Fixed #137

common/labels/str_fmt.py

@@ -10,6 +10,10 @@ class StringLabelsFormatter(object):
     def __init__(self, stol):
         assert(isinstance(stol, dict))
         self._stol = stol
+        self.__supported_labels = set(self._stol.itervalues())
+
+    def __is_label_supported(self, label):
+        return label in self.__supported_labels
 
     def str_to_label(self, value):
         assert(isinstance(value, unicode))
@@ -18,10 +22,18 @@ def str_to_label(self, value):
 
     def label_to_str(self, label):
         assert(isinstance(label, Label))
-        for value, l in self._stol.iteritems():
-            if l == label:
+
+        if not self.__is_label_supported(label):
+            raise Exception("Label {label} is not supported. Supported labels: [{values}]".format(
+                label=label, values=self.__supported_labels))
+
+        for value, supported_label in self._stol.iteritems():
+            if supported_label == label:
                 return value
 
+    def supports_label(self, label):
+        return label in self.__supported_labels
+
     def supports_value(self, value):
         assert(isinstance(value, unicode))
         return value in self._stol

contrib/source/rusentrel/opinions/formatter.py

@@ -14,6 +14,7 @@ class RuSentRelOpinionCollectionFormatter(OpinionCollectionsFormatter):
     def iter_opinions_from_file(self, filepath, labels_formatter):
         """
         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))
@@ -28,6 +29,9 @@ def iter_opinions_from_file(self, filepath, labels_formatter):
                 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
+        """
         assert(isinstance(collection, OpinionCollection))
         assert(isinstance(filepath, unicode))
         assert(isinstance(labels_formatter, StringLabelsFormatter))
@@ -43,20 +47,21 @@ def __opinion_key(opinion):
         with io.open(filepath, 'w') as f:
             for o in sorted_ops:
 
-                o_str = RuSentRelOpinionCollectionFormatter.__opinion_to_str(
+                str_value = RuSentRelOpinionCollectionFormatter.__try_opinion_to_str(
                     opinion=o,
                     labels_formatter=labels_formatter)
 
-                f.write(o_str)
+                if str_value is None:
+                    continue
+
+                f.write(str_value)
                 f.write(u'\n')
 
     def save_to_archive(self, collections_iter, labels_formatter):
         raise NotImplementedError()
 
     # endregion
 
-    # region private methods
-
     @staticmethod
     def _iter_opinions_from_file(input_file, labels_formatter):
         assert(isinstance(labels_formatter, StringLabelsFormatter))
@@ -68,22 +73,43 @@ def _iter_opinions_from_file(input_file, labels_formatter):
             if line == '\n':
                 continue
 
-            args = line.strip().split(',')
-            assert(len(args) >= 3)
+            str_opinion = RuSentRelOpinionCollectionFormatter.__try_str_to_opinion(
+                line=line,
+                labels_formatter=labels_formatter)
+
+            if str_opinion is None:
+                continue
+
+            yield str_opinion
+
+    # region private methods
+
+    @staticmethod
+    def __try_str_to_opinion(line, labels_formatter):
+        args = line.strip().split(',')
+        assert (len(args) >= 3)
+
+        source_value = args[0].strip()
+        target_value = args[1].strip()
+        str_label = args[2].strip()
 
-            source_value = args[0].strip()
-            target_value = args[1].strip()
-            sentiment = labels_formatter.str_to_label(args[2].strip())
+        if not labels_formatter.supports_value(str_label):
+            return None
 
-            yield Opinion(source_value=source_value,
-                          target_value=target_value,
-                          sentiment=sentiment)
+        return Opinion(source_value=source_value,
+                       target_value=target_value,
+                       sentiment=labels_formatter.str_to_label(str_label))
 
     @staticmethod
-    def __opinion_to_str(opinion, labels_formatter):
+    def __try_opinion_to_str(opinion, labels_formatter):
         assert(isinstance(opinion, Opinion))
         assert(isinstance(labels_formatter, StringLabelsFormatter))
 
+        label = opinion.Sentiment
+
+        if not labels_formatter.supports_label(label):
+            return None
+
         return u"{}, {}, {}, current".format(
             opinion.SourceValue,
             opinion.TargetValue,