#398 -- related simple support for brat formatting.

arekit/common/bound.py

@@ -17,3 +17,32 @@ def Length(self):
         return self.__length
 
     # endregion
+
+    def itersects_with(self, other):
+        begin = self.__pos
+        end = self.__pos + self.__length
+        other_begin = other.Position
+        other_end_included = other.Position + other.Length - 1
+        if end > other_begin >= begin:
+            return True
+        if end > other_end_included >= begin:
+            return True
+        if other_begin < begin and end <= other_end_included:
+            return True
+        return False
+
+    def intersect(self, other):
+        begin = self.__pos
+        end = self.__pos + self.__length
+        other_begin = other.Position
+        other_end = other.Position + other.Length
+        actual_begin = min(begin, other_begin)
+        actual_length = max(end, other_end) - actual_begin
+        return Bound(pos=actual_begin, length=actual_length)
+
+    def contains(self, other):
+        begin = self.__pos
+        end = self.__pos + self.__length
+        other_begin = other.Position
+        other_end = other.Position + other.Length
+        return begin <= other_begin and end >= other_end

arekit/contrib/source/brat/entities/compound.py

@@ -16,6 +16,13 @@ def __init__(self, id_in_doc, value, e_type, root, entities, index_begin, index_
         self.__entities = entities
         self.__root = root
 
+    @classmethod
+    def from_list(cls, root, childs):
+        assert(isinstance(root, BratEntity))
+        assert(isinstance(childs, list) and len(childs) > 0)
+        return cls(id_in_doc=root.ID, value=root.Value, e_type=root.Type, root=None,
+                   entities=childs, index_begin=root.IndexBegin, index_end=root.IndexEnd)
+
     @property
     def Root(self):
         return self.__root

arekit/contrib/source/brat/sentence.py

@@ -1,5 +1,9 @@
+from functools import cmp_to_key
+
 from arekit.common.bound import Bound
 from arekit.common.news.sentence import BaseNewsSentence
+from arekit.contrib.source.brat.entities.compound import BratCompoundEntity
+from arekit.contrib.source.brat.entities.entity import BratEntity
 
 
 class BratSentence(BaseNewsSentence):
@@ -21,16 +25,45 @@ def __init__(self, text, index_begin, entities):
         self.__index_begin = index_begin
         self.__entities = entities
 
-    def iter_entity_with_local_bounds(self, avoid_intersection=True):
-        last_position = -1
+    @staticmethod
+    def cmp_entities(a, b):
+        assert(isinstance(a, BratEntity))
+        assert(isinstance(b, BratEntity))
+        if a.IndexBegin != b.IndexBegin:
+            # Ordered by appearance
+            return a.IndexBegin - b.IndexBegin
+        else:
+            # Ordered by length first
+            b_length = b.IndexEnd - b.IndexBegin
+            a_length = a.IndexEnd - a.IndexBegin
+            return b_length - a_length
+
+    def iter_entity_with_local_bounds(self):
+        self.__entities.sort(key=cmp_to_key(lambda a, b: self.cmp_entities(a, b)))
 
+        bounds_and_entities = []
+
+        # Merging nested entities.
         for entity in self.__entities:
             start = entity.IndexBegin - self.__index_begin
             end = entity.IndexEnd - self.__index_begin
+            bound = Bound(pos=start, length=end - start)
+
+            updated = False
+            if len(bounds_and_entities) > 0:
+                last_bound, last_entities = bounds_and_entities[-1]
+                if bound.itersects_with(last_bound):
+                    # Update.
+                    last_entities.append(entity)
+                    bounds_and_entities[-1] = (bound.intersect(last_bound), last_entities)
+                    updated = True
 
-            if start <= last_position and avoid_intersection:
-                # intersected with the previous one.
-                continue
+            if not updated:
+                bounds_and_entities.append((bound, [entity]))
 
