appStore/rag.py

# import os
# # import json
# import numpy as np
# import pandas as pd
# import openai
# from haystack.schema import Document
# import streamlit as st
# from tenacity import retry, stop_after_attempt, wait_random_exponential


# # Get openai API key
# # openai.api_key = os.environ["OPENAI_API_KEY"]
# hf_token = os.environ["HF_API_KEY"]
# #model_select = "gpt-3.5-turbo-0125"
# model_select ="gpt-4"

# # define a special function for putting the prompt together (as we can't use haystack)
# def get_prompt(context, label):
#   base_prompt="Summarize the following context efficiently in bullet points, the less the better - but keep concrete goals. \
#   Summarize only elements of the context that address vulnerability of "+label+" to climate change. \
#   If there is no mention of "+label+" in the context, return nothing. \
#   Formatting example: \
#     - Bullet point 1 \
#     - Bullet point 2 \
# "

#   # Add the meta data for references
#   # context = ' - '.join([d.content for d in docs])
#   prompt = base_prompt+"; Context: "+context+"; Answer:"
  
#   return prompt

# # def get_prompt(context, label):
# #   base_prompt="Summarize the following context efficiently in bullet points, the less the better - but keep concrete goals. \
# #   Summarize only elements of the context that address vulnerability to climate change. \
# #   Formatting example: \
# #     - Bullet point 1 \
# #     - Bullet point 2 \
# # "

# #   # Add the meta data for references
# #   # context = ' - '.join([d.content for d in docs])
# #   prompt = base_prompt+"; Context: "+context+"; Answer:"
  
# #   return prompt

# #   base_prompt="Summarize the following context efficiently in bullet points, the less the better- but keep concrete goals. \
# #   Summarize only activities that address the vulnerability of "+label+" to climate change. \
# #   Formatting example: \
# #     - Collect and utilize gender-disaggregated data to inform and improve climate change adaptation efforts. \
# #     - Prioritize gender sensitivity in adaptation options, ensuring participation and benefits for women, who are more vulnerable to climate impacts. \
# # "
# # # convert df rows to Document object so we can feed it into the summarizer easily
# # def get_document(df):
# #     # we take a list of each extract
# #     ls_dict = []
# #     for index, row in df.iterrows():
# #         # Create a Document object for each row (we only need the text)
# #         doc = Document(
# #             row['text'],
# #             meta={
# #             'label': row['Vulnerability Label']}
# #         )
# #         # Append the Document object to the documents list
# #         ls_dict.append(doc)

# #     return ls_dict 


# # exception handling for issuing multiple API calls to openai (exponential backoff)
# @retry(wait=wait_random_exponential(min=1, max=60), stop=stop_after_attempt(6))
# def completion_with_backoff(**kwargs):
#     return openai.ChatCompletion.create(**kwargs)


# # construct RAG query, send to openai and process response
# def run_query(context, label):
#     '''
#     For non-streamed completion, enable the following 2 lines and comment out the code below
#     '''
#     # res = openai.ChatCompletion.create(model=model_select, messages=[{"role": "user", "content": get_prompt(docs)}])
#     # result = res.choices[0].message.content

#     # instantiate ChatCompletion as a generator object (stream is set to True)
#     response = completion_with_backoff(model=model_select, messages=[{"role": "user", "content": get_prompt(context, label)}], stream=True)
#     # iterate through the streamed output
#     report = []
#     res_box = st.empty()
#     for chunk in response:
#         # extract the object containing the text (totally different structure when streaming)
#         chunk_message = chunk['choices'][0]['delta']
#         # test to make sure there is text in the object (some don't have)
#         if 'content' in chunk_message:
#             report.append(chunk_message.content) # extract the message
#             # add the latest text and merge it with all previous
#             result = "".join(report).strip()
#             # res_box.success(result) # output to response text box
#             res_box.success(result)



import os
# import json
import numpy as np
import pandas as pd
import openai
from haystack.schema import Document
import streamlit as st
from tenacity import retry, stop_after_attempt, wait_random_exponential
from huggingface_hub import InferenceClient


