Managing the Sitecore xDB Contact Card without requiring the Tracker Revisited

Intro

Previously I mentioned how to update the contact card without the session.  I found that I was occasionally still getting locking issues, so I looked into it a bit further.  What I had was a little redundant for saving the update.

Saving Updates

When saving the updates really you do not need to call SaveAndReleaseContact, FlushContactToXdb should do the trick, but you should call it with the overload that takes two parameters and passin a new ContactSaveOptions with release set to true, and a new LeaseOwner.

var manager = Factory.CreateObject("tracking/contactManager", true) as ContactManager;

manager.FlushContactToXdb(contact, new ContactSaveOptions(true, new LeaseOwner(AnalyticsSettings.ClusterName, LeaseOwnerType.WebCluster)));

manager.SaveAndReleaseContact(contact);

Conclusion

While this is only a short note on the last post, be sure to wrap your contact manager to ensure you maintain the same contact card between your updates and when you flush the card or your changes will be lost, and only identify the contact when absolutely necessary to prevent a chance for potential conflict with locking.

Sitecore Caching Revisited

Previously I wrote about using the Sitecore Custom Cache.  Unfortunately I don’t think I put enough detail in that post.  Some may have been a little confused by it. 

Creating the Cache

Really creating a cache is almost too easy, you simply create a class that inherits from CustomCache and then implement SetObject  and GetObject as I said in my previous post.

public class MyCustomCache : CustomCache

 {

     public MyCustomCache(string name, long maxSize)

         : base(name, maxSize)

     {

     }

     public void SetObject(string key, object value)

     {

         base.SetObject(key, value, Sitecore.Reflection.TypeUtil.SizeOfObject());

     }

     public new object GetObject(object key)

     {

         return base.GetObject(key);

     }

 }

The Cache Manager

The problem with stopping here is though we have the cache we do not have an instance of it created.  This is also a very simple task.  The easiest way to implement it is just to create a static instance of it somewhere.  More properly though would be to create a static cache manager. 

public static class MyCacheManager

{

    private static readonly MyCustomCache Cache = new MyCustomCache(

        "My.NameSpace.Custom.Cache",

        StringUtil.ParseSizeString(Sitecore.Configuration.Settings.GetSetting("My.Custom.Cache.Size.Setting.Name", "5MB")));

    public static object Get(string key)

    {

        return Cache.GetObject(key);

    }

    public static void Set(string key, object value)

    {

        Cache.SetObject(key, value);

    }

}

Clearing the Cache

It really is that easy.  It may be wise though to add a cache clearing mechanism as well maybe add a new static method to the manager.

public static void Clear()

{

    Cache.Clear();

}

Of course you still need to trigger this call using the Sitecore Pipelines which is also very easy, just add the call anywhere Sitecore normally clears the cache.

Managing the Sitecore xDB Contact Card without requiring the Tracker

The Normal Way

When working with a user profile in Sitecore 8, we generally get the contact card by using Tracker.Current.Contact.  Of course if the tracker is null we will call Tracker.Start(), and of course that would also mean the Contact is null so we would also need to call Tracker.Identify().

This works out well in most situations since the tracker is maintained across the session.  If you don’t have a session though this could mean a big headache.  You would normally need to create a Session and an HttpContext, then create and push a tracker into the stack.  While this is not excessively resource intensive when you are doing a lot of calls it can add up quickly.

Getting the Card

There is an alternate way to get the card though.  You can use the ContactManager to load the contact by username, if there is not a contact identified by that username then it will return null and you can then create a new Contact and insert it.

var user = Context.User.Identity.Name;

if (string.IsNullOrEmpty(user)) return null;

var contactManager = Factory.CreateObject("tracking/contactManager", true) as ContactManager;

var contact = contactManager.LoadContactReadOnly(user);

Saving Updates

If you are updating this contact keep in mind that it will not persist across the session so you will need to flush it to the database before the end of the request.  This is also done with the Contact Manager by calling FlushContactToXdb and passing the contact, then by calling SaveAndReleaseContact also passing in the contact. 

var manager = Factory.CreateObject("tracking/contactManager", true) as ContactManager;

manager.FlushContactToXdb(contact);

manager.SaveAndReleaseContact(contact);

Submitting your Session

Just to be on the safe side of things it is not a bad idea to submit your context.  You may be thinking that that requires a session.  Fortunately Sitecore can create a very lightweight session context for this use.  You can use the SessionContextManager and call GetSession passing in the contact id, an empty guid, and two Boolean true values.  Once you have the session you can call Submit on the SessionContextManager and pass this session in.

var contextManager =

    Factory.CreateObject("tracking/sessionContextManager", true) as SessionContextManagerBase;

var session = contextManager.GetSession(c.ContactId, Guid.Empty, true, true);

contextManager.Submit(session);

While this solution is not perfect, you can expand on this solution to build a robust contact manager that does not depend on there being an active session.

