Core_Be/tools/Update-Md5Reference.ps1

param(
    [string]$OutputPath = "PROJECT_MD5_REFERENCE.md",
    [switch]$Verify
)

$ErrorActionPreference = "Stop"

$scriptDir = Split-Path -Parent $MyInvocation.MyCommand.Path
$fallbackRoot = Resolve-Path -LiteralPath (Join-Path $scriptDir "..")

try {
    $gitRoot = git -C $fallbackRoot rev-parse --show-toplevel 2>$null
    if ([string]::IsNullOrWhiteSpace($gitRoot)) {
        $repoRoot = $fallbackRoot.Path
    }
    else {
        $repoRoot = (Resolve-Path -LiteralPath $gitRoot.Trim()).Path
    }
}
catch {
    $repoRoot = $fallbackRoot.Path
}

Push-Location $repoRoot
try {
    $normalizedOutputPath = $OutputPath.Replace("\", "/")

    $files = @()
    try {
        $files = git ls-files --cached --modified --others --exclude-standard |
            ForEach-Object { $_.Replace("\", "/") } |
            Where-Object { $_ -and $_ -ne $normalizedOutputPath }
    }
    catch {
        $ignoredDirectories = @(".git", ".vs", "bin", "obj", "publish", "Migrations")
        $ignoredExtensions = @(".user", ".suo", ".userosscache", ".sln.docstates", ".log", ".zip")

        $files = Get-ChildItem -LiteralPath $repoRoot -Recurse -File |
            Where-Object {
                $relative = [System.IO.Path]::GetRelativePath($repoRoot, $_.FullName).Replace("\", "/")
                $parts = $relative -split "/"
                $extension = [System.IO.Path]::GetExtension($_.Name)

                ($ignoredDirectories | Where-Object { $parts -contains $_ }).Count -eq 0 -and
                ($ignoredExtensions -notcontains $extension) -and
                $relative -ne $normalizedOutputPath
            } |
            ForEach-Object { [System.IO.Path]::GetRelativePath($repoRoot, $_.FullName).Replace("\", "/") }
    }

    $files = $files | Sort-Object -Unique

    $hashRows = foreach ($file in $files) {
        $fullPath = Join-Path $repoRoot $file
        if (Test-Path -LiteralPath $fullPath -PathType Leaf) {
            $item = Get-Item -LiteralPath $fullPath
            $hash = Get-FileHash -LiteralPath $fullPath -Algorithm MD5

            [PSCustomObject]@{
                Path = $file
                Size = $item.Length
                MD5  = $hash.Hash.ToLowerInvariant()
            }
        }
    }

    $statusLines = @()
    try {
        $statusLines = git status --short |
            Where-Object {
                $statusPath = ($_ -replace "^[ MADRCU?!]{2}\s+", "").Replace("\", "/")
                $statusPath -ne $normalizedOutputPath
            }
    }
    catch {
        $statusLines = @()
    }

    if ($Verify) {
        if (-not (Test-Path -LiteralPath $OutputPath -PathType Leaf)) {
            throw "Cannot verify because $OutputPath does not exist."
        }

        $expectedRows = Get-Content -LiteralPath $OutputPath |
            ForEach-Object {
                if ($_ -match "^\| (?<Path>.+?) \| (?<Size>\d+) \| (?<MD5>[a-f0-9]{32}) \|$") {
                    [PSCustomObject]@{
                        Path = $Matches.Path
                        Size = [int64]$Matches.Size
                        MD5  = $Matches.MD5
                    }
                }
            }

        $expectedByPath = @{}
        foreach ($row in $expectedRows) {
            $expectedByPath[$row.Path] = $row
        }

        $currentByPath = @{}
        foreach ($row in $hashRows) {
            $currentByPath[$row.Path] = $row
        }

        $added = @($currentByPath.Keys | Where-Object { -not $expectedByPath.ContainsKey($_) } | Sort-Object)
        $removed = @($expectedByPath.Keys | Where-Object { -not $currentByPath.ContainsKey($_) } | Sort-Object)
        $changed = @(
            foreach ($path in $expectedByPath.Keys) {
                if ($currentByPath.ContainsKey($path) -and $currentByPath[$path].MD5 -ne $expectedByPath[$path].MD5) {
                    "$path ($($expectedByPath[$path].MD5) -> $($currentByPath[$path].MD5))"
                }
            }
        ) | Sort-Object

        if ($added.Count -eq 0 -and $removed.Count -eq 0 -and $changed.Count -eq 0) {
            Write-Host "MD5 reference verified. No changes detected across $($hashRows.Count) files."
            return
        }

