Fix milestone tag audit PR parsing

[PureWeen]

Note

Are you waiting for the changes in this PR to be merged?
It would be very helpful if you could test the resulting artifacts from this PR and let us know in a comment if this change resolves your issue. Thank you!

Description of Change

Root cause

Get-PrNumbersBetweenTags and Get-PrNumbersReachableFromTag treated every parenthesized (#N) reference in a commit subject as a merged PR. GitHub squash subjects can contain linked issue references before the actual PR suffix, so preview6 attempted to fetch issues #34716 and #34898 through the pulls API and logged two false 404 errors.

Fix

Add one shared Get-PrNumbersFromGitLog helper used by both tag paths. It accepts only the final numeric (#N) suffix (with optional trailing whitespace), ignores malformed or non-terminal references, and preserves sorted deduplication.

Preview6 proof

For 11.0.0-preview.5.26304.4..11.0.0-preview.6.26360.8:

• Before: 192 candidate references; false SourceGen: VisualStateManager.VisualStateGroups causes 'Names must be unique' at startup #34716 and Shell.Items.Clear() does not disconnect handlers correctly #34898 included.
• After: 190 candidate PRs; SourceGen: VisualStateManager.VisualStateGroups causes 'Names must be unique' at startup #34716 and Shell.Items.Clear() does not disconnect handlers correctly #34898 excluded while actual PRs Fix VisualStateGroups duplicate name crash with implicit styles (#34716) #34719 and Fix Shell.Items.Clear() memory leak by disconnecting child handlers on removal (#34898) #35031 remain.
• Real dry run exited 0 and resolved .NET 11.0-preview6 ([Enhancement] 3D graphics bindings #129).
• 39 relevant PRs checked, 151 wrong-branch PRs skipped, 15 linked issues checked, and 40 items already correct.
• All 14 legitimate milestone corrections remain, with zero report errors.
• -CloseFixedIssues remained a dry run: two open issues would close and 13 were already closed; no GitHub state was changed.

Tests

• New parser regression selection: 12 tests; restoring the former parser produces 6 failures.
• Fix-MilestoneDrift.Tests.ps1: 212 passed.
• MilestoneTrigger.Tests.ps1: 116 passed.

Non-goals

This fixes tag-audit parsing on main. It does not address the missed historical preview6 workflow launch: the preview6 tag was created from a net11.0 commit that did not yet contain the already-merged main tag-push trigger. Branch-flow/backport policy is intentionally unchanged.

Issues Fixed

N/A — this defect was discovered while auditing the preview6 tag.

[github-actions]

🚀 Dogfood this PR with:

⚠️ WARNING: Do not do this without first carefully reviewing the code of this PR to satisfy yourself it is safe.

curl -fsSL https://raw.githubusercontent.com/dotnet/maui/main/eng/scripts/get-maui-pr.sh | bash -s -- 36703

Or

• Run remotely in PowerShell:

iex "& { $(irm https://raw.githubusercontent.com/dotnet/maui/main/eng/scripts/get-maui-pr.ps1) } 36703"

[azure-pipelines]

Azure Pipelines:
1 pipeline(s) were filtered out due to trigger conditions.
There may be pipelines that require an authorized user to comment /azp run to run.

[Copilot]

Pull request overview

This PR fixes false-positive “merged PR” detections in the milestone drift/tag-audit tooling by tightening how PR numbers are extracted from git log output, preventing linked-issue references in commit subjects from being treated as PRs.

Changes:

• Added a shared Get-PrNumbersFromGitLog helper that only accepts a terminal (#N) PR suffix (with optional trailing whitespace) and deduplicates/sorts results.
• Updated both tag-walk paths (Get-PrNumbersBetweenTags, Get-PrNumbersReachableFromTag) to use the shared terminal-suffix parser.
• Added unit and git-backed integration tests covering multiple-parenthesis subjects, malformed suffixes, trailing whitespace, deduplication, and the two tag-based call paths.

Show a summary per file

File	Description
.github/scripts/Fix-MilestoneDrift.ps1	Centralizes PR-number parsing from git log and restricts matches to terminal squash-merge suffixes to avoid false PR lookups.
.github/scripts/Fix-MilestoneDrift.Tests.ps1	Adds regression and unmocked git integration tests ensuring both tag-based extraction paths share the corrected terminal-suffix behavior.

Copilot's findings

• Files reviewed: 2/2 changed files
• Comments generated: 0

[kubaflo]

🔍 AI-generated review — automated multi-model consensus (Claude Opus 4.8 · GPT-5.5 · Gemini 3.1 Pro), run on behalf of @kubaflo. Independent reviews → cross-pollinated → synthesized. This is a COMMENT (not an approval).

