In your app
A controller answers a route on your app: it callsimportBlogPost() and returns the answer.
Add the controller
app/controllers/import_blog_post_controller.ts
Add the route
start/routes.ts
.use(middleware.auth())): it acts with your key’s access.Call it
Your own page’s form or script sends
POST /blog/from-markdown and gets back its answer as JSON. Test it, below, calls it.text/markdown, at most 5 MB by default), or a multipart upload with the file and its images.
Images: featuredImage in the front matter and every image in the body may be an https:// URL (fetched, at most 15 MB by default), a library image’s slug, a filename posted with the file (image, up to 10), or generate: <description> with generateImages (at most 4 per import).
Media on an API key: a new image (a URL, a posted file or generate:) needs content:images:write, a new video content:videos:write and content:videos:publish; a library slug needs content:images:read / content:videos:read. The blog scopes don’t cover media, and the refusal names the missing scope.
Publishing: publish=true publishes only when the caller may (a key needs content:blog:publish); otherwise the post is saved as a draft. An MDX import is always a draft.
Removing an import: delete the post by the draft _id the import answered. On an API key that needs content:blog:delete.
Required parameters
Required parameters
One object, e.g.
importBlogPost({ markdown }); leave out the optional fields.Required. The whole document: YAML front matter between
--- lines, then the Markdown body. title is the one required front matter key; slug, publishedAt, authorName (or authorEmail), category, excerpt, featuredImage (+ featuredImageAlt), seoTitle, seoDescription and splitDepth are optional (the Markdown template, getBlogMarkdownTemplate, names every key). Body headings start at ##: the title is the post’s only <h1>.Optional parameters
Optional parameters
Optional. Leave it out (or pass
null) and it isn’t sent: the site’s default applies. Read a raw or JSON body as mdx. One of md, mdx.Optional. Leave it out (or pass
null) and it isn’t sent: the site’s default applies. The same as the body’s publish.Optional. Leave it out (or pass
null) and it isn’t sent: the site’s default applies. The same as the body’s generateImages.Returns
Returns
A
BlogImportResult. The new post. Read its properties (blogImportResult.id): each says whether the site always sends it. The example answer is the site’s wire JSON: some keys differ from the property names (_id → id, _url → url).Errors
Errors
Nothing to catch: the SDK retries what is safe, and throws the rest as the classes below. Your exception handler turns them into your app’s own answers, once for every call (see Map SDK errors to responses); catch only what you want to react to here.
The request is not valid; the message says which part and how to fix it.Retried first: No. Read:
errors (field → messages), siteMessage.The API key is missing, malformed, unknown, disabled, expired or archived, or a request its signing mode requires to be signed is not.Retried first: No. Read:
siteMessage, errorName, data.The API key lacks the scope this route needs (the message names it) or is used from outside its allowlist; or, on a route that answers 403 to everyone it refuses, the caller may not use it.Retried first: No. Read:
siteMessage, errorName, data.The document or a file is over its size limit (the message names it).Retried first: No. Read:
reason, status.The key used up its per-minute limit. Still, after the client’s own waits and retries.Retried first: Yes: the client waits
Retry-After (1 s when absent; over 60 s it gives up at once) and retries, up to retries times (2 by default). Read: retryAfter (seconds, or null), siteMessage.The site failed (or something in front of it did). Its logs say why.Retried first: No: a write may have landed. Read:
siteMessage, errorName, data.No answer: DNS, a refused connection, TLS, or the timeout (15 s per attempt by default).Retried first: Only when nothing reached the site (
neverConnected()); never after a timeout, when it may have landed. Read: reason, neverConnected.The client refused to send it (e.g. a path segment of
..); the site never saw it.Retried first: No. Read: reason (what to change).Any other status, or a success whose JSON doesn’t match this SDK version’s types (update the SDK).Retried first: No. Read:
reason, status.sites.forSite('acme') named a slug config/attract.ts doesn’t list (with the keys in the database: no row in attract_api_keys).Retried first: No: nothing was sent. Read: message, slug.Behavior
- Access: an API key with
content:blog:write, orcontent:blog:all, orcontent:all, oradmin:all, or a signed-in user holding theblog:import-markdownpermission. Other callers get 403. - Rate limit: the key’s own writes-per-minute limit, set when the key is created (default 120), and 30 imports and validations a minute per caller: whichever runs out first answers 429 with
Retry-After. The client waits and retries for you; you seeRateLimitedonly if that runs out. - Retries: a write is sent again only after a 429 (refused before it acted) or when it never reached the site. After a timeout or a 5xx it may have landed: check before you send it again.
- Timeout: the SDK gives this call 615 s on its own (your client’s own timeout when that is longer).
Test it
It calls the controller above with the SDK faked:fakeAttract() puts the fake in the container for this one test, and it answers every call with a valid example, so nothing leaves your machine and no key is read. Save it as tests/unit/blog_posts_import_blog_post.spec.ts, then run your tests:
tests/unit/blog_posts_import_blog_post.spec.ts
Terminal
blogPosts.importBlogPost with factories.BlogImportResult({...}), the fields you give over a valid example.
tests/unit/blog_posts_import_blog_post_answers.spec.ts
Related
List a blog post’s locales
site.blogPosts().getBlogLocales()Get the Markdown template
site.blogPosts().getBlogMarkdownTemplate()Get the MDX template
site.blogPosts().getBlogMdxTemplate()List blog posts
site.blogPosts().listBlogs()Create a blog post
site.blogPosts().createBlog()Get a blog post
site.blogPosts().getBlog()POST /@apostrophecms/blog/from-markdown
The endpoint this calls, in the API reference: the HTTP request and answer, and every rule the site checks.