Validate S3 metadata only contains ascii chars #861
Conversation
Otherwise any error message with non-ascii chars won't
generate properly.
The difference being that type('foo %s' % u'bar') -> unicode,
but type('foo {}'.format(u'bar')) -> str.
|
I fixed the failing test by defaulting to unicode literals in exceptions.py. That has the possibility for unintended side effects, but all the botocore/cli tests pass. The difference being that |
|
⛵ |
|
|
||
| """ | ||
| metadata = params.get('Metadata') | ||
| if not metadata or not isinstance(metadata, dict): |
There was a problem hiding this comment.
Yeah that is an interesting problem in that we cannot validate with the validator that the parameter passed by the user is correct prior to the handler. It would be interesting to expose a new event down the line passed the validation or a helper method to run a quick validation in the future if we run into more of these scenarios.
There was a problem hiding this comment.
I looked through several before-parameter-buildevent handlers, and there's several that are susceptible to this problem. I'm thinking it's probably worth having this event in the near future.
|
Looks good. Just had a comment to leave out there for future reference. 🚢 |
Fixes boto/boto3#478.
cc @kyleknap @JordonPhillips