add graylog config (#23)

* add additional chat gpt request error

* add graylog config

README.md

@@ -65,9 +65,9 @@ methods:
 ## Chat:
 
 ```shell
-cd bot_microservice
+docker run -p 8858:8858 -it --name freegpt --rm -e CHAT_PATH=/chat balshdocker/freegpt:latest
-python3 run.py
 ```
+Open http://localhost:8858/chat/
 
 
 ```bash
@@ -85,11 +85,13 @@ gunicorn main:create_app --workers 10 --bind 0.0.0.0:8083 --worker-class uvicorn
 
 ### Run local tests:
 ```bash
-poetry run pytest
+cd bot_microservice
+STAGE=runtests poetry run pytest
 ```
 
 ### Run tests in docker compose:
 ```bash
+cd bot_microservice
 STAGE=runtests docker compose run bot bash -c "coverage run -m pytest -vv --exitfirst && poetry run coverage report"
 ```
 
@@ -101,14 +103,17 @@ Docs can be found at
 
 on local start can be found at http://localhost/gpt/api/docs
 
+prod docs https://bot.mywistr.ru/gpt/api/docs/
+
 ## Help article
 
-[Пишем асинхронного Телеграм-бота](https://habr.com/ru/company/kts/blog/598575/)
+[Следить за обновлениями этого репозитория](https://github.com/fantasy-peak/cpp-freegpt-webui)
 
 
 ## TODO
 
-- [x] Добавить очередь сообщений
+- [] Добавить базу данных с моделями
-- [x] Исправить запуск локально
+- [] Добавить миграции через alembic
-- [x] Добавить тестов
+- [] Добавить веса моделей и их смену
-- [x] Close connection
+- [] Добавить тестов
+- [] Добавить сентри

bot_microservice/api/bot/controllers.py

@@ -2,8 +2,6 @@ from fastapi import APIRouter, Request
 from starlette import status
 from starlette.responses import Response
 
-from constants import INVALID_GPT_MODEL_MESSAGE
-from core.utils import ChatGptService
 from settings.config import settings
 
 router = APIRouter()
@@ -19,27 +17,3 @@ router = APIRouter()
 )
 async def process_bot_updates(request: Request) -> None:
     await request.app.state.queue.put_updates_on_queue(request)
-
-
-@router.get(
-    "/bot-healthcheck",
-    name="bot:gpt_healthcheck",
-    response_class=Response,
-    summary="bot healthcheck",
-    responses={
-        status.HTTP_500_INTERNAL_SERVER_ERROR: {"description": "Request to chat gpt not success"},
-        status.HTTP_200_OK: {"description": "Successful Response"},
-    },
-)
-async def gpt_healthcheck(response: Response) -> Response:
-    chatgpt_service = ChatGptService(chat_gpt_model=settings.GPT_MODEL)
-    data = chatgpt_service.build_request_data("Привет!")
-    response.status_code = status.HTTP_200_OK
-    try:
-        chatgpt_response = await chatgpt_service.do_request(data)
-        if chatgpt_response.status_code != status.HTTP_200_OK or chatgpt_response.text == INVALID_GPT_MODEL_MESSAGE:
-            response.status_code = status.HTTP_500_INTERNAL_SERVER_ERROR
-    except Exception:
-        response.status_code = status.HTTP_500_INTERNAL_SERVER_ERROR
-
-    return Response(status_code=response.status_code, content=None)

bot_microservice/api/system/controllers.py

@@ -1,6 +1,11 @@
 from fastapi import APIRouter
 from fastapi.responses import ORJSONResponse
 from starlette import status
+from starlette.responses import Response
+
+from constants import INVALID_GPT_REQUEST_MESSAGES
+from core.utils import ChatGptService
+from settings.config import settings
 
 router = APIRouter()
 
@@ -13,3 +18,30 @@ router = APIRouter()
 )
 async def healthcheck() -> ORJSONResponse:
     return ORJSONResponse(content=None, status_code=status.HTTP_200_OK)
+
+
+@router.get(
+    "/bot-healthcheck",
+    name="system:gpt_healthcheck",
+    response_class=Response,
+    summary="Проверяет доступность моделей и если они недоступны, то возвращает код ответа 500",
+    responses={
+        status.HTTP_500_INTERNAL_SERVER_ERROR: {"description": "Request to chat gpt not success"},
