Merge pull request #36 from TaskarCenterAtUW/bugfix-2065

Fixed BUG-2065

Author: sujata-m

CHANGELOG.md

@@ -1,5 +1,12 @@
 # Change log
 
+### 0.2.11
+
+- Fixed [BUG-2065](https://dev.azure.com/TDEI-UW/TDEI/_workitems/edit/2065/)
+- Added functionality to catch serialization errors
+- Added unit test cases for that
+- Added test file `test_serialization_error.zip` to test the serialization error
+
 ### 0.2.10
 
 - Added limit the message error when u_id and v_id are missing

src/python_osw_validation/__init__.py

@@ -161,12 +161,30 @@ def validate(self, max_errors=20) -> ValidationResult:
             # Validate OSW external extensions
             for file in validator.externalExtensions:
                 file_path = os.path.join(file)
+                file_name = os.path.basename(file)
                 extensionFile = gpd.read_file(file_path)
                 invalid_geojson = extensionFile[extensionFile.is_valid == False]
                 is_valid = len(invalid_geojson) == 0
                 if not is_valid:
-                    self.errors.append(
-                        f"Invalid geometries found in extension file {file}, list of invalid geometries: {invalid_geojson.to_json()}")
+                    try:
+                        # Safely extract invalid _id or fallback to index if _id is missing
+                        invalid_ids = list(set(invalid_geojson.get('_id', invalid_geojson.index)))
+                        num_invalid = len(invalid_ids)
+                        limit = min(num_invalid, 20)
+                        displayed_invalid = ', '.join(map(str, invalid_ids[:limit]))
+                        self.errors.append(
+                            f"Invalid geometries found in extension file `{file_name}`. Showing {limit if num_invalid > 20 else 'all'} of {num_invalid} invalid geometry IDs: {displayed_invalid}"
+                        )
+                    except Exception as e:
+                        self.errors.append(f"Invalid features found in `{file_name}`, but failed to extract IDs: {e}")
+
+                # Optional: Test serializability of extension file
+                try:
+                    for idx, row in extensionFile.drop(columns='geometry').iterrows():
+                        json.dumps(row.to_dict())
+                except Exception as e:
+                    self.errors.append(f"Extension file `{file_name}` has non-serializable properties: {e}")
+                    break
 
             if self.errors:
                 return ValidationResult(False, self.errors)

src/python_osw_validation/version.py

@@ -1 +1 @@
-__version__ = '0.2.10'
+__version__ = '0.2.11'

tests/assets/test_serialization_error.zip

@@ -30,6 +30,7 @@ def setUp(self):
         self.invalid_zones_file = os.path.join(ASSETS_PATH, 'UW.zones.invalid.zip')
         self.valid_osw_file = os.path.join(ASSETS_PATH, 'wa.bellevue.zip')
         self.invalid_v_id_file = os.path.join(ASSETS_PATH, '4151.zip')
+        self.serialization_file = os.path.join(ASSETS_PATH, 'test_serialization_error.zip')
         self.schema_file_path = SCHEMA_FILE_PATH
         self.invalid_schema_file_path = INVALID_SCHEMA_FILE_PATH
 
@@ -227,6 +228,13 @@ def test_invalid_zones_file(self):
         self.assertFalse(result.is_valid)
         self.assertIsNotNone(result.errors)
 
+    def test_invalid_serialization_file(self):
+        validation = OSWValidation(zipfile_path=self.serialization_file)
+        result = validation.validate()
+        self.assertFalse(result.is_valid)
+        self.assertIsNotNone(result.errors)
+        error_message = next((err for err in result.errors if 'non-serializable' in err.lower()), None)
+
     def test_unmatched_ids_limited_to_20(self):
         validation = OSWValidation(zipfile_path=self.invalid_v_id_file)
         result = validation.validate()