microsoft/hve-core
Publicmirrored from https://github.com/microsoft/hve-coreAvailable
scripts/collections/Validate-Collections.ps1
453lines · modecode
| 1 | #!/usr/bin/env pwsh |
| 2 | # Copyright (c) Microsoft Corporation. |
| 3 | # SPDX-License-Identifier: MIT |
| 4 | #Requires -Version 7.0 |
| 5 | |
| 6 | <# |
| 7 | .SYNOPSIS |
| 8 | Validates collection manifests for Copilot CLI plugin generation. |
| 9 | |
| 10 | .DESCRIPTION |
| 11 | Reads all .collection.yml files from collections/ and validates structure, |
| 12 | required fields, artifact path existence, and kind-suffix consistency. |
| 13 | |
| 14 | .EXAMPLE |
| 15 | ./Validate-Collections.ps1 |
| 16 | #> |
| 17 | |
| 18 | [CmdletBinding()] |
| 19 | param() |
| 20 | |
| 21 | $ErrorActionPreference = 'Stop' |
| 22 | |
| 23 | Import-Module (Join-Path $PSScriptRoot 'Modules/CollectionHelpers.psm1') -Force |
| 24 | Import-Module (Join-Path $PSScriptRoot '../lib/Modules/CIHelpers.psm1') -Force |
| 25 | |
| 26 | #region Validation Helpers |
| 27 | |
| 28 | function Test-KindSuffix { |
| 29 | <# |
| 30 | .SYNOPSIS |
| 31 | Validates that an item path matches its declared kind suffix. |
| 32 | |
| 33 | .DESCRIPTION |
| 34 | Checks kind-suffix consistency: agent files end with .agent.md, |
| 35 | prompt files with .prompt.md, instruction files with .instructions.md, |
| 36 | and skill items are directories containing a SKILL.md file. |
| 37 | |
| 38 | .PARAMETER Kind |
| 39 | The declared artifact kind (agent, prompt, instruction, skill). |
| 40 | |
| 41 | .PARAMETER ItemPath |
| 42 | The relative path from the collection manifest. |
| 43 | |
| 44 | .PARAMETER RepoRoot |
| 45 | Absolute path to the repository root for skill directory checks. |
| 46 | |
| 47 | .OUTPUTS |
| 48 | [string] Error message if validation fails, empty string if valid. |
| 49 | #> |
| 50 | [CmdletBinding()] |
| 51 | [OutputType([string])] |
| 52 | param( |
| 53 | [Parameter(Mandatory = $true)] |
| 54 | [string]$Kind, |
| 55 | |
| 56 | [Parameter(Mandatory = $true)] |
| 57 | [string]$ItemPath, |
| 58 | |
| 59 | [Parameter(Mandatory = $true)] |
| 60 | [string]$RepoRoot |
| 61 | ) |
| 62 | |
| 63 | switch ($Kind) { |
| 64 | 'agent' { |
| 65 | if ($ItemPath -notmatch '\.agent\.md$') { |
| 66 | return "kind 'agent' expects *.agent.md but got '$ItemPath'" |
| 67 | } |
| 68 | } |
| 69 | 'prompt' { |
| 70 | if ($ItemPath -notmatch '\.prompt\.md$') { |
| 71 | return "kind 'prompt' expects *.prompt.md but got '$ItemPath'" |
| 72 | } |
| 73 | } |
| 74 | 'instruction' { |
| 75 | if ($ItemPath -notmatch '\.instructions\.md$') { |
| 76 | return "kind 'instruction' expects *.instructions.md but got '$ItemPath'" |
| 77 | } |
| 78 | } |
| 79 | 'skill' { |
| 80 | $skillDir = Join-Path -Path $RepoRoot -ChildPath $ItemPath |
| 81 | $skillFile = Join-Path -Path $skillDir -ChildPath 'SKILL.md' |
| 82 | if (-not (Test-Path -Path $skillFile)) { |
| 83 | return "kind 'skill' expects SKILL.md inside '$ItemPath'" |
| 84 | } |
