Spaces:
Running
Running
| import base64 | |
| import requests | |
| import os | |
| import shutil | |
| from discord_webhook import DiscordWebhook | |
| from dotenv import load_dotenv | |
| load_dotenv() # take environment variables from .env. | |
| def empty_directory(directory): | |
| # Check if the directory exists | |
| if not os.path.exists(directory): | |
| return | |
| # Iterate over all files and directories within the specified directory | |
| for item in os.listdir(directory): | |
| item_path = os.path.join(directory, item) # Get the full path of the item | |
| # Check if the item is a file or directory and delete accordingly | |
| if os.path.isfile(item_path): | |
| os.remove(item_path) # Remove the file | |
| elif os.path.isdir(item_path): | |
| shutil.rmtree(item_path) # Remove the directory and all its contents | |
| def save_uploaded_file(directory, file): | |
| if not os.path.exists(directory): | |
| os.makedirs(directory) | |
| file_path = os.path.join(directory, file.name) | |
| with open(file_path, "wb") as f: | |
| f.write(file.getbuffer()) | |
| return file_path | |
| def make_discord_trace_multimodal(image_path, text_message): | |
| webhook = DiscordWebhook(url=os.environ["DISCORD_HOOK"], | |
| username="invoice parsing using GPT_Vision", | |
| content=text_message) | |
| # send two images | |
| with open(image_path, "rb") as f: | |
| webhook.add_file(file=f.read(), filename="input.jpg") | |
| webhook.execute() | |
| def pass_to_openai_vision_api(image): | |
| # OpenAI API Key | |
| api_key = os.environ["OPENAI_API_KEY"] | |
| # Getting the base64 string | |
| base64_image = base64.b64encode(image.read()).decode('utf-8') | |
| headers = { | |
| "Content-Type": "application/json", | |
| "Authorization": f"Bearer {api_key}" | |
| } | |
| gpt_prompt='''Above is the text extracted from an invoice. | |
| You are an assistant tasked with extracting information from the invoice. Do this step by step. | |
| 1. First extract the date and due date. | |
| 2. Then assign it a category (e.g Food). | |
| 3. Extract the invoice number and vendor account number. | |
| 4. Extract the total amount. | |
| 5. Extract the items along with their name, quantity and individual price. | |
| Output should only contain a dictionary in the following format | |
| { | |
| "Date": None, | |
| "Due Date": None, | |
| "Category": None, | |
| "Invoice Number": None, | |
| "Vendor Account Number": None, | |
| "Total Amount": None, | |
| "Items": [ | |
| { | |
| "Item": None, | |
| "Quantity": None, | |
| "Individual Price": None | |
| } | |
| ] | |
| } | |
| If a key is not mentioned in invoice or you dont understand, then make its value None | |
| ''' | |
| payload = { | |
| "model": "gpt-4-vision-preview", | |
| "messages": [ | |
| { | |
| "role": "user", | |
| "content": [ | |
| { | |
| "type": "text", | |
| "text": gpt_prompt | |
| }, | |
| { | |
| "type": "image_url", | |
| "image_url": { | |
| "url": f"data:image/jpeg;base64,{base64_image}" | |
| } | |
| } | |
| ] | |
| } | |
| ], | |
| "max_tokens": 300 # To be investigated | |
| } | |
| response = requests.post("https://api.openai.com/v1/chat/completions", headers=headers, json=payload) | |
| return response.json()['choices'][0]['message']['content'] |