app/prompts.py

from transformers import AutoTokenizer
from langchain_core.prompts import PromptTemplate
from typing import List
import models

tokenizer = AutoTokenizer.from_pretrained("meta-llama/Meta-Llama-3-8B-Instruct")

raw_prompt = "{question}"

history_prompt = """
Given the following conversation provide a helpful answer to the follow up question.

Chat History:
{chat_history}

Follow Up question: {question}
helpful answer:
"""

question_prompt = """
Given the following conversation and a follow up question, rephrase the 
follow up question to be a standalone question, in its original language.

Chat History:
{chat_history}

Follow Up Input: {question}

Standalone question:
"""

context_prompt = """
Answer the question based only on the following context:
{context}

Question: {standalone_question}
"""

map_prompt = """
Given the following list of file paths, return a comma separated list of the most likely files to have content that could potentially help answer the question. Return nothing if none of those would help. 
Make sure to return the complete full paths as it is writen in the original list

File list:
{file_list}

Question: {question}

Return a comma separated list of files and nothing else!
Comma separated list:
"""



def format_prompt(prompt):
    chat = [
        {"role": "system", "content": "You are a helpful AI assistant."},
        {"role": "user", "content": prompt},
    ]

    formatted_prompt = tokenizer.apply_chat_template(
        chat, 
        tokenize=False, 
        add_generation_prompt=True
    )

    return PromptTemplate.from_template(formatted_prompt)


def format_chat_history(messages: List[models.Message]):
    return '\n'.join([
        '{}: {}'.format(message.type, message.message)
        for message in messages
    ])

def format_context(docs: List[str]):
    return '\n\n'.join(docs)


raw_prompt_formatted = format_prompt(raw_prompt)
history_prompt_formatted = format_prompt(history_prompt)
question_prompt_formatted = format_prompt(question_prompt)
context_prompt_formatted = format_prompt(context_prompt)
map_prompt_formatted = format_prompt(map_prompt)