X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=main.ml;h=985a5b2fba0334288536e403a981fa8910e1e4c6;hb=d2cbab156c2b74051cf831dc3ac634aedc8f2c2f;hp=50dd048498a005aab27d8c06f8433d46bb720f18;hpb=5b4679e20761058f1e04c123da52631c0dd265cc;p=SXSI%2Fxpathcomp.git diff --git a/main.ml b/main.ml index 50dd048..985a5b2 100644 --- a/main.ml +++ b/main.ml @@ -8,7 +8,6 @@ INCLUDE "debug.ml" open Ata - let l = ref [] ;; let time f x = let t1 = Unix.gettimeofday () in @@ -20,88 +19,13 @@ let time f x = r ;; let total_time () = List.fold_left (+.) 0. !l;; - - -let test_slashslash tree k = - let test = - match k with "*" -> TagSet.remove (Tag.tag "") TagSet.star - | s -> TagSet.singleton (Tag.tag k) - in - let attorstring = TagSet.cup TagSet.pcdata TagSet.attribute in - let rec aux t acc = - if Tree.Binary.is_node t - then - let tag = Tree.Binary.tag t in - let l = Tree.Binary.first_child t - and r = Tree.Binary.next_sibling t - in - let acc = - if TagSet.mem tag test - then - TS.append t acc - else - acc - in - let rl = if TagSet.mem tag attorstring then acc else aux l acc - in aux r rl - else - acc - in - let _ = Printf.eprintf "Testing optimal //%s ... " k in - let r = time (aux tree ) TS.empty in - Printf.eprintf "Result set is %i nodes\n%!" (TS.length r) - - -let test_jump tree k = - let ttag = Tag.tag k in - - let rec loop acc tree = - if Tree.Binary.is_node tree - then - let acc = TS.cons tree acc in - loop acc (Tree.Binary.tagged_next tree ttag) - else - acc - - in - let _ = Printf.eprintf "Testing jumping for tag %s ... " k in - let r = time (loop TS.empty ) (Tree.Binary.tagged_next tree ttag) in - Printf.eprintf "Result set is %i nodes\n%!" (TS.length r) - - - -let test_traversal tree k = - let ttag = Tag.tag k in - let iid t = if Tree.Binary.is_node t then Tree.Binary.id t else -1 in - let rec aux t = - if Tree.Binary.is_node t - then - let tag = Tree.Binary.tag t in - 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%!" - (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; - - else - () - in - aux tree - -let test_count_subtree tree k = - let ttag = Tag.tag k in - let _ = Printf.eprintf "Counting subtrees with tag %s ... %!" k in - let r = time(Tree.Binary.subtree_tags tree) ttag in - Printf.eprintf "%i nodes \n%!" r +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 @@ -111,36 +35,61 @@ 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%!" +(* 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 -> + let r = Tree.count v s + in + 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 + end + else begin + Printf.eprintf "Naive contains\nCalling global contains : "; + time (Tree.init_naive_contains v) s + end 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 "Execution time %s : " (if !Options.count_only then "(counting only)" 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 + 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 + Printf.eprintf "Number of nodes in the result set : %i\n" rcount; + 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 "\n"; + TS.iter (fun t -> output_string oc "----------\n"; + Tree.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; + let _ = Gc.set enabled_gc in +(* let _ = Ata.dump Format.err_formatter auto in *) + Printf.eprintf "Total running time : %fms\n%!" (total_time()) ;; - Options.parse_cmdline();; let v = @@ -148,19 +97,19 @@ let v = then begin Printf.eprintf "Loading from file : "; - time (Tree.Binary.load ~sample:!Options.sample_factor ) + 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