Public/Get-SylviaWitteman.ps1
|
function Get-SylviaWitteman { [CmdletBinding()] param() $DutchCulture = New-Object -TypeName System.Globalization.CultureInfo -ArgumentList 'nl-NL' curl --silent --location 'https://www.parool.nl/auteur/sylvia-witteman' | Join-String -Separator ' ' | ConvertTo-HtmlDocument | ForEach-Object { $_.DocumentNode.SelectNodes('//a') } | Where-Object { $_.HasAttributes } | ForEach-Object { $_.GetAttributeValue('href', '') } | Where-Object { $_ -like 'https://www.parool.nl/columns-opinie/?*' } | ForEach-Object { $Document = curl --silent --location $_ | Join-String -Separator ' ' | ConvertTo-HtmlDocument $DateText = (($Document | Select-HtmlNode -XPath '//meta[@property="article:published_time"]').GetAttributeValue("content", "") -split 'T')[0] $Paragraphs = $Document | Select-HtmlNode -CssSelector 'p.z3lfzo5' -All | Get-HtmlNodeText -SkipRemoveLineBreaks $Body = ($Paragraphs | ForEach-Object { $_.Trim() } | Where-Object { $_ } | ForEach-Object { "<p>$($_)</p>" }) -join "`n" $Body = $Body -replace 'Reageren\? s\.witteman@parool\.nl\.?\s*', '' $Body = $Body -replace 'Sylvia Witteman \(1965\) is columniste en schrijfster\.\s*', '' $Body = $Body -replace 'Drie keer per week lees je hier haar column\.\s*', '' $Body = $Body -replace 'Lees alle columns van Sylvia Witteman terug\.\s*', '' [PSCustomObject][Ordered]@{ PSTypeName = 'UncommonSense.Parool.Article' Url = $_ Date = [DateTime]::ParseExact($DateText, 'yyyy\-MM\-dd', $DutchCulture) Title = $Document | Select-HtmlNode -CssSelector h1 | Get-HtmlNodeText Body = $Body } } } |