Compare commits

...

19 Commits

Author SHA1 Message Date
Kpa-clawbot fa5ac2751d fix: debounce distance index rebuild to prevent CPU hot loop
On busy meshes (325K+ transmissions, 50 observers), every ingest poll
triggers a full distance index rebuild (1M+ hop records) because
new observations frequently pick longer paths via pickBestObservation.
With 1-second poll intervals, the rebuild never finishes before the
next one starts, pegging CPU at 100% and starving the HTTP server.

Fix: mark the distance index dirty on path changes but only rebuild
at most every 30 seconds. The initial Load() rebuild still runs
synchronously, and distLast is set afterward to prevent an immediate
re-rebuild on the first ingest cycle.

Discovered on Cascadia Mesh instance (cascadiamesh.org) where the
server was completely unresponsive due to continuous distance index
rebuilds consuming all CPU.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-04-03 23:20:49 -07:00
you ddce26ff2d ci: pin build and deploy jobs to meshcore-vm runner 2026-04-04 04:21:48 +00:00
Kpa-clawbot ee29cc627f perf: parallelize expanded group fetches, use hashIndex Map lookup (#552)
## Summary
Fixes #388 — expanded groups were fetched sequentially with O(n)
`packets.find()` lookups.

## Changes
1. **Parallel fetch**: Replaced sequential `for...of + await` loop in
`loadPackets()` with `Promise.all()` so all expanded group children are
fetched concurrently.
2. **O(1) Map lookup**: Replaced 3 instances of `packets.find(p =>
p.hash === hash)` with `hashIndex.get(hash)`:
   - `loadPackets()` expanded group restore (~line 553)
   - `select-observation` click handler (~line 1015)
   - `pktToggleGroup()` (~line 2012)

## Perf justification
- **Before**: N expanded groups → N sequential API calls + N ×
O(packets.length) array scans
- **After**: N parallel API calls + N × O(1) Map lookups
- Typical N is 1-3 (minor severity as noted in issue), but the fix is
trivial and correct

## Tests
All existing tests pass: `test-packet-filter.js` (62), `test-aging.js`
(29), `test-frontend-helpers.js` (433).

Co-authored-by: you <you@example.com>
2026-04-03 21:09:17 -07:00
Kpa-clawbot f3caf42be4 feat: show transport badge in live packet feed (#551)
## Summary

Show the transport badge ("T") in the live packet feed, matching the
packets table (#337).

## Changes

- Add `transportBadge(pkt.route_type)` to all 4 feed rendering paths in
`live.js`:
  - Grouped feed items (initial history load)
  - `addFeedItemDOM()` (VCR replay)
  - Dedup new feed items (live WebSocket updates)
  - Node detail panel recent packets list
- Uses existing `transportBadge()` from `app.js` and `.badge-transport`
CSS from `style.css`

## Testing

- 2 new source-level assertions in `test-live.js` verifying
`transportBadge()` calls exist
- All existing tests pass (67 passed in test-live.js, no new failures)

Fixes #338

Co-authored-by: you <you@example.com>
2026-04-03 21:09:02 -07:00
Kpa-clawbot c34744247a fix: clean up nodeActivity in pruneStaleNodes to prevent memory leak (#553)
## Summary

`nodeActivity` (an object tracking per-node packet counts for heatmap
intensity) grows without bound — entries are added on every packet flash
but never removed, even when stale nodes are pruned.

## Changes

- **Delete `nodeActivity[key]`** alongside `nodeMarkers[key]` and
`nodeData[key]` when removing stale WS-only nodes in `pruneStaleNodes()`
- **Prune orphaned entries** — after the main prune loop, sweep
`nodeActivity` and delete any key that has no corresponding `nodeData`
entry (catches edge cases where nodes were removed by other code paths)
- Both run every 60s via the existing `pruneStaleNodes` interval timer

## Testing

- Added 2 regression tests in `test-frontend-helpers.js` verifying stale
node cleanup and orphan removal
- All 435 frontend helper tests pass, plus packet-filter (62) and aging
(29)

Fixes #390

---------

Co-authored-by: you <you@example.com>
2026-04-03 16:54:53 -07:00
Kpa-clawbot 10f712f9d7 fix: restructure scroll containers for iOS status bar tap-to-scroll (#330) (#554)
## Summary

Fixes #330 — iOS status bar tap-to-scroll broken because `#app` had
`overflow: hidden`, preventing `<body>` from being the scroll container.

## Approach: Option B from the issue

Instead of a JS polyfill, this restructures scroll containers so
`<body>` is the primary scroll container by default, which iOS Safari
requires for native status-bar tap-to-scroll.

### How it works

**`#app` default (body-scroll mode):** Uses `min-height` instead of
fixed `height`, no `overflow: hidden`. Content pushes beyond the
viewport and body scrolls naturally.

**`#app.app-fixed` (fixed-layout mode):** Restores the original `height:
calc(100dvh - 52px); overflow: hidden` for pages that need constrained
containers. The router in `app.js` toggles this class based on the
current page.

### Fixed-layout pages (`.app-fixed`)
These pages need fixed-height containers and are unchanged in behavior:
- **packets** — virtual scroll requires fixed-height `.panel-left` to
calculate visible rows
- **nodes** — split-panel layout with independently scrollable panels
- **map** — Leaflet requires fixed-dimension container
- **live** — Leaflet map (also has its own `#app:has(.live-page)`
override in live.css)
- **channels** — split-panel chat layout
- **audio-lab** — split-panel layout

### Body-scroll pages (no `.app-fixed`)
These pages now let the body scroll, enabling iOS tap-to-scroll:
- **analytics** — removed `overflow-y: auto; height: 100%`
- **observers** — removed `overflow-y: auto; height: calc(100vh - 56px)`
- **traces** — removed `overflow-y: auto; height: 100%`
- **home** — removed `#app:has(.home-hero)` override (no longer needed)
- **compare** — removed inline `overflow-y:auto; height:calc(100vh -
56px)`
- **perf** — removed inline `height:100%; overflow-y:auto`
- **observer-detail** — removed inline `overflow-y:auto;
height:calc(100vh - 56px)`
- **node-analytics** — removed inline `height:100%; overflow-y:auto`

### Files changed
| File | Change |
|------|--------|
| `public/style.css` | `#app` default → `min-height`; added `.app-fixed`
class |
| `public/app.js` | Router toggles `.app-fixed` based on page |
| `public/home.css` | Removed `#app:has()` workaround |
| `public/compare.js` | Removed inline overflow/height |
| `public/perf.js` | Removed inline overflow/height |
| `public/observer-detail.js` | Removed inline overflow/height |
| `public/node-analytics.js` | Removed inline overflow/height |

### What's preserved
- Sticky nav (`position: sticky; top: 0`) — works with body scroll
- Split-panel resize handles — unchanged, still in fixed containers
- Virtual scroll on packets page — unchanged, `.panel-left` still has
fixed height
- Leaflet maps — unchanged, containers still have fixed dimensions
- Mobile responsive overrides — unchanged

Co-authored-by: you <you@example.com>
2026-04-03 16:54:36 -07:00
Kpa-clawbot 412a8fdb8f feat: live map uses affinity-aware hop resolution (#528) (#550)
## Summary

Augments the shared `HopResolver` with neighbor-graph affinity data so
that when multiple nodes match a hop prefix, the resolver prefers
candidates that are known neighbors of the adjacent hop — instead of
relying solely on geo-distance.

Fixes #528

## Changes

### `public/hop-resolver.js`
- Added `affinityMap` — stores bidirectional neighbor adjacency with
scores
- Added `setAffinity(graph)` — ingests `/api/analytics/neighbor-graph`
edge data into O(1) Map lookups
- Added `getAffinity(pubkeyA, pubkeyB)` — returns affinity score between
two nodes (0 if not neighbors)
- Added `pickByAffinity(candidates, adjacentPubkey, anchor, ...)` —
picks best candidate: affinity-neighbor first (highest score), then
geo-distance fallback
- Modified forward and backward passes in `resolve()` to track the
previously-resolved pubkey and use `pickByAffinity` instead of raw
geo-sort

### `public/live.js`
- Added `fetchAffinityData()` — fetches `/api/analytics/neighbor-graph`
once and calls `HopResolver.setAffinity()`
- Added `startAffinityRefresh()` — refreshes affinity data every 60
seconds
- Both are called from `loadNodes()` after HopResolver is initialized

### `test-hop-resolver-affinity.js` (new)
- Affinity prefers neighbor candidate over geo-closest
- Cold start (no affinity data) falls back to geo-closest
- Null/undefined affinity doesn't crash
- Bidirectional score lookup
- Highest affinity score wins among multiple neighbors
- Unambiguous hops unaffected by affinity

## Performance

- API calls: 1 at load + 1 per 60s (no per-packet calls)
- Per-packet resolve: O(1) Map lookups, <0.5ms
- Memory: ~50KB for 2K-node graph

---------

Co-authored-by: you <you@example.com>
2026-04-03 16:32:53 -07:00
Kpa-clawbot 9a39198d92 fix: only count repeaters in hash collision analysis (#441) (#548)
Fixes #441

## Summary

Hash collision analysis was including ALL node types, inflating
collision counts with irrelevant data. Per MeshCore firmware analysis,
**only repeaters matter for collision analysis** — they're the only role
that forwards packets and appears in routing `path[]` arrays.

## Root Causes Fixed

1. **`hash_size==0` nodes counted in all buckets** — nodes with unknown
hash size were included via `cn.HashSize == bytes || cn.HashSize == 0`,
polluting every bucket
2. **Non-repeater roles included** — companions, rooms, sensors, and
observers were counted even though their hash collisions never cause
routing ambiguity

## Fix

Changed `computeHashCollisions()` filter from:
```go
// Before: include everything except companions
if cn.HashSize == bytes && cn.Role != "companion" {
```
To:
```go
// After: only include repeaters (per firmware analysis)
if cn.HashSize == bytes && cn.Role == "repeater" {
```

## Why only repeaters?

From [MeshCore firmware
analysis](https://github.com/Kpa-clawbot/CoreScope/issues/441#issuecomment-4185218547):
- Only repeaters override `allowPacketForward()` to return `true`
- Only repeaters append their hash to `path[]` during relay
- Companions, rooms, sensors, observers never forward packets
- Cross-role collisions are benign (companion silently drops, real
repeater still forwards)

## Tests
- `TestHashCollisionsOnlyRepeaters` — verifies companions, rooms,
sensors, and hash_size==0 nodes are all excluded

---------

Co-authored-by: you <you@example.com>
2026-04-03 14:23:13 -07:00
Kpa-clawbot 526ea8a1fc perf(live): chunk VCR replay packet processing to avoid UI freezes (#549)
## Summary

VCR replay functions (`vcrReplayFromTs`, `vcrRewind`,
`fetchNextReplayPage`) fetch up to 10K packets and process them all
synchronously on the main thread via `expandToBufferEntries`, causing
multi-second UI freezes — especially on mobile.

## Fix

- Added `expandToBufferEntriesAsync()` — processes packets in chunks of
200, yielding to the event loop via `setTimeout(0)` between chunks
- Updated all three VCR replay callers to use the async variant
- Kept the synchronous `expandToBufferEntries()` for backward
compatibility (tests, small datasets)
- Exposed `_liveExpandToBufferEntriesAsync` on window for test access

## Perf justification

- **Before:** 10K packets × ~2 observations = 20K+ objects created
synchronously, blocking the main thread for 1-3 seconds on mobile
- **After:** Same work split into chunks of 200 packets (~400 entries)
with event loop yields between chunks. Each chunk takes <5ms, keeping
the UI responsive (well under the 16ms frame budget)
- Chunk size of 200 is tunable via `VCR_CHUNK_SIZE`

## Tests

- Added regression test: sync expand correctness at scale (500 packets →
1000 entries)
- Added structural test: verifies `VCR_CHUNK_SIZE` exists and async
function yields via `setTimeout`
- All existing tests pass (`npm test`)

Fixes #395

---------

Co-authored-by: you <you@example.com>
2026-04-03 21:22:05 +00:00
Kpa-clawbot 8e42febc9c fix: virtual scroll height accounts for expanded group rows (#410) (#547)
## Summary

Fixes #410 — virtual scroll height miscalculation for expanded group
rows.

## Root Cause

When WebSocket messages add children to an already-expanded packet
group, `_rowCounts` becomes stale during the 200ms render debounce
window. Scroll events during this window call `renderVisibleRows()` with
stale row counts, causing wrong total height, spacer heights, and
visible range calculations.

## Changes

**public/packets.js:**
- Added `_rowCountsDirty` flag to track when row counts need
recomputation
- Added `_invalidateRowCounts()` — marks row counts as stale and clears
cumulative cache
- Added `_refreshRowCountsIfDirty()` — lazily recomputes `_rowCounts`
from `_displayPackets`
- Called `_invalidateRowCounts()` when WS handler adds children to
expanded groups (line ~402)
- Called `_refreshRowCountsIfDirty()` at top of `renderVisibleRows()`
before using row counts
- Reset `_rowCountsDirty` in all cleanup paths (destroy, empty display)

**test-packets.js:**
- Added 4 regression tests for `_invalidateRowCounts` /
`_refreshRowCountsIfDirty`

## Complexity

O(n) recomputation of `_rowCounts` when dirty (same as existing
`renderTableRows` path). Only triggers when WS modifies expanded group
children, which is infrequent relative to scroll events.

Co-authored-by: you <you@example.com>
2026-04-03 13:55:23 -07:00
Kpa-clawbot 59bff5462c fix: rate-limit cache invalidation to prevent 0% hit rate (#533) (#546)
## Summary

Fixes #533 — server cache hit rate always 0%.

## Root Cause

`invalidateCachesFor()` is called at the end of every
`IngestNewFromDB()` and `IngestNewObservations()` cycle (~2-5s). Since
new data arrives continuously, caches are cleared faster than any
analytics request can hit them, resulting in a permanent 0% cache hit
rate. The cache TTL (15s/60s) is irrelevant because entries are evicted
by invalidation long before they expire.

## Fix

Rate-limit cache invalidation with a 10-second cooldown:

- First call after cooldown goes through immediately
- Subsequent calls during cooldown accumulate dirty flags in
`pendingInv`
- Next call after cooldown merges pending + current flags and applies
them
- Eviction bypasses cooldown (data removal requires immediate clearing)

Analytics data may be at most ~10s stale, which is acceptable for a
dashboard.

## Changes

- **`store.go`**: Added `lastInvalidated`, `pendingInv`, `invCooldown`
fields. Refactored `invalidateCachesFor()` to rate-limit non-eviction
invalidation. Extracted `applyCacheInvalidation()` helper.
- **`cache_invalidation_test.go`**: Added 4 new tests:
- `TestInvalidationRateLimited` — verifies caches survive during
cooldown
  - `TestInvalidationCooldownAccumulatesFlags` — verifies flag merging
- `TestEvictionBypassesCooldown` — verifies eviction always clears
immediately
- `BenchmarkCacheHitDuringIngestion` — confirms 100% hit rate during
rapid ingestion (was 0%)

## Perf Proof

```
BenchmarkCacheHitDuringIngestion-16    3467889    1018 ns/op    100.0 hit%
```

Before: 0% hit rate under continuous ingestion. After: 100% hit rate
during cooldown periods.

Co-authored-by: you <you@example.com>
2026-04-03 13:53:58 -07:00
Kpa-clawbot 8c1cd8a9fe perf: track advert pubkeys incrementally, eliminate per-request JSON parsing (#360) (#544)
## Summary

`GetPerfStoreStats()` and `GetPerfStoreStatsTyped()` iterated **all**
ADVERT packets and called `json.Unmarshal` on each one — under a read
lock — on every `/api/perf` and `/api/health` request. With 5K+ adverts,
each health check triggered thousands of JSON parses.

