Files
multica/server/pkg/db/queries/inbox.sql
Jiayuan Zhang f0110da555 feat(inbox): mark a notification unread from the row context menu (MUL-5496) (#6137)
The inbox auto-marks a notification read the moment it is selected, so
"opened" and "handled" were the same signal — a row you glanced at and
meant to come back to was gone from the unread count with no way back.

Right-click any inbox row for a shared context menu: Mark as read /
Mark as unread, plus Archive (Unarchive in the archived view).

- POST /api/inbox/{id}/unread + MarkInboxUnread query, publishing
  inbox:unread. Item-scoped, mirroring mark-read: the list renders one
  row per issue carrying that group's newest item, so flipping the whole
  group would resurrect siblings the user already dealt with.
- useMarkInboxUnread patches both lists optimistically and re-pulls the
  cross-workspace unread summary on settle.
- One shared menu per list rather than a Base UI root per row (the same
  shape IssueContextMenuProvider uses): only one is ever open, and a
  per-row root would unmount with its menu when the row scrolls out of
  the virtualized viewport.
- The read toggle is main-view only — archived rows deliberately render
  as read and the unread count excludes them, so a toggle there would
  report success and change nothing on screen.
- Parking the row that is currently open holds the auto-read effect off
  that one item while it stays selected; re-opening it later marks it
  read again.
- Mobile subscribes to inbox:unread so the unread dots agree across
  clients.

Co-authored-by: Lambda <lambda@multica.ai>
Co-authored-by: multica-agent <github@multica.ai>
2026-07-30 19:18:36 +08:00

