libgrew.ml 14.4 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                                                   *)
(**********************************************************************************)

bguillaum's avatar
bguillaum committed
11
open Printf
pj2m's avatar
pj2m committed
12
open Log
13
open Conll
pj2m's avatar
pj2m committed
14

bguillaum's avatar
bguillaum committed
15
let libgrew_debug_mode () = Grew_base.Global.debug := true
bguillaum's avatar
bguillaum committed
16
let get_version () = VERSION
bguillaum's avatar
bguillaum committed
17

bguillaum's avatar
bguillaum committed
18
(* ==================================================================================================== *)
bguillaum's avatar
bguillaum committed
19
(** {2 Location} *)
bguillaum's avatar
bguillaum committed
20 21 22 23 24 25 26 27
(* ==================================================================================================== *)
module Loc = struct
  type t = Grew_base.Loc.t
  let to_string = Grew_base.Loc.to_string
  let to_line = Grew_base.Loc.to_line
end

(* ==================================================================================================== *)
bguillaum's avatar
bguillaum committed
28
(** {2 Exceptions} *)
bguillaum's avatar
bguillaum committed
29
(* ==================================================================================================== *)
30 31
exception Error of string
exception Bug of string
pj2m's avatar
pj2m committed
32

33 34
let handle ?(name="") ?(file="No file defined") fct () =
  try fct () with
bguillaum's avatar
typo  
bguillaum committed
35
    (* Raise again already caught exceptions *)
36 37
    | Error msg -> raise (Error msg)
    | Bug msg -> raise (Bug msg)
38 39

    (* Catch new exceptions *)
40 41 42 43 44 45
    | Grew_base.Error.Parse (msg, Some loc) -> raise (Error (sprintf "%s %s" (Grew_base.Loc.to_string loc) msg))
    | Grew_base.Error.Parse (msg, None) -> raise (Error (sprintf "%s" msg))
    | Grew_base.Error.Build (msg, Some loc) -> raise (Error (sprintf "%s %s" (Grew_base.Loc.to_string loc) msg))
    | Grew_base.Error.Build (msg, None) -> raise (Error (sprintf "%s" msg))
    | Grew_base.Error.Run (msg, Some loc) -> raise (Error (sprintf "%s %s" (Grew_base.Loc.to_string loc) msg))
    | Grew_base.Error.Run (msg, None) -> raise (Error (sprintf "%s" msg))
46
    | Conll.Error msg -> raise (Error (sprintf "Conll error: %s" msg))
47

48
    | Grew_base.Error.Bug (msg, Some loc) -> raise (Bug (sprintf "%s %s" (Grew_base.Loc.to_string loc) msg))
49 50
    | Grew_base.Error.Bug (msg, None) -> raise (Bug (sprintf "%s" msg))
    | exc -> raise (Bug (sprintf "[Libgrew.%s] UNCAUGHT EXCEPTION: %s" name (Printexc.to_string exc)))
pj2m's avatar
pj2m committed
51

bguillaum's avatar
bguillaum committed
52

bguillaum's avatar
bguillaum committed
53
(* ==================================================================================================== *)
54
(** {2 Domain} *)
bguillaum's avatar
bguillaum committed
55 56
(* ==================================================================================================== *)
module Domain = struct
57
  type t = Grew_domain.Domain.t
58

bguillaum's avatar
bguillaum committed
59 60
  let load filename =
    let ast = Grew_loader.Loader.domain filename in
bguillaum's avatar
bguillaum committed
61
    Grew_grs.Grs.domain_build ast
62

63
  let feature_names domain =  handle ~name:"feature_names" (fun () -> Grew_domain.Domain.feature_names domain) ()
bguillaum's avatar
bguillaum committed
64 65 66 67 68 69 70 71
end

(* ==================================================================================================== *)
(** {2 Patterns} *)
(* ==================================================================================================== *)
module Pattern = struct
  type t = Grew_rule.Rule.pattern

bguillaum's avatar
bguillaum committed
72 73
  let load ?domain file =
  handle ~name:"Pattern.load" (fun () -> Grew_rule.Rule.build_pattern ?domain (Grew_loader.Loader.pattern file)) ()
74 75 76

  let parse ?domain desc =
  handle ~name:"Pattern.load" (fun () -> Grew_rule.Rule.build_pattern ?domain (Grew_loader.Parser.pattern desc)) ()
77 78 79 80 81

  let pid_list pattern =
  handle ~name:"Pattern.pid_lits"
    (fun () -> List.map (fun x -> Grew_types.Pid.to_string x) (Grew_rule.Rule.pid_list pattern)
    ) ()
bguillaum's avatar
bguillaum committed
82 83 84 85 86 87 88
end

