_get_specific_type

opencitations · Nov 29, 2024 · e7a8e96 · e7a8e96
1 parent 8bd6298
commit e7a8e96
Show file tree

Hide file tree

Showing 3 changed files with 111 additions and 6 deletions.
diff --git a/oc_ocdm/graph/graph_entity.py b/oc_ocdm/graph/graph_entity.py
@@ -18,7 +18,7 @@
 from typing import TYPE_CHECKING
 
 from oc_ocdm.abstract_entity import AbstractEntity
-from rdflib import Graph, Namespace, URIRef
+from rdflib import RDF, Graph, Namespace, URIRef
 
 if TYPE_CHECKING:
     from typing import ClassVar, Dict, List, Optional, Tuple
@@ -260,6 +260,20 @@ def mark_as_to_be_deleted(self) -> None:
 
         self._to_be_deleted = True
 
+    def _get_specific_type(self) -> Optional[URIRef]:
+        """
+        Get the specific type of the entity (e.g., JournalArticle), if any.
+        Excludes the base Expression type.
+        
+        Returns:
+            The specific type URI if present, None otherwise
+        """
+        base_type = self.short_name_to_type_iri[self.short_name]
+        for _, _, type_uri in self.g.triples((self.res, RDF.type, None)):
+            if type_uri != base_type:
+                return type_uri
+        return None
+
     def merge(self, other: GraphEntity, prefer_self: bool = False) -> None:
         """
         **WARNING:** ``GraphEntity`` **is an abstract class that cannot be instantiated at runtime.
@@ -282,9 +296,22 @@ def merge(self, other: GraphEntity, prefer_self: bool = False) -> None:
                 new_triple = (triple[0], triple[1], self.res)
                 entity.g.add(new_triple)
 
-        types: List[URIRef] = other.get_types()
-        for cur_type in types:
-            self._create_type(cur_type)
+        self_specific_type = self._get_specific_type()
+        other_specific_type = other._get_specific_type()
+
+        final_specific_type = None
+        if prefer_self and self_specific_type:
+            final_specific_type = self_specific_type
+        elif other_specific_type:
+            final_specific_type = other_specific_type
+        elif self_specific_type:
+            final_specific_type = self_specific_type
+
+        self.g.remove((self.res, RDF.type, None))
+        base_type = self.short_name_to_type_iri[self.short_name]
+        self.g.add((self.res, RDF.type, base_type))
+        if final_specific_type:
+            self.g.add((self.res, RDF.type, final_specific_type))
 
         label: Optional[str] = other.get_label()
         if label is not None:
@@ -307,4 +334,4 @@ def commit_changes(self):
                 self.preexisting_graph.add(triple)
         self._to_be_deleted = False
         self._was_merged = False
-        self._merge_list = tuple()
+        self._merge_list = tuple()
diff --git a/oc_ocdm/test/graph/entities/bibliographic/test_bibliographic_resource.py b/oc_ocdm/test/graph/entities/bibliographic/test_bibliographic_resource.py
@@ -353,6 +353,84 @@ def test_assign_more_types_br(self):
         triple2 = self.br1.res, RDF.type, GraphEntity.iri_preprint
         self.assertIn(triple2, self.br1.g) and not self.assertIn(triple1, self.br1.g)
 
