1 (******************************************************************************)
2 (* SXSI : XPath evaluator *)
3 (* Kim Nguyen (Kim.Nguyen@nicta.com.au) *)
4 (* Copyright NICTA 2008 *)
5 (* Distributed under the terms of the LGPL (see LICENCE) *)
6 (******************************************************************************)
10 let () = init_timer();;
12 let enabled_gc = Gc.get()
13 let disabled_gc = { Gc.get() with
14 Gc.max_overhead = 1000000;
15 Gc.space_overhead = 100 }
16 let hash x = 131*x/(x-1+1)
18 let test_loop tree tag =
19 let t' = Tree.tagged_descendant tree tag Tree.root in
20 let f = Hashtbl.create 4096
22 let jump t _ = Tree.tagged_following_below tree tag t Tree.root in
24 if t == Tree.nil then 0
25 else 1+ ((Hashtbl.find f (hash 101)) (jump t ctx) ctx)
27 Hashtbl.add f (hash 101) g;
28 (Hashtbl.find f (hash 101)) t' Tree.root
31 let root = Tree.root in
32 let fin = Tree.closing tree root in
33 let rec loop t = if t <= fin then
34 let tag = Tree.tag tree t in
35 (* let _ = Tag.to_string tag in *)
36 if tag == Tag.pcdata then (ignore (Tree.get_text tree t));
37 let t = (Obj.magic ((Obj.magic t) + 1)) in
43 let test_loop2 tree tag =
44 let t' = Tree.tagged_descendant tree tag Tree.root in
45 let f = Hashtbl.create 4096
47 let jump t _ = Tree.tagged_following_below tree tag t Tree.root in
49 if t == Tree.nil then 0
50 else 1+ (match (Hashtbl.find f (hash 101)) with
51 `Foo ->g (jump t ctx) ctx
54 Hashtbl.add f (hash 101) `Foo;
58 external build_pointers : Tree.t -> pointers = "caml_build_pointers"
59 external iter_pointers : pointers -> int = "caml_iter_pointers"
60 external free_pointers : pointers -> unit = "caml_free_pointers"
63 let main v query_string output =
65 let _ = Tag.init (Tree.tag_pool v) in
66 Printf.eprintf "Parsing query : ";
69 XPath.Parser.parse_string query_string
71 Ulexer.Loc.Exc_located ((x,y),e) -> Printf.eprintf "character %i-%i %s\n" x y (Printexc.to_string e);exit 1
73 let _ = Printf.eprintf "Number of nodes %i\n%!" (Tree.size v) in
74 (* let _ = Tree.stats v in
75 let _ = Printf.eprintf "Timing first_child/next_sibling %!" in
76 let _ = time (Tree.benchmark_fcns) v in
77 let _ = Printf.eprintf "Timing last_child/prev_sibling %!" in
78 let _ = time (Tree.benchmark_lcps) v in
79 let _ = Printf.eprintf "Timing jump to a %!" in
80 let _ = time (Tree.benchmark_jump v) (Tag.tag "a") in
81 let _ = Printf.eprintf "Timing pointer allocation %!" in
82 let pointers = time (build_pointers) v in
83 let _ = Printf.eprintf "Timing pointer iteration %!" in
84 let i = time (iter_pointers) pointers in
85 let _ = Printf.eprintf "Traversed %i pointers\n\nTiming pointer deallocation %!" i in
86 let _ = time (free_pointers) pointers in *)
87 (* let _ = Printf.eprintf "Timing //keyword :" in
88 let r = time (test_loop v) (Tag.tag "keyword") in
89 let _ = Printf.eprintf "Count is %i\n%!" r in
90 let _ = Printf.eprintf "Timing //keyword 2:" in
91 let r = time (test_loop2 v) (Tag.tag "keyword") in
92 let _ = Printf.eprintf "Count is %i\n%!" r in
93 let _ = Printf.eprintf "Timing //node() :" in
94 let _ = time (test_full) v in *)
95 XPath.Ast.print Format.err_formatter query;
96 Format.fprintf Format.err_formatter "\n%!";
97 Printf.eprintf "Compiling query : ";
98 let auto,ltags,contains = time (XPath.Compile.compile ~querystring:query_string) query in
99 let _ = Ata.dump Format.err_formatter auto in
100 let _ = Printf.eprintf "%!" in
103 None -> (max_int,`NOTHING)
105 let r = Tree.count v s
107 Printf.eprintf "%i documents in the TextCollection\n" (Tree.text_size v);
108 Printf.eprintf "Global count is %i, using " r;
