Turn Paper Receipts into a CSV with Free Tesseract OCR

Why this works

Receipts fade, and expense apps charge monthly fees. Tesseract is a free, open-source OCR engine that reads printed text well. A short script and a spreadsheet are enough to turn a folder of photos into a budget.

Step by step

1. Photograph flat and well litPlace each receipt on a dark surface, shoot from above, and save as PNG or JPG.
2. Install TesseractUse your package manager (for example brew install tesseract or your Linux repository) and check with tesseract --version.
3. Extract the textRun tesseract receipt.png stdout -l eng for each photo and save the output to a text file per receipt.
4. Pull out the fieldsAsk a local model, or a short script, to extract date, merchant, total, and category into CSV columns. Use the prompt below.
5. Reconcile the totalCheck the extracted total against the printed total for every receipt. Fix the ones that differ.

Copy-paste prompt

From this OCR text of one receipt, return one CSV line with the columns: date (YYYY-MM-DD), merchant, total (numbers only, dot as decimal), currency, category (food, transport, household, health, other). If a field is not clear, write UNCLEAR in that cell. Output the header line once, then one data line.

OCR TEXT:
<paste>

Check your result

• Each total matches the printed receipt
• Dates are in one format
• UNCLEAR cells were fixed by hand

Pitfalls to avoid

• Thermal receipts fade over time; scan them soon after you get them.
• OCR confuses 0/O and 1/l. Check any total that looks wrong.