databrickslabs / databrickslabs/ontos
[Improvement]: Comments: attachments
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 212
- Forks
- 71
- Avg merge
- 4d 10h
- Merged PRs (30d)
- 43
Description
Parent PRD
#783
What to build
Add file attachments to comments. New table comment_attachments: id (UUID PK), comment_id (FK → comments.id, indexed), filename, content_type, size_bytes, storage_path (Databricks Volume path), created_by, created_at. Blobs live under a deterministic Volume path keyed by comment id, reusing the Volume storage model established for DocumentMetadataDb. Add a server-mediated multipart upload endpoint plus list/download/delete, under the existing comments feature permission. Validate against a content-type/extension allowlist (images: png/jpg/gif/webp; docs: pdf/csv/txt/xlsx/docx), a per-file size cap (default 25 MB), and a per-comment count cap (default 10); rejections return a clear error. Caps/allowlist are settings-driven. No AV scan in v1 (no scanner service — documented v2 follow-up). Downloads served with Content-Disposition: attachment, never rendered inline as HTML. Lifecycle: soft-delete (reply tombstone or root cascade) keeps blobs and rows (recoverable); hard-delete (admin) removes rows and blobs. Frontend: upload/preview UI (attachment chips) on sidebar and embedded variants, gated by a comments.attachments feature flag; flag off ⇒ endpoints unavailable and attach affordance hidden.
See PRD sections "Solution" (attachments), "Implementation decisions" (attachments model), "Further notes" (attachment safety posture), and user stories 22–26.
Acceptance criteria
- Migration adds
comment_attachmentstable (+ix_comment_attachments_comment) - Multipart upload writes blob to a deterministic Volume path keyed by comment id and records a row
- Upload rejects oversized file, over-count comment, and disallowed type with clear errors
- Allowlist and caps are settings-driven
- Download served as
Content-Disposition: attachment, never inline HTML - List/download respect the comment's (thread-root) audience
- Soft-delete keeps blobs + rows; hard-delete reclaims blobs + rows
-
comments.attachmentsflag gates endpoints + UI; flag off hides attach affordance - Attachment chips render on sidebar and embedded variants
- Pytest (route) covers accept-within-caps, reject oversized/over-count/disallowed, soft- vs hard-delete blob lifecycle (storage boundary mocked), audience-respecting list/download
Blocked by
None strictly, but sequence last per PRD ship order (resolve/close → threading → rich text → attachments); audience-respecting list/download aligns with thread-root audience from #785.
User stories addressed
22, 23, 24, 25, 26
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start with the PRD sections named in the issue, then inspect the existing comments feature, its permission and feature-flag handling, and the Volume storage model established for DocumentMetadataDb. Review the route-level Pytest coverage and the thread-root audience behavior from #785. Done means all listed migration, upload, validation, lifecycle, audience, flag, UI, and test acceptance criteria pass.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- api, backend, database, frontend, security
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100