Dev: some update

lib/propbank_analyzer.py

@@ -132,13 +132,60 @@ def find_role_in_roleset(roleset_data, role_number):
 
     return is_found, role_data
 
+
+#==============================================================================
+# Main Function(s)
+#==============================================================================
+
+def find_pb_role(amr_predicate, amr_role):
+    """
+    Find the probbank role in PropBank frame corresponding to a given AMR
+    predicate and a given AMR role.
+
+    Parameters
+    ----------
+    amr_predicate : STRING
+        AMR predicate (example: 'include-01').
+    amr_role : STRING
+        AMR core role (example: ':ARG0').
+
+    Returns
+    -------
+    PropBank role (example: 'PAG').
+
+    """
+    
+    # -- Intialize result
+    result = None
+    
+    # -- Analyze and adapt the target description
+    lemma = get_lemma_from_amr_predicate(amr_predicate)
+    roleset_id = get_roleset_id_from_amr_predicate(amr_predicate)
+    role_number = get_number_from_amr_role(amr_role)
+    
+    # -- Find the Frame XML data corresponding to a given lemma
+    frame_found, frame_filepath, frame_data = find_frame_of_lemma(lemma)
+           
+    if frame_found:      
+        # -- Analyze frame data to find the target role
+        rs_found, rs_data = find_roleset_in_frame(frame_data, lemma, roleset_id)
+        nb_roles = -1
+        
+        if rs_found:
+            nb_roles = len(rs_data.find_all('role'))
+            if role_number in range(nb_roles):
+                r_found, role_data = find_role_in_roleset(rs_data, role_number)
+                if r_found:
+                    result = role_data.get('f')   
+    
+    return result
     
 
 #==============================================================================
-# Main function
+# *** Dev Test ***
 #==============================================================================
 
-def main(amr_predicate, amr_role):
+def dev_analyze(amr_predicate, amr_role):
 
     print("\n" + "[CMT] PropBank Frame Analyzer")
     
@@ -190,12 +237,16 @@ def main(amr_predicate, amr_role):
             else:
                 print("----- role " + str(role_number) + " not found")    
     
+    # -- Test for main function(s)
+    print("-- Test for main function(s)")
+    pb_role = find_pb_role(amr_predicate, amr_role)
+    print("----- find_pb_role(amr_predicate, amr_role) = " + pb_role)
+    
     # -- Ending print
     print("\n" + "[SSC] Done")
     
-
-if __name__ == "__main__":
-    main(sys.argv[1], sys.argv[2])
+def dev_test_1():
+    dev_analyze('include-01', ':ARG0')
 
 
     

lib/re_test.py

 import regex as re
+import propbank_analyzer as pba
 
 print("[DEV] Regular Expression Test")
 
 # -- Données de test
 print("\n-- Données de test")
 
-graph = ''' (s / system
+GRAPH_INIT = ''' (s / system
       :domain (p / planet
             :name (n / name
                   :op1 "Solar"
@@ -22,7 +23,7 @@ graph = ''' (s / system
                               :op2 (d2 / direct-02
                                     :polarity -))))))'''
 
-print("----- graphe AMR traité : " + graph)
+print("----- graphe AMR traité : " + GRAPH_INIT)
 
 substitutions = []
 substitutions.append(('bind-01', ':ARG0', ':ARG0-AGT'))
@@ -54,6 +55,7 @@ ARGOF_PATTERN = ':ARG\d-of'
 
 print("\n-- Recherche des relations (predicat, argument)")
 
+graph_1 = GRAPH_INIT
 
 pred_arg_relation_list = []
 
@@ -70,7 +72,7 @@ def find_pred_arg_relations(graph, pred_arg_relation_list):
                            arg_pos_start, arg_pos_end))
     return pred_arg_relation_list
    
-pred_arg_relation_list = find_pred_arg_relations(graph, pred_arg_relation_list)
+pred_arg_relation_list = find_pred_arg_relations(graph_1, pred_arg_relation_list)
 
