app.py

import random
import time
import json
import re
import aiohttp
import requests
import asyncio
from flask import Flask, request, Response
from flask_cors import CORS
import uuid

app = Flask(__name__)
CORS(app)

MAGAI_TOKEN = {
    "cookie": "_fbp=fb.1.1731157547017.999175969945264246; _ga=GA1.1.1511565936.1731157547; soquick-mobile_live_u2main=bus|1731157558019x939242898939773400|1731157558072x230077743876915040; soquick-mobile_live_u2main.sig=O-AoJ9Vd_oxtGTWtsC31B_SsLlU; soquick-mobile_u1main=1731157558019x939242898939773400; _ga_N5J29RVHDJ=GS1.1.1731157547.1.1.1731157568.0.0.0",
    "app_last_change": "21388518093",
    "current_page_item": "1348695171700984260__LOOKUP__1726124636560x692535552825360400",
    "current_user": "1348695171700984260__LOOKUP__1722349236461x685414888067651300",
}

MAGAI_MAPPING = {
    "gpt-4o": "openai/gpt-4o",
    "claude-3.5-sonnet": "anthropic/claude-3.5-sonnet:beta",
    "claude-3-opus": "anthropic/claude-3-opus:beta",
    "gemini-1.5-pro": "google/gemini-pro-1.5"
}

UUID_LENGTH = 1e18
MODULO = 1e18

def generate_uuid():
    return f"{int(time.time() * 1000)}x{str(round(random.random() * UUID_LENGTH)).zfill(18)}"

def create_luid(separator="x"):
    timestamp = int(time.time() * 1000)
    return f"{timestamp}{separator}1"

def format_model_name(model_name):
    return re.sub(r"_+", "_", re.sub(r"[/:-]", "_", model_name))


def find_token_in_object(obj):
    if isinstance(obj, dict):
        for key, value in obj.items():
            if key == "token" and isinstance(value, str):
                return value
            token = find_token_in_object(value)
            if token:
                return token
    return None


def get_last_user_content(messages):
    for message in reversed(messages):
        if message["role"] == "user":
            return message["content"]
    return None


async def get_token(model, message):
    server_call_id = generate_uuid()
    created_id = MAGAI_TOKEN["current_page_item"].split("__")[0]
    user_id = MAGAI_TOKEN["current_user"].split("__")[2]
    model_id = "0060f9accd1dbade552f65ac646fb3da"
    item_id = "bUNih7"
    element_id = "bUNib7"

    body = {
        "app_last_change": MAGAI_TOKEN["app_last_change"],
        "calls": [
            {
                "client_state": {
                    "element_instances": {
                        "bUNib7": {
                            "dehydrated": f"{created_id}__LOOKUP__ElementInstance::bUNib7",
                            "parent_element_id": "bUMiq3",
                        },
                        "bTezP": {
                            "dehydrated": f"{created_id}__LOOKUP__ElementInstance::bTezP",
                            "parent_element_id": "bTezJ",
                        },
                        "bTezE": {
                            "dehydrated": f"{created_id}__LOOKUP__ElementInstance::bTezE",
                            "parent_element_id": "bTeqc",
                        },
                        "bTezJ": {
                            "dehydrated": f"{created_id}__LOOKUP__ElementInstance::bTezJ",
                            "parent_element_id": "bUKFL2",
                        },
                        "bTezQ": {
                            "dehydrated": f"{created_id}__LOOKUP__ElementInstance::bTezQ",
                            "parent_element_id": "bUKFL2",
                        },
                        "bUiru0": {
                            "dehydrated": f"{created_id}__LOOKUP__ElementInstance::bUiru0",
                            "parent_element_id": "bUjNK",
                        },
                        "bUDVj0": {
                            "dehydrated": f"{created_id}__LOOKUP__ElementInstance::bUDVj0",
                            "parent_element_id": "bUMiq3",
                        },
                        "bUXzm2": {
                            "dehydrated": f"{created_id}__LOOKUP__ElementInstance::bUXzm2",
                            "parent_element_id": "bUMhk3",
                        },
                        "bUifI1": {
                            "dehydrated": f"{created_id}__LOOKUP__ElementInstance::bUifI1",
                            "parent_element_id": "bTeqg",
                        },
                        "bUMiq3": {
                            "dehydrated": f"{created_id}__LOOKUP__ElementInstance::bUMiq3",
                            "parent_element_id": "bTezE",
                        },
                        "bTekm": {
                            "dehydrated": f"{created_id}__LOOKUP__ElementInstance::bTekm",
                            "parent_element_id": None,
                        },
                    },
                    "element_state": {
                        f"{created_id}__LOOKUP__ElementInstance::bTezP": {
                            "is_visible": True,
                            "value_that_is_valid": message,
                            "value": message,
                        },
                        f"{created_id}__LOOKUP__ElementInstance::bTezE": {
                            "custom.images_": None,
                            "custom.file_": None,
                            "custom.file_content_": None,
                            "custom.file_name_": None,
                            "custom.file_type_": None,
                        },
                        f"{created_id}__LOOKUP__ElementInstance::bTezJ": {
                            "custom.isrecording_": None,
                            "custom.prompt_": None,
                        },
                        f"{created_id}__LOOKUP__ElementInstance::bUiru0": {
                            "AAE": message
                        },
                        f"{created_id}__LOOKUP__ElementInstance::bUDVj0": {
                            "AAE": message
                        },
                        f"{created_id}__LOOKUP__ElementInstance::bUifI1": {
                            "custom.is_visible_": None,
                            "group_data": None,
                        },
                        f"{created_id}__LOOKUP__ElementInstance::bUMiq3": {
                            "group_data": None
                        },
                    },
                    "other_data": {
                        "Current Page Scroll Position": 0,
                        "Current Page Width": 661,
                    },
                    "cache": {
                        f"{model_id}": format_model_name(model),
                        "true": True,
                        "CurrentPageItem": MAGAI_TOKEN["current_page_item"],
                        "CurrentUser": MAGAI_TOKEN["current_user"],
                    },
                    "exists": {
                        f"{model_id}": True,
                        "true": True,
                        "CurrentPageItem": True,
                        "CurrentUser": True,
                    },
                },
                "run_id": generate_uuid(),
                "server_call_id": server_call_id,
                "item_id": item_id,
                "element_id": element_id,
                "page_id": "bTekm",
                "uid_generator": {
                    "timestamp": int(time.time() * 1000),
                    "seed": round(random.random() * UUID_LENGTH) % MODULO,
                },
                "random_seed": random.random(),
                "current_date_time": int(time.time() * 1000),
                "current_wf_params": {},
            }
        ],
        "client_breaking_revision": 5,
        "timezone_offset": -480,
        "timezone_string": "Asia/Shanghai",
        "user_id": user_id,
        "wait_for": [],
    }