| 85 | } |
| 86 | } |
| 87 | |
| 88 | return '' |
| 89 | } |
| 90 | |
| 91 | function Get-CollectionItemKey { |
| 92 | <# |
| 93 | .SYNOPSIS |
| 94 | Builds a stable uniqueness key for collection items. |
| 95 | |
| 96 | .DESCRIPTION |
| 97 | Uses kind and path to identify the same artifact across collections. |
| 98 | |
| 99 | .PARAMETER Kind |
| 100 | Artifact kind. |
| 101 | |
| 102 | .PARAMETER ItemPath |
| 103 | Artifact path. |
| 104 | |
| 105 | .OUTPUTS |
| 106 | [string] Composite key. |
| 107 | #> |
| 108 | [CmdletBinding()] |
| 109 | [OutputType([string])] |
| 110 | param( |
| 111 | [Parameter(Mandatory = $true)] |
| 112 | [string]$Kind, |
| 113 | |
| 114 | [Parameter(Mandatory = $true)] |
| 115 | [string]$ItemPath |
| 116 | ) |
| 117 | |
| 118 | return "$Kind|$ItemPath" |
| 119 | } |
| 120 | |
| 121 | #endregion Validation Helpers |
| 122 | |
| 123 | #region Orchestration |
| 124 | |
| 125 | function Invoke-CollectionValidation { |
| 126 | <# |
| 127 | .SYNOPSIS |
| 128 | Validates all collection manifests for correctness. |
| 129 | |
| 130 | .DESCRIPTION |
| 131 | Scans the collections/ directory for .collection.yml files and validates |
| 132 | each manifest for required fields (id, name, description, items), id |
| 133 | format, artifact path existence, kind-suffix consistency, and duplicate |
| 134 | ids across collections. |
| 135 | |
| 136 | .PARAMETER RepoRoot |
| 137 | Absolute path to the repository root directory. |
| 138 | |
| 139 | .OUTPUTS |
| 140 | Hashtable with Success bool, ErrorCount int, and CollectionCount int. |
| 141 | #> |
| 142 | [CmdletBinding()] |
| 143 | [OutputType([hashtable])] |
| 144 | param( |
| 145 | [Parameter(Mandatory = $true)] |
| 146 | [ValidateNotNullOrEmpty()] |
| 147 | [string]$RepoRoot |
| 148 | ) |
| 149 | |
| 150 | $collectionsDir = Join-Path -Path $RepoRoot -ChildPath 'collections' |
| 151 | $collectionFiles = Get-ChildItem -Path $collectionsDir -Filter '*.collection.yml' -File |
| 152 | |
| 153 | if ($collectionFiles.Count -eq 0) { |
| 154 | Write-Host ' WARN No collection manifests found in collections/' -ForegroundColor Yellow |
| 155 | return @{ Success = $true; ErrorCount = 0; CollectionCount = 0 } |
| 156 | } |
| 157 | |
| 158 | Write-Host 'Validating collections...' |
| 159 | |
| 160 | $errorCount = 0 |
| 161 | $seenIds = @{} |
| 162 | $validatedCount = 0 |
| 163 | $allowedMaturities = @('stable', 'preview', 'experimental', 'deprecated') |
| 164 | $canonicalCollectionId = 'hve-core-all' |
| 165 | $itemOccurrences = @{} |
| 166 | |
| 167 | $knownCollectionIds = @{} |
| 168 | foreach ($cf in $collectionFiles) { |
| 169 | $cfId = $cf.Name -replace '\.collection\.yml$', '' |
| 170 | $knownCollectionIds[$cfId] = $true |
| 171 | } |
| 172 | |
| 173 | foreach ($file in $collectionFiles) { |
| 174 | $baseName = $file.Name -replace '\.collection\.yml$', '' |
| 175 | $companionPath = Join-Path -Path $collectionsDir -ChildPath "$baseName.collection.md" |
