Pagination broken in get_tags for ECR when limit None is given

[natefaerber]

get_tags fails when limit is not give or is given as None (same thing.)

from oras.client import OrasClient
client = OrasClient('__SNIPPED_BY_ME__.dkr.ecr.us-west-2.amazonaws.com', auth_backend='basic')
pw = "..."
client.login(password=pw, username="AWS", hostname="__SNIPPED_BY_ME__.dkr.ecr.us-west-2.amazonaws.com")
from oras.logger import setup_logger, logger
setup_logger(quiet=False, debug=True)

client.get_tags('my/app')

my/app is a "namespaced" repository. This repository has over 1000 images which I suspect is part of the issue here since I don't have issues with smaller repos or issues when a pass N<=1000.

Errors seen after enabling debug logging.

Traceback (most recent call last):
  File "__SNIPPED_BY_ME__/.venv/lib/python3.12/site-packages/oras/decorator.py", line 40, in inner
    res = func(*args, **kwargs)
          ^^^^^^^^^^^^^^^^^^^^^
  File "__SNIPPED_BY_ME__/.venv/lib/python3.12/site-packages/oras/provider.py", line 967, in do_request
    response = self.session.request(
               ^^^^^^^^^^^^^^^^^^^^^
  File "__SNIPPED_BY_ME__/.venv/lib/python3.12/site-packages/requests/sessions.py", line 589, in request
    resp = self.send(prep, **send_kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "__SNIPPED_BY_ME__/.venv/lib/python3.12/site-packages/requests/sessions.py", line 703, in send
    r = adapter.send(request, **kwargs)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "__SNIPPED_BY_ME__/.venv/lib/python3.12/site-packages/requests/adapters.py", line 700, in send
    raise ConnectionError(e, request=request)
requests.exceptions.ConnectionError: HTTPSConnectionPool(host='__SNIPPED_BY_ME__.dkr.ecr.us-west-2.amazonaws.comhttps', port=443): Max retries exceeded with url: /__SNIPPED_BY_ME__.dkr.ecr.us-west-2.amazonaws.com/v2/my/app/tags/list?last=ukD72mdD%2FmC8b5xV3susmJzzaTgp3hKwR9nRUW1yZZ45rnWqcRvUcjdSjqGstiFS1nz2HtUlMNI14iKrrj%2F35AU1TdEkjIZ<snipped>A8VLpN5xHZ3%2BeRlKHJ7d%2FbioNLy3R5jOon7X61YbIG%2BRHzgQyJieYh5fCaZoH8fw%3D%3D (Caused by NameResolutionError("<urllib3.connection.HTTPSConnection object at 0x104e53ef0>: Failed to resolve '__SNIPPED_BY_ME__.dkr.ecr.us-west-2.amazonaws.comhttps' ([Errno 8] nodename nor servname provided, or not known)"))

Walking through the _do_paginated_request function, I can get a sense of the issue.

The response to the do_request will give me a links object like this:

{'next': {'url': 'https://__SNIPPED_BY_ME__.dkr.ecr.us-west-2.amazonaws.com/v2/my/app/tags/list?last=ukD72mdD%2FmC8b5xV3susmJzzaTgp3hKwR9nRUW1yZZ45rnWqcRvUcjdSjqGstiFS1nz2HtUlMNI14iKrrj%2F35AU1TdEkjIZ<snipped>A8VLpN5xHZ3%2BeRlKHJ7d%2FbioNLy3R5jOon7X61YbIG%2BRHzgQyJieYh5fCaZoH8fw%3D%3D', 'rel': 'next'}}

If I don't hit a condition to break out of the while loop, I get to this part of the code at https://github.com/oras-project/oras-py/blob/0.2.25/oras/provider.py#L394-L395.

# use link + base url to continue with next page
url = f"{base_url}{link}"

where link is https://__SNIPPED_BY_ME__.dkr.ecr.us-west-2.amazonaws.com/v2/my/app/tags/list?last=ukD72mdD%2FmC8b5xV3susmJzzaTgp3hKwR9nRUW1yZZ45rnWqcRvUcjdSjqGstiFS1nz2HtUlMNI14iKrrj%2F35AU1TdEkjIZ<snipped>A8VLpN5xHZ3%2BeRlKHJ7d%2FbioNLy3R5jOon7X61YbIG%2BRHzgQyJieYh5fCaZoH8fw%3D%3D which is going to be a problem when appended to {base_url} which is https://__SNIPPED_BY_ME__.dkr.ecr.us-west-2.amazonaws.com.

I suspect a solution could be as simple as changing

url = f"{base_url}{link}"

to

url = urllib.parse.urljoin(base_url, link)

However, I don't know that I have the bandwidth or context to know how to verify that this won't impact the repos for which this pagination was originally written and presumably working. Based on #68 (comment), I have a good feeling this will work as expected for repos that return the path as the link instead of the full URL.

I will followup with a PR.