## Fix

Added a refcounted `advertPubkeys map[string]int` to `PacketStore` that
tracks distinct pubkeys incrementally during `Load()`,
`IngestNewFromDB()`, and eviction. The perf/health handlers now just
read `len(s.advertPubkeys)` — O(1) with zero allocations.

## Benchmark Results (5K adverts, 200 distinct pubkeys)

| Method | ns/op | allocs/op |
|--------|-------|-----------|
| `GetPerfStoreStatsTyped` | **78** | **0** |
| `GetPerfStoreStats` | **2,565** | **9** |

Before this change, both methods performed O(N) JSON unmarshals per
call.

## Tests Added

- `TestAdvertPubkeyTracking` — verifies incremental tracking through
add/evict lifecycle
- `TestAdvertPubkeyPublicKeyField` — covers the `public_key` JSON field
variant
- `TestAdvertPubkeyNonAdvert` — ensures non-ADVERT packets don't affect
count
- `BenchmarkGetPerfStoreStats` — 5K adverts benchmark
- `BenchmarkGetPerfStoreStatsTyped` — 5K adverts benchmark

Fixes #360

---------

Co-authored-by: you <you@example.com>
2026-04-03 13:51:13 -07:00
Kpa-clawbot 29e8e37114 fix: mobile filter dropdown specificity prevents expansion (#534) (#541)
## Summary

Fixes #534 — mobile filter dropdown doesn't expand on packets page.

## Root Cause

CSS specificity battle in the mobile media query. The hide rule uses
`:not()` pseudo-classes which add specificity:

```css
/* Higher specificity due to :not() */
.filter-bar > *:not(.filter-toggle-btn):not(.col-toggle-wrap) { display: none; }

/* Lower specificity — loses even with .filters-expanded */
.filter-bar.filters-expanded > * { display: inline-flex; }
```

The JS toggle correctly adds/removes `.filters-expanded`, but the CSS
expanded rule could never win.

## Fix

Match the `:not()` selectors in the expanded rule so `.filters-expanded`
makes it strictly more specific:

```css
.filter-bar.filters-expanded > *:not(.filter-toggle-btn):not(.col-toggle-wrap) { display: inline-flex; }
```

Added a comment explaining the specificity dependency so future devs
don't repeat this.

## Tests

Added Playwright E2E test: mobile viewport (480×800), navigates to
packets page, clicks filter toggle, verifies filter inputs become
visible.

---------

Co-authored-by: you <you@example.com>
2026-04-03 13:50:10 -07:00
Kpa-clawbot 9b9f396af5 perf: replace O(n²) observation dedup with map-based O(n) (#355) (#543)
## Summary

Fixes #355 — replaces O(n²) observation dedup in `Load()`,
`IngestNewFromDB()`, and `IngestNewObservations()` with an O(1)
map-based lookup.

## Changes

- Added `obsKeys map[string]bool` field to `StoreTx` for O(1) dedup
keyed on `observerID + "|" + pathJSON`
- Replaced all 3 linear-scan dedup sites in `store.go` with map lookups
- Lazy-init `obsKeys` for transmissions created before this change (in
`IngestNewFromDB` and `IngestNewObservations`)
- Added regression test (`TestObsDedupCorrectness`) verifying dedup
correctness
- Added nil-map safety test (`TestObsDedupNilMapSafety`)
- Added benchmark comparing map vs linear scan

## Benchmark Results (ARM64, 16 cores)

| Observations | Map (O(1)) | Linear (O(n)) | Speedup |
|---|---|---|---|
| 10 | 34 ns/op | 41 ns/op | 1.2x |
| 50 | 34 ns/op | 186 ns/op | 5.5x |
| 100 | 34 ns/op | 361 ns/op | 10.6x |
| 500 | 34 ns/op | 4,903 ns/op | **146x** |

Map lookup is constant time regardless of observation count. The linear
scan degrades quadratically — at 500 observations per transmission
(realistic for popular packets seen by many observers), the old code is
146x slower per dedup check.

All existing tests pass.

---------

Co-authored-by: you <you@example.com>
2026-04-03 13:33:26 -07:00
Kpa-clawbot b472c8de30 perf: replace O(n²) selection sort with sort.Slice (#354) (#542)
## Summary

Fixes #354

Replaces the O(n²) selection sort in `sortedCopy()` with Go's built-in
`sort.Float64s()` (O(n log n)).

## Changes

- **`cmd/server/routes.go`**: Replaced manual nested-loop selection sort
with `sort.Float64s(cp)`
- **`cmd/server/helpers_test.go`**: Added regression test with
1000-element random input + benchmark

## Benchmark Results (ARM64)

```
BenchmarkSortedCopy/n=256     ~16μs/op    1 alloc
BenchmarkSortedCopy/n=1000    ~95μs/op    1 alloc
BenchmarkSortedCopy/n=10000   ~1.3ms/op   1 alloc
```

With the old O(n²) sort, n=10000 would take ~50ms+. The new
implementation scales as O(n log n).

## Testing

- All existing `TestSortedCopy` tests pass (unchanged behavior)
- New `TestSortedCopyLarge` validates correctness on 1000 random
elements
- `go test ./...` passes in `cmd/server`

Co-authored-by: you <you@example.com>
2026-04-03 13:11:59 -07:00
Kpa-clawbot 03e384bbc4 fix: null guard on pathHops prevents crash on ADVERT detail (#538) (#540)
## Summary

Fixes #538 — `null is not an object (evaluating 'pathHops.length')`
crash on ADVERT packet detail.

## Root Cause

`getParsedPath` caches its result as `p._parsedPath`. If another code
path (e.g., object spread, API response) sets `_parsedPath = null`, the
cache check (`!== undefined`) passes and returns `null` — causing
`.length` to crash.

Same pattern exists for `getParsedDecoded`.

## Changes

### `public/packet-helpers.js`
- `getParsedPath`: cached return now uses `|| []` to guard against null
cache
- `getParsedDecoded`: cached return now uses `|| {}` to guard against
null cache

### `public/packets.js`
- `renderDetail()` (line ~1440): defensive `|| []` / `|| {}` on
getParsedPath/getParsedDecoded calls
- `buildFlatRowHtml()` (line ~1103): same defensive guards

### `test-frontend-helpers.js`
- Added test: cached `_parsedPath = null` returns `[]`
- Added test: cached `_parsedDecoded = null` returns `{}`

## Testing

All 428 frontend helper tests pass. All 62 packet filter tests pass.

Co-authored-by: you <you@example.com>
2026-04-03 13:03:20 -07:00
Kpa-clawbot bf8c9e72ec fix: observer filter checks all observations in grouped mode (#537) (#539)
Fixes #537

## Problem
Observer filter in grouped mode only checked `p.observer_id` (the
primary observer), ignoring child observations. Grouped packets seen by
multiple observers would be hidden when filtering for a non-primary
observer.

## Fix
Two filter paths updated to also check `p._children`:

1. **Client-side display filter** (line ~1293): removed the
`!groupByHash` guard and added `_children` check so grouped packets are
included when any child observation matches
2. **WS real-time filter** (line ~360): added `_children` fallback check

The grouped row rendering (line ~1042) already correctly uses
`_observerFilterSet` for child filtering — no changes needed there.

## Tests
Added 5 tests in `test-frontend-helpers.js`:
- Grouped packet with matching child observer is shown
- Grouped packet with no matching observers is hidden  
- WS filter passes/rejects grouped packets correctly
- Source code assertions verifying both filter paths check `_children`

Co-authored-by: you <you@example.com>
2026-04-03 13:02:25 -07:00
Kpa-clawbot 48923db3d0 Add deep linking rule to AGENTS.md (#535)
Adds a rule to AGENTS.md requiring all new UI states to be
URL-addressable (deep-linkable). Part of #536.

Co-authored-by: you <you@example.com>
2026-04-03 13:01:31 -07:00
efiten 709e5a4776 fix: observer filter drops groups in grouped packets view (#464) (#531)
## Summary

- When `groupByHash=true`, each group only carries its representative
(best-path) `observer_id`. The client-side filter was checking only that
field, silently dropping groups that were seen by the selected observer
but had a different representative.
- `loadPackets` now passes the `observer` param to the server so
`filterPackets`/`buildGroupedWhere` do the correct "any observation
matches" check.
- Client-side observer filter in `renderTableRows` is skipped for
grouped mode (server already filtered correctly).
- Both `db.go` and `store.go` observer filtering extended to support
comma-separated IDs (multi-select UI).

## Test plan

- [ ] Set an observer filter on the Packets screen with grouping enabled
— all groups that have **any** observation from the selected observer(s)
should appear, not just groups where that observer is the representative
- [ ] Multi-select two observers — groups seen by either should appear
- [ ] Toggle to flat (ungrouped) mode — per-observation filter still
works correctly
- [ ] Existing grouped packets tests pass: `cd cmd/server && go test
./...`

