private string Describe(BrokenLinkType brokenLinkType)
{
switch (brokenLinkType)
{
case BrokenLinkType.External:
return(Localization.BrokenLink_External);
case BrokenLinkType.MediaLibrary:
return(Localization.BrokenLink_MediaLibrary);
case BrokenLinkType.Page:
return(Localization.BrokenLink_Page);
case BrokenLinkType.PageNotPublished:
return(Localization.BrokenLink_PageNotPublished);
case BrokenLinkType.Relative:
return(Localization.BrokenLink_Relative);
}
throw new InvalidOperationException("not supported link type " + brokenLinkType);
}
private XElement DescribeBrokenLink(XElement a, string link, BrokenLinkType brokenLinkType)
{
string previousnode = ToPreviewString(a.PreviousNode);
string nextnode = ToPreviewString(a.NextNode);
if (previousnode.Length > 50)
{
previousnode = "..." + previousnode.Substring(previousnode.Length - 40);
}
if (nextnode.Length > 50)
{
nextnode = nextnode.Substring(nextnode.Length - 40) + "...";
}
string errorText = Describe(brokenLinkType);
return(new XElement("invalidContent",
new XAttribute("previousNode", previousnode),
new XAttribute("originalText", a.Value),
new XAttribute("originalLink", link),
new XAttribute("nextNode", nextnode),
new XAttribute("errorType", errorText)));
}
//private object GetHostNameSyncObject(string hostname)
//{
// hostname = hostname.ToLowerInvariant();
// return _hostnameSync.GetOrAdd(hostname, h => new object());
//}
private bool SaveLinkCheckResult(string url, BrokenLinkType brokenLinkType)
{
_brokenLinks.TryAdd(url, brokenLinkType);
return(brokenLinkType == BrokenLinkType.None);
}
private RequestValidationInfo GetRequestValidationInfo(string url, string baseUrl, string serverUrl, BrokenLinkType brokenLinkType)
{
url = UrlHelper.ToAbsoluteUrl(url, baseUrl, serverUrl ?? _serverUrl);
string hostname;
try
{
hostname = new Uri(url).Host;
}
catch (UriFormatException)
{
return(null);
}
return(new RequestValidationInfo {
Hostname = hostname, LinkType = brokenLinkType, Url = url
});
}
public bool BuildBrokenLinksReport(XElement infoDocumentRoot)
{
using (new DataScope(PublicationScope.Published))
{
bool noInvalidLinksFound = true;
// Get all pages present in the console
List <IPage> actionRequiredPages = DataFacade.GetData <IPage>().ToList();
// Check security for each page (does the user have access - no need to bother the user with pages they do not have access to)
UserToken userToken = UserValidationFacade.GetUserToken();
var userPermissions = PermissionTypeFacade.GetUserPermissionDefinitions(userToken.Username).ToList();
var userGroupPermissions =
PermissionTypeFacade.GetUserGroupPermissionDefinitions(userToken.Username).ToList();
// Loop all pages and remove the ones the user has no access to
actionRequiredPages = actionRequiredPages.Where(page =>
PermissionTypeFacade.GetCurrentPermissionTypes(userToken, page.GetDataEntityToken(), userPermissions,
userGroupPermissions)
.Contains(PermissionType.Read)).ToList();
var pageIdsWithAccessTo = new HashSet <Guid>(actionRequiredPages.Select(p => p.Id));
var allSitemapElements = PageStructureInfo.GetSiteMap().DescendantsAndSelf();
var relevantElements =
allSitemapElements.Where(f => pageIdsWithAccessTo.Contains(new Guid(f.Attribute("Id").Value)));
var minimalTree =
relevantElements.AncestorsAndSelf().Where(f => f.Name.LocalName == "Page").Distinct().ToList();
var reportElements = new Hashtable <Guid, XElement>();
var linksToCheck = new List <LinkToCheck>();
// Rendering all the C1 pages and collecting links
foreach (XElement pageElement in minimalTree)
{
Guid pageId = new Guid(pageElement.Attribute("Id").Value);
IPage page = PageManager.GetPageById(pageId);
Verify.IsNotNull(page, "Failed to get the page");
string pageTitle = pageElement.Attribute("MenuTitle") != null
? pageElement.Attribute("MenuTitle").Value
: pageElement.Attribute("Title").Value;
var resultPageElement = new XElement(PageElementName,
new XAttribute("Id", pageId),
new XAttribute("Title", pageTitle));
