Get started

Quickstart

Send a document, wait for it, read the answer: the three calls every Northdoc integration is built on.

Three calls: send the file, wait until it is done, read the answer. This one asks for two fields of your own; leave out the options and you get the pages and pictures alone. Everything else in these docs is a variation on this. It uses the two shell variables from Authentication.

terminal

# 1. Send the file and keep its id
RESP=$(curl -s -X POST "$NORTHDOC_API/documents" \
  -H "Authorization: Bearer $NORTHDOC_KEY" \
  -F file=@invoice.pdf \
  -F 'options={"extraction": {"schema": {
        "type": "object",
        "properties": {"invoice_number": {"type": "string"}, "total": {"type": "number"}},
        "required": ["invoice_number", "total"]
      }}}')
DOC_ID=$(echo "$RESP" | jq -r '.id // empty')
[ -n "$DOC_ID" ] || { echo "$RESP"; exit 1; }

# 2. Check every two seconds until it is completed or failed
while :; do
  BODY=$(curl -s "$NORTHDOC_API/documents/$DOC_ID" -H "Authorization: Bearer $NORTHDOC_KEY")
  STATUS=$(echo "$BODY" | jq -r .status)
  echo "$STATUS"
  case "$STATUS" in
    completed|failed) break ;;
    queued|processing) sleep 2 ;;
    *) echo "$BODY"; exit 1 ;;
  esac
done

# 3. Read the data, the citations behind it, and any error
curl -s "$NORTHDOC_API/documents/$DOC_ID" \
  -H "Authorization: Bearer $NORTHDOC_KEY" \
  | jq '{data: .result.data, citations: .result.citations, error}'