OpenAI Assistant

openaiassist.py

#!/usr/bin/env python
import openai
import ast
import os


openai.api_key = 'sk-SAxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxj'

userprompt ="""
You can have a multi-line prompt here.

Feel free to use paragraphs, etc.

"""

if os.path.exists("conversation.txt"):
  
  # Read the previous conversation from the text file
  with open('conversation.txt', 'r') as file:
    content = file.read()
    messages = ast.literal_eval(content)
    new_user_prompt = {"role":"user", "content":userprompt}
    messages.append(new_user_prompt)

else:
  messages=[
    {"role": "system", "content": "You are a helpful and friendly assistant."},
    {"role": "user", "content": userprompt},
  ]

response = openai.ChatCompletion.create(
    model="gpt-4",
    messages=messages,
)

response_message = response["choices"][0]["message"]

total_tokens = response["usage"]["total_tokens"]

response_text = response_message["content"]
response_role = response_message["role"]

formatted_response = {"role": response_role, "content": response_text}

messages.append(formatted_response)

# Write the conversation to a text file
with open('conversation.txt', 'w') as file:
    file.write(str(messages))


#print(responselength)
#print(str(total_tokens))
#print(messages)
print(response_text)

Basic, command line python3 script to interface with OpenAI GPT.

• Set OpenAI key (line 7)

• userprompt is your prompt you want to send. (Line 9)

• Set any particulars for the role you want GPT to play ( Line 27)

• Set the GPT model, gpt-3.5-turbo or gpt-4 (Line 32)

• Creates an ongoing conversation history. This is written (raw) to conversation.txt, created in the same directory you run this script in. To reset your conversation history, delete or rename conversation.txt