Entity Framework - Update entities without Id

Question

I have found many questions here on SO and articles all over the internet but none really tackled my problem.

My model looks like this (I striped all non essential Properties):

Everyday or so "Play" gets updated (via a XML-file containing the information).

internal Play ParsePlayInfo(XDocument doc)
{
    Play play = (from p in doc.Descendants("Play")
    select new Play
    {
        Theatre = new Theatre()
        {
            //Properties
        },
        //Properties
        LastUpdate = DateTime.Now
    }).SingleOrDefault();

    var actors = (from a in doc.XPathSelectElement(".//Play//Actors").Nodes()
    select new Lecturer()
    {
        //Properties
    });

    var parts = (from p in doc.XPathSelectElement(".//Play//Parts").Nodes()
    select new Part()
    {
        //Properties
    }).ToList();

    foreach (var item in parts)
    {
        play.Parts.Add(item);
    }

    var reviews = (from r in doc.XPathSelectElement(".//Play//Reviews").Nodes()
    select new Review
    {
        //Properties
    }).ToList();

    for (int i = 0; i < reviews.Count(); i++)
    {
        PlayReviews pR = new PlayReviews()
        {
            Review = reviews[i],
            Play = play,
            //Properties
        };
        play.PlayReviews.Add(pR);
    }
    return play;
}

If I add this "play" via Add() every Childobject of Play will be inserted - regardless if some exist already. Since I need to update existing entries I have to do something about that. I guess the problem here is that I have no Id in the XML whatsoever.

As far as I can tell I have the following options:

1. add/update the child entities in my PlayRepositories Add-Method
2. restructure and rewrite ParsePlayInfo() so that get all the child entities first, add or update them and then create a new Play. The only problem I have here is that I wanted ParsePlayInfo() to be persistence ignorant, I could work around this by
3. creating multiple parse methods (eg ParseActors() ) and assign them to play in my controller (I'm using ASP.net MVC) after everything was parsed and added

Currently I am implementing option 1 - but it feels wrong.

What I want to do is update entities that are already in the database and insert new ones that aren't. Do I have to call SaveChanges() before I Attach/Add Play? There must be a (relatively) easy solution. I'd appreciate it if someone could guide me in the right direction on this one.

Answer 1

Well, since there is no answer yet I'll write one myself.

For those who are wondering, I got it to work - the code looks ugly as hell and I guess performance is even worse. But since there wont be many users and this method will only be called once a day at night anyway I am fine with it for now.

What did I do?

Well, I went with option 2 and 3.

private Play UpdatePlay()
{
    using (RepositoryContext context = new RepositoryContext())
    {
        HttpRequest http = new HttpRequest();
        PlayRepository rep = new PlayRepository(context);

        ActorRepository actRep = new ActorRepository(context);
    ReviewsRepository revRep = new ReviewsRepository(context);
        TheatreRepository insRep = new TheatreRepository(context);
        PartRepository partRep = new PartRepository(context);

        Parser p = new Parser();

        XDocument doc = http.GetPlayInfo();

        Theatre theatre = p.ParseTheatreInfo(doc);
        List<Actor> actors = p.ParseActorInfo(doc);
        List<PlayReviews> playReviews = p.ParseReviewsInfo(doc);

        for (int i = 0; i < actors.Count; i++)
        {
            actors[i] = actRep.AddOrUpdate(actors[i]);
        }
        for (int i = 0; i < playReviews.Count; i++)
        {
            playReviews[i].Reviews = revRep.AddOrUpdate(playReviews[i].Reviews);
        }

        theatre = insRep.AddOrUpdate(theatre);

        Play play = p.ParsePlayInfo(doc);

        List<Part> parts = GetParts(play);

        for (int i = 0; i < parts.Count; i++)
        {
            List<Actor> lec = (List<Actor>)parts[i].Actors;
            for (int j = 0; j < lec.Count; j++)
            {
                lec[j] = actRep.AddOrUpdate(lec[j]);
            }

        }

        play = rep.AddOrUpdate(play);
        context.LoadProperty(play, o => o.Theatre);
        context.LoadProperty(play, o => o.Actors);
        context.LoadProperty(play, o => o.PlayReviewss);
        context.LoadProperty(play, o => o.Parts);

        rep.Save();

        if (play.Theatre != theatre)
            play.Theatre = theatre;

        play = rep.AddParts(parts, play);

        play = rep.AddActor(actors, play);

        for (int i = 0; i < playReviews.Count; i++)
        {
            playReviews[i].Play = play;
            playReviews[i] = revRep.AddPlayInformation(playReviews[i]);
        }

        rep.Save();
        return play;
    }
}

(And on a side note, I just realized that I previously forgot to post this part of my code...)

As you can see, Save() gets called twice - and it gets worse when you consider whats going on in AddOrUpdate():

public Actor AddOrUpdate(Actor entity)
{
    Actor cur = context.Actors.Where(l => l.Name == entity.Name && l.Last_Name == entity.Last_Name).FirstOrDefault();
    if (cur == null)
    {
        context.Actors.AddObject(entity);
        return entity;
    }
    else
    {
        if (!entity.Mail.IsNullOrEmptyOrWhitespace() && cur.Mail != entity.Mail)
            cur.Mail = entity.Mail;
    //there are more of these...

        return cur;
    }
}

I can't believe that this is "the right" way to do this. It feels and looks just wrong. Maybe EF is to blame too, take

FirstEntityType first = new FirstEntityType();
first.Id = 2;
List<SecondType> list = CreateList(); //Let's say this returns a List with 10 elements
context.FirstEntityType.AddObject(first);

for (int i = 0; i < list.Count; i++)
{
    list[i].First = first;
}

//just for arguments sake a second for
for (int i = 0; i < list.Count; i++)
{
    context.SecondType.AddObject(list);
}

context.SaveChanges();

I haven't tested this particular piece of code but from what I experienced is that I'll end up with 10 new entries for SecondType and if I am not wrong 11 for FirstEntityType.

Why? Why isn't there a mechanism in EF that says "Hey, wait a minute - those are the same!"

Am I that wrong thinking that EF should behave like if I was using the db directly? In my example, I added "first" so I could assume that whenever I use "first" it is referenced. (I really hope my example works as described - don't have the time nor the desire to test it)