S3 sync command incorrectly compares keys of files with spaces in their names.

[johnboxall]

The aws s3 sync command incorrectly compares the keys of files with spaces:

$ mkdir test
$ echo "hellos s3" > "test/test .txt"
$ aws s3 mb s3://test
$ aws s3 sync test s3://test --debug 2>&1 | grep syncing
> ... syncing: test/test .txt -> www-test.mobify.com/test .txt, file does not exist at destination
$ aws s3 sync test s3://test --debug 2>&1 | grep syncing
> ... syncing: test/test .txt -> www-test.mobify.com/test .txt, file does not exist at destination
$ aws --version
> aws-cli/1.3.2 Python/2.7.6 Darwin/13.0.0

It seems as though in awscli.customizations.s3.comparator, the key that is generated for S3 encodes the space with "+", while the file from the local FS is not.

I believe that this is a different error than #614.

<3

[JRyanJones]

I can confirm this issue in the 1.3.2 release. Files with spaces are uploaded as such but when attempting to download, the spaces appear to be encoded as '+', yielding non-existent key errors.

[ErikEvenson]

I am seeing this issue as well: http://stackoverflow.com/questions/22652247/aws-cli-1-3-4-error-where-aws-cli-1-3-1-has-none.

[irfan-syed]

Same here. My sync operations from local disk to S3 are broken now because of this. This should be prioritised.

[jamesls]

Looking into this...

[anazar]

+1 also having this issue with sync

[jamesls]

PR linked above should fix this. Just need a quick code review for merging.

[anazar]

Thanks @jamesls !!

I'm not sure if there's a better place to ask this question but when using sync... how does it keep track of what files have been sycned? I'm planning to run a sync from one bucket to another to sync ~30million files. Sync seems to be the best solution for this that I've found. Any concerns or gotchas I should be aware of?

[irfan-syed]

Thanks jamesls. Is there going to be a new release for end users?