local-llm-server/llm_server/routes/v1/proxy.py

import time
from datetime import datetime

from flask import jsonify, request

from llm_server import opts
from llm_server.routes.v1.generate import concurrent_semaphore
from . import bp
from .. import stats
from ..cache import cache
from ..helpers.http import cache_control
from ..stats import SemaphoreCheckerThread
from ...llm.info import get_running_model


@bp.route('/stats', methods=['GET'])
@cache.cached(timeout=5, query_string=True)
def get_stats():
    model_list = get_running_model()  # will return False when the fetch fails
    if isinstance(model_list, bool):
        online = False
    else:
        online = True

    return jsonify({
        'stats': {
            'proompters_now': opts.concurrent_gens - concurrent_semaphore._value,
            'proompters_1_min': SemaphoreCheckerThread.proompters_1_min,
            'total_proompts': stats.get_count(),
            'uptime': int((datetime.now() - stats.start_time).total_seconds()),
        },
        'online': online,
        'mode': opts.mode,
        'model': get_running_model(),
        'endpoints': {
            'blocking': f'https://{request.headers.get("Host")}/{opts.frontend_api_client.strip("/")}',
        },
        'timestamp': int(time.time()),
        'openaiKeys': '∞',
        'anthropicKeys': '∞',
        'config': {
            'gatekeeper': 'none' if opts.auth_required is False else 'token',
            'context_size': opts.context_size,
        }
    }), 200