Fixes #464

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---------

Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: you <you@example.com>
2026-04-03 09:22:37 -07:00
27 changed files with 1479 additions and 176 deletions
+2 -2
View File
@@ -236,7 +236,7 @@ jobs:
build:
name: "🏗️ Build Docker Image"
needs: [e2e-test]
runs-on: [self-hosted, Linux]
runs-on: [self-hosted, meshcore-vm]
steps:
- name: Checkout code
uses: actions/checkout@v5
@@ -271,7 +271,7 @@ jobs:
name: "🚀 Deploy Staging"
if: github.event_name == 'push'
needs: [build]
runs-on: [self-hosted, Linux]
runs-on: [self-hosted, meshcore-vm]
steps:
- name: Checkout code
uses: actions/checkout@v5
+6
View File
@@ -362,6 +362,12 @@ One logical change per commit. Each commit is deployable. Each commit has its te
- Tests: `test-{feature}.js` in repo root
- No build step, no transpilation — write ES2020 for server, ES5/6 for frontend (broad browser support)
### Deep Linking
All new UI states that a user might want to share or bookmark MUST be reflected in the URL hash.
This includes: tabs, filters, selected items, view modes. Use query parameters on the hash
(e.g., `#/packets?observer=ABC&timeRange=24h`) for filter state.
Existing patterns: `#/nodes/{pubkey}?section=node-neighbors`, `#/analytics?tab=collisions`, `#/packets/{hash}`.
## What NOT to Do
- **Don't check in private information** — no names, API keys, tokens, passwords, IP addresses, personal data, or any identifying information. This is a PUBLIC repo.
- Don't add npm dependencies without asking
+181
View File
@@ -0,0 +1,181 @@
package main
import (
"encoding/json"
"fmt"
"testing"
)
// TestAdvertPubkeyTracking verifies that advertPubkeys is maintained
// incrementally during ingest and eviction, and that GetPerfStoreStats
// returns the correct count without per-request JSON parsing.
func TestAdvertPubkeyTracking(t *testing.T) {
ps := NewPacketStore(nil, nil)
ps.mu.Lock()
// Helper to create an ADVERT StoreTx with a given pubkey.
pt4 := 4
mkAdvert := func(id int, pubkey string) *StoreTx {
d := map[string]interface{}{"pubKey": pubkey}
j, _ := json.Marshal(d)
return &StoreTx{
ID: id,
Hash: fmt.Sprintf("hash%d", id),
PayloadType: &pt4,
DecodedJSON: string(j),
}
}
// Add 3 adverts: 2 distinct pubkeys
tx1 := mkAdvert(1, "pk_alpha")
tx2 := mkAdvert(2, "pk_beta")
tx3 := mkAdvert(3, "pk_alpha") // duplicate pubkey
for _, tx := range []*StoreTx{tx1, tx2, tx3} {
ps.packets = append(ps.packets, tx)
ps.byHash[tx.Hash] = tx
ps.byTxID[tx.ID] = tx
ps.byPayloadType[4] = append(ps.byPayloadType[4], tx)
ps.trackAdvertPubkey(tx)
}
ps.mu.Unlock()
// GetPerfStoreStats should report 2 distinct pubkeys
stats := ps.GetPerfStoreStats()
indexes := stats["indexes"].(map[string]interface{})
got := indexes["advertByObserver"].(int)
if got != 2 {
t.Errorf("advertByObserver = %d, want 2", got)
}
// GetPerfStoreStatsTyped should agree
typed := ps.GetPerfStoreStatsTyped()
if typed.Indexes.AdvertByObserver != 2 {
t.Errorf("typed AdvertByObserver = %d, want 2", typed.Indexes.AdvertByObserver)
}
// Evict tx3 (pk_alpha duplicate) — count should stay 2
ps.mu.Lock()
ps.untrackAdvertPubkey(tx3)
ps.mu.Unlock()
stats2 := ps.GetPerfStoreStats()
idx2 := stats2["indexes"].(map[string]interface{})
if idx2["advertByObserver"].(int) != 2 {
t.Errorf("after evicting duplicate: advertByObserver = %d, want 2", idx2["advertByObserver"].(int))
}
// Evict tx1 (last pk_alpha) — count should drop to 1
ps.mu.Lock()
ps.untrackAdvertPubkey(tx1)
ps.mu.Unlock()
stats3 := ps.GetPerfStoreStats()
idx3 := stats3["indexes"].(map[string]interface{})
if idx3["advertByObserver"].(int) != 1 {
t.Errorf("after evicting last pk_alpha: advertByObserver = %d, want 1", idx3["advertByObserver"].(int))
}
// Evict tx2 (last remaining) — count should be 0
ps.mu.Lock()
ps.untrackAdvertPubkey(tx2)
ps.mu.Unlock()
stats4 := ps.GetPerfStoreStats()
idx4 := stats4["indexes"].(map[string]interface{})
if idx4["advertByObserver"].(int) != 0 {
t.Errorf("after evicting all: advertByObserver = %d, want 0", idx4["advertByObserver"].(int))
}
}
// TestAdvertPubkeyPublicKeyField tests the "public_key" JSON field variant.
func TestAdvertPubkeyPublicKeyField(t *testing.T) {
ps := NewPacketStore(nil, nil)
ps.mu.Lock()
pt4 := 4
d, _ := json.Marshal(map[string]interface{}{"public_key": "pk_legacy"})
tx := &StoreTx{ID: 1, Hash: "h1", PayloadType: &pt4, DecodedJSON: string(d)}
ps.trackAdvertPubkey(tx)
ps.mu.Unlock()
stats := ps.GetPerfStoreStats()
idx := stats["indexes"].(map[string]interface{})
if idx["advertByObserver"].(int) != 1 {
t.Errorf("public_key field: advertByObserver = %d, want 1", idx["advertByObserver"].(int))
}
}
// TestAdvertPubkeyNonAdvert ensures non-ADVERT packets don't affect the count.
func TestAdvertPubkeyNonAdvert(t *testing.T) {
ps := NewPacketStore(nil, nil)
ps.mu.Lock()
pt2 := 2
d, _ := json.Marshal(map[string]interface{}{"pubKey": "pk_text"})
tx := &StoreTx{ID: 1, Hash: "h1", PayloadType: &pt2, DecodedJSON: string(d)}
ps.trackAdvertPubkey(tx)
ps.mu.Unlock()
stats := ps.GetPerfStoreStats()
idx := stats["indexes"].(map[string]interface{})
if idx["advertByObserver"].(int) != 0 {
t.Errorf("non-ADVERT should not be tracked: advertByObserver = %d, want 0", idx["advertByObserver"].(int))
}
}
// BenchmarkGetPerfStoreStats benchmarks the perf stats endpoint with many adverts.
// Before the fix, this did O(N) JSON unmarshals per call.
// After the fix, it's O(1) — just len(map).
func BenchmarkGetPerfStoreStats(b *testing.B) {
ps := NewPacketStore(nil, nil)
ps.mu.Lock()
pt4 := 4
for i := 0; i < 5000; i++ {
pk := fmt.Sprintf("pk_%04d", i%200) // 200 distinct pubkeys
d, _ := json.Marshal(map[string]interface{}{"pubKey": pk})
tx := &StoreTx{
ID: i + 1,
Hash: fmt.Sprintf("hash%d", i+1),
PayloadType: &pt4,
DecodedJSON: string(d),
}
ps.packets = append(ps.packets, tx)
ps.byHash[tx.Hash] = tx
ps.byTxID[tx.ID] = tx
ps.byPayloadType[4] = append(ps.byPayloadType[4], tx)
ps.trackAdvertPubkey(tx)
}
ps.mu.Unlock()
b.ResetTimer()
for i := 0; i < b.N; i++ {
ps.GetPerfStoreStats()
}
}
// BenchmarkGetPerfStoreStatsTyped benchmarks the typed variant.
func BenchmarkGetPerfStoreStatsTyped(b *testing.B) {
ps := NewPacketStore(nil, nil)
ps.mu.Lock()
pt4 := 4
for i := 0; i < 5000; i++ {
pk := fmt.Sprintf("pk_%04d", i%200)
d, _ := json.Marshal(map[string]interface{}{"pubKey": pk})
tx := &StoreTx{
ID: i + 1,
Hash: fmt.Sprintf("hash%d", i+1),
PayloadType: &pt4,
DecodedJSON: string(d),
}
ps.packets = append(ps.packets, tx)
ps.byHash[tx.Hash] = tx
ps.byTxID[tx.ID] = tx
ps.byPayloadType[4] = append(ps.byPayloadType[4], tx)
ps.trackAdvertPubkey(tx)
}
ps.mu.Unlock()
b.ResetTimer()
for i := 0; i < b.N; i++ {
ps.GetPerfStoreStatsTyped()
}
}
+162
View File
@@ -16,6 +16,7 @@ func newTestStore(t *testing.T) *PacketStore {
distCache: make(map[string]*cachedResult),
subpathCache: make(map[string]*cachedResult),
rfCacheTTL: 15 * time.Second,
invCooldown: 10 * time.Second,
}
}
@@ -169,3 +170,164 @@ func TestInvalidateCachesFor_NoFlags(t *testing.T) {
}
}
}
// TestInvalidationRateLimited verifies that rapid ingest cycles don't clear
// caches immediately — they accumulate dirty flags during the cooldown period
// and apply them on the next call after cooldown expires (fixes #533).
func TestInvalidationRateLimited(t *testing.T) {
s := newTestStore(t)
s.invCooldown = 100 * time.Millisecond // short cooldown for testing
// First invalidation should go through immediately
populateAllCaches(s)
s.invalidateCachesFor(cacheInvalidation{hasNewObservations: true})
state := cachePopulated(s)
if state["rf"] {
t.Error("rf cache should be cleared on first invalidation")
}
if !state["topo"] {
t.Error("topo cache should survive (no path changes)")
}
// Repopulate and call again within cooldown — should NOT clear
populateAllCaches(s)
s.invalidateCachesFor(cacheInvalidation{hasNewObservations: true})
state = cachePopulated(s)
if !state["rf"] {
t.Error("rf cache should survive during cooldown period")
}
// Wait for cooldown to expire
time.Sleep(150 * time.Millisecond)
// Next call should apply accumulated + current flags
populateAllCaches(s)
s.invalidateCachesFor(cacheInvalidation{hasNewPaths: true})
state = cachePopulated(s)
if state["rf"] {
t.Error("rf cache should be cleared (pending from cooldown)")
}
if state["topo"] {
t.Error("topo cache should be cleared (current call has hasNewPaths)")
}
if !state["hash"] {
t.Error("hash cache should survive (no transmission changes)")
}
}
// TestInvalidationCooldownAccumulatesFlags verifies that multiple calls during
// cooldown merge their flags correctly.
func TestInvalidationCooldownAccumulatesFlags(t *testing.T) {
s := newTestStore(t)
s.invCooldown = 200 * time.Millisecond
// Initial invalidation (goes through, starts cooldown)
s.invalidateCachesFor(cacheInvalidation{hasNewObservations: true})
// Several calls during cooldown with different flags
s.invalidateCachesFor(cacheInvalidation{hasNewPaths: true})
s.invalidateCachesFor(cacheInvalidation{hasNewTransmissions: true})
s.invalidateCachesFor(cacheInvalidation{hasChannelData: true})
// Verify pending has all flags
s.cacheMu.Lock()
if s.pendingInv == nil {
t.Fatal("pendingInv should not be nil during cooldown")
}
if !s.pendingInv.hasNewPaths || !s.pendingInv.hasNewTransmissions || !s.pendingInv.hasChannelData {
t.Error("all flags should be accumulated in pendingInv")
}
// hasNewObservations was applied immediately, not accumulated
if s.pendingInv.hasNewObservations {
t.Error("hasNewObservations was already applied, should not be in pending")
}
s.cacheMu.Unlock()
// Wait for cooldown, then trigger — all accumulated flags should apply
time.Sleep(250 * time.Millisecond)
populateAllCaches(s)
s.invalidateCachesFor(cacheInvalidation{}) // empty trigger
state := cachePopulated(s)
// Pending had paths, transmissions, channels — all those caches should clear
if state["topo"] {
t.Error("topo should be cleared (pending hasNewPaths)")
}
if state["hash"] {
t.Error("hash should be cleared (pending hasNewTransmissions)")
}
if state["chan"] {
t.Error("chan should be cleared (pending hasChannelData)")
}
}
// TestEvictionBypassesCooldown verifies eviction always clears immediately.
func TestEvictionBypassesCooldown(t *testing.T) {
s := newTestStore(t)
s.invCooldown = 10 * time.Second // long cooldown
// Start cooldown
s.invalidateCachesFor(cacheInvalidation{hasNewObservations: true})
// Eviction during cooldown should still clear everything
populateAllCaches(s)
s.invalidateCachesFor(cacheInvalidation{eviction: true})
state := cachePopulated(s)
for name, has := range state {
if has {
t.Errorf("%s cache should be cleared on eviction even during cooldown", name)
}
}
// pendingInv should be cleared
s.cacheMu.Lock()
if s.pendingInv != nil {
t.Error("pendingInv should be nil after eviction")
}
s.cacheMu.Unlock()
}
// BenchmarkCacheHitDuringIngestion simulates rapid ingestion and verifies
// that cache hits now occur thanks to rate-limited invalidation.
func BenchmarkCacheHitDuringIngestion(b *testing.B) {
s := &PacketStore{
rfCache: make(map[string]*cachedResult),
topoCache: make(map[string]*cachedResult),
hashCache: make(map[string]*cachedResult),
chanCache: make(map[string]*cachedResult),
distCache: make(map[string]*cachedResult),
subpathCache: make(map[string]*cachedResult),
rfCacheTTL: 15 * time.Second,
invCooldown: 50 * time.Millisecond,
}
// Trigger first invalidation to start cooldown timer
s.invalidateCachesFor(cacheInvalidation{hasNewObservations: true})
var hits, misses int64
for i := 0; i < b.N; i++ {
// Populate cache (simulates an analytics query filling the cache)
s.cacheMu.Lock()
s.rfCache["global"] = &cachedResult{
data: map[string]interface{}{"test": true},
expiresAt: time.Now().Add(time.Hour),
}
s.cacheMu.Unlock()
// Simulate rapid ingest invalidation (should be rate-limited)
s.invalidateCachesFor(cacheInvalidation{hasNewObservations: true})
// Check if cache survived the invalidation
s.cacheMu.Lock()
if len(s.rfCache) > 0 {
hits++
} else {
misses++
}
s.cacheMu.Unlock()
}
if hits == 0 {
b.Errorf("expected cache hits > 0 with rate-limited invalidation, got 0 hits / %d misses", misses)
}
b.ReportMetric(float64(hits)/float64(hits+misses)*100, "hit%")
}
+102
View File
@@ -3811,3 +3811,105 @@ func BenchmarkIndexByNode(b *testing.B) {
}
})
}
// --- Multi-observer comma-separated filter tests ---
func TestTransmissionsForObserverMultiCSV(t *testing.T) {
db := setupTestDB(t)
defer db.Close()
seedTestData(t, db)
store := NewPacketStore(db, nil)
store.Load()
t.Run("comma-separated returns union via index", func(t *testing.T) {
result := store.transmissionsForObserver("obs1,obs2", nil)
if len(result) == 0 {
t.Fatal("expected results for obs1,obs2")
}
// obs1 has transmissions 1,2,3; obs2 has transmission 1
// Union should include all unique transmissions
obs1Only := store.transmissionsForObserver("obs1", nil)
obs2Only := store.transmissionsForObserver("obs2", nil)
if len(result) < len(obs1Only) || len(result) < len(obs2Only) {
t.Errorf("union (%d) should be >= each individual set (obs1=%d, obs2=%d)",
len(result), len(obs1Only), len(obs2Only))
}
})
t.Run("comma-separated with spaces via index", func(t *testing.T) {
result := store.transmissionsForObserver("obs1, obs2", nil)
if len(result) == 0 {
t.Fatal("expected results for 'obs1, obs2' (with space)")
}
noSpace := store.transmissionsForObserver("obs1,obs2", nil)
if len(result) != len(noSpace) {
t.Errorf("with-space (%d) should equal no-space (%d)", len(result), len(noSpace))
}
})
t.Run("comma-separated returns union via filter path", func(t *testing.T) {
allTx := store.packets
result := store.transmissionsForObserver("obs1,obs2", allTx)
if len(result) == 0 {
t.Fatal("expected results for obs1,obs2 via filter path")
}
})
t.Run("comma-separated with spaces via filter path", func(t *testing.T) {
allTx := store.packets
withSpace := store.transmissionsForObserver("obs1, obs2", allTx)
noSpace := store.transmissionsForObserver("obs1,obs2", allTx)
if len(withSpace) != len(noSpace) {
t.Errorf("filter path: with-space (%d) should equal no-space (%d)", len(withSpace), len(noSpace))
}
})
}
func TestBuildTransmissionWhereMultiObserver(t *testing.T) {
db := setupTestDB(t)
defer db.Close()
seedTestData(t, db)
t.Run("comma-separated produces IN clause", func(t *testing.T) {
q := PacketQuery{Observer: "obs1,obs2"}
