I'm using Hibernate 4.3.1 final, Mysql 5.5 and I want to use an "order by case" order logic on some joined entities.
A pure sql representation of what I wish to achieve would look something like:
select adv.id, adv.published_date
from advert as adv
join account as act on act.id = adv.act_id
join account_status as aas on aas.id = act.current_aas_id
order by case aas.status
when 'pending' THEN 1
when 'approved' THEN 1
else 2
end, adv.published_date;
This orders the adverts of pending and approved accounts before those of deactive accounts.
I've managed to do all the select query using hibernate criteria, but I'm not sure how to specify the order by case statement using that api.
I found this post:
http://blog.tremend.ro/2008/06/10/how-to-order-by-a-custom-sql-formulaexpression-when-using-hibernate-criteria-api/
but I need to reference joined entity classes in the order by case and I'm not sure how to do that.
Any help or suggestions much appreciated.
I think I found a solution.
In the end I created my own subclass of Order and overrode the public String toSqlString(Criteria,CriteriaQuery) method:
#Override
public String toSqlString(Criteria criteria, CriteriaQuery criteriaQuery) {
final String[] columns = criteriaQuery.getColumnsUsingProjection( criteria, getPropertyName() );
final StringBuilder fragment = new StringBuilder();
fragment.append(" case ").append(columns[0]);
fragment.append(" when 'pending' then 1 ");
fragment.append(" when 'approved' then 1 ");
fragment.append(" else 2 end");
return fragment.toString();
}
the important call (which I found in the original implementation of the order class) is the
criteriaQuery.getColumnsUsingProjection(criteria, getPropertyName());
Which gave me access to the alias for the column I wanted to order on using the case statement.
If anyone else is looking at this, if you are ordering on a property that is not on the root object, then make sure that you use aliases in your criteria joins and that you then reference those aliases correctly in the your custom Order propertyName when you instantiate it.
In addition to https://stackoverflow.com/a/24077763/258559
It is probably worth mention that now you're adding Order subclass like this
criteria.addOrder(new CustomOrder());
and not
criteria.addOrder(CustomOrder.desc(fieldName));
because desc/asc are the static methods which would still create an instance of Order instead of your CustomOrder
Related
I want to write a simple query to retrieve a list of USER with a simple restriction on CUSTOMER joined table.
I'm only interested by the USER entity.
If I write it using HPQL :
public List<Users> getAssociatedAdminObs(Integer pCustId) {
Criteria crit = getCriteriaForObsAdmin("USER");
crit.createCriteria("clients").add(Restrictions.eq("idCustomer", pCustId));
return crit.list();
StringBuilder hqlQuery = new StringBuilder().append("select u from Users as u join u.customers as c where c.idCustomer=:idCustomer");
Query q = getSessionAndManageFilter().createQuery(hqlQuery.toString());
q.setInteger("idCustomer", pCustId);
return q.list();
}
The SQL generated only Select all the fields from USER entity, as expected.
Now if I write it through Hibernate criteria API :
public List<Users> getAssociatedAdminObs(Integer pCustId) {
Criteria crit = getSession().createCriteria(Users.class);
crit.setResultTransformer(Criteria.DISTINCT_ROOT_ENTITY);
crit.createCriteria("customers").add(Restrictions.eq("idCustomer", pCustId));
return crit.list();
}
The SQL generated Select all the fields from USER entity but also from the CUSTOMER entity.
I'm using hibernate-core 3.3.1.GA.
I know I can use projection to work around the issue but my object will be transient, I also validated it works using a subquery to do my restrictions.
But I'm not happy with using workarounds and I do not understand why it would behave differently between the 2 code examples ?
It doesn't make sense to me to select fields outside of the asked entity.
I am pretty new to hibernate again, so this might be a noobish question ;).
Without to_days, but clustered by timestamp it works like this:
CriteriaQuery<Tuple> query = criteriaBuilder.createQuery(Tuple.class);
Root<Session> sessionRoot = query.from(Session.class);
query.multiselect(
sessionRoot.get("time").alias("time"),
criteriaBuilder.count(sessionRoot).alias("count")
);
query.groupBy(sessionRoot.get("time"));
List<Tuple> results = this.executeQuery(query);
So I recieve:
time|count
13721938721|1
13721938722|2
13721938723|3
13721938724|4
13721938725|2
13721938726|1
13721938727|4
But this are all sessioncounts for each millisecond, but I need those clustered by day and not by timestamp: thus I use to_days in plain mysql.