+    def test_merge_specific_types(self):
+        """Test merging entities with specific types"""
+
+        # Case 1: Merge specific type into generic type
+        br_specific = self.graph_set.add_br(self.resp_agent)
+        br_generic = self.graph_set.add_br(self.resp_agent)
+
+        br_specific.create_journal_article()
+
+        # Verify initial types
+        self.assertIn((br_specific.res, RDF.type, GraphEntity.iri_journal_article), br_specific.g)
+        self.assertIn((br_specific.res, RDF.type, GraphEntity.iri_expression), br_specific.g)
+        self.assertIn((br_generic.res, RDF.type, GraphEntity.iri_expression), br_generic.g)
+
+        # Merge and verify types are preserved correctly
+        br_generic.merge(br_specific)
+        self.assertIn((br_generic.res, RDF.type, GraphEntity.iri_journal_article), br_generic.g)
+        self.assertIn((br_generic.res, RDF.type, GraphEntity.iri_expression), br_generic.g)
+
+        # Case 2: Merge generic type into specific type
+        br_specific_2 = self.graph_set.add_br(self.resp_agent)
+        br_generic_2 = self.graph_set.add_br(self.resp_agent)
+
+        br_specific_2.create_journal_article()
+
+        br_specific_2.merge(br_generic_2)
+        self.assertIn((br_specific_2.res, RDF.type, GraphEntity.iri_journal_article), br_specific_2.g)
+        self.assertIn((br_specific_2.res, RDF.type, GraphEntity.iri_expression), br_specific_2.g)
+
+        # Case 3: Merge between two specific types (prefer_self=True)
+        br_article = self.graph_set.add_br(self.resp_agent)
+        br_book = self.graph_set.add_br(self.resp_agent)
+
+        br_article.create_journal_article()
+        br_book.create_book()
+
+        br_article.merge(br_book, prefer_self=True)
+        self.assertIn((br_article.res, RDF.type, GraphEntity.iri_journal_article), br_article.g)
+        self.assertIn((br_article.res, RDF.type, GraphEntity.iri_expression), br_article.g)
+        self.assertNotIn((br_article.res, RDF.type, GraphEntity.iri_book), br_article.g)
+
+        # Case 4: Merge between two specific types (prefer_self=False)
+        br_article_2 = self.graph_set.add_br(self.resp_agent)
+        br_book_2 = self.graph_set.add_br(self.resp_agent)
+
+        br_article_2.create_journal_article()
+        br_book_2.create_book()
+
+        br_article_2.merge(br_book_2, prefer_self=False)
+        self.assertIn((br_article_2.res, RDF.type, GraphEntity.iri_book), br_article_2.g)
+        self.assertIn((br_article_2.res, RDF.type, GraphEntity.iri_expression), br_article_2.g)
+        self.assertNotIn((br_article_2.res, RDF.type, GraphEntity.iri_journal_article), br_article_2.g)
+
+        # Case 5: Merge a generic entity into a specific entity (prefer_self=True)
+        br_specific_3 = self.graph_set.add_br(self.resp_agent)
+        br_generic_3 = self.graph_set.add_br(self.resp_agent)
+
+        br_specific_3.create_journal_article()
+
+        # Verify initial types
+        self.assertIn((br_specific_3.res, RDF.type, GraphEntity.iri_journal_article), br_specific_3.g)
+        self.assertIn((br_specific_3.res, RDF.type, GraphEntity.iri_expression), br_specific_3.g)
+        self.assertIn((br_generic_3.res, RDF.type, GraphEntity.iri_expression), br_generic_3.g)
+
+        # Merge and verify types are preserved correctly
+        br_specific_3.merge(br_generic_3, prefer_self=True)
+        self.assertIn((br_specific_3.res, RDF.type, GraphEntity.iri_journal_article), br_specific_3.g)
+        self.assertIn((br_specific_3.res, RDF.type, GraphEntity.iri_expression), br_specific_3.g)
+
+        # Case 6: Merge a generic entity into a specific entity (prefer_self=False)
+        br_specific_4 = self.graph_set.add_br(self.resp_agent)
+        br_generic_4 = self.graph_set.add_br(self.resp_agent)
+
+        br_specific_4.create_journal_article()
+
+        br_specific_4.merge(br_generic_4, prefer_self=False)
+        self.assertIn((br_specific_4.res, RDF.type, GraphEntity.iri_expression), br_specific_4.g)
+        self.assertIn((br_specific_4.res, RDF.type, GraphEntity.iri_journal_article), br_specific_4.g)
 
 if __name__ == '__main__':
     unittest.main()
diff --git a/pyproject.toml b/pyproject.toml
@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "oc_ocdm"
-version = "9.2.0"
+version = "9.2.1"
 description = "Object mapping library for manipulating RDF graphs that are compliant with the OpenCitations datamodel."
 authors = [
     "Silvio Peroni <essepuntato@gmail.com>",