# Get openai API key
hf_token = os.environ["HF_API_KEY"]


# define a special function for putting the prompt together (as we can't use haystack)
def get_prompt(context, label):
  base_prompt="Summarize the following context efficiently in bullet points, the less the better - but keep concrete goals. \
  Summarize only elements of the context that address vulnerability of "+label+" to climate change. \
  If there is no mention of "+label+" in the context, return nothing. \
  Formatting example: \
    - Bullet point 1 \
    - Bullet point 2 \
"

  # Add the meta data for references
  # context = ' - '.join([d.content for d in docs])
  prompt = base_prompt+"; Context: "+context+"; Answer:"
  
  return prompt




# # exception handling for issuing multiple API calls to openai (exponential backoff)
# @retry(wait=wait_random_exponential(min=1, max=60), stop=stop_after_attempt(6))
# def completion_with_backoff(**kwargs):
#     return openai.ChatCompletion.create(**kwargs)


def get_prompt(context, label):
  base_prompt="Summarize the following context efficiently in bullet points, the less the better - but keep concrete goals. \
  Summarize only elements of the context that address vulnerability of "+label+" to climate change. \
  If there is no mention of "+label+" in the context, return nothing. \
  Do not include an introduction sentence, just the bullet points as per below. \
  Formatting example: \
    - Bullet point 1 \
    - Bullet point 2 \
"

  # Add the meta data for references
  # context = ' - '.join([d.content for d in docs])
  prompt = base_prompt+"; Context: "+context+"; Answer:"

  return prompt


# # construct RAG query, send to openai and process response
# def run_query(context, label, chatbot_role):
#     '''
#     For non-streamed completion, enable the following 2 lines and comment out the code below
#     '''
#     # res = openai.ChatCompletion.create(model=model_select, messages=[{"role": "user", "content": get_prompt(docs)}])
#     # result = res.choices[0].message.content

#     messages = [
#       ChatMessage(role="system", content=chatbot_role),
#       ChatMessage(role="user", content=get_prompt(context, label)),
#     ]
#     response = llm.chat(messages)
#     return(response)



# tokenizer = AutoTokenizer.from_pretrained(
#     "meta-llama/Meta-Llama-3.1-8B-Instruct",
#     token=hf_token,
# )

# stopping_ids = [
#     tokenizer.eos_token_id,
#     tokenizer.convert_tokens_to_ids("<|eot_id|>"),
# ]

# Define the role of the chatbot
# chatbot_role = """You are an analyst specializing in climate change impact assessments and producing insights from policy documents."""

# construct RAG query, send to openai and process response
def run_query(context, label):
    '''
    For non-streamed completion, enable the following 2 lines and comment out the code below
    '''
    chatbot_role = """You are an analyst specializing in climate change impact assessments and producing insights from policy documents."""

    # res = openai.ChatCompletion.create(model=model_select, messages=[{"role": "user", "content": get_prompt(docs)}])
    # result = res.choices[0].message.content

    # Initialize the client, pointing it to one of the available models
    client = InferenceClient()

    response = client.chat.completions.create(
        model="meta-llama/Meta-Llama-3.1-8B-Instruct",
        messages=[
          ChatMessage(role="system", content=chatbot_role),
          ChatMessage(role="user", content=get_prompt(context, label)),
        ],
        stream=True,
        max_tokens=500
    )

    # iterate and print stream
    # for message in chat_completion:
    #     print(message.choices[0].delta.content, end="")

    # instantiate ChatCompletion as a generator object (stream is set to True)
    # response = completion_with_backoff(model=model_select, messages=[{"role": "user", "content": get_prompt(context, label)}], stream=True)
    # iterate through the streamed output
    report = []
    res_box = st.empty()
    for chunk in client.chat_completion(messages, stream=True)::
        # extract the object containing the text (totally different structure when streaming)
        chunk_message = chunk['choices'][0]['delta']
        # test to make sure there is text in the object (some don't have)
        if 'content' in chunk_message:
            report.append(chunk_message.content) # extract the message
            # add the latest text and merge it with all previous
            result = "".join(report).strip()
            # res_box.success(result) # output to response text box
            res_box.success(result)