chore: add HTTPAdapter to request session

[sevbch]

Context

A customer reported this issue when scanning with ggshield, on a self-hosted instance (XXX= host url).

59089:605129 [E] ggshield.core.errors:209 status_code=None detail=HTTPSConnectionPool(host='XXX', port=443): Read timed out. (read
timeout=60)
2025-10-28 11:13:04 59089:605240 [W] urllib3.connectionpool:329 Connection pool is full, discarding connection: XXX. Connection pool size: 10
2025-10-28 11:13:23 59089:605238 [W] urllib3.connectionpool:329 Connection pool is full, discarding connection: XXX. Connection pool size: 10
Scanning... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╺━━ 93% 529 / 569
Error: Scanning failed: HTTPSConnectionPool(host='XXX', port=443): Read timed out. (read timeout=60)

What has been done

Adding a HTTPAdapter to the session configuration to better handle many parallel requests. The risk is that it could put additional pressure on the server. After reading this article, I left pool_connections to default value, as it shouldn't have much of an impact for our case, but increased pool_maxsize as we can see in the logs that this is the param that was blocking for the customer.

Another option was to increase the 60s timeout, but I'm afraid it would have too much impact. For instance, when a ggshield command fail, it could result in a longer waiting time for the caller to fetch the output.

I'm open to suggestions if you see another way of handling this.

Validation

Validated with the unit test.

PR check list

• As much as possible, the changes include tests (unit and/or functional)
• If the changes affect the end user (new feature, behavior change, bug fix) then the PR has a changelog entry (see doc/dev/getting-started.md). If the changes do not affect the end user, then the skip-changelog label has been added to the PR.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 91.95%. Comparing base (98f0c17) to head (2c04ebc).
⚠️ Report is 14 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1141   +/-   ##
=======================================
  Coverage   91.94%   91.95%           
=======================================
  Files         144      144           
  Lines        6196     6200    +4     
=======================================
+ Hits         5697     5701    +4     
  Misses        499      499           

Flag	Coverage Δ
unittests	91.95% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[sevbch]

@agateau-gg I updated the PR, I think it should make more sense now.

[agateau-gg]

This test does not test much: Session.adapters contains HTTPAdapter instances by default, for both http and https:

>>> from requests import Session
>>> s = Session()
>>> s.adapters
OrderedDict({'https://': <requests.adapters.HTTPAdapter object at 0x78f4aca515b0>, 'http://': <requests.adapters.HTTPAdapter object at 0x78f4ab918e30>})

I think it can be dropped.