Fix regression scanner attribution source

.github/scripts/Find-RegressionFixPRs.Tests.ps1

@@ -27,8 +27,9 @@ BeforeAll {
             'Get-UsableCandidateCount',
             'Invoke-GhJson',
             'Get-MergedRegressionFixPRs',
-            'Get-IssueAuthorAssociation',
+            'Get-FixPrContext',
             'Get-IssueContext',
+            'Get-RegressionAttributionContext',
             'Get-OpenRegressionCorpusPrTags',
             'Get-ExistingRegressionPrTags',
             'Get-IntroducingPrDetails',
@@ -413,7 +414,8 @@ Describe 'Get-MergedRegressionFixPRs' {
         Should -Invoke -CommandName Invoke-GhJson -Times 1 -Exactly -ParameterFilter {
             -not $AllowFailure -and
             ($GhArgs -join ' ') -match 'sort:created-asc' -and
-            ($GhArgs -join ' ') -match 'number,body,mergeCommit,mergedAt'
+            ($GhArgs -join ' ') -match 'number,mergeCommit,mergedAt' -and
+            ($GhArgs -join ' ') -notmatch '\bbody\b'
         }
     }
 }
@@ -476,33 +478,114 @@ Describe 'Get-IssueContext' {
     }
 }
 
-Describe 'Get-IssueAuthorAssociation' {
-    It 'reads the association from the issue REST resource' {
+Describe 'Get-FixPrContext' {
+    It 'reads the body and association atomically from the issue REST resource' {
         Mock Invoke-GhJson {
             param([string[]]$GhArgs)
-            $script:authorAssociationGhArgs = $GhArgs
-            return [PSCustomObject]@{ author_association = 'MEMBER' }
+            $script:fixPrContextGhArgs = $GhArgs
+            return [PSCustomObject]@{
+                number = 35803
+                body = "Fixes #35756`nRegression from #31931"
+                author_association = 'MEMBER'
+            }
         }
 
-        Get-IssueAuthorAssociation -Owner 'dotnet' -Repo 'maui' -Number 35803 |
-            Should -Be 'MEMBER'
-        $script:authorAssociationGhArgs | Should -Be @('api', 'repos/dotnet/maui/issues/35803')
+        $context = Get-FixPrContext -Owner 'dotnet' -Repo 'maui' -Number 35803
+
+        $context.Body | Should -Be "Fixes #35756`nRegression from #31931"
+        $context.AuthorAssociation | Should -Be 'MEMBER'
+        $context.IsTrustedAttribution | Should -BeTrue
+        $script:fixPrContextGhArgs | Should -Be @('api', 'repos/dotnet/maui/issues/35803')
         Should -Invoke -CommandName Invoke-GhJson -Times 1 -Exactly -ParameterFilter {
             $AllowFailure -and $GhArgs[0] -eq 'api'
         }
     }
     It 'treats an unavailable fix PR as untrusted' {
         Mock Invoke-GhJson { $null }
 
-        $association = Get-IssueAuthorAssociation -Owner 'dotnet' -Repo 'maui' -Number 35803
+        $context = Get-FixPrContext -Owner 'dotnet' -Repo 'maui' -Number 35803
 
-        ($null -eq $association) | Should -BeTrue
+        ($null -eq $context) | Should -BeTrue
         Should -Invoke -CommandName Invoke-GhJson -Times 1 -Exactly -ParameterFilter {
             $AllowFailure -and $GhArgs[0] -eq 'api'
         }
     }
 }
 