Sitecore Serialization Synchronization Service (S4)

I know most of my posts have been instructional so far, so it is about time I make a bit of an announcement.  I am working on a new tool for Sitecore Development.  I would love input from the community of what their biggest expectations for this tool would be.

This tool is called the Sitecore Serialization Synchronization Service or S4 for short.  Basically it is a synchronization tool that connects to your Sitecore instance and keeps serialized items in sync.  You update an item and the item in the file system will be updated as well, this also goes the other way around.  You pull the latest from source control and it adds or modifies files in your sync folder and it pushes those changes into Sitecore.

At this point the serialization engine is almost complete, the biggest key for me here is performance.  I do not want this tool to take much memory or processor power.  I also do not expect to have any sort of Visual Studio integration, at least not that is required for use of the tool.

I do not plan to implement any sort of Code Gen or Code Deployment in this tool.  Visual Studio (and other IDE's) already have this functionality, so though it might be easier with a tool it is not 100% necessary.  I plan to do a tutorial on Code Gen with serialized Sitecore Items or possibly with S4 as a data provider.

S4 also will not deploy on build since it is not integrated with Visual Studio.  While at first this may sound like a breaking feature, S4 is intended to keep your items in sync at all time so there is no need to deploy the items on build.

If there are any questions, comments or requests with this tool please let me know.  I appreciate all feedback.  If the response is large enough I will push the development of this tool a bit harder.  As it stands right now I should hopefully have a beta ready (likely just base functionality without a friendly UI or anything) in the next month or so, so please check back often for updates.

Leveraging Sitecore's Caching Mechanism

In the time I have been developing in Sitecore I have seen many different approaches to caching, but only once have I seen someone actually use caching based on Sitecore's cache service.  It is in fact almost how scary it is to implement.

Basically all you have to do is create a class that inherits from Sitecore.Caching.CustomCache.  Of course you need to implement a constructor that calls the base constructor, and override the SetObject and GetObject methods.  That is pretty much it though.

This cache will be handled in the same admin page as the other Sitecore caches, and if you install the Cache Admin or Cache Manager shared source modules you can easily use them to monitor and manage the caches without needing to build your own custom admin page.

I highly recommend at least investigating this approach to anyone considering putting in a custom cache on their site.  It is efficient enough for most scenarios as well as being very easy to setup potentially saving you hours of coding.

Getting Started with Sitecore Content Search using Lucene or Solr

Using Sitecore Content Search with Solr is basically identical to that with Lucene at the general use level, with the exception of the configuration.  Solr does have some more advanced features that Lucene does not have. 

To get started though with either all we need is a search context.  We can get this using the context item (or any item for that matter).  We simply cast the item to the SitecoreIndexableItem class and pass it into ContentSearchManager.CreatesearchContext and that will give us our context.

This is generally set up as:
using (var context = ContentSearchManager.CreateSearchContext((SitecoreIndexableItem)item)){
   //search code….
}

From inside this context we can get our search results by calling the generic method GetQueryable that takes all types that derive from SearchResultItem on our context and using Linq to filter our results.

An example where we get all items under the current item:
var results = context.GetQueryable<SearchResultItem>().Where(i => i.Paths.Contains(item.ID));

If you have custom fields you want to search on it is highly recommended that you create a class that derives from SearchResultItem and add your fields as properties.  What is really nice with this is that Sitecore content search will reflect on this class and do the search on those properties.

An example with an item with an author field:
public class MyCustomResultItem: SearchResultItem{
    public string Author { get; set; }
}

var results = context.GetQueryable<MyCustomResultItem>().Where(i => string.Equals(i.Author,”me”));

At this point you can iterate through results and get the items using .GetItem() on each resulting item.

Using SignalR with Sitecore

If you are unfamiliar with SignalR and need to set it up I will be posting an intro to SignalR on my other blog (Programming Medeley) in a few days.  Otherwise reading this post I will assume that you have a basic understanding of what SignalR is and how it works.

When developing in Sitecore with SignalR there is one major issue that you will run into (aside from basic routing).  That issue is the fact that you do not have a Sitecore site context within your SignalR context.  This causes issues with getting item URLs or using the Glass Mapper.