(* ==================================================================================================== *)
(** {2 Matching} *)
(* ==================================================================================================== *)
module Matching = struct
  type t = Grew_rule.Rule.matching
89 90

  let to_python pattern graph t = Grew_rule.Rule.to_python pattern graph t
bguillaum's avatar
bguillaum committed
91 92 93 94 95 96 97 98 99 100 101
end

(* ==================================================================================================== *)
(** {2 Deco} *)
(* ==================================================================================================== *)
module Deco = struct
  type t = Grew_graph.G_deco.t
  let build pattern matching = Grew_rule.Rule.match_deco pattern matching
end

(* ==================================================================================================== *)
bguillaum's avatar
bguillaum committed
102
(** {2 Graph} *)
bguillaum's avatar
bguillaum committed
103 104 105 106 107 108
(* ==================================================================================================== *)
module Graph = struct


type t = Grew_graph.G_graph.t

bguillaum's avatar
bguillaum committed
109
  let load_gr ?domain file =
bguillaum's avatar
bguillaum committed
110
    if not (Sys.file_exists file)
111
    then raise (Error ("File_not_found: " ^ file))
bguillaum's avatar
bguillaum committed
112 113 114 115
    else
      handle ~name:"Graph.load_gr" ~file
        (fun () ->
          let gr_ast = Grew_loader.Loader.gr file in
bguillaum's avatar
bguillaum committed
116
          Grew_graph.G_graph.build ?domain gr_ast
bguillaum's avatar
bguillaum committed
117
        ) ()
bguillaum's avatar
bguillaum committed
118

bguillaum's avatar
bguillaum committed
119
  let load_conll ?domain file =
bguillaum's avatar
bguillaum committed
120
    handle ~name:"Graph.load_conll" ~file
121
      (fun () ->
bguillaum's avatar
bguillaum committed
122
        Grew_graph.G_graph.of_conll ?domain (Conll.load file)
123
      ) ()
pj2m's avatar
pj2m committed
124

bguillaum's avatar
bguillaum committed
125
  let load_brown ?domain file =
bguillaum's avatar
bguillaum committed
126 127 128
    handle ~name:"Graph.load_brown"
      (fun () ->
        let brown = Grew_base.File.load file in
bguillaum's avatar
bguillaum committed
129
        Grew_graph.G_graph.of_brown ?domain brown
bguillaum's avatar
bguillaum committed
130
      ) ()
131

132
  let load_pst ?domain file =
133
    if not (Sys.file_exists file)
134
    then raise (Error ("File_not_found: " ^ file))
135
    else
136
      handle ~name:"load_pst" ~file
137
        (fun () ->
138 139
          let const_ast = Grew_loader.Loader.phrase_structure_tree file in
          Grew_graph.G_graph.of_pst ?domain const_ast
140 141
        ) ()

bguillaum's avatar
bguillaum committed
142
  let load ?domain file =
bguillaum's avatar
bguillaum committed
143 144 145
    handle ~name:"Graph.load_graph" ~file
      (fun () ->
        match Grew_base.File.get_suffix file with
bguillaum's avatar
bguillaum committed
146 147 148
        | Some ".gr" -> load_gr ?domain file
        | Some ".conll" -> load_conll ?domain file
        | Some ".br" | Some ".melt" -> load_brown ?domain file
149
        | Some ".cst" -> load_pst ?domain file
bguillaum's avatar
bguillaum committed
150 151 152 153
        | _ ->
            Log.fwarning "Unknown file format for input graph '%s', try to guess..." file;
            let rec loop = function
            | [] -> Log.fcritical "[Libgrew.load_graph] Cannot guess input file format of file '%s'. Use .gr or .conll file extension" file
bguillaum's avatar
bguillaum committed
154
            | load_fct :: tail -> try load_fct ?domain file with _ -> loop tail in
155
            loop [load_gr; load_conll; load_brown; load_pst]
bguillaum's avatar
bguillaum committed
156
      ) ()
bguillaum's avatar
bguillaum committed
157

bguillaum's avatar
bguillaum committed
158 159
  let of_gr ?domain ?(grewpy=false) gr_string =
    handle ~name:"Graph.of_gr" (fun () -> Grew_graph.G_graph.build ?domain ~grewpy (Grew_loader.Parser.gr gr_string)) ()
160

bguillaum's avatar
bguillaum committed
161 162
  let of_conll ?domain conll =
    handle ~name:"Graph.of_conll" (fun () -> Grew_graph.G_graph.of_conll ?domain conll) ()
163

