I have a complex query(using multiple joins and subqueries) written in HQL which I have used in a Repository class. Similar to one below -
#Repository
public interface DataRepository extends PagingAndSortingRepository<Data,String> {
public List<Data> findByService(#Param("service")Service service, Pageable page);
#Query("SELECT DISTINCT d from Data d "
+" WHERE (d.working in (SELECT d1 from Data d1 "
+" JOIN d1.working d1w "
+" JOIN d1.service s WITH (s in (:serviceList)))"
+" OR d.cleared IS NOT NULL) AND [..several other CRITERIA]")
public Page<Data> findForServices(#Param("serviceList")Set<Service> serviceList, Pageable page);
....
Now I need to add criteria to it dynamically. These criteria are flexible in number which is holding me from including it into the HQL straightaway. Is it anyhow possible?
Sifting through the internet I have come across solutions for dynamic query. But, I guess they would be working only for cases where I do not have a custom query i.e.- no #Query at the query in the repository.
There was another interesting question I found. But that also suits for a case where you have a single table to query.
I do not want to be switching over to raw SQL queries. How do I solve this?
The mentioned Criteria API with specifications and predicates is a little bit difficult to get used to but it is a good way to handle dynamic conditions.
I don't think it is possible to mix the annotation based query with programmatic query creation.
Related
Hi I use spring jpa to access my data in my spring boot project.I am wondering that is there any difference between #Query annatotation and critearia api in jpa.Are they totaly same or is there any difference(Their writing styles are different ,and I mean any performance or other issue between them)
Mostly I prefer #Query annotation it looks simple.Or any other option some one can advice like #Query or criteria api in spring jpa.And is there any disadvantages of #Query style?
#Query("SELECT u FROM User u WHERE u.status = 1")
Collection<User> findAllActiveUsers();
List<Book> findBooksByAuthorNameAndTitle(String authorName, String title) {
CriteriaBuilder cb = em.getCriteriaBuilder();
CriteriaQuery<Book> cq = cb.createQuery(Book.class);
Using #Query we can pass static query or pre compiled query so we can perform both select and non-select operations on the data
where as Criteria is suitable for executing Dynamic Queries such requirements occurs when data are know at run time
but using criteria api we can only perform select operations on the data.
For example
#Query(value = "SELECT u FROM User u")
List<User> findAllUsers(Sort sort);
We can also work with pre compiled query using #Query
For Example
#Query("SELECT u FROM User u WHERE u.status = :status and u.name = :name")
User findUserByStatusAndNameNamedParams(
#Param("status") Integer status,
#Param("name") String name);)
Dynamic queries like
Criteria cr = session.createCriteria(Employee.class);
// To get records having salary more than 2000
cr.add(Restrictions.gt("salary", 2000));
// To get records having salary less than 2000
cr.add(Restrictions.lt("salary", 2000));
Actual use of Dynamic queries comes when we'll encounter the need for building SQL statements based on conditions or data sets whose values are only known at runtime. And, in those cases, we can't just use a static query So we can't just use the #Query annotation since we can't provide a static SQL statement.In such case we use Criteria API
For more info follow the link provided
#Query and Criteria
I have a requirement to create a REST api. Api allows user to provide dynamic search criteria in URL. For example, let say I have NOTES table with column as Note_ID, NOTE_TEXT, STATUS, PERSON_ID. This table is used to keep notes of every person.
Now I want my REST api to be as https://server:host/MyApi/Notes?search=NoteText=='My Java adventure'. API should provide all notes having NOTE_TEXT as 'My Java adventure'. Similarly user can provide status also in url and also he can use operators as LIKE. I was able to do it via rsql parser as mentioned in https://www.baeldung.com/rest-api-search-language-rsql-fiql
Now I have additional requirement that based on user security person_id filter should be applied on query automatically.
I found that we can't have findBy method which can take Specification, Pageable and extra personId. For example I can't have a repository function as
findByPersonId(Specification spec, Pageable page, Long personId);
I thought of using SpEL to use it, but then I found that if we use #Query annotation on findBy method, Specifications are ignored.
Seems like there is no way I can have Specification and #Query both. I need to add more clauses using specification only. In reality my where clause is very complex which I have to append and getting it with Specification seems to be difficult. Its something like
Select * from NOTES where exists (select 'x' from ABC a where n.person_id = a.person_id)
Is there a way I can write #Query and also have Specification working on top of it?
Ideally I have achieve a query like
select * from test.SCH_FORUM_THREAD t
where exists (select 'x' from test.FORUM_THREAD_ACCESS fta, school.SCH_GROUP_PERSON gp
where gp.GROUP_ID = fta.GROUP_ID
and t.THREAD_ID = fta.THREAD_ID
and gp.PERSON_ID = :personId)
or exists (select 'x' from test.FORUM_THREAD_ACCESS fta
where fta.THREAD_ID = t.THREAD_ID
and fta.PERSON_ID = :personId);
So there are two exists clauses with or condition. I was able to make second exists by following How to write query(include subquery and exists) using JPA Criteria Builder
Now struggling with first exists as it has join also. Any idea how to do that with Specification.
