Composite pattern abstraction. Persistence. Querying hierarchical data

Question

Let's say I want to model my domain using the composite pattern. The model (gist) is fairly simple. At album controller I want to perform a search on a given name:

public ActionResult Edit(String name) {
    if(name == null) {
        ViewBag.ViewRoot = true;
        return View(this.albums.All());
    }

    var album = this.albums.Single(a => a.Name == name);
    if(album != null) {
        ViewBag.ViewRoot = false;
        return View(album.Yield());
    }

    return View("404", (Object)("Album:" + name));
}

The resulting (filtered) persisted object looks like this:

{
    "_id" : "52ab1a6a7b88c91e5cfc39ff",
    "ParentId" : null,
    "Name" : "sample01",
    "Items" : [{
        "_t" : "Album",
        "_id" : null,
        "ParentId" : "52ab1a6a7b88c91e5cfc39ff",
        "Name" : "sample02",
        "Items" : [{
            "_t" : "Album",
            "_id" : null,
            "ParentId" : null,
            "Name" : "sample03",
            "Items" : []
        }]
    }]
}

Mongodb doesn't perform recursive search out of the box. But it allows you to perform queries given that you provide an explicit direction (code sample is taken from MongoDB + C# driver + query array of elements where each array element contains sub-document to query on):

Query.ElemMatch("Children", Query.And(Query.EQ("StatusId", 1), Query.EQ("Active", true),Query.LT("SubChild.ExpiresOn", DateTime.UtcNow)));

I've been thinking about two solutions (which might keep it composite):

• tracking the node depth (then I should build complicated query);
• use parent sign (id, name), fetch entire graph, then perform recursive search at the client side.

What are your thoughts on this?

Answer 1

There are some design patterns for modelling tree structures in mongodb among them being :

• Parent References
• Child References
• Array of Ancestors
• Materialised Paths
• Nested Sets

for me it sounds like you need a tree structure with materialised paths, but you can have a look at MongoDB documentation for further information about the available design patterns used for similar issues