FAQ
Questions about the tool
Short answers. The longer version of the extract itself is on How it works.
Does TableJSON OCR a scanned PDF?
No. It reads the text layer only. A scan needs to be recognized somewhere else first. An invoice scan comes back as FAIL: no text layer, and the Excel file still downloads.
Do you keep my PDF?
A multipart upload is a temporary file for that request and is deleted when the response is sent. It is not added to a document library. URL fetches stay in memory for the request.
Do I need an account?
The playground on tablejson.com does not. A direct API call, Zapier, or RapidAPI does. There is no customer login on this site. Ask for a key on the contact page.
Why is the JSON one column, or a line_number table?
The text did not split into columns. That happens with scans, with words that are not spaced like a table, and with borders that are drawings rather than text. Check the PDF, or export CSV from the system that made it.
Can I get Excel or CSV?
Invoices download as .xlsx from the right-hand card or from POST /v1/invoice-xlsx. Add ?format=json for the checksum preview. Other tables return JSON. Turn the rows into CSV yourself. See JSON vs CSV.
What are the size limits?
500 MB for an upload. 50 MB when you send a URL. Twenty extract requests a minute per IP on the API.