This example shows how to extract structured invoice data from a folder of PDF / image invoices using the Parserdata API.

You point the script at a folder, it uploads each invoice to the /v1/extract endpoint, and prints back clean JSON with the extracted fields.

• API reference: https://parserdata.com/parserdata-api
• Pricing: https://parserdata.com/pricing

• Reads all files from a folder that match a pattern (e.g. ~/Downloads/invoices/*)
• Filters to supported file types: PDF, PNG, JPG, JPEG
• Sends each file to the Parserdata API with a natural language prompt
• Prints a structured JSON result for each invoice (invoice number, date, supplier, total, line items, etc.)
• Shows basic error handling and debugging output

• Python 3.8+
• A Parserdata API key (starts with pd_live_...)

git clone https://github.com/parserdata/parserdata-python-invoice-batch.git
cd parserdata-python-invoice-batch

We recommend using a virtual environment, but it's optional.

python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
pip install -r requirements.txt

You can set your API key as an environment variable:

macOS / Linux:

export PARSERDATA_API_KEY="pd_live_..."

Windows (PowerShell):

$env:PARSERDATA_API_KEY="pd_live_..."

Or you can hard-code it in the script (for quick tests only), but avoid committing your key to GitHub.

Create a folder with your test invoices, for example:

• C:\Users\YourName\Downloads\invoices\ (Windows)

• /Users/yourname/Downloads/invoices/ (macOS)

The script will pick up files with these extensions: .pdf, .png, .jpg, .jpeg

Open invoice_folder_extractor.py and adjust this line:

INPUT_GLOB = r"C:\Users\Admin\Downloads\invoices\*"

• Windows: INPUT_GLOB = r"C:\Users\YourName\Downloads\invoices*"

• macOS / Linux: INPUT_GLOB = "/Users/yourname/Downloads/invoices/*"

From the project root:

python invoice_folder_extractor.py

You should see output like:

=== invoice-001.pdf ===
Status: 200
{
  "invoice_number": "INV-001",
  "invoice_date": "2024-01-10",
  "supplier_name": "ACME Supplies",
  "total_amount": 1234.56,
  "line_items": [
    {
      "description": "Widget A",
      "quantity": 10,
      "unit_price": 12.34,
      "net_amount": 123.4
    }
  ]
}

If there's an error, you'll see:

=== invoice-001.pdf ===
Status: 400
Error body: { ... }

In this example, we use a natural language prompt:

PROMPT = (
    "Extract invoice number, invoice date, supplier name, total amount, and line items "
    "(description, quantity, unit price, net amount)."
)

You can change this to match your use case:

• Add/remove fields

• Request alternative names

• Ask for currency codes, tax amounts, etc.

For stricter outputs, you can switch to schema mode in the API. This example keeps it simple and just uses a prompt.

• 401 / 403 errors

Double-check your API key and that it's set in PARSERDATA_API_KEY.

• Timeouts

Default timeout is 300 seconds per request in this script. You can reduce it if needed:

r = requests.post(..., timeout=60)

• No files found

Check INPUT_GLOB and make sure the path and pattern match where your invoices actually are.

• Pipe results into a database or a CSV instead of printing

• Wrap the logic into a function and reuse it from another app

• Combine with automation tools (Zapier, Make, n8n, etc.) for fully automated flows

MIT – feel free to copy and adapt this example.

This repository is a reference example.

If you need help tailoring it to your workflow, or want advice on a more advanced Parserdata API integration (custom schemas, scale, or production use), reach out to us: support@parserdata.com