#332 refactoring. Done

arekit/contrib/experiment_rusentrel/exp_ds/opinions.py

@@ -1,5 +1,6 @@
 from arekit.common.experiment.api.ops_opin import OpinionOperations
 from arekit.common.experiment.data_type import DataType
+from arekit.contrib.experiment_rusentrel.labels.scalers.ruattitudes import ExperimentRuAttitudesLabelConverter
 from arekit.contrib.source.ruattitudes.news.helper import RuAttitudesNewsHelper
 
 
@@ -10,14 +11,17 @@ def __init__(self, ru_attitudes):
         super(RuAttitudesOpinionOperations, self).__init__()
 
         self.__ru_attitudes = ru_attitudes
+        self.__label_scaler = ExperimentRuAttitudesLabelConverter()
 
     # region private methods
 
     def __get_opinion_list_in_doc(self, doc_id, opinion_check=lambda _: True):
         news = self.__ru_attitudes[doc_id]
-        return [opinion
-                for opinion, _ in RuAttitudesNewsHelper.iter_opinions_with_related_sentences(news)
-                if opinion_check(opinion)]
+
+        data_it = RuAttitudesNewsHelper.iter_opinions_with_related_sentences(
+            news=news, label_scaler=self.__label_scaler)
+
+        return [opinion for opinion, _ in data_it if opinion_check(opinion)]
 
     # endregion
 

arekit/contrib/experiment_rusentrel/exp_ds/utils.py

@@ -1,7 +1,6 @@
 import logging
 
 from arekit.common.utils import progress_bar_iter
-from arekit.contrib.experiment_rusentrel.labels.scalers.ruattitudes import ExperimentRuAttitudesLabelConverter
 from arekit.contrib.source.ruattitudes.collection import RuAttitudesCollection
 from arekit.contrib.source.ruattitudes.io_utils import RuAttitudesVersions
 from arekit.contrib.source.ruattitudes.news.base import RuAttitudesNews
@@ -20,7 +19,6 @@ def read_ruattitudes_in_memory(version, keep_doc_ids_only, doc_id_func):
 
     it = RuAttitudesCollection.iter_news(version=version,
                                          get_news_index_func=doc_id_func,
-                                         label_convereter=ExperimentRuAttitudesLabelConverter(),
                                          return_inds_only=keep_doc_ids_only)
 
     it_formatted_and_logged = progress_bar_iter(

arekit/contrib/experiment_rusentrel/labels/scalers/ruattitudes.py

@@ -1,9 +1,9 @@
 from arekit.contrib.experiment_rusentrel.labels.types import ExperimentPositiveLabel, ExperimentNegativeLabel, \
     ExperimentNeutralLabel
-from arekit.contrib.source.ruattitudes.labels_scaler import RuAttitudesLabelConverter
+from arekit.contrib.source.ruattitudes.labels_scaler import RuAttitudesLabelScaler
 
 
-class ExperimentRuAttitudesLabelConverter(RuAttitudesLabelConverter):
+class ExperimentRuAttitudesLabelConverter(RuAttitudesLabelScaler):
 
     @classmethod
     def _neutral_label_instance(cls):

arekit/contrib/source/ruattitudes/collection.py

@@ -1,25 +1,23 @@
 from arekit.contrib.source.ruattitudes.io_utils import RuAttitudesVersions, RuAttitudesIOUtils
-from arekit.contrib.source.ruattitudes.labels_scaler import RuAttitudesLabelConverter
+from arekit.contrib.source.ruattitudes.labels_scaler import RuAttitudesLabelScaler
 from arekit.contrib.source.ruattitudes.reader import RuAttitudesFormatReader
 
 
 class RuAttitudesCollection(object):
 
     @staticmethod
-    def __get_reading_handler(input_file, read_inds_only, get_news_inds_func, label_converter):
+    def __get_reading_handler(input_file, read_inds_only, get_news_inds_func):
         assert(isinstance(read_inds_only, bool))
 
         if read_inds_only:
             return RuAttitudesFormatReader.iter_news_inds(input_file=input_file,
                                                           get_news_index_func=get_news_inds_func)
         else:
-            return RuAttitudesFormatReader.iter_news(
-                input_file=input_file,
-                label_converter=RuAttitudesLabelConverter() if label_converter is None else label_converter,
-                get_news_index_func=get_news_inds_func)
+            return RuAttitudesFormatReader.iter_news(input_file=input_file,
+                                                     get_news_index_func=get_news_inds_func)
 
