i have a Product table with two many-to-one references (Title & Description) to a single table named TextRef :
Product:
<many-to-one lazy="false" name="Title" class="TextRef" column="TitleRef" fetch="join" cascade="save-update"/>
<many-to-one lazy="false" name="Description" class="TextRef" column="DescriptionRef" fetch="join" cascade="save-update"/>
every TextRef has a one-to-many to TextRefItem table :
<bag lazy="false" name="Values" table="TextRefItem" cascade="save-update" inverse="true" fetch="join">
<key column="TextId"></key>
<one-to-many class="TextRefItem"/>
</bag>
now i want to load all of TextRefItem(s) for Title & Description in one go but NHibernate only create a join with the first reference (Title)
SELECT this_.ProductId as ProductId7_2_, this_.CategoryId as CategoryId7_2_,
this_.DescriptionRef as Descript3_7_2_,
textref2_.TextId as TextId8_0_, values3_.TextId as TextId4_,
values3_.TextItemId as TextItemId4_, values3_.TextItemId as TextItemId9_1_,values3_.LangId as LangId9_1_,
values3_.Text as Text9_1_, values3_.TextId as TextId9_1_
FROM
Product this_
inner join TextRef textref2_ on this_.DescriptionRef=textref2_.TextId
left outer join TextRefItem values3_ on textref2_.TextId=values3_.TextId
WHERE this_.ProductId = 1
for the other one(Description) it makes a separate select query
how could i tell NHibernate to avoid that ?
I don't know why a separate query is made for the second reference (Description). But it is possible to extend fetching strategies via criterias to load a product with all TextRefs and TextRefItems in one shot:
var criteria = session.CreateCriteria(typeof(Product))
.SetFetchMode("Description.Values", FetchMode.Join);
criteria.SetResultTransformer(new DistinctRootEntityResultTransformer());
criteria.Add(Restrictions.Eq("ProdId", 1));
var list = criteria.List<Product>();
I have following class
<class name="Product" table="Product">
<id name="ID" />
...
<map name="CustomFields" table="CustomFieldView">
<key column="RECORDID" />
<map-key column="CFName" type="String" />
<element column="CFValue" type="String" />
</map>
</class>
and SP to select product with CustomFields dictionary
<sql-query name="GetProducts">
<return alias="p" class="Product" />
<return-join alias="cf" property="p.CustomFields" />
SELECT {p.*}, {cf.*}
FROM Product p
INNER JOIN CustomFieldView cf ON p.ID = cf.RECORDID
// WHERE
</sql-query>
when I select single product like WHERE ID = 1234, then it works as expected - returns one Product with populated CustomFields Dictionary property.
But when I select not single Product like WHERE ID IN (18780, 21642) or other criterias then I get Products duplicated 'CustomFields.Count' times, e.g. 2 Products and each has 20 Custom Fields, then 40 Products and each has 20 valid custom fields.
Do I missed something in mapping ?
You are returning a Cartesian product and therefore your product is being returned x times for every custom field.
To get around this problem you will need to use something like:-
var query = Session
.GetNamedQuery("GetProducts")
.SetResultTransformer(new DistinctRootEntityResultTransformer());
return query.List<Product>();
Please note that you will send all the data down the wire and NHibernate will perform the distinct transformer client (meaning web server or desktop app) side.
I am not 100% sure if the return join will be populated as I have never done things this way, you will need to test this.
edit
I think you fetching strategy is not quite right. Do you really need a <sql-query...> Could you use another strategy e.g. HQL?
In nhibernate, I have two classes that are associated with a many-to-one mapping:
<class name="Employee" table="Employee">
..
<bag name="orgUnits">
<key column="id" />
<one-to-many name="OrgUnit" class="OrgUnit">
</bag>
..
</class>
I would like to use a criteria expression to get only Employees where the the collection is null (ie no orgunits) , something like this :
IList employeesWithNoOrgUnit = sess.CreateCriteria(typeof(Employee))
.Add( Expression.IsNull("OrgUnits") )
.List();
This doesn't filter the collection as I expect.
Colleague just found one way that works.
IList employeesWithNoOrgUnit = sess.CreateCriteria(typeof(Employee))
.Add( Restrictions.IsEmpty("OrgUnits") )
.List();
I'd like to map the following sql in NHibernate.
Will I need to make a separate entity object i.e RoomTypeVO mapped to tb_tags to do this?
Any help much appreciated.
SELECT
dbo.tb_rooms.id,
dbo.tb_rooms.name,
dbo.tb_tags.name AS 'roomType'
FROM
dbo.tb_rooms
INNER JOIN dbo.tb_tags ON (dbo.tb_rooms.typeID = dbo.tb_tags.id)
<id name="id" column="id">
<generator class="native" />
</id>
<property name="name" />
If you to a straight sql query you do not have to. If you want to use HQL you will have to work with an entity.
But, you can always do sql queries directly.
If you have a mapped entity then you could probably just do something like this:
FROM RoomType
When you refer to 'FROM', are you thinking of something like this?
