Use cache-busting URLs for proxied files (#294)

Migrates (in a backwards-compatible way) from `/proxy/identity_image/271/` to `/proxy/identity_image/271/f5d8e72f2b/`.
dently).

activities/models/emoji.py

@@ -169,7 +169,9 @@ class Emoji(StatorModel):
             if self.file:
                 return AutoAbsoluteUrl(self.file.url)
             elif self.remote_url:
-                return AutoAbsoluteUrl(f"/proxy/emoji/{self.pk}/")
+                return AutoAbsoluteUrl(
+                    f"/proxy/emoji/{self.pk}/", hash_tail_input=self.remote_url
+                )
         return StaticAbsoluteUrl("img/blank-emoji-128.png")
 
     def as_html(self):

activities/models/post_attachment.py

@@ -81,13 +81,17 @@ class PostAttachment(StatorModel):
         elif self.file:
             return RelativeAbsoluteUrl(self.file.url)
         else:
-            return AutoAbsoluteUrl(f"/proxy/post_attachment/{self.pk}/")
+            return AutoAbsoluteUrl(
+                f"/proxy/post_attachment/{self.pk}/", hash_tail_input=self.remote_url
+            )
 
     def full_url(self):
         if self.file:
             return RelativeAbsoluteUrl(self.file.url)
         else:
-            return AutoAbsoluteUrl(f"/proxy/post_attachment/{self.pk}/")
+            return AutoAbsoluteUrl(
+                f"/proxy/post_attachment/{self.pk}/", hash_tail_input=self.remote_url
+            )
 
     ### ActivityPub ###
 

core/uris.py

@@ -1,3 +1,4 @@
+import hashlib
 import sys
 from urllib.parse import urljoin
 
@@ -27,8 +28,19 @@ class AutoAbsoluteUrl(RelativeAbsoluteUrl):
     or a passed identity's URI domain.
     """
 
-    def __init__(self, relative: str, identity=None):
+    def __init__(
+        self,
+        relative: str,
+        identity=None,
+        hash_tail_input: str | None = None,
+        hash_tail_length: int = 10,
+    ):
         self.relative = relative
+        if hash_tail_input:
+            # When provided, attach a hash of the input (typically the proxied URL)
+            # SHA1 chosen as it generally has the best performance in modern python, and security is not a concern
+            # Hash truncation is generally fine, as in the typical use case the hash is scoped to the identity PK
+            self.relative += f"{hashlib.sha1(hash_tail_input.encode('ascii')).hexdigest()[:hash_tail_length]}/"
         if identity:
             absolute_prefix = f"https://{identity.domain.uri_domain}/"
         else:

docker/nginx.conf

@@ -104,8 +104,6 @@ http {
             proxy_cache_valid 301 307 12h;
             proxy_cache_valid 500 502 503 504 0s;
             proxy_cache_valid any 72h;
-            proxy_hide_header Cache-Control;
-            add_header Cache-Control "public, max-age=3600";
             add_header X-Cache $upstream_cache_status;
         }
 

mediaproxy/views.py

@@ -31,7 +31,7 @@ class BaseProxyView(View):
                 headers={
                     "X-Accel-Redirect": "/__takahe_accel__/",
                     "X-Takahe-RealUri": remote_url,
-                    "Cache-Control": "public, max-age=3600",
+                    "Cache-Control": "public",
                 },
             )
         else:
@@ -52,7 +52,9 @@ class BaseProxyView(View):
                     "Content-Type": remote_response.headers.get(
                         "Content-Type", "application/octet-stream"
                     ),
-                    "Cache-Control": "public, max-age=3600",
+                    "Cache-Control": remote_response.headers.get(
+                        "Cache-Control", "public, max-age=3600"
+                    ),
                 },
             )
 

takahe/urls.py

@@ -211,23 +211,23 @@ urlpatterns = [
     path("debug/500/", debug.ServerError.as_view()),
     path("debug/oauth_authorize/", debug.OauthAuthorize.as_view()),
     # Media/image proxy
-    path(
+    re_path(
-        "proxy/identity_icon/<identity_id>/",
+        "^proxy/identity_icon/(?P<identity_id>[^/]+)/((?P<image_hash>[^/]+)/)?$",
         mediaproxy.IdentityIconCacheView.as_view(),
         name="proxy_identity_icon",
     ),
-    path(
+    re_path(
-        "proxy/identity_image/<identity_id>/",
+        "^proxy/identity_image/(?P<identity_id>[^/]+)/((?P<image_hash>[^/]+)/)?$",
         mediaproxy.IdentityImageCacheView.as_view(),
         name="proxy_identity_image",
     ),
-    path(
+    re_path(
-        "proxy/post_attachment/<attachment_id>/",
+        "^proxy/post_attachment/(?P<attachment_id>[^/]+)/((?P<image_hash>[^/]+)/)?$",
         mediaproxy.PostAttachmentCacheView.as_view(),
         name="proxy_post_attachment",
     ),
-    path(
+    re_path(
-        "proxy/emoji/<emoji_id>/",
+        "^proxy/emoji/(?P<emoji_id>[^/]+)/((?P<image_hash>[^/]+)/)?$",
         mediaproxy.EmojiCacheView.as_view(),
         name="proxy_emoji",
     ),

users/models/identity.py

@@ -268,7 +268,9 @@ class Identity(StatorModel):
         if self.icon:
             return RelativeAbsoluteUrl(self.icon.url)
         elif self.icon_uri:
-            return AutoAbsoluteUrl(f"/proxy/identity_icon/{self.pk}/")
+            return AutoAbsoluteUrl(
+                f"/proxy/identity_icon/{self.pk}/", hash_tail_input=self.icon_uri
+            )
         else:
             return StaticAbsoluteUrl("img/unknown-icon-128.png")
 
@@ -279,7 +281,9 @@ class Identity(StatorModel):
         if self.image:
             return AutoAbsoluteUrl(self.image.url)
         elif self.image_uri:
-            return AutoAbsoluteUrl(f"/proxy/identity_image/{self.pk}/")
+            return AutoAbsoluteUrl(
+                f"/proxy/identity_image/{self.pk}/", hash_tail_input=self.image_uri
+            )
         return None
 
     @property