Email to Data (with an LLM)

Problem: I needed to file my taxes.

I had many email receipts (around 330) in .eml format. I needed to convert them to .pdf format before sending them to my accountant as a link to a cloud folder.

Issues:

• My email provider (FastMail) only allows me to download the emails in .eml format.
• The receipts follow a non-standard format (so getting a suitable regular expressions may be difficult).

My simple repo does the following:

1. Converts all email receipts (.eml files) to PDFs (The PDFs are largely plain text).
2. Extracts key financial data from those PDFs into a CSV file (which I can import neatly into Google Sheets)

Example Output

The PDF output looks like this (I manually redacted the actual data):

The CSV output looks like this:

File Name,Total Amount ($),Currency,Transaction Date,Descriptive Details
Receipt-2566-5568.pdf,47.42,USD,2050-06-01,"Render - Servers, PostgresDB, Redis usage for May 2050"
Receipt-2952-5288.pdf,9.52,EUR,2050-03-03,Twitter International ULC - Twitter Blue subscription

And the imported Google Sheets looks like this:

Use Cases

• Generally converting .eml files to PDFs
• Tax preparation
• Expense tracking
• Accounting reconciliation
• Digital receipt organization
• Audit preparation

Full repo here