reportElements[pageId] = resultPageElement;
string htmlDocument, errorCode;
string url = pageElement.Attribute("URL").Value;
string pageServerUrl = null;
if (url.StartsWith("http://") || (url.StartsWith("https://")))
{
pageServerUrl = new UrlBuilder(url).ServerUrl;
if (pageServerUrl == string.Empty)
{
pageServerUrl = url; /* Bug in versions < C1 4.0 beta 2 */
}
}
pageServerUrl = pageServerUrl ?? _serverUrl;
PageRenderingResult result = RenderPage(url, out htmlDocument, out errorCode);
if (result == PageRenderingResult.Failed)
{
resultPageElement.Add(GetRenderingErrorNode(errorCode));
continue;
}
if (result == PageRenderingResult.Redirect || result == PageRenderingResult.NotFound)
{
continue;
}
XDocument document;
try
{
document = XDocument.Parse(htmlDocument);
}
catch (Exception)
{
resultPageElement.Add(GetRenderingErrorNode(Localization.BrokenLinkReport_NotValidXhml));
continue;
}
linksToCheck.AddRange(CollectLinksToCheck(document, resultPageElement, url, pageServerUrl));
}
linksToCheck = linksToCheck.OrderBy(o => Guid.NewGuid()).ToList(); // Shuffling links
// Checking external and internall links in parrallel tasks - one per hostname
var linksGroupedByHostname = linksToCheck.Where(l => l.RequestValidationInfo != null)
.GroupBy(link => link.RequestValidationInfo.Hostname).ToList();
ParallelFacade.ForEach(linksGroupedByHostname, linkGroup =>
{
foreach (var linkToCheck in linkGroup)
{
linkToCheck.BrokenLinkType = ValidateByRequest(linkToCheck.RequestValidationInfo);
// linkToCheck.RequestValidationInfo = null;
}
});
// Having 100 tasks running in parallel would fill the app pool and make the site unresponsive
foreach (var link in linksToCheck)
{
if (!link.BrokenLinkType.HasValue)
{
Log.LogWarning(LogTitle, "Incorrectly processed link: " + link.Href);
link.BrokenLinkType = BrokenLinkType.Relative;
}
BrokenLinkType brokenLinkType = link.BrokenLinkType.Value;
if (brokenLinkType == BrokenLinkType.None)
{
continue;
}
var brokenLinkDescriptionElement = DescribeBrokenLink(link.LinkNode, link.Href, brokenLinkType);
link.ReportPageNode.Add(brokenLinkDescriptionElement);
noInvalidLinksFound = false;
}
BuildReportTreeRec(infoDocumentRoot, Guid.Empty, reportElements);
return(noInvalidLinksFound);
}
}
/// <summary>
/// Pasres the url, and checks if the urls can be validated without making an http request (for internal urls=.
/// </summary>
private UrlPreprocessResult PreprocessUrl(
string url,
string pageUrl,
string serverUrl,
out BrokenLinkType brokenLinkType,
out RequestValidationInfo requestValidationInfo)
{
BrokenLinkType cachedResult;
requestValidationInfo = null;
if (_brokenLinks.TryGetValue(url, out cachedResult))
{
brokenLinkType = cachedResult;
return(brokenLinkType == BrokenLinkType.None ? UrlPreprocessResult.Valid : UrlPreprocessResult.Broken);
}
// Trying to parse as a page url first
PageUrlData pageUrlData = null;
try
{
if (IsKnownHostname(url)) // Workaround "if" for early vesrions of 4.0 beta
{
pageUrlData = PageUrls.ParseUrl(url);
}
}
catch (UriFormatException)
{
}
if (pageUrlData != null)
{
Guid linkedPageId = pageUrlData.PageId;
IPage page;
using (new DataScope(pageUrlData.PublicationScope, pageUrlData.LocalizationScope))
{
page = PageManager.GetPageById(linkedPageId);
}
if (page == null)
{
if (pageUrlData.PublicationScope == PublicationScope.Published)
{
using (new DataScope(PublicationScope.Unpublished, pageUrlData.LocalizationScope))
{
if (PageManager.GetPageById(linkedPageId) != null)
{
brokenLinkType = BrokenLinkType.PageNotPublished;
return(SaveLinkCheckResult(url, brokenLinkType) ? UrlPreprocessResult.Valid : UrlPreprocessResult.Broken);
}
}
}
brokenLinkType = BrokenLinkType.Page;
return(SaveLinkCheckResult(url, brokenLinkType) ? UrlPreprocessResult.Valid : UrlPreprocessResult.Broken);
}
// If no PathInfo - page link is already valid
if (string.IsNullOrEmpty(pageUrlData.PathInfo))
{
brokenLinkType = BrokenLinkType.None;
