Serve artifacts of a type that can run script with CSP sandbox and nosniff

backend/fluksio/api/routes/artifacts.py

@@ -27,6 +27,11 @@ from fluksio.flow.artifacts import ArtifactStore
 #: trusting one — anything else is served as bytes.
 _MEDIA_TYPE = re.compile(r"^[\w.+-]+/[\w.+-]+$")
 
+#: What a browser shows without running anything, so it is served as it is.
+#: Every other type is served sandboxed: HTML, SVG and XML run script when
+#: opened as a page, and so may a type nobody listed here.
+_INERT = re.compile(r"^(audio|video)/|^image/(?!svg)|^application/(pdf|json)$")
+
 
 def artifact_caller(request: Request) -> str:
     """Who may move artifacts: a signed-in person, or an attached worker.
@@ -137,11 +142,14 @@ def get_artifact(
     path = store.path(digest)
     if path is None:
         raise HTTPException(status_code=404, detail="No such artifact")
-    return FileResponse(
-        path,
-        media_type=(
-            media_type if _MEDIA_TYPE.match(media_type) else "application/octet-stream"
-        ),
-        # The digest is the content, so it is also the perfect validator.
-        headers={"ETag": f'"{digest}"'},
-    )
+    if not _MEDIA_TYPE.match(media_type):
+        media_type = "application/octet-stream"
+    # The digest is the content, so it is also the perfect validator.
+    headers = {"ETag": f'"{digest}"'}
+    if not _INERT.match(media_type.lower()):
+        # The caller names the type and any member can upload, so an artifact
+        # asked for as text/html would otherwise run its script on this origin,
+        # beside the session, for whoever follows a link to it.
+        headers["Content-Security-Policy"] = "sandbox"
+        headers["X-Content-Type-Options"] = "nosniff"
+    return FileResponse(path, media_type=media_type, headers=headers)