Extract structured data from a document

Extract structured data from a document

Pull specific fields from a document into a typed schema.

Accepts either a file upload (multipart/form-data) or a document URL (JSON body), plus a JSON Schema (the schema field) describing what to extract.

Sync mode (default)

Blocks until extraction completes and returns 200 with the full result.

curl -X POST https://api.context212.com/api/v1/extract \  -H 'Authorization: Bearer $TOKEN' \  -F file=@invoice.pdf \  -F 'schema={"type":"object","properties":{"invoice_number":{"type":"string"}}}'

Scope (options.scope)

• page (default): apply the schema independently to each page. result.data has one object per page (fields missing on a page are null).
• document: apply the schema once to the full document (all pages joined). result.data is a single-element array with the consolidated object. Use this when fields and arrays span multiple pages (e.g. an auction notice with lots).

Async mode (options.async = true)

Not implemented yet — returns 501.

Embedded images (options.include_images, default true for extract)

Extract uploads every embedded PDF image, injects ![image N](url) into the parsed markdown before the LLM runs (so schema fields can capture URLs), and returns the same list on result.images.

Pass options.include_images: false to skip uploads.

For multipart: -F 'options={"scope":"document"}' (images on by default).

Supported file types: .pdf, .png, .jpg, .jpeg, .pptx, .docx, .xlsx, .html, .xhtml

Sync limits: 20 MB file size, 15 pages.

Header Parameters

Response Body