# PowerShell script to find binary files NOT tracked by Git LFS
# Run this from your repository root

param(
    [string]$RepoPath = ".",
    [switch]$ShowLFSStatus = $false
)

$ErrorActionPreference = "Continue"

# Common binary file extensions that should be in LFS
$binaryExtensions = @(
    # Images
    'bmp', 'exr', 'gif', 'hdr', 'jpg', 'jpeg', 'png', 'tga', 'tif', 'tiff', 'webp', 'heic', 'HEIC',
    'ico', 'cur', 'ani', 'xcf', 'psd', 'ai', 'af', 'eps', 'raw', 'cr2', 'nef', 'dng',
    # 3D/CG
    '3ds', 'abc', 'blend', 'blend1', 'blend2', 'bvh', 'c4d', 'dae', 'fbx', 'fbxkey',
    'ma', 'max', 'mb', 'obj', 'usd', 'usdz', 'vdb', 'bphys', 'bobj', 'bvel', 'bpointcache', 'blob',
    'unitypackage', 'skp', 'stl', 'ply', 'glb', 'gltf', '3mf', 'lwo', 'lws',
    # Character Creator / Reallusion
    'ccAvatar', 'ccProject', 'ccCloth', 'ccProp', 'ccScene', 'ccHair',
    'rlMotion', 'rlPose', 'iAvatar', 'iMotion', 'iProp', 'iMaterial', 'rlHead', 'rlHair', 'rlScene', 'rlSkintex',
    '3dxProfile',
    # Houdini
    'hiplc', 'bgeo', 'bgeo.sc', 'hdanc', 'otllc', 'otlnc',
    # Video
    'avi', 'mkv', 'mov', 'MOV', 'mp4', 'webm', 'm4v', 'mpg', 'mpeg', 'wmv', 'flv', 'f4v',
    # Audio
    'mp3', 'wav', 'm4a', 'aac', 'ogg', 'flac', 'wma', 'aiff', 'au', 'ra', 'ram', 'mid', 'midi', 'ac3', 'dts',
    # Archives
    '7z', 'bz2', 'gz', 'gzip', 'lzma', 'rar', 'tar', 'xz', 'z', 'zip', 'whl', 'npz',
    'pkl', 'pickle', 'joblib', 'parquet', 'feather',
    # Documents
    'doc', 'docx', 'xls', 'xlsx', 'ppt', 'pptx', 'pdf', 'odt', 'ods', 'odp', 'rtf',
    # Adobe
    'aegraphic', 'aep', 'prel', 'prin', 'prmf', 'prproj', 'ai', 'psd', 'pk', 'pkf',
    # Data/Cache
    'h5', 'hdf5', 'dat', 'db', 'sqlite', 'sqlite3', 'mdb', 'ldb', 'frm', 'ibd',
    'npy', 'pkl', 'pickle', 'joblib', 'caffemodel', 'pth', 'onnx', 'pb', 'tflite',
    # Binaries
    'exe', 'dll', 'so', 'dylib', 'pyd', 'bin', 'o', 'a', 'lib', 'obj', 'app',
    'msi', 'msp', 'msm', 'dmg', 'pkg', 'deb', 'rpm', 'snap', 'flatpak', 'AppImage',
    # Fonts
    'ttf', 'otf', 'woff', 'woff2', 'eot', 'pfm', 'afm', 'tfm',
    # Other
    'uni', 'pdn', 'pur', 'lnk', 'swf', 'fla', 'wasm', 'loom', 'mat', 'fbxkey',
    'ffp3', 'data', 'sqlite3', 'db', 'raw', 'dpx', 'drp', 'lut', 'cube', 'spi1d', 'spi3d', 'clf',
    # Substance
    'sbs', 'sbsar', 'spp', 'sbsprs', 'sbscfg',
    # Flip Fluids
    'bphys', 'bobj', 'bvel', 'bpointcache', 'blob', 'ffp3',
    # Game engines
    'uasset', 'umap', 'upk', 'udk', 'pak', 'unitypackage', 'prefab', 'asset', 'meta'
)

function Test-IsBinaryExtension {
    param([string]$extension)
    return $binaryExtensions -contains $extension
}

function Get-FileSize {
    param([string]$path)
    try {
        $file = Get-Item $path -ErrorAction SilentlyContinue
        if ($file) {
            if ($file.Length -gt 1MB) {
                return "{0:N2} MB" -f ($file.Length / 1MB)
            } elseif ($file.Length -gt 1KB) {
                return "{0:N2} KB" -f ($file.Length / 1KB)
            } else {
                return "$($file.Length) B"
            }
        }
    } catch {
        return "Unknown"
    }
    return "Unknown"
}

