与存储库相关的C＃设计模式

Question

I have a repository class say call FileRepository() working in the C# managed layer. There is a method call LoadRecordFromFile() which access to the C++ unmanaged COM DLL to load a large amount of records, as shown :

LoadRecordFromFile()
{
     for (uint index = 1; index <= this.File.Count; index++)
     {
      // create new record object here
     }
}

The information in the record object will later used to generate AnalysisViewModel object. Thus, There is another for loop used to generate a collection of AnalysisViewModel objects before it can be used in the View for display.

The problem is that the second for loop is using vast amount of loading time. I am trying to optimise the loading time by replacing two for loop by using one for loop in the LoadRecordFromFile(). ie create both record object and AnalysisViewModel() object together in the for loop in LoadRecordFromFile() in FileRepository class. However It breaks the encapsulation of repository design pattern. Please would someone propose how to redesign the two classes (AnalysisVeiwModel and FileRepository)to achieve using only one for loop to reduce the file loading time ? Thank you

Answer 1

I would store a collection of the records in your view model instead of just one.

So you could have a AnalysisRecordsViewModel which gets all records from the repository at once.

If you want to display or edit a single records in your view, your might use a second view model like AnalysisViewModel to display a single record.

Answer 2

Create a custom enumerator wrapping the API calls to the COM dll.

Start by declaring your collection. The collection is just a wrapper to create a enumerator:

public class MyCustomCollection : IEnumerable<YouRecordType>
{
    private readonly string _fileName;

    public MyCustomCollection(string fileName)
    {
        if (fileName == null) throw new ArgumentNullException(nameof(fileName));
        _fileName = fileName;
    }

    public IEnumerator<YouRecordType> GetEnumerator()
    {
        return new MyCustomEnumerator(_fileName);
    }

    IEnumerator IEnumerable.GetEnumerator()
    {
        return GetEnumerator();
    }
}

.. the collection should be returned by your repository.

The enumerator itself is the class that uses the COM object:

public class MyCustomEnumerator : IEnumerator<YouRecordType>
{
    private readonly string _fileName;
    private int _index;
    YouRecordType _current;

    public MyCustomEnumerator(string fileName)
    {
        _fileName = fileName;
        //open COM object here
    }

    public void Dispose()
    {
        //Dispose used COM resources.
    }

    public bool MoveNext()
    {
        //Read next row from the COM object
        var couldReadAnotherRow = true;
        _current = null; //comObject.Read(_index++); 
        return couldReadAnotherRow;
    }

    public void Reset()
    {
        //want to start over
    }

    public YouRecordType Current { get { return _current; } }

    object IEnumerator.Current
    {
        get { return Current; }
    }
}

You can see it as a lazy loaded collection. ie no rows are loaded until they are being called.

Thus returning that object from your repository should take no time at all. And you do not have to preload everything from start.

The enumerable works just like any other collection. You can use yourCollection.ToList() to load all rows or foreach(var yourType in yourCollection) to load a new row in every iteration.