where, args := db.buildTransmissionWhere(q)
if len(where) != 1 {
t.Fatalf("expected 1 WHERE clause, got %d", len(where))
}
clause := where[0]
if !strings.Contains(clause, "IN (?,?)") {
t.Errorf("expected IN (?,?) in clause, got: %s", clause)
}
if len(args) != 2 {
t.Fatalf("expected 2 args, got %d", len(args))
}
if args[0] != "obs1" || args[1] != "obs2" {
t.Errorf("expected [obs1, obs2], got %v", args)
}
})
t.Run("comma-separated with spaces trims IDs", func(t *testing.T) {
q := PacketQuery{Observer: "obs1, obs2"}
_, args := db.buildTransmissionWhere(q)
if len(args) != 2 {
t.Fatalf("expected 2 args, got %d", len(args))
}
if args[0] != "obs1" || args[1] != "obs2" {
t.Errorf("expected trimmed [obs1, obs2], got %v", args)
}
})
t.Run("single observer still works", func(t *testing.T) {
q := PacketQuery{Observer: "obs1"}
where, args := db.buildTransmissionWhere(q)
if len(where) != 1 {
t.Fatalf("expected 1 WHERE clause, got %d", len(where))
}
if !strings.Contains(where[0], "IN (?)") {
t.Errorf("expected IN (?) for single observer, got: %s", where[0])
}
if len(args) != 1 || args[0] != "obs1" {
t.Errorf("expected [obs1], got %v", args)
}
})
}
+8 -3
View File
@@ -608,12 +608,17 @@ func (db *DB) buildTransmissionWhere(q PacketQuery) ([]string, []interface{}) {
args = append(args, "%"+pk+"%")
}
if q.Observer != "" {
ids := strings.Split(q.Observer, ",")
placeholders := strings.Repeat("?,", len(ids))
placeholders = placeholders[:len(placeholders)-1]
if db.isV3 {
where = append(where, "EXISTS (SELECT 1 FROM observations oi JOIN observers obi ON obi.rowid = oi.observer_idx WHERE oi.transmission_id = t.id AND obi.id = ?)")
where = append(where, "EXISTS (SELECT 1 FROM observations oi JOIN observers obi ON obi.rowid = oi.observer_idx WHERE oi.transmission_id = t.id AND obi.id IN ("+placeholders+"))")
} else {
where = append(where, "EXISTS (SELECT 1 FROM observations oi WHERE oi.transmission_id = t.id AND oi.observer_id = ?)")
where = append(where, "EXISTS (SELECT 1 FROM observations oi WHERE oi.transmission_id = t.id AND oi.observer_id IN ("+placeholders+"))")
}
for _, id := range ids {
args = append(args, strings.TrimSpace(id))
}
args = append(args, q.Observer)
}
if q.Region != "" {
if db.isV3 {
+40
View File
@@ -2,6 +2,8 @@ package main
import (
"encoding/json"
"fmt"
"math/rand"
"net/http"
"net/http/httptest"
"os"
@@ -220,6 +222,44 @@ func TestSortedCopy(t *testing.T) {
}
}
func TestSortedCopyLarge(t *testing.T) {
// Regression: verify correct sort on larger input
rng := rand.New(rand.NewSource(42))
n := 1000
input := make([]float64, n)
for i := range input {
input[i] = rng.Float64() * 1000
}
result := sortedCopy(input)
if len(result) != n {
t.Fatalf("expected %d elements, got %d", n, len(result))
}
for i := 1; i < len(result); i++ {
if result[i] < result[i-1] {
t.Fatalf("not sorted at index %d: %v > %v", i, result[i-1], result[i])
}
}
// Original unchanged
if input[0] == result[0] && input[1] == result[1] && input[2] == result[2] {
// Could be coincidence but very unlikely with random data
}
}
func BenchmarkSortedCopy(b *testing.B) {
rng := rand.New(rand.NewSource(42))
for _, size := range []int{256, 1000, 10000} {
data := make([]float64, size)
for i := range data {
data[i] = rng.Float64() * 1000
}
b.Run(fmt.Sprintf("n=%d", size), func(b *testing.B) {
for i := 0; i < b.N; i++ {
sortedCopy(data)
}
})
}
}
func TestLastN(t *testing.T) {
arr := []map[string]interface{}{
{"id": 1}, {"id": 2}, {"id": 3}, {"id": 4}, {"id": 5},
+134
View File
@@ -0,0 +1,134 @@
package main
import (
"fmt"
"testing"
)
// TestObsDedupCorrectness verifies that the map-based dedup produces correct
// results: no duplicate observations (same observerID + pathJSON) on a single
// transmission.
func TestObsDedupCorrectness(t *testing.T) {
tx := &StoreTx{
ID: 1,
Hash: "abc123",
obsKeys: make(map[string]bool),
}
// Add 5 unique observations
for i := 0; i < 5; i++ {
obsID := fmt.Sprintf("obs-%d", i)
pathJSON := fmt.Sprintf(`["path-%d"]`, i)
dk := obsID + "|" + pathJSON
if tx.obsKeys[dk] {
t.Fatalf("observation %d should not be a duplicate", i)
}
tx.Observations = append(tx.Observations, &StoreObs{
ID: i,
ObserverID: obsID,
PathJSON: pathJSON,
})
tx.obsKeys[dk] = true
tx.ObservationCount++
}
if tx.ObservationCount != 5 {
t.Fatalf("expected 5 observations, got %d", tx.ObservationCount)
}
// Try to add duplicates of each — all should be rejected
for i := 0; i < 5; i++ {
obsID := fmt.Sprintf("obs-%d", i)
pathJSON := fmt.Sprintf(`["path-%d"]`, i)
dk := obsID + "|" + pathJSON
if !tx.obsKeys[dk] {
t.Fatalf("observation %d should be detected as duplicate", i)
}
}
// Same observer, different path — should NOT be a duplicate
dk := "obs-0" + "|" + `["different-path"]`
if tx.obsKeys[dk] {
t.Fatal("different path should not be a duplicate")
}
// Different observer, same path — should NOT be a duplicate
dk = "obs-new" + "|" + `["path-0"]`
if tx.obsKeys[dk] {
t.Fatal("different observer should not be a duplicate")
}
}
// TestObsDedupNilMapSafety ensures obsKeys lazy init works for pre-existing
// transmissions that may not have the map initialized.
func TestObsDedupNilMapSafety(t *testing.T) {
tx := &StoreTx{ID: 1, Hash: "abc"}
// obsKeys is nil — the lazy init pattern used in IngestNewFromDB/IngestNewObservations
if tx.obsKeys == nil {
tx.obsKeys = make(map[string]bool)
}
dk := "obs1|path1"
if tx.obsKeys[dk] {
t.Fatal("should not be duplicate on empty map")
}
tx.obsKeys[dk] = true
if !tx.obsKeys[dk] {
t.Fatal("should be duplicate after insert")
}
}
// BenchmarkObsDedupMap benchmarks the map-based O(1) dedup approach.
func BenchmarkObsDedupMap(b *testing.B) {
for _, obsCount := range []int{10, 50, 100, 500} {
b.Run(fmt.Sprintf("obs=%d", obsCount), func(b *testing.B) {
// Pre-populate a tx with obsCount observations
tx := &StoreTx{
ID: 1,
obsKeys: make(map[string]bool),
}
for i := 0; i < obsCount; i++ {
obsID := fmt.Sprintf("obs-%d", i)
pathJSON := fmt.Sprintf(`["hop-%d"]`, i)
dk := obsID + "|" + pathJSON
tx.Observations = append(tx.Observations, &StoreObs{
ObserverID: obsID,
PathJSON: pathJSON,
})
tx.obsKeys[dk] = true
}
// Benchmark: check dedup for a new observation (not duplicate)
newDK := "new-obs|new-path"
b.ResetTimer()
for i := 0; i < b.N; i++ {
_ = tx.obsKeys[newDK]
}
})
}
}
// BenchmarkObsDedupLinear benchmarks the old O(n) linear scan for comparison.
func BenchmarkObsDedupLinear(b *testing.B) {
for _, obsCount := range []int{10, 50, 100, 500} {
b.Run(fmt.Sprintf("obs=%d", obsCount), func(b *testing.B) {
tx := &StoreTx{ID: 1}
for i := 0; i < obsCount; i++ {
tx.Observations = append(tx.Observations, &StoreObs{
ObserverID: fmt.Sprintf("obs-%d", i),
PathJSON: fmt.Sprintf(`["hop-%d"]`, i),
})
}
newObsID := "new-obs"
newPath := "new-path"
b.ResetTimer()
for i := 0; i < b.N; i++ {
for _, existing := range tx.Observations {
if existing.ObserverID == newObsID && existing.PathJSON == newPath {
break
}
}
}
})
}
}
+1 -7
View File
@@ -1958,13 +1958,7 @@ func percentile(sorted []float64, p float64) float64 {
func sortedCopy(arr []float64) []float64 {
cp := make([]float64, len(arr))
copy(cp, arr)
for i := 0; i < len(cp); i++ {
for j := i + 1; j < len(cp); j++ {
if cp[j] < cp[i] {
cp[i], cp[j] = cp[j], cp[i]
}
}
}
sort.Float64s(cp)
return cp
}
+94 -3
View File
@@ -3059,11 +3059,11 @@ func TestHashCollisionsWithCollision(t *testing.T) {
now := time.Now().UTC()
recent := now.Add(-1 * time.Hour).Format(time.RFC3339)
// Two nodes with same first byte 'CC', no adverts so hash_size=0 (included in all buckets)
// Two repeater nodes with same first byte 'CC' and hash_size=1
db.conn.Exec(`INSERT INTO nodes (public_key, name, role, lat, lon, last_seen, first_seen, advert_count)
VALUES ('CC11223344556677', 'Node1', 'repeater', 37.5, -122.0, ?, '2026-01-01T00:00:00Z', 0)`, recent)
VALUES ('CC11223344556677', 'Node1', 'repeater', 37.5, -122.0, ?, '2026-01-01T00:00:00Z', 5)`, recent)
db.conn.Exec(`INSERT INTO nodes (public_key, name, role, lat, lon, last_seen, first_seen, advert_count)
VALUES ('CC99887766554433', 'Node2', 'repeater', 37.51, -122.01, ?, '2026-01-01T00:00:00Z', 0)`, recent)
VALUES ('CC99887766554433', 'Node2', 'repeater', 37.51, -122.01, ?, '2026-01-01T00:00:00Z', 5)`, recent)
cfg := &Config{Port: 3000}
hub := NewHub()
@@ -3072,6 +3072,14 @@ func TestHashCollisionsWithCollision(t *testing.T) {
if err := store.Load(); err != nil {
t.Fatalf("store.Load failed: %v", err)
}
// Inject hash_size=1 for both nodes so they appear in the 1-byte bucket
store.hashSizeInfoMu.Lock()
store.hashSizeInfoCache = map[string]*hashSizeNodeInfo{
"CC11223344556677": {HashSize: 1, AllSizes: map[int]bool{1: true}},
"CC99887766554433": {HashSize: 1, AllSizes: map[int]bool{1: true}},
}
store.hashSizeInfoAt = time.Now()
store.hashSizeInfoMu.Unlock()
srv.store = store
router := mux.NewRouter()
srv.RegisterRoutes(router)
@@ -3186,3 +3194,86 @@ func TestHashCollisionsMissingCoordinates(t *testing.T) {
}
}
}
// TestHashCollisionsOnlyRepeaters verifies that only repeater nodes
// are included in collision analysis. Companions, rooms, sensors, and
// hash_size==0 nodes are excluded — per firmware analysis, only repeaters
// forward packets and appear in path[] arrays. (#441)
func TestHashCollisionsOnlyRepeaters(t *testing.T) {
db := setupTestDB(t)
// Insert nodes sharing the same 1-byte prefix "AA":
// 1. repeater with hash_size=1 → should be counted
// 2. repeater with hash_size=0 (unknown) → should be excluded
// 3. companion with hash_size=1 → should be excluded
// 4. room with hash_size=1 → should be excluded
// 5. sensor with hash_size=1 → should be excluded
now := time.Now().Format("2006-01-02 15:04:05")
db.conn.Exec(`INSERT INTO nodes (public_key, name, role, last_seen) VALUES
('aa11223344556677', 'Repeater1', 'repeater', ?),
('aa99887766554433', 'UnknownNode', 'repeater', ?),
('aadeadbeefcafe01', 'Companion1', 'companion', ?),
('aabbcc1122334455', 'Room1', 'room', ?),
('aabbcc9988776655', 'Sensor1', 'sensor', ?)`, now, now, now, now, now)
// We also need a second repeater with hash_size=1 and same prefix to
// confirm that genuine collisions ARE still detected.
db.conn.Exec(`INSERT INTO nodes (public_key, name, role, last_seen) VALUES
('aa00112233445566', 'Repeater2', 'repeater', ?)`, now)
cfg := &Config{Port: 3000}
hub := NewHub()
srv := NewServer(db, cfg, hub)
store := NewPacketStore(db, nil)
store.Load()
srv.store = store
// Inject hash size info directly into the cache
store.hashSizeInfoMu.Lock()
store.hashSizeInfoCache = map[string]*hashSizeNodeInfo{
"aa11223344556677": {HashSize: 1, AllSizes: map[int]bool{1: true}},
"aa00112233445566": {HashSize: 1, AllSizes: map[int]bool{1: true}},
"aa99887766554433": {HashSize: 0, AllSizes: map[int]bool{}}, // unknown
"aadeadbeefcafe01": {HashSize: 1, AllSizes: map[int]bool{1: true}}, // companion
"aabbcc1122334455": {HashSize: 1, AllSizes: map[int]bool{1: true}}, // room
"aabbcc9988776655": {HashSize: 1, AllSizes: map[int]bool{1: true}}, // sensor
}
store.hashSizeInfoAt = time.Now()
store.hashSizeInfoMu.Unlock()
result := store.computeHashCollisions("")
bySize, ok := result["by_size"].(map[string]interface{})
if !ok {
t.Fatal("missing by_size")
}
size1, ok := bySize["1"].(map[string]interface{})
if !ok {
t.Fatal("missing by_size[1]")
}
stats, ok := size1["stats"].(map[string]interface{})
if !ok {
t.Fatal("missing stats")
}
// Only Repeater1 and Repeater2 should be in nodesForByte (hash_size=1, role=repeater).
// UnknownNode (hash_size=0), Companion1, Room1, Sensor1 must all be excluded.
nodesForByte := stats["nodes_for_byte"]
if nodesForByte != 2 {
t.Errorf("expected nodes_for_byte=2 (only repeaters with hash_size=1), got %v", nodesForByte)
}
// They share prefix "AA", so there should be exactly 1 collision entry.
collisions, ok := size1["collisions"].([]collisionEntry)
if !ok {
t.Fatalf("collisions is not []collisionEntry")
}
if len(collisions) != 1 {
t.Errorf("expected 1 collision entry, got %d", len(collisions))
}
if len(collisions) == 1 && len(collisions[0].Nodes) != 2 {
t.Errorf("expected 2 nodes in collision, got %d", len(collisions[0].Nodes))
}
}
+158 -98
View File
@@ -43,6 +43,8 @@ type StoreTx struct {
// Cached parsed fields (set once, read many)
parsedPath []string // cached parsePathJSON result
pathParsed bool // whether parsedPath has been set
// Dedup map: "observerID|pathJSON" → true for O(1) duplicate checks
obsKeys map[string]bool
}
// StoreObs is a lean in-memory observation (no duplication of transmission fields).
@@ -88,6 +90,10 @@ type PacketStore struct {
collisionCacheTTL time.Duration
cacheHits int64
cacheMisses int64
// Rate-limited invalidation (fixes #533: caches cleared faster than hit)
lastInvalidated time.Time
pendingInv *cacheInvalidation // accumulated dirty flags during cooldown
invCooldown time.Duration // minimum time between invalidations
// Short-lived cache for QueryGroupedPackets (avoids repeated full sort)
groupedCacheMu sync.Mutex
groupedCacheKey string
@@ -111,12 +117,18 @@ type PacketStore struct {
// computed during Load() and incrementally updated on ingest.
distHops []distHopRecord
distPaths []distPathRecord
distDirty bool // set when paths change; cleared after rebuild
distLast time.Time // last time distance index was rebuilt
// Cached GetNodeHashSizeInfo result — recomputed at most once every 15s
hashSizeInfoMu sync.Mutex
hashSizeInfoCache map[string]*hashSizeNodeInfo
hashSizeInfoAt time.Time
// Precomputed distinct advert pubkey count (refcounted for eviction correctness).
// Updated incrementally during Load/Ingest/Evict — avoids JSON parsing in GetPerfStoreStats.
advertPubkeys map[string]int // pubkey → number of advert packets referencing it
// Eviction config and stats
retentionHours float64 // 0 = unlimited
maxMemoryMB int // 0 = unlimited
@@ -182,7 +194,9 @@ func NewPacketStore(db *DB, cfg *PacketStoreConfig) *PacketStore {
subpathCache: make(map[string]*cachedResult),
rfCacheTTL: 15 * time.Second,
collisionCacheTTL: 60 * time.Second,
invCooldown: 10 * time.Second,
spIndex: make(map[string]int, 4096),
advertPubkeys: make(map[string]int),
}
if cfg != nil {
ps.retentionHours = cfg.RetentionHours
@@ -253,6 +267,7 @@ func (s *PacketStore) Load() error {
RouteType: nullIntPtr(routeType),
PayloadType: nullIntPtr(payloadType),
DecodedJSON: nullStrVal(decodedJSON),
obsKeys: make(map[string]bool),
