#340 . Provide checks for entity borders intersections with the left …

…and right sentence borders.
nicolay-r · Jun 18, 2022 · 877b8e2 · 877b8e2
1 parent bfa8493
commit 877b8e2
Show file tree

Hide file tree

Showing 2 changed files with 25 additions and 8 deletions.
diff --git a/arekit/contrib/source/brat/news_reader.py b/arekit/contrib/source/brat/news_reader.py
@@ -7,8 +7,8 @@ class BratDocumentSentencesReader(object):
 
     @staticmethod
     def from_file(input_file, entities, line_handler=None, skip_entity_func=None):
-        assert(isinstance(entities, EntityCollection))
-        assert(callable(skip_entity_func) or skip_entity_func is None)
+        assert (isinstance(entities, EntityCollection))
+        assert (callable(skip_entity_func) or skip_entity_func is None)
 
         sentences = BratDocumentSentencesReader.__parse_sentences(input_file=input_file,
                                                                   line_handler=line_handler)
@@ -18,7 +18,7 @@ def from_file(input_file, entities, line_handler=None, skip_entity_func=None):
 
         while s_ind < len(sentences) and e_ind < len(entities):
             e = entities.get_entity_by_index(e_ind)
-            assert(isinstance(e, BratEntity))
+            assert (isinstance(e, BratEntity))
 
             s = sentences[s_ind]
 
@@ -35,12 +35,21 @@ def from_file(input_file, entities, line_handler=None, skip_entity_func=None):
                 e_ind += 1
                 continue
 
-            raise Exception("e_i:{} e:('{}',{},{}), s_i:{}".format(
-                e_ind,
-                e.Value.encode('utf-8'), e.CharIndexBegin, e.CharIndexEnd,
-                s_ind))
+            if e.CharIndexEnd > s.EndBound:
+                # Intersects with the right border of sentence
+                s_ind += 1
+                continue
 
-        assert(e_ind == len(entities))
+            if e.CharIndexBegin < s.BeginBound:
+                # Intersects with the left border of sentence
+                e_ind += 1
+                continue
+
+            raise Exception("e_i:{} e:('{}',{},{}), s_i:{}, s_b: [{} {}]".format(
+                e_ind,
+                e.Value, e.CharIndexBegin, e.CharIndexEnd,
+                s_ind,
+                s.BeginBound, s.EndBound))
 
         return sentences
 

diff --git a/arekit/contrib/source/brat/sentence.py b/arekit/contrib/source/brat/sentence.py
@@ -43,6 +43,14 @@ def is_entity_goes_after(self, entity):
         assert(isinstance(entity, BratEntity))
         return entity.CharIndexBegin > self.__end
 
+    @property
+    def BeginBound(self):
+        return self.__begin
+
+    @property
+    def EndBound(self):
+        return self.__end
+
     # endregion
 
     # region overriden methods