Files
tibi-svelte-starter/api/hooks/ssr/AGENTS.md
T
apairon 4020ad62c5 feat: enhance medialib image handling and add asset URL resolution
- Implemented `resolveApiAssetUrl` function to normalize asset URLs based on API base.
- Updated `MedialibImage` component to utilize new asset URL resolution and added support for alt text and class properties.
- Enhanced image loading behavior with improved width measurement and focal point handling.
- Added placeholder image handling and improved accessibility with alt text.
- Introduced new test script for auditing broken links in skill documentation.
- Expanded seeded test content to include medialib entries and updated related tests for pagebuilder previews.
- Improved global setup and teardown logging for clarity on seeded content management.
2026-05-17 00:52:41 +00:00

56 lines
3.1 KiB
Markdown

# SSR and Caching
Server-side rendering via goja (Go JS runtime) with HTML caching.
For the full workflow, prefer the `tibi-ssr-caching` skill.
## Related skills
- `frontend-architecture` when SSR changes interact with route loading, i18n, or `App.svelte` data flow.
- `media-seo-publishing` when SSR output includes image URLs, SEO metadata, or publication timing.
- `tibi-hook-authoring` when SSR changes depend on neighboring hook behavior such as public filtering or cache invalidation.
## Request flow
1. `get_read.js` receives the request and calls `lib/ssr-server.js`.
2. `get_read.js` loads `lib/app.server.js` and calls `app.default.render({ url })`.
3. `frontend/src/ssr.ts` stays thin and only initializes locale state before rendering `App.svelte`.
4. `frontend/src/App.svelte` is responsible for actual page data loading for both browser and SSR.
5. During SSR, `App.svelte` calls the same `loadContent(...)` path directly inside `typeof window === "undefined"`.
6. Rendered HTML is stored in the `ssr` collection together with dependency tracking strings.
Keep browser and SSR content loading on the same application path whenever possible. Do not fork a second data-loading model for SSR unless the current architecture explicitly requires it.
## Build
- SSR bundle is built via `yarn build:server` and outputs to `lib/app.server.js`.
- The project no longer uses Babel for SSR.
- goja-compatible transforms are configured in `esbuild.config.server.js` via `supported`.
- The server build must remove frontend-only splitting/outdir options inherited from the shared esbuild config.
## Cache invalidation
- `clear_cache.js` hook invalidates SSR cache entries based on collection dependencies.
- Dependencies are stored as strings like `content:<id>` or `content:*`.
- `DELETE` invalidation must be robust even when `context.data.id` is missing.
## Route validation
- SSR route validation is active in `config.js`.
- Public page URLs are language-prefixed (`/de/...`, `/en/...`), while `content.path` in the DB is stored without that prefix.
- `ssrValidatePath()` must strip the language prefix before querying content and return a canonical language-prefixed URL when needed.
- If route validation changes, inspect `api/hooks/config.js`, `filter_public.js`, and the frontend route-loading path together instead of patching only one side.
## 404 signaling
The SSR hook (`get_read.js`) checks `context.is404` after rendering to determine the HTTP status code. If `true`, it returns HTTP 404 with the rendered 404 page HTML (and does not cache the result).
`NotFound.svelte` sets `context.is404 = true` during SSR. When the component renders (only when the page is not found), its top-level script sets the flag. The goja runtime provides `context` as a global during SSR, so it's available from the compiled frontend code.
When adding a 404 page or changing the not-found logic, ensure `context.is404` is still set during SSR.
## Related validation
- After SSR changes, run `yarn build:server` plus the narrowest reachable SSR/public-read check.
- If caching or publication behavior changed, also rerun the relevant mutation-side invalidation check instead of relying on HTML diffing alone.