public override void Parse()
{
string linkText = string.Empty;
string linkURL = string.Empty;
try
{
byte[] fileByteArray = Utility.GetFileByteArray(FileUrl);
if (fileByteArray != null)
{
using (MemoryStream fileStream = new MemoryStream(fileByteArray, false))
{
Mail_Message mime = Mail_Message.ParseFromStream(fileStream);
HtmlDocument doc = new HtmlDocument();
string html = string.IsNullOrEmpty(mime.BodyHtmlText) ? mime.BodyText : mime.BodyHtmlText;
if (!string.IsNullOrEmpty(html))
{
doc.LoadHtml(html);
foreach (HtmlNode link in doc.DocumentNode.SelectNodesOrEmpty("//a[@href]"))
{
if (link.Attributes["href"].Value.StartsWith("#", StringComparison.InvariantCultureIgnoreCase) == false &&
link.Attributes["href"].Value.StartsWith("javascript:", StringComparison.InvariantCultureIgnoreCase) == false &&
link.Attributes["href"].Value.StartsWith("mailto:", StringComparison.InvariantCultureIgnoreCase) == false)
{
linkURL = link.Attributes["href"].Value;
if (link.FirstChild == link.LastChild)
{
linkText = link.InnerText;
}
else
{
linkText = link.LastChild.InnerText;
}
linkText = new string(linkText.ToCharArray()).Replace("\r\n", " ");
FileLink objLink = new FileLink()
{
ParentFileUrl = FileUrl,
LinkText = linkText,
LinkAddress = linkURL
};
Results.Add(objLink);
}
}
}
}
}
}
catch (Exception ex)
{
FileLink objLink = new FileLink()
{
ParentFileUrl = FileUrl,
LinkText = "Error occurred when parsing this file.",
LinkAddress = ex.Message,
hasError = true
};
Results.Add(objLink);
}
}
public override void Parse()
{
StringBuilder hyperlinkText = null;
string hyperlinkRelationshipId = string.Empty;
try
{
byte[] fileByteArray = Utility.GetFileByteArray(FileUrl);
if (fileByteArray != null)
{
using (MemoryStream fileStream = new MemoryStream(fileByteArray, false))
{
using (WordprocessingDocument doc = WordprocessingDocument.Open(fileStream, false))
{
Document mainDocument = doc.MainDocumentPart.Document;
// Iterate through the hyperlink elements in the
// main document part.
foreach (DocumentFormat.OpenXml.Wordprocessing.Hyperlink hyperlink in mainDocument.Descendants <DocumentFormat.OpenXml.Wordprocessing.Hyperlink>())
{
if (hyperlink.Id != null)
{
hyperlinkText = new StringBuilder();
// Get the text in the document that is associated
// with the hyperlink. The text could be spread across
// multiple text elements so process all the text
// elements that are descendants of the hyperlink element.
foreach (DocumentFormat.OpenXml.Wordprocessing.Text text in hyperlink.Descendants <DocumentFormat.OpenXml.Wordprocessing.Text>())
{
hyperlinkText.Append(text.InnerText);
}
// The hyperlink element has an explicit relationship
// with the actual hyperlink. Get the relationship id
// via the hyperlink element's Id attribute.
hyperlinkRelationshipId = hyperlink.Id.Value;
// Get the hyperlink uri via the explicit relationship Id.
