feat: Remove truncation when stream_gen_ai_spans is enabled

[alexander-alderman-webb]

Note: Depends on #6079

Description

Remove message popping and binary blob redaction when the experimental stream_gen_ai_spans is enabled.

Stop parametrizing truncation tests that fail when the stream_gen_ai_spans option is enabled.

Issues

Reminders

• Please add tests to validate your changes, and lint your code using tox -e linters.
• Add GH Issue ID & Linear ID (if applicable)
• PR title should use conventional commit style (feat:, fix:, ref:, meta:)
• For external contributors: CONTRIBUTING.md, Sentry SDK development docs, Discord community

[github-actions]

Codecov Results 📊

✅ 282 passed | Total: 282 | Pass Rate: 100% | Execution Time: 42.74s

All tests are passing successfully.

❌ Patch coverage is 45.00%. Project has 14761 uncovered lines.

Files with missing lines (10)

File	Patch %	Lines
langchain.py	3.18%	⚠️ 578 Missing
langgraph.py	5.24%	⚠️ 217 Missing
utils.py	80.74%	⚠️ 99 Missing and 56 partials
ai_client.py	0.00%	⚠️ 148 Missing
openai.py	90.84%	⚠️ 63 Missing and 71 partials
anthropic.py	86.68%	⚠️ 59 Missing and 42 partials
utils.py	14.56%	⚠️ 88 Missing
invoke_agent.py	0.00%	⚠️ 82 Missing
invoke_agent.py	0.00%	⚠️ 47 Missing
litellm.py	87.92%	⚠️ 18 Missing and 16 partials

---

Generated by Codecov Action

[sentrivana]

Nice! Two things:

1. Does message truncation exclusively affect gen_ai spans currently? In other words, is it guaranteed that all spans that will become very large because of this change will be sent as v2 spans, or do we need an extra check to only lift the truncation for gen_ai spans specifically?
2. If we remove the parametrization from the truncation tests, will we have test coverage for the "stream_gen_ai_spans=True should not truncate messages" code path?

[alexander-alderman-webb]

1. That's a possible edge case, and since it's unexpected for truncate_and_annotate_messages to not truncate in some cases, I moved the logic to the call sites in the integrations: 4ec3ff7
2. By removing the parametrization we do not assert that there is no truncation when stream_gen_ai_spans=True. I don't have strong feelings here, and removed the parametrization because I try to avoid having to many tests that assert the absence of something 🤷 . Happy to parametrize the tests again as well.

[sentrivana]

2. By removing the parametrization we _do not_ assert that there is no truncation when `stream_gen_ai_spans=True`. I don't have strong feelings here, and removed the parametrization because I try to avoid having to many tests that assert the absence of something 🤷 . Happy to parametrize the tests again as well.

I'm fine with removing the parametrization, but we should have some test coverage for the messages being there as expected, without truncation, in the stream_gen_ai_spans=True case. If this is already tested by another existing test not touched by this PR, fine to leave as is. If not, would be good to add a new test for that specifically.

[alexander-alderman-webb]

Yes, that makes sense. I'll update other tests to accept two inputs messages and assert that both are present when the option is enabled (demonstrating that input truncation is disabled in this case).

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 962fd65. Configure here.}

[sentrivana]