-            yield entity, Bound(pos=start, length=end - start)
-            last_position = end
+        # Returning result.
+        for item in bounds_and_entities:
+            bound, entities = item
+            entity = entities[0] if len(entities) == 1 else \
+                BratCompoundEntity.from_list(root=entities[0], childs=entities[1:])
+            yield entity, bound

tests/contrib/source/test_brat_compound_entities.py

@@ -3,6 +3,7 @@
 from arekit.common.bound import Bound
 from arekit.common.entities.collection import EntityCollection
 from arekit.common.synonyms.grouping import SynonymsCollectionValuesGroupingProviders
+from arekit.contrib.source.brat.entities.compound import BratCompoundEntity
 from arekit.contrib.source.brat.entities.entity import BratEntity
 from arekit.contrib.source.brat.sentences_reader import BratDocumentSentencesReader
 from arekit.contrib.utils.processing.lemmatization.mystem import MystemWrapper
@@ -14,9 +15,9 @@ class TestCompoundEntites(unittest.TestCase):
     text = "мама мыла раму"
     entities = [
         BratEntity(id_in_doc="T1", e_type="PERSON", index_begin=0, index_end=4, value="мама"),
-        BratEntity(id_in_doc="T2", e_type="VERB", index_begin=5, index_end=9, value="мыла"),
-        BratEntity(id_in_doc="T3", e_type="OBJECT", index_begin=10, index_end=14, value="раму"),
-        BratEntity(id_in_doc="T3", e_type="ACTION", index_begin=0, index_end=9, value="мама мыла")
+        BratEntity(id_in_doc="T2", e_type="VERB", index_begin=5, index_end=8, value="мыл"),
+        BratEntity(id_in_doc="T3", e_type="OBJECT", index_begin=9, index_end=13, value="раму"),
+        BratEntity(id_in_doc="T3", e_type="ACTION", index_begin=0, index_end=8, value="мама мыл")
     ]
 
     def test(self):
@@ -38,7 +39,11 @@ def test(self):
         for sentence in sentences:
             for e, b in sentence.iter_entity_with_local_bounds():
                 assert(isinstance(b, Bound))
-                print(e.Value, b.Position, b.Position + b.Length)
+                print(type(e))
+                print("{} ({}, {})".format(e.Value, b.Position, b.Position + b.Length))
+                if isinstance(e, BratCompoundEntity):
+                    for ee in e.iter_childs():
+                        print("\t{} ({}, {}) [{}]".format(ee.Value, ee.IndexBegin, ee.IndexEnd, ee.Type))
 
 
 if __name__ == '__main__':

tests/tutorials/test_tutorial_collection_binding.py

@@ -5,6 +5,7 @@
 from arekit.common.entities.collection import EntityCollection
 from arekit.common.synonyms.grouping import SynonymsCollectionValuesGroupingProviders
 from arekit.contrib.source.brat.annot import BratAnnotationParser
+from arekit.contrib.source.brat.entities.compound import BratCompoundEntity
 from arekit.contrib.source.brat.news import BratNews
 from arekit.contrib.source.brat.sentences_reader import BratDocumentSentencesReader
 from arekit.contrib.source.zip_utils import ZipArchiveUtils
@@ -96,9 +97,16 @@ def test_reading(self):
         for sentence in news.iter_sentences():
             print(sentence.Text.strip())
             for entity, bound in sentence.iter_entity_with_local_bounds():
-                print("{}: ['{}',{}, {}]".format(
+                print("{}: ['{}',{}, {}] {}".format(
                     entity.ID, entity.Value, entity.Type,
-                    "-".join([str(bound.Position), str(bound.Position + bound.Length)])))
+                    "-".join([str(bound.Position), str(bound.Position + bound.Length)]),
+                    "[COMPOUND]" if isinstance(entity, BratCompoundEntity) else ""))
+
+                if not isinstance(entity, BratCompoundEntity):
+                    continue
+
+                for child in entity.iter_childs():
+                    print("\t{}: ['{}',{}]".format(child.ID, child.Value, child.Type))
 
         for brat_relation in news.Relations:
             print(brat_relation.SourceID, brat_relation.TargetID, brat_relation.Type)