libgrew.ml 8.4 KB
Newer Older
bguillaum's avatar
bguillaum committed
1 2 3 4 5 6 7 8 9 10
(**********************************************************************************)
(*    Libcaml-grew - a Graph Rewriting library dedicated to NLP applications      *)
(*                                                                                *)
(*    Copyright 2011-2013 Inria, Université de Lorraine                           *)
(*                                                                                *)
(*    Webpage: http://grew.loria.fr                                               *)
(*    License: CeCILL (see LICENSE folder or "http://www.cecill.info")            *)
(*    Authors: see AUTHORS file                                                   *)
(**********************************************************************************)

pj2m's avatar
pj2m committed
11 12
include Grew_types

bguillaum's avatar
bguillaum committed
13
open Printf
pj2m's avatar
pj2m committed
14 15
open Log

16
open Grew_fs
17
open Libgrew_utils
bguillaum's avatar
bguillaum committed
18 19 20 21
open Grew_graph
open Grew_rule
open Grew_grs

pj2m's avatar
pj2m committed
22
open Grew_parser
bguillaum's avatar
bguillaum committed
23
open Grew_html
pj2m's avatar
pj2m committed
24 25


26 27 28 29 30 31
let css_file = Filename.concat DATA_DIR "style.css"

let empty_grs = Grs.empty

let set_timeout t = Timeout.timeout := t

bguillaum's avatar
bguillaum committed
32

pj2m's avatar
pj2m committed
33 34
exception File_dont_exists of string

bguillaum's avatar
bguillaum committed
35
exception Parsing_err of string
pj2m's avatar
pj2m committed
36 37
exception Build of string * (string * int) option
exception Run of string * (string * int) option
bguillaum's avatar
bguillaum committed
38
exception Bug of string * (string * int) option
pj2m's avatar
pj2m committed
39

40 41
let handle ?(name="") ?(file="No file defined") fct () =
  try fct () with
42 43 44 45 46 47 48
    (* Raise again already catched exceptions *)
    | Parsing_err msg -> raise (Parsing_err msg)
    | Build (msg,loc) -> raise (Build (msg,loc))
    | Bug (msg, loc) -> raise (Bug (msg,loc))
    | Run (msg, loc) -> raise (Run (msg,loc))

    (* Catch new exceptions *)
49 50 51 52 53 54 55
    | Grew_parser.Parse_error (msg,Some (sub_file,l)) ->
        raise (Parsing_err (sprintf "[file:%s, line:%d] %s" sub_file l msg))
    | Grew_parser.Parse_error (msg,None) ->
        raise (Parsing_err (sprintf "[file:%s] %s" file msg))
    | Error.Build (msg,loc) -> raise (Build (msg,loc))
    | Error.Bug (msg, loc) -> raise (Bug (msg,loc))
    | Error.Run (msg, loc) -> raise (Run (msg,loc))
56

57
    | exc -> raise (Bug (sprintf "[Libgrew.%s] UNCATCHED EXCEPTION: %s" name (Printexc.to_string exc), None))
pj2m's avatar
pj2m committed
58

59
let is_empty rh =
bguillaum's avatar
bguillaum committed
60
  handle ~name:"is_empty" (fun () -> Rewrite_history.is_empty rh) ()
bguillaum's avatar
bguillaum committed
61

62 63
let num_sol rh =
  handle ~name:"num_sol" (fun () -> Rewrite_history.num_sol rh) ()
pj2m's avatar
pj2m committed
64

bguillaum's avatar
bguillaum committed
65

66
IFDEF DEP2PICT THEN
67
let build_html_doc ?(corpus=false) dir grs =
68
  handle ~name:"build_doc [with Dep2pict]"
69
    (fun () ->
70
      Html_doc.build ~corpus ~dep:true dir grs;
71 72 73 74

      (* draw pattern graphs for all rules and all filters *)
      let fct module_ rule_ =
        let dep_code = Rule.to_dep rule_ in
75 76
        let dep_png_file = sprintf "%s/%s_%s-patt.png" dir module_ (Rule.get_name rule_) in
        ignore (Dep2pict.Dep2pict.fromDepStringToPng dep_code dep_png_file) in
77 78 79
      Grs.rule_iter fct grs;
      Grs.filter_iter fct grs
    ) ()
80
ELSE
81 82
let build_html_doc ?(corpus=false) dir grs =
  handle ~name:"build_doc [without Dep2pict]" (fun () -> Html_doc.build ~corpus ~dep:false grs_ dir) ()
83 84
END

