fix(pairedData): return null source for deleted parent projects DEV-2034

[platreth]

📣 Summary

When a source project used for a connected (paired) dataset is deleted, it is now clearly reported as deleted rather than still appearing as a normal source.

📖 Description

Connected projects pull data from a "source" project. If that source project was deleted, the paired-data endpoint kept listing it as if it were a live source, which the connected-projects UI couldn't reliably detect. The deleted source's source link is now returned as null, giving the frontend a dependable signal.

💭 Notes

paired_data is stored as a JSON blob on the destination Asset keyed by source uid, with no cascading FK, so deleting a source asset leaves a dangling reference (DEV-2034). PairedDataSerializer.to_representation now emits source: null when the source Asset row no longer exists, detected by reusing the source__names mapping the viewset already builds (uid absent ⇒ row deleted) — no extra query on the list endpoint, with a direct .exists() fallback when context is absent. Runtime-only change: no OpenAPI schema drift, no migration.

👀 Preview steps

1. ℹ️ have an account with two projects; enable data sharing on project A (the source).
2. Connect project B to A (paired data) and GET /api/v2/assets/<B_uid>/paired-data/.
3. Delete project A.
4. GET /api/v2/assets/<B_uid>/paired-data/ again.
5. 🔴 [on main] the deleted source still appears with a populated "source" URL (only "source__name" is null).
6. 🟢 [on PR] that entry's "source" is null.

[greptile-apps]

Greptile Summary

This PR fixes a bug where deleting a source (parent) project left its paired-data entry in a zombie state — the list endpoint kept returning a populated source URL instead of null, making the frontend unable to reliably detect the deletion. The fix adds a _source_is_deleted helper that reuses the source__names dict the viewset already builds (uid absent → row deleted), and guards both get_source__name and __get_source_asset_url so they emit null when the source is gone.

• _source_is_deleted is decorated with @cache_for_request so the two callers in to_representation share a single lookup; a direct .exists() fallback handles the rare case when context is absent.
• The added test (test_list_with_deleted_source) covers the full list-endpoint flow: create pairing → delete source → assert both source and source__name are null.
• A pre-existing NameError lurks in the except KeyError fallback of get_source__name (line 64: source.name with source undefined) — the viewset always populates source__names so the path is practically unreachable today, but it should still be cleaned up.

Confidence Score: 4/5

Safe to merge for the paired-data list endpoint; a latent NameError in the context-fallback branch of get_source__name would affect callers that instantiate the serializer without the viewset context.

The viewset always injects source__names into context, so the broken else: source__name = source.name branch (undefined source) is unreachable through normal usage. However, it is a real defect: any test or integration that bypasses the viewset context would hit a NameError rather than returning the asset name.

The get_source__name fallback path in kpi/serializers/v2/paired_data.py (lines 62-64) contains the source.name reference that needs cleanup.

Sequence Diagram

%%{init: {'theme': 'neutral'}}%%
sequenceDiagram
    participant Client
    participant Viewset as PairedDataViewset
    participant Serializer as PairedDataSerializer
    participant DB as Database

    Client->>Viewset: GET paired-data list
    Viewset->>DB: Query uid+name for all source_uids
    DB-->>Viewset: Rows for existing assets only
    Viewset->>Serializer: serialize(context with source__names dict)

    loop For each PairedData entry
        Serializer->>Serializer: _source_is_deleted(instance)
        Note over Serializer: uid NOT in source__names means deleted
        alt Source still exists
            Serializer-->>Client: source URL + source__name populated
        else Source was deleted
            Serializer-->>Client: source null + source__name null
        end
    end

    Note over Serializer: cache_for_request on _source_is_deleted shared between callers

Loading

%%{init: {'theme': 'base', 'themeVariables': {"darkMode": true, "background": "#0d1117", "primaryColor": "#21262d", "primaryTextColor": "#e6edf3", "primaryBorderColor": "#8b949e", "lineColor": "#8b949e", "textColor": "#e6edf3", "edgeLabelBackground": "#161b22", "actorBkg": "#21262d", "actorBorder": "#8b949e", "actorTextColor": "#e6edf3", "actorLineColor": "#8b949e", "signalColor": "#8b949e", "signalTextColor": "#e6edf3", "noteBkgColor": "#373320", "noteBorderColor": "#d4a72c", "noteTextColor": "#f0e6c0", "labelBoxBkgColor": "#21262d", "labelBoxBorderColor": "#8b949e", "labelTextColor": "#e6edf3", "loopTextColor": "#e6edf3", "activationBkgColor": "#30363d", "activationBorderColor": "#8b949e"}}}%%
sequenceDiagram
    participant Client
    participant Viewset as PairedDataViewset
    participant Serializer as PairedDataSerializer
    participant DB as Database

    Client->>Viewset: GET paired-data list
    Viewset->>DB: Query uid+name for all source_uids
    DB-->>Viewset: Rows for existing assets only
    Viewset->>Serializer: serialize(context with source__names dict)

    loop For each PairedData entry
        Serializer->>Serializer: _source_is_deleted(instance)
        Note over Serializer: uid NOT in source__names means deleted
        alt Source still exists
            Serializer-->>Client: source URL + source__name populated
        else Source was deleted
            Serializer-->>Client: source null + source__name null
        end
    end

    Note over Serializer: cache_for_request on _source_is_deleted shared between callers

Loading

Comments Outside Diff (1)

1. kpi/serializers/v2/paired_data.py, line 62-64 (link)

   Undefined source variable in fallback path

   In the except KeyError fallback block, Asset.objects.values_list('name', flat=True).get(...) already assigns the name string directly to source__name. The else clause then overwrites it with source.name, but source is never defined in this scope — this would raise NameError. The else block was likely a remnant from before the query was changed to use values_list. The else branch (lines 63-64) can simply be removed; source__name is already correctly set by the try block.

_{Reviews (2): Last reviewed commit: "refactor(pairedData): push deleted-sourc..." | Re-trigger Greptile}

[noliveleger]

Let's move this line inside __get_source_asset_url() and get_source__name() and return None (change their signature accordingly).

You can decorate _source_is_deleted with @cache_for_request to avoid calling the DB twice.

in both methods, if self._source_is_deleted() returns None, we return right away, in the representation, no need to add conditions anymore. It would be handled inside other methods.

On the side note. That's not great. I think we should move away from the JSON Field and give PairedData its own db model and use FKs. Would you mind to create a linear issue for that?

[noliveleger]

LGTM