Missing Site Context

The fix for this is actually quite simple though.  You simply need to set the Sitecore site context on every request. 

To get the SiteContext you simply call Sitecore.Sites.SiteContextFactory.GetSiteContext and pass in the hostname (which can be found within the context of the Hub as this.Context.Request.Url.Host) and an empty path and it will give you a SiteContext. 

Once we have this all we have to do is set the current site to this context by setting Sitecore.Context.Site to this SiteContext. 

I use the single line:
Sitecore.Context.Site = Sitecore.Context.Site ?? SitecontextFactory.GetSiteContext(this.Context.Request.Url.Host, “/”);

Routing

As for the routing issue that you may run into, this is the same as setting up WebApi.  You need to ensure the route is registered in the Application_Start of the Global.asax. 

The command for this is (using the Global Configuration):
GlobalConfiguration.Configuration.Routes.MapHttpRoute(“signalrHub”, “signalr/{controller}”);          

Saving Form data with the Web Forms For Marketers MVC Form API

As many people have noticed with the Web Forms For Marketers MVC Forms the data does not serialize correctly so using custom save actions you cannot get the data from the form correctly.

There are a few rather complex workarounds for this mostly involving parsing the form from the form values in the http response.

The easiest solution to this issue is to create your own Form Processor which you can inject with a configuration file.

<configuration xmlns:patch="http://www.sitecore.net/xmlconfig/">
  <sitecore>
    <wffm>
      <formProcessors>
        <processor type="MyNamespace.Forms.MVC.Processors.FormProcessor, MyNamespace.Forms.MVC" />
      </formProcessors>
    </wffm>
  </sitecore>
</configuration>

The processor itself is actually quite easy to write you simply need a class that inherits Sitecore.Forms.MVC.Processors.FormProcessor, then you can implement your save action in the FormSuccessValidation event.

namespace MyNamespace.Forms.MVC.Processors
{
    public class FormProcessor : FormProcessor
    {
        public FormProcessor()
        {
            SuccessValidation += FormSuccessValidation;
        }
 
        private void SaveForm(FormModel form)
        {
            var Fields = form.Sections.SelectMany(i => i.Fields);
        }
 
        public void FormSuccessValidation(object source, FormEventArgs args)
        {
            SaveForm(args.Form);
        }
    }
}

You can see from my code that you can access the fields through the given model in the form.Sections property.  

I find that this is much easier to implement than a custom save action (though it will not show up in the drop down list which may be a problem for some people, you can use this to store the data to access in a custom save action).  


If you have any questions related to Web Forms For Marketers please don't hesitate to ask as I have spent a lot of time in that code lately.

Error 500.19 Configuration File Too Long Sitecore Web.config

It is quite well known that the web.config file in Sitecore is monstrous.  It takes very little to make it large enough to get the 500.19 .NET error with the file being too large.  Most resolutions to this involve modifying the registry, but it does not really have to be that difficult.

The easiest solution is to take the Sitecore node out of the file and put it in a separate file similar to how the connection strings are pulled out.

The easiest way to get the whole node selected would be to use a text editor that supports regular expressions and use the expression (<sitecore)[\s\S]*(</sitecore>) to select it.  Then you simply cut it and past it into a file (I generally put it in a file called Sitecore.config in the App_Config folder).

After this you need to put an empty Sitecore node where you removed the old one from and add a configsource attribute that points to the new config file.

<sitecore configSource="App_Config/Sitecore.config" />

Language Fallback: A word of caution

The concept of Language Fallback in Sitecore is great, actually having the ability to have Sitecore return content in a different language if none exists in the current context is tremendously helpful.  Even the concept behind this functionality is pretty straight forward. 

The module simply checks if there is a version in the current language.  If not it checks if there is a fallback for the current language and repeats the process on that language if there is until it either finds a version of the item or creates an empty item to return. 

Due to the simplicity of this module it has continued to work with every version of Sitecore since it was introduced, and will likely continue working for years to come.

Unfortunately, if you think about the recursion in this module there is a large problem that can happen.  Endless loops can occur when languages eventually fall back to the original language.

The current module in the marketplace written by Alex Shyba does not have checks to ensure that you do not have an endless loop.  Nikola Gotsev and I ran into this problem a while back where we had a language that would fall back to another language that fell back to the first language.  Not knowing this content error had happened led us into a bit of a nightmare of disabling modules and checking changes to find what brought our environment to its’ knees.