     @staticmethod
-    def iter_news(version, get_news_index_func, return_inds_only, label_convereter=None):
+    def iter_news(version, get_news_index_func, return_inds_only):
         """
         RuAttitudes collection reader from zip archive
         """
@@ -32,7 +30,6 @@ def iter_news(version, get_news_index_func, return_inds_only, label_convereter=N
             process_func=lambda input_filepath: RuAttitudesCollection.__get_reading_handler(
                 input_file=input_filepath,
                 read_inds_only=return_inds_only,
-                label_converter=label_convereter,
                 get_news_inds_func=get_news_index_func),
             version=version)
 

arekit/contrib/source/ruattitudes/conts.py

@@ -1,10 +1,6 @@
 from arekit.common.labels.base import NoLabel
 from arekit.contrib.source.common.labels import PositiveLabel, NegativeLabel
 
-# Default label scales.
-NEU_INT_VALUE = 0
-POS_INT_VALUE = 1
-NEG_INT_VALUE = -1
 
 # Default label instances.
 RU_ATTITUDES_NEU_LABEL = NoLabel()

arekit/contrib/source/ruattitudes/labels_scaler.py

@@ -1,17 +1,26 @@
+from collections import OrderedDict
+
+from arekit.common.labels.scaler.base import BaseLabelScaler
 from arekit.contrib.source.ruattitudes.conts import \
-    NEG_INT_VALUE, POS_INT_VALUE, \
-    RU_ATTITUDES_POS_LABEL, RU_ATTITUDES_NEG_LABEL, NEU_INT_VALUE, RU_ATTITUDES_NEU_LABEL
+    RU_ATTITUDES_POS_LABEL, RU_ATTITUDES_NEG_LABEL, RU_ATTITUDES_NEU_LABEL
 
 
-class RuAttitudesLabelConverter:
+class RuAttitudesLabelScaler(BaseLabelScaler):
 
     def __init__(self):
 
-        self.__int_to_label_dict = {
-                NEU_INT_VALUE: self._neutral_label_instance(),
-                POS_INT_VALUE: self._positive_label_instance(),
-                NEG_INT_VALUE: self._negative_label_instance()
-        }
+        self.__int_to_label_dict = OrderedDict([
+            (self._neutral_label_instance(), 0),
+            (self._positive_label_instance(), 1),
+            (self._negative_label_instance(), -1)])
+
+        self.__uint_to_label_dict = OrderedDict([
+            (self._neutral_label_instance(), 0),
+            (self._positive_label_instance(), 1),
+            (self._negative_label_instance(), 2)])
+
+        super(RuAttitudesLabelScaler, self).__init__(int_dict=self.__int_to_label_dict,
+                                                     uint_dict=self.__uint_to_label_dict)
 
     @classmethod
     def _neutral_label_instance(cls):
@@ -24,6 +33,3 @@ def _positive_label_instance(cls):
     @classmethod
     def _negative_label_instance(cls):
         return RU_ATTITUDES_NEG_LABEL
-
-    def int_to_label(self, int_value):
-        return self.__int_to_label_dict[int_value]

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

@@ -1,6 +1,8 @@
+from arekit.common.labels.scaler.base import BaseLabelScaler
 from arekit.common.linkage.text_opinions import TextOpinionsLinkage
 from arekit.common.news.base import News
 from arekit.common.opinions.base import Opinion
+from arekit.contrib.source.ruattitudes.news.opin_converter import RuAttitudesSentenceOpinionConverter
 from arekit.contrib.source.ruattitudes.sentence.base import RuAttitudesSentence
 
 
@@ -50,28 +52,33 @@ def get_objects_declared_before(self, sentence_index):
 
