pytest-json-ctrf 0.6.2__tar.gz → 0.6.3__tar.gz

{pytest_json_ctrf-0.6.2/pytest_json_ctrf.egg-info → pytest_json_ctrf-0.6.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pytest-json-ctrf
-Version: 0.6.2
+Version: 0.6.3
 Summary: Pytest plugin to generate json report in CTRF (Common Test Report Format)
 Author: Oleksii Ostapov
 License-Expression: MIT
@@ -132,6 +132,9 @@ The `pytest_runtest_logreport` hook in the controller node is used to collect th
 - https://github.com/infopulse/Playwright-course-python -> The report will be used in the demo project as +1 report option
 
 ## Changelog
+- 0.6.3 - fixed issue 7 - test and retry attempt `duration` is the sum of pytest's monotonic phase durations, so it stays correct when a time-travel library such as time-machine moves the wall clock between tests
+  - `start` and `stop` are still wall-clock timestamps, so `duration` can differ from `stop - start`
+  - time between phases (report hooks, failure formatting) is no longer counted
 - 0.6.2 - minimum versions are now pytest 7.4 and Python 3.9 (drops Python 3.8). pytest 6.x failed at startup, even without `--ctrf`; 7.0–7.2 failed when `--ctrf` was used; 7.3 worked, but reported `pytest.fail(..., pytrace=False)` messages without the `Failed: ` prefix
 - 0.6.1 - proper license added to the package to fix build pipeline warnings
 - 0.6.0 - fixed issue 12 - report now validates against the CTRF schema. Thanks to [@jamesarosen](https://github.com/jamesarosen) for the detailed report!

{pytest_json_ctrf-0.6.2 → pytest_json_ctrf-0.6.3}/README.md

@@ -115,6 +115,9 @@ The `pytest_runtest_logreport` hook in the controller node is used to collect th
 - https://github.com/infopulse/Playwright-course-python -> The report will be used in the demo project as +1 report option
 
 ## Changelog
+- 0.6.3 - fixed issue 7 - test and retry attempt `duration` is the sum of pytest's monotonic phase durations, so it stays correct when a time-travel library such as time-machine moves the wall clock between tests
+  - `start` and `stop` are still wall-clock timestamps, so `duration` can differ from `stop - start`
+  - time between phases (report hooks, failure formatting) is no longer counted
 - 0.6.2 - minimum versions are now pytest 7.4 and Python 3.9 (drops Python 3.8). pytest 6.x failed at startup, even without `--ctrf`; 7.0–7.2 failed when `--ctrf` was used; 7.3 worked, but reported `pytest.fail(..., pytrace=False)` messages without the `Failed: ` prefix
 - 0.6.1 - proper license added to the package to fix build pipeline warnings
 - 0.6.0 - fixed issue 12 - report now validates against the CTRF schema. Thanks to [@jamesarosen](https://github.com/jamesarosen) for the detailed report!

{pytest_json_ctrf-0.6.2 → pytest_json_ctrf-0.6.3}/ctrf/TestObject.py

@@ -18,6 +18,7 @@ class TestObject:
     raw_status: Optional[str]
     start: int
     stop: int
+    _duration_s: float
     retries: int
     message: Optional[str]
 
@@ -48,6 +49,7 @@ class TestObject:
         self.trace = None
         self.start = 0
         self.stop = 0
+        self._duration_s = 0.0
         self._setup_seen = False
 
     @staticmethod
@@ -73,7 +75,7 @@ class TestObject:
         attempt = {
             'attempt': len(self.retry_attempts) + 1,
             'status': self._status.value,
-            'duration': max(self.stop - self.start, 0),
+            'duration': round(self._duration_s * 1000),
             'start': self.start,
             'stop': self.stop,
             'message': self.message,
@@ -133,6 +135,11 @@ class TestObject:
                 self._finish_attempt()
             self._setup_seen = True
             self.start = int(report.start * 1000)
+        # report.duration is monotonic, unlike the wall-clock start/stop.
+        # SubtestReport (pytest>=9) and pytest-subtests' SubTestReport carry a context;
+        # their time is already counted in the parent test's call phase
+        if not hasattr(report, 'context'):
+            self._duration_s += report.duration
         self.set_status(report)
         if report.when == "teardown":
             self.stop = int(report.stop * 1000)
@@ -146,7 +153,7 @@ class TestObject:
             'name': self.name,
             'status': self._status.value,
             'rawStatus': self.raw_status,
-            'duration': self.stop - self.start,
+            'duration': round(self._duration_s * 1000),
             'start': self.start,
             'stop': self.stop,
             'suite': self.suite,

{pytest_json_ctrf-0.6.2 → pytest_json_ctrf-0.6.3}/pyproject.toml

@@ -10,7 +10,7 @@ exclude = [
 
 [project]
 name = "pytest-json-ctrf"
-version = "0.6.2"
+version = "0.6.3"
 
 # minimum pytest and Python versions: keep in sync with .github/workflows/deploy.yml and the README
 dependencies = [

{pytest_json_ctrf-0.6.2 → pytest_json_ctrf-0.6.3/pytest_json_ctrf.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pytest-json-ctrf
-Version: 0.6.2
+Version: 0.6.3
 Summary: Pytest plugin to generate json report in CTRF (Common Test Report Format)
 Author: Oleksii Ostapov
 License-Expression: MIT
@@ -132,6 +132,9 @@ The `pytest_runtest_logreport` hook in the controller node is used to collect th
 - https://github.com/infopulse/Playwright-course-python -> The report will be used in the demo project as +1 report option
 
 ## Changelog
+- 0.6.3 - fixed issue 7 - test and retry attempt `duration` is the sum of pytest's monotonic phase durations, so it stays correct when a time-travel library such as time-machine moves the wall clock between tests
+  - `start` and `stop` are still wall-clock timestamps, so `duration` can differ from `stop - start`
+  - time between phases (report hooks, failure formatting) is no longer counted
 - 0.6.2 - minimum versions are now pytest 7.4 and Python 3.9 (drops Python 3.8). pytest 6.x failed at startup, even without `--ctrf`; 7.0–7.2 failed when `--ctrf` was used; 7.3 worked, but reported `pytest.fail(..., pytrace=False)` messages without the `Failed: ` prefix
 - 0.6.1 - proper license added to the package to fix build pipeline warnings
 - 0.6.0 - fixed issue 12 - report now validates against the CTRF schema. Thanks to [@jamesarosen](https://github.com/jamesarosen) for the detailed report!