Add type annotations to storage API #1410
Conversation
Codecov Report
@@ Coverage Diff @@
## trunk #1410 +/- ##
==========================================
+ Coverage 86.46% 86.46% +<.01%
==========================================
Files 366 366
Lines 76791 76794 +3
Branches 7529 7529
==========================================
+ Hits 66396 66399 +3
Misses 7527 7527
Partials 2868 2868
Continue to review full report at Codecov.
|
libcloud/storage/base.py
Outdated
| """ | ||
| raise NotImplementedError( | ||
| 'download_object_as_stream not implemented for this driver') | ||
|
|
||
| def upload_object(self, file_path, container, object_name, extra=None, | ||
| verify_hash=True, headers=None): | ||
| # type: (str, Container, str, Optional[dict], Optional[bool], Optional[dict]) -> Object # noqa: E501 |
There was a problem hiding this comment.
Since verify_hash has a default value, I don't think it needs Optional annotation.
There was a problem hiding this comment.
And for headers we could probably use Dict[str, str].
|
Nice work 👍 What's your take on utilizing Python 3 native syntax for type annotations vs using comments? In theory, using comments, allows the trunk to still work with Python 2.7, but since we don't officially support it any more, I'm not sure if that matters or not since sooner or later we will likely start utilizing other Python 3 only features as well (and we have v2.8.x branch we can base v2.8.x bug fix release on, if needed). |
@Kami Personally I'm not a huge fan of the type comments for function signatures and would be in favor of eventually switching to the native type annotations to increase readability. E.g. compare the following two snippets: def upload_object_via_stream(self, iterator, object_name, extra=None,
headers=None):
# type: (Iterator[bytes], str, Optional[dict], Optional[Dict[str, str]]) -> Object # noqa: E501def upload_object_via_stream(self,
iterator: Iterator[bytes],
object_name: str,
extra: Optional[dict] = None,
headers: Optional[Dict[str, str]] = None,
) -> Object:From my point of view, the former is harder to read since the function arguments must be matched with the type declaration based on index/position as opposed to being attached to each argument explicitly. |
|
The type annotations uncovered a bug: most drivers were discarding the base API's |
|
@c-w Good catch on the headers issue 👍 As far as comments notation goes - in-line annotations can also be used when using comments, but I also agree that native notation is nicer. So perhaps now that we don't need to support Python 2.7 anymore, we can just switch to that notation. |
Add type annotations to storage API
Description
As discussed in #1408 (comment), this PR adds type annotations to the StorageDriver API.
For now, mypy errors are ignored for the s3, google_storage and cloudfiles drivers since the use of inheritance on the connection classes is causing issues which will require a more thorough strategy to address.
Status
Checklist