fix: grab the first value if the server response contains a comma

[MarcoP9]

Overview

After encountering a 429 Too Many Requests Error, the server returns an incorrect Retry-After header of "30,120", resulting in an exception in the retry handler. As suggested by @musale in the comments, the function is now splitting the string by the comma and returning the first value.

Related Issue

Contributes to fix: microsoftgraph/msgraph-sdk-python#1159

[baywet]

Thanks for the contribution!
Can you please add a unit test for that scenario to prevent future regressions?

[MarcoP9]

@microsoft-github-policy-service agree

[MarcoP9]

Hi @baywet, I have added a test, I hope it is conforming to what you were expecting, let me know if I need to change something :)!

[baywet]

Thank you for making the changes!

[baywet]

@MarcoP9 can you have a look at the failure for the http unit tests please?

https://github.com/microsoft/kiota-python/actions/runs/15881670608/job/44846104287?pr=482

Disregard the failing serialization CI. I've authored #483 to address this.

[MarcoP9]

Hi @baywet, I have fixed the unit test that was failing. My bad, I did not consider that other http dates might contain commas. I have now made the check more specific, the first value is grabbed only if a regex pattern is matched:
(\d{2}),\d{3}$

Let me know if it sounds good!

[baywet]

@MarcoP9 this looks like a narrow implementation. What of we don't exactly have the same number of digits?
Also regex are usually expensive in terms of compute.
Couldn't we simply try to parse the string as a float instead of an int?

[MarcoP9]

@baywet I thought about that too, however, the server is returning a retry-after value 30,120, containing a comma, and not a dot. A float casting would fail in that case :/. I thought about other possibilities as:

• Replacing commas with dots, but it would probably fail with http date as Tue, 15 Nov 1994 08:12:31 GMT
• We can split by comma and check if in the first element is all digits: value.split(",")[0] if "," in value and value.split(",")[0].isdigit()

What do you think?

[baywet]

what if we kept the code as is, but relaxed the regex a little instead?
either "\d+,\d+$" (you don't need the capturing group)
or even "[\d,]+$"

[sonarqubecloud]

Quality Gate passed

Issues
1 New issue
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud