UncommonSense.HaarlemsDagblad

0.0.1

Public/Get-JoostPrinsen.ps1

                                function Get-JoostPrinsen

{

    $DutchCulture = [cultureinfo]::GetCultureInfo('nl-NL')

    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 = $_

            Invoke-WebRequest $CurrentUrl

            | Select-Object -ExpandProperty Content

            | pup 'script json{}'

            | ConvertFrom-Json

            | Where-Object text -Like 'window`["__PRE*'

            | Select-Object -First 1 -ExpandProperty Text

            | ForEach-Object { $_ -replace '^window\["__PRELOADED_STATE_GRAPH__[^"]*"\] = ', '' }

            | jq '..|.article? // empty' # | { title: .title, intro: ([.intro.p]|join(\" \")), body: ([.body.json[].p]|join(\" \"))}'

            | ConvertFrom-Json

            | Select-Object -Property Title, Intro, Body, PublishedAt

            | ForEach-Object {

                $Intro = ($_.Intro).p -join "`n"

                $Body = ($_.Body).p -join "`n"

                [PSCustomObject]@{

                    PSTypeName = 'UncommonSense.HaarlemsDagblad.Article'

                    Url        = $CurrentUrl

                    Date       = $_.PublishedAt

                    Title      = $_.Title -replace '\s\|\scolumn$', ''

                    Body       = @($Intro, $Body) -join "`n`n"

                }

            }

        }

}