X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=main.ml;h=ab043b7749872a731780c28e981492d7da1be9d4;hb=d4342e4bb9c853114de295567cd91ec86bb9e68f;hp=7cc631b0f1e9c6b0191eeeb7728145f438f11007;hpb=4a66518948bf6356b5cb72ba30b4d48a8c9e833a;p=SXSI%2Fxpathcomp.git diff --git a/main.ml b/main.ml index 7cc631b..ab043b7 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 @@ -21,176 +20,6 @@ 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 = - 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_foll 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\n%!" - (Tree.Binary.id t) (Tag.to_string tag) (k) - (iid (Tree.Binary.tagged_desc t ttag)) - (iid (Tree.Binary.tagged_foll 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 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 @@ -209,16 +38,35 @@ let main v query output = time (fill_hashtag) v; *) Printf.eprintf "Compiling query : "; - let auto,_ = time XPath.Compile.compile query in + 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 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); + 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 -> () @@ -227,7 +75,7 @@ let main v query output = time( fun () -> let oc = open_out f in output_string oc "\n"; - TS.iter (fun t -> output_string oc "----------\n"; + TS.rev_iter (fun t -> output_string oc "----------\n"; Tree.Binary.print_xml_fast oc t; output_char oc '\n') result) (); end;