    url = "https://app.magai.co/workflow/start"
    async with aiohttp.ClientSession() as session:
        async with session.post(
            url,
            headers={
                "x-bubble-fiber-id": generate_uuid(),
                "x-bubble-pl": create_luid(),
                "accept": "application/json, text/javascript, */*; q=0.01",
                "cookie": MAGAI_TOKEN["cookie"],
            },
            json=body,
        ) as response:
            response_data = await response.json()

    if "error_class" in response_data:
        raise Exception(response_data)

    server_call_data = response_data.get(server_call_id)
    if not server_call_data or "step_results" not in server_call_data:
        return None
    for step_result in server_call_data["step_results"].values():
        if isinstance(step_result.get("return_value"), dict):
            token = find_token_in_object(step_result["return_value"])
            if token:
                return token


async def get_request_data(model, messages):
    if model not in MAGAI_MAPPING:
        return Response(
            json.dumps(
                {
                    "error": {
                        "message": "This model is currently unavailable. Please try again later or choose another model.",
                        "code": "model_not_exists",
                    }
                }
            ),
            status=400,
            mimetype="application/json",
        )

    last_user_message = get_last_user_content(messages)
    token = await get_token(MAGAI_MAPPING[model], last_user_message)
    headers = {
        "Content-Type": "application/json",
        "HTTP-Referer": "https://magai.co",
        "Origin": "https://app.magai.co",
        "Pragma": "no-cache",
        "Referer": "https://app.magai.co/",
        "Token": token,
        "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/500.00 (KHTML, like Gecko) Chrome/100.0.0.0 Safari/500.00",
    }

    json_data = {
        "model": MAGAI_MAPPING[model],
        "messages": [{"role": "system", "content": "You are a helpful assistant."}]
        + messages,
        "tools": [
            {
                "type": "function",
                "function": {
                    "name": "get_actual_time_info",
                    "description": "Returns actual information from web about prompt theme.",
                    "parameters": {
                        "type": "object",
                        "properties": {
                            "query": {
                                "type": "string",
                                "description": "The query string based on users prompt to search information about.",
                            }
                        },
                        "required": ["query"],
                    },
                },
            },
            {
                "type": "function",
                "function": {
                    "name": "generate_image",
                    "description": "Returns generated image URL.",
                    "parameters": {
                        "type": "object",
                        "properties": {
                            "query": {
                                "type": "string",
                                "description": "Prompt to image generation AI model, that describes what image to generate.",
                            }
                        },
                        "required": ["query"],
                    },
                },
            },
        ],
        "provider": {"data_collection": "deny"},
        "tool_choice": "auto",
        "stream": True,
    }

    response = requests.post(
        "https://live.proxy.magai.co:4430/opr/api/v1/chat/completions",
        headers=headers,
        json=json_data,
    )
    return response


def format_response(response):
    content = ""
    for line in response.iter_lines():
        if line:
            decoded_line = line.decode("utf-8")
            if decoded_line.startswith("data:"):
                try:
                    data = json.loads(decoded_line[5:].strip())
                    if "choices" in data and len(data["choices"]) > 0:
                        delta = data["choices"][0].get("delta", {})
                        if "content" in delta:
                            content += delta["content"]
                except json.JSONDecodeError:
                    pass
    return content


@app.route("/hf/v1/chat/completions", methods=["POST"])
def chat_completions():
    data = request.json
    messages = data.get("messages", [])
    model = data.get("model", "claude-3.5-sonnet")

    async def process_request():
        response = await get_request_data(model, messages)
        return format_response(response)

    loop = asyncio.new_event_loop()
    asyncio.set_event_loop(loop)
    result = loop.run_until_complete(process_request())

    event_stream_response = ""
    for part in result:
        part = part.replace("\n", "\\n")
        event_stream_response += f'data:{{"id":"{uuid.uuid4()}","object":"chat.completion.chunk","created":{int(time.time())},"model":"{model}","system_fingerprint":"fp_45ah8ld5a7","choices":[{{"index":0,"delta":{{"content":"{part}"}},"logprobs":null,"finish_reason":null}}]}}\n\n'
    event_stream_response += "data:[DONE]\n"

    return Response(event_stream_response, mimetype="text/event-stream")


if __name__ == "__main__":
    app.run(host="0.0.0.0", port=7860)