Skip to content

Self-evolving Context Database for AI Agents. Unify Agent Memory, Knowledge RAG and Skills.

fix(vectordb): fill all missing schema fields in fix_fields_data (#3674)

bianbiandashen60a02927be86
diff --git a/openviking/storage/vectordb/utils/validation.py b/openviking/storage/vectordb/utils/validation.py
index 366eaa528..f58a7afbd 100644
--- a/openviking/storage/vectordb/utils/validation.py
+++ b/openviking/storage/vectordb/utils/validation.py
@@ -408,9 +408,6 @@ def is_valid_fields_data(field_data_dict: dict, field_meta_dict: dict) -> bool:
def fix_fields_data(field_data_dict: dict, field_meta_dict: dict) -> dict:
- if len(field_data_dict) >= len(field_meta_dict):
- return field_data_dict
-
for field_name, field_meta in field_meta_dict.items():
if field_name not in field_data_dict:
# Handle both dict access and object access if field_meta is a Model (though here it's likely a dict)
diff --git a/tests/vectordb/test_fix_fields_data.py b/tests/vectordb/test_fix_fields_data.py
new file mode 100644
index 000000000..38dac2c50
--- /dev/null
+++ b/tests/vectordb/test_fix_fields_data.py
@@ -0,0 +1,53 @@
+"""Regression test: fix_fields_data must fill every missing schema field.
+
+The function backfills schema fields that are absent from a row's data with
+their defaults. It previously short-circuited when
+``len(field_data_dict) >= len(field_meta_dict)``, using field *count* as a
+proxy for "all schema fields present". That proxy is wrong: a row can have as
+many keys as the schema (or more) while still missing a specific field — for
+example a row written before a new field was added that also carries an extra
+non-schema/internal key. In that case the missing field was silently dropped
+instead of defaulted.
+"""
+
+from openviking.storage.vectordb.utils.validation import fix_fields_data
+
+
+def test_missing_field_is_filled_even_when_counts_match():
+ meta = {
+ "a": {"FieldType": "string", "DefaultValue": None},
+ "b": {"FieldType": "int64", "DefaultValue": None},
+ "vec": {"FieldType": "string", "DefaultValue": None},
+ }
+ # 3 keys == len(meta), but "b" is absent and an extra non-schema key exists.
+ data = {"a": "hello", "vec": "x", "persisted": 5}
+
+ result = fix_fields_data(dict(data), meta)
+
+ assert "b" in result, "missing schema field must be backfilled"
+ assert result["b"] == 0 # int64 type default
+
+
+def test_default_value_is_used_when_present():
+ meta = {
+ "a": {"FieldType": "string", "DefaultValue": None},
+ "status": {"FieldType": "string", "DefaultValue": "pending"},
+ "extra": {"FieldType": "string", "DefaultValue": None},
+ }
+ data = {"a": "hello", "extra": "x", "legacy": "y"}
+
+ result = fix_fields_data(dict(data), meta)
+
+ assert result["status"] == "pending"
+
+
+def test_complete_data_is_returned_unchanged():
+ meta = {
+ "a": {"FieldType": "string", "DefaultValue": None},
+ "b": {"FieldType": "int64", "DefaultValue": None},
+ }
+ data = {"a": "hello", "b": 7}
+
+ result = fix_fields_data(dict(data), meta)
+
+ assert result == {"a": "hello", "b": 7}