| 176 | if (-not (Test-Path -Path $companionPath)) { |
| 177 | Write-Host " WARN $($file.Name): missing companion '$baseName.collection.md'" -ForegroundColor Yellow |
| 178 | } |
| 179 | |
| 180 | if (Test-Path -Path $companionPath) { |
| 181 | $mdContent = Get-Content -Path $companionPath -Raw |
| 182 | $hasBegin = $mdContent.Contains($CollectionMdBeginMarker) |
| 183 | $hasEnd = $mdContent.Contains($CollectionMdEndMarker) |
| 184 | |
| 185 | if ($hasBegin -xor $hasEnd) { |
| 186 | Write-Host " WARN $($file.Name): $baseName.collection.md has mismatched auto-generation markers" -ForegroundColor Yellow |
| 187 | } |
| 188 | |
| 189 | if ($hasBegin -and $hasEnd) { |
| 190 | $beginIdx = $mdContent.IndexOf($CollectionMdBeginMarker) |
| 191 | $endIdx = $mdContent.IndexOf($CollectionMdEndMarker) |
| 192 | if ($endIdx -le $beginIdx) { |
| 193 | Write-Host " WARN $($file.Name): $baseName.collection.md has markers in wrong order" -ForegroundColor Yellow |
| 194 | } |
| 195 | } |
| 196 | } |
| 197 | |
| 198 | $manifest = Get-CollectionManifest -CollectionPath $file.FullName |
| 199 | $fileErrors = @() |
| 200 | $seenItemKeys = @{} |
| 201 | |
| 202 | # Required fields |
| 203 | $requiredFields = @('id', 'name', 'description', 'items') |
| 204 | foreach ($field in $requiredFields) { |
| 205 | if (-not $manifest.ContainsKey($field) -or $null -eq $manifest[$field]) { |
| 206 | $fileErrors += "missing required field '$field'" |
| 207 | } |
| 208 | } |
| 209 | |
| 210 | # Skip further checks if required fields are absent |
| 211 | if ($fileErrors.Count -gt 0) { |
| 212 | foreach ($err in $fileErrors) { |
| 213 | Write-Host " x $($file.Name): $err" -ForegroundColor Red |
| 214 | } |
| 215 | $errorCount += $fileErrors.Count |
| 216 | continue |
| 217 | } |
| 218 | |
| 219 | $id = $manifest.id |
| 220 | |
| 221 | # Id format |
| 222 | if ($id -notmatch '^[a-z0-9-]+$') { |
| 223 | $fileErrors += "id '$id' must match ^[a-z0-9-]+$" |
| 224 | } |
| 225 | |
| 226 | # Duplicate id check |
| 227 | if ($seenIds.ContainsKey($id)) { |
| 228 | $fileErrors += "duplicate id '$id' (also in $($seenIds[$id]))" |
| 229 | } |
| 230 | else { |
| 231 | $seenIds[$id] = $file.Name |
| 232 | } |
| 233 | |
| 234 | # Validate collection-level maturity if present |
| 235 | if ($manifest.ContainsKey('maturity') -and -not [string]::IsNullOrWhiteSpace([string]$manifest.maturity)) { |
| 236 | $collMaturity = [string]$manifest.maturity |
| 237 | if ($allowedMaturities -notcontains $collMaturity) { |
| 238 | $fileErrors += "invalid collection maturity '$collMaturity' (allowed: $($allowedMaturities -join ', '))" |
| 239 | } |
| 240 | } |
| 241 | |
| 242 | # Validate each item |
| 243 | $itemCount = $manifest.items.Count |
| 244 | foreach ($item in $manifest.items) { |
| 245 | $itemPath = $item.path |
| 246 | $kind = $item.kind |
| 247 | $absolutePath = Join-Path -Path $RepoRoot -ChildPath $itemPath |
| 248 | $itemMaturity = $null |
| 249 | if ($item.ContainsKey('maturity')) { |
