Add sitemap

Author: koldakov

futuramaapi/apps/app.py

@@ -1,7 +1,7 @@
 import mimetypes
 from abc import ABC, abstractmethod
 from contextlib import asynccontextmanager
-from typing import TYPE_CHECKING, Self
+from typing import TYPE_CHECKING, Literal, Self
 
 import sentry_sdk
 from fastapi import FastAPI
@@ -25,6 +25,19 @@
 mimetypes.add_type("image/webp", ".webp")
 
 
+BOTS_FORBIDDEN_URLS: tuple[str, ...] = (
+    "/favicon.ico",
+    "/openapi.json",
+    "/robots.txt",
+    "/sitemap.xml",
+    "/static",
+    "/health",
+    "/logout",
+    "/api/",
+    "/s/",
+)
+
+
 class BaseAPI(ABC):
     version: str = __version__
 
@@ -70,6 +83,27 @@ def get_app(
     def build(self) -> None: ...
 
 
+def _is_route_public(
+    url: BaseRoute,
+    /,
+    *,
+    allowed_methods: list[Literal["GET", "POST", "HEAD", "PUT"]] | None = None,
+) -> bool:
+    if url.path.startswith(BOTS_FORBIDDEN_URLS):
+        return False
+
+    if allowed_methods is None:
+        allowed_methods = ["GET"]
+
+    if not hasattr(url, "methods"):
+        return True
+
+    if not any(x in allowed_methods for x in url.methods):
+        return False
+
+    return True
+
+
 class FuturamaAPI(BaseAPI):
     def get_app(
         self,
@@ -118,6 +152,15 @@ def build(self) -> None:
     def urls(self) -> list[BaseRoute]:
         return self.app.routes
 
+    @property
+    def public_urls(self) -> list[BaseRoute]:
+        urls: list[BaseRoute] = []
+        for route in self.app.routes:
+            if _is_route_public(route) and route.path not in [u.path for u in urls]:
+                urls.append(route)
+
+        return urls
+
 
 @asynccontextmanager
 async def _lifespan(_: FastAPI):

futuramaapi/routers/root/api.py

@@ -11,7 +11,7 @@
 from futuramaapi.routers.users.dependencies import cookie_user_from_form_data, user_from_cookies
 from futuramaapi.routers.users.schemas import Link, User
 
-from .schemas import About, Root, UserAuth
+from .schemas import About, Root, SiteMap, UserAuth
 
 router = APIRouter()
 
@@ -178,3 +178,14 @@ async def user_link_redirect(
         counter=add(link.counter, 1),
     )
     return RedirectResponse(link.url)
+
+
+@router.get(
+    "/sitemap.xml",
+    include_in_schema=False,
+)
+async def get_sitemap(
+    request: Request,
+) -> Response:
+    obj: SiteMap = await SiteMap.from_request(request)
+    return await obj.get_response()

futuramaapi/routers/root/schemas.py

@@ -1,8 +1,11 @@
 from typing import ClassVar, Self
 
+from fastapi import Response
+from pydantic import HttpUrl
 from sqlalchemy.ext.asyncio import AsyncSession
 from starlette.requests import Request
 
+from futuramaapi.core import settings
 from futuramaapi.helpers.pydantic import BaseModel, Field
 from futuramaapi.mixins.pydantic import BaseModelTemplateMixin
 from futuramaapi.repositories.base import FilterStatementKwargs
@@ -46,3 +49,34 @@ class UserAuth(BaseModel, BaseModelTemplateMixin):
     @classmethod
     async def from_request(cls, session: AsyncSession, request: Request, /) -> Self:
         return cls()
+
+
+class SiteMap(BaseModel):
+    _header: ClassVar[str] = '<?xml version="1.0" encoding="UTF-8"?>'
+    _media_type: ClassVar[str] = "application/xml"
+    _url_tag: ClassVar[str] = """<url><loc>%s</loc></url>"""
+    _url_set_tag: ClassVar[str] = """<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">%s</urlset>"""
+
+    base_url: HttpUrl = settings.build_url(is_static=False)
+    urls: list[str]
+
+    async def get_response(self) -> Response:
+        urls: str = ""
+        for url in self.urls:
+            urls += self._url_tag % settings.build_url(
+                path=url,
+                is_static=False,
+            )
+        url_set: str = self._url_set_tag % urls
+        return Response(
+            content=f"""{self._header}{url_set}""",
+            media_type=self._media_type,
+        )
+
+    @classmethod
+    async def from_request(cls, request: Request) -> Self:
+        from futuramaapi.apps.app import futurama_api
+
+        return cls(
+            urls=[url.path for url in futurama_api.public_urls],
+        )