convert-mdx-to-md.ps1

$total = 0
$errors = @()
$bunDir = "C:\Users\Antonio\Documents\Obsidian\Recursos\003-Documentación\bun"

if (-not (Test-Path $bunDir)) {
    $parent = Get-ChildItem -Path "C:\Users\Antonio\Documents\Obsidian\Recursos" -Filter "003-Doc*" | Select-Object -First 1
    if ($parent) { $bunDir = Join-Path $parent.FullName "bun" }
    else { Write-Host "Cannot find bun directory!" -ForegroundColor Red; exit 1 }
}

$files = Get-ChildItem -Path $bunDir -Recurse -Filter "*.mdx"
Write-Host "Found $($files.Count) MDX files" -ForegroundColor Yellow

function Strip-MdxComments($text) {
    $text = [regex]::Replace($text, '\{/\*.*?\*/\}', '', 'Singleline')
    $text = [regex]::Replace($text, '<!--.*?-->', '', 'Singleline')
    return $text
}

function Strip-Frontmatter($text) {
    return [regex]::Replace($text, '(?ms)^---\s*\n.*?\n---\s*\n', '')
}

function Strip-Imports($text) {
    return [regex]::Replace($text, '(?m)^import\s+.+?;\s*$', '')
}

function Normalize-JsxTags($text) {
    # Join multi-line JSX opening tags: <Tag\n  attr="val"\n> -> <Tag attr="val">
    for ($i = 0; $i -lt 15; $i++) {
        $text = [regex]::Replace($text, '(?s)(<\w[\w-]*)\s*\r?\n\s*([^>]*\>)', '$1 $2')
    }
    return $text
}

function Fix-CodeFences($text) {
    # Clean code fence headers (including indented ones inside Tab/CodeGroup)
    $text = [regex]::Replace($text, '(?m)^[ \t]*(`{3,}\w*)\s+.*$', '$1')
    $text = [regex]::Replace($text, '(?m)^[ \t]*(`{3,})\s+\w+.*$', '$1')
    return $text
}

function Strip-CodeAnnotations($text) {
    return [regex]::Replace($text, '(?m)\s*//\s*\[!code\s+.*?\]\s*$', '')
}

function Convert-Callouts($text) {
    $replacements = @{
        '<Tip>' = '> [!TIP]'
        '</Tip>' = ''
        '<Warning>' = '> [!WARNING]'
        '</Warning>' = ''
        '<Note>' = '> [!NOTE]'
        '</Note>' = ''
        '<Info>' = '> [!INFO]'
        '</Info>' = ''
        '<Note type="warning">' = '> [!WARNING]'
        '<Note type="tip">' = '> [!TIP]'
    }
    foreach ($key in $replacements.Keys) {
        $text = $text -replace [regex]::Escape($key), $replacements[$key]
    }
    return $text
}

function Strip-Wrappers($text) {
    $wrappers = @('CardGroup', 'Tabs', 'CodeGroup', 'Steps')
    foreach ($w in $wrappers) {
        $text = [regex]::Replace($text, "<$w[^>]*>", '')
        $text = [regex]::Replace($text, "</$w>", '')
    }
    return $text
}

function Convert-Components($text) {
    # Card with title -> **title**: content
    $text = [regex]::Replace($text, '<Card\s+title="([^"]*)"[^>]*>(.*?)</Card>', '**$1**: $2', 'Singleline')
    # Tab -> ### title
    $text = [regex]::Replace($text, '<Tab\s+title="([^"]*)"[^>]*>', '### $1')
    $text = [regex]::Replace($text, '</Tab>', '')
    # Step -> #### title
    $text = [regex]::Replace($text, '<Step\s+title="([^"]*)"[^>]*>', '#### $1')
    $text = [regex]::Replace($text, '</Step>', '')
    # Accordion -> ### title
    $text = [regex]::Replace($text, '<Accordion\s+title="([^"]*)"[^>]*>', '### $1')
    $text = [regex]::Replace($text, '</Accordion>', '')
    # Frame
    $text = [regex]::Replace($text, '<Frame[^>]*>', '')
    $text = [regex]::Replace($text, '</Frame>', '')
    # Badge -> inline code
    $text = [regex]::Replace($text, '<Badge[^>]*>(.*?)</Badge>', '`$1`')
    # img with caption
    $text = [regex]::Replace($text, '<img\s+[^>]*src="([^"]*)"[^>]*caption="([^"]*)"[^>]*>', '![$2]($1)')
    $text = [regex]::Replace($text, '<img\s+[^>]*caption="([^"]*)"[^>]*src="([^"]*)"[^>]*>', '![$1]($2)')
    $text = [regex]::Replace($text, '<img\s+[^>]*src="([^"]*)"[^>]*>', '![]($1)')
    return $text
}

function Strip-RemainingTags($text) {
    # Remove self-closing <TagName ... />
    $text = [regex]::Replace($text, '<\w[\w-]*(\s+[^>]*?)?\s*/>', '')
    # Remove any remaining HTML-like tags
    $text = [regex]::Replace($text, '</?\w[\w-]*(\s+[^>]*?)?\s*/?>', '')
    return $text
}

function Cleanup-Indent($text) {
    # Strip 2+ leading spaces from non-code lines (remnants of Tab nesting)
    $inCode = $false
    $result = @()
    foreach ($line in $text -split "`n") {
        if ($line -match '^[ \t]*```') { $inCode = -not $inCode }
        if (-not $inCode -and $line.Length -gt 0) {
            $stripped = $line -replace '^  +', ''
            $result += $stripped
        } else {
            $result += $line
        }
    }
    return ($result -join "`n")
}

function Cleanup-Whitespace($text) {
    # Remove horizontal rules (---) left after tag removal
    $text = [regex]::Replace($text, '(?m)^\s*[-]{3,}\s*$', '')
    # Remove whitespace-only lines
    $text = [regex]::Replace($text, '(?m)^[ \t]+$', '')
    # Collapse 3+ newlines to 2
    $text = [regex]::Replace($text, "`n{3,}", "`n`n")
    return $text.Trim()
}

$files | ForEach-Object {
    $file = $_.FullName
    $total++

    try {
        $content = Get-Content -Path $file -Raw -ErrorAction Stop

        $content = Strip-MdxComments $content
        $content = Strip-Frontmatter $content
        $content = Strip-Imports $content
        $content = Strip-CodeAnnotations $content
        $content = Normalize-JsxTags $content
        $content = Fix-CodeFences $content
        $content = Convert-Callouts $content
        $content = Strip-Wrappers $content
        $content = Convert-Components $content
        $content = Strip-RemainingTags $content
        $content = Cleanup-Indent $content
        $content = Cleanup-Whitespace $content

        $mdFile = $file -replace '\.mdx$', '.md'
        [System.IO.File]::WriteAllText($mdFile, $content, [System.Text.Encoding]::UTF8)
        Remove-Item -Path $file -Force

        Write-Host "OK $($_.BaseName).md" -ForegroundColor Green
    }
    catch {
        Write-Host "ERR $($_.Name): $_" -ForegroundColor Red
        $errors += "$($_.Name): $_"
    }
}

Write-Host "`n=== Done: $total files processed, $($errors.Count) errors ===" -ForegroundColor Yellow
if ($errors.Count -gt 0) { $errors | ForEach-Object { Write-Host "  - $_" -ForegroundColor Red } }