}
s.byHash[hashStr] = tx
s.packets = append(s.packets, tx)
@@ -262,6 +277,7 @@ func (s *PacketStore) Load() error {
pt := *tx.PayloadType
s.byPayloadType[pt] = append(s.byPayloadType[pt], tx)
}
s.trackAdvertPubkey(tx)
}
if obsID.Valid {
@@ -269,15 +285,9 @@ func (s *PacketStore) Load() error {
obsIDStr := nullStrVal(observerID)
obsPJ := nullStrVal(pathJSON)
// Dedup: skip if same observer + same path already loaded
isDupe := false
for _, existing := range tx.Observations {
if existing.ObserverID == obsIDStr && existing.PathJSON == obsPJ {
isDupe = true
break
}
}
if isDupe {
// Dedup: skip if same observer + same path already loaded (O(1) map lookup)
dk := obsIDStr + "|" + obsPJ
if tx.obsKeys[dk] {
continue
}
@@ -295,6 +305,7 @@ func (s *PacketStore) Load() error {
}
tx.Observations = append(tx.Observations, obs)
tx.obsKeys[dk] = true
tx.ObservationCount++
if obs.Timestamp > tx.LatestSeen {
tx.LatestSeen = obs.Timestamp
@@ -320,6 +331,7 @@ func (s *PacketStore) Load() error {
// Precompute distance analytics (hop distances, path totals)
s.buildDistanceIndex()
s.distLast = time.Now()
s.loaded = true
elapsed := time.Since(t0)
@@ -392,6 +404,52 @@ func (s *PacketStore) indexByNode(tx *StoreTx) {
}
}
// trackAdvertPubkey increments the advertPubkeys refcount for ADVERT packets.
// Must be called under s.mu write lock.
func (s *PacketStore) trackAdvertPubkey(tx *StoreTx) {
if tx.PayloadType == nil || *tx.PayloadType != 4 || tx.DecodedJSON == "" {
return
}
var d map[string]interface{}
if json.Unmarshal([]byte(tx.DecodedJSON), &d) != nil {
return
}
pk := ""
if v, ok := d["pubKey"].(string); ok {
pk = v
} else if v, ok := d["public_key"].(string); ok {
pk = v
}
if pk != "" {
s.advertPubkeys[pk]++
}
}
// untrackAdvertPubkey decrements the advertPubkeys refcount for ADVERT packets.
// Must be called under s.mu write lock.
func (s *PacketStore) untrackAdvertPubkey(tx *StoreTx) {
if tx.PayloadType == nil || *tx.PayloadType != 4 || tx.DecodedJSON == "" {
return
}
var d map[string]interface{}
if json.Unmarshal([]byte(tx.DecodedJSON), &d) != nil {
return
}
pk := ""
if v, ok := d["pubKey"].(string); ok {
pk = v
} else if v, ok := d["public_key"].(string); ok {
pk = v
}
if pk != "" {
if s.advertPubkeys[pk] <= 1 {
delete(s.advertPubkeys, pk)
} else {
s.advertPubkeys[pk]--
}
}
}
// QueryPackets returns filtered, paginated packets from memory.
func (s *PacketStore) QueryPackets(q PacketQuery) *PacketResult {
atomic.AddInt64(&s.queryCount, 1)
@@ -579,30 +637,8 @@ func (s *PacketStore) GetPerfStoreStats() map[string]interface{} {
nodeIdx := len(s.byNode)
ptIdx := len(s.byPayloadType)
// Count distinct pubkeys with ADVERT observations (matches Node.js _advertByObserver.size)
advertByObsCount := 0
if adverts, ok := s.byPayloadType[4]; ok {
seen := make(map[string]bool)
for _, tx := range adverts {
if tx.DecodedJSON == "" {
continue
}
var d map[string]interface{}
if json.Unmarshal([]byte(tx.DecodedJSON), &d) != nil {
continue
}
pk := ""
if v, ok := d["pubKey"].(string); ok {
pk = v
} else if v, ok := d["public_key"].(string); ok {
pk = v
}
if pk != "" && !seen[pk] {
seen[pk] = true
advertByObsCount++
}
}
}
// Distinct advert pubkey count — precomputed incrementally (see trackAdvertPubkey).
advertByObsCount := len(s.advertPubkeys)
s.mu.RUnlock()
// Realistic estimate: ~5KB per packet + ~500 bytes per observation
@@ -690,15 +726,16 @@ type cacheInvalidation struct {
}
// invalidateCachesFor selectively clears only the analytics caches affected
// by the kind of data that changed. This avoids the previous behaviour of
// wiping every cache on every ingest cycle, which defeated caching under
// continuous ingestion (issue #375).
// by the kind of data that changed. To prevent continuous ingestion from
// defeating caching entirely (issue #533), invalidation is rate-limited:
// if called within invCooldown of the last invalidation, the flags are
// accumulated in pendingInv and applied on the next call after cooldown.
func (s *PacketStore) invalidateCachesFor(inv cacheInvalidation) {
s.cacheMu.Lock()
defer s.cacheMu.Unlock()
// Eviction bypasses rate-limiting — data was removed, caches must clear.
if inv.eviction {
// Eviction can affect any analytics — clear everything
s.rfCache = make(map[string]*cachedResult)
s.topoCache = make(map[string]*cachedResult)
s.hashCache = make(map[string]*cachedResult)
@@ -709,9 +746,40 @@ func (s *PacketStore) invalidateCachesFor(inv cacheInvalidation) {
s.channelsCacheMu.Lock()
s.channelsCacheRes = nil
s.channelsCacheMu.Unlock()
s.lastInvalidated = time.Now()
s.pendingInv = nil
return
}
now := time.Now()
if now.Sub(s.lastInvalidated) < s.invCooldown {
// Within cooldown — accumulate dirty flags
if s.pendingInv == nil {
s.pendingInv = &cacheInvalidation{}
}
s.pendingInv.hasNewObservations = s.pendingInv.hasNewObservations || inv.hasNewObservations
s.pendingInv.hasNewPaths = s.pendingInv.hasNewPaths || inv.hasNewPaths
s.pendingInv.hasNewTransmissions = s.pendingInv.hasNewTransmissions || inv.hasNewTransmissions
s.pendingInv.hasChannelData = s.pendingInv.hasChannelData || inv.hasChannelData
return
}
// Cooldown expired — merge any pending flags and apply
if s.pendingInv != nil {
inv.hasNewObservations = inv.hasNewObservations || s.pendingInv.hasNewObservations
inv.hasNewPaths = inv.hasNewPaths || s.pendingInv.hasNewPaths
inv.hasNewTransmissions = inv.hasNewTransmissions || s.pendingInv.hasNewTransmissions
inv.hasChannelData = inv.hasChannelData || s.pendingInv.hasChannelData
s.pendingInv = nil
}
s.applyCacheInvalidation(inv)
s.lastInvalidated = now
}
// applyCacheInvalidation performs the actual cache clearing. Must be called
// with cacheMu held.
func (s *PacketStore) applyCacheInvalidation(inv cacheInvalidation) {
if inv.hasNewObservations {
s.rfCache = make(map[string]*cachedResult)
}
@@ -726,7 +794,6 @@ func (s *PacketStore) invalidateCachesFor(inv cacheInvalidation) {
}
if inv.hasChannelData {
s.chanCache = make(map[string]*cachedResult)
// Also invalidate the separate channels list cache
s.channelsCacheMu.Lock()
s.channelsCacheRes = nil
s.channelsCacheMu.Unlock()
@@ -742,29 +809,7 @@ func (s *PacketStore) GetPerfStoreStatsTyped() PerfPacketStoreStats {
observerIdx := len(s.byObserver)
nodeIdx := len(s.byNode)
advertByObsCount := 0
if adverts, ok := s.byPayloadType[4]; ok {
seen := make(map[string]bool)
for _, tx := range adverts {
if tx.DecodedJSON == "" {
continue
}
var d map[string]interface{}
if json.Unmarshal([]byte(tx.DecodedJSON), &d) != nil {
continue
}
pk := ""
if v, ok := d["pubKey"].(string); ok {
pk = v
} else if v, ok := d["public_key"].(string); ok {
pk = v
}
if pk != "" && !seen[pk] {
seen[pk] = true
advertByObsCount++
}
}
}
advertByObsCount := len(s.advertPubkeys)
s.mu.RUnlock()
estimatedMB := math.Round(float64(totalLoaded*5120+totalObs*500)/1048576*10) / 10
@@ -779,7 +824,7 @@ func (s *PacketStore) GetPerfStoreStatsTyped() PerfPacketStoreStats {
SqliteOnly: false,
MaxPackets: 2386092,
EstimatedMB: estimatedMB,
MaxMB: 1024,
MaxMB: s.maxMemoryMB,
Indexes: PacketStoreIndexes{
ByHash: hashIdx,
ByObserver: observerIdx,
@@ -1061,6 +1106,7 @@ func (s *PacketStore) IngestNewFromDB(sinceID, limit int) ([]map[string]interfac
RouteType: r.routeType,
PayloadType: r.payloadType,
DecodedJSON: r.decodedJSON,
obsKeys: make(map[string]bool),
}
s.byHash[r.hash] = tx
s.packets = append(s.packets, tx) // oldest-first; new items go to tail
@@ -1072,6 +1118,7 @@ func (s *PacketStore) IngestNewFromDB(sinceID, limit int) ([]map[string]interfac
// so GetChannelMessages reverse iteration stays correct
s.byPayloadType[pt] = append(s.byPayloadType[pt], tx)
}
s.trackAdvertPubkey(tx)
if _, exists := broadcastTxs[r.txID]; !exists {
broadcastTxs[r.txID] = tx
@@ -1081,15 +1128,12 @@ func (s *PacketStore) IngestNewFromDB(sinceID, limit int) ([]map[string]interfac
if r.obsID != nil {
oid := *r.obsID
// Dedup
isDupe := false
for _, existing := range tx.Observations {
if existing.ObserverID == r.observerID && existing.PathJSON == r.pathJSON {
isDupe = true
break
}
// Dedup (O(1) map lookup)
dk := r.observerID + "|" + r.pathJSON
if tx.obsKeys == nil {
tx.obsKeys = make(map[string]bool)
}
if isDupe {
if tx.obsKeys[dk] {
continue
}
@@ -1106,6 +1150,7 @@ func (s *PacketStore) IngestNewFromDB(sinceID, limit int) ([]map[string]interfac
Timestamp: normalizeTimestamp(r.obsTS),
}
tx.Observations = append(tx.Observations, obs)
tx.obsKeys[dk] = true
tx.ObservationCount++
if obs.Timestamp > tx.LatestSeen {
tx.LatestSeen = obs.Timestamp
@@ -1326,15 +1371,12 @@ func (s *PacketStore) IngestNewObservations(sinceObsID, limit int) []map[string]
continue // transmission not yet in store
}
// Dedup by observer + path
isDupe := false
for _, existing := range tx.Observations {
if existing.ObserverID == r.observerID && existing.PathJSON == r.pathJSON {
isDupe = true
break
}
// Dedup by observer + path (O(1) map lookup)
dk := r.observerID + "|" + r.pathJSON
if tx.obsKeys == nil {
tx.obsKeys = make(map[string]bool)
}
if isDupe {
if tx.obsKeys[dk] {
continue
}
@@ -1351,6 +1393,7 @@ func (s *PacketStore) IngestNewObservations(sinceObsID, limit int) []map[string]
Timestamp: normalizeTimestamp(r.timestamp),
}
tx.Observations = append(tx.Observations, obs)
tx.obsKeys[dk] = true
tx.ObservationCount++
if obs.Timestamp > tx.LatestSeen {
tx.LatestSeen = obs.Timestamp
@@ -1430,13 +1473,19 @@ func (s *PacketStore) IngestNewObservations(sinceObsID, limit int) []map[string]
}
}
// Rebuild distance index if any paths changed (distances depend on path hops)
// Mark distance index dirty if any paths changed (rebuild is debounced)
for txID, tx := range updatedTxs {
if tx.PathJSON != oldPaths[txID] {
s.buildDistanceIndex()
s.distDirty = true
break
}
}
// Rebuild at most every 30s to avoid hot-looping on busy meshes
if s.distDirty && time.Since(s.distLast) > 30*time.Second {
s.buildDistanceIndex()
s.distDirty = false
s.distLast = time.Now()
}
if len(updatedTxs) > 0 {
// Targeted cache invalidation: new observations always affect RF
@@ -1572,32 +1621,36 @@ func (s *PacketStore) filterPackets(q PacketQuery) []*StoreTx {
}
// transmissionsForObserver returns unique transmissions for an observer.
func (s *PacketStore) transmissionsForObserver(observerID string, from []*StoreTx) []*StoreTx {
func (s *PacketStore) transmissionsForObserver(observerIDs string, from []*StoreTx) []*StoreTx {
ids := strings.Split(observerIDs, ",")
idSet := make(map[string]bool, len(ids))
for i, id := range ids {
ids[i] = strings.TrimSpace(id)
idSet[ids[i]] = true
}
if from != nil {
return filterTxSlice(from, func(tx *StoreTx) bool {
for _, obs := range tx.Observations {
if obs.ObserverID == observerID {
if idSet[obs.ObserverID] {
return true
}
}
return false
})
}
// Use byObserver index
observations := s.byObserver[observerID]
if len(observations) == 0 {
return nil
}
seen := make(map[int]bool, len(observations))
// Use byObserver index: union transmissions for all IDs
seen := make(map[int]bool)
var result []*StoreTx
for _, obs := range observations {
if seen[obs.TransmissionID] {
continue
}
seen[obs.TransmissionID] = true
tx := s.byTxID[obs.TransmissionID]
if tx != nil {
result = append(result, tx)
for _, id := range ids {
for _, obs := range s.byObserver[id] {
if seen[obs.TransmissionID] {
continue
}
seen[obs.TransmissionID] = true
tx := s.byTxID[obs.TransmissionID]
if tx != nil {
result = append(result, tx)
}
}
}
return result
@@ -1939,6 +1992,7 @@ func (s *PacketStore) EvictStale() int {
}
// Remove from byPayloadType
s.untrackAdvertPubkey(tx)
if tx.PayloadType != nil {
pt := *tx.PayloadType
ptList := s.byPayloadType[pt]
@@ -4473,10 +4527,16 @@ func (s *PacketStore) computeHashCollisions(region string) map[string]interface{
// Compute collisions for each byte size (1, 2, 3)
collisionsBySize := make(map[string]interface{})
for _, bytes := range []int{1, 2, 3} {
// Filter nodes relevant to this byte size
// Filter nodes relevant to this byte size.
// - Exclude hash_size==0 nodes: no adverts seen, so actual hash
// size is unknown. Including them in every bucket inflates
// collision counts.
// - Exclude companions: they are mobile/temporary and don't form
// the mesh backbone, so collisions with them aren't meaningful.
// (Fixes #441)
var nodesForByte []collisionNode
for _, cn := range allCNodes {
if cn.HashSize == bytes || cn.HashSize == 0 {
if cn.HashSize == bytes && cn.Role == "repeater" {
nodesForByte = append(nodesForByte, cn)
}
}
+3
View File
@@ -463,6 +463,9 @@ function navigate() {
currentPage = basePage;
const app = document.getElementById('app');
// Pages with fixed-height containers (maps, virtual-scroll, split-panels)
const fixedPages = { packets: 1, nodes: 1, map: 1, live: 1, channels: 1, 'audio-lab': 1 };
app.classList.toggle('app-fixed', basePage in fixedPages);
if (pages[basePage]?.init) {
const t0 = performance.now();
pages[basePage].init(app, routeParam);
+1 -1
View File
@@ -48,7 +48,7 @@ if (typeof window !== 'undefined') window.comparePacketSets = comparePacketSets;
packetsB = [];
currentView = 'summary';
app.innerHTML = '<div class="compare-page" style="overflow-y:auto;height:calc(100vh - 56px);padding:16px">' +
app.innerHTML = '<div class="compare-page" style="padding:16px">' +
'<div class="page-header" style="display:flex;align-items:center;gap:12px;margin-bottom:16px">' +
'<a href="#/observers" class="btn-icon" title="Back to Observers" aria-label="Back">\u2190</a>' +
'<h2 style="margin:0">\uD83D\uDD0D Observer Comparison</h2>' +
+1 -2
View File
@@ -1,7 +1,6 @@
/* === CoreScope — home.css === */
/* Override #app overflow:hidden for home page scrolling */
#app:has(.home-hero), #app:has(.home-chooser) { overflow-y: auto; }
/* Home page now uses body scroll (no #app override needed — see style.css) */
/* Chooser */
.home-chooser {
+79 -17
View File
@@ -11,6 +11,7 @@ window.HopResolver = (function() {
let nodesList = [];
let observerIataMap = {}; // observer_id → iata
let iataCoords = {}; // iata → {lat, lon}
let affinityMap = {}; // pubkey → { neighborPubkey → score }
function dist(lat1, lon1, lat2, lon2) {
return Math.sqrt((lat1 - lat2) ** 2 + (lon1 - lon2) ** 2);
@@ -67,6 +68,34 @@ window.HopResolver = (function() {
return null; // no GPS — can't geo-filter client-side
}
/**
* Pick the best candidate using affinity first, then geo-distance fallback.
