PDF Parser

Extract the text from a PDF, as one string or one string per page.

What it does

PDF Parser reads a PDF held in a field of the incoming item and returns its text and its number of pages. The PDF can be a base64 string, a data: URL with base64 content, or raw bytes. Everything happens in your browser; the file is not uploaded.

With Merge pages on, you get the whole document as one string. Turn it off to get a list with one string per page.

When to use it

• Summarise or question a PDF with an AI node, using its text.
• Search a PDF’s text with Regex, for example to find an invoice number.
• Handle each page separately: turn Merge pages off, then use Split Out.
• For a PDF downloaded by an Http Request, use Extract From File instead: it reads the downloaded file directly.

Inputs and settings

Setting	Type	What it does
Input field	Text	The path to the field holding the PDF data (base64 string, data URL, ArrayBuffer/Uint8Array, or array of bytes). Default: data
Output field name	Text	The name of the output field that will hold the extracted text. Default: text
Merge pages	Toggle	When enabled, output one merged string for the whole document. When disabled, output an array with one string per page. Default: On
Options	Group	The options for the node.
Options › Include input fields	Toggle	Whether to include the original input fields in the output. Default: Off

Input field is a field name, not an expression. Use dots for nested fields, for example file.content. After Local File Read, set it to content and read the file as Data URL or Base64.

Outputs

• Output

The output has two fields: the text under the Output field name (text by default) and pageCount. Turn on Include input fields to keep the incoming fields too, such as the file name from Local File Read. Use {{ $input.text }} in the next node.

Dependencies and credentials

None. This node needs no connection and has no slots for other nodes.

Example workflow

1. Local File Read with Read as Data URL and Accepted file types .pdf.
2. PDF Parser with Input field content.
3. Summarization with Text set to {{ $input.text }}, using a chat model such as Chat OpenAI.

Common issues

• “Field “data” does not exist in the input object …“ No field has the name in Input field. Set it to the field that holds the PDF, for example content.
• “Field “content” could not be decoded as base64 PDF data.“ The field holds plain text, not a PDF encoding. Read the file as Data URL or Base64.
• “… is a non-base64 data URL, which is not supported for PDF data.” The data URL is not base64. Produce a base64 data URL instead.
• The text is empty or very short. The PDF is a scan: its pages are images, not text. This node does not read text from images.

Related nodes

• Extract From File: read PDFs, CSV, JSON and more from a downloaded file.
• Local File Read: pick a PDF from your computer.
• Split Out: one item per page when pages are not merged.