fix(cli): accept --swift-test-xunit-paths as the only report source

[dfrankland]

The bug

--swift-test-xunit-paths is absent from every required_unless_present_any list, so the usage its own help text documents does not parse:

$ trunk-analytics-cli upload --swift-test-xunit-paths=xunit-swift-testing.xml --org-url-slug=o --token=t
error: the following required arguments were not provided:
  --junit-paths <JUNIT_PATHS>
  --bazel-bep-path <BAZEL_BEP_PATH>
  --test-reports <TEST_REPORTS>

It is a real report source — coalesce_junit_path_wrappers handles it on its own branch and produces junit path wrappers from it — so it belongs in those lists alongside the other three. Adding it there is the whole fix.

Why the workaround is worse than the error

Callers have to supply a junit glob they do not want. The obvious value, the report they are already uploading, silently doubles every test: the two lists are appended rather than reconciled, so each test arrives once carrying the file a language server resolved and once carrying none.

Measured against 0.15.5-beta.1 on a 3-case suite, from a checkout with no Swift toolchain so nothing resolved:

Invocation	Cases bundled
--junit-paths '<report glob>'	3
--swift-test-xunit-paths <report> + a glob matching nothing	3
both pointed at the real report	6

file feeds gen_info_id, so on a real checkout those six are two distinct tests rather than one test twice. Every consumer hitting this requirement has to pick between a glob they invented and a doubled upload.

The test could not have caught either problem

upload_bundle_using_swift_test_xunit inherits CommandBuilder's default --junit-paths ./*. That hid the missing requirement, and it also matches the two xunit files the test writes into the repo root — so the test has been exercising the double-upload path since it landed. Its assertions keyed a HashMap by test name, and a duplicate carries the same name, so the second copy overwrote the first and nothing looked wrong.

It now passes the reports through a new PathsState::SwiftTestXunitPaths variant and asserts the exact case count. PathsState is the report-source selector, so putting the flag there is the harness stating the same thing this change does — a fourth source alongside junit, BEP, and xcresult — rather than smuggling it past the selector through extra_args. Selecting it also suppresses the default glob, which is what makes this a genuine test of standalone usage, and the count gives the duplication somewhere to surface if it returns.

Not fixed here

If a caller passes an explicit --junit-paths glob that happens to match a file also named in --swift-test-xunit-paths, the double upload still happens. That is a live footgun for a monorepo using a broad glob like **/*.xml next to a Swift suite.

Fixing it properly means excluding the swift-handled files from the expanded junit set, which has to happen inside FileSetBuilder — the swift path uploads a rewritten temp copy, so the duplicate pair have different paths and a canonical-path dedup would not see them. That is a threaded exclusion parameter through build_file_sets, and a separate concern from this argument-parsing fix. Happy to follow up if you want it.

Verification

cargo check and the test suite have not been run locally — this environment has neither the pinned nightly-2026-06-11 toolchain nor the system OpenSSL openssl-sys needs, and I would rather say so than imply a green run. The change is four clap attribute lists, one enum variant with its match arm, and test assertions. CI is the gate; please don't merge on my word that it compiles.

The parsing behaviour above was verified by running the released 0.15.5-beta.1 binary directly.

🤖 Generated with Claude Code

[trunk-io]

😎 Merged successfully - details.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 83.85%. Comparing base (8396875) to head (63fcdc8).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1198      +/-   ##
==========================================
+ Coverage   83.63%   83.85%   +0.22%     
==========================================
  Files          74       74              
  Lines       17573    17573              
==========================================
+ Hits        14697    14736      +39     
+ Misses       2876     2837      -39     

☔ 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.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[trunk-staging-io]

   

Failed Test	Failure Summary	Logs
pending_quarantine_test should be quarantined when run with variant	A test marked as pending was expected to fail but unexpectedly passed.	Logs ↗︎
variant_quarantine_test should be quarantined when run with variant	A test expected the sum of 2 + 2 to be 5, but it was actually 4, indicating a failing assertion.	Logs ↗︎

View Full Report ↗︎ ⋅ Docs

[trunk-io]

   

View Full Report ↗︎ ⋅ Docs