85
let load_grs file =
86 87 88 89
  handle ~name:"load_grs" ~file
    (fun () ->
      if not (Sys.file_exists file)
      then raise (File_dont_exists file)
90
      else Grs.build file
91
    ) ()
bguillaum's avatar
bguillaum committed
92

bguillaum's avatar
bguillaum committed
93
let to_sentence ?main_feat gr =
94 95 96 97 98 99 100 101 102 103 104
  handle ~name:"to_sentence"
    (fun () ->
      let graph = gr.Instance.graph in
      G_graph.to_sentence ?main_feat graph
    ) ()

let get_sequence_names grs =
  handle ~name:"get_sequence_names"
    (fun () ->
      Grs.sequence_names grs
    ) ()
bguillaum's avatar
bguillaum committed
105 106

let load_gr file =
107 108 109 110 111
  if not (Sys.file_exists file)
  then raise (File_dont_exists file)
  else
    handle ~name:"load_gr" ~file
      (fun () ->
bguillaum's avatar
bguillaum committed
112 113
        let gr_ast = Grew_parser.gr_of_file file in
        Instance.from_graph (G_graph.build gr_ast)
114
      ) ()
pj2m's avatar
pj2m committed
115

bguillaum's avatar
bguillaum committed
116
let load_conll file =
117 118 119 120 121 122
  handle ~name:"load_conll" ~file
    (fun () ->
      let graph = G_graph.of_conll ~loc:(file,-1) (Conll.load file) in
      Instance.from_graph graph
    ) ()

bguillaum's avatar
bguillaum committed
123 124 125 126 127 128 129
let of_conll file_name line_list =
  handle ~name:"of_conll"
    (fun () ->
      let graph = G_graph.of_conll (Conll.parse file_name line_list) in
      Instance.from_graph graph
    ) ()

130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146
let load_graph file =
  handle ~name:"load_graph" ~file
    (fun () ->
      if Filename.check_suffix file ".gr"
      then load_gr file
      else if Filename.check_suffix file ".conll"
      then load_conll file
      else
        begin
          Log.fwarning "Unknown file format for input graph '%s', try to guess..." file;
          try load_gr file with
              Parsing_err _ ->
                try load_conll file with
                    Parsing_err _ ->
                      Log.fcritical "[Libgrew.load_graph] Cannot guess input file format of file '%s'. Use .gr or .conll file extension" file
        end
    ) ()
147

bguillaum's avatar
bguillaum committed
148
let xml_graph xml =
149
  handle ~name:"xml_graph" (fun () -> Instance.from_graph (G_graph.of_xml xml)) ()
bguillaum's avatar
bguillaum committed
150

bguillaum's avatar
bguillaum committed
151
let raw_graph instance =
152
  handle ~name:"raw_graph" (fun () -> G_graph.to_raw instance.Instance.graph) ()
bguillaum's avatar
bguillaum committed
153

154 155
let rewrite ~gr ~grs ~seq =
  handle ~name:"rewrite" (fun () -> Grs.rewrite grs seq gr) ()
156

bguillaum's avatar
bguillaum committed
157
let display ~gr ~grs ~seq =
158
  handle ~name:"display" (fun () -> Grs.build_rew_display grs seq gr) ()
bguillaum's avatar
bguillaum committed
159

160 161
let write_stat filename rew_hist =
  handle ~name:"write_stat" (fun () -> Gr_stat.save filename (Gr_stat.from_rew_history rew_hist)) ()
bguillaum's avatar
bguillaum committed
162

163 164 165
let write_annot static_dir annot_dir base_name_rew_hist_list =
  handle ~name:"write_annot" (fun () -> Html_annot.build static_dir annot_dir base_name_rew_hist_list) ()

bguillaum's avatar
bguillaum committed
166
let save_index ~dirname ~base_names =
167 168 169 170 171
  handle ~name:"save_index" (fun () ->
    let out_ch = open_out (Filename.concat dirname "index") in
    List.iter (fun f -> fprintf out_ch "%s\n" f) base_names;
    close_out out_ch
  ) ()
172

bguillaum's avatar
bguillaum committed
173
let save_graph_conll filename graph =
174 175 176 177 178
  handle ~name:"save_graph_conll" (fun () ->
    let out_ch = open_out filename in
    fprintf out_ch "%s" (Instance.to_conll graph);
    close_out out_ch
  ) ()
bguillaum's avatar
bguillaum committed
179

