#495 experimental extension

arekit/common/entities/base.py

@@ -1,14 +1,16 @@
 class Entity(object):
 
-    def __init__(self, value, e_type, display_value=None, group_index=None):
+    def __init__(self, value, e_type, childs=None, display_value=None, group_index=None):
         assert(isinstance(value, str) and len(value) > 0)
         assert(isinstance(e_type, str) or e_type is None)
         assert(isinstance(display_value, str) or display_value is None)
         assert(isinstance(group_index, int) or group_index is None)
+        assert(isinstance(childs, list) or childs is None)
         self.__value = value.lower()
         self.__type = e_type
         self.__display_value = display_value
         self.__group_index = group_index
+        self.__childs = childs
 
     @property
     def GroupIndex(self):
@@ -40,3 +42,11 @@ def set_group_index(self, value):
         assert(isinstance(value, int) and value >= -1)
         assert(self.__group_index is None)
         self.__group_index = value
+
+    def iter_childs(self):
+        if self.__childs is None:
+            return
+            yield
+        for child in self.__childs:
+            yield child
+

arekit/common/news/entity.py

@@ -3,12 +3,13 @@
 
 class DocumentEntity(Entity):
 
-    def __init__(self, value, display_value, e_type, id_in_doc, group_index):
+    def __init__(self, value, display_value, e_type, childs, id_in_doc, group_index):
         """ id_in_doc: Id, utilized witin the internal services
         """
         super(DocumentEntity, self).__init__(value=value,
                                              e_type=e_type,
                                              display_value=display_value,
+                                             childs=childs,
                                              group_index=group_index)
         self.__id = id_in_doc
 

arekit/common/news/parsed/providers/base.py

@@ -1,3 +1,5 @@
+from collections import Counter
+
 from arekit.common.entities.base import Entity
 from arekit.common.news.entity import DocumentEntity
 from arekit.common.news.parsed.base import ParsedNews
@@ -25,18 +27,45 @@ def init_parsed_news(self, parsed_news):
         self._doc_entities = []
         self.__entity_map.clear()
 
-        for index, entity in enumerate(parsed_news.iter_entities()):
+        c = Counter()
+        for _, entity in enumerate(parsed_news.iter_entities()):
+            assert(isinstance(entity, Entity))
+
+            # Register childrens.
+            doc_childs = {}
+            for ce in entity.iter_childs():
+                de = DocumentEntity(id_in_doc=c["entities"],
+                                    value=ce.Value,
+                                    e_type=ce.Type,
+                                    childs=None,
+                                    display_value=ce.DisplayValue,
+                                    group_index=ce.GroupIndex)
+                c["entities"] += 1
+                doc_childs[ce] = de
 
-            doc_entity = DocumentEntity(id_in_doc=index,
+                self._doc_entities.append(de)
+
+            # Register Root node.
+            doc_entity = DocumentEntity(id_in_doc=c["entities"],
                                         value=entity.Value,
                                         e_type=entity.Type,
+                                        childs=list(doc_childs.values()) if len(doc_childs) > 0 else None,
                                         display_value=entity.DisplayValue,
                                         group_index=entity.GroupIndex)
+            c["entities"] += 1
 
             self._doc_entities.append(doc_entity)
 
             if self.__entity_index_func is not None:
+                # For root node.
+                assert(self.__entity_index_func(entity) not in self.__entity_map)
                 self.__entity_map[self.__entity_index_func(entity)] = doc_entity
+                # For children.
+                for ce, de in doc_childs.items():
+                    assert(self.__entity_index_func(ce) not in self.__entity_map)
+                    self.__entity_map[self.__entity_index_func(ce)] = de
+
+        #print("Document entites registred:", len(self._doc_entities))
 
     def get_document_entity(self, entity):
         """ Maps entity to the related one with DocumentEntity type

arekit/common/news/parsed/providers/entity_service.py

@@ -56,7 +56,7 @@ def init_parsed_news(self, parsed_news):
         super(EntityServiceProvider, self).init_parsed_news(parsed_news)
         assert(isinstance(parsed_news, ParsedNews))
         self.__iter_raw_terms_func = lambda: parsed_news.iter_terms(filter_func=None, term_only=False)
-        self.__init_entity_positions()
+        self.__entity_positions = self.__calculate_entity_positions()
 
     # region public 'extract' methods
 
@@ -158,6 +158,15 @@ def __calculate_entity_positions(self):
         for s_ind, t_ind_in_sent, term in self.__iter_raw_terms_func():
 
             if isinstance(term, Entity):
+
+                # First childs
+                for _ in term.iter_childs():
+                    position = TermPosition(term_ind_in_doc=t_ind_in_doc,
+                                            term_ind_in_sent=t_ind_in_sent,
+                                            s_ind=s_ind)
+                    positions.append(position)
+
+                # Then actual root.
                 position = TermPosition(term_ind_in_doc=t_ind_in_doc,
                                         term_ind_in_sent=t_ind_in_sent,
                                         s_ind=s_ind)

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

@@ -7,9 +7,9 @@ class BratCompoundEntity(BratEntity):
 
     def __init__(self, id_in_doc, value, e_type, root, entities, index_begin, index_end,
                  display_value=None, group_index=None):
-        assert(isinstance(entities, list))
         assert(isinstance(root, BratCompoundEntity) or root is None)
         super(BratCompoundEntity, self).__init__(value=value, e_type=e_type,
+                                                 childs=entities,
                                                  id_in_doc=id_in_doc,
                                                  index_begin=index_begin,
                                                  index_end=index_end,
@@ -28,6 +28,3 @@ def from_list(cls, root, childs):
     @property
     def Root(self):
         return self.__root
-
-    def iter_childs(self):
-        return iter(self.__entities)

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

@@ -6,7 +6,8 @@ class BratEntity(Entity):
         Provides bounds, i.e. char indices in related sentence.
     """
 
-    def __init__(self, id_in_doc, e_type, index_begin, index_end, value, display_value=None, group_index=None):
+    def __init__(self, id_in_doc, e_type, index_begin, index_end, value,
+                 childs=None, display_value=None, group_index=None):
         """ index_begin: int
                 - char index (in case of string type of `text`)
                 - term index (in case of list type of `text`)
@@ -17,7 +18,7 @@ def __init__(self, id_in_doc, e_type, index_begin, index_end, value, display_val
         assert(isinstance(e_type, str))
         assert(isinstance(index_begin, int))
         assert(isinstance(index_end, int))
-        super(BratEntity, self).__init__(value=value, e_type=e_type,
+        super(BratEntity, self).__init__(value=value, e_type=e_type, childs=childs,
                                          display_value=display_value, group_index=group_index)
 
         self.__e_type = e_type