libgrew.ml 9.25 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                                                   *)
(**********************************************************************************)

11
include Libgrew_types
pj2m's avatar
pj2m committed
12

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

16
open Grew_fs
17
open Grew_base
18 19
open Grew_types

bguillaum's avatar
bguillaum committed
20 21 22 23
open Grew_graph
open Grew_rule
open Grew_grs

pj2m's avatar
pj2m committed
24
open Grew_parser
bguillaum's avatar
bguillaum committed
25
open Grew_html
pj2m's avatar
pj2m committed
26 27


28 29 30 31 32 33
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
34 35
type loc = Loc.t
let string_of_loc = Loc.to_string
36
let line_of_loc = Loc.to_line
bguillaum's avatar
bguillaum committed
37

38
type instance = Instance.t
bguillaum's avatar
bguillaum committed
39

pj2m's avatar
pj2m committed
40 41
exception File_dont_exists of string

bguillaum's avatar
bguillaum committed
42 43 44 45
exception Parsing_err of string * loc option
exception Build of string * loc option
exception Run of string * loc option
exception Bug of string * loc option
pj2m's avatar
pj2m committed
46

47 48
let handle ?(name="") ?(file="No file defined") fct () =
  try fct () with
49
    (* Raise again already catched exceptions *)
bguillaum's avatar
bguillaum committed
50 51 52 53
    | Parsing_err (msg,loc_opt) -> raise (Parsing_err (msg,loc_opt))
    | Build (msg,loc_opt) -> raise (Build (msg,loc_opt))
    | Bug (msg, loc_opt) -> raise (Bug (msg,loc_opt))
    | Run (msg, loc_opt) -> raise (Run (msg,loc_opt))
54 55

    (* Catch new exceptions *)
bguillaum's avatar
bguillaum committed
56 57 58 59
    | Grew_parser.Parse_error (msg, loc_opt) -> raise (Parsing_err (msg, loc_opt))
    | Error.Build (msg, loc_opt) -> raise (Build (msg, loc_opt))
    | Error.Bug (msg, loc_opt) -> raise (Bug (msg,loc_opt))
    | Error.Run (msg, loc_opt) -> raise (Run (msg,loc_opt))
60

61
    | exc -> raise (Bug (sprintf "[Libgrew.%s] UNCATCHED EXCEPTION: %s" name (Printexc.to_string exc), None))
pj2m's avatar
pj2m committed
62

63
let is_empty rh =
bguillaum's avatar
bguillaum committed
64
  handle ~name:"is_empty" (fun () -> Rewrite_history.is_empty rh) ()
bguillaum's avatar
bguillaum committed
65

66 67
let num_sol rh =
  handle ~name:"num_sol" (fun () -> Rewrite_history.num_sol rh) ()
pj2m's avatar
pj2m committed
68

bguillaum's avatar
bguillaum committed
69

70
IFDEF DEP2PICT THEN
71
let build_html_doc ?(corpus=false) dir grs =
72
  handle ~name:"build_doc [with Dep2pict]"
73
    (fun () ->
74
      Html_doc.build ~corpus ~dep:true dir grs;
75 76 77 78

      (* draw pattern graphs for all rules and all filters *)
      let fct module_ rule_ =
        let dep_code = Rule.to_dep rule_ in
79 80
        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
81 82 83
      Grs.rule_iter fct grs;
      Grs.filter_iter fct grs
    ) ()
84
ELSE
85 86
let build_html_doc ?(corpus=false) dir grs =
  handle ~name:"build_doc [without Dep2pict]" (fun () -> Html_doc.build ~corpus ~dep:false grs_ dir) ()
87 88
END

89
let load_grs file =
90 91 92 93
  handle ~name:"load_grs" ~file
    (fun () ->
      if not (Sys.file_exists file)
      then raise (File_dont_exists file)
94
      else Grs.build file
95
    ) ()
bguillaum's avatar
bguillaum committed
96

bguillaum's avatar
bguillaum committed
97
let to_sentence ?main_feat gr =
98 99 100 101 102 103 104 105 106 107 108
  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
109 110

let load_gr file =
111 112 113 114 115
  if not (Sys.file_exists file)
  then raise (File_dont_exists file)
  else
    handle ~name:"load_gr" ~file
      (fun () ->
bguillaum's avatar
bguillaum committed
116 117
        let gr_ast = Grew_parser.gr_of_file file in
        Instance.from_graph (G_graph.build gr_ast)
118
      ) ()
pj2m's avatar
pj2m committed
119

bguillaum's avatar
bguillaum committed
120
let load_conll file =
121 122
  handle ~name:"load_conll" ~file
    (fun () ->
bguillaum's avatar
bguillaum committed
123
      let graph = G_graph.of_conll ~loc:(Loc.file file) (Conll.load file) in
124 125 126
      Instance.from_graph graph
    ) ()

bguillaum's avatar
bguillaum committed
127 128 129
let of_conll file_name line_list =
  handle ~name:"of_conll"
    (fun () ->
130
      let graph = G_graph.of_conll ~loc:(Loc.file file_name) (Conll.parse file_name line_list) in
bguillaum's avatar
bguillaum committed
131 132 133
      Instance.from_graph graph
    ) ()

134
let of_brown ?sentid brown =
135 136
  handle ~name:"of_brown"
    (fun () ->
137
      let graph = G_graph.of_brown ?sentid brown in
138 139 140
      Instance.from_graph graph
    ) ()

bguillaum's avatar
bguillaum committed
141 142 143 144 145 146 147 148
let load_brown file =
  handle ~name:"load_brown"
    (fun () ->
      let brown = File.load file in
      let graph = G_graph.of_brown brown in
      Instance.from_graph graph
    ) ()