# Get absolute path
$absPath = Resolve-Path $RepoPath
Write-Host "Scanning repository for untracked binaries: $absPath" -ForegroundColor Cyan
Write-Host ""

# Get list of ALL files tracked by Git
Write-Host "Getting files tracked by Git..." -ForegroundColor Yellow
$allGitFiles = git ls-files 2>$null

# Get list of files tracked by LFS
Write-Host "Getting files tracked by LFS..." -ForegroundColor Yellow
$lfsFiles = git lfs ls-files 2>$null | ForEach-Object { 
    # Parse the LFS output format: <oid> * <filename> or <oid> - <filename>
    if ($_ -match '^[a-f0-9]+\s+[-*]\s+(.+)$') {
        $matches[1]
    }
}

# Create a hashtable for fast lookup of LFS files
$lfsFileLookup = @{}
foreach ($file in $lfsFiles) {
    $lfsFileLookup[$file] = $true
}

Write-Host "Total Git-tracked files: $($allGitFiles.Count)" -ForegroundColor Gray
Write-Host "Total LFS-tracked files: $($lfsFiles.Count)" -ForegroundColor Gray
Write-Host ""

# Find binary files that are NOT in LFS
$untrackedBinaries = @()
$extensionSummary = @{}

foreach ($file in $allGitFiles) {
    # Skip if empty
    if ([string]::IsNullOrWhiteSpace($file)) { continue }
    
    # Skip if already in LFS
    if ($lfsFileLookup.ContainsKey($file)) { continue }
    
    # Get extension (with error handling for weird filenames)
    try {
        $extension = [System.IO.Path]::GetExtension($file).TrimStart('.')
    } catch {
        continue  # Skip files with problematic names
    }
    
    # Check if it's a known binary extension
    if (Test-IsBinaryExtension -extension $extension) {
        $fullPath = Join-Path $absPath $file
        $fileSize = Get-FileSize -path $fullPath
        
        $untrackedBinaries += [PSCustomObject]@{
            File = $file
            Extension = $extension
            Size = $fileSize
        }
        
        # Track summary
        if (-not $extensionSummary.ContainsKey($extension)) {
            $extensionSummary[$extension] = 0
        }
        $extensionSummary[$extension]++
    }
}

# Display results
if ($untrackedBinaries.Count -eq 0) {
    Write-Host "No untracked binary files found!" -ForegroundColor Green
    Write-Host "All binary files are properly tracked by Git LFS." -ForegroundColor Green
} else {
    Write-Host "Found $($untrackedBinaries.Count) binary files tracked by Git (not LFS):" -ForegroundColor Red
    Write-Host ""
    
    # Group by extension for display
    $groupedByExt = $untrackedBinaries | Group-Object -Property Extension | Sort-Object Count -Descending
    
    foreach ($group in $groupedByExt) {
        $groupName = $group.Name
        $groupCount = $group.Count
        Write-Host "  .$groupName ($groupCount files):" -ForegroundColor Yellow
        foreach ($item in $group.Group | Select-Object -First 10) {
            Write-Host "    - $($item.File) ($($item.Size))" -ForegroundColor Gray
        }
        if ($groupCount -gt 10) {
            Write-Host "    ... and $($groupCount - 10) more" -ForegroundColor DarkGray
        }
    }
    
    Write-Host ""
    Write-Host "========================================" -ForegroundColor Red
    Write-Host "SUGGESTED .gitattributes ADDITIONS:" -ForegroundColor Cyan
    Write-Host "========================================" -ForegroundColor Cyan
    
    # Generate suggested gitattributes entries
    $sortedExts = $extensionSummary.GetEnumerator() | Sort-Object Value -Descending
    foreach ($ext in $sortedExts) {
        $extName = $ext.Key
        Write-Host "*.$extName filter=lfs diff=lfs merge=lfs -text" -ForegroundColor White
    }
    
    Write-Host ""
    Write-Host "To fix these files, add the above lines to your .gitattributes file," -ForegroundColor Yellow
    Write-Host "then run: git add --renormalize ." -ForegroundColor Yellow
}

Write-Host ""

# Optional: Show full LFS status
if ($ShowLFSStatus) {
    Write-Host "========================================" -ForegroundColor Cyan
    Write-Host "Current LFS Status:" -ForegroundColor Cyan
    Write-Host "========================================" -ForegroundColor Cyan
    git lfs status 2>&1 | ForEach-Object { Write-Host $_ }
}