tsc -b (used in npm run build) flagged TS2774 because LucideIcon is a
function and therefore always truthy. Cast the lookup result to
LucideIcon | undefined so the falsy branch becomes meaningful.
Add a second tab in the node Icon picker to choose from the ~2250 brand
icons hosted by homarr-labs/dashboard-icons (Plex, Sonarr, Home
Assistant, etc.) served via jsDelivr CDN. The Generic tab keeps the
existing lucide picker unchanged.
Storage uses a 'brand:<slug>' prefix on custom_icon, so existing nodes
referencing lucide keys keep working with zero migration. A new
resolveCustomIcon helper returns a discriminated union (lucide | brand)
and a NodeIcon component centralizes rendering for both kinds.
Includes a manifest fetch script (scripts/fetch-dashboard-icons.mjs)
and a checked-in dashboardIcons.json snapshot.
Modal grew taller than viewport when icon picker expanded, hiding
header and footer buttons. Constrain DialogContent to 90vh and add
overflow-y-auto so all controls stay reachable.
Add zigbee_coordinator, zigbee_router, zigbee_enddevice to the
CustomStyleModal editable list with Radio/Zap/Lightbulb icons so users
can tweak color, opacity, and size like any other node type.
Code review flagged silent corruption when dragging a mixed selection
of a parent + one of its children: the child's ids stayed in
pendingSnap even though nodeBox excluded it from the bbox. On drag
stop we shifted the child's parent-relative position by the same delta
the parent already moves by — double-snapping the child off-screen.
Restrict the pendingSnap id set to nodes that contributed a box
(top-level only). Children follow their parent's move automatically;
no extra shift is needed.
Code review flagged a drag race + deps issues:
- Race: setNodes during onNodeDrag fights React Flow's internal drag
handler, which derives positions from the cursor offset captured at
drag start. Snap delta could be overwritten on the next frame.
- Deps: useCallback closed over guides.length, causing recreation on
every guide change and pulling stale references into onNodeDrag.
Fix: compute guides live during drag for visual feedback, store the
pending snap delta in a ref, and apply it via setNodes only on
onNodeDragStop. Replace the guides.length dep with a ref mirror updated
via useEffect (not during render).
CI build failed with TS2305: NodeDragHandler is not exported. Local
typecheck missed it because tsc-noEmit is more permissive than tsc -b.
Use the exported OnNodeDrag<NodeType> type and alias it with our
NodeData so the callback params are inferred (no implicit any).
Draw.io / Figma style: while dragging a node, show dashed cyan guide
lines when its edges (left / center / right / top / middle / bottom)
align with another node's, and snap the position to the matched line
within a configurable threshold.
- utils/alignment.ts: pure snap math, returns delta + guide segments.
Same-size boxes show all aligned guides simultaneously.
- canvas/AlignmentGuides.tsx: SVG overlay locked to the React Flow
viewport (panned/zoomed correctly).
- hooks/useAlignmentGuides.ts: wires onNodeDrag/onNodeDragStop, applies
snap via setNodes, listens for Alt to temporarily disable.
- utils/alignmentSettings.ts: localStorage-backed prefs (enabled,
threshold 2-16px) with a tiny CustomEvent pub-sub so the SettingsPanel
and the drag hook stay in sync without a global store.
- Sidebar settings panel: toggle + threshold slider.
Multi-selection drag uses the union bounding box. Children with
parentId are skipped for v1 to avoid mixing absolute and parent-relative
coordinates. Hold Alt to escape snap.
text_content is not in the API serializer schema, so text node content
was dropped on save and the node came back empty after reload.
Store text in label instead — already persisted, and TextNode + the
edit modal already fall back to label, so existing data stays
compatible. Clear stale text_content on update.
Regression test added for the text-node save/load roundtrip.
Backend sends ISO timestamps with +00:00 offset. Code blindly appended
'Z' when the string didn't end with 'Z', producing '...+00:00Z' which
parses to Invalid Date.
Detect any timezone marker (Z, z, or numeric offset like +00:00) and
only append 'Z' for naive datetimes.
Co-authored-by:
Approving a pending device created a Node with check_method=NULL, so
the status scheduler silently skipped it (scheduler.py:64 filters
falsy check_method) and the bubble stayed grey forever.
Default to 'ping' when the device has an IP, in both single and bulk
approve paths. Also persist caller-supplied check_method/check_target
in the single-approve path.
Add regression tests asserting default check_method='ping' after
approval for both endpoints.
ping(8) -W flag is milliseconds on macOS, seconds on Linux. Code used
-W 1 for both, giving macOS a 1ms timeout that fails any host with
RTT >1ms (typical wifi/mesh is 2-10ms).
Use -W 1000 on darwin to match the intended 1s timeout. Add regression
test asserting the platform-specific value.
- Skip NodeModal on double-click for 'group'/'groupRect' (use inline rename)
- Exempt 'group' from properties-clears-height rule in updateNode
(was collapsing group + hiding children with extent: 'parent')
- Enable ws: true on Vite /api proxy so /api/v1/status/ws/status upgrades
(status bubbles stayed grey because WS handshake failed)
- Add regression tests for group/groupRect height preservation
Build failed because Record<NodeType, ...> in nodeIcons.ts and
themes.ts requires every NodeType key. Add Type icon for text and
neutral border/icon palette per theme.