* @param {Array} candidates - candidates with lat/lon/pubkey/name
* @param {string|null} adjacentPubkey - pubkey of the previously/next resolved hop
* @param {Object|null} anchor - {lat, lon} for geo fallback
* @param {number|null} fallbackLat - fallback anchor lat (e.g. observer)
* @param {number|null} fallbackLon - fallback anchor lon
* @returns {Object} best candidate
*/
function pickByAffinity(candidates, adjacentPubkey, anchor, fallbackLat, fallbackLon) {
// If we have affinity data and an adjacent hop, prefer neighbors
if (adjacentPubkey && Object.keys(affinityMap).length > 0) {
const withAffinity = candidates
.map(c => ({ ...c, affinity: getAffinity(adjacentPubkey, c.pubkey) }))
.filter(c => c.affinity > 0);
if (withAffinity.length > 0) {
withAffinity.sort((a, b) => b.affinity - a.affinity);
return withAffinity[0];
}
}
// Fallback: geo-distance sort (existing behavior)
const effectiveAnchor = anchor || (fallbackLat != null ? { lat: fallbackLat, lon: fallbackLon } : null);
if (effectiveAnchor) {
candidates.sort((a, b) => dist(a.lat, a.lon, effectiveAnchor.lat, effectiveAnchor.lon) - dist(b.lat, b.lon, effectiveAnchor.lat, effectiveAnchor.lon));
}
return candidates[0];
}
/**
* Resolve an array of hex hop prefixes to node info.
* Returns a map: { hop: {name, pubkey, lat, lon, ambiguous, unreliable} }
@@ -139,40 +168,50 @@ window.HopResolver = (function() {
// Forward pass
let lastPos = (originLat != null && originLon != null) ? { lat: originLat, lon: originLon } : null;
let lastResolvedPubkey = null;
for (let i = 0; i < hops.length; i++) {
const hop = hops[i];
if (hopPositions[hop]) { lastPos = hopPositions[hop]; continue; }
if (hopPositions[hop]) {
lastPos = hopPositions[hop];
lastResolvedPubkey = resolved[hop] ? resolved[hop].pubkey : null;
continue;
}
const r = resolved[hop];
if (!r || !r.ambiguous) continue;
const withLoc = r.candidates.filter(c => c.lat && c.lon && !(c.lat === 0 && c.lon === 0));
if (!withLoc.length) continue;
let anchor = lastPos;
if (!anchor && i === hops.length - 1 && observerLat != null) {
anchor = { lat: observerLat, lon: observerLon };
}
if (anchor) {
withLoc.sort((a, b) => dist(a.lat, a.lon, anchor.lat, anchor.lon) - dist(b.lat, b.lon, anchor.lat, anchor.lon));
}
r.name = withLoc[0].name;
r.pubkey = withLoc[0].pubkey;
hopPositions[hop] = { lat: withLoc[0].lat, lon: withLoc[0].lon };
// Affinity-aware: prefer candidates that are neighbors of the previous hop
const picked = pickByAffinity(withLoc, lastResolvedPubkey, lastPos, i === hops.length - 1 ? observerLat : null, i === hops.length - 1 ? observerLon : null);
r.name = picked.name;
r.pubkey = picked.pubkey;
hopPositions[hop] = { lat: picked.lat, lon: picked.lon };
lastPos = hopPositions[hop];
lastResolvedPubkey = picked.pubkey;
}
// Backward pass
let nextPos = (observerLat != null && observerLon != null) ? { lat: observerLat, lon: observerLon } : null;
let nextResolvedPubkey = null;
for (let i = hops.length - 1; i >= 0; i--) {
const hop = hops[i];
if (hopPositions[hop]) { nextPos = hopPositions[hop]; continue; }
if (hopPositions[hop]) {
nextPos = hopPositions[hop];
nextResolvedPubkey = resolved[hop] ? resolved[hop].pubkey : null;
continue;
}
const r = resolved[hop];
if (!r || !r.ambiguous) continue;
const withLoc = r.candidates.filter(c => c.lat && c.lon && !(c.lat === 0 && c.lon === 0));
if (!withLoc.length || !nextPos) continue;
withLoc.sort((a, b) => dist(a.lat, a.lon, nextPos.lat, nextPos.lon) - dist(b.lat, b.lon, nextPos.lat, nextPos.lon));
r.name = withLoc[0].name;
r.pubkey = withLoc[0].pubkey;
hopPositions[hop] = { lat: withLoc[0].lat, lon: withLoc[0].lon };
// Affinity-aware: prefer candidates that are neighbors of the next hop
const picked = pickByAffinity(withLoc, nextResolvedPubkey, nextPos, null, null);
r.name = picked.name;
r.pubkey = picked.pubkey;
hopPositions[hop] = { lat: picked.lat, lon: picked.lon };
nextPos = hopPositions[hop];
nextResolvedPubkey = picked.pubkey;
}
// Sanity check: drop hops impossibly far from neighbors
@@ -203,5 +242,28 @@ window.HopResolver = (function() {
return nodesList.length > 0;
}
return { init: init, resolve: resolve, ready: ready, haversineKm: haversineKm };
/**
* Load neighbor-graph affinity data.
* @param {Object} graph - { edges: [{source, target, score, weight}, ...] }
*/
function setAffinity(graph) {
affinityMap = {};
if (!graph || !graph.edges) return;
for (const e of graph.edges) {
if (!affinityMap[e.source]) affinityMap[e.source] = {};
affinityMap[e.source][e.target] = e.score || e.weight || 1;
if (!affinityMap[e.target]) affinityMap[e.target] = {};
affinityMap[e.target][e.source] = e.score || e.weight || 1;
}
}
/**
* Get the affinity score between two pubkeys (0 if not neighbors).
*/
function getAffinity(pubkeyA, pubkeyB) {
if (!pubkeyA || !pubkeyB || !affinityMap[pubkeyA]) return 0;
return affinityMap[pubkeyA][pubkeyB] || 0;
}
return { init: init, resolve: resolve, ready: ready, haversineKm: haversineKm, setAffinity: setAffinity, getAffinity: getAffinity };
})();
+99 -14
View File
@@ -43,6 +43,7 @@
timelineScope: 3600000, // 1h default ms
timelineTimestamps: [], // historical timestamps from DB for sparkline
timelineFetchedScope: 0, // last fetched scope to avoid redundant fetches
replayGen: 0, // generation counter — incremented on each replay/rewind to discard stale async results
};
// ROLE_COLORS loaded from shared roles.js (includes 'unknown')
@@ -116,6 +117,7 @@
function vcrResumeLive() {
stopReplay();
VCR.replayGen++; // invalidate any in-flight async chunk processing
VCR.playhead = -1;
VCR.speed = 1;
VCR.missedCount = 0;
@@ -142,6 +144,8 @@
function vcrReplayFromTs(targetTs) {
const fetchFrom = new Date(targetTs).toISOString();
stopReplay();
VCR.replayGen++;
var gen = VCR.replayGen;
vcrSetMode('REPLAY');
// Reload map nodes to match the replay time
@@ -153,7 +157,10 @@
.then(r => r.json())
.then(data => {
const pkts = data.packets || [];
const replayEntries = expandToBufferEntries(pkts);
return expandToBufferEntriesAsync(pkts);
})
.then(function(replayEntries) {
if (gen !== VCR.replayGen) return; // stale async result — user changed mode
if (replayEntries.length === 0) {
vcrSetMode('PAUSED');
return;
@@ -202,6 +209,8 @@
function vcrRewind(ms) {
stopReplay();
VCR.replayGen++;
var gen = VCR.replayGen;
// Fetch packets from DB for the time window
const now = Date.now();
const from = new Date(now - ms).toISOString();
@@ -212,8 +221,11 @@
// Prepend to buffer (avoid duplicates by ID)
const existingIds = new Set(VCR.buffer.map(b => b.pkt.id).filter(Boolean));
const filtered = pkts.filter(p => !existingIds.has(p.id));
const newEntries = expandToBufferEntries(filtered);
VCR.buffer = [...newEntries, ...VCR.buffer];
return expandToBufferEntriesAsync(filtered);
})
.then(function(newEntries) {
if (gen !== VCR.replayGen) return; // stale async result
VCR.buffer = [].concat(newEntries, VCR.buffer);
VCR.playhead = 0;
VCR.speed = 1;
vcrSetMode('REPLAY');
@@ -274,15 +286,18 @@
// Get timestamp of last packet in buffer to fetch the next page
const last = VCR.buffer[VCR.buffer.length - 1];
if (!last) return Promise.resolve(false);
var gen = VCR.replayGen;
const since = new Date(last.ts + 1).toISOString(); // +1ms to avoid dupe
return fetch(`/api/packets?limit=10000&grouped=false&expand=observations&since=${encodeURIComponent(since)}&order=asc`)
.then(r => r.json())
.then(data => {
const pkts = data.packets || [];
if (pkts.length === 0) return false;
const newEntries = expandToBufferEntries(pkts);
VCR.buffer = VCR.buffer.concat(newEntries);
return true;
return expandToBufferEntriesAsync(pkts).then(function(newEntries) {
if (gen !== VCR.replayGen) return false; // stale
VCR.buffer = VCR.buffer.concat(newEntries);
return true;
});
})
.catch(() => false);
}
@@ -449,11 +464,53 @@
}
// Expand a DB packet (with optional observations[]) into VCR buffer entries
/**
* Process packets into buffer entries in chunks to avoid blocking the main thread.
* Returns a Promise that resolves with the entries array.
* Each chunk processes CHUNK_SIZE packets, then yields to the event loop via setTimeout(0).
*/
var VCR_CHUNK_SIZE = 200;
function expandToBufferEntriesAsync(pkts) {
return new Promise(function(resolve) {
var entries = [];
var i = 0;
function processChunk() {
var end = Math.min(i + VCR_CHUNK_SIZE, pkts.length);
for (; i < end; i++) {
var p = pkts[i];
if (p.observations && p.observations.length > 0) {
for (var j = 0; j < p.observations.length; j++) {
var obs = p.observations[j];
entries.push({
ts: new Date(obs.timestamp || p.timestamp || p.created_at).getTime(),
pkt: dbPacketToLive(Object.assign({}, p, obs, { hash: p.hash, raw_hex: p.raw_hex, decoded_json: p.decoded_json }))
});
}
} else {
entries.push({
ts: new Date(p.timestamp || p.created_at).getTime(),
pkt: dbPacketToLive(p)
});
}
}
if (i < pkts.length) {
setTimeout(processChunk, 0);
} else {
resolve(entries);
}
}
processChunk();
});
}
// Synchronous version kept for small datasets and backward compat (tests)
function expandToBufferEntries(pkts) {
const entries = [];
for (const p of pkts) {
var entries = [];
for (var k = 0; k < pkts.length; k++) {
var p = pkts[k];
if (p.observations && p.observations.length > 0) {
for (const obs of p.observations) {
for (var j = 0; j < p.observations.length; j++) {
var obs = p.observations[j];
entries.push({
ts: new Date(obs.timestamp || p.timestamp || p.created_at).getTime(),
pkt: dbPacketToLive(Object.assign({}, p, obs, { hash: p.hash, raw_hex: p.raw_hex, decoded_json: p.decoded_json }))
@@ -1286,7 +1343,7 @@
html += `<h4 style="font-size:12px;margin:12px 0 6px;color:var(--text-muted);">Recent Packets</h4>
<div style="font-size:11px;max-height:200px;overflow-y:auto;">` +
recent.slice(0, 10).map(p => `<div style="padding:2px 0;display:flex;justify-content:space-between;">
<a href="#/packets/${encodeURIComponent(p.hash || '')}" style="color:var(--accent);text-decoration:none;">${escapeHtml(p.payload_type || '?')}${p.observation_count > 1 ? ' <span class="badge badge-obs" style="font-size:9px">👁 ' + p.observation_count + '</span>' : ''}</a>
<a href="#/packets/${encodeURIComponent(p.hash || '')}" style="color:var(--accent);text-decoration:none;">${escapeHtml(p.payload_type || '?')}${transportBadge(p.route_type)}${p.observation_count > 1 ? ' <span class="badge badge-obs" style="font-size:9px">👁 ' + p.observation_count + '</span>' : ''}</a>
<span style="color:var(--text-muted)">${formatLiveTimestampHtml(p.timestamp)}</span>
</div>`).join('') +
'</div>';
@@ -1359,9 +1416,29 @@
const _el2 = document.getElementById('liveNodeCount'); if (_el2) _el2.textContent = Object.keys(nodeMarkers).length;
// Initialize shared HopResolver with loaded nodes
if (window.HopResolver) HopResolver.init(list);
// Fetch affinity data for hop disambiguation
fetchAffinityData();
startAffinityRefresh();
} catch (e) { console.error('Failed to load nodes:', e); }
}
let _affinityInterval = null;
async function fetchAffinityData() {
try {
const resp = await fetch('/api/analytics/neighbor-graph');
const graph = await resp.json();
if (window.HopResolver && HopResolver.setAffinity) {
HopResolver.setAffinity(graph);
}
} catch (e) { console.warn('Failed to fetch affinity data:', e); }
}
function startAffinityRefresh() {
if (_affinityInterval) clearInterval(_affinityInterval);
_affinityInterval = setInterval(fetchAffinityData, 60000);
}
function clearNodeMarkers() {
if (nodesLayer) nodesLayer.clearLayers();
if (animLayer) animLayer.clearLayers();
@@ -1471,7 +1548,7 @@
item.innerHTML = `
<span class="feed-icon" style="color:${color}">${icon}</span>
<span class="feed-type" style="color:${color}">${typeName}</span>
${hopStr}${obsBadge}
${transportBadge(pkt.route_type)}${hopStr}${obsBadge}
<span class="feed-text">${escapeHtml(preview)}</span>
<span class="feed-time">${formatLiveTimestampHtml(group.latestTs || Date.now())}</span>
`;
@@ -1573,6 +1650,7 @@
}
delete nodeMarkers[key];
delete nodeData[key];
delete nodeActivity[key];
pruned = true;
}
} else if (marker && marker._staleDimmed) {
@@ -1588,15 +1666,21 @@
if (_el2) _el2.textContent = Object.keys(nodeMarkers).length;
if (window.HopResolver) HopResolver.init(Object.values(nodeData));
}
// Prune orphaned nodeActivity entries (nodes removed above or never tracked)
for (var aKey in nodeActivity) {
if (!(aKey in nodeData)) delete nodeActivity[aKey];
}
}
// Expose for testing
window._livePruneStaleNodes = pruneStaleNodes;
window._liveNodeMarkers = function() { return nodeMarkers; };
window._liveNodeData = function() { return nodeData; };
window._liveNodeActivity = function() { return nodeActivity; };
window._vcrFormatTime = vcrFormatTime;
window._liveDbPacketToLive = dbPacketToLive;
window._liveExpandToBufferEntries = expandToBufferEntries;
window._liveExpandToBufferEntriesAsync = expandToBufferEntriesAsync;
window._liveSEG_MAP = SEG_MAP;
window._liveBufferPacket = bufferPacket;
window._liveVCR = function() { return VCR; };
@@ -2406,7 +2490,7 @@
item.innerHTML = `
<span class="feed-icon" style="color:${color}">${icon}</span>
<span class="feed-type" style="color:${color}">${typeName}</span>
${hopStr}${obsBadge}
${transportBadge(pkt.route_type)}${hopStr}${obsBadge}
<span class="feed-text">${escapeHtml(preview)}</span>
<span class="feed-time">${formatLiveTimestampHtml(pkt._ts || Date.now())}</span>
`;
@@ -2474,7 +2558,7 @@
item.innerHTML = `
<span class="feed-icon" style="color:${color}">${icon}</span>
<span class="feed-type" style="color:${color}">${typeName}</span>
${hopStr}${obsBadge}
${transportBadge(pkt.route_type)}${hopStr}${obsBadge}
<span class="feed-text">${escapeHtml(preview)}</span>
<span class="feed-time">${formatLiveTimestampHtml(pkt._ts || Date.now())}</span>
`;
@@ -2552,6 +2636,7 @@
if (_lcdClockInterval) { clearInterval(_lcdClockInterval); _lcdClockInterval = null; }
if (_rateCounterInterval) { clearInterval(_rateCounterInterval); _rateCounterInterval = null; }
if (_pruneInterval) { clearInterval(_pruneInterval); _pruneInterval = null; }
if (_affinityInterval) { clearInterval(_affinityInterval); _affinityInterval = null; }
if (ws) { ws.onclose = null; ws.close(); ws = null; }
if (map) { map.remove(); map = null; }
if (_onResize) {
@@ -2584,7 +2669,7 @@
packetCount = 0; activeAnims = 0;
nodeActivity = {}; pktTimestamps = [];
feedDedup.clear();
