fix: Make ungrounded types have a default name when sending to the frontend

[raunakab]

Description

When compiling stats on entities, ungrounded entities were throwing errors because the logic expected that grounded_source_name to be non-null. However, it is possible for grounded_source_name to be null (hence the "ungrounded source"), which caused an error in pydantic serialization.

This PR fixes that issue by assigning a default name to all ungrounded entities when compiling the stats for the frontend.

Addresses: https://linear.app/danswer/issue/DAN-2255/fix-kg-admin-deserialization-serialization-issue.

How Has This Been Tested?

Untested.

---

Summary by cubic

Fixed errors when compiling entity stats by assigning a default name to ungrounded types before sending data to the frontend, ensuring proper serialization.

• Bug Fixes
  • Ungrounded entities now use "Ungrounded" as a default name if no grounded source name is present.

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
internal-search	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Jul 30, 2025 7:06pm

[greptile-apps]

Greptile Summary

This PR fixes a Pydantic serialization error in the Knowledge Graph (KG) admin functionality. The issue occurred when compiling entity statistics for the frontend - ungrounded entities in the database can have null grounded_source_name values, but the SourceStatistics Pydantic model expects a non-null string for its source_name field, causing serialization failures.

The fix involves three coordinated changes:

1. Making the constant public: The _UNGROUNDED_SOURCE_NAME constant in entity_type.py is renamed to UNGROUNDED_SOURCE_NAME to make it publicly accessible across modules.

2. Adding fallback logic: In entities.py, the get_entity_stats_by_grounded_source_name() function now uses the null coalescing operator (or) to provide the default "Ungrounded" name when grounded_source_name is null.

3. Improving variable naming: The API endpoint code in kg/api.py uses more descriptive variable names (source_name instead of key) for better code clarity.

This approach maintains consistency with existing patterns in the codebase - the same UNGROUNDED_SOURCE_NAME constant is already used in the get_configured_entity_types() function. The fix ensures that entity statistics can be properly compiled and serialized for the frontend while preserving the semantic meaning that some entities are "ungrounded" (not associated with a specific source).

Confidence score: 5/5

• This is a safe, well-targeted fix that addresses a specific serialization bug without breaking existing functionality.
• The solution follows established patterns in the codebase and maintains consistency across the KG system.
• No files need additional attention - the changes are straightforward and the logic is sound.

_{3 files reviewed, 1 comment}

_{Edit Code Review Bot Settings | Greptile}

[greptile-apps]

syntax: Extra backtick in the comment - should be 'row.grounded_source_name' instead of 'row.grounded_source_name``'

Suggested change

	# `row.grounded_source_name`` is NULLABLE in the database schema.
	# Thus, for all "ungrounded" entity-types, we use a default name.
	# `row.grounded_source_name` is NULLABLE in the database schema.
	# Thus, for all "ungrounded" entity-types, we use a default name.

[cubic-dev-ai]

cubic analysis

1 issue found across 3 files • Review in cubic

_{React with 👍 or 👎 to teach cubic. You can also tag @cubic-dev-ai to give feedback, ask questions, or re-run the review.}

[evan-onyx]

LGTM