Advanced Search

Return to Snippet

Revision: 50500
at August 25, 2011 00:15 by danfsmith
Initial Code 
param ($path, $urlpath)

add-type -Path f:\dan\tools\html-agility-pack\HtmlAgilityPack.dll
$files = Get-ChildItem -Include *.htm,*.aspx -Path $path -Recurse
$doc = New-Object HtmlAgilityPack.HtmlDocument 
$result = $files | % { 
	Write-Host "Checking $_"
	$name = $_.FullName.Replace($path,$urlpath).Replace("\", "/")
	$htmldoc = $doc.Load($_.FullName) 
	$linknodes = $doc.DocumentNode.SelectNodes("//a")
	if ($linknodes) {
		foreach ($node in $linknodes) {	
			if ($node.GetAttributeValue("href", "").ToLower().Contains("pdf"))
			{
				Write-Host "Found" $node.GetAttributeValue("href", "")
				$pdflink = $node.GetAttributeValue("href", "")
				$line = $node.Line
				New-Object PsObject -Property @{PdfLink = $pdflink; FileName = $name; LineNumber = $line;}
			}
		}
	}
}
$result | Sort PdfLink
Initial URL 

                                

                                                            

                                    

                                        

                                        Initial Description
                                        

                                    

                                    

                                

                                                            

                                    

                                        

                                        Initial Title
                                        

                                    

                                    
Get All PDF links from HTML files

                                

                                                            

                                    

                                        

                                        Initial Tags
                                        

                                    

                                    
html

                                

                                                            

                                    

                                        

                                        Initial Language
                                        

                                    

                                    
Windows PowerShell