libgrew.ml 5.96 KB
Newer Older
pj2m's avatar
pj2m committed
1 2
include Grew_types

bguillaum's avatar
bguillaum committed
3
open Printf
pj2m's avatar
pj2m committed
4 5
open Log

bguillaum's avatar
bguillaum committed
6 7 8 9 10
open Grew_utils
open Grew_graph
open Grew_rule
open Grew_grs

pj2m's avatar
pj2m committed
11
open Grew_parser
bguillaum's avatar
bguillaum committed
12
open Grew_html
pj2m's avatar
pj2m committed
13 14


bguillaum's avatar
bguillaum committed
15

pj2m's avatar
pj2m committed
16 17
exception File_dont_exists of string

bguillaum's avatar
bguillaum committed
18
exception Parsing_err of string
pj2m's avatar
pj2m committed
19 20
exception Build of string * (string * int) option
exception Run of string * (string * int) option
bguillaum's avatar
bguillaum committed
21
exception Bug of string * (string * int) option
pj2m's avatar
pj2m committed
22 23 24

type grs = Grs.t
type gr = Instance.t
bguillaum's avatar
bguillaum committed
25
type rew_history = Rewrite_history.t
pj2m's avatar
pj2m committed
26

bguillaum's avatar
bguillaum committed
27 28
let is_empty = Rewrite_history.is_empty

bguillaum's avatar
bguillaum committed
29 30
let num_sol = Rewrite_history.num_sol

pj2m's avatar
pj2m committed
31 32
let empty_grs = Grs.empty

bguillaum's avatar
bguillaum committed
33 34
let set_timeout t = Timeout.timeout := t

35 36
IFDEF DEP2PICT THEN
let build_doc file dir grs_ast grs =
bguillaum's avatar
bguillaum committed
37
  Html_doc.build ~dep:true file dir grs_ast;
38 39 40 41 42 43 44 45 46 47
  
  (* draw pattern graphs for all rules and all filters *)
  let fct module_ rule_ = 
    let dep_code = Rule.to_dep rule_ in
    let dep_svg_file = sprintf "%s/%s_%s-patt.png" dir module_ (Rule.get_name rule_) in
    ignore (Dep2pict.Dep2pict.fromDepStringToPng dep_code dep_svg_file) in
  Grs.rule_iter fct grs;
  Grs.filter_iter fct grs
ELSE
let build_doc file dir grs_ast grs =
bguillaum's avatar
bguillaum committed
48
  Html_doc.build ~dep:false file dir grs_ast
49 50
END

bguillaum's avatar
bguillaum committed
51 52 53 54
let load_grs ?doc_output_dir file =
  if not (Sys.file_exists file)
  then raise (File_dont_exists file)
  else
55
    try
bguillaum's avatar
bguillaum committed
56
      let grs_ast = Grew_parser.grs_of_file file in
bguillaum's avatar
bguillaum committed
57
      let grs = Grs.build grs_ast in
bguillaum's avatar
bguillaum committed
58
      (match doc_output_dir with
59
        | None -> ()
60
        | Some dir -> build_doc file dir grs_ast grs);
bguillaum's avatar
bguillaum committed
61
      grs
bguillaum's avatar
bguillaum committed
62
    with
bguillaum's avatar
bguillaum committed
63 64
    | Grew_parser.Parse_error (msg,Some (sub_file,l)) -> 
        raise (Parsing_err (sprintf "[file:%s, line:%d] %s" sub_file l msg))
bguillaum's avatar
bguillaum committed
65 66
    | Grew_parser.Parse_error (msg,None) -> 
        raise (Parsing_err (sprintf "[file:%s] %s" file msg))
bguillaum's avatar
bguillaum committed
67 68
    | Error.Build (msg,loc) -> raise (Build (msg,loc))
    | Error.Bug (msg, loc) -> raise (Bug (msg,loc))
bguillaum's avatar
bguillaum committed
69
    | exc -> raise (Bug (sprintf "[Libgrew.load_grs] UNCATCHED EXCEPTION: %s" (Printexc.to_string exc), None))
bguillaum's avatar
bguillaum committed
70

bguillaum's avatar
bguillaum committed
71 72 73
let to_sentence ?main_feat gr =
  let graph = gr.Instance.graph in
  G_graph.to_sentence ?main_feat graph
bguillaum's avatar
bguillaum committed
74

