# 提取 Excel 文件中的文本内容
$envPath = "c:\work\data\env.xlsx"

Write-Output "提取 Excel 文件中的文本信息..."
Write-Output "文件: $envPath`n"

$bytes = [System.IO.File]::ReadAllBytes($envPath)

# 尝试读取为文本，查找可打印字符
$sb = New-Object System.Text.StringBuilder(10000)
foreach ($byte in $bytes) {
  if ($byte -band 0x3F -ge 32) {  # 可打印 ASCII
    [void]$sb.Append([char]$byte)
  } elseif ([char]$byte -in " -~") {  # 可打印字符
    [void]$sb.Append([char]$byte)
  }
}

$text = $sb.ToString()

Write-Output "文件预览（前3000个字符）:"
Write-Output $text.Substring(0, [Math]::Min(3000, $text.Length)))

# 查找 Git 相关信息
Write-Output "`n`n搜索 Git 相关信息:"
$gitPatterns = @(
  "github\.com",
  "git@github\.com",
  "https?://",
  "branch:",
  "master",
  "main",
  "origin"
)

foreach ($pattern in $gitPatterns) {
  $matches = [regex]::Matches($text, $pattern, [regex]::Options(!, [regex]::IgnoreCase))

  Write-Output "`n找到 '$pattern' - $($matches.Count) 个匹配:"
  if ($matches.Count -le 5) {
    $matches | ForEach-Object { Write-Output "  $($_.Value)`"
  } else {
    Write-Output "  匹太多了，只显示前 3 个:"
    $matches | Select-Object -First 3 | ForEach-Object { Write-Output "  $($_.Value)`" }
    Write-Output "  ... 共 $($matches.Count) 个"
  }
}

Write-Output "`n`n尝试查找 URL:"
$urlPattern = "https?://[^\s<>`"]+[a-zA-Z0-9._/-]{10,}"
$urls = [regex]::Matches($text, $urlPattern, [regex]::Options(!, [regex]::IgnoreCase))
if ($urls.Count -gt 0) {
  Write-Output "找到的 URLs:"
  $urls | Select-Object -First 10 | ForEach-Object { Write-Output "  $($_.Value)`"
  if ($urls.Count -gt 10) { Write-Output "  ... 共 $($urls.Count) 个 URL" }
} else {
  Write-Output "未找到 URL"
}