Minimal update un Sequoia simplification script

cm-code/bert-minimal.py

@@ -6,8 +6,8 @@ from transformers import AutoModel, AutoTokenizer
 name  = 'almanach/camembert-base'
 #sent  = "Des poids lourds et engins en feu \
 #         dans une entreprise en Vendée ."
-#sent = "La gare routière attend toujours ses illuminations ."
-sent = "Quelle surprise ! Arturo a la covid"
+sent = "La gare routière attend toujours ses illuminations ."
+#sent = "Quelle surprise ! Arturo a la covid"
 tok   = AutoTokenizer.from_pretrained(name)
 model = AutoModel.from_pretrained(name)
 

cm-code/petits.conllu

+# global.columns = ID FORM LEMMA UPOS XPOS FEATS HEAD DEPREL DEPS MISC
 # text = Les petits ruisseaux font les grandes rivières.
 1	Les	le	DET	_	Definite=Def|Number=Plur|PronType=Art	3	det	_	_
 2	petits	petit	ADJ	_	Gender=Masc|Number=Plur	3	amod	_	_

lib/conllulib.py

@@ -538,8 +538,8 @@ class TransBasedConfig(object):
     `next_act` is a string among "SHIFT", "RIGHT-ARC-X" or "LEFT-ARC-X" where
     "X" is the name of any valid syntactic relation label (deprel).
     Returns a new syntactic relation added by the action, or None for "SHIFT"        
-    Returned relation is a triple (mod, head, deprel) with modifier, head, and 
-    deprel label if `add_deprel=True` (default), or a pair (mod, head) if 
+    Returned relation is a triple (dep, head, deprel) with dependent, head, and 
+    deprel label if `add_deprel=True` (default), or a pair (dep, head) if 
     `add_deprel=False`.
     """    
     if next_act == "SHIFT":

sequoia/bin/simplify_sequoia.py

@@ -171,7 +171,7 @@ with open(sys.argv[1], "r", encoding="UTF=8") as f:
       print(sent.serialize(), end="")
     else:
       np_counter += 1
-      np_ids.append(sent.metadata["sent_id"])
+      np_ids.append((sent.metadata["sent_id"],len(sent)))
             
 print( "{} range tokens removed.\n".format(range_counter), file=sys.stderr)
 
@@ -182,4 +182,4 @@ print( "{} supersense tags modified (complex operators).\n".format(mod_ssense_co
 
 #print( "{} subrelations removed from deprel.".format(subrel_counter), file=sys.stderr)
 print( "{} non-projective sentences removed:".format(np_counter), file=sys.stderr)
-print(", ".join(np_ids), file=sys.stderr)
+print("\n".join([f"{np_id} -> {lgth}" for (np_id, lgth) in np_ids]), file=sys.stderr)