75
let get_sequence_names grs = Grs.sequence_names grs
bguillaum's avatar
bguillaum committed
76 77

let load_gr file =
bguillaum's avatar
bguillaum committed
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93
  if Sys.file_exists file
  then
    begin
      try
        let gr_ast = Grew_parser.gr_of_file file in
        Instance.from_graph (G_graph.build gr_ast)
      with
        | 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))
        | exc -> raise (Bug (sprintf "[Libgrew.load_gr] UNCATCHED EXCEPTION: %s" (Printexc.to_string exc), None))
    end
  else raise (File_dont_exists file)
pj2m's avatar
pj2m committed
94

bguillaum's avatar
bguillaum committed
95
let load_conll file =
96
  try
97 98
    let graph = G_graph.of_conll ~loc:(file,-1) (Conll.load file) in
    Instance.from_graph graph
99
  with
bguillaum's avatar
bguillaum committed
100 101
    | Grew_parser.Parse_error (msg,Some (sub_file,l)) -> 
        raise (Parsing_err (sprintf "[file:%s, line:%d] %s" sub_file l msg))
bguillaum's avatar
bguillaum committed
102 103
    | Grew_parser.Parse_error (msg,None) -> 
        raise (Parsing_err (sprintf "[file:%s] %s" file msg))
104 105
  | Error.Build (msg,loc) -> raise (Build (msg,loc))
  | Error.Bug (msg, loc) -> raise (Bug (msg,loc))
bguillaum's avatar
bguillaum committed
106
  | exc -> raise (Bug (sprintf "[Libgrew.load_conll] UNCATCHED EXCEPTION: %s" (Printexc.to_string exc), None))
107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122

let load_graph file = 
  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

bguillaum's avatar
bguillaum committed
123 124 125
let raw_graph instance =
  G_graph.to_raw instance.Instance.graph

126
let rewrite ~gr ~grs ~seq = 
127
  try Grs.rewrite grs seq gr
128
  with
bguillaum's avatar
bguillaum committed
129 130
  | Error.Run (msg,loc) -> raise (Run (msg,loc))
  | Error.Bug (msg, loc) -> raise (Bug (msg,loc))
bguillaum's avatar
bguillaum committed
131
  | exc -> raise (Bug (sprintf "[Libgrew.rewrite] UNCATCHED EXCEPTION: %s" (Printexc.to_string exc), None))
132

bguillaum's avatar
bguillaum committed
133
let display ~gr ~grs ~seq =
134
  try Grs.build_rew_display grs seq gr
pj2m's avatar
pj2m committed
135
  with
bguillaum's avatar
bguillaum committed
136 137
  | Error.Run (msg,loc) -> raise (Run (msg,loc))
  | Error.Bug (msg, loc) -> raise (Bug (msg,loc))
138
  | Error.Build (msg, loc) -> raise (Build (msg,loc))
bguillaum's avatar
bguillaum committed
139
  | exc -> raise (Bug (sprintf "[Libgrew.display] UNCATCHED EXCEPTION: %s" (Printexc.to_string exc), None))
bguillaum's avatar
bguillaum committed
140 141 142

let write_stat filename rew_hist = Gr_stat.save filename (Gr_stat.from_rew_history rew_hist) 

bguillaum's avatar
bguillaum committed
143 144 145 146 147
let save_index ~dirname ~base_names =
  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

148 149
let save_gr base rew_hist = 
  Rewrite_history.save_gr base rew_hist
150

151 152 153
let save_det_gr base rew_hist =
  Rewrite_history.save_det_gr base rew_hist

bguillaum's avatar
bguillaum committed
154
let write_html 
155 156 157
    ?(no_init=false)
    ?(out_gr=false)
    ?main_feat 
158
    ?dot
bguillaum's avatar
bguillaum committed
159
    ~header
160
    ~graph_file
bguillaum's avatar
bguillaum committed
161 162
    rew_hist
    output_base =
163
  ignore (
bguillaum's avatar
bguillaum committed
164
  Html_rh.build
165
    ?main_feat
166
    ?dot
167
    ~out_gr
bguillaum's avatar
bguillaum committed
168 169
    ~init_graph: (not no_init)
    ~header
170
    ~graph_file
bguillaum's avatar
bguillaum committed
171 172
    output_base rew_hist
    )
173 174

let error_html 
175 176 177
    ?(no_init=false) 
    ?main_feat 
    ?dot
178
    ~header
179 180
    msg 
    ?init
181 182
    output_base =
  ignore (
bguillaum's avatar
bguillaum committed
183
  Html_rh.error
184 185
    ?main_feat
    ?dot
186 187 188 189 190
    ~init_graph: (not no_init)
    ~header
    output_base msg init
    )

191
let make_index ~title ~grs_file ~html ~grs ~seq ~input_dir ~output_dir ~base_names  =
bguillaum's avatar
bguillaum committed
192
  let init = Corpus_stat.empty grs seq in
bguillaum's avatar
bguillaum committed
193 194 195 196 197
  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
198
  Corpus_stat.save_html title grs_file input_dir output_dir corpus_stat
bguillaum's avatar
bguillaum committed
199

bguillaum's avatar
bguillaum committed
200
let html_sentences ~title = Html_sentences.build ~title
bguillaum's avatar
bguillaum committed
201

pj2m's avatar
pj2m committed
202 203
let get_css_file = Filename.concat DATA_DIR "style.css"

204
let graph_of_instance instance = instance.Instance.graph