109 if r < !Options.tc_threshold then begin
110 Printf.eprintf "TextCollection contains\nCalling global contains : ";
111 time (Tree.init_textfun op v) s;
114 Printf.eprintf "Naive contains\nCalling global contains : ";
115 time (Tree.init_naive_contains v) s
118 let test_list = jump_to in
121 if (!Options.backward) then begin
122 Printf.eprintf "Finding min occurences : ";
124 ( List.fold_left (fun ((min_occ,kind)as acc) (tag,_) ->
125 let numtags = Tree.subtree_tags v tag Tree.root in
126 if ((numtags < min_occ) && numtags >= 2)
127 then (numtags,`TAG(tag))
128 else acc) jump_to) ltags
130 else (max_int,`NOTHING)
132 let _ = if (snd test_list) != `NOTHING then
133 let occ,s1,s2 = match test_list with
134 | (x,`TAG (tag)) -> (x, "tag", (Tag.to_string tag))
135 | (x,`CONTAINS(s)) -> (x, "contains()", s)
138 Printf.eprintf "Will jump to %s %s occuring %i time\n%!" s1 s2 occ
140 Printf.eprintf "Execution time %s : "
141 (if !Options.count_only then "(counting only)" else if !Options.backward then "(bottomup)" else "");
143 let _ = Gc.full_major();Gc.compact() in
144 let _ = Printf.eprintf "%!" in
145 (* let _ = Gc.set (disabled_gc) in *)
146 if !Options.backward && ((snd test_list) != `NOTHING )then
147 if !Options.count_only then
148 let r = time_mem (bottom_up_count auto v )(snd test_list) in
149 let _ = Printf.eprintf "Number of nodes in the result set : %i\n%!" r
152 let r = time_mem (bottom_up auto v )(snd test_list) in
153 let _ = Printf.eprintf "Number of nodes in the result set : %i\n%!" (IdSet.length r)
159 Printf.eprintf "Serializing results : ";
161 (*let oc = open_out f in *)
162 let oc = Unix.openfile f [ Unix.O_WRONLY;Unix.O_TRUNC;Unix.O_CREAT] 0o644 in
163 (*output_string oc "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";*)
165 Tree.print_xml_fast3 v t oc;
166 (*output_char oc '\n'; *)
172 if !Options.backward then Printf.eprintf "WARNING: couldn't find a jumping point, running top-down\n"
174 if !Options.count_only then
175 let r = time_mem ( top_down_count auto ) v in
176 let _ = Printf.eprintf "Number of nodes in the result set : %i\n%!" r
179 let module GR = Ata.Test(struct let doc = v end) in
180 let result = time_mem (GR.top_down auto) v in
181 let _ = Printf.eprintf "Counting results " in
182 let rcount = time (GR.Results.length) result in
183 Printf.eprintf "Number of nodes in the result set : %i\n" rcount;
184 Printf.eprintf "\n%!";
189 Printf.eprintf "Serializing results : ";
191 (*let oc = open_out f in *)
192 let oc = Unix.openfile f [ Unix.O_WRONLY;Unix.O_TRUNC;Unix.O_CREAT] 0o644 in
193 (*output_string oc "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";*)
194 let t1 = ref (Unix.gettimeofday()) in
196 let old_count = ref 1 in
197 GR.Results.iter (fun t ->
200 if (!count mod 15) == 0
202 let t2 = Unix.gettimeofday() in
203 let _ = Printf.eprintf "Printing %i elements in %f ms\n"
204 (!count - !old_count) (1000. *.(t2 -. !t1))
206 ( old_count := !count; t1 := Unix.gettimeofday())
208 Tree.print_xml_fast3 v t oc;
209 (*output_char oc '\n'; *)
213 let _ = Gc.set enabled_gc in
214 Printf.eprintf "Total running time : %fms\n%!" (total_time())
217 Options.parse_cmdline();;
220 if (Filename.check_suffix !Options.input_file ".srx")
223 Printf.eprintf "Loading from file : ";
224 time (Tree.load ~sample:!Options.sample_factor ~load_text:(not !Options.count_only))
229 time (fun () -> let v = Tree.parse_xml_uri !Options.input_file;
230 in Printf.eprintf "Parsing document : %!";v
233 if !Options.save_file <> ""
235 Printf.eprintf "Writing file to disk : ";
236 time (Tree.save v) !Options.save_file;
240 main v !Options.query !Options.output_file;;