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 (******************************************************************************)
12 let t1 = Unix.gettimeofday () in
14 let t2 = Unix.gettimeofday () in
15 let t = (1000. *. (t2 -. t1)) in
17 Printf.eprintf " %fms\n%!" t ;
20 let total_time () = List.fold_left (+.) 0. !l;;
21 let enabled_gc = Gc.get()
22 let disabled_gc = { Gc.get() with
23 Gc.max_overhead = 1000000;
24 Gc.space_overhead = 100 }
26 let main v query_string output =
27 let _ = Tag.init (Tree.tag_pool v) in
28 Printf.eprintf "Parsing query : ";
31 XPath.Parser.parse_string query_string
33 Ulexer.Loc.Exc_located ((x,y),e) -> Printf.eprintf "character %i-%i %s\n" x y (Printexc.to_string e);exit 1
35 XPath.Ast.print Format.err_formatter query;
36 Format.fprintf Format.err_formatter "\n%!";
37 Printf.eprintf "Compiling query : ";
38 let auto,ltags,contains = time (XPath.Compile.compile ~querystring:query_string) query in
39 let _ = Ata.dump Format.err_formatter auto in
40 let _ = Printf.eprintf "%!" in
43 None -> (max_int,`NOTHING)
45 let r = Tree.count v s
47 Printf.eprintf "%i documents in the TextCollection\n" (Tree.text_size v);
48 Printf.eprintf "Global count is %i, using " r;
49 if r < !Options.tc_threshold then begin
50 Printf.eprintf "TextCollection contains\nCalling global contains : ";
51 time (Tree.init_contains v) s;
54 Printf.eprintf "Naive contains\nCalling global contains : ";
55 time (Tree.init_naive_contains v) s
59 if (!Options.backward) then begin
60 Printf.eprintf "Finding min occurences : ";
62 ( List.fold_left (fun ((min_occ,kind)as acc) (tag,_) ->
63 let numtags = Tree.subtree_tags v tag Tree.root in
64 if ((numtags < min_occ) && numtags >= 2)
65 then (numtags,`TAG(tag))
66 else acc) jump_to) ltags
68 else (max_int,`NOTHING)
70 let _ = if (snd test_list) != `NOTHING then
71 let occ,s1,s2 = match test_list with
72 | (x,`TAG (tag)) -> (x, "tag", (Tag.to_string tag))
73 | (x,`CONTAINS(s)) -> (x, "contains()", s)
76 Printf.eprintf "Will jump to %s %s occuring %i time\n%!" s1 s2 occ
78 Printf.eprintf "Execution time %s : "
79 (if !Options.count_only then "(counting only)" else if !Options.backward then "(bottomup)" else "");
81 let _ = Gc.full_major();Gc.compact() in
82 (* let _ = Gc.set (disabled_gc) in *)
83 if !Options.backward && ((snd test_list) != `NOTHING )then
85 let r = time (bottom_up_count auto v )(snd test_list) in
86 let _ = Printf.eprintf "Number of nodes in the result set : %i\n%!" r
90 if !Options.backward then Printf.eprintf "WARNING: couldn't find a jumping point, running top-down\n"
92 if !Options.count_only then
93 let r = time ( top_down_count auto ) v in
94 let _ = Printf.eprintf "Number of nodes in the result set : %i\n%!" r
97 let result = time (top_down auto) v in
98 let rcount = IdSet.length result in
99 Printf.eprintf "Number of nodes in the result set : %i\n" rcount;
100 Printf.eprintf "\n%!";
105 Printf.eprintf "Serializing results : ";
107 let oc = open_out f in
108 output_string oc "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";
110 Tree.print_xml_fast oc v t;
112 output_string oc "----------\n";
116 let _ = Gc.set enabled_gc in
117 Printf.eprintf "Total running time : %fms\n%!" (total_time())
120 Options.parse_cmdline();;
123 if (Filename.check_suffix !Options.input_file ".srx")
126 Printf.eprintf "Loading from file : ";
127 time (Tree.load ~sample:!Options.sample_factor )
128 (Filename.chop_suffix !Options.input_file ".srx");
132 time (fun () -> let v = Tree.parse_xml_uri !Options.input_file;
133 in Printf.eprintf "Parsing document : %!";v
136 if !Options.save_file <> ""
138 Printf.eprintf "Writing file to disk : ";
139 time (Tree.save v) !Options.save_file;
143 main v !Options.query !Options.output_file;;