How to Read Project Properties that are not Available in the Client Object Model?

Recently I had a development task that at the first sight seemed to be trivial, but it turned out quickly to be rather a challenge. I had to display some basic project information on a page in our Project Web Site (PWS), like project start and finish date, remaining work and percent complete. The web page was built using client-side technologies, like the Client-side object model (CSOM) for Project 2013 and using the AngularJS library, and we did not plan to change the architecture to server side code.

If you check the properties of the PublishedProject (either on the client side in namespace / assembly Microsoft.ProjectServer.Client or on the server side in Microsoft.ProjectServer), you see that it has properties like StartDate and FinishDate, and it inherits its PercentComplete property from the Project base class, however there is no property for RemainingWork or PercentWorkComplete defined, although both of these values are available as fields if you manage a Project Server view (see screenshot below). This information is not available via REST / OData either.

You should know, that in the case of  Project Server, the server side OM is simply a wrapper around the PSI, for example, the PercentComplete property in the Project class is defined:

public int PercentComplete
{
  get
  {
    ProjectDataSet.TaskRow summaryTaskRow = this.SummaryTaskRow;
    if (summaryTaskRow != null && !summaryTaskRow.IsTASK_PCT_COMPNull())
      return summaryTaskRow.TASK_PCT_COMP;
    else
      return 0;
  }
}

Client side OMs (either managed or ECMAScript) and REST calls invoke the server side OM, so at the background the good old PSI is still in action.

It seems that the developers of Project Server remained simply not enough time to map all of the fields available via PSI to the object models on the server side and the client side.

You should know either, that the project properties we need are stored as task properties for the project summary task of the current project. In the Project Server database the tasks of the published projects (so the project summary tasks as well) are stored in the [pub].[MSP_TASKS] table. If you run the following query (where ProjectWebApp is the name of the database and the Guid in the [PROJ_UID] filter is the ID of your project), you find some specific field values that may help to identify the summary task record of a project:

SELECT [TASK_UID]    
      ,[TASK_PARENT_UID]
      ,[TASK_ID]
      ,[TASK_OUTLINE_NUM]
      ,[TASK_OUTLINE_LEVEL]
      ,[TASK_NAME]
      ,[TASK_START_DATE]
      ,[TASK_FINISH_DATE]
      ,[TASK_PCT_COMP]
      ,[TASK_PCT_WORK_COMP]
      ,[TASK_REM_WORK]
  FROM [ProjectWebApp].[pub].[MSP_TASKS]
  WHERE [PROJ_UID] = ‘d0ae5086-be7a-e411-9568-005056b45654′

The project summary task record – at least, based on my experimental results – , matches the following conditions:

[TASK_ID] = 0

[TASK_OUTLINE_NUM] = 0

[TASK_OUTLINE_LEVEL] = 0

[TASK_UID] = [TASK_PARENT_UID]

But as said, we need a solution on the client side, and obviously one that does not tamper with the Project Server database. What options are there to achieve the missing information?

The Project class has a property called SummaryTaskId, but if you have this value already, and would like to query the project tasks via REST (for example: http://YourProjServer/PWA/_api/ProjectServer/Projects(‘d0ae5086-be7a-e411-9568-005056b45654′)/Tasks(‘FFAE5086-BE7A-E411-9568-005056B45654′)) or via the client object model, the result is empty. The description of the SummaryTaskId property says: “Gets the GUID for the hidden project summary task”. Yes, it is so hidden, that it simply not included in the Tasks collection of the Project class! The Tasks property of the PublishedProject class is of type PublishedTaskCollection, and on the server side the record for the project summary task is simply filtered out, when initializing the internal Dictionary used for the storage of the Task records. If you don’t believe me, or need more details on that, see the constructor method of Microsoft.ProjectServer.PublishedTaskCollection class below:

internal PublishedTaskCollection()
{
    Func<Dictionary<Guid, PublishedTask>> valueFactory = null;
    if (valueFactory == null)
    {
        valueFactory = () => base.ProjectData.Task.OfType<ProjectDataSet.TaskRow>().Where<ProjectDataSet.TaskRow>(delegate (ProjectDataSet.TaskRow r) {
            if (!r.IsTASK_PARENT_UIDNull())
            {
                return (r.TASK_PARENT_UID != r.TASK_UID);
            }
            return true;
        }).ToDictionary<ProjectDataSet.TaskRow, Guid, PublishedTask>(r => r.TASK_UID, r => this.CreateTask(r));
    }
    this._tasks = new Lazy<Dictionary<Guid, PublishedTask>>(valueFactory);
}