+        status.HTTP_200_OK: {"description": "Successful Response"},
+    },
+)
+async def gpt_healthcheck(response: Response) -> Response:
+    chatgpt_service = ChatGptService(chat_gpt_model=settings.GPT_MODEL)
+    data = chatgpt_service.build_request_data("Привет!")
+    response.status_code = status.HTTP_200_OK
+    try:
+        chatgpt_response = await chatgpt_service.do_request(data)
+        if chatgpt_response.status_code != status.HTTP_200_OK:
+            response.status_code = status.HTTP_500_INTERNAL_SERVER_ERROR
+        for message in INVALID_GPT_REQUEST_MESSAGES:
+            if message in chatgpt_response.text:
+                response.status_code = status.HTTP_500_INTERNAL_SERVER_ERROR
+    except Exception:
+        response.status_code = status.HTTP_500_INTERNAL_SERVER_ERROR
+
+    return Response(status_code=response.status_code, content=None)

bot_microservice/constants.py

@@ -4,7 +4,7 @@ AUDIO_SEGMENT_DURATION = 120 * 1000
 
 API_PREFIX = "/api"
 CHAT_GPT_BASE_URI = "/backend-api/v2/conversation"
-INVALID_GPT_MODEL_MESSAGE = "Invalid request model"
+INVALID_GPT_REQUEST_MESSAGES = ("Invalid request model", "return unexpected http status code")
 
 
 class BotStagesEnum(StrEnum):

bot_microservice/core/logging.py

@@ -3,6 +3,7 @@ import sys
 from types import FrameType
 from typing import TYPE_CHECKING, Any, cast
 
+import graypy
 from loguru import logger
 from sentry_sdk.integrations.logging import EventHandler
 
@@ -40,20 +41,35 @@ def configure_logging(*, level: LogLevelEnum, enable_json_logs: bool, enable_sen
 
     intercept_handler = InterceptHandler()
 
-    logging.basicConfig(handlers=[intercept_handler], level=logging_level)
-
     formatter = _json_formatter if enable_json_logs else _text_formatter
-    logger.configure(
+
-        handlers=[
+    base_config_handlers = [intercept_handler]
+
+    loguru_handlers = [
+        {
+            "sink": sys.stdout,
+            "level": logging_level,
+            "serialize": enable_json_logs,
+            "format": formatter,
+            "colorize": True,
+        }
+    ]
+
+    if settings.GRAYLOG_HOST:
+        graylog_handler = graypy.GELFTCPHandler(settings.GRAYLOG_HOST, 12201)
+        base_config_handlers.append(graylog_handler)
+        loguru_handlers.append(
             {
-                "sink": sys.stdout,
+                "sink": graylog_handler,
                 "level": logging_level,
                 "serialize": enable_json_logs,
                 "format": formatter,
-                "colorize": True,
+                "colorize": False,
             }
-        ],
+        )
-    )
+
+    logging.basicConfig(handlers=base_config_handlers, level=logging_level)
+    logger.configure(handlers=loguru_handlers)
 
     # sentry sdk не умеет из коробки работать с loguru, нужно добавлять хандлер
     # https://github.com/getsentry/sentry-python/issues/653#issuecomment-788854865

bot_microservice/core/utils.py

@@ -20,7 +20,7 @@ from speech_recognition import (
 from constants import (
     AUDIO_SEGMENT_DURATION,
     CHAT_GPT_BASE_URI,
-    INVALID_GPT_MODEL_MESSAGE,
+    INVALID_GPT_REQUEST_MESSAGES,
 )
 from settings.config import settings
 
@@ -124,10 +124,11 @@ class ChatGptService:
         try:
             response = await self.do_request(chat_gpt_request)
             status = response.status_code
-            if response.text == INVALID_GPT_MODEL_MESSAGE:
+            for message in INVALID_GPT_REQUEST_MESSAGES:
-                message = f"{INVALID_GPT_MODEL_MESSAGE}: {settings.GPT_MODEL}"
+                if message in response.text:
-                logger.info(message, data=chat_gpt_request)
