#276 - Cannot deserialize from JSON Cas if child type comes before super type

- Use toposort when deserializing JSON type systems

Author: jcklie

cassis/json.py

@@ -76,9 +76,19 @@ def deserialize(self, source: Union[IO, str], typesystem: Optional[TypeSystem] =
 
         embedded_typesystem = TypeSystem()
         json_typesystem = data.get(TYPES_FIELD)
-        # First load all the types but no features since features of a type X might be of a later loaded type Y
+
+        # First, build a dependency graph to support cases where a child type is defined before its super type
+        type_dependencies = defaultdict(set)
         for type_name, json_type in json_typesystem.items():
-            self._parse_type(embedded_typesystem, type_name, json_type)
+            type_dependencies[type_name].add(json_type[SUPER_TYPE_FIELD])
+
+        # Second, load all the types but no features since features of a type X might be of a later loaded type Y
+        for type_name in toposort_flatten(type_dependencies):
+            if is_predefined(type_name):
+                continue
+
+            self._parse_type(embedded_typesystem, type_name, json_typesystem[type_name])
+
         # Now we are sure we know all the types, we can create the features
         for type_name, json_type in json_typesystem.items():
             self._parse_features(embedded_typesystem, type_name, json_type)

tests/test_files/json/child_type_before_parent.json

@@ -0,0 +1,43 @@
+{
+    "%TYPES": {
+        "example.ChildType": {
+            "%NAME": "example.ChildType",
+            "%SUPER_TYPE": "example.ParentType",
+            "childFeature": {
+                "%NAME": "childFeature",
+                "%RANGE": "uima.cas.Integer"
+            }
+        },
+        "example.ParentType": {
+            "%NAME": "example.ParentType",
+            "%SUPER_TYPE": "uima.tcas.Annotation",
+            "parentFeature": {
+                "%NAME": "parentFeature",
+                "%RANGE": "uima.cas.String"
+            }
+        }
+    },
+    "%FEATURE_STRUCTURES": [
+        {
+            "%ID": 1,
+            "%TYPE": "uima.cas.Sofa",
+            "sofaNum": 1,
+            "sofaID": "_InitialView"
+        },
+        {
+            "%ID": 2,
+            "%TYPE": "example.ChildType",
+            "childFeature": "child",
+            "parentFeature": "parent",
+            "@sofa": 1
+        }
+    ],
+    "%VIEWS": {
+        "_InitialView": {
+            "%SOFA": 1,
+            "%MEMBERS": [
+                2
+            ]
+        }
+    }
+}

tests/test_json.py

@@ -5,15 +5,16 @@
 from tests.test_files.test_cas_generators import MultiFeatureRandomCasGenerator, MultiTypeRandomCasGenerator
 from tests.util import assert_json_equal
 
-FIXTURE_DIR = os.path.join(os.path.dirname(os.path.realpath(__file__)), "test_files", "json", "fs_as_array", "ser-ref")
+FIXTURE_DIR = os.path.join(os.path.dirname(os.path.realpath(__file__)), "test_files", "json")
+SER_REF_DIR = os.path.join(FIXTURE_DIR, "fs_as_array", "ser-ref")
 
 FIXTURES = [
-    (os.path.join(FIXTURE_DIR, "casWithSofaDataArray"), []),
-    (os.path.join(FIXTURE_DIR, "casWithSofaDataURI"), []),
-    (os.path.join(FIXTURE_DIR, "casWithFloatingPointSpecialValues"), []),
-    (os.path.join(FIXTURE_DIR, "casWithText"), [["uima.tcas.DocumentAnnotation", 0, 15, "This is a test."]]),
+    (os.path.join(SER_REF_DIR, "casWithSofaDataArray"), []),
+    (os.path.join(SER_REF_DIR, "casWithSofaDataURI"), []),
+    (os.path.join(SER_REF_DIR, "casWithFloatingPointSpecialValues"), []),
+    (os.path.join(SER_REF_DIR, "casWithText"), [["uima.tcas.DocumentAnnotation", 0, 15, "This is a test."]]),
     (
-        os.path.join(FIXTURE_DIR, "casWithoutTextButWithAnnotations"),
+        os.path.join(SER_REF_DIR, "casWithoutTextButWithAnnotations"),
         [
             ["uima.tcas.Annotation", 0, 4, None],
             ["uima.tcas.Annotation", 5, 7, None],
@@ -22,7 +23,7 @@
         ],
     ),
     (
-        os.path.join(FIXTURE_DIR, "casWithTextAndAnnotations"),
+        os.path.join(SER_REF_DIR, "casWithTextAndAnnotations"),
         [
             ["uima.tcas.Annotation", 0, 4, "This"],
             ["uima.tcas.Annotation", 5, 7, "is"],
@@ -32,7 +33,7 @@
         ],
     ),
     (
-        os.path.join(FIXTURE_DIR, "casWithEmojiUnicodeTextAndAnnotations"),
+        os.path.join(SER_REF_DIR, "casWithEmojiUnicodeTextAndAnnotations"),
         [
             ["uima.tcas.Annotation", 0, 1, "🥳", b"\xf0\x9f\xa5\xb3"],
             ["uima.tcas.Annotation", 2, 6, "This"],
@@ -59,15 +60,15 @@
         ],
     ),
     (
-        os.path.join(FIXTURE_DIR, "casWithLeftToRightTextAndAnnotations"),
+        os.path.join(SER_REF_DIR, "casWithLeftToRightTextAndAnnotations"),
         [
             ["uima.tcas.Annotation", 0, 3, "هذا"],
             ["uima.tcas.Annotation", 4, 10, "اختبار"],
             ["uima.tcas.DocumentAnnotation", 0, 10, "هذا اختبار"],
         ],
     ),
     (
-        os.path.join(FIXTURE_DIR, "casWithTraditionalChineseTextAndAnnotations"),
+        os.path.join(SER_REF_DIR, "casWithTraditionalChineseTextAndAnnotations"),
         [
             ["uima.tcas.Annotation", 0, 1, "這"],
             ["uima.tcas.Annotation", 1, 2, "是"],
@@ -172,3 +173,8 @@ def test_recursive_type_system():
     assert target_type_a.get_feature("typeB").rangeType.name == target_type_b.name
     assert target_type_b is not None
     assert target_type_b.get_feature("typeA").rangeType.name == target_type_a.name
+
+
+def test_deserializing_type_system_if_child_type_is_defined_before_supertype():
+    with open(os.path.join(FIXTURE_DIR, "child_type_before_parent.json"), "rb") as f:
+        load_cas_from_json(f)