version 0.29:

* renumbering of conll output
* ignore UDT lines with "i-j" number in UDT Conll

git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/semagramme/libcaml-grew/trunk@8779 7838e531-6607-4d57-9587-6c381814729c

config/Makefile.in

-VERSION = 0.28.1
+VERSION = 0.29
 
 DATA_DIR = @prefix@@datarootdir@/libgrew/
 DOC_DIR = @prefix@@datarootdir@/libgrew-doc/

src/grew_graph.ml

@@ -791,8 +791,15 @@ module G_graph = struct
     let nodes = Gid_map.fold
       (fun gid node acc -> (gid,node)::acc)
       graph.map [] in
+
+    (* sort nodes wrt position *)
     let snodes = List.sort (fun (_,n1) (_,n2) -> G_node.position_comp n1 n2) nodes in
 
+    (* renumbering of nodes to have a consecutive sequence of int 1 --> n, in case of node deletion or addition *)
+    let snodes = List_.mapi 
+      (fun i (gid,node) -> (gid, G_node.set_position (float i) node)
+      ) snodes in
+
     let get_num gid =
       let gnode = List.assoc gid snodes in
       if G_node.is_conll_root gnode
@@ -836,8 +843,8 @@ module G_graph = struct
             (get_num gid)
             (match G_fs.get_string_atom "phon" fs with Some p -> p | None -> "_e_")
             (match G_fs.get_string_atom "lemma" fs with Some p -> p | None -> "_e_")
-            (match G_fs.get_string_atom "cat" fs with Some p -> p | None -> "X")
-            (match G_fs.get_string_atom "pos" fs with Some p -> p | None -> "X")
+            (match G_fs.get_string_atom "cat" fs with Some p -> p | None -> "_")
+            (match G_fs.get_string_atom "pos" fs with Some p -> p | None -> "_")
             (G_fs.to_conll ~exclude: ["phon"; "lemma"; "cat"; "pos"; "position"] fs)
             (String.concat "|" govs)
             (String.concat "|" labs)

src/grew_types.ml

@@ -388,22 +388,30 @@ module Conll = struct
         ) (Str.split (Str.regexp "|") morph)
 
   let underscore s = if s = "" then "_" else s
+
+  exception Dash
+  let contain_dash s =
+    try String.iter (function '-' -> raise Dash | _ -> ()) s; false
+    with Dash -> true
+
   let parse_line file_name (line_num, line) =
     match Str.split (Str.regexp "\t") line with
+      | num :: _ when contain_dash num -> None
       | [ num; phon; lemma; pos1; pos2; morph; govs; dep_labs; _; _ ] ->
         begin
           try
             let gov_list = if govs = "_" then [] else Str.split (Str.regexp "|") govs
             and lab_list = if dep_labs = "_" then [] else Str.split (Str.regexp "|") dep_labs in
             let deps = List.combine gov_list lab_list in
-            {line_num = line_num;
-             num = num;
-             phon = underscore phon;
-             lemma = underscore lemma;
-             pos1 = underscore pos1;
-             pos2 = underscore pos2;
-             morph = parse_morph file_name line_num morph;
-             deps = deps;
+            Some {
+              line_num = line_num;
+              num = num;
+              phon = underscore phon;
+              lemma = underscore lemma;
+              pos1 = underscore pos1;
+              pos2 = underscore pos2;
+              morph = parse_morph file_name line_num morph;
+              deps = deps;
             }
           with exc -> Error.build ~loc:(Loc.file_line file_name line_num) "[Conll.load] illegal line, exc=%s\n>>>>>%s<<<<<<" (Printexc.to_string exc) line
         end
@@ -411,9 +419,9 @@ module Conll = struct
 
   let load file_name =
     let lines = File.read_ln file_name in
-    List.map (parse_line file_name) lines
+    List_.opt_map (parse_line file_name) lines
 
-  let parse file_name lines = List.map (parse_line file_name) lines
+  let parse file_name lines = List_.opt_map (parse_line file_name) lines
 
     (* We would prefer to compare the float equivalent of l1.num l2.num but this would break the dicho_find function *)
   let compare l1 l2 = Pervasives.compare ((* float_of_string *) l1.num) ((* float_of_string *) l2.num)