+Describe 'Get-RegressionAttributionContext' {
+    It 'uses the REST fix body when structural search results omit body' {
+        Mock Invoke-GhJson {
+            param([string[]]$GhArgs)
+            if ($GhArgs[0] -eq 'pr') {
+                return @([PSCustomObject]@{
+                        number = 35768
+                        mergeCommit = [PSCustomObject]@{ oid = 'fix-sha' }
+                        mergedAt = '2026-07-17T00:00:00Z'
+                    })
+            }
+            return [PSCustomObject]@{
+                number = 35768
+                body = "Fixes #35756`nRegression from #31931"
+                author_association = 'MEMBER'
+            }
+        }
+        Mock Get-IssueContext {
+            return [PSCustomObject]@{
+                Number = 35756
+                Labels = @('regressed-in-10.0.70')
+                Body = ''
+                CommentText = ''
+                IsTrustedAttribution = $false
+            }
+        }
+
+        $fixPr = @(Get-MergedRegressionFixPRs -Owner dotnet -Repo maui -LookbackDays 60 -Limit 1)[0]
+        $fixPr.PSObject.Properties.Name | Should -Not -Contain 'body'
+        $fixContext = Get-FixPrContext -Owner dotnet -Repo maui -Number $fixPr.number
+
+        $attribution = Get-RegressionAttributionContext `
+            -Owner dotnet `
+            -Repo maui `
+            -FixPr $fixPr.number `
+            -FixPrContext $fixContext
+
+        $attribution.LinkedIssueNumbers | Should -Be @(35756)
+        $attribution.IntroducingPr | Should -Be 31931
+        $attribution.AttributionSource | Should -Be 'pr-body'
+        $attribution.RegressionIssues.Count | Should -Be 1
+        Should -Invoke -CommandName Get-IssueContext -Times 1 -Exactly -ParameterFilter {
+            $Number -eq 35756
+        }
+    }
+
+    It 'uses an untrusted REST body for issue discovery but not attribution' {
+        $fixContext = [PSCustomObject]@{
+            Body = "Fixes #35756`nRegression from #31931"
+            AuthorAssociation = 'CONTRIBUTOR'
+            IsTrustedAttribution = $false
+        }
+        Mock Get-IssueContext {
+            return [PSCustomObject]@{
+                Number = 35756
+                Labels = @()
+                Body = ''
+                CommentText = ''
+                IsTrustedAttribution = $false
+            }
+        }
+
+        $attribution = Get-RegressionAttributionContext `
+            -Owner dotnet `
+            -Repo maui `
+            -FixPr 35768 `
+            -FixPrContext $fixContext
+
+        $attribution.LinkedIssueNumbers | Should -Be @(35756)
+        $attribution.IntroducingPr | Should -BeNullOrEmpty
+        $attribution.AttributionSource | Should -BeNullOrEmpty
+    }
+}
+
 Describe 'Get-OpenRegressionCorpusPrTags' {
     It 'includes tags from pending scanner draft PRs' {
         Mock Invoke-GhJson {

.github/scripts/Find-RegressionFixPRs.ps1

@@ -279,19 +279,26 @@ function Get-MergedRegressionFixPRs {
     $prs = Invoke-GhJson -GhArgs @(
         'pr', 'list', '--repo', "$Owner/$Repo",
         '--state', 'merged', '--search', $search, '--limit', "$Limit",
-        '--json', 'number,body,mergeCommit,mergedAt'
+        '--json', 'number,mergeCommit,mergedAt'
     )
     return @($prs | Where-Object { $null -ne $_ } | Sort-Object mergedAt, number)
 }
 
-function Get-IssueAuthorAssociation {
-    # REST exposes author_association for both issues and PRs (which are issues).
+function Get-FixPrContext {
+    # REST exposes a PR through the issue resource. Fetch the editable body and its
+    # trust signal together so attribution never combines differently filtered data.
     param([string]$Owner, [string]$Repo, [int]$Number)
     $issue = Invoke-GhJson -GhArgs @(
         'api', "repos/$Owner/$Repo/issues/$Number"
     ) -AllowFailure
     if (-not $issue) { return $null }
-    return [string]$issue.author_association
+    $association = [string]$issue.author_association
+    return [PSCustomObject]@{
+        Number               = $issue.number
+        Body                 = [string]$issue.body
+        AuthorAssociation    = $association
+        IsTrustedAttribution = Test-IsTrustedAssociation $association
+    }
 }
 
 function Get-IssueContext {
@@ -323,6 +330,70 @@ function Get-IssueContext {
     }
 }
 
+function Get-RegressionAttributionContext {
+    # Deterministic orchestration for linked-issue expansion and trusted-text
+    # attribution. Text stays internal; only structural fields reach candidates.
+    param(
+        [string]$Owner,
+        [string]$Repo,
+        [int]$FixPr,
+        $FixPrContext
+    )
+
+    $fixPrBody = if ($FixPrContext) { [string]$FixPrContext.Body } else { '' }
+    $linkedIssues = @(Get-LinkedIssueNumbers -PRBody $fixPrBody -Owner $Owner -Repo $Repo)
+    $commentSources = New-Object System.Collections.Generic.List[object]
+    $bodySources = New-Object System.Collections.Generic.List[object]
+    if ($FixPrContext -and $FixPrContext.IsTrustedAttribution -and $fixPrBody) {
+        $bodySources.Add([PSCustomObject]@{ Source = 'pr-body'; Text = $fixPrBody }) | Out-Null
+    }
+
+    $regressionIssues = New-Object System.Collections.Generic.List[object]
+    foreach ($issueNum in $linkedIssues) {
+        $ctx = Get-IssueContext -Owner $Owner -Repo $Repo -Number $issueNum
+        if (-not $ctx) { continue }
+        $regressedIn = @(Get-RegressedInLabels $ctx.Labels)
+        $regressionIssues.Add([PSCustomObject]@{
+                number            = $ctx.Number
+                regressedInLabels = $regressedIn
+            }) | Out-Null
+        if ($ctx.CommentText) {
+            $commentSources.Add([PSCustomObject]@{ Source = 'issue-comment'; Text = $ctx.CommentText }) | Out-Null
+        }
+        if ($ctx.IsTrustedAttribution) {
+            $bodySources.Add([PSCustomObject]@{ Source = 'issue-body'; Text = $ctx.Body }) | Out-Null
+        }
+    }
+
+    $attributionSources = New-Object System.Collections.Generic.List[object]
+    foreach ($source in $commentSources) {
+        $attributionSources.Add($source) | Out-Null
+    }
+    foreach ($source in $bodySources) {
+        $attributionSources.Add($source) | Out-Null
+    }
+
+    $introducingPr = $null
+    $attributionSource = $null
+    foreach ($entry in $attributionSources) {
+        $references = @(Get-IntroducingPrReferences $entry.Text | Where-Object {
+                $_ -ne $FixPr -and $_ -notin $linkedIssues
+            })
+        if ($references.Count -gt 0) {
+            $introducingPr = [int]$references[0]
+            $attributionSource = $entry.Source
+            break
+        }
+    }
+
+    return [PSCustomObject]@{
+        LinkedIssueNumbers = @($linkedIssues)
+        RegressionIssues   = $regressionIssues
+        IntroducingPr      = $introducingPr
+        AttributionSource  = $attributionSource
+    }
+}
+
 function Get-OpenRegressionCorpusPrTags {
     # Draft corpus PRs are not in main yet, so include their tags in deduplication.
     param([string]$Owner, [string]$Repo)
@@ -436,56 +507,16 @@ foreach ($pr in $fixPRs) {
     if ((Get-UsableCandidateCount -Candidates $candidates) -ge $MaxPRs) { break }
 