| 250 | $itemMaturity = [string]$item.maturity |
| 251 | } |
| 252 | $effectiveMaturity = Resolve-CollectionItemMaturity -Maturity $itemMaturity |
| 253 | |
| 254 | # Repo-specific path exclusion |
| 255 | if (Test-HveCoreRepoRelativePath -Path $itemPath) { |
| 256 | $fileErrors += "repo-specific path not allowed in collections: $itemPath (root-level artifacts under .github/{type}/ are excluded from distribution)" |
| 257 | } |
| 258 | |
| 259 | # Path existence |
| 260 | if (-not (Test-Path -Path $absolutePath)) { |
| 261 | $fileErrors += "path not found: $itemPath" |
| 262 | } |
| 263 | |
| 264 | # Kind-suffix consistency |
| 265 | if ($kind) { |
| 266 | $suffixError = Test-KindSuffix -Kind $kind -ItemPath $itemPath -RepoRoot $RepoRoot |
| 267 | if ($suffixError) { |
| 268 | $fileErrors += $suffixError |
| 269 | } |
| 270 | } |
| 271 | else { |
| 272 | $fileErrors += "item missing 'kind': $itemPath" |
| 273 | } |
| 274 | |
| 275 | if (-not [string]::IsNullOrWhiteSpace($itemMaturity) -and ($allowedMaturities -notcontains $itemMaturity)) { |
| 276 | $fileErrors += "invalid maturity '$itemMaturity' for item '$itemPath' (allowed: $($allowedMaturities -join ', '))" |
| 277 | } |
| 278 | |
| 279 | # Check 2: intra-collection duplicate detection |
| 280 | if (-not [string]::IsNullOrWhiteSpace($itemPath) -and -not [string]::IsNullOrWhiteSpace($kind)) { |
| 281 | $dupKey = Get-CollectionItemKey -Kind $kind -ItemPath $itemPath |
| 282 | if ($seenItemKeys.ContainsKey($dupKey)) { |
| 283 | $fileErrors += "duplicate item '$dupKey' appears more than once in collection '$id'" |
| 284 | } else { |
| 285 | $seenItemKeys[$dupKey] = $true |
| 286 | } |
| 287 | } |
| 288 | |
| 289 | # Check 3: collection-id to folder name consistency |
| 290 | if ($id -ne 'hve-core-all') { |
| 291 | $pathSegments = $itemPath -split '[/\\]' |
| 292 | # Expected pattern: .github/{type}/{collection-id}/{file-or-deeper} |
| 293 | if ($pathSegments.Count -ge 4 -and $pathSegments[0] -eq '.github') { |
| 294 | $folderName = $pathSegments[2] |
| 295 | if ($folderName -ne 'shared' -and -not $knownCollectionIds.ContainsKey($folderName)) { |
| 296 | Write-Host " WARN collection '$id': item folder '$folderName' does not match any known collection ID: $itemPath" -ForegroundColor Yellow |
| 297 | } |
| 298 | } |
| 299 | } |
| 300 | |
| 301 | if (-not [string]::IsNullOrWhiteSpace($itemPath) -and -not [string]::IsNullOrWhiteSpace($kind)) { |
| 302 | $itemKey = Get-CollectionItemKey -Kind $kind -ItemPath $itemPath |
| 303 | if (-not $itemOccurrences.ContainsKey($itemKey)) { |
| 304 | $itemOccurrences[$itemKey] = @() |
| 305 | } |
| 306 | |
| 307 | $itemOccurrences[$itemKey] += @{ |
| 308 | CollectionId = $id |
| 309 | CollectionFile = $file.Name |
| 310 | Kind = $kind |
| 311 | Path = $itemPath |
| 312 | Maturity = $effectiveMaturity |
| 313 | } |
| 314 | } |
| 315 | |
| 316 | # Informational log for instruction items |
| 317 | if ($kind -eq 'instruction') { |
| 318 | Write-Verbose " instruction: $itemPath" |
