Refactoring #178

arekit/common/news/base.py

@@ -1,15 +1,10 @@
-from arekit.common.news.objects_parser import BaseObjectsParser
-
-
 class News(object):
 
-    def __init__(self, news_id, sentences, entities_parser):
+    def __init__(self, news_id, sentences):
         assert(isinstance(news_id, int))
         assert(isinstance(sentences, list))
-        assert(isinstance(entities_parser, BaseObjectsParser))
-        self.__news_id = news_id
-        self.__entities_parser = entities_parser
 
+        self.__news_id = news_id
         self._sentences = sentences
 
     # region properties
@@ -29,13 +24,10 @@ def SentencesCount(self):
 
     # endregion
 
-    def parse_sentence(self, sent_ind):
+    def sentence_to_terms_list(self, sent_ind):
         assert(isinstance(sent_ind, int))
         sentence = self._sentences[sent_ind]
-        return self.__entities_parser.parse(sentence)
-
-    def get_entities_collection(self):
-        raise NotImplementedError("Document does not support entities collection generation.")
+        return self._sentence_to_terms_list_core(sentence)
 
     def iter_sentences(self, return_text):
         """
@@ -56,3 +48,13 @@ def extract_linked_text_opinions(self, opinion):
             is an iterable opinions that should be used to find a related text_opinion entries.
         """
         raise NotImplementedError()
+
+    def get_entities_collection(self):
+        raise NotImplementedError("Document does not support entities collection generation.")
+
+    @staticmethod
+    def _sentence_to_terms_list_core(sentence):
+        """
+        pipeline processing application towards the particular sentence.
+        """
+        raise NotImplementedError()

arekit/common/news/objects_parser.py

@@ -36,3 +36,9 @@ def parse(self, sentence):
         entries.extend(last_part)
 
         return entries
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        pass

arekit/contrib/source/rusentrel/news/base.py

@@ -17,9 +17,7 @@ def __init__(self, doc_id, sentences, entities):
         assert(isinstance(sentences, list))
         assert(isinstance(entities, RuSentRelDocumentEntityCollection))
 
-        super(RuSentRelNews, self).__init__(news_id=doc_id,
-                                            sentences=sentences,
-                                            entities_parser=RuSentRelTextEntitiesParser())
+        super(RuSentRelNews, self).__init__(news_id=doc_id, sentences=sentences)
 
         self.__entities = entities
 
@@ -142,4 +140,9 @@ def extract_linked_text_opinions(self, opinion):
     def get_entities_collection(self):
         return self.__entities
 
+    @staticmethod
+    def _sentence_to_terms_list_core(sentence):
+        with RuSentRelTextEntitiesParser() as parser:
+            return parser.parse(sentence)
+
     # endregion

arekit/processing/text/parser.py

@@ -63,7 +63,7 @@ def __parse_sentence(news, sent_ind, parse_options):
 
         if parse_options.ParseEntities:
             # Providing a modified list with parsed unicode terms.
-            terms_list = news.parse_sentence(sent_ind)
+            terms_list = news.sentence_to_terms_list(sent_ind)
             return TextParser.__parse_string_list(terms_iter=terms_list,
                                                   skip_term=lambda term: isinstance(term, Entity),
                                                   # TODO. Declare Stemmer within a derived parse options.

tests/contrib/experiment_rusentrel/test_news_entities_parsing.py

@@ -27,7 +27,7 @@ def test_ruattitudes_news_text_parsing(self):
 
             # Parse single sentence.
             assert(isinstance(news, RuAttitudesNews))
-            parsed_text = news.parse_sentence(0)
+            parsed_text = news.sentence_to_terms_list(0)
             self.__print_parsed_text(parsed_text)
 
             # Parse news via external parser.
@@ -46,7 +46,7 @@ def test_rusentrel_news_text_parsing(self):
                                            version=version)
 
         assert(isinstance(news, RuSentRelNews))
-        parsed_text = news.parse_sentence(8)
+        parsed_text = news.sentence_to_terms_list(8)
         self.__print_parsed_text(parsed_text)
 
         # Parse news via external parser.