149 150 151
let load_graph file =
  handle ~name:"load_graph" ~file
    (fun () ->
bguillaum's avatar
bguillaum committed
152 153 154 155 156
      match File.get_suffix file with
      | Some ".gr" -> load_gr file
      | Some ".conll" -> load_conll file
      | Some ".br" | Some ".melt" -> load_brown file 
      | _ ->
157
          Log.fwarning "Unknown file format for input graph '%s', try to guess..." file;
bguillaum's avatar
bguillaum committed
158 159 160 161
          let rec loop = function
          | [] -> Log.fcritical "[Libgrew.load_graph] Cannot guess input file format of file '%s'. Use .gr or .conll file extension" file
          | load_fct :: tail -> try load_fct file with _ -> loop tail in
          loop [load_gr; load_conll; load_brown]
162
    ) ()
163

bguillaum's avatar
bguillaum committed
164
let xml_graph xml =
165
  handle ~name:"xml_graph" (fun () -> Instance.from_graph (G_graph.of_xml xml)) ()
bguillaum's avatar
bguillaum committed
166

bguillaum's avatar
bguillaum committed
167
let raw_graph instance =
168
  handle ~name:"raw_graph" (fun () -> G_graph.to_raw instance.Instance.graph) ()
bguillaum's avatar
bguillaum committed
169

170 171
let rewrite ~gr ~grs ~seq =
  handle ~name:"rewrite" (fun () -> Grs.rewrite grs seq gr) ()
172

bguillaum's avatar
bguillaum committed
173
let display ~gr ~grs ~seq =
174
  handle ~name:"display" (fun () -> Grs.build_rew_display grs seq gr) ()
bguillaum's avatar
bguillaum committed
175

176 177
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
178

179 180
let write_annot ~title static_dir annot_dir base_name_rew_hist_list =
  handle ~name:"write_annot" (fun () -> Html_annot.build ~title static_dir annot_dir base_name_rew_hist_list) ()
181

bguillaum's avatar
bguillaum committed
182
let save_index ~dirname ~base_names =
183 184 185 186 187
  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
  ) ()
188

bguillaum's avatar
bguillaum committed
189
let save_graph_conll filename graph =
190 191 192 193 194
  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
195

196 197
let save_gr base rew_hist =
  handle ~name:"save_gr" (fun () -> Rewrite_history.save_gr base rew_hist) ()
198

199 200
let save_conll base rew_hist =
  handle ~name:"save_conll" (fun () -> Rewrite_history.save_conll base rew_hist) ()
bguillaum's avatar
bguillaum committed
201

202 203 204
let save_full_conll base rew_hist =
  handle ~name:"save_full_conll" (fun () -> Rewrite_history.save_full_conll base rew_hist) ()

205 206
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
207

208
let save_det_conll ?header base rew_hist =
209
  handle ~name:"save_deeeet_conll" (fun () -> Rewrite_history.save_det_conll ?header base rew_hist) ()
210

211 212
let det_dep_string rew_hist =
  handle ~name:"det_dep_string" (fun () -> Rewrite_history.det_dep_string rew_hist) ()
bguillaum's avatar
bguillaum committed
213

bguillaum's avatar
bguillaum committed
214 215 216
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
217
let write_html
218 219
    ?(no_init=false)
    ?(out_gr=false)
220 221
    ?filter
    ?main_feat
222
    ?dot
bguillaum's avatar
bguillaum committed
223
    ~header
bguillaum's avatar
bguillaum committed
224
    ?graph_file
bguillaum's avatar
bguillaum committed
225 226
    rew_hist
    output_base =
227 228 229 230 231 232 233 234 235
  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
236
        ?graph_file
237
        output_base rew_hist
bguillaum's avatar
bguillaum committed
238
    )
239
  ) ()
240

bguillaum's avatar
bguillaum committed
241 242 243
let error_html
    ?(no_init=false)
    ?main_feat
244
    ?dot
245
    ~header
bguillaum's avatar
bguillaum committed
246
    msg
247
    ?init
248
    output_base =
249 250 251 252 253 254 255 256
  handle ~name:"error_html" (fun () ->
    ignore (
      Html_rh.error
        ?main_feat
        ?dot
        ~init_graph: (not no_init)
        ~header
        output_base msg init
257
    )
258
  ) ()
259

260
let make_index ~title ~grs_file ~html ~grs ~seq ~input_dir ~output_dir ~base_names  =
261 262 263 264 265 266 267 268 269 270 271 272 273 274 275
  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
276

277 278
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
279

280 281
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
282

283 284
let to_gr_graph graph =
  handle ~name:"to_gr_graph" (fun () -> G_graph.to_gr graph) ()
285

286 287
let to_conll_graph graph =
  handle ~name:"to_conll_graph" (fun () -> G_graph.to_conll graph) ()
bguillaum's avatar
bguillaum committed
288

289 290 291
type pattern = Rule.pattern
type matching = Rule.matching

bguillaum's avatar
bguillaum committed
292
let load_pattern file =
293 294 295 296
  handle ~name:"load_pattern" (fun () -> Rule.build_pattern (Grew_parser.load_pattern file)) ()

let match_in_graph pattern graph = Rule.match_in_graph pattern graph

bguillaum's avatar
bguillaum committed
297
let match_deco pattern matching = Rule.match_deco pattern matching