Fix convert_to_dapr_duration double-counting sub-second time

[amdadulbari]

Description

convert_to_dapr_duration() produces a Dapr duration string that is ~1 second too large for any timedelta whose sub-second part is >= 0.5.

from datetime import timedelta
from dapr.serializers.util import convert_to_dapr_duration, convert_from_dapr_duration

convert_to_dapr_duration(timedelta(milliseconds=1500))
# -> '0h0m2s500ms0μs'   (2.5s, expected 1.5s)
convert_from_dapr_duration('0h0m2s500ms0μs')
# -> timedelta(seconds=2, microseconds=500000)   # round-trip is off by 1s

The seconds field comes from divmod(td.total_seconds(), 60), so it still carries the sub-second fraction — which is also emitted via the ms/μs fields (from td.microseconds). Formatting it with {:.0f} then rounds a fraction >= 0.5 up into an extra whole second, so the fraction is both double-counted and rounded.

This is on the serialization hot path (DaprJSONEncoder serializes every timedelta), so any duration with a >= 0.5s fraction — e.g. actor timer/reminder periods or state TTLs — is encoded incorrectly.

Fix

Truncate the whole-seconds field with int() so the sub-second fraction is only counted once (via the ms/μs fields). hours/mins are always whole, so only seconds needs this.

Tests

Added test_convert_timedelta_to_dapr_duration_subsecond in tests/serializers/test_util.py (asserts the exact string and an exact round-trip). It fails before this change and passes after. Existing tests, ruff, and mypy all pass.

[Copilot]

Pull request overview

This PR fixes convert_to_dapr_duration() in the Dapr Python SDK serializer utilities so that timedelta values with a sub-second fraction >= 0.5s no longer round up and double-count the fractional part when emitting the Go/Dapr duration string format.

Changes:

• Truncate the whole-seconds component when formatting durations so sub-second fractions are emitted only via the ms/μs fields.
• Add a regression test asserting both the exact encoded string and an exact round-trip for a timedelta(milliseconds=1500) case.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
dapr/serializers/util.py	Fixes duration formatting by truncating the seconds field to avoid double-counting/rounding sub-second fractions.
tests/serializers/test_util.py	Adds a regression test covering sub-second rounding/double-counting and validates exact round-trip behavior.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 82.77%. Comparing base (e8c506b) to head (5a7eef4).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1192   +/-   ##
=======================================
  Coverage   82.77%   82.77%           
=======================================
  Files         123      123           
  Lines       10130    10130           
=======================================
  Hits         8385     8385           
  Misses       1745     1745           

☔ View full report in Codecov by Harness.
📢 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.

[acroca]

Nit: I think it'd be nice to cover 1999 and 1001 milliseconds too.