✅ LGTM (3-model unanimous, high confidence)

A tightly-scoped, correct fix. The root cause is real and the anchored regex is the right remedy. Every claim was independently verified — I ran the Pester suite locally and all three models fetched fresh at head 16c4946:

Check	Result
Anchored regex \(#(\d+)\)\s*$ captures only the terminal squash suffix	✅ Opus scanned ~200 real maui subjects; the regression signature is real — e.g. [Regression] Fix TypedBinding nested property re-subscription (#34428) (#34449): old greedy code queried linked-ref #34428 via the pulls API (→ false 404), new code keeps only #34449
Overflow safety ([int]::TryParse vs old throwing [int] cast)	✅ strict improvement — (#999999999999999999999999999999) is skipped, not thrown; \d++TryParse cannot throw
CRLF edge — trailing \r after -split "n"`	✅ non-issue: \r ∈ \s, so \s*$ still anchors; verified directly and with a real repo at core.autocrlf true → returns only the terminal PR
Behavioral equivalence of both tag functions	✅ both delegate to the shared Get-PrNumbersFromGitLog; git exit-code throw, HashSet dedup, Sort-Object all preserved
No external callers relying on old greedy behavior	✅ both functions are internal to Fix-MilestoneDrift.ps1; all call sites use the result only for -in/count membership; the two other file-matches are comment-only
Test suite	✅ 212/212 pass (incl. the 12 new cases); I re-ran the new blocks: 10 unit + 2 unmocked git init integration tests, all green

The new tests are genuinely good — they cover the exact bug (issue-ref-before-PR), malformed/non-numeric/overflow refs, trailing whitespace, dedup+sort, and prove parity on real git log output via the unmocked integration block.

Non-blocking (low, pre-existing — @PureWeen your call)

• The .Tests.ps1 file isn't wired into CI. No Invoke-Pester exists in any of the 37 workflows, and fix-milestone-drift.yml runs the script but not its tests — so these (valuable) tests give no automated regression protection. This is pre-existing (all sibling *.Tests.ps1 are equally un-wired) and not introduced by this PR, so it doesn't block. Optional: add a workflow step running Invoke-Pester over .github/scripts/*.Tests.ps1 on PRs touching that folder.

ℹ️ Per policy this bot posts a comment, not a formal approval — merge is @kubaflo's call.

_{Models: claude-opus-4.8 (LGTM) · gpt-5.5 (LGTM) · gemini-3.1-pro-preview (LGTM) · consensus: LGTM · confidence: high · findings: 1 low pre-existing / 0 blocking}

[kubaflo]

LGTM — approving.

Independent 3-model review (Claude Opus 4.8 / GPT-5.5 / Gemini 3.1 Pro) reached unanimous LGTM with no findings. The fix is correct and well-scoped:

• Terminal-anchor is right. \(#(\d+)\)\s*$ targets exactly the GitHub squash-merge terminal PR token, so backport subjects like … (#34898) (#35031) now resolve to only the PR that landed on the branch (35031) instead of also fetching the linked issue via the pulls API — which was the source of the false 404s on preview6.
• Real latent-bug fix. Replacing [int]$m.Groups[1].Value with [int]::TryParse prevents an OverflowException crash of the whole audit on a huge digit run (covered by the (#999999999999999999999999999999) test).
• CRLF-safe. \s*$ natively consumes any \r left by -split "n", so no RegexOptions.Multiline` is needed.
• No downstream risk. Dropping non-terminal references is exactly the intended milestone-drift behavior; the two callers delegate to the shared helper with unchanged signatures.

CI-gate: no required checks are red (maui-pr is correctly path-skipped for a .github/scripts change; only the advisory Build Analysis is pending). I also ran the added Pester block locally — 12/12 new tests pass (10 unit + 2 unmocked temp-git-repo integration asserting 34719,35031 between-tags and 100,34719,35031 reachable).

🤖 AI-generated review (GitHub Copilot CLI, on behalf of @kubaflo).

[PureWeen]

Adversarial review: no issues found

Three independent reviewers examined the two authoritative changed files, full parser call paths, boundary/overflow behavior, and test isolation. No correctness, security, or regression findings survived consensus.

The added coverage exercises the original issue-reference-before-PR defect, malformed and oversized references, trailing whitespace, deduplication/sorting, and both git-log callers. Existing reviews contained no unresolved error-level findings.