是否可以在不获取所有链接的情况下获取链接表?
好吧,所以我首先要说的是我正在为项目使用NHibernate,在这个项目中,我们还有一个同步功能(用于从中央MSSQL数据库同步到本地SQLite). .现在我知道NHibernate并不是用来同步数据库的,但是我还是想这样做.
Ok, so first of I would like to say that I'm using NHibernate for my project, and in this project we have (among other things) a sync function (to sync from a central MSSQL database to a local SQLite). Now I know that NHibernate was not made to sync databases, but I would like to do this anyways.
我有一个中型的大型数据库模型,所以不能在这里添加它,但是问题是我有两个数据表和一个链接表来链接这两个表.
I have a medium large database model so I can't add it here, but the problem is that I have two datatables, and one link table to link them both.
数据库模型:
| Product | | ProductLinkProducer | | Producer |
|--------------------| |---------------------| |---------------------|
| Id | | LinkId | | Id |
| Name | | Product | | Name |
| ProductLinkProducer| | Producer | | ProductLinkProducer |
数据库:
| Product | | ProductLinkProducer | | Producer |
|---------| |---------------------| |----------|
| Id | | LinkId | | Id |
| Name | | ProductId | | Name |
| | | ProducerId | | |
因此,在同步期间,我首先从Product表中复制所有数据,然后再从Producer表中复制所有数据(基本上是var products = session.Query<Products>().ToList()
).这是由NHibernate在每个语句中完成的:
So during the sync, I first copy all data from the Product table, and then from the Producer table (basically var products = session.Query<Products>().ToList()
). This is done by NHibernate in a single statement each:
select
product0_.id as id2_,
product0_.name as name2_
from
Product product0_
现在我必须从第一届会议(products.ForEach(x => session.Evict(x));
)逐出所有项目
Now I have to evict all items from the first session (products.ForEach(x => session.Evict(x));
)
然后保存(products.ForEach(x => syncSession.save(x));
)是每行插入一次(如预期的那样).
And then the save (products.ForEach(x => syncSession.save(x));
) is one insert per row (as expected).
因此,当将数据保存在链接表中时,我希望也只有一个 select .但是事实并非如此.因为首先它会像上面一样制作select ...
.但是现在,在每行插入之前,它会为产品和生产者进行更多的选择.
So when saving the data in the link table I would have wished that there also would be just a single select. However that is not the case. Because first it makes a select ...
as above. But now before every row to insert it does even more select for the Product and for the Producer.
所以它看起来像:
产品:
- 选择
- 插入(id 1)
- 插入(id 2)
制作人:
- 选择
- 插入(id 101)
- 插入(id 102)
ProdLinkProducer:
ProdLinkProducer:
- 选择
- 从产品中选择ID 1
- 从产品中选择ID 1
- 从生产者中选择id 101
- 从产品中选择ID 2
- 从产品中选择ID 2
- 从生产者中选择ID 102
- 从生产者中选择ID 102
- 插入
- 插入
- select
- select id 1 from Products
- select id 1 from Products
- select id 101 from Producer
- select id 2 from Products
- select id 2 from Products
- select id 102 from Producer
- select id 102 from Producer
- insert
- insert
那么有什么方法可以避免这种行为吗?
So is there anyway avoiding this behavior?
编辑
为了更好地解释我所做的事情,我创建了一个小测试项目.可以在这里找到: https://github.com/tb2johm/NHibernateSync (我宁愿只添加一个ghist,但我认为它可能遗漏了很多数据,对不起...)
To better explain what I have done, I have created a small test project. It can be found here: https://github.com/tb2johm/NHibernateSync (I would have preferred to add only a ghist, but I think that it might have left out to much data, sorry...)
EDIT2
我已经找到一种使它起作用的方法,但是我不喜欢它. 此解决方案的工作方式是在数据库模型中创建一个ProductLinkProducerSync表,该表不包含任何链接,而仅包含值,并且避免同步普通链接表,而仅同步"sync"表.但是正如我说的那样,我不喜欢这个主意,因为如果我更改数据库中的任何内容,那么我需要在两个地方更新相同的数据.
I have found out one way to make it work, but I don't like it. The way this solution works is to in the database model create a ProductLinkProducerSync table, that doesn't contain any links, but just the values, and avoid synchronizing the ordinary link tables, but just the "sync" tables. But as I said I don't like this idea, since if I change anything in the database, I have kind of the same data in two places that I need to update.
我无法以开箱即用的方式找到NHibernate.
I was unable to find NHibernate out of the box way of doing what you are asking.
但是,通过手动将FK引用(代理类)重新绑定到新会话上,我能够获得所需的行为(我想有些事情总比没有好:)
However I was able to get the desired behavior (I guess something is better than nothing:) by manually rebinding the FK references (proxy classes) to the new session:
var links = session.Query<ProductLinkProducer>().ToList();
links.ForEach(x => session.Evict(x));
foreach (var link in links)
{
link.Product = syncSession.Get<Product>(link.Product.Id);
link.Producer = syncSession.Get<Producer>(link.Producer.Id);
syncSession.Save(link);
}
syncSession.Flush();
以下是使用NHibernate元数据服务的通用版本:
Here is the generalized version using NHibernate metadata services:
static IEnumerable<Action<ISession, T>> GetRefBindActions<T>(ISessionFactory sessionFactory)
{
var classMeta = sessionFactory.GetClassMetadata(typeof(T));
var propertyNames = classMeta.PropertyNames;
var propertyTypes = classMeta.PropertyTypes;
for (int i = 0; i < propertyTypes.Length; i++)
{
var propertyType = propertyTypes[i];
if (propertyType.IsAssociationType && !propertyType.IsCollectionType)
{
var propertyName = propertyNames[i];
var propertyClass = propertyType.ReturnedClass;
var propertyClassMeta = sessionFactory.GetClassMetadata(propertyClass);
yield return (session, target) =>
{
var oldValue = classMeta.GetPropertyValue(target, propertyName, EntityMode.Poco);
var id = propertyClassMeta.GetIdentifier(oldValue, EntityMode.Poco);
var newValue = session.Get(propertyClass, id);
classMeta.SetPropertyValue(target, propertyName, newValue, EntityMode.Poco);
};
}
}
}
并将其应用于您的Sync
方法:
and applying it to your Sync
method:
private static void Sync<T>(string tableName, ISession session, ISession syncSession)
{
Console.WriteLine("Fetching data for ####{0}####...", tableName);
var sqlLinks = session.Query<T>();
var links = sqlLinks.ToList();
Console.WriteLine("...Done");
Console.WriteLine("Evicting data...");
links.ForEach(x => session.Evict(x));
Console.WriteLine("...Done");
Console.WriteLine("Saving data...");
var bindRefs = GetRefBindActions<T>(syncSession.SessionFactory).ToList();
foreach (var link in links)
{
foreach (var action in bindRefs) action(syncSession, link);
syncSession.Save(link);
}
Console.WriteLine("...Flushing data...");
syncSession.Flush();
Console.WriteLine("...Done");
Console.WriteLine("\n\n\n");
}