main.py

from flask import Flask, request, jsonify, render_template, Response
import os
import requests
import json
from scipy import spatial
from flask_cors import CORS
import random
import numpy as np
from langchain_chroma import Chroma
from chromadb import Documents, EmbeddingFunction, Embeddings


app = Flask(__name__)
CORS(app)


class MyEmbeddingFunction(EmbeddingFunction):

    def embed_documents(self, input: Documents) -> Embeddings:
        for i in range(5):
            try:
                embeddings = []
                url = "https://api.deepinfra.com/v1/inference/BAAI/bge-large-en-v1.5"

                payload = json.dumps({
                    "inputs": input
                })
                headers = {
                    'Accept': 'application/json, text/plain, */*',
                    'Accept-Language': 'en-US,en;q=0.9,gu;q=0.8,ru;q=0.7,hi;q=0.6',
                    'Connection': 'keep-alive',
                    'Content-Type': 'application/json',
                    'Origin': 'https://deepinfra.com',
                    'Referer': 'https://deepinfra.com/',
                    'Sec-Fetch-Dest': 'empty',
                    'Sec-Fetch-Mode': 'cors',
                    'Sec-Fetch-Site': 'same-site',
                    'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/127.0.0.0 Safari/537.36',
                    'sec-ch-ua': '"Not)A;Brand";v="99", "Google Chrome";v="127", "Chromium";v="127"',
                    'sec-ch-ua-mobile': '?0',
                    'sec-ch-ua-platform': '"Windows"'
                }

                response = requests.request("POST", url, headers=headers, data=payload)
                return response.json()["embeddings"]
            except:
                pass

    def embed_query(self, input: Documents) -> Embeddings:
        print(input)
        for i in range(5):
            try:
                embeddings = []
                url = "https://api.deepinfra.com/v1/inference/BAAI/bge-large-en-v1.5"

                payload = json.dumps({
                    "inputs": [input]
                })
                headers = {
                    'Accept': 'application/json, text/plain, */*',
                    'Accept-Language': 'en-US,en;q=0.9,gu;q=0.8,ru;q=0.7,hi;q=0.6',
                    'Connection': 'keep-alive',
                    'Content-Type': 'application/json',
                    'Origin': 'https://deepinfra.com',
                    'Referer': 'https://deepinfra.com/',
                    'Sec-Fetch-Dest': 'empty',
                    'Sec-Fetch-Mode': 'cors',
                    'Sec-Fetch-Site': 'same-site',
                    'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/127.0.0.0 Safari/537.36',
                    'sec-ch-ua': '"Not)A;Brand";v="99", "Google Chrome";v="127", "Chromium";v="127"',
                    'sec-ch-ua-mobile': '?0',
                    'sec-ch-ua-platform': '"Windows"'
                }

                response = requests.request("POST", url, headers=headers, data=payload)
                return response.json()["embeddings"][0]
            except:
                pass

CHROMA_PATH = "chroma"
custom_embeddings = MyEmbeddingFunction()
db = Chroma(
    persist_directory=CHROMA_PATH, embedding_function=custom_embeddings
)


def embeddingGen(query):
    url = "https://api.deepinfra.com/v1/inference/BAAI/bge-large-en-v1.5"

    payload = json.dumps({
        "inputs": [query]
    })
    headers = {
        'Accept': 'application/json, text/plain, */*',
        'Accept-Language': 'en-US,en;q=0.9,gu;q=0.8,ru;q=0.7,hi;q=0.6',
        'Connection': 'keep-alive',
        'Content-Type': 'application/json',
        'Origin': 'https://deepinfra.com',
        'Referer': 'https://deepinfra.com/',
        'Sec-Fetch-Dest': 'empty',
        'Sec-Fetch-Mode': 'cors',
        'Sec-Fetch-Site': 'same-site',
        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/127.0.0.0 Safari/537.36',
        'sec-ch-ua': '"Not)A;Brand";v="99", "Google Chrome";v="127", "Chromium";v="127"',
        'sec-ch-ua-mobile': '?0',
        'sec-ch-ua-platform': '"Windows"'
    }

    response = requests.request("POST", url, headers=headers, data=payload)
    return response.json()


def strings_ranked_by_relatedness(query, df, top_n=5):
    def relatedness_fn(x, y):
        x_norm = np.linalg.norm(x)
        y_norm = np.linalg.norm(y)
        return np.dot(x, y) / (x_norm * y_norm)

    query_embedding_response = embeddingGen(query)
    query_embedding = query_embedding_response["embeddings"][0]
    strings_and_relatednesses = [
        (row["text"], relatedness_fn(query_embedding, row["embedding"])) for row in df
    ]
    strings_and_relatednesses.sort(key=lambda x: x[1], reverse=True)
    strings, relatednesses = zip(*strings_and_relatednesses)
    return strings[:top_n], relatednesses[:top_n]


