X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=main.ml;h=7cc631b0f1e9c6b0191eeeb7728145f438f11007;hb=eda0e1062076c3343ab0cfdc10e4d19d6e23c570;hp=50dd048498a005aab27d8c06f8433d46bb720f18;hpb=5b4679e20761058f1e04c123da52631c0dd265cc;p=SXSI%2Fxpathcomp.git diff --git a/main.ml b/main.ml index 50dd048..7cc631b 100644 --- a/main.ml +++ b/main.ml @@ -21,6 +21,23 @@ let time f x = ;; let total_time () = List.fold_left (+.) 0. !l;; +let poa = TagSet.add Tag.pcdata (TagSet.singleton Tag.attribute) + +let rec fill_hashtag t = + if Tree.Binary.is_node t then + begin + let tag = Tree.Binary.tag t in + let a = + if TagSet.mem tag poa + then 0 + else + fill_hashtag (Tree.Binary.first_child t) + in + let b = fill_hashtag (Tree.Binary.next_sibling t) + in a+b+1 + end + else 0 + let test_slashslash tree k = let test = @@ -59,7 +76,7 @@ let test_jump tree k = if Tree.Binary.is_node tree then let acc = TS.cons tree acc in - loop acc (Tree.Binary.tagged_next tree ttag) + loop acc (Tree.Binary.tagged_foll tree ttag) else acc @@ -80,11 +97,10 @@ let test_traversal tree k = let l = Tree.Binary.first_child t and r = Tree.Binary.next_sibling t in - let _ = Printf.eprintf "Tree with id %i and tag=%s, tagged_desc %s is %i tagged_foll is %i, tagged_next is %i\n%!" + let _ = Printf.eprintf "Tree with id %i and tag=%s, tagged_desc %s is %i tagged_foll is %i\n%!" (Tree.Binary.id t) (Tag.to_string tag) (k) (iid (Tree.Binary.tagged_desc t ttag)) (iid (Tree.Binary.tagged_foll t ttag)) - (iid (Tree.Binary.tagged_next t ttag)) in aux l; aux r; @@ -93,6 +109,8 @@ let test_traversal tree k = () in aux tree + + let test_count_subtree tree k = let ttag = Tag.tag k in @@ -100,6 +118,80 @@ let test_count_subtree tree k = let r = time(Tree.Binary.subtree_tags tree) ttag in Printf.eprintf "%i nodes \n%!" r + +let test_contains tree s = + let _ = Printf.eprintf "Fetching DocIds containing %s ... %!" s in + time (fun s -> let r = Tree.Binary.contains tree s in + Tree.Binary.DocIdSet.iter + (fun t -> output_string stderr + (Tree.Binary.get_string tree t); + output_char stderr '\n') r ) s + + +let test_count_contains tree s = + let _ = Printf.eprintf "Counting DocIds containing %s ... %!" s in + let r = time (Tree.Binary.count_contains tree) s in + Printf.eprintf "%i documents ids\n%!" (r) + +let test_contains_old tree s = + let _ = Printf.eprintf "Fetching (old) DocIds containing %s ... %!" s in + let r = time (Tree.Binary.contains_old tree) s in + Printf.eprintf "%i documents ids\n%!" (Tree.Binary.DocIdSet.cardinal r) + +let test_contains_iter tree s = + let _ = Printf.eprintf "Fetching (old) DocIds containing %s ... %!" s in + let r = time (Tree.Binary.contains_iter tree) s in + Printf.eprintf "%i documents ids\n%!" (Tree.Binary.DocIdSet.cardinal r) + +module Stack = +struct + type t = { mutable table: Tree.Binary.t array; + mutable top : int } + + let empty = { table = Array.make 0 (Obj.magic 0); + top = 0 } + let cons e s = + let ls = Array.length s.table in + if ls > s.top + then + begin + s.table.(s.top) <- e; + s.top <- s.top + 1; + s + end + else + + let a = Array.make (ls * 2 + 1) (Tree.Binary.root e) + in + Array.blit s.table 0 a 0 ls; + s.table <- a; + s.table.(s.top) <- e; + s.top <- s.top + 1; + s + +end + +let test_fast tree = + let rec aux t acc = + if Tree.Binary.is_node t + then + aux (Tree.Binary.right t)( aux (Tree.Binary.left t) (Stack.cons t acc)) + else acc + in let _ = Printf.eprintf "Fast traversal ...%!" in + time (aux tree) Stack.empty + +let test_cps tree = + let rec aux t acc cont = + if Tree.Binary.is_node t + then aux (Tree.Binary.left t) (Stack.cons t acc) ((Tree.Binary.right t)::cont) + else match cont with + | [] -> acc + | p::r -> aux p acc r + in + let _ = Printf.eprintf "CPS traversal ...%!" in + time (aux tree Stack.empty) [] + + let main v query output = let _ = Tag.init (Tree.Binary.tag_pool v) in Printf.eprintf "Parsing query : "; @@ -111,36 +203,38 @@ let main v query output = in XPath.Ast.print Format.err_formatter query; Format.fprintf Format.err_formatter "\n%!"; - Printf.eprintf "Compiling query : "; - let auto = time XPath.Compile.compile query in - - let _ = Ata.dump Format.err_formatter auto ; - Format.fprintf Format.err_formatter "\n%!" - in - let _ = test_count_subtree v "keyword" in - let _ = test_jump v "keyword" in - - Printf.eprintf "Execution time : "; - let result = time (BottomUpNew.run auto) v in - Printf.eprintf "Number of nodes in the result set : %i\n" (TS.length result); - begin - match output with - | None -> () - | Some f -> - - Printf.eprintf "Serializing results : "; - time( fun () -> - let oc = open_out f in - output_string oc "\n"; - TS.iter (fun t -> Tree.Binary.print_xml_fast oc t; +(* 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 + Printf.eprintf "Execution time %s : " (if !Options.count_only then "(counting only)" else ""); + begin + if !Options.count_only then + let result = time (BottomUpNew.run_count auto) v + in + Printf.eprintf "Number of nodes in the result set : %i\n" result + else + let result = time (BottomUpNew.run auto) v in + Printf.eprintf "Number of nodes in the result set : %i\n" (TS.length result); + begin + match output with + | None -> () + | Some f -> + Printf.eprintf "Serializing results : "; + time( fun () -> + let oc = open_out f in + output_string oc "\n"; + TS.iter (fun t -> output_string oc "----------\n"; + Tree.Binary.print_xml_fast oc t; output_char oc '\n') result) (); - end; - (* let _ = Ata.dump Format.err_formatter auto in - Format.fprintf Format.err_formatter "\n%!"; *) - Printf.eprintf "Total time : %fms\n%!" (total_time()) + end; + end; + Printf.eprintf "Total running time : %fms\n%!" (total_time()) ;; - Options.parse_cmdline();; let v =