feat: add knowledge_ids field to ApplicationVersion and populate from ResourceMapping

[shaohuzhang1]

feat: add knowledge_ids field to ApplicationVersion and populate from ResourceMapping

[shaohuzhang1]

Seems you are using me but didn't get OPENAI_API_KEY seted in Variables/Secrets for this repo. you could follow readme for more information

[Copilot]

Pull request overview

Adds a persisted knowledge_ids snapshot to ApplicationVersion so simple-application chat startup can read dataset bindings from the published version (instead of querying ResourceMapping on every request), with a migration to backfill existing data.

Changes:

• Add knowledge_ids JSONField to ApplicationVersion.
• Backfill knowledge_ids for existing SIMPLE applications from ResourceMapping via a data migration.
• Populate knowledge_ids during SIMPLE application publish; use ApplicationVersion.knowledge_ids for non-debug simple chat open/reopen paths.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.

File	Description
apps/chat/serializers/chat.py	Switches simple chat open/reopen to use ApplicationVersion.knowledge_ids when not in debug mode.
apps/application/serializers/application.py	Stores knowledge_ids into the newly created ApplicationVersion for SIMPLE app publishes; minor formatting changes.
apps/application/models/application.py	Adds the knowledge_ids field to the ApplicationVersion model.
apps/application/migrations/0014_applicationversion_knowledge_ids.py	Adds the DB column and backfills knowledge_ids for SIMPLE apps from ResourceMapping.

Comments suppressed due to low confidence (1)

apps/application/serializers/application.py:1853

• This field definition exceeds the repository's configured Python max line length (120) (see .editorconfig). Please wrap it to match the codebase style.

    file_clean_time = serializers.IntegerField(required=True, min_value=1, max_value=100000, label=_("File clean time"))

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[shaohuzhang1]

Seems you are using me but didn't get OPENAI_API_KEY seted in Variables/Secrets for this repo. you could follow readme for more information

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated no new comments.

Comments suppressed due to low confidence (6)

apps/chat/serializers/chat.py:507

• application_version is re-queried here and then dereferenced without a null check. Even though the caller checks for a published version, this extra query introduces a small race window (or alternate call path) that can raise an AttributeError when application_version is None. Add a guard and normalize knowledge_ids to a list.

            application_version = QuerySet(ApplicationVersion).filter(application_id=application.id).order_by(
                '-create_time')[0:1].first()
            knowledge_id_list = application_version.knowledge_ids

apps/chat/serializers/chat.py:600

• Same issue as above: application_version is queried again and then used without checking for None. If the version is deleted between the earlier publication check and this query, this will crash. Add a guard and normalize knowledge_ids to a list.

            application_version = QuerySet(ApplicationVersion).filter(application_id=application_id).order_by(
                '-create_time')[0:1].first()
            knowledge_id_list = application_version.knowledge_ids

apps/application/serializers/application.py:1795

• This refactor collapses a multi-line queryset chain into a long single line, which is inconsistent with the surrounding style in this file (e.g., application.py:302-315 wraps long calls) and makes it harder to read/review. Consider restoring the multi-line formatting.

        id_list = list(
            QuerySet(Application).filter(id__in=id_list, workspace_id=workspace_id).values_list("id", flat=True)
        )

apps/application/serializers/application.py:1280

• This persists knowledge_ids directly from ResourceMapping without de-duplication. Since resource_mapping has no uniqueness constraint, duplicates can be stored and later expand __in queries / downstream lists unnecessarily. Consider de-duping while preserving order (migration already de-dupes via dict.fromkeys).

        if application.type == ApplicationTypeChoices.SIMPLE:
            work_flow_version.knowledge_ids = [
                str(row.target_id)
                for row in QuerySet(ResourceMapping).filter(
                    source_id=str(application.id), source_type="APPLICATION", target_type="KNOWLEDGE"
                )
            ]

apps/application/migrations/0014_applicationversion_knowledge_ids.py:44

• This data migration materializes all matching ApplicationVersion rows into an in-memory updates list before a single bulk_update, and iterates over all versions even though only SIMPLE applications are relevant. On large datasets this can consume a lot of memory/time during migrations. Filter to the relevant versions and flush bulk_update in batches.

    updates = []
    for obj in ApplicationVersion.objects.using(db_alias).iterator():
        app_id = str(obj.application_id)
        if app_id not in simple_application_ids:
            continue
        knowledge_ids = mapping.get(app_id)
        if knowledge_ids:
            obj.knowledge_ids = knowledge_ids
            updates.append(obj)
    if updates:
        ApplicationVersion.objects.using(db_alias).bulk_update(
            updates, ["knowledge_ids"], batch_size=500
        )

apps/application/serializers/application.py:1853

• This line exceeds typical line-length/readability patterns used elsewhere in this file (e.g., multi-line serializer field declarations at application.py:302-304). Keeping it multi-line avoids overly long lines and matches the existing style.

    file_clean_time = serializers.IntegerField(required=True, min_value=1, max_value=100000, label=_("File clean time"))