@wesports/cms-client

Pure-TypeScript reader for the Thevea CMS, consumed by WeSports storefronts in frontend-suite/. No React, no Next.js — just typed fetch wrappers.

Install

npm install github:Geborek/cms-client

Usage

import { getSingleton, listEntries, type CmsConfig } from '@wesports/cms-client';

const config: CmsConfig = {
  apiUrl: process.env.CMS_API_URL!,
  locale: 'sv-SE',
  token: process.env.CMS_API_TOKEN,
  workspace: process.env.CMS_WORKSPACE,
};

const home = await getSingleton('home-page', config);
const banners = await listEntries('banner', config, { limit: 10 });

Each storefront's lib/cms.ts is the recommended construction site for CmsConfig (reads env + Next.js cookies).

Workspace scoping

The CMS enforces per-workspace scoping on all entry CRUD as of 2026-05-24. Set workspace on CmsConfig so every call sends the X-Workspace header.

Migrate content between CMS instances

scripts/migrate.ts walks a source CMS (locales → content types → fields → entries) and upserts everything onto a target CMS for one workspace. Idempotent; never deletes anything on the target.

Run from any consuming repo (the package's scripts/ lands in node_modules via the github: install):

set -a
SOURCE_CMS_URL=http://localhost:9000
SOURCE_CMS_USERNAME=admin
SOURCE_CMS_PASSWORD=admin
TARGET_CMS_URL=https://cms.ecomatic.se
TARGET_CMS_USERNAME=...
TARGET_CMS_PASSWORD=...
WORKSPACE=outdoorexperten
set +a

# Dry run first (logs every action, writes nothing)
DRY_RUN=true npx tsx node_modules/@wesports/cms-client/scripts/migrate.ts

# Real run
npx tsx node_modules/@wesports/cms-client/scripts/migrate.ts

Both sides use admin JWT auth (workaround for the known cms_*-token-on-list-endpoints bug). Workspace + locales must already exist on the target — see infrastructure/deploy/runbooks/cms-secrets.md.

Entry matching: singletons by content type; collection entries by their natural key (key or slug field in data). Entries without either are skipped with a warning.

Out of scope: media uploads (binary files), workspace provisioning. Media must be re-uploaded separately.

Env vars:

• SOURCE_WORKSPACE / TARGET_WORKSPACE — workspace slugs (independent on each side). Falls back to WORKSPACE if you want the same on both.
• LOCALE — defaults to sv-SE. Critical — list endpoints default-filter to ?locale=en, so the wrong locale means "0 entries found" silently.

Flags:

• DRY_RUN=true — log all actions, no writes
• ONLY=home-page,banner — restrict to specific content-type slugs
• SKIP_LOCALES=true — don't touch /api/locales

Caveat — verify migrations with a cache flush

The CMS Redis cache for list endpoints has no write-through invalidation, so a GET /api/<ct>/entries?... immediately after the migrate may return the pre-migration view for a while. The migrate itself uses per-call cache-busting (_t=<now> on every request) so the migration logic is correct — but if you're querying with a different client to verify, you'll see stale data.

Either query with a unique query-param every time, or flush:

docker exec main-redis-1 redis-cli FLUSHALL

Tracked in frontend-suite/shared/memory/reference_debt.md under CMS-side debt.