148 lines
6.1 KiB
SQL

-- name: ListInboxItems :many
SELECT i.*,
iss.status as issue_status
FROM inbox_item i
LEFT JOIN issue iss ON iss.id = i.issue_id
WHERE i.workspace_id = $1 AND i.recipient_type = $2 AND i.recipient_id = $3 AND i.archived = false
ORDER BY i.created_at DESC;
-- name: ListArchivedInboxItems :many
-- Archived counterpart of ListInboxItems, backing the inbox's "Archived"
-- sub-view (MUL-3736).
--
-- An issue whose group still has an active row is excluded: archiving is
-- issue-level, so a NEW notification on an already-archived issue leaves the
-- old archived rows in place alongside the fresh active one. The issue belongs
-- in the main inbox at that point, and the two lists must stay mutually
-- exclusive per issue group — otherwise the same issue renders in both. The
-- exclusion lives here rather than in the client so neither list depends on
-- the other's cache being loaded. Items without an issue_id group on their own
-- id and can never have an active sibling, hence the IS NULL short-circuit.
--
-- LIMIT bounds the response while the archive grows without end (v1 ships no
-- pagination). Rows are newest-first, so truncation drops the OLDEST rows and
-- can never hide a group's newest row — the one the deduplicated UI renders.
SELECT i.*,
iss.status as issue_status
FROM inbox_item i
LEFT JOIN issue iss ON iss.id = i.issue_id
WHERE i.workspace_id = $1 AND i.recipient_type = $2 AND i.recipient_id = $3 AND i.archived = true
AND (i.issue_id IS NULL OR NOT EXISTS (
SELECT 1
FROM inbox_item active
WHERE active.workspace_id = i.workspace_id
AND active.recipient_type = i.recipient_type
AND active.recipient_id = i.recipient_id
AND active.issue_id = i.issue_id
AND active.archived = false
))
ORDER BY i.created_at DESC
LIMIT 200;
-- name: GetInboxItem :one
SELECT * FROM inbox_item
WHERE id = $1;
-- name: GetInboxItemInWorkspace :one
SELECT * FROM inbox_item
WHERE id = $1 AND workspace_id = $2;
-- name: CreateInboxItem :one
INSERT INTO inbox_item (
workspace_id, recipient_type, recipient_id,
type, severity, issue_id, title, body,
actor_type, actor_id, details
) VALUES ($1, $2, $3, $4, $5, $6, $7, $8, $9, $10, $11)
RETURNING *;
-- name: MarkInboxRead :one
UPDATE inbox_item SET read = true
WHERE id = $1
RETURNING *;
-- name: MarkInboxUnread :one
-- Exact inverse of MarkInboxRead, and item-level for the same reason it is:
-- the inbox renders one row per issue carrying that group's NEWEST item, and
-- the group's read state is that item's read state. Flipping the whole group
-- unread would resurrect older siblings the user already dealt with and
-- inflate CountUnreadInbox (which counts raw rows), while changing nothing the
-- UI shows.
UPDATE inbox_item SET read = false
WHERE id = $1
RETURNING *;
-- name: ArchiveInboxItem :one
UPDATE inbox_item SET archived = true
WHERE id = $1
RETURNING *;
-- name: ArchiveInboxByIssue :execrows
UPDATE inbox_item SET archived = true
WHERE workspace_id = $1 AND recipient_type = $2 AND recipient_id = $3 AND issue_id = $4 AND archived = false;
-- name: UnarchiveInboxItem :one
-- Deliberately does not touch `read`: unarchiving restores an item to the main
-- inbox in the exact read/unread state it was archived in, so restoring an
-- unread item legitimately raises the unread badge again (MUL-3736).
UPDATE inbox_item SET archived = false
WHERE id = $1
RETURNING *;
-- name: UnarchiveInboxByIssue :execrows
-- Issue-level restore, mirroring ArchiveInboxByIssue: archiving one item
-- archives every sibling for the same issue, so unarchiving must bring the
-- whole group back. Leaves `read` untouched for the same reason as above.
UPDATE inbox_item SET archived = false
WHERE workspace_id = $1 AND recipient_type = $2 AND recipient_id = $3 AND issue_id = $4 AND archived = true;
-- name: ArchiveInboxByIssueAndType :many
UPDATE inbox_item SET archived = true
WHERE workspace_id = $1 AND issue_id = $2 AND type = $3 AND archived = false
RETURNING recipient_type, recipient_id;
-- name: CountUnreadInbox :one
SELECT count(*) FROM inbox_item
WHERE workspace_id = $1 AND recipient_type = $2 AND recipient_id = $3 AND read = false AND archived = false;
-- name: CountUnreadInboxByWorkspace :many
-- Per-workspace unread inbox counts for a recipient member, matching the
-- inbox UI's deduplicated view: notifications are grouped per issue
-- (Linear-style, one row per issue) and an issue counts as unread only when
-- its NEWEST non-archived item is unread. Opening an issue marks just that
-- newest item read, so counting raw unread rows would keep older siblings
-- alive and light the switcher dot for a workspace whose inbox the user sees
-- as empty (MUL-3695). Items without an issue group on their own id. The
-- member join keeps counts scoped to workspaces the user still belongs to,
-- so a stale item left behind in a workspace the user has since left cannot
-- light the dot.
SELECT newest.workspace_id, count(*) AS count
FROM (
SELECT DISTINCT ON (i.workspace_id, COALESCE(i.issue_id, i.id))
i.workspace_id, i.read
FROM inbox_item i
JOIN member m ON m.workspace_id = i.workspace_id AND m.user_id = i.recipient_id
WHERE i.recipient_type = 'member'
AND i.recipient_id = $1
AND i.archived = false
ORDER BY i.workspace_id, COALESCE(i.issue_id, i.id), i.created_at DESC
) newest
WHERE newest.read = false
GROUP BY newest.workspace_id;
-- name: MarkAllInboxRead :execrows
UPDATE inbox_item SET read = true
WHERE workspace_id = $1 AND recipient_type = 'member' AND recipient_id = $2 AND archived = false AND read = false;
-- name: ArchiveAllInbox :execrows
UPDATE inbox_item SET archived = true
WHERE workspace_id = $1 AND recipient_type = 'member' AND recipient_id = $2 AND archived = false;
-- name: ArchiveAllReadInbox :execrows
UPDATE inbox_item SET archived = true
WHERE workspace_id = $1 AND recipient_type = 'member' AND recipient_id = $2 AND read = true AND archived = false;
-- name: ArchiveCompletedInbox :execrows
UPDATE inbox_item i SET archived = true
WHERE i.workspace_id = $1 AND i.recipient_type = 'member' AND i.recipient_id = $2 AND i.archived = false
AND i.issue_id IN (SELECT id FROM issue WHERE status IN ('done', 'cancelled'));