164 165
  let of_pst ?domain pst_string =
    handle ~name:"of_pst"
166
      (fun () ->
167 168 169 170 171 172 173 174 175 176
        let pst_ast = Grew_loader.Parser.phrase_structure_tree pst_string in
        (Grew_graph.G_graph.of_pst ?domain pst_ast)
      ) ()

  let sentence_of_pst ?domain pst_string =
    handle ~name:"of_pst"
      (fun () ->
        let pst_ast = Grew_loader.Parser.phrase_structure_tree pst_string in
        let word_list = Grew_ast.Ast.word_list pst_ast in
        Sentence.fr_clean_spaces (String.concat " " word_list)
177 178
      ) ()

bguillaum's avatar
bguillaum committed
179 180
  let of_brown ?domain ?sentid brown =
    handle ~name:"Graph.of_brown" (fun () -> Grew_graph.G_graph.of_brown ?domain ?sentid brown) ()
bguillaum's avatar
bguillaum committed
181

bguillaum's avatar
bguillaum committed
182 183
  let to_dot ?domain ?main_feat ?(deco=Grew_graph.G_deco.empty) graph =
    handle ~name:"Graph.to_dot" (fun () -> Grew_graph.G_graph.to_dot ?domain ?main_feat graph ~deco) ()
bguillaum's avatar
bguillaum committed
184

bguillaum's avatar
bguillaum committed
185 186
  let to_dep ?domain ?filter ?main_feat ?(deco=Grew_graph.G_deco.empty) graph =
    handle ~name:"Graph.to_dep" (fun () -> Grew_graph.G_graph.to_dep ?domain ?filter ?main_feat ~deco graph) ()
bguillaum's avatar
bguillaum committed
187

bguillaum's avatar
bguillaum committed
188 189
  let to_gr ?domain graph =
    handle ~name:"Graph.to_gr" (fun () -> Grew_graph.G_graph.to_gr ?domain graph) ()
bguillaum's avatar
bguillaum committed
190

bguillaum's avatar
bguillaum committed
191 192
  let to_conll_string ?domain graph =
    handle ~name:"Graph.to_conll_string" (fun () -> Grew_graph.G_graph.to_conll_string ?domain graph) ()
bguillaum's avatar
bguillaum committed
193

bguillaum's avatar
bguillaum committed
194 195 196 197 198
  let to_sentence ?main_feat gr =
    handle ~name:"Graph.to_sentence"
      (fun () ->
        Grew_graph.G_graph.to_sentence ?main_feat gr
      ) ()
bguillaum's avatar
bguillaum committed
199

bguillaum's avatar
bguillaum committed
200
  let save_conll ?domain filename graph =
bguillaum's avatar
bguillaum committed
201 202
    handle ~name:"Graph.save_conll" (fun () ->
      let out_ch = open_out filename in
bguillaum's avatar
bguillaum committed
203
      fprintf out_ch "%s" (Grew_graph.G_graph.to_conll_string ?domain graph);
bguillaum's avatar
bguillaum committed
204
      close_out out_ch
bguillaum's avatar
bguillaum committed
205 206
    ) ()

207 208 209 210
  let search_pattern ?domain pattern graph =
    handle ~name:"Graph.search_pattern" (fun () ->
      Grew_rule.Rule.match_in_graph ?domain pattern graph
    ) ()
211