        Write-Host "MD5 reference verification found differences."
        if ($added.Count -gt 0) {
            Write-Host ""
            Write-Host "Added files:"
            $added | ForEach-Object { Write-Host "  $_" }
        }
        if ($removed.Count -gt 0) {
            Write-Host ""
            Write-Host "Removed files:"
            $removed | ForEach-Object { Write-Host "  $_" }
        }
        if ($changed.Count -gt 0) {
            Write-Host ""
            Write-Host "Changed files:"
            $changed | ForEach-Object { Write-Host "  $_" }
        }

        exit 1
    }

    $projectName = Split-Path -Leaf $repoRoot
    $generatedAt = Get-Date -Format "yyyy-MM-dd HH:mm:ss zzz"
    $targetFramework = ""
    if (Test-Path -LiteralPath "Fitness_core_BE.csproj") {
        $projectXml = [xml](Get-Content -Raw -LiteralPath "Fitness_core_BE.csproj")
        $targetFramework = $projectXml.Project.PropertyGroup.TargetFramework
    }

    $topLevelCounts = $hashRows |
        Group-Object {
            if ($_.Path.Contains("/")) {
                ($_.Path -split "/")[0]
            }
            else {
                "."
            }
        } |
        Sort-Object Name

    $lines = [System.Collections.Generic.List[string]]::new()
    $lines.Add("# Project MD5 Reference")
    $lines.Add("")
    $lines.Add("Generated: $generatedAt")
    $lines.Add("")
    $lines.Add("Project: $projectName")
    if (-not [string]::IsNullOrWhiteSpace($targetFramework)) {
        $lines.Add("")
        $lines.Add("Target framework: $targetFramework")
    }
    $lines.Add("")
    $lines.Add("Purpose: baseline MD5 hashes for future implementation checks. MD5 is used here for change detection, not for security-sensitive integrity verification.")
    $lines.Add("")
    $lines.Add('Manifest scope: Git tracked files plus untracked files not ignored by `.gitignore`; build outputs, IDE files, ignored migrations, and this manifest are excluded.')
    $lines.Add("")
    $lines.Add("Regenerate command:")
    $lines.Add("")
    $lines.Add('```powershell')
    $lines.Add('.\tools\Update-Md5Reference.ps1')
    $lines.Add('```')
    $lines.Add("")
    $lines.Add("Verify command:")
    $lines.Add("")
    $lines.Add('```powershell')
    $lines.Add('.\tools\Update-Md5Reference.ps1 -Verify')
    $lines.Add('```')
    $lines.Add("")
    $lines.Add("## Project Shape")
    $lines.Add("")
    $lines.Add("| Area | Files |")
    $lines.Add("| --- | ---: |")
    foreach ($group in $topLevelCounts) {
        $lines.Add("| $($group.Name) | $($group.Count) |")
    }
    $lines.Add("")
    $lines.Add("## Git Status At Baseline")
    $lines.Add("")
    if ($statusLines.Count -gt 0) {
        $lines.Add('```text')
        foreach ($status in $statusLines) {
            $lines.Add($status)
        }
        $lines.Add('```')
    }
    else {
        $lines.Add("Clean working tree or Git status unavailable.")
    }
    $lines.Add("")
    $lines.Add("## MD5 Hashes")
    $lines.Add("")
    $lines.Add("| Path | Size Bytes | MD5 |")
    $lines.Add("| --- | ---: | --- |")
    foreach ($row in $hashRows) {
        $lines.Add("| $($row.Path) | $($row.Size) | $($row.MD5) |")
    }
    $lines.Add("")

    Set-Content -LiteralPath $OutputPath -Value $lines -Encoding UTF8
    Write-Host "Wrote $OutputPath with $($hashRows.Count) MD5 hashes."
}
finally {
    Pop-Location
}