Nikola actually posted a fix for this last month.  This fix simply includes checking if we have already fallen back to a specific language (by enumerating fallback languages in a list).  This solves the problem no matter how complex the chain is.

Dynamic Key Placeholders Revisited

A little over a year ago I did a post on Dynamic Key Placeholders, since then I have noticed there was some information lacking, such as what versions of Sitecore it works with, and how to get Placeholder Settings to work.

From my testing this should work in every version of Sitecore from 6.5 to 8.0, and likely will continue working in previous and future versions as well.

The Old Way  

By default when using dynamic key placeholders you can create placeholder settings for each index of the key.  Such as if the original key was Placeholder, you would need to create settings for Placeholder1, Placeholder2, etc… This can lead to a lot of work, and a cluttered collection of placeholder settings.

 

A New Way to Do It

An easier way would be to create an override of the GetAllowedRenderings processor to allow for an experience more like the original when creating placeholder settings.

A Quick Modification to the Original

In the original source for the dynamic key placeholders we need to make a simple modification.  In the part of the code that we set the key I have added the # symbol in there to allow us to split it apart far easier to create an override for Placeholder Settings.  You can use any symbol you want, but keep I will continue with this one for the rest of the post.

                // Count represents how many of the same container have already been added to the page

                if (renderings.Any())

                {

                    _dynamicKey = _key + "#" + renderings.Count();

                }

Our New Processor

The big changes though are in the addition of overriding the GetAllowedRenderings processor.  The change is not too complex as we are basically mirroring what Sitecore is already doing, but with the exception of overriding the placeholder key.

To override the key we want to split the key first into the segments by the / character and then for each segment we use the split on the # to remove our index from the dynamic key giving us the original key.

After that we want to try to get the placeholder item using our cleaned up key.  Of course if there is a device specified we should get the item in the context of that device.

If the Item exists we should get the list of renderings it supports.

    public class GetAllowedRenderings : Sitecore.Pipelines.GetPlaceholderRenderings.GetAllowedRenderings

    {

        /// <summary>

        /// This is a method based on the original Sitecore method to allow for dynamic placeholder keys

        /// </summary>

        /// <param name="args">The Placeholder Rendering Args (Device, Key, Database, etc...)</param>

        public new void Process(GetPlaceholderRenderingsArgs args)

        {

            Assert.IsNotNull((object)args, "args");

            //Remove the dynamic indexing from the key

            var key = string.Join("/", args.PlaceholderKey.Split('/').Select(i => i.Split('#')[0]));

            //Get the placeholder item, if there is a Device selected get it using that Device Context

            Item placeholderItem = null;

            if (ID.IsNullOrEmpty(args.DeviceId))

            {

                placeholderItem = Client.Page.GetPlaceholderItem(key, args.ContentDatabase, args.LayoutDefinition);

            }

            else

            {

                using (new DeviceSwitcher(args.DeviceId, args.ContentDatabase))

                {

                    placeholderItem = Client.Page.GetPlaceholderItem(key, args.ContentDatabase, args.LayoutDefinition);

                }

            }

           

            //If there was a placeholder item (Placeholder Settings) set the allowed renderings

            if (placeholderItem != null)

            {

                args.HasPlaceholderSettings = true;

                bool allowedControlsSpecified;

                args.PlaceholderRenderings = this.GetRenderings(placeholderItem, out allowedControlsSpecified);

                if (allowedControlsSpecified)

                    args.Options.ShowTree = false;

            }

        }

    }

Our Configuration Change

Finally we have the configuration file to enable this override, this is simply done by replacing the processor in the getPlaceholderRenderings pipeline.

<configuration xmlns:patch="http://www.sitecore.net/xmlconfig/">

  <sitecore>

    <pipelines>

      <getPlaceholderRenderings>                              

        <processor patch:instead="*[@type='Sitecore.Pipelines.GetPlaceholderRenderings.GetAllowedRenderings, Sitecore.Kernel']" type="[Your Method], [Your Assembly] />

      </getPlaceholderRenderings>

    </pipelines>

  </sitecore>   

</configuration>

You can find the source from this post at https://github.com/musicman89/Sitecore-Dynamic-Key-Placeholders

Update (4/10/2015):

You can find a Shared Source Module created at the Sitecore Hackathon 2015 that incorporates this concept at https://marketplace.sitecore.net/en/Modules/Integrated_Dynamic_Placeholders.aspx by Jamie Stump and Mark Servais, do note that I had no involvement in the creation of this module.

Thank you Jamie Stump for pointing this out.