X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=main.ml;h=c8ed40e1cfed841f0c95f99ba0da467a8ab10a13;hb=05d12de7ce75866e7e7e19fba37720fda99c38ad;hp=c446ef39f2eef515f619a41ed90e13490fdbe743;hpb=dc91851aaeac91a71eba2c266d0227adea0c5815;p=SXSI%2Fxpathcomp.git

diff --git a/main.ml b/main.ml
index c446ef3..c8ed40e 100644
--- a/main.ml
+++ b/main.ml
@@ -13,16 +13,19 @@ 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
 ;;
 let total_time () =  List.fold_left (+.) 0. !l;;
-
+let enabled_gc = Gc.get()
+let disabled_gc = { Gc.get() with
+		      Gc.max_overhead = 1000000; 
+		      Gc.space_overhead = 100 }
 
 let main v query output =
-    let _ = Tag.init (Tree.Binary.tag_pool v) in
+    let _ = Tag.init (Tree.tag_pool v) in
       Printf.eprintf "Parsing query : ";    
       let query = try
 	time
@@ -32,27 +35,56 @@ 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 -> ()
-	  | Some s -> Tree.Binary.init_contains v s
+	  
+	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\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;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
-	    if !Options.count_only then
-	      failwith "Count only not implemented in this version"
+	    let _ = Gc.full_major();Gc.compact() in
+	    let _ = Gc.set (disabled_gc) 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 _ = Gc.set ({ Gc.get() with Gc.max_overhead = 1000000; Gc.space_overhead = 100 }) in
-	      let result = time (if !Options.time then run_time auto else run auto) v in	  
-	      	Printf.eprintf "Number of nodes in the result set : %i\n" (TS.length result);
+	      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
  		match output with
@@ -62,11 +94,15 @@ 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.rev_iter (fun t -> output_string oc "----------\n";
-					   Tree.Binary.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
+(*	  let _ = Ata.dump Format.err_formatter auto in *)
 	  Printf.eprintf "Total running time : %fms\n%!" (total_time())
 ;;
 		
@@ -77,19 +113,19 @@ let v =
   then 
     begin
       Printf.eprintf "Loading from file : ";
-      time (Tree.Binary.load  ~sample:!Options.sample_factor )
-	(Filename.chop_suffix !Options.input_file ".srx");
-    end
+      time (Tree.load  ~sample:!Options.sample_factor )
+    (Filename.chop_suffix !Options.input_file ".srx");
+	end
   else 
     let v = 
-      time (fun () -> let v = Tree.Binary.parse_xml_uri !Options.input_file;
+      time (fun () -> let v = Tree.parse_xml_uri !Options.input_file;
 	    in Printf.eprintf "Parsing document : %!";v
 	   ) () 
     in
       if !Options.save_file <> ""
       then begin
 	Printf.eprintf "Writing file to disk : ";
-	time (Tree.Binary.save v) !Options.save_file;
+	time (Tree.save v) !Options.save_file;
       end;
       v
 in