<property name="totalPrice"
formula="( SELECT SUM (li.quantity*p.price) FROM LineItem li, Product p
WHERE li.productId = p.productId
AND li.customerId = customerId
AND li.orderNumber = orderNumber )"/>
I have a 3-leveled hierarchy of entities: Customer-Order-Line, which I would like to retrieve in entirety for a given customer, using ISession.Get(id). I have the following XML fragments:
customer.hbm.xml:
<bag name="Orders" cascade="all-delete-orphan" inverse="false" fetch="join">
<key column="CustomerID" />
<one-to-many class="Order" />
</bag>
order.hbm.xml:
<bag name="Lines" cascade="all-delete-orphan" inverse="false" fetch="join">
<key column="OrderID" />
<one-to-many class="Line" />
</bag>
I have used the fetch="join" attribute to indicate that I want to fetch the child entities for each parent, and this has constructed the correct SQL:
SELECT
customer0_.ID AS ID8_2_,
customer0_.Name AS Name8_2_,
orders1_.CustomerID AS CustomerID__4_,
orders1_.ID AS ID4_,
orders1_.ID AS ID9_0_,
orders1_.PostalAddress AS PostalAd2_9_0_,
orders1_.OrderDate AS OrderDate9_0_,
lines2_.OrderID AS OrderID__5_,
lines2_.ID AS ID5_,
lines2_.ID AS ID10_1_,
lines2_.[LineNo] AS column2_10_1_,
lines2_.Quantity AS Quantity10_1_,
lines2_.ProductID AS ProductID10_1_
FROM Customer customer0_
LEFT JOIN [Order] orders1_
ON customer0_.ID=orders1_.CustomerID
LEFT JOIN Line lines2_
ON orders1_.ID=lines2_.OrderID
WHERE customer0_.ID=1
So far, this looks good - SQL returns the correct set of records (with only one distinct orderid), but when I run a test to confirm the correct number of entities (from NH) for Orders and Lines, I get the wrong results
I should be getting (from my test data), 1xOrder and 4xLine, however, I am getting 4xOrder and 4xLine. It appears that NH is not recognising the 'repeating' group of Order information in the result set, nor correctly 'reusing' the Order entity.
I am using all integer IDs (PKs), and I've tried implementing IComparable of T and IEquatable of T using this ID, in the hope that NH will see the equality of these entities. I've also tried overridding Equals and GetHashCode to use the ID. Neither of these 'attempts' have succeeded.
Is "multiple leveled fetch" a supported operation for NH, and if so, is there an XML setting required (or some other mechanism) to support it?
NB: I used sirocco's solution with a few changes to my own code to finally solve this one. the xml needs to be changed from bag to set, for all collections, and the entitities themselves were changed to implement IComparable<>, which is a requirement of a set for uniqueness to be established.
public class BaseEntity : IComparable<BaseEntity>
{
...
private Guid _internalID { get; set; }
public virtual Guid ID { get; set; }
public BaseEntity()
{
_internalID = Guid.NewGuid();
}
#region IComparable<BaseEntity> Members
public int CompareTo( BaseEntity other )
{
if ( ID == Guid.Empty || other.ID == Guid.Empty )
return _internalID.CompareTo( other._internalID );
return ID.CompareTo( other.ID );
}
#endregion
...
}
Note the use of an InternalID field. This is required for new (transient) entities, other wise they won't have an ID initially (my model has them supplied when saved).
You're getting 4XOrder and 4XLines because the join with lines doubles the results . You can set a Transformer on the ICriteria like :
.SetResultTransformer(new DistinctRootEntityResultTransformer())
I just read Ayende's Blogpost where he used the following Example:
session.CreateCriteria(typeof(Post))
.SetFetchMode("Comments", FetchMode.Eager)
.List();
In a Criteria Query to avoid Lazy Loading on one particular Query
Maybe that can help you.
If you need to keep your one-to-manys as bags, then you can issue 2 queries, each with only 1 level of hierarchy. eg something like this:
var temp = session.CreateCriteria( typeof( Order ) )
.SetFetchMode( "Lines", NHibernate.FetchMode.Eager )
.Add( Expression.Eq( "Customer.ID", id ) )
.List();
var customer = session.CreateCriteria( typeof( Customer ) )
.SetFetchMode( "Orders", NHibernate.FetchMode.Eager )
.Add( Expression.Eq( "ID", id ) )
.UniqueResult();
Lines get loaded into the NH cache in the first query, so they won't need lazy loading when later accessing eg customer.Orders[0].Lines[0].
#Tigraine: your query only returns Post with Comments. This brings All posts with all Comments (2 levels). What Ben asking is Customer to Order To LineItem (3 level).
#Ben: to my knowledge nHibernate doesn't support eager loading upto 3 level yet. Hibernate does support it thou.
I was having the same problem. See this thread. I didn't get a solution but a hint from Fabio. Use Set instead of bag. And it worked.
So my suggestion is try to use set. You don't have to use Iesi collection use IDictonary and NH is happy
public override IEnumerable<Baseline> GetAll()
{
var baselines = Session.CreateQuery(#" from Baseline b
left join fetch b.BaselineMilestones bm
left join fetch bm.BaselineMilestonePrevious ")
.SetResultTransformer(Transformers.DistinctRootEntity)
.List<Baseline>();
return baselines;
}