Commit Graph

6 Commits

Author SHA1 Message Date
cproudlock
23dc9fa379 ADR-013 Phase 4: relocate 4 self-contained plugin frontends
Relocate applications, geenforce, knowledgebase, and machines - each owns only
its own views dir, so a clean move to plugins/<name>/frontend/ (views/ +
routes.js, core imports rewritten to @/). geenforce's entryForm.js helper + its
vitest spec move with it (ManifestEditor imports it as a sibling).

Machinery fixes this batch surfaced:
- routes.gen.js codegen uses namespace imports (import * as p_x). A route file
  without a `toplevel` export is undefined on the namespace instead of a strict-
  ESM missing-binding build error.
- vitest gains a `pretest` stage so plugin-frontend specs (now under
  plugins/<name>/frontend/) run from their staged copy in src/.plugins-staged/.

Verified live: GE-Enforce (the most complex, uses the entryForm sibling helper)
renders fully from its staged frontend. Build + 58 vitest + naming green.
2026-07-18 23:51:23 -04:00
cproudlock
cd353b6432 Review safe-polish: docs accuracy, dead imports, no-emoji, geenforce robustness
Some checks failed
CI / backend (push) Has been cancelled
CI / naming (push) Has been cancelled
CI / frontend (push) Has been cancelled
From the full multi-agent review (0 high, 7 medium, 17 low findings). Applies
the mechanical, low-risk items; design/policy findings left for a decision.

Docs accuracy: CLAUDE.md contract 0.10.0 -> 0.11.0 and both stale Alembic head
citations -> 7d24_customfield_searchable / 31 migrations; Dockerfile bundled-
plugin comment fixed (drop nonexistent "equipment", add machines +
measuringtools, count eleven).

Style/naming (LOCKED rules): remove a CSS-escaped pushpin emoji before location
search results (no-emoji policy); rename ManifestEditor shareRoot -> shareroot
(variable mirrors the API field verbatim).

Dead code: remove confirmed-unused imports across ~20 modules (require_role/
require_permission scaffold residue, stray db/Vendor/Model/current_user/Optional/
error_response); drop unused build_scope import + a stale GEENFORCE_API_KEY
docstring clause in geenforce. Migration files left untouched.

Correctness: geenforce ingest robustness - record_enforcement_report now 400s
on a non-dict counts / non-list results instead of 500; _apply_app_link ignores
a non-numeric appid per its docstring instead of 500. Regression tests added.

Backend query.get sweep finished: auth.py refresh -> db.session.get (last one).

910 backend tests pass; pyflakes clean; naming green; frontend build green.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-13 08:02:43 -04:00
cproudlock
7dfbe7bf8a Add the get_permissions plugin hook (contract 0.10.0)
All checks were successful
CI / backend (push) Successful in 1m20s
CI / naming (push) Successful in 2s
CI / frontend (push) Successful in 8s
Plugins declare their own RBAC permissions instead of core accumulating
them: 36 permissions moved out of the core catalog into the 9 owning
plugins (core keeps the 19 its own blueprints enforce). The catalog is
resolved dynamically (core + enabled plugins) and feeds the roles grid,
the token scope picker and ceiling, and flask seed permissions;
installing or enabling a plugin seeds its permissions automatically. A
disabled plugin drops out of the assignable catalog while existing role
links keep working. New plugins - bundled or external - now bring their
permissions with zero core edits.

781 tests pass; live-verified with a machines.edit-scoped token.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-12 09:29:55 -04:00
cproudlock
b130ef43f3 Add the dualpath-as-single-machine site toggle
All checks were successful
CI / backend (push) Successful in 1m13s
CI / naming (push) Successful in 1s
CI / frontend (push) Successful in 7s
Most facilities consider a Dualpath pair one physical dual-bay machine.
New site setting dualpath_single_machine (default on): the machines
list, dashboard counts, machines-by-type report, and floor map collapse
each pair to its primary bay (lower assetnumber), with combined
2007 / 2008 labels; pagination totals stay honest. Detail pages remain
per-bay and always show a dual-bay sibling banner linking the partner.
Pair resolution lives in core services and joins the plugin contract
surface (0.8.0 -> 0.9.0).

On the WJ dataset: 31 pairs collapse, machine counts 262 -> 231, map
470 assets. Toggle verified live in both states, left on.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-12 06:23:39 -04:00
cproudlock
46e50c07ff Add the API import surface for legacy migrations (contract 0.8.0)
All checks were successful
CI / backend (push) Successful in 1m4s
CI / naming (push) Successful in 1s
CI / frontend (push) Successful in 7s
Goal: an LLM or script can migrate an entire legacy database using only
the HTTP API - original history preserved, safely re-runnable.

- X-Import-Mode header (admin only): create/update endpoints across 15
  timestamped entity types accept original createddate/modifieddate;
  helper exposed via shopdb.api (contract 0.7.0 -> 0.8.0).
- Exact-match natural-key lookup filters on 13 list endpoints for the
  lookup-then-upsert recipe.
- Selfhosted USB checkout/checkin accept backdated event times in
  import mode.
- docs/IMPORT-API.md: operator manual grounded in the real legacy
  schema - order of operations, full table-by-table mapping including
  the machines fan-out, idempotent Python importer with dry-run, parity
  checks, and decided dispositions for unmigrated tables (DNC config
  stays live-fed via the collector; supportteams/appowners map to the
  upcoming supportteams model).

635 tests pass; naming green; frontend untouched.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-11 20:10:46 -04:00
cproudlock
48d3160bc5 Rename the equipment domain to machines; retype the models catalog (ADR-011)
All checks were successful
CI / backend (push) Successful in 23s
CI / naming (push) Successful in 2s
CI / frontend (push) Successful in 8s
The equipment plugin is now the machines plugin, ending the UI-vs-code
vocabulary split while the contract is pre-1.0 and nothing external
depends on the old names.

- plugins/equipment -> plugins/machines: manifest, class, /api/machines,
  machines.* permissions, registry key (with an auto-migrating load shim
  for existing installs).
- Tables: equipment -> machines (equipmentid -> machineid) and
  equipmenttypes -> machinetypes, renamed in the plugin's own migration
  chain (machines0002rename), idempotent for both upgrading and fresh
  installs.
- The legacy core machinetypes lookup actually types the vendor MODELS
  catalog, so it is renamed losslessly to modeltypes
  (models.modeltypeid, /api/modeltypes, Model Types settings page)
  rather than collapsed, freeing the machinetypes name. Core migration
  7d17_machines_rename also flips data in place: assettypes row
  equipment -> machine, auditlog entitytype, identifier_/search_
  settings keys, permission rows, and renames alembic_version_equipment.
- Frontend: machinesApi/modeltypesApi, item.machine response shape,
  assettype value compares 'equipment' -> 'machine' (map, search,
  custom fields, relationships), routes machines.js with plugin gating
  retagged, /print/machine-badge, Machine Types (subtypes) and Model
  Types (catalog) settings pages, machines-by-type report id.
- Docs swept; ADRs left as history per the authoring rule.

Upgrade: flask db upgrade then flask plugin upgrade-all.

Verified: dev DB flipped live (262 machines, 35 modeltypes, 95 models
retyped, zero equipment tables remain); fresh scratch-MySQL install
produces the new names; 341 tests green; naming/style green; frontend
builds; live E2E on machines list/detail, PC relationships, map,
reports, and both settings pages.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-11 15:17:42 -04:00