Improve cache busting proxy URLs with file exts

Fixes #287.

activities/models/emoji.py

@@ -17,7 +17,12 @@ from core.html import strip_html
 from core.ld import format_ld_date
 from core.models import Config
 from core.uploads import upload_emoji_namer
-from core.uris import AutoAbsoluteUrl, RelativeAbsoluteUrl, StaticAbsoluteUrl
+from core.uris import (
+    AutoAbsoluteUrl,
+    ProxyAbsoluteUrl,
+    RelativeAbsoluteUrl,
+    StaticAbsoluteUrl,
+)
 from stator.models import State, StateField, StateGraph, StatorModel
 from users.models import Domain
 
@@ -169,8 +174,9 @@ class Emoji(StatorModel):
             if self.file:
                 return AutoAbsoluteUrl(self.file.url)
             elif self.remote_url:
-                return AutoAbsoluteUrl(
+                return ProxyAbsoluteUrl(
-                    f"/proxy/emoji/{self.pk}/", hash_tail_input=self.remote_url
+                    f"/proxy/emoji/{self.pk}/",
+                    remote_url=self.remote_url,
                 )
         return StaticAbsoluteUrl("img/blank-emoji-128.png")
 

activities/models/post_attachment.py

@@ -3,7 +3,7 @@ from functools import partial
 from django.db import models
 
 from core.uploads import upload_namer
-from core.uris import AutoAbsoluteUrl, RelativeAbsoluteUrl
+from core.uris import ProxyAbsoluteUrl, RelativeAbsoluteUrl
 from stator.models import State, StateField, StateGraph, StatorModel
 
 
@@ -81,16 +81,18 @@ class PostAttachment(StatorModel):
         elif self.file:
             return RelativeAbsoluteUrl(self.file.url)
         else:
-            return AutoAbsoluteUrl(
+            return ProxyAbsoluteUrl(
-                f"/proxy/post_attachment/{self.pk}/", hash_tail_input=self.remote_url
+                f"/proxy/post_attachment/{self.pk}/",
+                remote_url=self.remote_url,
             )
 
     def full_url(self):
         if self.file:
             return RelativeAbsoluteUrl(self.file.url)
         else:
-            return AutoAbsoluteUrl(
+            return ProxyAbsoluteUrl(
-                f"/proxy/post_attachment/{self.pk}/", hash_tail_input=self.remote_url
+                f"/proxy/post_attachment/{self.pk}/",
+                remote_url=self.remote_url,
             )
 
     ### ActivityPub ###

core/uris.py

@@ -32,15 +32,8 @@ class AutoAbsoluteUrl(RelativeAbsoluteUrl):
         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:
@@ -48,6 +41,31 @@ class AutoAbsoluteUrl(RelativeAbsoluteUrl):
         self.absolute = urljoin(absolute_prefix, self.relative)
 
 
+class ProxyAbsoluteUrl(AutoAbsoluteUrl):
+    """
+    AutoAbsoluteUrl variant for proxy paths, that also attaches a remote URI hash
+    plus extension to the end if it can.
+    """
+
+    def __init__(
+        self,
+        relative: str,
+        identity=None,
+        remote_url: str | None = None,
+    ):
+        if remote_url:
+            # See if there is a file extension we can grab
+            extension = "bin"
+            remote_filename = remote_url.split("/")[-1]
+            if "." in remote_filename:
+                extension = remote_filename.split(".")[-1]
+            # When provided, attach a hash of the remote 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.
+            relative += f"{hashlib.sha1(remote_url.encode('ascii')).hexdigest()[:10]}.{extension}"
+        super().__init__(relative, identity)
+
+
 class StaticAbsoluteUrl(RelativeAbsoluteUrl):
     """
     Creates static URLs given only the static-relative path

takahe/urls.py

@@ -212,22 +212,22 @@ urlpatterns = [
     path("debug/oauth_authorize/", debug.OauthAuthorize.as_view()),
     # Media/image proxy
     re_path(
-        "^proxy/identity_icon/(?P<identity_id>[^/]+)/((?P<image_hash>[^/]+)/)?$",
+        "^proxy/identity_icon/(?P<identity_id>[^/]+)/((?P<image_hash>[^/]+))?$",
         mediaproxy.IdentityIconCacheView.as_view(),
         name="proxy_identity_icon",
     ),
     re_path(
-        "^proxy/identity_image/(?P<identity_id>[^/]+)/((?P<image_hash>[^/]+)/)?$",
+        "^proxy/identity_image/(?P<identity_id>[^/]+)/((?P<image_hash>[^/]+))?$",
         mediaproxy.IdentityImageCacheView.as_view(),
         name="proxy_identity_image",
     ),
     re_path(
-        "^proxy/post_attachment/(?P<attachment_id>[^/]+)/((?P<image_hash>[^/]+)/)?$",
+        "^proxy/post_attachment/(?P<attachment_id>[^/]+)/((?P<image_hash>[^/]+))?$",
         mediaproxy.PostAttachmentCacheView.as_view(),
         name="proxy_post_attachment",
     ),
     re_path(
-        "^proxy/emoji/(?P<emoji_id>[^/]+)/((?P<image_hash>[^/]+)/)?$",
+        "^proxy/emoji/(?P<emoji_id>[^/]+)/((?P<image_hash>[^/]+))?$",
         mediaproxy.EmojiCacheView.as_view(),
         name="proxy_emoji",
     ),

users/models/identity.py

@@ -23,7 +23,12 @@ from core.ld import (
 from core.models import Config
 from core.signatures import HttpSignature, RsaKeys
 from core.uploads import upload_namer
-from core.uris import AutoAbsoluteUrl, RelativeAbsoluteUrl, StaticAbsoluteUrl
+from core.uris import (
+    AutoAbsoluteUrl,
+    ProxyAbsoluteUrl,
+    RelativeAbsoluteUrl,
+    StaticAbsoluteUrl,
+)
 from stator.models import State, StateField, StateGraph, StatorModel
 from users.models.domain import Domain
 from users.models.system_actor import SystemActor
@@ -268,8 +273,9 @@ class Identity(StatorModel):
         if self.icon:
             return RelativeAbsoluteUrl(self.icon.url)
         elif self.icon_uri:
-            return AutoAbsoluteUrl(
+            return ProxyAbsoluteUrl(
-                f"/proxy/identity_icon/{self.pk}/", hash_tail_input=self.icon_uri
+                f"/proxy/identity_icon/{self.pk}/",
+                remote_url=self.icon_uri,
             )
         else:
             return StaticAbsoluteUrl("img/unknown-icon-128.png")
@@ -281,8 +287,9 @@ class Identity(StatorModel):
         if self.image:
             return AutoAbsoluteUrl(self.image.url)
         elif self.image_uri:
-            return AutoAbsoluteUrl(
+            return ProxyAbsoluteUrl(
-                f"/proxy/identity_image/{self.pk}/", hash_tail_input=self.image_uri
+                f"/proxy/identity_image/{self.pk}/",
+                remote_url=self.image_uri,
             )
         return None