pytest-Jira-xray-plus

An independent Apache-2.0 fork of pytest-jira-xray, based on 6ac1117585da590001cdbc2bf261b78683c19dcb (0.9.4 line). Experimental alpha 0.1.0a1, prepared for community testing. Native per-case Xray integration is unverified; this is not production-ready. No native iteration or dataset support is claimed. Publication status must be checked on PyPI.

Requires Python 3.12+ and pytest 9.1.1–9.x; validated on Python 3.12 / pytest 9.1.1 only. Older inherited compatibility claims have been removed for this initial alpha. Report bugs with synthetic reproducers via GitHub issues.

Start with the two-stage validation kit: one offline command, then an optional experimental sandbox-only native capability probe.

Preserve every executed case offline

pytest --xray-case-report=cases.json now produces a transport-independent JSON report: unique node IDs, effective parameter IDs, phase outcomes and all observed attempts. A failed attempt survives a later successful retry. No Jira markers, credentials or uploads are needed. Parameter values are excluded; --xray-case-values separately opts into potentially sensitive scalar values. IDs themselves can contain data, so use safe explicit IDs or opt into --xray-case-redact-identities for per-run aliases in the offline file.

See report schema and aggregation rules. The report is not Xray import JSON and does not populate Jira's native iteration UI.

Optional comment fallback

--xray-parameter-details retains case details in the legacy Xray Test Run comment. This remains a fallback, separate from the complete local case report. It does not create native Xray iterations or datasets, and it explicitly opts into including scalar values. Without either feature flag, upstream behavior is unchanged.

Upstream already rejects duplicate Jira IDs by default. With --allow-duplicate-ids, it merges results and preserves failure over pass. The baseline probe reproduces loss of individual passing-case detail, not a last-result-wins status bug.

Local use (no Jira required)

Use a fresh virtual environment. Do not install this fork alongside pytest-jira-xray: both intentionally provide pytest_xray and the xray pytest entry point for compatibility.

python -m venv .venv
. .venv/bin/activate
python -m pip install -e . pytest-httpserver pytest-xdist pytest-rerunfailures
pytest examples/synthetic.py --xray-case-report=cases.json

# Optional legacy Xray JSON comment fallback (also offline):
pytest examples/synthetic.py --jira-xray --allow-duplicate-ids \
  --xray-parameter-details --xraypath=xray.json

The example intentionally has one failure followed by a pass. Exit code 1 is expected. Add --cloud for Cloud status strings (PASSED/FAILED); the default remains upstream's Server/DC spelling (PASS/FAIL). When enabling --jira-xray, also use --xraypath for offline evaluation; omitting it enables the inherited upload path. --xray-case-report alone never enables that publisher.

The comment fallback leaves non-parametrized Xray results unchanged. Existing captures, defects and evidences continue to merge. Setup/teardown errors are separate phase records, not extra parameter cases. xdist report serialization is covered by integration tests; parallel completion order is not guaranteed.

Values are opt-in: scalars and node IDs may contain sensitive data. Use synthetic inputs for evaluation. Non-scalar objects receive a placeholder without calling user repr; choose descriptive pytest IDs for those. Indirect fixture values are the input parameters, not the fixture's runtime return values.

Validation

pytest -q
python tools/reproduce.py src --details
python -m build
python -m twine check dist/*

The regression probe can also be run against an unmodified upstream checkout: python tools/reproduce.py /path/to/upstream/src. It deliberately fails when both passing parameter cases are absent from the report, after first proving that a failure survives a later pass. See validation scope. Regression tests block HTTP transport; inherited integration tests use only a localhost fake server with synthetic credentials.

Scope and next decisions

Native iterations, dataset creation, per-iteration evidence, cross-session attempt identity, parameter redaction and comment-size limits need further design. See the unexecuted native validation plan. For the legacy Xray payload only, the original status hierarchy is retained: ABORTED outranks FAIL, so a group containing a skip and a failure is ABORTED, with the failure visible in comments. This prototype does not silently turn that group into PASS, but does not change upstream's semantics to fail-dominant either. Expected failures and unexpected passes retain upstream mappings. Publishing error exit behavior is also inherited. The independent offline report uses the conservative failure-first rules documented in CASE-REPORT.md.

Read API decision notes, release proposal, and NOTICE. Original upstream instructions remain in README.rst.