import gradio as gr
from huggingface_hub import InferenceClient
from sentence_transformers import SentenceTransformer
import torch
import numpy as np
import requests

#LLM we are using
client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")

#adding text file
with open("be_a_better_you.txt", "r", encoding="utf-8") as file1, open("journal_prompts.txt", "r", encoding="utf-8") as file2, open("workout.txt", "r", encoding="utf-8") as file3:
    wellness_text = file1.read() + "\n" + file2.read() +"\n" + file3.read()

#cleaning up the text
cleaned_text = wellness_text.strip()
chunks = cleaned_text.split("\n")
cleaned_chunks = []

#putting text in chunks
for chunk in chunks:
    stripped_chunk = chunk.strip()
    if stripped_chunk:
        cleaned_chunks.append(stripped_chunk)

#import model for embeddings
model = SentenceTransformer('all-MiniLM-L6-v2')

chunk_embeddings = model.encode(cleaned_chunks, convert_to_tensor=True)

def get_top_chunks(query):
  # creating a function taking query as my parameter
    query_embedding = model.encode(query, convert_to_tensor=True)
  # encode query to vector embedding for comparison
    query_embedding_normalized = query_embedding / query_embedding.norm()
  # normalize query to 1: allows for comparison of meaning
    chunk_embeddings_normalized = chunk_embeddings / chunk_embeddings.norm(dim=1, keepdim=True)
  # normalizing chunks for comparison of meaning
    similarities = torch.matmul(chunk_embeddings_normalized, query_embedding_normalized)
  # using matmul (matrix multiplication) method to compare query to chunks
    top_indices = torch.topk(similarities, k=3).indices
  # get the indices of the chunks thart are most similar to my query

    top_chunks = []

    for i in top_indices:
        chunk = cleaned_chunks[i]
    # for each index number in top_indices, get back the text
        top_chunks.append(chunk)

    return top_chunks

def get_nutrition_info(food_query):
    url = "https://trackapi.nutritionix.com/v2/natural/nutrients"
    headers = {
        "x-app-id": "5e0843",
        "x-app-key": "00948483e0e3540ea8dc5297ad8216d9",
        "Content-Type": "application/json"
    }
    body = {
        "query": food_query
    }

    response = requests.post(url, headers=headers, json=body)
    if response.status_code == 200:
        data = response.json()
        return data
    else:
        return None


def get_motivational_quote():
    url = "https://zenquotes.io/api/random"
    response = requests.get(url)
    if response.status_code == 200:
        quote = response.json()['q'] + " -" + response.json()[0]['a']
        return quote
    else:
        return "Keep going! You're doing great."



def respond(message, history):
    messages = [{
        "role": "system", 
        "content": "You are a big sister chatbot named Nessie. You help people feel better in a simple manner. Always reply in 3 sentences or less. Do NOT stop mid-sentence, as you may confuse them even more. When the user is asking for ideas give 5 max, as you don't want to overwhelm them"
    }]
    # change the personality
    context = get_top_chunks(message)
    
    if history:
        messages.extend(history)
    messages.append({"role": "user", "content": message})

    user_context = f"{message}\nInformation: {context}"
    messages.append({"role": "user", "content": user_context})

    response = ""
    for messages in client.chat_completion(
        messages,
        max_tokens = 200,
        stream = True,
    ):
        token = messages.choices[0].delta.content
        response+= token 
        yield response

theme = gr.themes.Soft(
    primary_hue="rose",
    secondary_hue="zinc",
    neutral_hue="pink",
)

with gr.Blocks(theme=theme) as demo:
    chatbot = gr.ChatInterface(
        fn=respond,
        type='messages',
        title="Hi! I'm Nessie, your personal wellness assistant. What can I assist you with today?",
        examples=[
            "Can you help me with my dietary goals? I want to track my calories, macros, and get advice based on myself.",
            "Can you help me reach my fitness goals? I would like guidance and recommendations on workouts based on my goals.",
            "Can you give me some journal prompts? I want to start journaling to help myself reflect on my goals and have some daily affirmations. "
        ]
    )
    #chatbot = gr.ChatInterface(respond, type = 'messages', title= "Hi! I'm Nessie, your personal wellness assistant. What can I assist you with today?",examples=["Can I help you with your dietary goals? I can help you track your calories, macros, and give advice based on personal goals, height, and weight.","Can I help you with your physical health and help you reach your fitness goals? I can give guidance and recommendations for specific workouts based on personal goals.","If you are struggling, I am here. You are so beautiful and so loved! I'm here for whatever you need. "])
    demo.launch(debug=True)