Scripting Jekyll pages with PowerShell

The travel site gets a new page for every day of a trip, so I wrote New-TravelDay.ps1 to create them. Two details mattered more than I expected.

Slugs from titles with accents

Day titles include places like Plaça Reial and Rüdesheim. A plain -replace '[^a-z0-9]+' turns those into pla-a-reial. Normalising to FormD first splits each accented letter into the letter plus a combining mark, which you can then drop:

function ConvertTo-Slug {
    param([string]$Value)
    $normalised = $Value.Normalize([Text.NormalizationForm]::FormD)
    $plain = -join ($normalised.ToCharArray() | Where-Object {
        [Globalization.CharUnicodeInfo]::GetUnicodeCategory($_) -ne
            [Globalization.UnicodeCategory]::NonSpacingMark
    })
    ($plain.ToLowerInvariant() -replace '[^a-z0-9]+', '-').Trim('-')
}

ConvertTo-Slug 'Plaça Reial'   # placa-reial

UTF-8 without a BOM

In Windows PowerShell 5.1, Set-Content -Encoding UTF8 writes a byte order mark. Jekyll expects the file to start with ---, and a BOM in front of it can stop the front matter being read. Writing through .NET avoids it in both 5.1 and 7:

$utf8NoBom = New-Object System.Text.UTF8Encoding($false)
[IO.File]::WriteAllText($path, $content, $utf8NoBom)

I also convert line endings to LF before writing, so Git doesn’t warn about CRLF on every new file.