PDF Tax Extract icon

PDF Tax Extract

Extract tax form fields and PS Form 1583 applicant names from PDFs using embedded text positions. No OCR.

Actions4

Form 1120 Line 1c

AI-generated

Summary

Extracts line 1c (Balance, line 1a minus 1b) from a Form 1120 PDF by reading embedded text positions (no OCR) and adds it to each item as a number in line1c.

Inputs

  • Input Binary Property (required) — Name of the binary field on each input item that holds the PDF file. Defaults to data. If the field is missing, the item errors (“Binary property … was not found”).

Output shape

One output item per input item. It copies the input JSON and adds line1c (number, or null on failure), error (empty string on success), and pdfTaxExtract: { operation: 'form1120Line1c', success }.

The node scans pages for one containing 'U.S. Corporation Income Tax Return' and 'Balance. Subtract line 1b from line 1a'. It then takes the rightmost numeric value on the same row as the 'Balance.' label. Amounts are parsed to numbers: $ and commas are stripped, and parentheses or a leading minus give a negative. If the label is found but no number is on its row, line1c is 0 with no error. Failures set line1c to null and fill error: 'Form 1120 page not found', password-protected or unreadable PDF, or an amount that can't be parsed. Scanned image-only PDFs without embedded text won't work. With Continue On Fail enabled, thrown errors (such as a missing binary) return an item containing only error.

Examples

Example 1: Pull gross receipts net of returns from a corporate return PDF downloaded by an HTTP Request node.

Operation: Form 1120 Line 1c; Input Binary Property: data. Read {{ $json.line1c }} downstream.

Example 2: Route failed extractions for manual review.

Run the operation, then add an IF node on {{ $json.pdfTaxExtract.success }} being false.

Links

Discussion