ARC

0.0.1

Public/Network/Get-WebPortalHTMLCode.ps1

FUNCTION Get-WebPortalHTMLCode {

    [CmdletBinding()] 
    PARAM (        
        [Parameter(Mandatory)]
        $IP,
        [ValidateSet('HTTP', 'HTTPS')]
        $Protocol = "HTTPS",               
        $Port = 443,
        $TimeoutSeconds = 5
    )

    $Matchfound = $False
    $NewURL = $null
    $NewPort = $null
    $NewPage = $null
    $SinglePage = $null
    $RegexChecks =  @('(?i)content\s*=\s*["'']?\d+;\s*url\s*=\s*''([^"''>\s]+)',
                      '(?i)top\.document\.location.href\s*=\s*["'']([^"''>\s]+)',
                      '(?i)\<script\>window.onload=function\(\){ url\s*=\s*["'']([^"''>\s]+)',
                      '(?i)http-equiv=Refresh content="0; URL=([^"]+)',
                      '(?i)onload="location.replace\({*["'']([^"''>\s]+)',
                      '(?i)onload="init\({*["'']([^"''>\s]+)',
                      '(?i)window.location ={*["'']([^"''>\s]+)'
                      '(?i)path.indexOf\({*["'']([^"''>\s]+)',
                      '(?i)window.location.href = {*["'']([^"''>\s]+)')

    IF ($PSVersionTable.PSVersion.Major -lt 7) { Write-Warning "This PowerShell Command requires PowerShell 7 to run properly."; RETURN }

    $PageTotal = $null
    
    TRY { 
        $SinglePage = (Invoke-WebRequest -Uri "$($Protocol)://$($IP):$Port" -UseBasicParsing -TimeoutSec $TimeoutSeconds -ErrorAction SilentlyContinue -SkipCertificateCheck).RawContent
    }
    CATCH { }
    
    IF ($SinglePage -eq $null) { Return $null }
    
    
    $PageTotal += $SinglePage
    $Count = 0
    
    WHILE ($true) {
        $Count++
        $Matchfound = $False

        FOREACH ($Check in $RegexChecks) {
            Write-Verbose "Checking for Match $Check"
            IF ($SinglePage -match $Check) { 

                $NewURL = $matches[1]

                ### Remove Leading '/'
                IF ($NewURL.Substring(0,1) -eq "/") { $NewURL = $NewURL.Substring(1,$NewURL.Length-1) }

                ### Get New Port If there is one.
                # IF ($NewURL -match '(?i)https?:\/\/[^:\/]+:(\d+)') { $NewPort = $Matches[1] } ELSE { $NewPort = $Port }

                $NewPage = $null

                TRY {
                $NewPage += (Invoke-WebRequest -Uri "$($Protocol)://$($IP):$Port/$NewURL" -UseBasicParsing -TimeoutSec $TimeoutSeconds -ErrorAction SilentlyContinue -SkipCertificateCheck).RawContent
                } CATCH { }

                IF ($NewPage -ne $null) { 

                    $PageTotal += $NewPage
                    $SinglePage = $NewPage
                    $Matchfound = $True

                    break 
                }
            }
        }

        IF ($Matchfound -eq $False) { break }
        IF ($Count -ge 4) { break }
    }

    RETURN $PageTotal
}