X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=main.ml;h=ab043b7749872a731780c28e981492d7da1be9d4;hb=d4342e4bb9c853114de295567cd91ec86bb9e68f;hp=48f81bd78b1ae7639fedfff5f1460768ebbeac9e;hpb=95367aa932a9e179976e59ea326542c50905f5b3;p=SXSI%2Fxpathcomp.git

diff --git a/main.ml b/main.ml
index 48f81bd..ab043b7 100644
--- a/main.ml
+++ b/main.ml
@@ -6,8 +6,7 @@
 (******************************************************************************)
 INCLUDE "debug.ml"
 
-open Automaton
-
+open Ata
 
 let l = ref [] ;;
 let time f x =
@@ -22,56 +21,97 @@ let time f x =
 let total_time () =  List.fold_left (+.) 0. !l;;
 
 
-let main filename query output =
-
-    (* Just a trick to allow the C++ code to print debugging stuff first *)  
-    let v = time (fun () -> let v = Tree.Binary.parse_xml_uri filename;
-		  in Printf.eprintf "Parsing document : %!";v
-		 ) () 
-    in
-      MM(v,__LOCATION__);
+let main v query output =
+    let _ = Tag.init (Tree.Binary.tag_pool v) in
       Printf.eprintf "Parsing query : ";    
       let query = try
 	time
 	  XPath.Parser.parse_string query
       with
 	  Ulexer.Loc.Exc_located ((x,y),e) -> Printf.eprintf "character %i-%i %s\n" x y (Printexc.to_string e);exit 1
-      in      
+      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 = time XPath.Compile.compile  query in
-	  XPath.Ast.print Format.err_formatter query;
-	  Format.eprintf "\n%!";
-	  (*		Format.eprintf "Internal rep of the tree is :\n%!";
-			Tree.Binary.dump v;			      *)
-	  Printf.eprintf "Execution time : ";
-	  time (fun v -> ignore (TopDown.accept auto v)) v;
-	  Printf.eprintf "Number of nodes in the result set : %i\n" (BST.cardinal auto.result);
+	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 -> 
+	      let r = Tree.Binary.count v s 
+	      in
+		Printf.eprintf "Global count is %i, using " r;
+		if r < 60000 then begin
+		  Printf.eprintf "TextCollection contains\nCalling global contains : ";
+		  time (Tree.Binary.init_contains v) s
+		end
+		else begin
+		  Printf.eprintf "Naive contains\nCalling global contains : ";
+		  time (Tree.Binary.init_naive_contains v) s
+		end
+	in
+	  Printf.eprintf "Execution time %s : " (if !Options.count_only then "(counting only)" else "");
 	  begin
- 	    match output with
-	      | None -> ()
-	      | 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";
-			  BST.iter (fun t -> Tree.Binary.print_xml_fast oc t;
-				      output_char oc '\n') auto.result) ();
-	end;
-	Printf.eprintf "Total time : %fms\n Coherence : %i\n%!" (total_time())
+	    if !Options.count_only then
+	      let r = time ( run_count auto  )v in
+	      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);
+		Printf.eprintf "\n%!";
+	      begin
+ 		match output with
+		  | None -> ()
+		  | 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";				
+				TS.rev_iter (fun t -> output_string oc "----------\n";
+					   Tree.Binary.print_xml_fast oc t;
+					   output_char oc '\n') result) ();
+	      end;
+	  end;
+	  Printf.eprintf "Total running time : %fms\n%!" (total_time())
 ;;
 		
-let argc = Array.length Sys.argv;;
-if (argc < 3 || argc >4)
-then
-  (prerr_endline ("usage : " ^ Sys.argv.(0) ^ " <document> \'query\'[ <output> ]");
-   exit 1)
-;;
-
+Options.parse_cmdline();;
 
-main Sys.argv.(1) Sys.argv.(2) (if argc == 4 then Some Sys.argv.(3) else None) ;; 
+let v = 
+  if (Filename.check_suffix !Options.input_file ".srx")
+  then 
+    begin
+      Printf.eprintf "Loading from file : ";
+      time (Tree.Binary.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;
+	    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;
+      end;
+      v
+in
+  main v !Options.query !Options.output_file;;
 
+IFDEF DEBUG
+THEN
 Printf.eprintf "\n=================================================\nDEBUGGING\n%!";
-Tree.DEBUGTREE.print_stats Format.err_formatter;;
-
 
+Tree.DEBUGTREE.print_stats Format.err_formatter;;
+Gc.full_major()
+ENDIF