PDFs and other data types

Extract structured data from a PDF

This example demonstrates how to extract structured data from a PDF invoice using the Gemini API and Pydantic.

Import necessary libraries. You'll need Pydantic for this one.

import os
import requests
import json
import re
from pydantic import BaseModel, Field
from typing import List, Union
from google import genai
from google.genai import types

Define Pydantic models for structured data

class Item(BaseModel):
    name: str
    price_per_kg: Union[float, str] = Field(..., alias="price/kg")
    quantity_kg: Union[float, int] = Field(..., alias="quantity (kg)")


class InvoiceContents(BaseModel):
    sender: str
    recipient: str
    address: str
    full_total: Union[float, str]
    subtotal: Union[float, str]
    gst_value: Union[float, str] = Field(..., alias="GST")
    items: List[Item]

Load the PDF invoice inline from a URL. For PDFs larger than 20MB, you'll need to use the Files API for uploading

pdf_url = "https://www.princexml.com/samples/invoice/invoicesample.pdf"
response = requests.get(pdf_url)
response.raise_for_status()  # Ensure the download was successful
pdf_data = response.content

Initialize the Gemini client with your API key

client = genai.Client(api_key=os.environ.get("GEMINI_API_KEY", "YOUR_API_KEY"))

Configure the model for structured output. Specify the prompt text.

model = "gemini-2.5-pro-preview-03-25"
prompt_text = (
    "Extract the following information from the invoice: "
    "sender, recipient, address, full_total, subtotal, GST, "
    "and a list of items (name, price/kg, quantity (kg))."
)

Call the Gemini API to extract structured data from the PDF

response = client.models.generate_content(
    model=model,
    contents=[
        types.Part.from_bytes(data=pdf_data, mime_type="application/pdf"),
        prompt_text,
    ],
    config=genai.types.GenerateContentConfig(temperature=0.0),
)

Extract JSON from response text which might be formatted as a markdown code block. Check if the response is a markdown code block and extract the JSON content.

response_text = response.text
json_match = re.search(r"```(?:json)?\n(.*?)```", response_text, re.DOTALL)
if json_match:
    json_str = json_match.group(1).strip()
else:
    json_str = response_text.strip()

Parse the JSON response into the Pydantic model

invoice_data = json.loads(json_str)
invoice = InvoiceContents(**invoice_data)

Print the extracted data as a JSON string

print(invoice.model_dump_json(indent=2))

Running the Example

Install the Google Generative AI library and Pydantic

$ pip install google-genai pydantic requests

Then run the program with Python

$ python structured-data-extraction.py
{
  "sender": "SUNNY FARM",
  "recipient": "Denny Gunawan",
  "address": "221 Queen St\nMelbourne VIC 3000",
  "full_total": "$39.60",
  "subtotal": "$36.00",
  "gst_value": "$3.60",
  "items": [
    {
      "name": "Apple",
      "price_per_kg": "$5.00",
      "quantity_kg": 1
    },
    {
      "name": "Orange",
      "price_per_kg": "$1.99",
      "quantity_kg": 2
    },
    {
      "name": "Watermelon",
      "price_per_kg": "$1.69",
      "quantity_kg": 3
    },
    {
      "name": "Mango",
      "price_per_kg": "$9.56",
      "quantity_kg": 2
    },
    {
      "name": "Peach",
      "price_per_kg": "$2.99",
      "quantity_kg": 1
    }
  ]
}

Extract structured data from a PDF

Running the Example

Further Information