HyperlinkRelationship hyperlinkRelationship = doc
.MainDocumentPart.HyperlinkRelationships
.Single(c => c.Id == hyperlinkRelationshipId);
if (hyperlinkRelationship != null &&
hyperlinkRelationship.IsExternal)
{
if (hyperlinkRelationship.Uri.IsAbsoluteUri == false)
{
FileLink objLink = new FileLink()
{
ParentFileUrl = FileUrl,
LinkText = hyperlinkText.ToString(),
LinkAddress = hyperlinkRelationship.Uri.OriginalString,
hasError = true
};
Results.Add(objLink);
}
else
{
if (hyperlinkRelationship.Uri.Scheme != Uri.UriSchemeMailto)
{
FileLink objLink = new FileLink()
{
ParentFileUrl = FileUrl,
LinkText = hyperlinkText.ToString(),
LinkAddress = hyperlinkRelationship.Uri.AbsoluteUri
};
Results.Add(objLink);
}
}
}
}
}
}
}
}
}
catch (Exception ex)
{
FileLink objLink = new FileLink()
{
ParentFileUrl = FileUrl,
LinkText = "Error occurred when parsing this file.",
LinkAddress = ex.Message,
hasError = true
};
Results.Add(objLink);
}
}
private bool UrlIsValid(FileLink fileLink)
{
if (fileLink.hasError)
{
return(false);
}
string formatedURL = fileLink.LinkAddress;
Uri uri = null;
Uri.TryCreate(formatedURL, UriKind.Absolute, out uri);
if (uri == null)
{
if (formatedURL.StartsWith("/"))
{
Uri originalURI = null;
Uri.TryCreate(fileLink.ParentFileUrl, UriKind.Absolute, out originalURI);
if (originalURI != null)
{
formatedURL = originalURI.Scheme + "://" + originalURI.Host + ":" + originalURI.Port + formatedURL;
}
}
}
Uri.TryCreate(formatedURL, UriKind.Absolute, out uri);
if (uri == null)
{
return(false);
}
try
{
if (uri.Scheme == Uri.UriSchemeHttp || uri.Scheme == Uri.UriSchemeHttps)
{
if (errorHost.Contains(uri.Host))
{
return(false);
}
if (uri.Scheme == Uri.UriSchemeHttps)
{
ServicePointManager.ServerCertificateValidationCallback += (sender, certificate, chain, sslPolicyErrors) => true;
}
try
{
HttpWebRequest hwReq = HttpWebRequest.Create(uri) as HttpWebRequest;
hwReq.AutomaticDecompression = DecompressionMethods.GZip;
hwReq.UseDefaultCredentials = true;
hwReq.UserAgent = "uKnowlive Link Checker Tool";
hwReq.Timeout = 15000;
hwReq.Method = "HEAD";
hwReq.KeepAlive = false;
using (HttpWebResponse hwRes = hwReq.GetResponse() as HttpWebResponse)
{
return(hwRes == null ? false : hwRes.StatusCode == HttpStatusCode.OK);
}
}
catch (WebException wex)
{
if (wex.Status == WebExceptionStatus.NameResolutionFailure)
{
if (errorHost.Contains(uri.Host) == false)
{
errorHost.Add(uri.Host);
}
}
else if (wex.Status == WebExceptionStatus.ProtocolError)
{
if (wex.Response != null)
{
HttpWebResponse response = wex.Response as HttpWebResponse;
if (response.StatusCode == HttpStatusCode.NotFound)
{
try
{
HttpWebRequest hwReqNew = HttpWebRequest.Create(uri) as HttpWebRequest;
hwReqNew.AutomaticDecompression = DecompressionMethods.GZip;
hwReqNew.UseDefaultCredentials = true;
hwReqNew.UserAgent = "uKnowlive Link Checker Tool";
hwReqNew.Timeout = 15000;
hwReqNew.Method = "GET";
hwReqNew.KeepAlive = false;
using (HttpWebResponse hwResNew = hwReqNew.GetResponse() as HttpWebResponse)
{
try
{
hwReqNew.Abort();
}
catch {}
return(hwResNew == null ? false : hwResNew.StatusCode == HttpStatusCode.OK);
}
}
catch {}
}
//else if (response.StatusCode == HttpStatusCode.Unauthorized)
//{
// try
// {
