Files
meshcore-analyzer/tests
Alex BandClaude Opus 5.5 deb55dc82e feat(scope-audit): transport view on the Scope Audit page (#2142, page) (#2169)
> **Requires #2167 to be merged first.** This page calls the transport
view API that #2167 adds (`/api/scope-audit?mode=transport`), so it
cannot be split from it; until #2167 merges this PR also shows its
commits.

## Summary

Page half of #2142. **Stacked on #2167** (the API); review that first.
#2171 adds the observer region filter on top. Partial fix for #2142
until all three are in.

Adds a **Declared / Transport** toggle to `#/scope-audit`. The transport
view calls `?mode=transport` and lists every repeater seen forwarding in
the window:

- **Seen carrying:** one chip per region scope with its packet count
(first/last seen in the tooltip);
- **Declared:** the repeater's declared regions coloured by whether they
were seen carried (the declared view's own chips), or *not asked* when
no declared-regions answer exists. Never asked is worded differently
from declaring nothing;
- **Other traffic:** unscoped, unmatched and ambiguous-hop counts.

A **Region** box (suggestions from the scopes seen) adds a *Carries X*
column, lists the repeaters not yet carrying it first, and summarises
carriers against active non-carriers: the rollout report the issue asks
for. Window, view and region are deep-linked
(`#/scope-audit?window=7d&mode=transport&scope=be`) and survive a
reload. The page says plainly that *transported* means seen carrying,
not configured, and that over a short window a quiet region looks the
same as a missing one. The intro text changes with the view.

## Screenshot

The transport view with scope `be` typed in: every repeater seen
forwarding in the last 24 h, whether it carries `be` (non-carriers
first), the scopes it was seen carrying with counts, and its declared
regions or *not asked*. Taken against `test-fixtures/e2e-fixture.db`
with fictional `be` / `be-van` / `nl` scopes.

![Scope Audit transport
view](https://raw.githubusercontent.com/A13xB0/CoreScope/pr-screenshots/2026-10/scope-audit-transport-view.png)

## Browser validation (staging, real data)

Headless Chromium against a staging copy of a production database with
its live feed:

- `#/scope-audit?window=24h`: the declared view, 16 rows, as before.
- Clicking **Transport**: 385 rows; the hash gains `mode=transport`.
- Typing a scope as `#XX-YY` in capitals: normalised to `xx-yy`; the
hash gains it (now `scope=xx-yy`); summary "385 repeaters seen
forwarding in the last 24h · 16 with a declared-regions answer · 51
carried xx-yy · 334 active but not seen carrying it"; the first rows are
the non-carriers.
- Reloading that hash restores view and region. Switching back to
Declared hides the Region box and restores the declared intro. No page
errors.
- Checked at 1280 px and 390 px wide.

Two layout bugs found this way were fixed before opening: the region box
inherited `width: 100%` from `.nodes-search`, and `display: flex`
overrode the `hidden` attribute on the region bar.

## Tests

`tests/unit/test-frontend-helpers.js` gains 10 cases on rendered markup
and state, using the page's exported internals like the existing
scope-audit tests: hash and API paths (and that a region never leaks
into the declared view's link), region normalisation (`#`, capitals,
empty), *not asked* vs declared chips (with the `*` wildcard),
carried-scope chips with counts and the filtered region marked, the
carries column only with a region, HTML escaping of names, the summary
counts, region suggestions, and the per-view intro. 768 passing in that
file.

Net-new UI surface, so per TDD.md the tests land in this PR rather than
as a preceding red commit. Frontend suites on this branch: 193 of 194;
the one failure, `test-issue-1956-release-routing.js`, needs CI's
`GITHUB_REF_NAME` and fails identically on `master` locally (#2149 and
#2160 address that).

## Review notes

- Colours only through existing CSS variables (no hex literals); the
filtered region is marked with the link colour, not a new status colour.
- The window note's wording is in the same spirit as the declared view's
`windowHonestyNote`.
- No API calls per row: one request per view/window/region, cached
client-side for 30 s like the declared view.

## SonarLint

SonarQube Community rules (the SonarLint engine) were run on the files
this PR touches, and only issues on lines it adds are listed.

- **Fixed** in `df53e770`:
- one `modeQuery` helper for the page hash and the API path (S3358; the
two copies were identical);
  - the carries-scope cell built before the row (S3358);
- `localeCompare` for the region suggestions (S2871), `for…of` and
`.dataset` (S4138, S7761), optional chaining (S6582);
  - object spread in the helper tests (S6661).
- Optional chaining and spread are already used throughout `public/`. No
behaviour change.

## Checked against the repo rules

- Rebased onto `master` (`673442cc`). Each red commit was re-run and
still fails on its assertions; each green commit passes (TDD.md).
- `make test`: all 16 modules pass. Frontend suites: all pass except
`test-issue-1956-release-routing.js`, which needs CI's `GITHUB_REF_NAME`
and fails the same way on `master` locally (#2160 addresses it).
- No private data: test fixtures and examples use the codes the existing
tests use (SJC/SFO/LAX, `#be`-style scopes).
- No new `map[string]interface{}`, no hard-coded colours. `git add` was
explicit, and there is one logical change per commit.

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

https://claude.ai/code/session_01KaR4xu8gVykQBkH8yG8m6H

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 21:22:37 +02:00
..