     # region base News
 
-    def extract_text_opinions_linkages(self, opinion):
+    def extract_text_opinions_linkages(self, opinion, label_scaler):
         """
         Note: Complexity is O(N)
         """
         assert(isinstance(opinion, Opinion))
-        return TextOpinionsLinkage(self.__iter_all_text_opinions_in_sentences(opinion=opinion))
+        assert(isinstance(label_scaler, BaseLabelScaler))
+
+        return TextOpinionsLinkage(self.__iter_all_text_opinions_in_sentences(
+            opinion=opinion, label_scaler=label_scaler))
 
     # endregion
 
     # region Private methods
 
-    def __iter_all_text_opinions_in_sentences(self, opinion):
+    def __iter_all_text_opinions_in_sentences(self, opinion, label_scaler):
         for sentence in self.iter_sentences():
             assert(isinstance(sentence, RuAttitudesSentence))
 
             sentence_opin = sentence.find_sentence_opin_by_key(key=opinion.Tag)
             if sentence_opin is None:
                 continue
 
-            yield sentence_opin.to_text_opinion(
+            yield RuAttitudesSentenceOpinionConverter.to_text_opinion(
+                sentence_opinion=sentence_opin,
                 doc_id=sentence.Owner.ID,
                 end_to_doc_id_func=lambda sent_level_id: sentence.get_doc_level_text_object_id(sent_level_id),
-                text_opinion_id=None)
+                text_opinion_id=None,
+                label_scaler=label_scaler)
 
     # endregion

arekit/contrib/source/ruattitudes/news/helper.py

@@ -1,4 +1,6 @@
+from arekit.common.labels.scaler.base import BaseLabelScaler
 from arekit.contrib.source.ruattitudes.news.base import RuAttitudesNews
+from arekit.contrib.source.ruattitudes.news.opin_converter import RuAttitudesSentenceOpinionConverter
 from arekit.contrib.source.ruattitudes.sentence.base import RuAttitudesSentence
 from arekit.contrib.source.ruattitudes.sentence.opinion import SentenceOpinion
 
@@ -12,8 +14,9 @@ def build_opinion_dict(news):
         return RuAttitudesNewsHelper.__build_opinion_dict(news)
 
     @staticmethod
-    def iter_opinions_with_related_sentences(news):
+    def iter_opinions_with_related_sentences(news, label_scaler):
         assert(isinstance(news, RuAttitudesNews))
+        assert(isinstance(label_scaler, BaseLabelScaler))
 
         doc_opinions = RuAttitudesNewsHelper.build_opinion_dict(news=news)
         assert(isinstance(doc_opinions, dict))
@@ -22,7 +25,8 @@ def iter_opinions_with_related_sentences(news):
 
             opinion, related_sentences = RuAttitudesNewsHelper.__extract_opinion_with_related_sentences(
                 news=news,
-                sentence_opin_tag=sentence_opin_tag)
+                sentence_opin_tag=sentence_opin_tag,
+                label_scaler=label_scaler)
 
             if opinion is None:
                 continue
@@ -34,7 +38,7 @@ def iter_opinions_with_related_sentences(news):
     # region private methods
 
     @staticmethod
-    def __extract_opinion_with_related_sentences(news, sentence_opin_tag):
+    def __extract_opinion_with_related_sentences(news, sentence_opin_tag, label_scaler):
         opinion = None
         related_sentences = []
 
@@ -54,9 +58,11 @@ def __extract_opinion_with_related_sentences(news, sentence_opin_tag):
 
             source, target = sentence.get_objects(sentence_opin)
 
-            opinion = sentence_opin.to_opinion(
+            opinion = RuAttitudesSentenceOpinionConverter.to_opinion(
+                sentence_opinion=sentence_opin,
                 source_value=source.Value,
-                target_value=target.Value)
+                target_value=target.Value,
+                label_scaler=label_scaler)
 