212 213 214 215
  let node_matching pattern graph matching =
    handle ~name:"Graph.node_matching" (fun () ->
      Grew_rule.Rule.node_matching pattern graph matching
    ) ()
bguillaum's avatar
bguillaum committed
216
end
217

bguillaum's avatar
bguillaum committed
218
(* ==================================================================================================== *)
219
(** {2 Graph Rewriting System} *)
bguillaum's avatar
bguillaum committed
220 221 222
(* ==================================================================================================== *)
module Grs = struct
  type t = Grew_grs.Grs.t
223

bguillaum's avatar
bguillaum committed
224
  let empty = Grew_grs.Grs.empty
225

bguillaum's avatar
bguillaum committed
226 227 228 229
  let load file =
    handle ~name:"Grs.load" ~file
      (fun () ->
        if not (Sys.file_exists file)
230
        then raise (Error ("File_not_found: " ^ file))
bguillaum's avatar
bguillaum committed
231 232
        else Grew_grs.Grs.build file
      ) ()
233

bguillaum's avatar
bguillaum committed
234 235 236 237 238
  let get_sequence_names grs =
    handle ~name:"Grs.get_sequence_names"
      (fun () ->
        Grew_grs.Grs.sequence_names grs
      ) ()
239

bguillaum's avatar
bguillaum committed
240 241 242 243 244 245 246
  let build_html_doc ?(corpus=false) dir grs =
    handle ~name:"Grs.build_doc [with Dep2pict]"
      (fun () ->
        Grew_html.Html_doc.build ~corpus ~dep:true dir grs;

        (* draw pattern graphs for all rules and all filters *)
        let fct module_ rule_ =
bguillaum's avatar
bguillaum committed
247
          let dep_code = Grew_rule.Rule.to_dep ?domain:(Grew_grs.Grs.get_domain grs) rule_ in
bguillaum's avatar
bguillaum committed
248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264
          let dep_png_file = sprintf "%s/%s_%s-patt.png" dir module_ (Grew_rule.Rule.get_name rule_) in
          let d2p = Dep2pict.Dep2pict.from_dep ~dep:dep_code in
          Dep2pict.Dep2pict.save_png ~filename:dep_png_file d2p in
        Grew_grs.Grs.rule_iter fct grs;
        Grew_grs.Grs.filter_iter fct grs
      ) ()

  let get_domain grs = Grew_grs.Grs.get_domain grs
end

(* ==================================================================================================== *)
(** {2 Rewrite} *)
(* ==================================================================================================== *)
module Rewrite = struct
  type display = Libgrew_types.rew_display
  type history = Grew_grs.Rewrite_history.t

265 266 267
  let set_max_depth_det value = Grew_rule.Rule.set_max_depth_det value
  let set_max_depth_non_det value = Grew_rule.Rule.set_max_depth_non_det value

bguillaum's avatar
bguillaum committed
268 269
  let set_debug_loop () = Grew_rule.Rule.set_debug_loop ()

bguillaum's avatar
bguillaum committed
270 271 272 273 274 275 276 277
  let display ~gr ~grs ~seq =
    handle ~name:"Rewrite.display" (fun () -> Grew_grs.Grs.build_rew_display grs seq gr) ()

  let set_timeout t = Grew_base.Timeout.timeout := t

  let rewrite ~gr ~grs ~seq =
    handle ~name:"Rewrite.rewrite" (fun () -> Grew_grs.Grs.rewrite grs seq gr) ()

278 279 280
  let simple_rewrite ~gr ~grs ~strat =
    handle ~name:"Rewrite.simple_rewrite" (fun () -> Grew_grs.Grs.simple_rewrite grs strat gr) ()

bguillaum's avatar
bguillaum committed
281 282 283 284 285 286 287 288 289
  let is_empty rh =
    handle ~name:"Rewrite.is_empty" (fun () -> Grew_grs.Rewrite_history.is_empty rh) ()

  let num_sol rh =
    handle ~name:"Rewrite.num_sol" (fun () -> Grew_grs.Rewrite_history.num_sol rh) ()

  let write_stat filename rew_hist =
    handle ~name:"Rewrite.write_stat" (fun () -> Grew_html.Gr_stat.save filename (Grew_html.Gr_stat.from_rew_history rew_hist)) ()

bguillaum's avatar
bguillaum committed
290 291
  let write_annot ?domain ~title static_dir annot_dir base_name_rew_hist_list =
    handle ~name:"Rewrite.write_annot" (fun () -> Grew_html.Html_annot.build ?domain ~title static_dir annot_dir base_name_rew_hist_list) ()
bguillaum's avatar
bguillaum committed
292 293 294 295

  let save_index ~dirname ~base_names =
    handle ~name:"Rewrite.save_index" (fun () ->
      let out_ch = open_out (Filename.concat dirname "index") in
296
      Array.iter (fun f -> fprintf out_ch "%s\n" f) base_names;
bguillaum's avatar
bguillaum committed
297
      close_out out_ch
298 299
    ) ()

