Register wasb[s] scheme for url manipulations #912

datacube/utils/uris.py

@@ -222,10 +222,12 @@ def register_scheme(*schemes):
     urllib.parse.uses_params.extend(schemes)
 
 
-# s3:// not recognised by python by default
-#  without this `urljoin` might be broken for s3 urls
-register_scheme('s3')
-
-# gs:// not recognised by python by default
-#  without this `urljoin` might be broken for google storage urls
-register_scheme('gs')
+# `urljoin`, that we use for relative path computation, needs to know which url
+# schemes support relative offsets. By default only well known types are
+# understood. So here we register more common blob store url protocols.
+register_scheme(
+    's3',         # `s3://...`      -- AWS S3 Object Store
+    'gs',         # `gs://...`      -- Google Cloud Storage
+    'wasb',       # `wasb[s]://...` -- Windows Azure Storage Blob
+    'wasbs'
+)

tests/test_utils_other.py

@@ -107,6 +107,8 @@ def test_uri_to_local_path():
 @pytest.mark.parametrize("base", [
     "s3://foo",
     "gs://foo",
+    "wasb://foo",
+    "wasbs://foo",
     "/vsizip//vsicurl/https://host.tld/some/path",
 ])
 def test_uri_resolve(base):
@@ -511,6 +513,8 @@ def test_testutils_geobox():
     ("test.bar", False),
     ("s3://mybucket/objname.tiff", True),
     ("gs://mybucket/objname.tiff", True),
+    ("wasb://mybucket/objname.tiff", True),
+    ("wasbs://mybucket/objname.tiff", True),
     ("ftp://host.name/filename.txt", True),
     ("https://host.name.com/path/file.txt", True),
     ("http://host.name.com/path/file.txt", True),