# Campus review workflow

The campus has not been field-tested. `data/reviewed-campus.json` intentionally starts empty. No new entrance, shortcut, official name, architectural geometry or building photo is represented as verified by this change.

## Student features

Select a place and choose **Save place**. Saved places appear in the guide's **Saved places** section and remain on the same browser/site until removed or site data is cleared.

**Report incorrect information** includes the selected place and ID. Reports can be saved, downloaded and deleted locally. Storage failure leaves the report downloadable. Reports are never sent automatically. Moving hosts does not copy them; export any reports you still need before leaving the old site.

## Review and publish

1. Open `editor.html`, also linked as **Review campus data** in the map footer. This is a local drafting tool, not an authenticated shared admin panel. It has no server write capability.
2. Import exported walkpath GeoJSON, a local report JSON, a legacy report CSV, or a workspace backup. Existing locally saved reports appear automatically. Mark reports pending, investigating, resolved or rejected. Report status does not change routing.
3. Select a place to correct its name, description or coordinates. Add actual building photos with credit and permission/source. Files must be JPEG, PNG or WebP, at most 2 MB each and five per place. Photos are embedded in the exported bundle. Large photo collections may exceed browser storage; download the workspace when warned.
4. For an entrance, provide field-checked coordinates and select the physically connected graph node. The reviewed connection must be within 15 m. Enter evidence and the date checked. Do not substitute a building centre or the nearest road across a wall.
5. Review each continuous recorded segment visually. Select its actual start and end connections, inspect gates/access, choose permitted direction, and record evidence. The validator rejects endpoint gaps over 15 m, internal gaps over 100 m and out-of-campus coordinates. Crossings do not create junctions automatically. Rejected/pending segments never enter the public routing overlay.
6. Download a workspace backup. **Export reviewed changes** produces the public overlay only, excluding reports and pending/rejected candidates. Exported review evidence and photos become public when deployed, so keep personal details out of them.
7. In the source folder run `node tools/apply-reviewed-data.mjs <downloaded-bundle.json>` for validation. Review the counts and evidence; add `--apply` to install it locally. This replaces the review overlay and backs up the previous version as `data/reviewed-campus.backup.json`. Keep existing published changes in your exported bundle.
8. Run `npm test` and `npm run build`. Then publish `../campus-explorer-site/out` with `npm run deploy`. Building or approving in the editor does not deploy. To roll back, validate and apply the backup file, then deploy again.

After replacing the base models or imported place lists, regenerate `data/place-catalog.json` using `node tools/build-place-catalog.mjs`. Place IDs currently follow model/import IDs. Review their identity before reusing an old bundle after a source import changes ordering.

## First field test

Choose two nearby familiar buildings and walk the displayed route. Note which entrance is actually used, missing footpaths, closed gates, road crossings and inaccurate names. Record only continuous public shortcut segments, and export each recording. Photograph the entrance and exterior if permitted; retain a source/permission note. Compare the result in the editor before approving anything. GPS recordings alone do not prove public access.

Detailed architecture remains a separate modelling task requiring trustworthy exterior references and dimensions. Current building geometry remains approximate.