VCR.buffer = []; VCR.playhead = -1; VCR.mode = 'LIVE'; VCR.missedCount = 0; VCR.speed = 1;
VCR.buffer = []; VCR.playhead = -1; VCR.mode = 'LIVE'; VCR.missedCount = 0; VCR.speed = 1; VCR.replayGen = 0;
}
let _themeRefreshHandler = null;
+1 -1
View File
@@ -51,7 +51,7 @@
const nodeName = escapeHtml(n.name || n.public_key.slice(0, 12));
container.innerHTML = `
<div style="max-width:1000px;margin:0 auto;padding:12px 16px;height:100%;overflow-y:auto">
<div style="max-width:1000px;margin:0 auto;padding:12px 16px">
<div style="margin-bottom:12px">
<a href="#/nodes/${encodeURIComponent(n.public_key)}" style="color:var(--accent);text-decoration:none;font-size:12px"> Back to ${nodeName}</a>
<h2 style="margin:4px 0 2px;font-size:18px">📊 ${nodeName} Analytics</h2>
+1 -1
View File
@@ -37,7 +37,7 @@
}
app.innerHTML = `
<div class="observer-detail-page" style="overflow-y:auto;height:calc(100vh - 56px);padding:16px">
<div class="observer-detail-page" style="padding:16px">
<div class="page-header" style="display:flex;align-items:center;gap:12px;margin-bottom:16px">
<a href="#/observers" class="btn-icon" title="Back to Observers" aria-label="Back"></a>
<h2 style="margin:0" id="obsTitle">Observer Detail</h2>
+2 -2
View File
@@ -10,7 +10,7 @@
*/
window.getParsedPath = function getParsedPath(p) {
if (p._parsedPath !== undefined) return p._parsedPath;
if (p._parsedPath !== undefined) return p._parsedPath || [];
var raw = p.path_json;
if (typeof raw !== 'string') {
p._parsedPath = Array.isArray(raw) ? raw : [];
@@ -32,7 +32,7 @@ window.clearParsedCache = function clearParsedCache(p) {
};
window.getParsedDecoded = function getParsedDecoded(p) {
if (p._parsedDecoded !== undefined) return p._parsedDecoded;
if (p._parsedDecoded !== undefined) return p._parsedDecoded || {};
var raw = p.decoded_json;
if (typeof raw !== 'string') {
p._parsedDecoded = (raw && typeof raw === 'object') ? raw : {};
+54 -19
View File
@@ -53,6 +53,7 @@
let _displayPackets = []; // filtered packets for current view
let _displayGrouped = false; // whether _displayPackets is in grouped mode
let _rowCounts = []; // per-entry DOM row counts (1 for flat, 1+children for expanded groups)
let _rowCountsDirty = false; // set when _rowCounts may be stale (e.g. WS added children) (#410)
let _cumulativeOffsetsCache = null; // cached cumulative offsets, invalidated on _rowCounts change
let _lastVisibleStart = -1; // last rendered start index (for dirty checking)
let _lastVisibleEnd = -1; // last rendered end index (for dirty checking)
@@ -357,7 +358,7 @@
if (pktTime && pktTime < cutoff) return false;
}
if (filters.type) { const types = filters.type.split(',').map(Number); if (!types.includes(p.payload_type)) return false; }
if (filters.observer) { const obsSet = new Set(filters.observer.split(',')); if (!obsSet.has(p.observer_id)) return false; }
if (filters.observer) { const obsSet = new Set(filters.observer.split(',')); if (!obsSet.has(p.observer_id) && !(p._children && p._children.some(c => obsSet.has(String(c.observer_id))))) return false; }
if (filters.hash && p.hash !== filters.hash) return false;
if (RegionFilter.getRegionParam()) {
const selectedRegions = RegionFilter.getRegionParam().split(',');
@@ -396,6 +397,9 @@
existing._children.unshift(p);
if (existing._children.length > 200) existing._children.length = 200;
sortGroupChildren(existing);
// Invalidate row counts — child count changed, so virtual scroll
// heights are stale until next renderTableRows() (#410)
_invalidateRowCounts();
}
} else {
// New group
@@ -442,6 +446,7 @@
clearTimeout(_wsRenderTimer);
_displayPackets = [];
_rowCounts = [];
_rowCountsDirty = false;
_cumulativeOffsetsCache = null;
_observerFilterSet = null;
_lastVisibleStart = -1;
@@ -488,6 +493,7 @@
if (regionParam) params.set('region', regionParam);
if (filters.hash) params.set('hash', filters.hash);
if (filters.node) params.set('node', filters.node);
if (filters.observer) params.set('observer', filters.observer);
params.set('groupByHash', 'true'); // always fetch grouped
const data = await api('/packets?' + params.toString());
@@ -541,19 +547,22 @@
// Ambiguous hops are already resolved by HopResolver client-side
// No need for per-observer server API calls
// Restore expanded group children
// Restore expanded group children (parallel fetch, Map lookup)
if (groupByHash && expandedHashes.size > 0) {
for (const hash of expandedHashes) {
const group = packets.find(p => p.hash === hash);
if (group) {
try {
const childData = await api(`/packets?hash=${hash}&limit=20`);
group._children = childData.packets || [];
sortGroupChildren(group);
} catch {}
} else {
// Group no longer in results — remove from expanded
const expandedArr = [...expandedHashes];
const results = await Promise.all(expandedArr.map(hash => {
const group = hashIndex.get(hash);
if (!group) return { hash, group: null, data: null };
return api(`/packets?hash=${hash}&limit=20`)
.then(data => ({ hash, group, data }))
.catch(() => ({ hash, group, data: null }));
}));
for (const { hash, group, data } of results) {
if (!group) {
expandedHashes.delete(hash);
} else if (data) {
group._children = data.packets || [];
sortGroupChildren(group);
}
}
}
@@ -1006,7 +1015,7 @@
}
else if (action === 'select-observation') {
const parentHash = row.dataset.parentHash;
const group = packets.find(p => p.hash === parentHash);
const group = hashIndex.get(parentHash);
const child = group?._children?.find(c => String(c.id) === String(value));
if (child) {
const parentData = group._fetchedData;
@@ -1099,8 +1108,8 @@
// Build HTML for a single flat (ungrouped) packet row
function buildFlatRowHtml(p) {
const decoded = getParsedDecoded(p);
const pathHops = getParsedPath(p);
const decoded = getParsedDecoded(p) || {};
const pathHops = getParsedPath(p) || [];
const region = p.observer_id ? (observerMap.get(p.observer_id)?.iata || '') : '';
const typeName = payloadTypeName(p.payload_type);
const typeClass = payloadTypeColor(p.payload_type);
@@ -1122,6 +1131,21 @@
</tr>`;
}
// Mark _rowCounts as stale so renderVisibleRows() recomputes them lazily.
// Called when expanded group children change outside renderTableRows() (#410).
function _invalidateRowCounts() {
_rowCountsDirty = true;
_cumulativeOffsetsCache = null;
}
// Recompute _rowCounts from _displayPackets if they've been invalidated.
function _refreshRowCountsIfDirty() {
if (!_rowCountsDirty || !_displayPackets.length) return;
_rowCounts = _displayPackets.map(function(p) { return _getRowCount(p); });
_cumulativeOffsetsCache = null;
_rowCountsDirty = false;
}
// Compute the number of DOM <tr> rows a single entry produces.
// Used by both row counting and renderVisibleRows to avoid divergence (#424).
function _getRowCount(p) {
@@ -1160,6 +1184,9 @@
const scrollContainer = document.getElementById('pktLeft');
if (!scrollContainer) return;
// Recompute row counts if they were invalidated (e.g. WS added children) (#410)
_refreshRowCountsIfDirty();
// Compute total DOM rows accounting for expanded groups
const offsets = _cumulativeRowOffsets();
const totalDomRows = offsets[offsets.length - 1];
@@ -1291,7 +1318,11 @@
}
if (filters.observer) {
const obsIds = new Set(filters.observer.split(','));
displayPackets = displayPackets.filter(p => obsIds.has(p.observer_id));
displayPackets = displayPackets.filter(p => {
if (obsIds.has(p.observer_id)) return true;
if (p._children) return p._children.some(c => obsIds.has(String(c.observer_id)));
return false;
});
}
// Packet Filter Language
@@ -1312,6 +1343,7 @@
if (!displayPackets.length) {
_displayPackets = [];
_rowCounts = [];
_rowCountsDirty = false;
_cumulativeOffsetsCache = null;
_observerFilterSet = null;
_lastVisibleStart = -1;
@@ -1331,6 +1363,7 @@
_displayGrouped = groupByHash;
_observerFilterSet = filters.observer ? new Set(filters.observer.split(',')) : null;
_rowCounts = displayPackets.map(p => _getRowCount(p));
_rowCountsDirty = false;
_cumulativeOffsetsCache = null;
attachVScrollListener();
@@ -1436,8 +1469,8 @@
const pkt = data.packet;
const breakdown = data.breakdown || {};
const ranges = breakdown.ranges || [];
const decoded = getParsedDecoded(pkt);
const pathHops = getParsedPath(pkt);
const decoded = getParsedDecoded(pkt) || {};
const pathHops = getParsedPath(pkt) || [];
// Resolve sender GPS — from packet directly, or from known node in DB
let senderLat = decoded.lat != null ? decoded.lat : (decoded.latitude || null);
@@ -1979,7 +2012,7 @@
const data = await api(`/packets/${hash}`);
const pkt = data.packet;
if (!pkt) return;
const group = packets.find(p => p.hash === hash);
const group = hashIndex.get(hash);
if (group && data.observations) {
group._children = data.observations.map(o => clearParsedCache({...pkt, ...o, _isObservation: true}));
group._fetchedData = data;
@@ -2039,6 +2072,8 @@
renderPath,
_getRowCount,
_cumulativeRowOffsets,
_invalidateRowCounts,
_refreshRowCountsIfDirty,
buildGroupRowHtml,
buildFlatRowHtml,
};
+1 -1
View File
@@ -5,7 +5,7 @@
let interval = null;
async function render(app) {
app.innerHTML = '<div id="perfWrapper" style="height:100%;overflow-y:auto;padding:16px 24px;"><h2>⚡ Performance Dashboard</h2><div id="perfContent">Loading...</div></div>';
app.innerHTML = '<div id="perfWrapper" style="padding:16px 24px;"><h2>⚡ Performance Dashboard</h2><div id="perfContent">Loading...</div></div>';
await refresh();
}
+12 -5
View File
@@ -181,7 +181,12 @@ a:focus-visible, button:focus-visible, input:focus-visible, select:focus-visible
}
/* === Layout === */
#app { height: calc(100vh - 52px); height: calc(100dvh - 52px); overflow: hidden; }
/* Default: body-scroll mode content pushes beyond viewport, iOS status-bar
tap-to-scroll works because <body> is the scroll container. Pages that need
a fixed-height container (maps, virtual-scroll, split-panels) add
.app-fixed via the router so their children can use height:100%. */
#app { min-height: calc(100vh - 52px); min-height: calc(100dvh - 52px); }
#app.app-fixed { height: calc(100vh - 52px); height: calc(100dvh - 52px); min-height: 0; overflow: hidden; }
.split-layout {
display: flex; height: 100%; overflow: hidden;
@@ -674,7 +679,7 @@ button.ch-item.selected { background: var(--selected-bg); }
.advert-info { font-size: 12px; line-height: 1.5; }
/* === Traces Page === */
.traces-page { padding: 16px; max-width: var(--trace-max-width, 95vw); margin: 0 auto; overflow-y: auto; height: 100%; }
.traces-page { padding: 16px; max-width: var(--trace-max-width, 95vw); margin: 0 auto; }
.trace-search {
display: flex; gap: 8px; margin-bottom: 20px;
}
@@ -746,7 +751,7 @@ button.ch-item.selected { background: var(--selected-bg); }
::-webkit-scrollbar-thumb:hover { background: var(--text-muted); }
/* === Observers Page === */
.observers-page { padding: 20px; max-width: 1200px; margin: 0 auto; overflow-y: auto; height: calc(100vh - 56px); }
.observers-page { padding: 20px; max-width: 1200px; margin: 0 auto; }
.obs-summary { display: flex; gap: 20px; margin-bottom: 16px; flex-wrap: wrap; }
.obs-stat { display: flex; align-items: center; gap: 6px; font-size: 14px; color: var(--text-muted); }
.health-dot { width: 10px; height: 10px; border-radius: 50%; display: inline-block; flex-shrink: 0; }
@@ -947,7 +952,9 @@ button.ch-item.selected { background: var(--selected-bg); }
.filter-bar { flex-direction: row; flex-wrap: wrap; gap: 4px; }
.filter-toggle-btn { display: inline-flex !important; }
.filter-bar > *:not(.filter-toggle-btn):not(.col-toggle-wrap) { display: none; }
.filter-bar.filters-expanded > * { display: inline-flex; }
/* Must match :not() specificity of the hide rule above, otherwise .filters-expanded loses
the specificity battle and filter children stay hidden (see issue #534). */
.filter-bar.filters-expanded > *:not(.filter-toggle-btn):not(.col-toggle-wrap) { display: inline-flex; }
.filter-bar.filters-expanded > .col-toggle-wrap { display: inline-block; }
.filter-bar.filters-expanded input { width: 100%; }
.filter-bar.filters-expanded select { width: 100%; }
@@ -1136,7 +1143,7 @@ button.ch-item.ch-item-encrypted .ch-badge { filter: grayscale(0.6); }
.node-activity-time { color: var(--text-muted); white-space: nowrap; min-width: 70px; font-size: 12px; }
/* Analytics page */
.analytics-page { padding: 16px 24px; max-width: 1600px; margin: 0 auto; overflow-y: auto; height: 100%; }
.analytics-page { padding: 16px 24px; max-width: 1600px; margin: 0 auto; }
.analytics-header { margin-bottom: 20px; }
.analytics-header h2 { margin: 0 0 4px; }
.analytics-card { background: var(--card-bg); border: 1px solid var(--border); border-radius: 8px; padding: 16px; margin-bottom: 16px; }
+41
View File
@@ -1573,6 +1573,47 @@ async function run() {
// ─── End affinity debug tests ─────────────────────────────────────────────
// ─── Mobile filter dropdown tests (#534) ──────────────────────────────────
await test('Mobile: filter toggle expands filter bar on packets page (#534)', async () => {
// Use a mobile viewport
await page.setViewportSize({ width: 480, height: 800 });
await page.goto(`${BASE}/#/packets`);
await page.waitForTimeout(500);
const filterBar = await page.$('.filter-bar');
assert(filterBar, 'Filter bar should exist on packets page');
// Before clicking toggle, filter inputs should be hidden
const toggleBtn = await page.$('.filter-toggle-btn');
assert(toggleBtn, 'Filter toggle button should exist on mobile');
await toggleBtn.click();
await page.waitForTimeout(300);
// After clicking, .filters-expanded should be on the filter bar
const expanded = await filterBar.evaluate(el => el.classList.contains('filters-expanded'));
assert(expanded, 'Filter bar should have filters-expanded class after toggle');
// Filter inputs should now be visible
const filterInput = await page.$('.filter-bar input');
if (filterInput) {
const display = await filterInput.evaluate(el => getComputedStyle(el).display);
assert(display !== 'none', `Filter input should be visible when expanded, got display: ${display}`);
}
const filterSelect = await page.$('.filter-bar select');
if (filterSelect) {
const display = await filterSelect.evaluate(el => getComputedStyle(el).display);
assert(display !== 'none', `Filter select should be visible when expanded, got display: ${display}`);
}
// Reset viewport
await page.setViewportSize({ width: 1280, height: 720 });
});
// ─── End mobile filter tests ──────────────────────────────────────────────