     $fixNumber = [int]$pr.number
-    $fixPrAssociation = Get-IssueAuthorAssociation -Owner $Owner -Repo $Repo -Number $fixNumber
-    $linkedIssues = Get-LinkedIssueNumbers -PRBody $pr.body -Owner $Owner -Repo $Repo
-
-    # Keep only maintainer-authored editable bodies as attribution. The fix body
-    # still drives linked-issue discovery above, regardless of author association.
-    $commentSources = New-Object System.Collections.Generic.List[object]
-    $bodySources = New-Object System.Collections.Generic.List[object]
-    if (Test-IsTrustedAssociation $fixPrAssociation) {
-        $bodySources.Add([PSCustomObject]@{ Source = 'pr-body'; Text = [string]$pr.body }) | Out-Null
-    }
-
-    $regressionIssues = New-Object System.Collections.Generic.List[object]
-    foreach ($issueNum in $linkedIssues) {
-        $ctx = Get-IssueContext -Owner $Owner -Repo $Repo -Number $issueNum
-        if (-not $ctx) { continue }
-        $regressedIn = @(Get-RegressedInLabels $ctx.Labels)
-        $regressionIssues.Add([PSCustomObject]@{
-            number            = $ctx.Number
-            regressedInLabels = $regressedIn
-        }) | Out-Null
-        if ($ctx.CommentText) {
-            $commentSources.Add([PSCustomObject]@{ Source = 'issue-comment'; Text = $ctx.CommentText }) | Out-Null
-        }
-        if ($ctx.IsTrustedAttribution) {
-            $bodySources.Add([PSCustomObject]@{ Source = 'issue-body'; Text = $ctx.Body }) | Out-Null
-        }
-    }
-
-    # An explicit trusted maintainer comment takes precedence over editable bodies.
-    $attribSources = New-Object System.Collections.Generic.List[object]
-    foreach ($source in $commentSources) {
-        $attribSources.Add($source) | Out-Null
-    }
-    foreach ($source in $bodySources) {
-        $attribSources.Add($source) | Out-Null
-    }
-
-    # Find the introducing PR reference, excluding the fix PR and its linked issues.
-    $introducingPr = $null
-    $attributionSource = $null
-    foreach ($entry in $attribSources) {
-        $refs = @(Get-IntroducingPrReferences $entry.Text | Where-Object {
-            $_ -ne $fixNumber -and $_ -notin $linkedIssues
-        })
-        if ($refs.Count -gt 0) {
-            $introducingPr = [int]$refs[0]
-            $attributionSource = $entry.Source
-            break
-        }
-    }
+    $fixPrContext = Get-FixPrContext -Owner $Owner -Repo $Repo -Number $fixNumber
+    $attribution = Get-RegressionAttributionContext `
+        -Owner $Owner `
+        -Repo $Repo `
+        -FixPr $fixNumber `
+        -FixPrContext $fixPrContext
+    $linkedIssues = @($attribution.LinkedIssueNumbers)
+    $regressionIssues = $attribution.RegressionIssues
+    $introducingPr = $attribution.IntroducingPr
+    $attributionSource = $attribution.AttributionSource
 
     if (-not (Test-CandidateIsNew -IntroducingPr $introducingPr -FixPr $fixNumber -ExistingNumbers @($existingNumbers))) {
         Write-Host "  ⏭️ PR #$fixNumber → introducing #$introducingPr already in corpus; skipping."