What the parser extracts

Receipts look simple to a person, but their layouts vary widely. ClearLedger normalizes the most useful fields into a consistent result instead of returning a block of OCR text.

FIELDEXAMPLEWHY REVIEW IT
merchantCedar & Finch KitchenLogos, headers, and addresses can compete for the merchant name.
date2026-08-18Order dates, payment dates, and print dates may all appear.
subtotal / tax / total25.00 / 2.06 / 27.06Tax, tips, service fees, discounts, and totals need reconciliation.
line itemsMarket salad · 1 · 12.00Wrapped descriptions and quantity formats vary by merchant.
payment methodVisa •••• 4242A receipt may show several payment references.

How PDF receipt parsing works

  1. Validate the PDF

    The service checks the PDF signature, file size, encryption state, readability, and page count before extraction.

  2. Read text or run OCR

    Pages with an embedded text layer are read directly. Sparse scanned pages are rendered and processed with OCR inside the ClearLedger service.

  3. Structure fields and rows

    Extraction rules identify receipt labels, values, line items, confidence signals, and normalized source coordinates.

  4. Review and export

    You correct uncertain fields, reconcile the amounts, and export the reviewed result to UTF-8 CSV.

Supported PDF receipts

ClearLedger currently accepts one PDF at a time. It supports text-based receipts and scanned PDFs, including receipts that use common English or Chinese labels.

  • Maximum file size: 20 MB
  • Maximum length: 30 pages
  • Embedded-text PDF: read directly
  • Scanned PDF: OCR fallback
  • Password-protected or unreadable PDF: rejected
NOT CURRENTLY SUPPORTED

JPG/PNG uploads, bulk folders, password removal, and unattended accounting submission are outside the current product boundary.

Why source-linked review matters

OCR accuracy is only one part of the problem. The parser also has to decide which date is the transaction date, whether a fee belongs with tax, and which amount is the final total. ClearLedger stores source coordinates and confidence with each field so the reviewer can inspect the corresponding area on the first-page preview.

The parser reads text from every page. The visual source preview currently shows the first page, so a field found on a later page is identified by page number rather than highlighted on-screen.

Use the result as reviewed data

ClearLedger organizes receipt information; it does not provide accounting, financial, legal, or tax advice. Before using an export for reimbursement, bookkeeping, or tax work, confirm the merchant, date, currency, amounts, and line items against the original document.