+                    message = f"{message}: {settings.GPT_MODEL}"
-                return message
+                    logger.info(message, data=chat_gpt_request)
+                    return message
             if status != httpx.codes.OK:
                 logger.info(f"got response status: {status} from chat api", data=chat_gpt_request)
                 return "Что-то пошло не так, попробуйте еще раз или обратитесь к администратору"

bot_microservice/settings/.env.template

@@ -6,6 +6,8 @@ APP_PORT="8000"
 # SENTRY_DSN=
 SENTRY_TRACES_SAMPLE_RATE="0.95"
 
+# GRAYLOG_HOST=
+
 USER="web"
 TZ="Europe/Moscow"
 

bot_microservice/settings/config.py

@@ -48,8 +48,11 @@ class AppSettings(SentrySettings, BaseSettings):
     DOMAIN: str = "https://localhost"
     URL_PREFIX: str = ""
 
+    GRAYLOG_HOST: str | None = None
+
     GPT_MODEL: str = "gpt-3.5-turbo-stream-AItianhuSpace"
     GPT_BASE_HOST: str = "http://chat_service:8858"
+
     # quantity of workers for uvicorn
     WORKERS_COUNT: int = 1
     # Enable uvicorn reloading

bot_microservice/tests/integration/test_system.py

@@ -32,13 +32,13 @@ async def test_bot_healthcheck_is_ok(
         assert response.status_code == httpx.codes.OK
 
 
+@pytest.mark.parametrize("text", ["Invalid request model", "return unexpected http status code"])
 async def test_bot_healthcheck_invalid_request_model(
-    rest_client: AsyncClient,
+    rest_client: AsyncClient, test_settings: AppSettings, text: str
-    test_settings: AppSettings,
 ) -> None:
     with mocked_ask_question_api(
         host=test_settings.GPT_BASE_HOST,
-        return_value=Response(status_code=httpx.codes.OK, text="Invalid request model"),
+        return_value=Response(status_code=httpx.codes.OK, text=text),
     ):
         response = await rest_client.get("/api/bot-healthcheck")
         assert response.status_code == httpx.codes.INTERNAL_SERVER_ERROR

docker-compose.yml

@@ -19,11 +19,12 @@ services:
       dockerfile: deploy/Dockerfile
       target: bot-service
       args:
-        STAGE: ${STAGE}
+        STAGE: ${STAGE:-production}
     restart: unless-stopped
     env_file:
       - bot_microservice/settings/.env
     volumes:
+      - ./bot_microservice/settings/.env:/app/settings/.env:ro
       - /etc/localtime:/etc/localtime:ro
     networks:
       chat-gpt-network:

poetry.lock

@@ -999,6 +999,21 @@ gitdb = ">=4.0.1,<5"
 [package.extras]
 test = ["black", "coverage[toml]", "ddt (>=1.1.1,!=1.4.3)", "mypy", "pre-commit", "pytest", "pytest-cov", "pytest-sugar"]
 
+[[package]]
+name = "graypy"
+version = "2.1.0"
+description = "Python logging handlers that send messages in the Graylog Extended Log Format (GELF)."
+optional = false
+python-versions = "*"
+files = [
+    {file = "graypy-2.1.0-py2.py3-none-any.whl", hash = "sha256:5df0102ed52fdaa24dd579bc1e4904480c2c9bbb98917a0b3241ecf510c94207"},
+    {file = "graypy-2.1.0.tar.gz", hash = "sha256:fd8dc4a721de1278576d92db10ac015e99b4e480cf1b18892e79429fd9236e16"},
+]
+
+[package.extras]
+amqp = ["amqplib (==1.0.2)"]
+docs = ["sphinx (>=2.1.2,<3.0.0)", "sphinx-autodoc-typehints (>=1.6.0,<2.0.0)", "sphinx-rtd-theme (>=0.4.3,<1.0.0)"]
+
 [[package]]
 name = "greenlet"
 version = "2.0.2"
@@ -3110,4 +3125,4 @@ dev = ["doc8", "flake8", "flake8-import-order", "rstcheck[sphinx]", "sphinx"]
 [metadata]
 lock-version = "2.0"
 python-versions = "^3.11"
-content-hash = "f8faa71d22eb911772b7607eb35d2feb1e5dbe0b0bf2c602373b1e31bffaf820"
+content-hash = "ab644b9882ee200392911afc6b71bf87fdb413e4fdd9f06a460ce33da98687d7"

pyproject.toml

@@ -22,6 +22,7 @@ sentry-sdk = "^1.31.0"
 SpeechRecognition = "^3.8"
 pydub = "^0.25"
 greenlet = "^2.0.2"
+graypy = "^2.1.0"
 
 
 [tool.poetry.dev-dependencies]