Trash button in PendingDevicesModal now removes only filtered devices
instead of wiping the whole pending list. Falls back to clearPending
when no filters are active.
Mirrors IP scan flow: POST /zigbee/import-pending now creates a
ScanRun(kind=zigbee, status=running) and returns immediately.
Networkmap fetch + pending upsert run in the background, status
transitions to done/error when finished.
Frontend: import modal closes on submit, scan history shows the
run with a ZIG/IP kind chip and toasts on completion. Pending
modal auto-refreshes when run finishes.
scan_runs.kind column added (default 'ip', idempotent migration).
Existing zigbee tests refactored to exercise _persist_pending_import
directly (background tasks don't see the test session); route test
verifies the run is created with kind=zigbee.
Replaces sidebar pending/hidden panels with a wide modal showing
devices as cards. Adds search, segmented source/status filters,
type filter, select mode for bulk approve/hide/restore, and
keyboard shortcuts. Hidden cards click-to-restore (no approval
detour); approval no longer pops the edit modal.
Backend: new restore + bulk-restore endpoints (hidden -> pending).
- SearchModal.tsx: same null-IP guard as SearchBar; also search by
friendly_name and ieee_address. Fixes Docker CI typecheck.
- python-multipart 0.0.26 -> 0.0.27 (CVE-2026-42561). Fixes
Security pip-audit.
Pre-push code review flagged the inline `with suppress(OperationalError)`
ALTERs as silently eating real errors. The project still uses inline
init_db migrations (CLAUDE.md aspires to Alembic but the runtime path
is inline), so as a minimum: add a `_try_migrate` helper that
distinguishes already-applied (debug log) from genuine failures
(warning log). Apply it to the Zigbee-related migrations only — the
legacy ALTERs above remain unchanged in this commit's scope.
Coordinator auto-approves to a canvas Node; routers/end devices land in
pending_devices keyed by IEEE. Discovered parent->child edges are stored
in pending_device_links so that approving a pending device later
auto-creates the Edge once both endpoints exist as canvas Nodes.
- new POST /api/v1/zigbee/import-pending (default mode in modal)
- new pending_device_links table; ieee_address on nodes + pending_devices
- pending_devices.ip migrated to nullable (table rebuild on existing DBs)
- approve / bulk-approve return auto-created edges; sidebar pushes them
into the canvas store with bottom -> top-t handles
- ZigbeeImportModal: radio toggle pending vs canvas; reset on close
- PendingDeviceModal: zigbee badge, IEEE/LQI/vendor/model rows, services
hidden for zigbee
- Sidebar pending row: ZIG source badge, LQI badge, friendly_name fallback
- SearchBar: null-safe IP, also searches friendly_name and ieee_address
- Tooltip trigger uses asChild to avoid nested-button hydration error
- Modal: small italic note under Test/Fetch buttons warning users that
large meshes can take several minutes
- Service: _NETWORKMAP_TIMEOUT 180s -> 300s (5 min) for very large meshes
Edges
- Z2M `links` is bidirectional and includes router mesh paths, which
caused duplicate edges and edges entering the coordinator from the
bottom. Walk `links` only to derive parent_id + LQI; build final
edges strictly from the parent->child hierarchy (one edge per
non-coordinator node). Result: parent bottom -> child top, every time.
- Tests: bidirectional pair collapses to one edge, router-mesh siblings
dropped, coordinator never receives an edge.
Auto-select
- After import, deselect existing canvas nodes and mark only the
freshly-imported ones as selected, so the user can drag the whole
subtree as a group.
The previous parser read `data.routes` which is just an echo of the
`routes` request flag (a boolean). On real brokers this caused
`TypeError: 'bool' object is not iterable` and 500s during /import.
- Rewrite parse_networkmap to read data.value.nodes + data.value.links
with fallback to data.{nodes,links} for legacy variants
- Defensive: drop links to unknown nodes, propagate lqi from link to
target node, extract model/vendor from definition block
- Bump networkmap timeout 10s -> 180s (large meshes are slow)
- Tests: rewrite fixture builders + sample payload to real Z2M shape;
add cases for legacy shape, routes:false echo (regression), malformed
list, link to unknown node, lqi propagation, definition extraction
- Update docs to mention 60s+ wait window
53 backend tests pass, mypy + ruff clean.
aiomqtt/paho exception strings can include the broker URI with embedded
credentials (mqtt://user:pass@host) or auth detail. The 502 response
from /import and the message field on /test-connection echoed these
verbatim via str(exc).
- Add _sanitize_mqtt_error() that maps known patterns (auth, refused,
DNS, TLS, timeout) to coarse, credential-free categories
- Original exception still logged at WARNING level for operator debug
- Drop hostname:port from TimeoutError messages
- /test-connection unexpected-error path no longer interpolates exc
Tests: 6 new (auth/refused/DNS/TLS/unknown sanitization, end-to-end
fetch_networkmap leak check).
Remy