From 1fe36ab574bb8212d284acdf4df4fed8c83c2581 Mon Sep 17 00:00:00 2001
From: ZhymabekRoman <robanokssamit@yandex.ru>
Date: Fri, 7 Jun 2024 15:20:40 +0500
Subject: [PATCH] misc: anti bot

---
 server/middlewares/logger.py |  4 ++++
 server/utils/anti_bot.py     | 38 ++++++++++++++++++++++++++++++++++++
 2 files changed, 42 insertions(+)
 create mode 100644 server/utils/anti_bot.py

diff --git a/server/middlewares/logger.py b/server/middlewares/logger.py
index 1a68ce4..be744b5 100644
--- a/server/middlewares/logger.py
+++ b/server/middlewares/logger.py
@@ -9,6 +9,7 @@ from starlette.responses import Response, StreamingResponse
 from starlette.types import Message
 
 from server import transponder_code_correlation, url_correlation, xkcd_passwd, xp, config, home_page_process
+from server.utils.anti_bot import filter_bots
 from server.utils.notify import send_message
 from server.utils.error import generate_error
 from server.utils.utils import string_to_number_ascii
@@ -28,6 +29,9 @@ async def get_body(request: Request) -> bytes:
 
 class LoggerMiddleware(BaseHTTPMiddleware):
     async def dispatch(self, request: Request, call_next: Callable[[Request], Awaitable[StreamingResponse]]) -> Response:  # type: ignore
+        if filter_bots(request.headers.get("User-Agent")):
+            return Response(status_code=403)
+
         start_time = time.time()
         generated_id = xp.generate_xkcdpassword(xkcd_passwd, delimiter="-", numwords=3)
         transponder_code = string_to_number_ascii(generated_id)
diff --git a/server/utils/anti_bot.py b/server/utils/anti_bot.py
new file mode 100644
index 0000000..bf4c7b1
--- /dev/null
+++ b/server/utils/anti_bot.py
@@ -0,0 +1,38 @@
+from loguru import logger
+
+
+bot_signatures = {
+    "facebookexternalhit",
+    "/apple-touch-icon",
+    "Googlebot",
+    "AdsBot-Google",
+    "Google-Adwords-Instant",
+    "Apache-HttpClient",
+    "SafeDNSBot",
+    "RevueBot",
+    "MetaURI API",
+    "redback/v",
+    "Slackbot",
+    "HTTP_Request2/",
+    "python-requests/",
+    "LightspeedSystemsCrawler/",
+    "CipaCrawler/",
+    "Twitterbot/",
+    "Go-http-client/",
+    "/cheese_service",
+    "undici",
+    "database",
+    "node-fetch",
+    "Java/",
+    "axios/",
+    "okhttp/",
+    "go-resty/",
+    "Faraday v",
+    # "curl/"
+}
+
+
+def filter_bots(details: str) -> bool:
+    is_bot = any(bot_signature in details for bot_signature in bot_signatures)
+    logger.debug(f"{details} is a bot: {is_bot}")
+    return is_bot