UncommonSense.TheGuardian

0.0.3

Public/Get-TomGauld.ps1

function Get-TomGauld
{
    param
    (
        [switch]$All
    )

    $NextUrl = 'https://www.theguardian.com/profile/tom-gauld'

    do
    {
        $Links = Invoke-WebRequest -Uri $NextUrl | Select-Object -ExpandProperty Links

        $Links
        | Where-Object HRef -Like 'https://www.theguardian.com/books/picture/*'
        | Select-Object -ExpandProperty HRef -Unique
        | ForEach-Object {
            $ImagePageContent = Invoke-WebRequest -Uri $_ | Select-Object -ExpandProperty Content
            $ImageUrl = ($ImagePageContent | pup 'source:first-of-type attr{srcset}' --plain) -replace '1880w$', ''
            $Title = ($ImagePageContent | pup 'h1 text{}' | ForEach-Object { $_.Trim() } | Where-Object { $_ }) -join ' '
            $DateText = ($ImagePageContent | pup 'time:first-of-type attr{datetime}') -replace '[+-](\d{4})$', ''

            [PSCustomObject][Ordered]@{
                PSTypeName = 'UncommonSense.TheGuardian.Article'
                Url        = $_
                Date       = [DateTime]::ParseExact($DateText, 's', $null)
                Title      = $Title -replace '\s–\scartoon$', ''
                Body       = $ImageUrl
            }
        }

        if ($All)
        {
            $NextUrl = $Links | Where-Object Rel -EQ Next | Select-Object -ExpandProperty HRef
        }
        else
        {
            $NextUrl = $null
        }
    }
    while ($NextUrl)
}