return(UrlPreprocessResult.Valid);
}
// If there's pathInfo -> making a request to check whether the link is actually broken
requestValidationInfo = GetRequestValidationInfo(url, pageUrl, serverUrl, BrokenLinkType.Page);
if (requestValidationInfo == null)
{
brokenLinkType = BrokenLinkType.Page;
return(UrlPreprocessResult.Broken);
}
brokenLinkType = BrokenLinkType.None;
return(UrlPreprocessResult.NeedToBeValidatedByRequest);
}
MediaUrlData mediaUrlData = MediaUrls.ParseUrl(url);
if (mediaUrlData != null)
{
Guid mediaId = mediaUrlData.MediaId;
string mediastore = mediaUrlData.MediaStore;
bool mediaExist = DataFacade.GetData <IMediaFile>().Any(f => f.StoreId == mediastore && f.Id == mediaId);
brokenLinkType = mediaExist ? BrokenLinkType.None : BrokenLinkType.MediaLibrary;
return(SaveLinkCheckResult(url, brokenLinkType) ? UrlPreprocessResult.Valid : UrlPreprocessResult.Broken);
}
var linkType = UrlHelper.IsAbsoluteLink(url) ? BrokenLinkType.External : BrokenLinkType.Relative;
requestValidationInfo = GetRequestValidationInfo(url, pageUrl, serverUrl, linkType);
if (requestValidationInfo == null)
{
brokenLinkType = linkType;
return(UrlPreprocessResult.Broken);
}
brokenLinkType = BrokenLinkType.None;
return(UrlPreprocessResult.NeedToBeValidatedByRequest);
}
private RequestValidationInfo GetRequestValidationInfo(string url, string baseUrl, string serverUrl, BrokenLinkType brokenLinkType)
{
url = UrlHelper.ToAbsoluteUrl(url, baseUrl, serverUrl ?? _serverUrl);
string hostname;
try
{
hostname = new Uri(url).Host;
}
catch (UriFormatException)
{
return null;
}
return new RequestValidationInfo { Hostname = hostname, LinkType = brokenLinkType, Url = url};
}
//private object GetHostNameSyncObject(string hostname)
//{
// hostname = hostname.ToLowerInvariant();
// return _hostnameSync.GetOrAdd(hostname, h => new object());
//}
private bool SaveLinkCheckResult(string url, BrokenLinkType brokenLinkType)
{
_brokenLinks.TryAdd(url, brokenLinkType);
return brokenLinkType == BrokenLinkType.None;
}
private string Describe(BrokenLinkType brokenLinkType)
{
switch (brokenLinkType)
{
case BrokenLinkType.External:
return Localization.BrokenLink_External;
case BrokenLinkType.MediaLibrary:
return Localization.BrokenLink_MediaLibrary;
case BrokenLinkType.Page:
return Localization.BrokenLink_Page;
case BrokenLinkType.PageNotPublished:
return Localization.BrokenLink_PageNotPublished;
case BrokenLinkType.Relative:
return Localization.BrokenLink_Relative;
}
throw new InvalidOperationException("not supported link type " + brokenLinkType);
}
/// <summary>
/// Pasres the url, and checks if the urls can be validated without making an http request (for internal urls=.
/// </summary>
private UrlPreprocessResult PreprocessUrl(
string url,
string pageUrl,
string serverUrl,
out BrokenLinkType brokenLinkType,
out RequestValidationInfo requestValidationInfo)
{
BrokenLinkType cachedResult;
requestValidationInfo = null;
if (_brokenLinks.TryGetValue(url, out cachedResult))
{
brokenLinkType = cachedResult;
return brokenLinkType == BrokenLinkType.None ? UrlPreprocessResult.Valid : UrlPreprocessResult.Broken;
}
// Trying to parse as a page url first
PageUrlData pageUrlData = null;
try
{
if (IsKnownHostname(url)) // Workaround "if" for early vesrions of 4.0 beta
{
pageUrlData = PageUrls.ParseUrl(url);
}
}
catch (UriFormatException)
{
}
if (pageUrlData != null)
{
Guid linkedPageId = pageUrlData.PageId;
IPage page;
using (new DataScope(pageUrlData.PublicationScope, pageUrlData.LocalizationScope))
{
page = PageManager.GetPageById(linkedPageId);
}
if (page == null)
{
if (pageUrlData.PublicationScope == PublicationScope.Published)
{
using (new DataScope(PublicationScope.Unpublished, pageUrlData.LocalizationScope))
{
if (PageManager.GetPageById(linkedPageId) != null)
{
brokenLinkType = BrokenLinkType.PageNotPublished;
return SaveLinkCheckResult(url, brokenLinkType) ? UrlPreprocessResult.Valid : UrlPreprocessResult.Broken;
}
}
}
