PDF Parse

Parses the text layer of a PDF into text, Markdown, or `flowyn.pdf_parse.v1` JSON. Headings, tables, and reading order are preserved. Pages without a text layer come back empty; Flowyn does not perform OCR.

What pdf parse does

Parses the text layer of a PDF into text, Markdown, or flowyn.pdf_parse.v1 JSON. Headings, tables, and reading order are preserved. Pages without a text layer come back empty; Flowyn does not perform OCR.

When to use pdf parse

• You need the content of invoices, reports, or contracts as structured data
• You feed documents to an LLM or an agent and want Markdown
• You index PDFs for search

Authentication

Send an API key as a bearer token. Keys are shown once, belong in a server environment, and are bound to one environment: flw_test_ keys act in test and flw_live_ keys in live.

Authorization: Bearer flw_test_••••

Endpoint

POST https://api.flowyn.dev/v1/pdf/parse

Request

Send an Idempotency-Key so a retried request cannot create a duplicate. The body is JSON.

curl -X POST https://api.flowyn.dev/v1/pdf/parse \
  -H "Authorization: Bearer $FLOWYN_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: parse-quarterly-001" \
  -d '{
    "file_id": "9c1e5b7a-2d4f-4a6b-8e3c-1f0a7d2b5c64",
    "output": "markdown"
  }'

Response

Flowyn returns HTTP 202. The response is abridged here; the OpenAPI document lists every field.

{
  "id": "2b7d9e1f-4a6c-4b8d-9f2e-7c1a3e5d8b90",
  "object": "job",
  "operation": "pdf_parse",
  "status": "queued",
  "file_id": "9c1e5b7a-2d4f-4a6b-8e3c-1f0a7d2b5c64",
  "request_id": "req_5f2a9c1e7b3d4a6f8e0c2b1d"
}

Lifecycle

1. Upload the file with POST /v1/files (tus resumable upload). The response is the file with its id.
2. Call POST /v1/pdf/parse with the file_id. Flowyn stores the operation and returns HTTP 202 with a job id before any work starts.
3. A managed worker claims the job, runs the operation, and records an attempt.
4. Read GET /v1/file-jobs/{id}. When status is complete, outputs lists the produced files; download one with GET /v1/files/{id}/download.

Status values

Retries

A file operation gets up to 3 attempts. Worker failures and timeouts are retried; an unreadable or unsupported file fails without a retry.

Idempotency

Send Idempotency-Key. The same key and body return the original job.

Errors

Limits

• output is text, markdown, or json
• Files up to 50 MB and 200 pages
• No OCR: image-only pages return empty content

Upload the file first

Uploads use the tus resumable upload protocol. One request can create the upload and send the bytes.

curl -i -X POST https://api.flowyn.dev/v1/files \
  -H "Authorization: Bearer $FLOWYN_API_KEY" \
  -H "Tus-Resumable: 1.0.0" \
  -H "Upload-Length: $(wc -c < report.pdf | tr -d ' ')" \
  -H "Upload-Metadata: filename $(printf report.pdf | base64),filetype $(printf application/pdf | base64)" \
  -H "Content-Type: application/offset+octet-stream" \
  --data-binary @report.pdf

Read it back

Keep the id and read the resource whenever you need its current state.

curl https://api.flowyn.dev/v1/file-jobs/2b7d9e1f-4a6c-4b8d-9f2e-7c1a3e5d8b90 \
  -H "Authorization: Bearer $FLOWYN_API_KEY"

Related

• DOCX Parse
• Intelligent PDF Splitter
• REST reference

Next step

Try it in the Playground with a sample PDF.