 # ----- prédicat pour chaque ARGi-of
 def find_argof_pred_relations(graph, pred_arg_relation_list):
@@ -84,7 +86,7 @@ def find_argof_pred_relations(graph, pred_arg_relation_list):
                        arg_pos_start, arg_pos_end))
     return pred_arg_relation_list
 
-find_argof_pred_relations(graph, pred_arg_relation_list)
+find_argof_pred_relations(graph_1, pred_arg_relation_list)
     
 print("----- Resultat (matchs trouvés) :")
 for r in pred_arg_relation_list:
@@ -95,6 +97,8 @@ for r in pred_arg_relation_list:
 
 print("\n-- Substitution des arguments dans le graphe")
 
+graph_2 = GRAPH_INIT
+
 def sub_betwenn_pos(text, start, end, new_str):
     result = text[:start]
     result += new_str
@@ -103,8 +107,8 @@ def sub_betwenn_pos(text, start, end, new_str):
 
 
 # ----- argument pour chaque prédicat 
-def sub_pred_arg_relations(graph):
-    for (pred, old_arg, new_arg) in substitutions:
+def sub_pred_arg_relations(graph, sub_list):
+    for (pred, old_arg, new_arg) in sub_list:
         for pred_match in re.finditer(PRED_PATTERN, graph):
             for arg_match in rx.finditer(graph[pred_match.end():]):
                 arg_pos_start = pred_match.end() + arg_match.start()
@@ -119,11 +123,11 @@ def sub_pred_arg_relations(graph):
                                             new_arg)
     return graph
    
-graph = sub_pred_arg_relations(graph)
+graph_2 = sub_pred_arg_relations(graph_2, substitutions)
 
 # ----- prédicat pour chaque ARGi-of
-def sub_argof_pred_relations(graph):
-    for (pred, old_arg, new_arg) in substitutions:
+def sub_argof_pred_relations(graph, sub_list):
+    for (pred, old_arg, new_arg) in sub_list:
         for arg_match in re.finditer(ARGOF_PATTERN, graph):
             pred_match = re.findall(PRED_PATTERN, graph[arg_match.end():])
             arg_pos_start = arg_match.start()
@@ -138,7 +142,40 @@ def sub_argof_pred_relations(graph):
                                         new_arg)
     return graph
 
-graph = sub_argof_pred_relations(graph)
+graph_2 = sub_argof_pred_relations(graph_2, substitutions)
  
     
-print("----- Résultat (graphe après substitutions) :" + graph)
\ No newline at end of file
+print("----- Résultat (graphe après substitutions) :" + graph_2)
+    
+# -- Substitution des arguments dans le graphe
+
+print("\n-- Test avec l'analyseur des cadres ProbBank (pba)")
+
+graph_3 = GRAPH_INIT
+
+init_relations = pred_arg_relation_list
+
+substitutions_from_pb = []
+
+for (pred, orig_arg, _, _) in init_relations:
+    orig_role = orig_arg[0:5]
+    print("----- find pb role for: " + pred + " and " + orig_role)
+    new_role = pba.find_pb_role(pred, orig_role)
+    if new_role is not None:
+        print("----- pb role found: " + new_role)
+        new_arg = orig_arg[0:5] + '-' + new_role
+        if len(orig_arg) >= 8:
+            new_arg += orig_arg[5:8]
+        print("----- substitution add: " + pred + 
+              ", " + orig_arg + ", " + new_arg)
+        substitutions_from_pb.append((pred, orig_arg, new_arg))
+    else:
+        print("----- pb role not found")
+        
+print("----- origin relations: " + str(init_relations))
+print("----- substitutions list: " + str(substitutions_from_pb))
+
+graph_3 = sub_pred_arg_relations(graph_3, substitutions_from_pb)
+result_graph = sub_argof_pred_relations(graph_3, substitutions_from_pb)
+
+print("----- Result: " + result_graph)
\ No newline at end of file