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
25 let set_index_type = function
26 | "default" -> text_index_type := 0
27 | "swcsa" -> text_index_type := 1
28 | "rlcsa" -> text_index_type := 2
29 | s -> raise (Arg.Bad(s))
31 let usage_msg = Printf.sprintf "%s [options] <input.{xml|srx}> 'query' [output]" Sys.argv.(0)
35 fun s -> match !pos with
36 | 0 -> input_file:= s;incr pos
37 | 1 -> query := s; incr pos
38 | 2 -> output_file := Some s; incr pos
39 | _ -> raise (Arg.Bad(s))
45 match String.explode t ':' with
47 let l = try int_of_string lvl with _ -> raise (Arg.Bad (lvl)) in
48 if Logger.is_logger tr then Logger.activate tr l
49 else raise (Arg.Bad (t))
50 | _ -> raise (Arg.Bad (t))
51 ) (String.explode s ',')
53 let pretty_loggers () =
54 ignore(flush_str_formatter());
57 (fun f s -> fprintf f "%s" s)
59 (Logger.available ());
60 flush_str_formatter ()
63 [ "-c", Arg.Set(count_only),
64 " counting only (don't materialize the result set)";
66 "-two", Arg.Set(twopass),
67 " Use twopass algorithm";
69 "-f", Arg.Set_int(sample_factor),
70 "<n> sample factor [default=64]";
72 "-ne", Arg.Clear(index_empty_texts),
73 " don't index empty texts [default=index]";
75 "-d", Arg.Set(disable_text_collection),
76 " disable text collection[default=false]";
78 "-s", Arg.Set_string(save_file),
79 "<save_file> save the intermediate representation into file.srx";
81 "-b", Arg.Set(bottom_up), " real bottom up run";
83 "-nj", Arg.Set(no_jump), " disable jumping";
85 "-nc", Arg.Set(no_cache), " disable caching";
88 "-p", Arg.Set(do_perf), " dump perf counters (Linux only)";
90 "-index-type", Arg.Symbol ([ "default"; "swcsa"; "rlcsa" ],
92 " choose text index type";
94 "-r", Arg.Set_int(repeat),
95 " repeat query execution n time (benchmarking only, default 1)";
97 "-doc-stats", Arg.Set(docstats),
98 " Compute document statistics (performs full traversal)";
101 "-v", Arg.Set(verbose), " verbose mode"; ] @
104 "-log", Arg.String (set_logger),
105 "<logger1:l1,...,loggern:ln> enable logging with the specified level. Valid loggers are: "
106 ^ (pretty_loggers ())
112 let parse_cmdline() =
113 let _ = Arg.parse spec anon_fun usage_msg
115 if (!pos > 3 || !pos < 2)
116 then begin Arg.usage spec usage_msg; exit 1 end