microsoft/hve-core
Publicmirrored fromhttps://github.com/microsoft/hve-coreAvailable
scripts/evals/Modules/ModerationRunner.psm1
158lines · modecode
| 1 | # Copyright (c) Microsoft Corporation. |
| 2 | # SPDX-License-Identifier: MIT |
| 3 | # ModerationRunner.psm1 |
| 4 | # Purpose: Helpers for content moderation batch processing and orchestration |
| 5 | #Requires -Version 7.0 |
| 6 | |
| 7 | <# |
| 8 | .SYNOPSIS |
| 9 | Builds a JSON-lines input file from a batch of records. |
| 10 | |
| 11 | .DESCRIPTION |
| 12 | Accepts an array of hashtables with 'id' and 'text' keys and writes them |
| 13 | as JSON-lines to a temporary file. Returns the file path. |
| 14 | |
| 15 | .PARAMETER Records |
| 16 | Array of hashtables, each with 'id' and 'text' keys. |
| 17 | |
| 18 | .PARAMETER OutFile |
| 19 | Path to the output JSON-lines file. Defaults to a temp file. |
| 20 | |
| 21 | .OUTPUTS |
| 22 | System.String - Path to the JSON-lines file. |
| 23 | |
| 24 | .EXAMPLE |
| 25 | $records = @( |
| 26 | @{ id = 'rec1'; text = 'Hello world' }, |
| 27 | @{ id = 'rec2'; text = 'Test content' } |
| 28 | ) |
| 29 | $inputFile = New-ModerationInputFile -Records $records |
| 30 | #> |
| 31 | function New-ModerationInputFile { |
| 32 | [CmdletBinding()] |
| 33 | [OutputType([string])] |
| 34 | param( |
| 35 | [Parameter(Mandatory = $true)] |
| 36 | [hashtable[]]$Records, |
| 37 | |
| 38 | [Parameter(Mandatory = $false)] |
| 39 | [string]$OutFile |
| 40 | ) |
| 41 | |
| 42 | if (-not $OutFile) { |
| 43 | $OutFile = [System.IO.Path]::GetTempFileName() |
| 44 | } |
| 45 | |
| 46 | $jsonLines = $Records | ForEach-Object { |
| 47 | ConvertTo-Json $_ -Compress -Depth 1 |
| 48 | } |
| 49 | $jsonLines | Set-Content -Path $OutFile -Encoding utf8NoBOM |
| 50 | |
| 51 | Write-Verbose "Wrote $($Records.Count) records to $OutFile" |
| 52 | return $OutFile |
| 53 | } |
| 54 | |
| 55 | <# |
| 56 | .SYNOPSIS |
| 57 | Reads files from a file list and builds moderation records. |
| 58 | |
| 59 | .DESCRIPTION |
| 60 | Accepts an array of file paths, reads each file, and constructs a hashtable |
| 61 | record with 'id' (relative path) and 'text' (file content). |
| 62 | |
| 63 | .PARAMETER FileList |
| 64 | Array of file paths to read. |
| 65 | |
| 66 | .PARAMETER RepoRoot |
| 67 | Repository root for relativizing file paths. Defaults to the current directory. |
| 68 | |
| 69 | .OUTPUTS |
| 70 | System.Collections.Hashtable[] - Array of records with id and text keys. |
| 71 | |
| 72 | .EXAMPLE |
| 73 | $files = Get-ChildItem *.md |
| 74 | $records = ConvertTo-ModerationRecords -FileList $files.FullName |
| 75 | #> |
| 76 | function ConvertTo-ModerationRecords { |
| 77 | [CmdletBinding()] |
| 78 | [OutputType([hashtable[]])] |
| 79 | param( |
| 80 | [Parameter(Mandatory = $true)] |
| 81 | [string[]]$FileList, |
| 82 | |
| 83 | [Parameter(Mandatory = $false)] |
| 84 | [string]$RepoRoot = $PWD |
| 85 | ) |
| 86 | |
| 87 | $records = @() |
| 88 | foreach ($filePath in $FileList) { |
| 89 | if (-not (Test-Path -LiteralPath $filePath)) { |
| 90 | Write-Warning "File not found: $filePath" |
| 91 | continue |
| 92 | } |
| 93 | $relativePath = (Resolve-Path -LiteralPath $filePath -Relative -RelativeBasePath $RepoRoot).TrimStart('.', '\', '/') |
| 94 | $content = Get-Content -LiteralPath $filePath -Raw -Encoding utf8 |
| 95 | $records += @{ |
| 96 | id = $relativePath |
| 97 | text = $content |
| 98 | } |
| 99 | } |
| 100 | Write-Verbose "Built $($records.Count) records from $($FileList.Count) files" |
| 101 | return $records |
| 102 | } |
| 103 | |
| 104 | <# |
| 105 | .SYNOPSIS |
| 106 | Parses moderate.py JSON output and surfaces structured error messages. |
| 107 | |
| 108 | .DESCRIPTION |
| 109 | Reads the JSON output from moderate.py, extracts flagged records, and emits |
| 110 | GitHub Actions error annotations for each flagged item. |
| 111 | |
| 112 | .PARAMETER OutputPath |
| 113 | Path to the moderate.py JSON output file. |
| 114 | |
| 115 | .OUTPUTS |
| 116 | System.Boolean - Returns $true if any records were flagged, $false otherwise. |
| 117 | |
| 118 | .EXAMPLE |
| 119 | if (Test-ModerationOutput -OutputPath logs/moderation-corpus.json) { |
| 120 | Write-Error "Content moderation failed" |
| 121 | } |
| 122 | #> |
| 123 | function Test-ModerationOutput { |
| 124 | [CmdletBinding()] |
| 125 | [OutputType([bool])] |
| 126 | param( |
| 127 | [Parameter(Mandatory = $true)] |
| 128 | [string]$OutputPath |
| 129 | ) |
| 130 | |
| 131 | if (-not (Test-Path $OutputPath)) { |
| 132 | Write-Error "Moderation output file not found: $OutputPath" |
| 133 | return $true |
| 134 | } |
| 135 | |
| 136 | $output = Get-Content -Path $OutputPath -Raw | ConvertFrom-Json |
| 137 | $flaggedCount = $output.summary.flaggedCount |
| 138 | |
| 139 | if ($flaggedCount -eq 0) { |
| 140 | Write-Verbose "Content moderation passed: all $($output.summary.total) records clean" |
| 141 | return $false |
| 142 | } |
| 143 | |
| 144 | Write-Warning "Content moderation failed: $flaggedCount/$($output.summary.total) records flagged" |
| 145 | foreach ($record in $output.records) { |
| 146 | if ($record.flagged) { |
| 147 | $labels = $record.flaggedLabels -join ', ' |
| 148 | Write-Host "::error file=$($record.id)::Content moderation flag: $labels" |
| 149 | } |
| 150 | } |
| 151 | return $true |
| 152 | } |
| 153 | |
| 154 | Export-ModuleMember -Function @( |
| 155 | 'New-ModerationInputFile', |
| 156 | 'ConvertTo-ModerationRecords', |
| 157 | 'Test-ModerationOutput' |
| 158 | ) |
| 159 | |