In mysql I perform this query:
SELECT TO_DAYS(`time`) AS `days`, COUNT(*) as `count` FROM sessions WHERE 1 GROUP BY `days`
This gives me:
days|count
777594|123
777595|60
777596|61
777597|74
But I have no idea, yet: how to achieve the same thing with javax.persistence.criteria.CriteriaBuilder and CriteriaQuery in hibernate?
I dont know how to do it with criteriaBuilder, but i do know how in Hibernate 4 criteria api:
query.setProjection(
Projections.sqlProjection(
"TO_DAYS(time) as days",
new String[]{"days"},
new Type[]{StandardBasicTypes.INTEGER}
)
);
sqlProjection allows you to cast or convert data types, but careful, using a projection will only retrieve the fileds you specify in it, and the resulting list will come up like this:
List<Object[]> results = this.executeQuery(query);
But you can make hibernate do a alias match with the properties using a result transformer:
query.setResultTransformer(new AliasToBeanResultTransformer(Session.class));
and the list comes out like it normally does:
List<Session.class> results = this.executeQuery(query);
Sorry i could not provide a criteriaBuilder solution, but i hope this gets you in the right track.
After some investigation, it turned out, that HQL does not support TO_DAYS. Since I want to make it possible for MySQL and other databases, this is my final solution:
Query q = entityManager.createQuery("SELECT concat(day(e.time), '-', month(e.time), '-', year(e.time)) AS days, COUNT(*) FROM Event e GROUP BY concat(day(e.time), '-', month(e.time), '-', year(e.time))");
The result is:
3-5-2012|980
4-5-2012|200
10-6-2012|123
12-6-2012|144
13-11-2012|500
So afterwards I convert all ugly date strings into proper milliseconds in java and have the data, which I need.
i've got an sql statement that works pretty well. but on implementing in my webapp working with play 2.1 i get this error:
javax.persistence.PersistenceException: Query threw SQLException:Column Index out of range, 0 < 1.
i found this question here: Error executing MySQL query via ebean using RawSql
but then i got other exceptions.
i'm trying to get tagged threads that contains a list of tags (same as stack overflow does).
here the sql statement
SELECT t.topic
FROM topic t
WHERE 3 = (SELECT COUNT( DISTINCT ta.id )
FROM topic_tag tt
INNER JOIN tag ta ON ta.id = tt.tag_id
WHERE ta.name IN ('children', 'spain','new')
AND tt.topic_id = t.id )
in play i do this:
RawSql rawSql = RawSqlBuilder.unparsed(sqlString).create();
result = find.setRawSql(rawSql).findList();
then, i got the out of bounds exception. after that i try to set column mappings:
RawSql rawSql = RawSqlBuilder.unparsed(sqlString)
.columnMapping("t.topic","topic")
.columnMapping("t.id","id")
.columnMapping("ta.name","tagList.name")
.columnMapping("ta.id","tagList.id")
.create();
now i get a null pointer exception. probably because ebean can't create a query from that.
here some code from my models:
#Entity
public class Topic extends Model{
#Id
public Long id;
#Required
public String topic;
#ManyToMany
public List<Tag> tagList;
}
#Entity
public class Tag extends Model {
#Id
public long id;
#Required
public String name;
}
after a lot of trying and frustrating i hope that somebody got a hint or a solution for this.
I just wasted few hours with similar problem, I actually managed to solve it by only mapping id field for certain kind of model and selecting lesser amount of fields, other values were automatically loaded after that - So basically, error occurred if I tried to select values like:
.. select e.id, e.name, e.description from exampleTable e .. and use mappings like:
RawSql rawSql = RawSqlBuilder.parse(sql)
// map the sql result columns to bean properties
.columnMapping("e.id", "exampleModel.id")
.columnMapping("e.name", "exampleModel.name")
.columnMapping("e.description", "exampleModel.description")
.create();
When I changed to select only e.id and map:
RawSql rawSql = RawSqlBuilder.parse(sql)
// map the sql result columns to bean properties
.columnMapping("e.id", "exampleModel.id")
.create();
It loaded also e.name and e.description to model values and errors disappeared.