         return opinion, related_sentences
 

arekit/contrib/source/ruattitudes/news/opin_converter.py

@@ -0,0 +1,41 @@
+from arekit.common.labels.scaler.base import BaseLabelScaler
+from arekit.common.opinions.base import Opinion
+from arekit.common.text_opinions.base import TextOpinion
+from arekit.contrib.source.ruattitudes.sentence.opinion import SentenceOpinion
+
+
+class RuAttitudesSentenceOpinionConverter:
+
+    @staticmethod
+    def to_text_opinion(sentence_opinion, doc_id, end_to_doc_id_func, text_opinion_id, label_scaler):
+        """
+        Converts opinion into document-level referenced opinion
+        """
+        assert(isinstance(sentence_opinion, SentenceOpinion))
+        assert (isinstance(label_scaler, BaseLabelScaler))
+
+        return TextOpinion(doc_id=doc_id,
+                           text_opinion_id=text_opinion_id,
+                           source_id=end_to_doc_id_func(sentence_opinion.SourceID),
+                           target_id=end_to_doc_id_func(sentence_opinion.TargetID),
+                           owner=None,
+                           label=label_scaler.int_to_label(sentence_opinion.Label))
+
+    @staticmethod
+    def to_opinion(sentence_opinion, source_value, target_value, label_scaler):
+        """
+        Converts onto document, non referenced opinion
+        (non bounded to the text).
+        """
+        assert(isinstance(sentence_opinion, SentenceOpinion))
+        assert (isinstance(label_scaler, BaseLabelScaler))
+
+        opinion = Opinion(source_value=source_value,
+                          target_value=target_value,
+                          sentiment=label_scaler.int_to_label(sentence_opinion.Label))
+
+        # Using this tag allows to perform a revert operation,
+        # i.e. to find opinion_ref by opinion.
+        opinion.set_tag(sentence_opinion.Tag)
+
+        return opinion

arekit/contrib/source/ruattitudes/reader.py

@@ -1,5 +1,5 @@
 from arekit.common.utils import split_by_whitespaces
-from arekit.contrib.source.ruattitudes.labels_scaler import RuAttitudesLabelConverter
+from arekit.contrib.source.ruattitudes.labels_scaler import RuAttitudesLabelScaler
 from arekit.contrib.source.ruattitudes.news.base import RuAttitudesNews
 from arekit.contrib.source.ruattitudes.sentence.base import RuAttitudesSentence
 from arekit.contrib.source.ruattitudes.sentence.opinion import SentenceOpinion
@@ -52,11 +52,8 @@ def iter_news_inds(input_file, get_news_index_func):
                                                               local_index=local_news_ind)
 
     @staticmethod
-    def iter_news(input_file, get_news_index_func, label_converter):
+    def iter_news(input_file, get_news_index_func):
         assert(callable(get_news_index_func))
-        # TODO. #322 -- remove label converter and adopt the latter only when we need to perform
-        # TODO. #322 a conversion towards the opinion (text_opinion).
-        assert(isinstance(label_converter, RuAttitudesLabelConverter))
 
         reset = False
         title = None
@@ -79,8 +76,7 @@ def iter_news(input_file, get_news_index_func, label_converter):
                 objects_list.append(object)
 
             if RuAttitudesFormatReader.OPINION_KEY in line:
-                sentence_opin = RuAttitudesFormatReader.__parse_sentence_opin(
-                    line=line, label_converter=label_converter)
+                sentence_opin = RuAttitudesFormatReader.__parse_sentence_opin(line)
                 opinions_list.append(sentence_opin)
 
             if RuAttitudesFormatReader.FRAMEVAR_TITLE in line:
@@ -175,14 +171,12 @@ def __parse_sentence(line, is_title):
         return text.strip()
 
     @staticmethod
-    def __parse_sentence_opin(line, label_converter):
-        assert(isinstance(label_converter, RuAttitudesLabelConverter))
-
+    def __parse_sentence_opin(line):
         line = line[len(RuAttitudesFormatReader.OPINION_KEY):]
 
         s_from = line.index('b:(')
         s_to = line.index(')', s_from)
-        label = label_converter.int_to_label(int(line[s_from + 3:s_to]))
+        label = int(line[s_from + 3:s_to])
 
         o_from = line.index('oi:[')
         o_to = line.index(']', o_from)
@@ -197,7 +191,7 @@ def __parse_sentence_opin(line, label_converter):
 
         sentence_opin = SentenceOpinion(source_id=source_object_id_in_sentence,
                                         target_id=target_object_id_in_sentence,
-                                        label=label,
+                                        label_int=label,
                                         tag=opninion_key)
 
         return sentence_opin