UncommonSense.Cinema

0.1.0

Public/Get-FilmtheaterHilversum.ps1

function Get-FilmtheaterHilversum
{
    [CmdletBinding()]
    param()

    $DutchCulture = Get-Culture 'nl-NL'
    $Today = Get-Date -Hour 0 -Minute 0 -Second 0 -Millisecond 0
    $Tomorrow = $Today.AddDays(1)

    Invoke-WebRequest -Uri https://www.filmtheaterhilversum.nl/films/ `
    | Select-Object -ExpandProperty Links `
    | Where-Object Href -Like 'https://filmtheaterhilversum.nl/film/*'
    | Select-Object -ExpandProperty HRef `
    | ForEach-Object {
        Write-Verbose "Processing movie url $($_)"
        $Url = $_
        $Document = ConvertTo-HtmlDocument -Uri $Url

        $Title = $Document | Select-HtmlNode -CssSelector 'h1' | Get-HtmlNodeText
        Write-Verbose "- Title is $Title"

        $Body = $Document | Select-HtmlNode -CssSelector '.content p strong' | Get-HtmlNodeText
        Write-Verbose "- Body is $Body"

        $Document
        | Select-HtmlNode -CssSelector '.movie-time' -All
        | ForEach-Object {
            $CurrentMovieTime = $_
            $ScreenDay = $CurrentMovieTime | Select-HtmlNode -CssSelector '.movie-screenday' | Get-HtmlNodeText
            Write-Verbose "- Screen day as text '$($ScreenDay)'"

            $ScreenDay = switch -regex ($ScreenDay)
            {
                'Vandaag' { $Today; break; }
                'Morgen' { $Tomorrow; break; }
                '^\w{2}\s(\d{1,2})\s(\w{3})$'
                {
                    # It seems that the nl-NL culture expects a period after the abbreviated Dutch month name
                    # on Ubuntu in GitHub actions, but not on MacOS? Moved away from DateTime.ParseExact to
                    # a more involved but hopefully more robust way of parsing.
                    $Day = [int]::Parse($Matches[1])

                    $Month = switch($Matches[2])
                    {
                        'jan' { 1; break; }
                        'feb' { 2; break; }
                        'mrt' { 3; break; }
                        'apr' { 4; break; }
                        'mei' { 5; break; }
                        'jun' { 6; break; }
                        'jul' { 7; break; }
                        'aug' { 8; break; }
                        'sep' { 9; break; }
                        'okt' { 10; break; }
                        'nov' { 11; break; }
                        'dec' { 12; break; }
                    }

                    Get-Date -Day $Day -Month $Month -Hour 0 -Minute 0 -Second 0 -Millisecond 0
                }
            }
            Write-Verbose "- Screen day as date $($ScreenDay)"

            if ($ScreenDay -lt $Today) # assume next year if day/month before today (end of December/start of January)
            {
                $ScreenDay = $ScreenDay.AddYears(1)
            }

            Write-Verbose "- Corrected screen day '$($ScreenDay.ToShortDateString())'"

            $CurrentMovieTime
            | Select-HtmlNode -CssSelector '.movie-timeblock' -All
            | ForEach-Object { $_ | Get-HtmlNodeText }
            | ForEach-Object { Write-Verbose "- Time '$($_)'"; $_ }
            | ForEach-Object { [TimeSpan]::ParseExact("$($_):00", 'g', $DutchCulture) }
            | ForEach-Object { $ScreenDay.Add($_) }
        }
        | ForEach-Object {
            [PSCustomObject]@{
                PSTypeName = 'UncommonSense.Cinema.Film'
                Url        = $Url
                Date       = $_
                Title      = $Title
                Body       = $Body
            }
        }
    }
}