Handle ceph bucket names

Fix an issue where multi-tenant ceph S3 bucket names would not be correctly parsed, since they include a colon.

Author: dhirving

python/lsst/resources/s3.py

@@ -200,7 +200,15 @@ def profile(self) -> str | None:
 
     @property
     def bucket(self) -> str:
-        bucket = self._uri.hostname
+        split = self._uri.netloc.split("@")
+        num_components = len(split)
+        if num_components == 2:
+            bucket = split[1]
+        elif num_components == 1:
+            bucket = split[0]
+        else:
+            raise ValueError(f"Unexpected extra '@' in S3 URI: '{str(self)}'")
+
         if not bucket:
             raise ValueError(f"S3 URI does not include bucket name: '{str(self)}'")
 

tests/test_s3.py

@@ -266,6 +266,23 @@ def test_s3_endpoint_url(self):
             path.write(data)
             self.assertEqual(path.read(), data)
 
+    def test_uri_syntax(self):
+        path1 = ResourcePath("s3://profile@bucket/path")
+        self.assertEqual(path1.bucket, "bucket")
+        self.assertEqual(path1.profile, "profile")
+        path2 = ResourcePath("s3://bucket2/path")
+        self.assertEqual(path2.bucket, "bucket2")
+        self.assertEqual(path2.profile, None)
+
+    def test_ceph_uri_syntax(self):
+        # The Ceph S3 'multi-tenant' syntax for buckets can include colons.
+        path1 = ResourcePath("s3://profile@ceph:bucket/path")
+        self.assertEqual(path1.bucket, "ceph:bucket")
+        self.assertEqual(path1.profile, "profile")
+        path2 = ResourcePath("s3://ceph:bucket2/path")
+        self.assertEqual(path2.bucket, "ceph:bucket2")
+        self.assertEqual(path2.profile, None)
+
 
 if __name__ == "__main__":
     unittest.main()