180 181
let save_gr base rew_hist =
  handle ~name:"save_gr" (fun () -> Rewrite_history.save_gr base rew_hist) ()
182

183 184
let save_conll base rew_hist =
  handle ~name:"save_conll" (fun () -> Rewrite_history.save_conll base rew_hist) ()
bguillaum's avatar
bguillaum committed
185

186 187
let save_det_gr base rew_hist =
  handle ~name:"save_det_gr" (fun () -> Rewrite_history.save_det_gr base rew_hist) ()
bguillaum's avatar
bguillaum committed
188

189
let save_det_conll ?header base rew_hist =
190
  handle ~name:"save_deeeet_conll" (fun () -> Rewrite_history.save_det_conll ?header base rew_hist) ()
191

192 193
let det_dep_string rew_hist =
  handle ~name:"det_dep_string" (fun () -> Rewrite_history.det_dep_string rew_hist) ()
bguillaum's avatar
bguillaum committed
194

bguillaum's avatar
bguillaum committed
195 196 197
let conll_dep_string ?keep_empty_rh rew_hist =
  handle ~name:"conll_dep_string" (fun () -> Rewrite_history.conll_dep_string ?keep_empty_rh rew_hist) ()

bguillaum's avatar
bguillaum committed
198
let write_html 
199 200
    ?(no_init=false)
    ?(out_gr=false)
201 202
    ?filter
    ?main_feat
203
    ?dot
bguillaum's avatar
bguillaum committed
204
    ~header
bguillaum's avatar
bguillaum committed
205
    ?graph_file
bguillaum's avatar
bguillaum committed
206 207
    rew_hist
    output_base =
208 209 210 211 212 213 214 215 216
  handle ~name:"write_html" (fun () ->
    ignore (
      Html_rh.build
        ?filter
        ?main_feat
        ?dot
        ~out_gr
        ~init_graph: (not no_init)
        ~header
bguillaum's avatar
bguillaum committed
217
        ?graph_file
218
        output_base rew_hist
bguillaum's avatar
bguillaum committed
219
    )
220
  ) ()
221 222

let error_html 
223 224 225
    ?(no_init=false) 
    ?main_feat 
    ?dot
226
    ~header
227 228
    msg 
    ?init
229
    output_base =
230 231 232 233 234 235 236 237
  handle ~name:"error_html" (fun () ->
    ignore (
      Html_rh.error
        ?main_feat
        ?dot
        ~init_graph: (not no_init)
        ~header
        output_base msg init
238
    )
239
  ) ()
240

241
let make_index ~title ~grs_file ~html ~grs ~seq ~input_dir ~output_dir ~base_names  =
242 243 244 245 246 247 248 249 250 251 252 253 254 255 256
  handle ~name:"make_index" (fun () ->
    let init = Corpus_stat.empty grs seq in
    let corpus_stat =
      List.fold_left
        (fun acc base_name ->
          Corpus_stat.add_gr_stat base_name (Gr_stat.load (Filename.concat output_dir (base_name^".stat"))) acc
        ) init base_names in
    Corpus_stat.save_html title grs_file input_dir output_dir corpus_stat
  ) ()

let html_sentences ~title = handle ~name:"html_sentences" (fun () -> Html_sentences.build ~title) ()

let graph_of_instance instance = handle ~name:"graph_of_instance" (fun () -> instance.Instance.graph) ()

let feature_names () =  handle ~name:"feature_names" (fun () -> Domain.feature_names ()) ()
bguillaum's avatar
bguillaum committed
257

258 259
let to_dot_graph ?main_feat ?(deco=G_deco.empty) graph =
  handle ~name:"to_dot_graph" (fun () -> G_graph.to_dot ?main_feat graph ~deco) ()
bguillaum's avatar
bguillaum committed
260

261 262
let to_dep_graph ?filter ?main_feat ?(deco=G_deco.empty) graph =
  handle ~name:"to_dep_graph" (fun () -> G_graph.to_dep ?filter ?main_feat ~deco graph) ()
pj2m's avatar
pj2m committed
263

264 265
let to_gr_graph graph =
  handle ~name:"to_gr_graph" (fun () -> G_graph.to_gr graph) ()
266

267 268
let to_conll_graph graph =
  handle ~name:"to_conll_graph" (fun () -> G_graph.to_conll graph) ()