Safety before Techfest

[SXSI/xpathcomp.git] / main.ml

diff --git a/main.ml b/main.ml
index dc2cb3e..c889833 100644 (file)
--- a/main.ml
+++ b/main.ml
@@ -81,14 +81,14 @@ let main v query_string output =
       let jump_to = 
        match contains with
           None -> (max_int,`NOTHING)
-         | Some s -> 
+         | Some ((op,s)) -> 
              let r = Tree.count v s 
              in
              Printf.eprintf "%i documents in the TextCollection\n" (Tree.text_size v);
              Printf.eprintf "Global count is %i, using " r;
              if r < !Options.tc_threshold then begin             
                Printf.eprintf "TextCollection contains\nCalling global contains : ";
-               time (Tree.init_contains v) s;
+               time (Tree.init_textfun op v) s;
              end
              else begin
                Printf.eprintf "Naive contains\nCalling global contains : ";
@@ -124,21 +124,40 @@ let main v query_string output =
            let _ = Printf.eprintf "%!" in
 (*         let _ = Gc.set (disabled_gc) in *)
              if !Options.backward && ((snd test_list) != `NOTHING )then 
-               
-               let r = time (bottom_up_count auto v )(snd test_list)  in
+               if !Options.count_only then
+               let r = time_mem (bottom_up_count auto v )(snd test_list)  in
                let _ = Printf.eprintf "Number of nodes in the result set : %i\n%!" r 
                in ()
+               else begin
+               let r = time_mem (bottom_up auto v )(snd test_list)  in
+               let _ = Printf.eprintf "Number of nodes in the result set : %i\n%!" (IdSet.length r)
+               in
+                 match output with
+
+                   | None -> ()
+                   | Some f ->               
+                       Printf.eprintf "Serializing results : ";
+                       time( fun () ->
+                               (*let oc = open_out f in *)
+                               let oc = Unix.openfile f [ Unix.O_WRONLY;Unix.O_TRUNC;Unix.O_CREAT] 0o644 in
+                               (*output_string oc "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";*)
+                               IdSet.iter (fun t -> 
+                                             Tree.print_xml_fast3 v t oc;
+                                             (*output_char oc '\n'; *)                         
+                                          ) r) ();
+               end
+                    
              else
                let _ = 
                  if !Options.backward then Printf.eprintf "WARNING: couldn't find a jumping point, running top-down\n" 
                in
                if !Options.count_only then
-                 let r = time ( top_down_count auto ) v in 
+                 let r = time_mem ( top_down_count auto ) v in 
                  let _ = Printf.eprintf "Number of nodes in the result set : %i\n%!" r
                  in ()
                else      
                let module GR = Ata.Test(struct let doc = v end) in
-                 let result = time (GR.top_down auto) v in
+                 let result = time_mem (GR.top_down auto) v in
                  let _ = Printf.eprintf "Counting results " in
                  let rcount = time (GR.Results.length) result in
                    Printf.eprintf "Number of nodes in the result set : %i\n" rcount;
@@ -152,10 +171,23 @@ let main v query_string output =
                                    (*let oc = open_out f in *)
                                    let oc = Unix.openfile f [ Unix.O_WRONLY;Unix.O_TRUNC;Unix.O_CREAT] 0o644 in
                                      (*output_string oc "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";*)
+                                   let t1 = ref (Unix.gettimeofday()) in
+                                   let count = ref 1 in
+                                   let old_count = ref 1 in
                                     GR.Results.iter (fun t -> 
-                                                 Tree.print_xml_fast3 v t oc;
+                                                       incr count;
+                                                       begin
+                                                         if (!count mod 15) == 0
+                                                         then
+                                                         let t2 =  Unix.gettimeofday() in
+                                                         let _ = Printf.eprintf "Printing %i elements in %f ms\n"
+                                                           (!count - !old_count) (1000. *.(t2 -. !t1))
+                                                         in
+                                                         ( old_count := !count; t1 := Unix.gettimeofday())
+                                                       end;
+                                                       Tree.print_xml_fast3 v t oc;
                                                        (*output_char oc '\n'; *)                               
-                                              ) result) ();
+                                                    ) result) ();
                    end;
          end;
          let _ = Gc.set enabled_gc in
@@ -169,7 +201,7 @@ let v =
   then 
     begin
       Printf.eprintf "Loading from file : ";
-      time (Tree.load  ~sample:!Options.sample_factor )
+      time (Tree.load  ~sample:!Options.sample_factor ~load_text:(not !Options.count_only))
        !Options.input_file;
        end
   else