// Extract frontend coverage if instrumented server is running
try {
const coverage = await page.evaluate(() => window.__coverage__);
+117
View File
@@ -998,6 +998,56 @@ console.log('\n=== live.js: pruneStaleNodes ===');
assert.ok(markers['apiNode'], 'API stale node should NOT be removed');
assert.ok(data['apiNode'], 'API stale node data should NOT be removed');
});
test('pruneStaleNodes cleans up nodeActivity for removed nodes', () => {
const { ctx } = makeLiveSandbox();
const prune = ctx.window._livePruneStaleNodes;
const markers = ctx.window._liveNodeMarkers();
const data = ctx.window._liveNodeData();
const activity = ctx.window._liveNodeActivity();
// WS-only stale node
markers['staleNode'] = { _glowMarker: null };
data['staleNode'] = { public_key: 'staleNode', role: 'companion', _liveSeen: Date.now() - 48 * 3600000 };
activity['staleNode'] = 5;
// Active node
markers['activeNode'] = { setStyle: function() {}, _glowMarker: null };
data['activeNode'] = { public_key: 'activeNode', role: 'companion', _liveSeen: Date.now() };
activity['activeNode'] = 3;
prune();
assert.ok(!markers['staleNode'], 'stale node marker removed');
assert.ok(!data['staleNode'], 'stale node data removed');
assert.ok(!activity['staleNode'], 'stale node activity removed');
assert.ok(markers['activeNode'], 'active node marker preserved');
assert.ok(data['activeNode'], 'active node data preserved');
assert.strictEqual(activity['activeNode'], 3, 'active node activity preserved');
});
test('pruneStaleNodes removes orphaned nodeActivity entries', () => {
const { ctx } = makeLiveSandbox();
const prune = ctx.window._livePruneStaleNodes;
const markers = ctx.window._liveNodeMarkers();
const data = ctx.window._liveNodeData();
const activity = ctx.window._liveNodeActivity();
// Add an active node
markers['existingNode'] = { setStyle: function() {}, _glowMarker: null };
data['existingNode'] = { public_key: 'existingNode', role: 'companion', _liveSeen: Date.now() };
activity['existingNode'] = 2;
// Add orphaned activity (no corresponding nodeData)
activity['ghostNode'] = 10;
prune();
assert.ok(markers['existingNode'], 'existing node preserved');
assert.ok(data['existingNode'], 'existing node data preserved');
assert.strictEqual(activity['existingNode'], 2, 'existing node activity preserved');
assert.ok(!activity['ghostNode'], 'orphaned activity entry removed');
});
}
// ===== live.js: vcrFormatTime respects UTC/local setting =====
@@ -2695,6 +2745,63 @@ console.log('\n=== packets.js: savedTimeWindowMin defaults ===');
'buildGroupRowHtml should use hoisted _observerFilterSet');
});
test('observer filter in grouped mode includes packet when child matches (#537)', () => {
// The display filter should keep a grouped packet whose primary observer_id
// does NOT match, but one of its _children does.
const obsIds = new Set(['OBS_B']);
const packets = [
{ observer_id: 'OBS_A', _children: [{ observer_id: 'OBS_A' }, { observer_id: 'OBS_B' }] },
{ observer_id: 'OBS_C', _children: [{ observer_id: 'OBS_C' }] },
];
const result = packets.filter(p => {
if (obsIds.has(p.observer_id)) return true;
if (p._children) return p._children.some(c => obsIds.has(String(c.observer_id)));
return false;
});
assert.strictEqual(result.length, 1, 'should keep packet with matching child observer');
assert.strictEqual(result[0].observer_id, 'OBS_A');
});
test('observer filter in grouped mode hides packet with no matching observations (#537)', () => {
const obsIds = new Set(['OBS_X']);
const packets = [
{ observer_id: 'OBS_A', _children: [{ observer_id: 'OBS_A' }, { observer_id: 'OBS_B' }] },
];
const result = packets.filter(p => {
if (obsIds.has(p.observer_id)) return true;
if (p._children) return p._children.some(c => obsIds.has(String(c.observer_id)));
return false;
});
assert.strictEqual(result.length, 0, 'should hide packet with no matching observers');
});
test('WS observer filter checks children for grouped packets (#537)', () => {
const filters = { observer: 'OBS_B' };
const obsSet = new Set(filters.observer.split(','));
const p = { observer_id: 'OBS_A', _children: [{ observer_id: 'OBS_B' }] };
const passes = obsSet.has(p.observer_id) || (p._children && p._children.some(c => obsSet.has(String(c.observer_id))));
assert.ok(passes, 'WS filter should pass grouped packet when child matches');
const p2 = { observer_id: 'OBS_C', _children: [{ observer_id: 'OBS_D' }] };
const passes2 = obsSet.has(p2.observer_id) || (p2._children && p2._children.some(c => obsSet.has(String(c.observer_id))));
assert.ok(!passes2, 'WS filter should reject grouped packet with no matching observers');
});
test('packets.js display filter checks _children for observer match (#537)', () => {
// Verify the actual source code has the children check
assert.ok(
packetsSource.includes('p._children) return p._children.some(c => obsIds.has(String(c.observer_id))'),
'display filter should check _children for observer match'
);
});
test('packets.js WS filter checks _children for observer match (#537)', () => {
assert.ok(
packetsSource.includes('p._children && p._children.some(c => obsSet.has(String(c.observer_id)))'),
'WS filter should check _children for observer match'
);
});
test('buildFlatRowHtml has null-safe decoded_json', () => {
const flatBuilderMatch = packetsSource.match(/function buildFlatRowHtml[\s\S]*?(?=\n function )/);
assert.ok(flatBuilderMatch, 'buildFlatRowHtml should exist');
@@ -4126,7 +4233,17 @@ console.log('\n=== app.js: routeTypeName/payloadTypeName edge cases ===');
assertJsonEqual(getParsedPath(p), []);
});
test('getParsedPath: cached null _parsedPath returns empty array (#538)', () => {
const p = { path_json: '["a"]', _parsedPath: null };
assertJsonEqual(getParsedPath(p), []);
});
// --- getParsedDecoded ---
test('getParsedDecoded: cached null _parsedDecoded returns empty object (#538)', () => {
const p = { decoded_json: '{"x":1}', _parsedDecoded: null };
assertJsonEqual(getParsedDecoded(p), {});
});
test('getParsedDecoded: valid JSON object', () => {
const p = { decoded_json: '{"type":"GRP_TXT","text":"hello"}' };
const result = getParsedDecoded(p);
+99
View File
@@ -0,0 +1,99 @@
/**
* Unit tests for HopResolver affinity-aware hop resolution.
*/
'use strict';
const fs = require('fs');
const vm = require('vm');
// Load hop-resolver.js in a sandboxed context
const code = fs.readFileSync(__dirname + '/public/hop-resolver.js', 'utf8');
const sandbox = { window: {}, console, Math, Object, Array, Number, Date, Map, Set, parseInt, parseFloat, encodeURIComponent };
vm.createContext(sandbox);
vm.runInContext(code, sandbox);
const HopResolver = sandbox.window.HopResolver;
let passed = 0;
let failed = 0;
function assert(condition, msg) {
if (condition) { passed++; console.log(' ✓ ' + msg); }
else { failed++; console.error(' ✗ ' + msg); }
}
// ── Test nodes ──
// Two nodes share the same 1-byte prefix "ab"
const nodeA = { public_key: 'ab1111', name: 'NodeA', lat: 37.0, lon: -122.0 };
const nodeB = { public_key: 'ab2222', name: 'NodeB', lat: 38.0, lon: -123.0 };
const nodeC = { public_key: 'cd3333', name: 'NodeC', lat: 37.5, lon: -122.5 };
console.log('\n=== HopResolver Affinity Tests ===\n');
// Test 1: Affinity prefers neighbor candidate over geo-closest
console.log('Test 1: Affinity prefers neighbor over geo-closest');
HopResolver.init([nodeA, nodeB, nodeC]);
HopResolver.setAffinity({
edges: [
{ source: 'cd3333', target: 'ab2222', score: 0.8 }
// NodeC is a neighbor of NodeB but NOT NodeA
]
});
// Resolve hop "ab" after NodeC was resolved — should pick NodeB (neighbor) not NodeA (geo-closer)
// Origin at NodeC's position so forward pass runs with NodeC as anchor
const result1 = HopResolver.resolve(['cd33', 'ab'], nodeC.lat, nodeC.lon, null, null, null);
assert(result1['ab'].name === 'NodeB', 'Should pick NodeB (affinity neighbor of NodeC) — got: ' + result1['ab'].name);
// Test 2: Without affinity, falls back to geo-closest
console.log('\nTest 2: Cold start (no affinity) falls back to geo-closest');
HopResolver.init([nodeA, nodeB, nodeC]);
HopResolver.setAffinity({}); // No edges
// With anchor at NodeC's position, NodeA is closer to NodeC than NodeB
const result2 = HopResolver.resolve(['cd33', 'ab'], nodeC.lat, nodeC.lon, null, null, null);
// NodeA (37, -122) is closer to NodeC (37.5, -122.5) than NodeB (38, -123)
assert(result2['ab'].name === 'NodeA', 'Should pick NodeA (geo-closest) — got: ' + result2['ab'].name);
// Test 3: setAffinity with null/undefined doesn't crash
console.log('\nTest 3: setAffinity with null/undefined is safe');
HopResolver.setAffinity(null);
HopResolver.setAffinity(undefined);
HopResolver.setAffinity({});
assert(true, 'No crash on null/undefined/empty affinity');
// Test 4: getAffinity returns correct scores
console.log('\nTest 4: getAffinity returns correct scores');
HopResolver.setAffinity({
edges: [
{ source: 'aaa', target: 'bbb', score: 0.95 },
{ source: 'ccc', target: 'ddd', weight: 5 }
]
});
assert(HopResolver.getAffinity('aaa', 'bbb') === 0.95, 'aaa→bbb = 0.95');
assert(HopResolver.getAffinity('bbb', 'aaa') === 0.95, 'bbb→aaa = 0.95 (bidirectional)');
assert(HopResolver.getAffinity('ccc', 'ddd') === 5, 'ccc→ddd = 5 (weight fallback)');
assert(HopResolver.getAffinity('aaa', 'zzz') === 0, 'unknown pair = 0');
assert(HopResolver.getAffinity(null, 'bbb') === 0, 'null pubkey = 0');
// Test 5: Affinity with multiple neighbors — highest score wins
console.log('\nTest 5: Highest affinity score wins among neighbors');
HopResolver.init([nodeA, nodeB, nodeC]);
HopResolver.setAffinity({
edges: [
{ source: 'cd3333', target: 'ab1111', score: 0.3 },
{ source: 'cd3333', target: 'ab2222', score: 0.9 }
]
});
const result5 = HopResolver.resolve(['cd33', 'ab'], nodeC.lat, nodeC.lon, null, null, null);
assert(result5['ab'].name === 'NodeB', 'Should pick NodeB (highest affinity 0.9) — got: ' + result5['ab'].name);
// Test 6: Unambiguous hops are not affected by affinity
console.log('\nTest 6: Unambiguous hops unaffected by affinity');
const nodeD = { public_key: 'ee4444', name: 'NodeD', lat: 36.0, lon: -121.0 };
HopResolver.init([nodeA, nodeB, nodeC, nodeD]);
HopResolver.setAffinity({ edges: [] });
const result6 = HopResolver.resolve(['ee44'], null, null, null, null, null);
assert(result6['ee44'].name === 'NodeD', 'Unique prefix resolves directly — got: ' + result6['ee44'].name);
assert(!result6['ee44'].ambiguous, 'Should not be marked ambiguous');
console.log('\n' + (passed + failed) + ' tests, ' + passed + ' passed, ' + failed + ' failed\n');
process.exit(failed > 0 ? 1 : 0);
+53
View File
@@ -272,6 +272,48 @@ console.log('\n=== live.js: expandToBufferEntries ===');
});
}
// ===== expandToBufferEntriesAsync (chunked, non-blocking) =====
console.log('\n=== live.js: expandToBufferEntriesAsync ===');
{
// Build a sandbox with packet-helpers loaded so expandToBufferEntries can call dbPacketToLive
const ctx = makeSandbox();
addLiveGlobals(ctx);
loadInCtx(ctx, 'public/roles.js');
loadInCtx(ctx, 'public/packet-helpers.js');
try { loadInCtx(ctx, 'public/live.js'); } catch (e) {
for (const k of Object.keys(ctx.window)) ctx[k] = ctx.window[k];
}
const expandSync = ctx.window._liveExpandToBufferEntries;
const expandAsync = ctx.window._liveExpandToBufferEntriesAsync;
assert.ok(expandAsync, '_liveExpandToBufferEntriesAsync must be exposed');
const pkts = [];
for (let i = 0; i < 500; i++) {
pkts.push({
id: i, hash: 'h' + i, timestamp: new Date(1700000000000 + i * 1000).toISOString(),
decoded_json: '{"type":"GRP_TXT"}', path_json: '[]',
observations: [
{ timestamp: new Date(1700000000000 + i * 1000 + 100).toISOString(), snr: 5, observer_name: 'O1' },
{ timestamp: new Date(1700000000000 + i * 1000 + 200).toISOString(), snr: 8, observer_name: 'O2' },
],
});
}
test('sync expand handles 500 packets (1000 entries) correctly', () => {
const result = expandSync(pkts);
assert.strictEqual(result.length, 1000, '500 packets * 2 observations = 1000 entries');
assert.strictEqual(result[0].pkt.hash, 'h0');
assert.strictEqual(result[999].pkt.hash, 'h499');
});
test('VCR_CHUNK_SIZE is defined and async function yields via setTimeout', () => {
const src = fs.readFileSync(__dirname + '/public/live.js', 'utf8');
assert.ok(src.includes('VCR_CHUNK_SIZE'), 'VCR_CHUNK_SIZE constant must exist');
assert.ok(src.includes('expandToBufferEntriesAsync'), 'async version must exist');
assert.ok(src.includes('setTimeout(processChunk, 0)'), 'must yield via setTimeout between chunks');
});
}
// ===== SEG_MAP (7-segment display) =====
console.log('\n=== live.js: SEG_MAP ===');
{
@@ -839,6 +881,17 @@ console.log('\n=== live.js: source-level safety checks ===');
assert.ok(src.includes('const existingIds = new Set(VCR.buffer.map(b => b.pkt.id)'),
'vcrRewind should dedup by packet ID');
});
test('feed items include transport badge', () => {
const count = (src.match(/transportBadge\(pkt\.route_type\)/g) || []).length;
assert.ok(count >= 3,
`feed rendering should call transportBadge(pkt.route_type) in at least 3 places (found ${count})`);
});
test('node detail recent packets include transport badge', () => {
assert.ok(src.includes('transportBadge(p.route_type)'),
'node detail recent packets should call transportBadge(p.route_type)');
});
}
// ===== SUMMARY =====
+27
View File
@@ -757,6 +757,33 @@ console.log('\n=== packets.js: page registration ===');
});
}
console.log('\n=== packets.js: _invalidateRowCounts / _refreshRowCountsIfDirty (#410) ===');
{
const ctx = loadPacketsSandbox();
const api = ctx._packetsTestAPI;
test('_invalidateRowCounts and _refreshRowCountsIfDirty are exported', () => {
assert(typeof api._invalidateRowCounts === 'function');
assert(typeof api._refreshRowCountsIfDirty === 'function');
});
test('_invalidateRowCounts does not throw', () => {
api._invalidateRowCounts();
});
test('_refreshRowCountsIfDirty does not throw when no display packets', () => {
api._invalidateRowCounts();
api._refreshRowCountsIfDirty();
});
test('_cumulativeRowOffsets returns valid offsets after invalidation cycle', () => {
// Even with no display packets, should return valid array
const offsets = api._cumulativeRowOffsets();
assert(Array.isArray(offsets));
assert(offsets[0] === 0);
});
}
// ===== SUMMARY =====
console.log(`\n${'='.repeat(40)}`);
console.log(`packets.js tests: ${passed} passed, ${failed} failed`);