81               (*            let _ = Gc.set (disabled_gc) in  *)
  82               if !Options.backward && ((snd test_list) != `NOTHING )then
  83
  84                 let r = time (bottom_up_count auto v )(snd test_list)  in
  85                 let _ = Printf.eprintf "Number of nodes in the result set : %i\n%!" r
  86                 in ()
  87               else
  88                 let _ =
  89                   if !Options.backward then Printf.eprintf "WARNING: couldn't find a jumping point, running top-down\n"
  90                 in
  91                 if !Options.count_only then
  92                   let r = time ( top_down_count auto ) v in
  93                   let _ = Printf.eprintf "Number of nodes in the result set : %i\n%!" r
  94                   in ()
  95                 else
  96                   let result = time (top_down auto) v in
  97                   let rcount = GResult.length result in
  98                     Printf.eprintf "Number of nodes in the result set : %i\n" rcount;
  99                     Printf.eprintf "\n%!";
 100                     begin
 101                       match output with
 102                         | None -> ()
 103                         | Some f ->
 104                             Printf.eprintf "Serializing results : ";
 105                             time( fun () ->
 106                                     let oc = open_out f in
 107                                       output_string oc "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";
 108                                       GResult.iter (fun t ->
 109                                                     Tree.print_xml_fast oc v t;
 110                                                     output_char oc '\n';
 111
 112                                                  ) result) ();
 113                     end;
 114           end;
 115           let _ = Gc.set enabled_gc in
 116             Printf.eprintf "Total running time : %fms\n%!" (total_time())
 117 ;;
 118
 119 Options.parse_cmdline();;
 120
 121 let v =
 122   if (Filename.check_suffix !Options.input_file ".srx")
 123   then
 124     begin
 125       Printf.eprintf "Loading from file : ";
 126       time (Tree.load  ~sample:!Options.sample_factor )
 127         !Options.input_file;
 128         end
 129   else
 130     let v =
 131       time (fun () -> let v = Tree.parse_xml_uri !Options.input_file;
 132             in Printf.eprintf "Parsing document : %!";v
 133            ) ()
 134     in
 135       if !Options.save_file <> ""
 136       then begin
 137         Printf.eprintf "Writing file to disk : ";
 138         time (Tree.save v) !Options.save_file;
 139       end;
 140       v
 141 in
 142   main v !Options.query !Options.output_file;;
 143