(Of course my own query had several joins and were bit more complicated than this, but basics are the same.)
So to summarize: when this problem occurs, check that you are not loading anything twice (columnMapping), use System.out.println(""); or similar to check which values are already loaded for your model. Remember to also check annotations such as "#JoinColumn" which might load more data under same model - just based on given e.id value. If you dont select and set e.id as columnMapping value, then you might need to list all needed fields separately as .. e.name, e.description ..
Hopefully these findings helps someone out.
I am new to linq to sql
I wrote this function:
public ICollection<ICustomer> GetAll()
{
DataClasses1DataContext context = new DataClasses1DataContext();
var customers = from customer in context.Customers select customer;
return customers.ToList().Cast<ICustomer>().ToList();
}
But it always return list of null values.
The database contain 3 records "filled with data" but this function return 3 nulls.
how to fix that?
It may not be able to cast the results properly, have you made your partial Customer object implement ICustomer? If not, that is the reason.
Also you don't have to bring it to a list twice, or even once for that matter since you aren't returning a list, it might be more appropriate to change your signature to List or IEnumerable depending on your usage.
You can test whether or not the cast is succeeding by doing a simple test.
DataClasses1DataContext context = new DataClasses1DataContext();
var customers = from customer in context.Customers select customer;
int numberOfCustomers = customers.Count();
var myCustomers = customers.Cast<ICustomer>(); //you could also do .OfType<ICustomer>();
int numberOfICustomers = myCustomers.Count();
If numberOfCustomers is 3 and numberOfICustomers is 0 then you know that was the issue.
Your problem is almost certainly at the .Cast() method (confirm this by stepping through your code & ensuring that customers is populated correctly).
Does the Customer object implement the ICustomer interface? It sounds like an obvious thing to check but that would be a likely problem.
I am hoping you can help. I am developing a tiered website using Linq to Sql. I created a new class(or object) in DBML designer called memberState. This object is not an actual table in the database. I have this method in my middle layer:
public override IEnumerable(memberState) GetMembersByState(string #state)
{
using (BulletinWizardDataContext context = DataContext)
{
IEnumerable(memberState) mems = (from m in context.Members
join ma in context.MemberAddresses
on m.UserId equals ma.UserId
join s in context.States
on ma.StateId equals s.StateId
where s.StateName == #state
select new memberState
{
userId = m.UserID,
firstName = m.FirstName,
middleInitial = m.MiddleInitial,
lastName = m.LastName,
createDate = m.CreateDate,
modifyDate = m.ModifyDate
}).ToArray(memberState)();
return mems;
}
}
The tables in my joins (Members, States, and MemberAddresses are actual tables in my Database). I created the object memberStates so I could use it in the query above (notice the Select New memberState. When the data is updated on the web page how do I persist the changes back to the Member Table? My Member Table consists of the following columns: UserId, FirstName, MiddleInitial, LastName, CreateDate, ModifyDate. I am not sure how save the changes back to the database.
Thanks,
If I remember correctly, you can create a view from the different tables (Members, States, and MemberAddresses) and add that to the data context. Then any modifications to data in the view object can be saved, and linq to sql will handle the commit correctly as long as all the relationships are clearly setup/defined in both the database and in the data context.
If you have a Member table, the dbml will most likely contain a Member class. To update a member in the database, you will have to create a new Member object, and the Attach it to the BulletinWizardDataContext.Members collection. Something similar to the following code should the trick (I have not tested the code):
using (BulletinWizardDataContext context = DataContext)
{
Member m = new Member() { UserId = userId };
context.Members.Attach(m);
m.FirstName = firstName;
// Set other properties
context.SubmitChanges();
}
Attach must be called before setting the properties. Also, Linq2Sql has some issues with Attach in the case where the properties of your object are set to default values (i.e. 0 for numeric values, false for booleans, null for string etc.). In this case Attach will not generate the correct SQL.
var m = myContext.Members.Single(m=> m.UserID == myMemState.userID);
m.FirstName = myMemState.firstName;
m.MiddleInitial = myMemState.middleInitial;
...
That would be the quick way. It does an additional roundtrip to the db, but will work well. If that's an issue for you, then do Attach like Jakob suggested. For that you have to have to do some extra steps, like reviewing the configuration for optimistic updates and make sure you have the original fields when doing the attach.