Fix 12 Pester 6 milestone issues

[Copilot]

Summary

Fixes 12 open issues from the Pester 6 milestone across several areas of the codebase.

Changes

Assertions

• The -Exist parameter of Should cannot find the file if the file name contains a wildcard character #2628 — Should -Exist fails when path contains wildcard characters ([, ]): Changed Test-Path $ActualValue to Test-Path -LiteralPath $ActualValue so paths with literal bracket characters are found correctly.

• Infinite loop for FileInfo and DirectoryInfo due to Root properties in DirectoryInfo. Should we restrict depth as a general fix in addition to specialcasing DirectoryInfo in Get-DisplayProperty2? #2474 — Infinite loop when formatting FileInfo/DirectoryInfo objects (via Root circular property): Added System.IO.DirectoryInfo and System.IO.FileInfo to the display property map in Get-DisplayProperty2.

Test Results / Coverage Output

• TestResults.OutputEncoding option ignored #2452 — TestResults.OutputEncoding configuration option was completely ignored (always wrote UTF-8): Added $Encoding parameter to Export-PesterResult and Export-XmlReport; XmlWriterSettings.Encoding is now set from the configured encoding name.

• Invoke-Pester breaks on invalid CodeCoverage.OutputEncoding format #2451 — Invalid CodeCoverage.OutputEncoding caused an ugly exception at the very end of Invoke-Pester: Added early encoding validation in Resolve-CodeCoverageConfiguration. Also added the same early validation to Resolve-TestResultConfiguration for TestResults.OutputEncoding.

• Get-CimInstance (non failing) error when running Pester not as Administrator #2678 — Get-CimInstance : Access denied error printed when Pester runs without Administrator rights: Added -ErrorAction Ignore to both Get-CimInstance and Get-WmiObject calls in Get-RunTimeEnvironment, with a null fallback for the OS info.

• Error during handling of output-element with state based ToString() Implementation #2649 — NUnit3 report writing crashes when a test output object's ToString() method throws: Wrapped the ToString() call in Format-CDataString with a try/catch that substitutes a descriptive fallback string.

Runtime / Discovery

• Error in BeforeAll of Data driven tests leads to missing info in the error output #2571 — Data-driven block (-ForEach) name template (e.g. <_>) is not expanded in error output when BeforeAll fails: Moved the block-name expansion scriptblock in OuterSetup to run before OneTimeTestSetup (BeforeAll), so the name is always expanded even on BeforeAll failure.

• Test discovery on linux fails to find tests in hidden folders #2515 — Test discovery on Linux skips hidden (dot-prefixed) folders like .github: Added -Force to both Get-ChildItem calls in Find-File.

• Relative paths for reports are relative to current folder after all tests completed, not to the original pester folder. #2641 — Relative output report paths resolve against the current directory after all tests run (which may differ from the invocation directory if tests call Set-Location): Added a Start plugin step to Get-TestResultPlugin that resolves the output path to an absolute path early, before tests execute.

• Pester clears test drives inefficiently #2657 — Clear-TestDrive was extremely slow with many files because it called Remove-Item on every individual file, including files already deleted by earlier recursive calls: Replaced the naive per-item loop with a HashSet-based approach that only calls Remove-Item -Recurse on "root" new items (items whose parent is not also new).

Configuration / Usability

• setting any option to non default on testResult or code coverage should enable them #2682 — Users must explicitly set Enabled = $true even when configuring other TestResult/CodeCoverage options: Added auto-enable logic in Invoke-Pester that uses the IsModified property to detect user-configured options and automatically enables the feature.

• Container is reported as Passed/Successful when discovery fails #2538 — Container and block containing a discovery error are reported as Passed/Successful in the result object: Reordered the result-computation logic in PostProcess-RSpecTestRun so that a non-empty ErrorRecord always yields Failed, regardless of Passed.

Security Summary

No security vulnerabilities were introduced or identified in these changes.