apache / apache/libcloud

Deletion of S3 objects fails for object paths starting with a slash

Open
#1,654 7 comments 0 reactions 0 assignees View on GitHub
api: storage drivers: aws s3
Dominant language
Python
Stars
2.1k
Forks
931
Avg merge
1d 2h
Merged PRs (30d)
4

Description

## Summary

When trying to delete all objects in an S3 bucket, the deletion fails for objects with a leading slash in their path.
The `delete_object` method returns `True` for these objects, even though they are not deleted.

## Detailed Information

To delete an S3 bucket we use `Container.list_objects()` and `delete_object` to first delete all objects in the bucket. This doesn't work for objects, where the object path starts with a slash. The `delete_object` function returns `True` for these objects even though they are not deleted because the DELETE request is sent to the wrong URL.
For example to delete the object `s3://container-name//path/to/object` the DELETE request is sent to
https://s3-eu-west-1.amazonaws.com/container-name/path/to/object
but it should be sent to this path instead:
https://s3-eu-west-1.amazonaws.com//container-name/path/to/object
The problem is that double slashes are replaced in the request URL in the method `Connection.morph_action_hook`. This is a similar issue to #1529.

To reproduce the issue, you can use the following code (replace credentials and bucket name):
```python
from libcloud.storage.types import Provider
from libcloud.storage.providers import get_driver

driver = get_driver(Provider.S3)
client = driver("key", "secret", region="eu-west-1")

container = client.get_container("container-name")
objects = container.list_objects()
for obj in objects:
client.delete_object(obj)
client.delete_container(container)
```
This will result in an `ContainerIsNotEmptyError` if the bucket "container-name" contains objects where the object path starts with a slash.

Libcloud version: 3.4.1
Python version: 3.7.12

Contributor guide

Open the contributing guide

Research direction

Start at Connection.morph_action_hook and reproduce the issue with the S3 example using an object path that begins with a slash. Trace how the DELETE request URL is normalized; done means the leading slash is preserved, the object is actually deleted, and the bucket can then be removed.

Written by the indexing model from the issue text.

Assessment

Tech stack
aws, python
Domain
cloud
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Clearly specified
Newbie friendliness
42/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.