Of course, we get the same, empty result if we would like to filter the tasks for one the special conditions we found in the database (like [TASK_OUTLINE_LEVEL] = 0):
http://YourProjServer/PWA/_api/ProjectServer/Projects(‘d0ae5086-be7a-e411-9568-005056b45654&#8242;)/Tasks?$filter=OutlineLevel eq 0 

The project reporting data contains the project summary tasks as well, so we could invoke the ProjectData OData endpoint from the client side to query the required information. The problem with this approach is that it would require extra permissions on the reporting data and one cannot limit this permission to the summary tasks of a specific project, to summary tasks, or just to tasks at all. If you grant your users the Access Project Server Reporting Service global permission, they can query all of the reporting data. It is sure not our goal, but you can test it if you wish.

Once you have the ID of the project summary task (for example via the SummaryTaskId property), the task is available via a query like this one:

http://YourProjServer/PWA/_api/ProjectData/Tasks(ProjektID=guid’d0ae5086-be7a-e411-9568-005056b45654&#8242;,TaskID=guid’FFAE5086-BE7A-E411-9568-005056B45654′)

When using PSI, we can access the required information via the TASK_REM_WORK and TASK_PCT_WORK_COMP fields in ProjectDataSet.TaskRow, that means, rows in the Task property (type of  ProjectDataSet.TaskDataTable) of the ProjectDataSet. The first row in the record set contains the information about the project summary task.

We could create our own extensions for the client object model (wrapping around just this piece of  PSI), as I illustrated for the managed, and for the ECMAScript object model as well, but it would require a lot of work, so I ignored this option for now. Instead of this, I’ve created a simple .NET console application utilizing the PSI (see the most important part of the code below). Unfortunately, I have not found a method that returns only a specific task of a specific project, so I had to call the ReadProjectEntities method to read all of the tasks of the project.

I’ve captured the request and the response using Fiddler:

Then extended my JavaScript code with methods that assemble the request in the same format, submit it to the server, then parse the required fields out of the response.

First, I needed a helper method to format strings:

In my Angular controller I defined this function to format dates:

Next, in the controller we get the ID of the project for the current PWS, and we read project properties that are available via the client object model, and finally the ones, that are available only via PSI:

As you can see, the value we receive in the remainingWork property should be divided by 600 and 100 to get the value in hours.

In our custom ProjService service I’ve implemented the corresponding methods.

The project ID is stored in the property bag of the PWS in a property called MSPWAPROJUID (see this post about how to read property bags from the client object model):

Having the project ID, reading project properties via the client object model should be straightforward as well:

Reading the ‘extra’ properties via PSI is a bit more complex. First, we assemble the request XML as we captured it with Fiddler when used the console application mentioned above, and post it to the server. Next, we process the response (see the code of the helper method buildXMLFromString farther below), and parse out the necessary properties from the project summary task (that is the Task node having rowOrder = 0) using XPath queries.

These are the helper methods I used for processing the response text as XML:

Having an HTML template like this one:

the result should be displayed similar to this one:

A drawback of this approach (not to mention the fact that it is pretty hacky) is, that due the ReadProjectEntities method, all of the fields of all of the project tasks should be downloaded to the client, although we need only a few fields of a single task, the project summary task. So it would make sense to implement some kind of  caching on the client side, but it is out of the scope of this post. But as long as Microsoft does not provide all the project fields in the client object model, I have not found any better solution that would require a relative small effort.