#259 - Wrong entities are removed when iterating through CAS

cassis/cas.py

@@ -788,9 +788,9 @@ def _copy(self) -> "Cas":
         return result
 
 
-def _sort_func(a: FeatureStructure) -> Tuple[int, int]:
+def _sort_func(a: FeatureStructure) -> Tuple[int, int, int]:
     d = a.__slots__
     if "begin" in d and "end" in d:
-        return (a.begin, a.end)
+        return a.begin, a.end, id(a)
     else:
-        return (sys.maxsize, sys.maxsize)
+        return sys.maxsize, sys.maxsize, id(a)

tests/test_cas.py

@@ -6,6 +6,7 @@
     TYPE_NAME_ANNOTATION,
     TYPE_NAME_INTEGER,
     TYPE_NAME_INTEGER_ARRAY,
+    TYPE_NAME_STRING,
     TYPE_NAME_TOP,
     AnnotationHasNoSofa,
 )
@@ -450,6 +451,28 @@ def test_removing_throws_if_fs_in_other_view(small_typesystem_xml, tokens, sente
         view.remove(tokens[0])
 
 
+def test_removing_many_annotations():
+    typesystem = TypeSystem()
+    NamedEntity = typesystem.create_type(name="NamedEntity", supertypeName=TYPE_NAME_ANNOTATION)
+    typesystem.create_feature(domainType=NamedEntity, name="source", rangeType=TYPE_NAME_STRING)
+
+    count_to_generate = 100
+
+    cas = Cas(typesystem)
+    for i in range(count_to_generate):
+        cas.add(NamedEntity(source=("A" if (i % 2) else "B")))
+
+    assert len(cas.select(NamedEntity.name)) == count_to_generate
+    assert sum(1 for e in cas.select(NamedEntity.name) if e.source == "A") == count_to_generate / 2
+    assert sum(1 for e in cas.select(NamedEntity.name) if e.source == "B") == count_to_generate / 2
+
+    for e in cas.select(NamedEntity.name):
+        if e.source == "A":
+            cas.remove(e)
+    assert sum(1 for e in cas.select(NamedEntity.name) if e.source == "B") == count_to_generate / 2
+    assert sum(1 for e in cas.select(NamedEntity.name) if e.source == "A") == 0
+
+
 def test_fail_on_duplicate_fs_id(small_typesystem_xml):
     cas = Cas(typesystem=load_typesystem(small_typesystem_xml))