brokenLinkType = BrokenLinkType.Page;
return SaveLinkCheckResult(url, brokenLinkType) ? UrlPreprocessResult.Valid : UrlPreprocessResult.Broken;
}
// If no PathInfo - page link is already valid
if (string.IsNullOrEmpty(pageUrlData.PathInfo))
{
brokenLinkType = BrokenLinkType.None;
return UrlPreprocessResult.Valid;
}
// If there's pathInfo -> making a request to check whether the link is actually broken
requestValidationInfo = GetRequestValidationInfo(url, pageUrl, serverUrl, BrokenLinkType.Page);
if (requestValidationInfo == null)
{
brokenLinkType = BrokenLinkType.Page;
return UrlPreprocessResult.Broken;
}
brokenLinkType = BrokenLinkType.None;
return UrlPreprocessResult.NeedToBeValidatedByRequest;
}
MediaUrlData mediaUrlData = MediaUrls.ParseUrl(url);
if (mediaUrlData != null)
{
Guid mediaId = mediaUrlData.MediaId;
string mediastore = mediaUrlData.MediaStore;
bool mediaExist = DataFacade.GetData<IMediaFile>().Any(f => f.StoreId == mediastore && f.Id == mediaId);
brokenLinkType = mediaExist ? BrokenLinkType.None : BrokenLinkType.MediaLibrary;
return SaveLinkCheckResult(url, brokenLinkType) ? UrlPreprocessResult.Valid : UrlPreprocessResult.Broken;
}
var linkType = UrlHelper.IsAbsoluteLink(url) ? BrokenLinkType.External : BrokenLinkType.Relative;
requestValidationInfo = GetRequestValidationInfo(url, pageUrl, serverUrl, linkType);
if (requestValidationInfo == null)
{
brokenLinkType = linkType;
return UrlPreprocessResult.Broken;
}
brokenLinkType = BrokenLinkType.None;
return UrlPreprocessResult.NeedToBeValidatedByRequest;
}
private XElement DescribeBrokenLink(XElement a, string link, BrokenLinkType brokenLinkType)
{
string previousnode = ToPreviewString(a.PreviousNode);
string nextnode = ToPreviewString(a.NextNode);
if (previousnode.Length > 50)
{
previousnode = "..." + previousnode.Substring(previousnode.Length - 40);
}
if (nextnode.Length > 50)
{
nextnode = nextnode.Substring(nextnode.Length - 40) + "...";
}
string errorText = Describe(brokenLinkType);
return new XElement("invalidContent",
new XAttribute("previousNode", previousnode),
new XAttribute("originalText", a.Value),
new XAttribute("originalLink", link),
new XAttribute("nextNode", nextnode),
new XAttribute("errorType", errorText));
}
public BrokenLink(Uri BrokenUrl, BrokenLinkType TypeOfLink)
{
Url = BrokenUrl;
LinkType = TypeOfLink;
}
/// <summary>
/// Validates a link by checking the cache to see if the response has already been checked and recording broken links
/// </summary>
/// <param name="PageUri">The Uri of the page the link is on</param>
/// <param name="UriToValidate">The Uri of the link to check</param>
/// <param name="LinkType">The type of the link being checked</param>
private void ValidateLink(Uri PageUri, Uri UriToValidate, string originalUrl, BrokenLinkType LinkType)
{
//Validate if original url is absolute and its pointing to authoring or publishing site instead of relative link
Uri originalUri = new Uri(originalUrl, UriKind.RelativeOrAbsolute);
if (originalUri.IsAbsoluteUri)
{
string originalAuthority = originalUri.GetLeftPart(UriPartial.Authority).ToLower();
string pageAuthority = PageUri.GetLeftPart(UriPartial.Authority).ToLower();
if (originalAuthority.Equals(WWWUrl.ToLower()))
{
//Absolute URL for www site
}
else if (originalAuthority.Equals(pageAuthority))
{
//Absolute URL for publishing site
}
else
{
//External Url
LinkType = BrokenLinkType.ExternalLink;
}
}
if (!CheckedLinks.Contains(UriToValidate.ToString()))
{
if (UriToValidate.ToString().ToLower().Contains("fixupredirect.aspx"))
{
CheckedLinks.Add(UriToValidate.ToString(), true);
}
else if (!LinkCheckerUtilities.LinkIsValid(UriToValidate))
{
Utility.LogInformation("Found broken link: " + UriToValidate.ToString());
AddBrokenLink(PageUri, new BrokenLink(UriToValidate, LinkType));
CheckedLinks.Add(UriToValidate.ToString(), false);
}
else
{
CheckedLinks.Add(UriToValidate.ToString(), LinkType != BrokenLinkType.ExternalLink);
}
}
}
