X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=main.ml;h=c8ed40e1cfed841f0c95f99ba0da467a8ab10a13;hb=05d12de7ce75866e7e7e19fba37720fda99c38ad;hp=48f81bd78b1ae7639fedfff5f1460768ebbeac9e;hpb=95367aa932a9e179976e59ea326542c50905f5b3;p=SXSI%2Fxpathcomp.git diff --git a/main.ml b/main.ml index 48f81bd..c8ed40e 100644 --- a/main.ml +++ b/main.ml @@ -6,72 +6,135 @@ (******************************************************************************) INCLUDE "debug.ml" -open Automaton - +open Ata let l = ref [] ;; 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 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.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 "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 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 if !Options.backward then "(bottomup)" else ""); begin - match output with - | None -> () - | Some f -> - - Printf.eprintf "Serializing results : "; - time( fun () -> - let oc = open_out f in - output_string oc "\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()) + 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 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 + | None -> () + | Some f -> + Printf.eprintf "Serializing results : "; + time( fun () -> + let oc = open_out f in + output_string oc "\n"; + 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()) ;; -let argc = Array.length Sys.argv;; -if (argc < 3 || argc >4) -then - (prerr_endline ("usage : " ^ Sys.argv.(0) ^ " \'query\'[ ]"); - exit 1) -;; +Options.parse_cmdline();; +let v = + if (Filename.check_suffix !Options.input_file ".srx") + then + begin + Printf.eprintf "Loading from file : "; + time (Tree.load ~sample:!Options.sample_factor ) + (Filename.chop_suffix !Options.input_file ".srx"); + end + else + let v = + 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.save v) !Options.save_file; + end; + v +in + main v !Options.query !Options.output_file;; -main Sys.argv.(1) Sys.argv.(2) (if argc == 4 then Some Sys.argv.(3) else None) ;; - +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