| 319 | } |
| 320 | } |
| 321 | |
| 322 | if ($fileErrors.Count -gt 0) { |
| 323 | Write-Host " FAIL $id ($itemCount items) - $($fileErrors.Count) error(s)" -ForegroundColor Red |
| 324 | foreach ($err in $fileErrors) { |
| 325 | Write-Host " $err" -ForegroundColor Red |
| 326 | } |
| 327 | $errorCount += $fileErrors.Count |
| 328 | } |
| 329 | else { |
| 330 | Write-Host " OK $id ($itemCount items)" |
| 331 | } |
| 332 | |
| 333 | $validatedCount++ |
| 334 | } |
| 335 | |
| 336 | $canonicalManifestFound = ($collectionFiles | Where-Object { |
| 337 | ($_.Name -replace '\.collection\.yml$', '') -eq $canonicalCollectionId |
| 338 | }).Count -gt 0 |
| 339 | if (-not $canonicalManifestFound) { |
| 340 | Write-Host " WARN '$canonicalCollectionId.collection.yml' not found; skipping orphan and cross-collection coverage checks" -ForegroundColor Yellow |
| 341 | } |
| 342 | |
| 343 | # Duplicate artifact key detection across all collections |
| 344 | $artifactKeyMap = @{} |
| 345 | foreach ($itemKey in $itemOccurrences.Keys) { |
| 346 | $occurrences = $itemOccurrences[$itemKey] |
| 347 | $first = $occurrences[0] |
| 348 | $artifactKey = Get-CollectionArtifactKey -Kind $first.Kind -Path $first.Path |
| 349 | $compositeKey = "$($first.Kind)|$artifactKey" |
| 350 | |
| 351 | if (-not $artifactKeyMap.ContainsKey($compositeKey)) { |
| 352 | $artifactKeyMap[$compositeKey] = @() |
| 353 | } |
| 354 | if ($artifactKeyMap[$compositeKey] -notcontains $first.Path) { |
| 355 | $artifactKeyMap[$compositeKey] += $first.Path |
| 356 | } |
| 357 | } |
| 358 | |
| 359 | foreach ($compositeKey in $artifactKeyMap.Keys) { |
| 360 | $paths = $artifactKeyMap[$compositeKey] |
| 361 | if ($paths.Count -gt 1) { |
| 362 | $kindLabel = ($compositeKey -split '\|')[0] |
| 363 | $nameLabel = ($compositeKey -split '\|')[1] |
| 364 | $pathList = ($paths | Sort-Object) -join ', ' |
| 365 | Write-Host " FAIL duplicate $kindLabel artifact key '$nameLabel' found at distinct paths: $pathList" -ForegroundColor Red |
| 366 | $errorCount++ |
| 367 | } |
| 368 | } |
| 369 | |
| 370 | foreach ($itemKey in $itemOccurrences.Keys) { |
| 371 | $occurrences = $itemOccurrences[$itemKey] |
| 372 | $canonicalMatches = @($occurrences | Where-Object { $_.CollectionId -eq $canonicalCollectionId }) |
| 373 | $themedMatches = @($occurrences | Where-Object { $_.CollectionId -ne $canonicalCollectionId }) |
| 374 | |
| 375 | # Check 4: item in one or more themed collections but absent from hve-core-all |
| 376 | if ($canonicalManifestFound -and $themedMatches.Count -gt 0 -and $canonicalMatches.Count -eq 0) { |
| 377 | $themedCollections = ($themedMatches | ForEach-Object { $_.CollectionId } | Sort-Object -Unique) -join ', ' |
| 378 | Write-Host " FAIL item '$itemKey' exists in themed collection(s) [$themedCollections] but is absent from '$canonicalCollectionId'" -ForegroundColor Red |
| 379 | $errorCount++ |
| 380 | continue |
| 381 | } |
| 382 | |
