4 let index_empty_texts = ref true
5 let sample_factor = ref 64
6 let disable_text_collection = ref false
7 let tc_threshold = ref 60000
10 let input_file = ref ""
11 let output_file = ref None
12 let save_file = ref ""
13 let count_only = ref false
15 let bottom_up = ref false
16 let no_jump = ref false
17 let no_cache = ref false
18 let verbose = ref false
19 let text_index_type = ref 0
20 let do_perf = ref false
21 let twopass = ref false
23 let docstats = ref false
24 let no_wrap_results = ref false
26 let set_index_type = function
27 | "default" -> text_index_type := 0
28 | "swcsa" -> text_index_type := 1
29 | "rlcsa" -> text_index_type := 2
30 | s -> raise (Arg.Bad(s))
32 let usage_msg = Printf.sprintf "%s [options] <input.{xml|srx}> 'query' [output]" Sys.argv.(0)
36 fun s -> match !pos with
37 | 0 -> input_file:= s;incr pos
38 | 1 -> query := s; incr pos
39 | 2 -> output_file := Some s; incr pos
40 | _ -> raise (Arg.Bad(s))
46 match String.explode t ':' with
48 let l = try int_of_string lvl with _ -> raise (Arg.Bad (lvl)) in
49 if Logger.is_logger tr then Logger.activate tr l
50 else raise (Arg.Bad (t))
51 | _ -> raise (Arg.Bad (t))
52 ) (String.explode s ',')
54 let pretty_loggers () =
55 ignore(flush_str_formatter());
58 (fun f s -> fprintf f "%s" s)
60 (Logger.available ());
61 flush_str_formatter ()
64 [ "-c", Arg.Set(count_only),
65 " counting only (don't materialize the result set)";
67 "-two", Arg.Set(twopass),
68 " Use twopass algorithm";
70 "-f", Arg.Set_int(sample_factor),
71 "<n> sample factor [default=64]";
73 "-ne", Arg.Clear(index_empty_texts),
74 " don't index empty texts [default=index]";
76 "-d", Arg.Set(disable_text_collection),
77 " disable text collection[default=false]";
79 "-s", Arg.Set_string(save_file),
80 "<save_file> save the intermediate representation into file.srx";
82 "-b", Arg.Set(bottom_up), " real bottom up run";
84 "-nj", Arg.Set(no_jump), " disable jumping";
86 "-nc", Arg.Set(no_cache), " disable caching";
88 "-nw", Arg.Set(no_wrap_results), " do not wrap results in <xml_results/>";
90 "-p", Arg.Set(do_perf), " dump perf counters (Linux only)";
92 "-index-type", Arg.Symbol ([ "default"; "swcsa"; "rlcsa" ],
94 " choose text index type";
96 "-r", Arg.Set_int(repeat),
97 " repeat query execution n time (benchmarking only, default 1)";
99 "-doc-stats", Arg.Set(docstats),
100 " Compute document statistics (performs full traversal)";
102 "-v", Arg.Set(verbose), " verbose mode"; ] @
105 "-log", Arg.String (set_logger),
106 "<logger1:l1,...,loggern:ln> enable logging with the specified level. Valid loggers are: "
107 ^ (pretty_loggers ())
114 "-", Arg.Unit (fun () -> anon_fun "-"), "";
118 let parse_cmdline() =
121 Arg.parse_argv Sys.argv real_spec anon_fun usage_msg;
122 if (!pos > 3 || !pos < 2) then raise (Arg.Bad "Invalid number of arguments")
124 Arg.Bad s | Arg.Help s ->
125 eprintf "%s\n" (Arg.usage_string spec usage_msg);
128 Logger.set_verbose !verbose