Make contents of __init__.py equal across projects (#304)

---------

Signed-off-by: Pieter Noordhuis <pieter.noordhuis@databricks.com>
Signed-off-by: Jesse Whitehouse <jesse.whitehouse@databricks.com>
Co-authored-by: Jesse Whitehouse <jesse.whitehouse@databricks.com>

Author: pietern

src/databricks/__init__.py

@@ -1,4 +1,7 @@
-# https://packaging.python.org/guides/packaging-namespace-packages/#pkgutil-style-namespace-packages
-# This file should only contain the following line. Otherwise other sub-packages databricks.* namespace
-# may not be importable.
+# See: https://packaging.python.org/guides/packaging-namespace-packages/#pkgutil-style-namespace-packages
+#
+# This file must only contain the following line, or other packages in the databricks.* namespace
+# may not be importable. The contents of this file must be byte-for-byte equivalent across all packages.
+# If they are not, parallel package installation may lead to clobbered and invalid files.
+# Also see https://github.com/databricks/databricks-sdk-py/issues/343.
 __path__ = __import__("pkgutil").extend_path(__path__, __name__)

tests/unit/test_init_file.py

@@ -0,0 +1,19 @@
+import hashlib
+
+
+class TestInitFile:
+    """
+    Micro test to confirm the contents of `databricks/__init__.py` does not change.
+
+    Also see https://github.com/databricks/databricks-sdk-py/issues/343#issuecomment-1866029118.
+    """
+
+    def test_init_file_contents(self):
+        with open("src/databricks/__init__.py") as f:
+            init_file_contents = f.read()
+
+        # This hash is the expected hash of the contents of `src/databricks/__init__.py`.
+        # It must not change, or else parallel package installation may lead to clobbered and invalid files.
+        expected_sha1 = "2772edbf52e517542acf8c039479c4b57b6ca2cd"
+        actual_sha1 = hashlib.sha1(init_file_contents.encode("utf-8")).hexdigest()
+        assert expected_sha1 == actual_sha1