X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=main.ml;h=c88983303c0227f2919be9402640e019f22b1df8;hb=cf6d366b25132eea7b0f1966c11d034d748af0fa;hp=b30eff286dd6cd14f69de1f29ae8937fe01ab4c1;hpb=ede0fe9faff0bffe5f9f010d8942a057186d1ce1;p=SXSI%2Fxpathcomp.git

diff --git a/main.ml b/main.ml
index b30eff2..c889833 100644
--- a/main.ml
+++ b/main.ml
@@ -14,6 +14,7 @@ let disabled_gc = { Gc.get() with
 		      Gc.max_overhead = 1000000; 
 		      Gc.space_overhead = 100 }
 let hash x = 131*x/(x-1+1)
+
 let test_loop tree tag =
   let t' = Tree.tagged_desc tree tag  Tree.root in
   let f = Hashtbl.create 4096
@@ -25,6 +26,20 @@ let test_loop tree tag =
   in
   Hashtbl.add f (hash 101) g;
   (Hashtbl.find f (hash 101)) t' Tree.root
+
+let test_full tree = 
+  let root = Tree.root in
+  let fin = Tree.closing tree root in
+  let rec loop t = if t <= fin then
+  let tag = Tree.tag tree t in
+(*  let _ = Tag.to_string tag in *)
+  if tag == Tag.pcdata then (ignore (Tree.get_text tree t)); 
+  let t = (Obj.magic ((Obj.magic t) + 1)) in
+  loop t
+  in
+  loop root
+
+
 let test_loop2 tree tag =
   let t' = Tree.tagged_desc tree tag  Tree.root in
   let f = Hashtbl.create 4096
@@ -51,10 +66,12 @@ let main v query_string output =
       in
 (*      let _ = Printf.eprintf "Timing //keyword :" in
       let r = time (test_loop v) (Tag.tag "keyword") in
-      let _ = Printf.eprintf "Count is %i\n%!" r in
+      let _ = Printf.eprintf "Count is %i\n%!" r in 
       let _ = Printf.eprintf "Timing //keyword 2:" in
       let r = time (test_loop2 v) (Tag.tag "keyword") in
-      let _ = Printf.eprintf "Count is %i\n%!" r in *)
+      let _ = Printf.eprintf "Count is %i\n%!" r in  
+      let _ = Printf.eprintf "Timing //node() :" in
+      let _ = time (test_full)  v in      *)
       XPath.Ast.print Format.err_formatter query;
       Format.fprintf Format.err_formatter "\n%!";
       Printf.eprintf "Compiling query : ";
@@ -64,14 +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 : ";
@@ -105,23 +122,44 @@ let main v query_string output =
 	  begin
 	    let _ = Gc.full_major();Gc.compact() in
 	    let _ = Printf.eprintf "%!" in
-	    let _ = Gc.set (disabled_gc) 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 result = time (top_down auto) v in	  
-		  let rcount = IdSet.length result in
+		let module GR = Ata.Test(struct let doc = v end) 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;
 		    Printf.eprintf "\n%!";
 		    begin
@@ -130,13 +168,26 @@ let main v query_string output =
 			| Some f ->		      
 			    Printf.eprintf "Serializing results : ";
 			    time( fun () ->
-				    let oc = open_out f in
-				      output_string oc "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";				
-				      IdSet.iter (fun t -> 
-						    Tree.print_xml_fast oc v t;
-						    output_char oc '\n';
-
-						 ) result) ();
+				    (*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 -> 
+							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) ();
 		    end;
 	  end;
 	  let _ = Gc.set enabled_gc in
@@ -150,7 +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