utils.py

from openai import OpenAI
import streamlit as st
from PIL import Image
from io import BytesIO
import base64
import json
import re


# Convert Image to Base64 
def im_2_b64(image):
    image = Image.open(image)
    buff = BytesIO()
    image.save(buff, format="JPEG")
    img_str = base64.b64encode(buff.getvalue())
    return img_str


RANDOM_SEED = 42
client = OpenAI(api_key=st.secrets["OPENAI_KEY"])


with_prompt = """
In the following task, you will be presented with a image and a story that is, in some manner, related to that goal. You will be given a specific  goal and entity (generally, person or object), and be asked to identify condition necessary for that goal and the alternate condition that could prevent that goal.

Conditions for the output:
1. Condition: The condition is the necessary condition for the goal to be achieved. If the condition is not met, the goal cannot be achieved. The condition must be related to the entity.

2. Alternate Condition: The alternate condition is a different version of the condition that would prevent the goal from being achieved. It is likely that this alternate condition will contradict information provided in the images and/or story.

Output in a python dictionary where it should have the following keys: 'condition', 'alternate_condition'.

Story: {story}

Entity: {entity}

Goal: {goal}
"""

wo_prompt = """
In the following task, you will be presented with a image and a story that is, in some manner, related to that goal. You will be given a specific  goal, and be asked to identify an entity (person or object), condition necessary for that goal and the alternate condition that could prevent that goal.

Conditions for the output:
1. Entity: The entity is the person or object that the goal is related to. The entity should be a crucial part for achieving the goal.

2. Condition: The condition is the necessary condition for the goal to be achieved. If the condition is not met, the goal cannot be achieved. The condition must be related to the entity.

3. Alternate Condition: The alternate condition is a different version of the condition that would prevent the goal from being achieved. It is likely that this alternate condition will contradict information provided in the images and/or story.

Output in a python dictionary where it should have the following keys: 'entity', 'condition', 'alternate_condition'.

Story: {story}

Goal: {goal}
"""

data = {
    "Story id": [],
    "Prompt": [],
    "entity": [],
    "agent": [],
    "story": [],
    "Image1": [],
    "Image2": [],
    "Image3": [],
    "GPT-4 Output": [],
}


def get_gpt4V_response(story, goal, entity, image, temperature=0.5):
    # Convert image to base64
    image_b64 = im_2_b64(image)
    image_url = f"data:image/jpeg;base64,{image_b64.decode('utf-8')}"
    st.write("✅ Image converted")

    if entity:
        prompt = with_prompt
        now_prompt = prompt.format(story=story, goal=goal, entity=entity)
    else:
        prompt = wo_prompt
        now_prompt = prompt.format(story=story, goal=goal)

    content = [
        {"type": "text", "text": now_prompt},
    ]

    st.write("✅ Prompt created")

    content.append({
        "type": "image_url",
        "image_url": {
            "url": image_url,
        },
    })
    
    st.write("🚀 Getting Response from GPT4V")
    response = client.chat.completions.create(
        model="gpt-4-vision-preview",
        seed=RANDOM_SEED,
        messages=[
            {
                "role": "user",
                "content": content
            }
        ],
        temperature=temperature,
        max_tokens=1024,
        # top_p=1,
        # frequency_penalty=0,
        # presence_penalty=0,
    )
    print(response)
    print("Prompt:")
    print(now_prompt)
    out = response.choices[0].message.content
    print("OUTPUT:", out)
    print("====================================")
    print()

    st.write("✅ Response generated")

    return out

def get_str_to_json(st):
    st = re.sub(r"```python", "", st)
    st = re.sub(r"```", "", st)
    st = re.sub(r"'(.*)': '(.*)'", r'"\1": "\2"', st)
    st = re.sub(r"'(.*)': (.*)", r'"\1": \2', st)
    st = re.sub(r"(.*): '(.*)'", r'\1: "\2"', st)
    st = st.replace("\\'", "'")
    st = st.replace('\\"', '"')
    st = st.replace("True", "true")
    st = st.replace("False", "false")
    st = st.replace("None", "null")
    st = st.replace("nan", "null")
    st = st.replace("inf", "null")
    st = st.replace("-inf", "null")
    st = st.replace(",,", ",null,")
    st = st.replace(",]", ",null]")
    st = st.replace(",}", ",null}")
    st = st.replace(",,", ",null,")
    st = st.strip()
    st = json.loads(st)
    return st