feat(quart): Add span streaming support to Quart integration

[ericapisani]

Add span streaming support for the Quart integration when the
trace_lifecycle stream experiment is enabled. Sets HTTP request
attributes (method, headers, URL, query, client IP) on the segment
span and uses the correct source constant from sentry_sdk.traces
for span-first mode.

Depends on #6501
being merged first.

Fixes PY-2352
Fixes #6050

[linear-code]

PY-2352

[github-actions]

Codecov Results 📊

✅ 89523 passed | ⏭️ 6013 skipped | Total: 95536 | Pass Rate: 93.71% | Execution Time: 311m 35s

📊 Comparison with Base Branch

Metric	Change
Total Tests	📈 +112
Passed Tests	📈 +112
Failed Tests	—
Skipped Tests	—

All tests are passing successfully.

✅ Patch coverage is 91.84%. Project has 2393 uncovered lines.
✅ Project coverage is 89.81%. Comparing base (base) to head (head).

Files with missing lines (1)

File	Patch %	Lines
sentry_sdk/integrations/quart.py	91.84%	⚠️ 4 Missing and 7 partials

Coverage diff

@@            Coverage Diff             @@
##          main       #PR       +/-##
==========================================
+ Coverage    89.80%    89.81%    +0.01%
==========================================
  Files          192       192         —
  Lines        23460     23495       +35
  Branches      8062      8084       +22
==========================================
+ Hits         21069     21102       +33
- Misses        2391      2393        +2
- Partials      1328      1332        +4

---

Generated by Codecov Action

[sentrivana]

We'll need a streaming counterpart for this piece of code, something like:

client = sentry_sdk.get_client()

if has_span_streaming_enabled(client.options):
    span = get_current_span()
    if span is not None:
        span._segment._update_active_thread()

[sentrivana]

We should also set user.id if we have it (essentially port _add_user_to_event).

It's one of the attributes that should be set on all spans, not just the segment, so it should be set on the scope.

Maybe extract the code that fetches the user id from _add_user_to_event to a helper func, get rid of _add_user_to_event and set the user id in the event processor directly, and in the streaming path set it on the current scope?

def _fetch_user_id() -> None:
    if quart_auth is None:
        return

    user = quart_auth.current_user
    if user is None:
        return

    try:
        return quart_auth.current_user._auth_id
    except Exception:
        return None

and then here

user_id = _fetch_user_id()
if user_id is not None:
    sentry_sdk.get_current_scope().set_attribute("user.id", user_id)

and in the event processor

user_id = _fetch_user_id()
if user_id is not None:
    user_info = event.setdefault("user", {})
    user_info["id"] = user_id

[sentrivana]

Please also add a test for this if there is none or add to an existing one.

[ericapisani]

Sounds good. Regarding needing to set the user.id on the scope - would that mean that we also need to make an adjustment for the user.ip_address as well? I noticed it also appears in that list that you had linked 👀

[sentrivana]

That's a really good point that made me realize we had a bug in the attribute setting code, fix here. Yup we should absolutely also be setting ip_address.

I need to amend my previous comment though: we should port setting the user but not the way I wrote. There's an existing mechanism for it (set_user) that we should use -- it sets the user data on the scope and then this code materializes it on the span eventually. The only thing to be mindful of is that set_user accepts a dict where the keys are different from the final attribute names (see the linked code), so it's e.g. name instead of user.name etc.

[sentrivana]

Another thing to note: set_user will completely overwrite whatever was there previously (it doesn't merge the dicts), so we need to collect all the fields we have and update once.

[cursor]

Missing MySQL host env variable

Medium Severity

The DB CI job sets SENTRY_PYTHON_TEST_POSTGRES_HOST to postgres for Python 3.6/3.7 container runs but does not set SENTRY_PYTHON_TEST_MYSQL_HOST the same way for the new MySQL service. tests/integrations/aiomysql reads that variable and defaults to localhost, which is unreachable from the job container network.

 

^{Reviewed by Cursor Bugbot for commit dde12e6. Configure here.}

[cursor]

Missing MySQL host container env

Medium Severity

The new MySQL CI env sets SENTRY_PYTHON_TEST_MYSQL_USER and password but not SENTRY_PYTHON_TEST_MYSQL_HOST. For Python 3.6/3.7 jobs that run inside a service-linked container, tests/integrations/aiomysql defaults to localhost, unlike Postgres and Redis which use the mysql service hostname in that layout.

 

^{Reviewed by Cursor Bugbot for commit e5eb0b3. Configure here.}

[cursor]

Cursor Bugbot has reviewed your changes and found 2 potential issues.

There are 4 total unresolved issues (including 2 from previous reviews).

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

^{Reviewed by Cursor Bugbot for commit 717731d. Configure here.}

[cursor]

Debug print left in test

Low Severity

A print("SEGMENT ATTRS", ...) call remains in test_span_streaming_quart_auth_user_id, which will emit noisy output whenever that test runs under PII with a logged-in user.

 

^{Reviewed by Cursor Bugbot for commit 717731d. Configure here.}

[cursor]

set_user drops existing scope user

Medium Severity

The span-streaming path calls set_user with only newly built user_properties (IP and Quart auth id) on the current scope, so any user fields already on the isolation scope are replaced instead of merged before one update.

 

^{Reviewed by Cursor Bugbot for commit 717731d. Configure here.}