bottom up run works for text nodes

[SXSI/xpathcomp.git] / main.ml

diff --git a/main.ml b/main.ml
index 985a5b2..c8ed40e 100644 (file)
--- a/main.ml
+++ b/main.ml
@@ -13,7 +13,7 @@ let time f x =
   let t1 = Unix.gettimeofday () in
   let r = f x in
   let t2 = Unix.gettimeofday () in 
-  let t = (1000. *.(t2 -. t1)) in
+  let t = (1000. *. (t2 -. t1)) in
     l:= t::!l;
     Printf.eprintf "  %fms\n%!" t ;
     r
@@ -35,41 +35,55 @@ let main v query output =
       in
        XPath.Ast.print Format.err_formatter query;
        Format.fprintf Format.err_formatter "\n%!";
-(*     Printf.eprintf "Dummy iteration : ";
-       time (fill_hashtag) v;
-       Printf.eprintf "Dummy iteration (tag access cached) : ";
-       time (fill_hashtag) v;
-*)
        Printf.eprintf "Compiling query : ";
        let auto,ltags,contains = time XPath.Compile.compile  query in 
        let _ = Ata.dump Format.err_formatter auto in
        let _ = Printf.eprintf "%!" in
-       let _ = match contains with
-           None -> ()
+         
+       let do_contains = match contains with
+           None -> false
          | Some 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
+               if r < !Options.tc_threshold then begin           
+                 Printf.eprintf "TextCollection contains\nTiming call to raw global contains (1st time): ";
+                 time (Tree.unsorted_contains v) s;              
+                 Printf.eprintf "Calling global contains : ";
+                 time (Tree.init_contains v) s;
+                 Printf.eprintf "Timing call to global count contains : ";
+                 let r = time (Tree.count_contains v) s
+                 in
+                   Printf.eprintf " number of matching nodes %i \n%!" r;
+                   Printf.eprintf "Timing call to raw global contains (2nd time): ";
+                   time (Tree.unsorted_contains v) s;
                end
                else begin
                  Printf.eprintf "Naive contains\nCalling global contains : ";
                  time (Tree.init_naive_contains v) s
-               end
+               end;true
        in
-         Printf.eprintf "Execution time %s : " (if !Options.count_only then "(counting only)" else "");
+         Printf.eprintf "Execution time %s : "
+           (if !Options.count_only then "(counting only)" else if !Options.backward then "(bottomup)" else "");
          begin
            let _ = Gc.full_major();Gc.compact() in
            let _ = Gc.set (disabled_gc) in 
-           if !Options.count_only then
-             let r = time ( run_count auto  )v in
+             if !Options.backward then
+               let tag,set = List.hd ltags in
+               let r = if do_contains 
+               then time (bottom_up_count_contains auto) v
+               else time (bottom_up_count auto v) tag in
+               let _ = Printf.eprintf "Number of nodes in the result set : %i\n%!" r 
+               in ()
+             else
+               if !Options.count_only then
+                 let r = time ( top_down_count auto ) v in (* not clean *)
              let _ = Printf.eprintf "Number of nodes in the result set : %i\n%!" r
              in ()
            else      
-
-             let result,rcount = time (if !Options.time then run_time auto else run auto) v in   
+             let result = time (top_down auto) v in      
+             let rcount = IdSet.length result in
                Printf.eprintf "Number of nodes in the result set : %i\n" rcount;
                Printf.eprintf "\n%!";
              begin
@@ -80,9 +94,11 @@ let main v query output =
                      time( fun () ->
                              let oc = open_out f in
                                output_string oc "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";                                
-                               TS.iter (fun t -> output_string oc "----------\n";
-                                          Tree.print_xml_fast oc t;
-                                          output_char oc '\n') result) ();
+                               IdSet.iter (fun t -> 
+                                             Tree.print_xml_fast oc t;
+                                             output_char oc '\n';
+                                             output_string oc "----------\n";
+                                          ) result) ();
              end;
          end;
          let _ = Gc.set enabled_gc in
@@ -98,8 +114,8 @@ let v =
     begin
       Printf.eprintf "Loading from file : ";
       time (Tree.load  ~sample:!Options.sample_factor )
-       (Filename.chop_suffix !Options.input_file ".srx");
-    end
+    (Filename.chop_suffix !Options.input_file ".srx");
+       end
   else 
     let v = 
       time (fun () -> let v = Tree.parse_xml_uri !Options.input_file;