bguillaum's avatar
bguillaum committed
300 301
  let save_gr ?domain base rew_hist =
    handle ~name:"Rewrite.save_gr" (fun () -> Grew_grs.Rewrite_history.save_gr ?domain base rew_hist) ()
bguillaum's avatar
bguillaum committed
302

bguillaum's avatar
bguillaum committed
303 304
  let save_conll ?domain base rew_hist =
    handle ~name:"Rewrite.save_conll" (fun () -> Grew_grs.Rewrite_history.save_conll ?domain base rew_hist) ()
bguillaum's avatar
bguillaum committed
305

bguillaum's avatar
bguillaum committed
306 307
  let save_full_conll ?domain base rew_hist =
    handle ~name:"Rewrite.save_full_conll" (fun () -> Grew_grs.Rewrite_history.save_full_conll ?domain base rew_hist) ()
bguillaum's avatar
bguillaum committed
308

bguillaum's avatar
bguillaum committed
309 310
  let save_det_gr ?domain base rew_hist =
    handle ~name:"Rewrite.save_det_gr" (fun () -> Grew_grs.Rewrite_history.save_det_gr ?domain base rew_hist) ()
bguillaum's avatar
bguillaum committed
311

bguillaum's avatar
bguillaum committed
312 313
  let save_det_conll ?domain ?header base rew_hist =
    handle ~name:"Rewrite.save_det_conll" (fun () -> Grew_grs.Rewrite_history.save_det_conll ?domain ?header base rew_hist) ()
bguillaum's avatar
bguillaum committed
314

bguillaum's avatar
bguillaum committed
315 316
  let det_dep_string ?domain rew_hist =
    handle ~name:"Rewrite.det_dep_string" (fun () -> Grew_grs.Rewrite_history.det_dep_string ?domain rew_hist) ()
bguillaum's avatar
bguillaum committed
317

bguillaum's avatar
bguillaum committed
318 319
  let conll_dep_string ?domain ?keep_empty_rh rew_hist =
    handle ~name:"Rewrite.conll_dep_string" (fun () -> Grew_grs.Rewrite_history.conll_dep_string ?domain ?keep_empty_rh rew_hist) ()
bguillaum's avatar
bguillaum committed
320

bguillaum's avatar
bguillaum committed
321
  let write_html ?domain ?(no_init=false) ?(out_gr=false) ?filter ?main_feat ?dot ~header ?graph_file rew_hist output_base =
bguillaum's avatar
bguillaum committed
322 323
    handle ~name:"Rewrite.write_html" (fun () ->
      ignore (
bguillaum's avatar
bguillaum committed
324
        Grew_html.Html_rh.build ?domain ?filter ?main_feat ?dot ~out_gr ~init_graph: (not no_init) ~header ?graph_file output_base rew_hist
bguillaum's avatar
bguillaum committed
325 326 327
      )
    ) ()

bguillaum's avatar
bguillaum committed
328
  let error_html ?domain ?(no_init=false) ?main_feat ?dot ~header msg ?init output_base =
bguillaum's avatar
bguillaum committed
329 330
    handle ~name:"Rewrite.error_html" (fun () ->
      ignore (
bguillaum's avatar
bguillaum committed
331
        Grew_html.Html_rh.error ?domain ?main_feat ?dot ~init_graph: (not no_init) ~header output_base msg init
bguillaum's avatar
bguillaum committed
332 333 334 335 336 337 338
      )
    ) ()

  let make_index ~title ~grs_file ~html ~grs ~seq ~input_dir ~output_dir ~base_names  =
    handle ~name:"Rewrite.make_index" (fun () ->
      let init = Grew_html.Corpus_stat.empty grs seq in
      let corpus_stat =
339
        Array.fold_left
bguillaum's avatar
bguillaum committed
340 341 342 343 344
          (fun acc base_name ->
            Grew_html.Corpus_stat.add_gr_stat base_name (Grew_html.Gr_stat.load (Filename.concat output_dir (base_name^".stat"))) acc
          ) init base_names in
      Grew_html.Corpus_stat.save_html title grs_file input_dir output_dir corpus_stat
    ) ()
bguillaum's avatar
bguillaum committed
345

bguillaum's avatar
bguillaum committed
346 347
  let html_sentences ~title = handle ~name:"Rewrite.html_sentences" (fun () -> Grew_html.Html_sentences.build ~title) ()
end