// CredentialCache credCache = new CredentialCache();
// credCache.Add(uri, "NTLM", CredentialCache.DefaultNetworkCredentials);
// HttpWebRequest hwReqNew = HttpWebRequest.Create(uri) as HttpWebRequest;
// hwReqNew.AutomaticDecompression = DecompressionMethods.GZip;
// hwReqNew.Credentials = credCache;
// hwReqNew.UserAgent = "uKnowlive Link Checker Tool";
// hwReqNew.Timeout = 15000;
// hwReqNew.Method = "HEAD";
// hwReqNew.KeepAlive = false;
// using (HttpWebResponse hwResNew = hwReqNew.GetResponse() as HttpWebResponse)
// {
// return (hwResNew == null ? false : hwResNew.StatusCode == HttpStatusCode.OK);
// }
// }
// catch { }
//}
}
}
}
}
else if (uri.Scheme == Uri.UriSchemeFtp)
{
FtpWebRequest ftpWebRequest = FtpWebRequest.Create(uri) as FtpWebRequest;
ftpWebRequest.Method = WebRequestMethods.Ftp.ListDirectory;
ftpWebRequest.KeepAlive = false;
using (FtpWebResponse ftpRes = ftpWebRequest.GetResponse() as FtpWebResponse)
{
return(ftpRes != null);
}
}
else if (uri.Scheme == Uri.UriSchemeFile)
{
if (System.IO.Path.HasExtension(uri.LocalPath))
{
FileInfo fi = new FileInfo(uri.LocalPath);
return(fi.Exists);
}
else if (uri.Segments.Length > 1)
{
DirectoryInfo di = new DirectoryInfo(uri.LocalPath);
return(di.Exists);
}
else
{
IPHostEntry iph = Dns.GetHostEntry(uri.Host);
if (iph != null && iph.AddressList.Length > 0)
{
return(true);
}
}
}
else if (uri.Scheme == Uri.UriSchemeMailto)
{
return(true);
}
}
catch {}
return(false);
}
public override void Parse()
{
string hyperlinkText = string.Empty;
string hyperlinkRelationshipId;
try
{
byte[] fileByteArray = Utility.GetFileByteArray(FileUrl);
if (fileByteArray != null)
{
using (MemoryStream fileStream = new MemoryStream(fileByteArray, false))
{
using (PresentationDocument document = PresentationDocument.Open(fileStream, false))
{
// Iterate through all the slide parts in the presentation part.
foreach (SlidePart slidePart in document.PresentationPart.SlideParts)
{
IEnumerable <DocumentFormat.OpenXml.Drawing.HyperlinkType> links = slidePart.Slide.Descendants <DocumentFormat.OpenXml.Drawing.HyperlinkType>();
// Iterate through all the links in the slide part.
foreach (DocumentFormat.OpenXml.Drawing.HyperlinkType hyperlink in links)
{
if (hyperlink.Id != null)
{
hyperlinkText = Utility.GetPPTHyperlinkText(hyperlink);
hyperlinkRelationshipId = hyperlink.Id.Value;
HyperlinkRelationship hyperlinkRelationship = slidePart
.HyperlinkRelationships
.Single(c => c.Id == hyperlinkRelationshipId);
if (hyperlinkRelationship != null &&
hyperlinkRelationship.IsExternal)
{
if (hyperlinkRelationship.Uri.IsAbsoluteUri == false)
{
FileLink objLink = new FileLink()
{
ParentFileUrl = FileUrl,
LinkText = hyperlinkText,
LinkAddress = hyperlinkRelationship.Uri.OriginalString,
hasError = true
};
Results.Add(objLink);
}
else
{
if (hyperlinkRelationship.Uri.Scheme != Uri.UriSchemeMailto)
{
FileLink objLink = new FileLink()
{
ParentFileUrl = FileUrl,
LinkText = hyperlinkText,
LinkAddress = hyperlinkRelationship.Uri.AbsoluteUri
};
Results.Add(objLink);
}
}
}
}
}
}
}
}
}
}
catch (Exception ex)
{
FileLink objLink = new FileLink()
{
ParentFileUrl = FileUrl,
LinkText = "Error occurred when parsing this file.",
LinkAddress = ex.Message,
hasError = true
};
Results.Add(objLink);
}
}