@app.route("/api/gpt", methods=["POST"])
def gptRes():
    data = request.get_json()
    messages = data["messages"]
    def inference():
        url = "https://api.deepinfra.com/v1/openai/chat/completions"

        payload = json.dumps({
            "model": "meta-llama/Meta-Llama-3.1-8B-Instruct",
            "messages": messages,
            "stream": True,
            "max_tokens": 1024,
        })
        headers = {
            'Accept-Language': 'en-US,en;q=0.9,gu;q=0.8,ru;q=0.7,hi;q=0.6',
            'Connection': 'keep-alive',
            'Content-Type': 'application/json',
            'Origin': 'https://deepinfra.com',
            'Referer': 'https://deepinfra.com/',
            'Sec-Fetch-Dest': 'empty',
            'Sec-Fetch-Mode': 'cors',
            'Sec-Fetch-Site': 'same-site',
            'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/127.0.0.0 Safari/537.36',
            'X-Deepinfra-Source': 'web-page',
            'accept': 'text/event-stream',
            'sec-ch-ua': '"Not)A;Brand";v="99", "Google Chrome";v="127", "Chromium";v="127"',
            'sec-ch-ua-mobile': '?0',
            'sec-ch-ua-platform': '"Windows"'
        }

        response = requests.request("POST", url, headers=headers, data=payload, stream=True)

        for line in response.iter_lines(decode_unicode=True):
            if line:
                # try:
                #     line = line.split("data:")[1]
                #     line = json.loads(line)
                #     yield line["choices"][0]["delta"]["content"]
                # except:
                #     yield ""
                yield line

    return Response(inference(), content_type='text/event-stream')


@app.route("/", methods=["GET"])
def index():
    return render_template("index.html")


@app.route("/api/getAPI", methods=["POST"])
def getAPI():
    return jsonify({"API":  random.choice(apiKeys)})


@app.route("/api/getContext", methods=["POST"])
def getContext():
    global db
    question = request.form["question"]
    results = db.similarity_search_with_score(question, k=5)
    context = "\n\n---\n\n".join([doc.page_content for doc, _score in results])
    sources = [doc.metadata.get("id", None) for doc, _score in results]
    return jsonify({"context": context, "sources": sources})


@app.route("/api/audioGenerate", methods=["POST"])
def audioGenerate():
    answer = request.form["answer"]
    audio = []
    for i in answer.split("\n"):
        url = "https://deepgram.com/api/ttsAudioGeneration"

        payload = json.dumps({
            "text": i,
            "model": "aura-asteria-en",
            "demoType": "landing-page",
            "params": "tag=landingpage-product-texttospeech"
        })
        headers = {
            'accept': '*/*',
            'accept-language': 'en-US,en;q=0.9,gu;q=0.8,ru;q=0.7,hi;q=0.6',
            'content-type': 'application/json',
            'origin': 'https://deepgram.com',
            'priority': 'u=1, i',
            'referer': 'https://deepgram.com/',
            'sec-ch-ua': '"Not/A)Brand";v="8", "Chromium";v="126", "Google Chrome";v="126"',
            'sec-ch-ua-mobile': '?0',
            'sec-ch-ua-platform': '"Windows"',
            'sec-fetch-dest': 'empty',
            'sec-fetch-mode': 'cors',
            'sec-fetch-site': 'same-origin',
            'user-agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36'
        }

        response = requests.request("POST", url, headers=headers, data=payload)
        audio.append(response.json()["data"])
    return jsonify({"audio": audio})


if __name__ == "__main__":
    # app.run(debug=True)
    from waitress import serve

    serve(app, host="0.0.0.0", port=7860)