app.py

import gradio as gr
import boto3
import json
import os
import numpy as np
import botocore
import time
from scipy.spatial.distance import cosine as cosine_similarity

theme = gr.themes.Base(text_size='sm')

# Retrieve AWS credentials from environment variables

AWS_ACCESS_KEY_ID = os.getenv('AWS_ACCESS_KEY_ID')
AWS_SECRET_ACCESS_KEY = os.getenv('AWS_SECRET_ACCESS_KEY')
AWS_REGION = os.getenv('REGION_NAME')
AWS_SESSION = os.getenv('AWS_SESSION')
BUCKET_NAME = os.getenv('BUCKET_NAME')
EXTRACTIONS_PATH = os.getenv('EXTRACTIONS_PATH')

employee_type = None
division = None
authenticated = False
extractions = {}

# Create AWS Bedrock client using environment variables
def create_bedrock_client():

    return boto3.client(
        'bedrock-runtime',
        region_name=AWS_REGION,
        aws_access_key_id=AWS_ACCESS_KEY_ID,
        aws_secret_access_key=AWS_SECRET_ACCESS_KEY,
        aws_session_token=AWS_SESSION
    )

# Create AWS S3 client using environment variables
def create_s3_client():
    
    return boto3.client(
        's3',
        aws_access_key_id=AWS_ACCESS_KEY_ID,
        aws_secret_access_key=AWS_SECRET_ACCESS_KEY,
        aws_session_token=AWS_SESSION
    )

# Read JSON directly into mem from S3
def read_json_from_s3():

    response = s3_client.get_object(Bucket=BUCKET_NAME, Key=EXTRACTIONS_PATH)
    file_content = response['Body'].read().decode('utf-8')
    json_content = json.loads(file_content)
    
    return json_content

# Get AWS Titan embedding of text
def get_titan_embedding(bedrock_client, doc_name, text, attempt=0, cutoff=10000):
    """
    Retrieves a text embedding for a given document using the Amazon Titan Embedding model.

    This function sends the provided text to the Amazon Titan text embedding model
    and retrieves the resulting embedding. It handles retries for throttling exceptions
    and input size limitations by recursively calling itself with adjusted parameters.

    Parameters:
        doc_name (str): The name of the document, used for logging and error messages.
        text (str): The text content to be processed by the Titan embedding model.
        attempt (int): The current attempt number (used in recursive calls to handle retries). Defaults to 0.
        cutoff (int): The maximum number of words to include from the input text if a ValidationException occurs due to input size limits. Defaults to 5000.

    Returns:
        dict or None: The embedding response from the Titan model as a dictionary, or None if the operation fails or exceeds the retry limits.
    """
    
    retries = 5
    
    model_id = 'amazon.titan-embed-text-v1' 
    accept = 'application/json' 
    content_type = 'application/json'
    
    body = json.dumps({
        "inputText": text,
    })

    # Invoke model 
    response = bedrock_client.invoke_model(
        body=body, 
        modelId=model_id, 
        accept=accept, 
        contentType=content_type
    )
    
    # Print response
    response_body = json.loads(response['body'].read())
    return response_body.get('embedding')

# Main Chat
def ask_ds(message, history):
    global employee_type
    global division
    global authenticated
    global extractions
    
    if len(message) == 0:
        return

    if authenticated == False:
        
        if division == None:
            if message.lower().strip() in ['ime', 'peer disability', 'pas']:
                division = message.lower().strip().replace(' ', '_')
                return "[1] CSR\n[2] QA"
            else:
                return "Please select a valid choice."
        elif employee_type == None:
            if message.lower().strip() in ['csr', 'qa']:
                employee_type = message.lower().strip()
                authenticated = True
                EXTRACTIONS_PATH = EXTRACTIONS_PATH.replace('{employee_type}', employee_type).replace('{division}', division[:3])
                extractions = read_json_from_s3()
                return "Welcome to Ask Dane Street! Whether you're new to the team or just looking for some quick information, I'm here to guide you through our company's literature and platform. Simply ask your question, and I'll provide you with the most relevant information I can."
            else:
                return "Please select a valid choice."    

    question = message

    # RAG
    question_embedding = get_titan_embedding(bedrock_client, 'question', question)
    
    similar_documents = []
    for file, data in extractions.items():
        similarity = cosine_similarity(question_embedding, np.array(data['embedding']))
        similar_documents.append((file, similarity))

    similar_documents.sort(key=lambda x: x[1], reverse=False)
    top_docs = similar_documents[:5]

    similar_content = ''
    for file, _ in top_docs:
        similar_content += extractions[file]['content'] + '\n'

    # Invoke
    response = bedrock_client.invoke_model_with_response_stream(
        modelId="anthropic.claude-3-sonnet-20240229-v1:0",
        body=json.dumps(
            {
                "anthropic_version": "bedrock-2023-05-31",
                "max_tokens": 4096,
                "system": f"""Here is some relevant information that may help answer the user's upcoming question:

                <relevant_information>
                {similar_content}
                </relevant_information>
                
                The user's question is:
                <question>{question}</question>

                Please carefully review the relevant information provided above. 

                Your task is to review the provided relevant information and answer the user's question to the best of your ability. 
                Aim to use information from the relevant information section to directly address the question asked, and refrain from saying 
                things like 'According to the relevant information provided'.

                Format your output nicely with sentences that are not too long, in a professional and kind tone. You should prefer lists or 
                bullet points when applicable. Begin by thanking the user for their question, and at the end of your answer, say "Thank you for using Ask Dane Street!"
                Remember, aim to only use information from the relevant information section in your response, without explicitly referring 
                to that section. Return your answer immediately and without preamble. 
                </Task>
                
                <Relevant Information>
                {similar_content}
                </Relevant Information>""",
                "messages": [
                        {
                        "role": "user",
                        "content": [
                            {
                                "type": "text",
                                "text": message
                            }
                        ]
                    }
                ],
            }
        ),
    )

    # Stream the response
    all_text = ''
    stream = response.get('body')
    if stream:
        for event in stream:
            chunk = event.get('chunk')
            if chunk and json.loads(chunk.get('bytes').decode()):
                # check if delta is present
                try:
                    this_text = json.loads(chunk.get('bytes').decode()).get('delta').get('text')
                    all_text += this_text
                    yield all_text  # Stream the text back to the UI
                except:
                    pass

    # Print relevant files
    output = '\n\nCheck out the following documents for more information:\n'
    for file, sim in top_docs:
        output += f"\n{file.replace('.txt', '.pdf')}"

    yield all_text + output


# Create necessary services and collect data
bedrock_client = create_bedrock_client()
s3_client = create_s3_client()

demo = gr.ChatInterface(fn=ask_ds, title="Ask DS", multimodal=False, chatbot=gr.Chatbot(value=[(None, "Select your division:\n[1] IME \n[2] PAS\n[3] Peer Disability")],),theme=theme)
demo.launch()