| 383 | # Maturity conflict: only when item appears in canonical AND at least one themed |
| 384 | if ($canonicalMatches.Count -gt 0 -and $themedMatches.Count -gt 0) { |
| 385 | $canonical = $canonicalMatches[0] |
| 386 | foreach ($occurrence in $themedMatches) { |
| 387 | if ($occurrence.Maturity -ne $canonical.Maturity) { |
| 388 | Write-Host " FAIL maturity conflict for '$itemKey': canonical '$canonicalCollectionId'='$($canonical.Maturity)', '$($occurrence.CollectionId)'='$($occurrence.Maturity)'" -ForegroundColor Red |
| 389 | $errorCount++ |
| 390 | } |
| 391 | } |
| 392 | } |
| 393 | } |
| 394 | |
| 395 | if ($canonicalManifestFound) { |
| 396 | # Check 1: Orphan artifact detection |
| 397 | $onDiskArtifacts = Get-ArtifactFiles -RepoRoot $RepoRoot |
| 398 | foreach ($artifact in $onDiskArtifacts) { |
| 399 | $diskKey = Get-CollectionItemKey -Kind $artifact.kind -ItemPath $artifact.path |
| 400 | $occurrences = if ($itemOccurrences.ContainsKey($diskKey)) { $itemOccurrences[$diskKey] } else { @() } |
| 401 | |
| 402 | $inCanonical = @($occurrences | Where-Object { $_.CollectionId -eq $canonicalCollectionId }).Count -gt 0 |
| 403 | $inThemed = @($occurrences | Where-Object { $_.CollectionId -ne $canonicalCollectionId }).Count -gt 0 |
| 404 | |
| 405 | if (-not $inCanonical) { |
| 406 | Write-Host " FAIL orphan: '$diskKey' is on disk but absent from '$canonicalCollectionId'" -ForegroundColor Red |
| 407 | $errorCount++ |
| 408 | } elseif (-not $inThemed) { |
| 409 | Write-Host " WARN '$diskKey' exists in '$canonicalCollectionId' but is not in any themed collection" -ForegroundColor Yellow |
| 410 | } |
| 411 | } |
| 412 | } |
| 413 | |
| 414 | Write-Host '' |
| 415 | Write-Host "$validatedCount collections validated, $errorCount errors" |
| 416 | |
| 417 | return @{ |
| 418 | Success = ($errorCount -eq 0) |
| 419 | ErrorCount = $errorCount |
| 420 | CollectionCount = $validatedCount |
| 421 | } |
| 422 | } |
| 423 | |
| 424 | #endregion Orchestration |
| 425 | |
| 426 | #region Main Execution |
| 427 | if ($MyInvocation.InvocationName -ne '.') { |
| 428 | try { |
| 429 | # Verify PowerShell-Yaml module |
| 430 | if (-not (Get-Module -ListAvailable -Name PowerShell-Yaml)) { |
| 431 | throw "Required module 'PowerShell-Yaml' is not installed." |
| 432 | } |
| 433 | Import-Module PowerShell-Yaml -ErrorAction Stop |
| 434 | |
| 435 | # Resolve paths |
| 436 | $ScriptDir = Split-Path -Parent $MyInvocation.MyCommand.Path |
| 437 | $RepoRoot = (Get-Item "$ScriptDir/../..").FullName |
| 438 | |
| 439 | $result = Invoke-CollectionValidation -RepoRoot $RepoRoot |
| 440 | |
| 441 | if (-not $result.Success) { |
| 442 | throw "Validation failed with $($result.ErrorCount) error(s)." |
| 443 | } |
| 444 | |
| 445 | exit 0 |
| 446 | } |
| 447 | catch { |
| 448 | Write-Error "Collection validation failed: $($_.Exception.Message)" |
| 449 | Write-CIAnnotation -Message $_.Exception.Message -Level Error |
| 450 | exit 1 |
| 451 | } |
| 452 | } |
| 453 | #endregion |
| 454 | |