private UriResourceItem AnalyzeUri(Uri parentUri)
{
var uriResourceItem = new UriResourceItem(parentUri);
if (uriResourceItem.SchemeType == WebToolsCore.Enumerations.UriSchemeType.HTTP || uriResourceItem.SchemeType == WebToolsCore.Enumerations.UriSchemeType.HTTPS)
{
//
// Analyze elements that could contain hrefs
var htmlDocumentContent = WebUtil.GetHtmlContent(uriResourceItem.CleanUri);
if (htmlDocumentContent != default)
{
// Analyze links
foreach (string attributeValue in htmlDocumentContent.GetAllNodes(new[] { "*[@href]", "*[@src]" }).GetAttributeValues(new[] { "href", "src" }))
{
try
{
var childUri = UriUtil.ToAbsoluteUri(uriResourceItem.CleanUri, attributeValue);
if (uriResourceItem.ReferencedUris.Contains(childUri) == false) // If we don't already have this uri, lets add it
{
uriResourceItem.ReferencedUris.Add(childUri);
}
}
catch { }
}
// Are we deep scanning?
if (this.deepScanCheckbox.Checked == true)
{
// Raw regex check
foreach (var match in RegexUtil.FindLinkMatches(htmlDocumentContent.Text))
{
try
{
var childUri = UriUtil.ToAbsoluteUri(uriResourceItem.CleanUri, match);
if (uriResourceItem.ReferencedUris.Contains(childUri) == false) // If we don't already have this uri, lets add it
{
uriResourceItem.ReferencedUris.Add(childUri);
}
}
catch { }
}
}
}
else
{
// The current link isnt a html parsable document
// Is deep scan enabled?
if (this.deepScanCheckbox.Checked == true)
{
var genericDocumentContent = WebUtil.GetDocumentContent(parentUri);
if (genericDocumentContent != default)
{
foreach (var match in RegexUtil.FindLinkMatches(genericDocumentContent))
{
try
{
var childUri = UriUtil.ToAbsoluteUri(uriResourceItem.CleanUri, match);
if (uriResourceItem.ReferencedUris.Contains(childUri) == false) // If we don't already have this uri, lets add it
{
uriResourceItem.ReferencedUris.Add(childUri);
}
}
catch { }
}
}
}
}
}
return(uriResourceItem);
}