Dage 18: Persistent Storage JSON

[Intrinsical-AI]

TASK TICKET

DAGE-18 - Includes a mini-research on mocks vs real I/O for files).

Changes

• Improved docstrings on previous code
• Improved logs information
• Fixed OLD tests ("is" replaced by "==")
• Added fns:

          save_queries_and_docs(self, filepath: str | Path) -> None:
          load_queries_and_docs(self, filepath: str | Path) -> None:
          save_rating_triples(self, filepath: str | Path) -> None:
          load_rating_triples(self, filepath: str | Path) -> None:

• Added new tests to /tests/unit/test_data_store.py

[nseidan]

typo -> throw*

[Intrinsical-AI]

Thanks for the spot! Comming in the next commit

[nseidan]

Please resolve when you commit the change.

[Intrinsical-AI]

Update: this should be solved

[nseidan]

I approved this, but please make changes on the things mentioned and feel free to resolve the conversations once you update .

[Intrinsical-AI]

@dantuzi should make more sense in general now

[dantuzi]

what does model_dump() do?

[Intrinsical-AI]

Doc it's a Pydantic model. model_dump it's the internal fn to serialize (to dict / json)

[Intrinsical-AI]

feat: datastore: JSON persistence + QueryContext API cleanup

Added on-disk persistence (./tmp/datastore.json) to make runs reproducible. Cleaned up QueryRatingContext with proper (de)serialization and safer access to ratings.

Included

• DataStore.save/load_tmp_file_content() with custom serializer.
• Stable IDs in QueryRatingContext; .to_dict() / .from_dict() methods.
• get_query() → get_query_text(); get_rating_score() raises if missing.
• Field validator in Document to reject empty fields.
• Writers/tests updated to use the new API; TMP file path patchable for tests.

[dantuzi]

can we use the naming convention test_expect

[Intrinsical-AI]

Added naming conventions