Scanned pdf with API and ask questions

API
,
1

Hello, I want to give the ChatGPT API multiple scanned PDF files and ask questions about them. From what I’ve seen, I was only able to send images to the API. Can anyone help?

1 Like
2

Hi there and welcome to the Community!

Your understanding is correct. You would have to supply the PDF pages as images to one of the models that support vision (i.e. gpt-4-turbo or the newer gpt-4o models) in order to achieve that.

3 
import openai
import base64
from PIL import Image
from io import BytesIO
from openai import OpenAI

# client = OpenAI()

# Load image from file
def load_image_as_base64(image_path):
    with open(image_path, "rb") as image_file:
        return base64.b64encode(image_file.read()).decode("utf-8")

# Example usage
image_path = "./Photos/10_I.png"  # Replace with the path to your local image
encoded_image = f"data:image/jpeg;base64,{load_image_as_base64(image_path)}"

# Use this encoded image as part of your request
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[
        {
            "role": "user",
            "content": [
                {
                    "type": "text",
                    "text": "tell me this fields:Sendername, sender taxcode, sender address, Date, Receiverbankcountry, Receiver client address, Receiverbankcode, Receiverbankname, Receiveraccount, Receivername, Currency, Amount, Details of payment, Invoice number.",
                },
                {
                    "type": "image_url",
                    "image_url": {"url": encoded_image},
                }
            ],
        }
    ],
    max_tokens=500,
)

print(response.choices[0].message.content)

This is the code I’m using now, but I want to provide PDF file input because there are multiple images in the PDF. I want to upload more than one PDF. Is it not possible? i have to send images one by one?

4

I’d recommend having a look at this OpenAI cookbook:

2 Likes