UncommonSense.HaarlemsDagblad

0.0.4

Public/Get-JoostPrinsen.ps1

function Get-JoostPrinsen
{
    Invoke-WebRequest -Uri 'https://www.haarlemsdagblad.nl/zoeken?keyword=joost+prinsen'
    | Select-Object -ExpandProperty Links
    | Select-Object -ExpandProperty Href
    | Where-Object { $_ -like '/cnt/dmf*' }
    | ForEach-Object { "https://haarlemsdagblad.nl$($_)" }
    | ForEach-Object {
        $CurrentUrl = $_
        $Content = Invoke-WebRequest -Uri $CurrentUrl | Select-Object -ExpandProperty Content

        $ArticleContent = ConvertTo-HtmlDocument -Text $Content
        | Select-HtmlNode -CssSelector 'script[data-fragment-type=ArticleContent]'
        | Select-Object -ExpandProperty InnerText
        | ForEach-Object { $_ -replace '^window\["__PRELOADED_STATE_GRAPH__[^"]*"\] = ', '' }
        | ConvertFrom-Json

        $FirstProperty = $ArticleContent.PSObject.Properties | Select-Object -First 1 -ExpandProperty Name

        $ArticleContent
        | Select-Object -ExpandProperty $FirstProperty
        | Select-Object -ExpandProperty data
        | Select-Object -ExpandProperty article
        | ForEach-Object {
            $Intro = $_.Intro | Select-Object -ExpandProperty p | Join-String -Separator "`n"
            $Body = $_.Body | Select-Object -ExpandProperty p | Join-String -Separator "`n"

            [PSCustomObject]@{
                PSTypeName = 'UncommonSense.HaarlemsDagblad.Article'
                Url        = $CurrentUrl
                Date       = $_.PublishedAt
                Title      = $_.Title -replace '\s\|\scolumn$', ''
                Body       = @($Intro, $Body) -join "`n`n"
            }
        }
    }
}