Done #139, related to #137.

common/opinions/formatter.py

@@ -1,10 +1,10 @@
 
 class OpinionCollectionsFormatter(object):
 
-    def iter_opinions_from_file(self, filepath, labels_formatter):
+    def iter_opinions_from_file(self, filepath, labels_formatter, error_on_non_supported):
         raise NotImplementedError()
 
-    def save_to_file(self, collection, filepath, labels_formatter):
+    def save_to_file(self, collection, filepath, labels_formatter, error_on_non_supported):
         raise NotImplementedError()
 
     def save_to_archive(self, collections_iter, labels_formatter):

contrib/bert/run_evaluation.py

@@ -140,7 +140,8 @@ def _handle_iteration(self, iter_index):
                         self._experiment.DataIO.OpinionFormatter.save_to_file(
                             collection=collection,
                             filepath=filepath,
-                            labels_formatter=self.__labels_formatter))
+                            labels_formatter=self.__labels_formatter,
+                            error_on_non_supported=False))
 
                 # evaluate
                 result = self._experiment.evaluate(data_type=self.__data_type,

contrib/experiment_rusentrel/exp_sl/opinions.py

@@ -109,7 +109,8 @@ def read_result_opinion_collection(self, data_type, doc_id, epoch_index):
 
     def __custom_read(self, filepath, labels_fmt):
         opinions = self.__opinion_formatter.iter_opinions_from_file(filepath=filepath,
-                                                                    labels_formatter=labels_fmt)
+                                                                    labels_formatter=labels_fmt,
+                                                                    error_on_non_supported=False)
 
         return self.__create_collection(opinions)
 

contrib/networks/core/callback/utils_model_eval.py

@@ -130,7 +130,8 @@ def __convert_output_to_opinion_collections(exp_io, opin_ops, doc_ops, labels_sc
                                                                                          epoch_index=epoch_index),
         save_to_file_func=lambda filepath, collection: opin_fmt.save_to_file(collection=collection,
                                                                              filepath=filepath,
-                                                                             labels_formatter=labels_formatter))
+                                                                             labels_formatter=labels_formatter,
+                                                                             error_on_non_supported=False))
 
 
 def __log_wrap_samples_iter(it):

contrib/source/rusentrel/opinions/collection.py

@@ -29,5 +29,6 @@ def iter_opinions_from_doc(doc_id,
             inner_path=RuSentRelIOUtils.get_sentiment_opin_filepath(doc_id),
             process_func=lambda input_file: RuSentRelOpinionCollectionFormatter._iter_opinions_from_file(
                 input_file=input_file,
-                labels_formatter=labels_fmt),
+                labels_formatter=labels_fmt,
+                error_on_non_supported=True),
             version=version)

contrib/source/rusentrel/opinions/formatter.py

@@ -11,30 +11,29 @@ class RuSentRelOpinionCollectionFormatter(OpinionCollectionsFormatter):
 
     # region protected public methods
 
-    def iter_opinions_from_file(self, filepath, labels_formatter):
+    def iter_opinions_from_file(self, filepath, labels_formatter, error_on_non_supported=True):
         """
         Important: For externaly saved collections (using save_to_file method) and related usage
-        NOTE: We consider only those opinions which labels could be obtained by provided labels_formatter
         """
         assert(isinstance(filepath, unicode))
         assert(isinstance(labels_formatter, StringLabelsFormatter))
+        assert(isinstance(error_on_non_supported, bool))
 
         with open(filepath, 'r') as input_file:
 
             it = RuSentRelOpinionCollectionFormatter._iter_opinions_from_file(
                 input_file=input_file,
-                labels_formatter=labels_formatter)
+                labels_formatter=labels_formatter,
+                error_on_non_supported=error_on_non_supported)
 
             for opinion in it:
                 yield opinion
 
-    def save_to_file(self, collection, filepath, labels_formatter):
-        """
-        NOTE: We consider only those opinions which labels could be obtained by provided labels_formatter
-        """
+    def save_to_file(self, collection, filepath, labels_formatter, error_on_non_supported=True):
         assert(isinstance(collection, OpinionCollection))
         assert(isinstance(filepath, unicode))
         assert(isinstance(labels_formatter, StringLabelsFormatter))
+        assert(isinstance(error_on_non_supported, bool))
 
         def __opinion_key(opinion):
             assert(isinstance(opinion, Opinion))
@@ -52,7 +51,11 @@ def __opinion_key(opinion):
                     labels_formatter=labels_formatter)
 
                 if str_value is None:
-                    continue
+                    if error_on_non_supported:
+                        raise Exception("Opinion label `{label}` is not supported by formatter".format(
+                            label=o.Sentiment))
+                    else:
+                        continue
 
                 f.write(str_value)
                 f.write(u'\n')
@@ -63,8 +66,9 @@ def save_to_archive(self, collections_iter, labels_formatter):
     # endregion
 
     @staticmethod
-    def _iter_opinions_from_file(input_file, labels_formatter):
+    def _iter_opinions_from_file(input_file, labels_formatter, error_on_non_supported):
         assert(isinstance(labels_formatter, StringLabelsFormatter))
+        assert(isinstance(error_on_non_supported, bool))
 
         for line in input_file.readlines():
 
@@ -78,7 +82,10 @@ def _iter_opinions_from_file(input_file, labels_formatter):
                 labels_formatter=labels_formatter)
 
             if str_opinion is None:
-                continue
+                if error_on_non_supported:
+                    raise Exception("Line '{line}' has non supported label")
+                else:
+                    continue
 
             yield str_opinion
 

tests/contrib/experiment_rusentrel/test_rusentrel_eval.py

@@ -80,7 +80,8 @@ def iter_doc_opinions(doc_id, result_version, labels_formatter):
             inner_path=path.join(u"{}.opin.txt".format(doc_id)),
             process_func=lambda input_file: RuSentRelOpinionCollectionFormatter._iter_opinions_from_file(
                 input_file=input_file,
-                labels_formatter=labels_formatter),
+                labels_formatter=labels_formatter,
+                error_on_non_supported=True),
             version=result_version)