# PowerShell 脚本：检查文件编码是否为 UTF-8
# 用法: .\check_encoding.ps1 [文件路径或目录路径]
# 输出编码: UTF-8

param(
    [Parameter(Mandatory=$false)]
    [string]$Path = "."
)

# 设置控制台和输出编码为 UTF-8，修复中文乱码
if ([Console]::OutputEncoding.CodePage -ne 65001) {
    chcp 65001 | Out-Null
}
[Console]::OutputEncoding = [System.Text.Encoding]::UTF8
$PSDefaultParameterValues['*:Encoding'] = 'utf8'
$OutputEncoding = [System.Text.Encoding]::UTF8

# 定义代码文件扩展名列表
$codeFileExtensions = @(
    # 源代码文件
    '.cs', '.js', '.ts', '.jsx', '.tsx', '.java', '.py', '.cpp', '.c', '.h', '.hpp', 
    '.go', '.rs', '.rb', '.php', '.swift', '.kt', '.scala', '.vb', '.fs', '.dart',
    # 配置文件/标记文件
    '.json', '.xml', '.html', '.htm', '.css', '.scss', '.sass', '.less', 
    '.yaml', '.yml', '.toml', '.ini', '.config', '.properties', '.conf',
    # 脚本文件
    '.ps1', '.psm1', '.psd1', '.sh', '.bash', '.zsh', '.fish', '.bat', '.cmd',
    # 数据/标记文件
    '.md', '.txt', '.log', '.csv', '.sql', '.r', '.m', '.mm',
    # 其他文本格式
    '.vue', '.svelte', '.tsx', '.jsx', '.dts', '.map'
)

function Test-IsCodeFile {
    param([string]$FilePath)
    
    $extension = [System.IO.Path]::GetExtension($FilePath).ToLower()
    return $codeFileExtensions -contains $extension
}

function Test-IsUtf8 {
    param([string]$FilePath)
    
    try {
        # 读取文件的前几个字节检查 BOM
        $bytes = [System.IO.File]::ReadAllBytes($FilePath)
        
        # 检查 UTF-8 BOM (EF BB BF)
        if ($bytes.Length -ge 3 -and $bytes[0] -eq 0xEF -and $bytes[1] -eq 0xBB -and $bytes[2] -eq 0xBF) {
            return @{ IsUtf8 = $true; HasBom = $true; Encoding = "UTF-8 with BOM" }
        }
        
        # 尝试用 UTF-8 解码整个文件
        $content = [System.IO.File]::ReadAllText($FilePath, [System.Text.Encoding]::UTF8)
        
        # 检查是否包含无效的 UTF-8 序列（通过重新编码比较）
        $utf8Bytes = [System.Text.Encoding]::UTF8.GetBytes($content)
        $originalBytes = [System.IO.File]::ReadAllBytes($FilePath)
        
        # 如果重新编码后的字节与原始文件相同（忽略 BOM），则为有效 UTF-8
        if ($bytes.Length -ge 3 -and $bytes[0] -eq 0xEF -and $bytes[1] -eq 0xBB -and $bytes[2] -eq 0xBF) {
            $originalBytes = $originalBytes[3..($originalBytes.Length-1)]
        }
        
        # 简单检查：尝试用 UTF-8 解码，如果成功且能再编码回去，可能是 UTF-8
        # 更准确的方法是用 chardet 或类似库，但 PowerShell 内置方法如下：
        try {
            $decoded = [System.Text.Encoding]::UTF8.GetString($originalBytes)
            $reencoded = [System.Text.Encoding]::UTF8.GetBytes($decoded)
            
            # 如果字节数组匹配（或接近），很可能是 UTF-8
            $match = $true
            if ($reencoded.Length -ne $originalBytes.Length) {
                $match = $false
            } else {
                for ($i = 0; $i -lt [Math]::Min($reencoded.Length, $originalBytes.Length); $i++) {
                    if ($reencoded[$i] -ne $originalBytes[$i]) {
                        $match = $false
                        break
                    }
                }
            }
            
            return @{ IsUtf8 = $match; HasBom = $false; Encoding = if ($match) { "UTF-8 without BOM" } else { "Unknown/Not UTF-8" } }
        } catch {
            return @{ IsUtf8 = $false; HasBom = $false; Encoding = "Not UTF-8" }
        }
    } catch {
        return @{ IsUtf8 = $false; HasBom = $false; Encoding = "Error: $($_.Exception.Message)" }
    }
}

# 主逻辑
if (Test-Path $Path -PathType Leaf) {
    # 单个文件
    if (-not (Test-IsCodeFile -FilePath $Path)) {
        Write-Host "Skipping non-code file: $Path" -ForegroundColor Gray
        exit 0
    }
    $result = Test-IsUtf8 -FilePath $Path
    if (-not $result.IsUtf8) {
        Write-Host "[X] Not UTF-8: $Path" -ForegroundColor Red
        Write-Host "    Encoding: $($result.Encoding)" -ForegroundColor Yellow
    }
} elseif (Test-Path $Path -PathType Container) {
    # 目录
    Write-Host "Checking directory: $Path" -ForegroundColor Cyan
    Write-Host ""
    
    $allFiles = Get-ChildItem -Path $Path -File -Recurse
    $files = $allFiles | Where-Object { Test-IsCodeFile -FilePath $_.FullName }
    
    if ($files.Count -eq 0) {
        Write-Host "No code files found in the directory." -ForegroundColor Yellow
        exit 0
    }
    
    Write-Host "Found $($files.Count) code file(s) (filtered from $($allFiles.Count) total files)" -ForegroundColor Gray
    Write-Host ""
    
    $utf8Count = 0
    $nonUtf8Count = 0
    $nonUtf8Files = @()
    
    foreach ($file in $files) {
        $result = Test-IsUtf8 -FilePath $file.FullName
        if ($result.IsUtf8) {
            $utf8Count++
        } else {
            $nonUtf8Count++
            $nonUtf8Files += $file.FullName
            Write-Host "[X] Not UTF-8: $($file.FullName)" -ForegroundColor Red
            Write-Host "    Encoding: $($result.Encoding)" -ForegroundColor Yellow
            Write-Host ""
        }
    }
    
    Write-Host "Summary:" -ForegroundColor Cyan
    Write-Host "  Total code files: $($files.Count)"
    Write-Host "  UTF-8 files: $utf8Count" -ForegroundColor Green
    if ($nonUtf8Count -gt 0) {
        Write-Host "  Non UTF-8 files: $nonUtf8Count" -ForegroundColor Red
    } else {
        Write-Host "  Non UTF-8 files: 0" -ForegroundColor Green
        Write-Host ""
        Write-Host "All code files are UTF-8 encoded!" -ForegroundColor Green
    }
} else {
    Write-Host "Error: Path does not exist: $Path" -ForegroundColor Red
    exit 1
}