Also as there are two exists, does that mean I need two specifications. Can I achieve it in one specification.
I was able to resolve it by creating a complex specification code. Something like
#Override
public Predicate toPredicate(Root<ForumThread> root, CriteriaQuery<?> query, CriteriaBuilder builder) {
Subquery<ForumThread> subQuery = query.subquery(ForumThread.class);
Root<ForumThread> subRoot = subQuery.from(ForumThread.class);
Join<ForumThreadAccess, GroupPerson> fragpjoin = subRoot.join("groupPersons");
Predicate threadPredicate = builder.equal(root.get("threadId"), subRoot.get("threadId"));
Predicate personPredicate = builder.equal(fragpjoin.get("personId"), personId);
subQuery.select(subRoot).where(threadPredicate, personPredicate);
Predicate existsGroupPredicate = builder.exists(subQuery);
Subquery<ForumThreadAccess> subQuery1 = query.subquery(ForumThreadAccess.class);
Root<ForumThreadAccess> subRoot1 = subQuery1.from(ForumThreadAccess.class);
Predicate threadPredicate1 = builder.equal(root.get("threadId"), subRoot1.get("threadId"));
Predicate personPredicate1 = builder.equal(subRoot1.get("personId"), personId);
subQuery1.select(subRoot1).where(threadPredicate1, personPredicate1);
Predicate existsPersonPredicate = builder.exists(subQuery1);
return builder.or(existsGroupPredicate,existsPersonPredicate);
}
To make it work your entities should also have proper #OneToMany and #ManyToMany in place.
Thanks
I have created two entities in JPA, Listing and ItemType - these exist in a many-to-many relationship (Hibernate auto-generates a junction table). I'm trying to find the best way to create a query which accepts a dynamic list of item type Strings and returns the IDs of all listings which match the specified item types, but I am a recent initiate in JPA.
At present I'm using JpaRepository to create relatively simple queries. I've been trying to do this using CriteriaQuery but some close-but-not-quite answers I've read elsewhere seem to suggest that because this is in Spring, this may not be the best approach and that I should be handling this using the JpaRepository implementation itself. Does that seem reasonable?
I have a query which doesn't feel a million miles away (based on Baeldung's example and my reading on WikiBooks) but for starters I'm getting a Raw Type warning on the Join, not to mention that I'm unsure if this will run and I'm sure there's a better way of going about this.
public List<ListingDTO> getListingsByItemType(List<String> itemTypes) {
List<ListingDTO> listings = new ArrayList<>();
CriteriaQuery<Listing> criteriaQuery = criteriaBuilder.createQuery(Listing.class);
Root<Listing> listing = criteriaQuery.from(Listing.class);
//Here Be Warnings. This should be Join<?,?> but what goes in the diamond?
Join itemtype = listing.join("itemtype", JoinType.LEFT);
In<String> inClause = criteriaBuilder.in(itemtype.get("name"));
for (String itemType : itemTypes) {
inClause.value(itemType);
}
criteriaQuery.select(listing).where(inClause);
TypedQuery<Listing> query = entityManager.createQuery(criteriaQuery);
List<Listing> results = query.getResultList();
for (Listing result : results) {
listings.add(convertListingToDto(result));
}
return listings;
}
I'm trying to understand how best to pass in a dynamic list of names (the field in ItemType) and return a list of unique ids (the PK in Listing) where there is a row which matches in the junction table. Please let me know if I can provide any further information or assistance - I've gotten the sense that JPA and its handling of dynamic queries like this is part of its bread and butter!
The criteria API is useful when you need to dynamically create a query based on various... criteria.
All you need here is a static JPQL query:
select distinct listing from Listing listing
join listing.itemTypes itemType
where itemType.name in :itemTypes
Since you're using Spring-data-jpa, you just need to define a method and annotate it with #Query in your repository interface:
#Query("<the above query>")
List<Listing> findByItemTypes(List<String> itemTypes)
I came to a problem where I need to put ordering by multiplication of two columns of entity, for the sake of imagination entity is:
#Entity
public class Entity {
#Column(name="amount")
private BigDecimal amount;
#Column(name="unitPprice")
private BigDecimal unitPrice;
.
.
.
many more columns
}
My repo interface implements JpaRepository and QuerydslPredicateExecutor,
but I am struggling to find a way to order my data by "amount*unitPrice",
as I can't find a way to put it into
PageRequest (new Sort.Order(ASC, "amount * unitPrice"))
without having PropertyReferenceException: No property amount * unitPrice... thrown.
I can't user named query, as my query takes quite massive filter based on user inputs (can't put where clause into query, because if user hasn't selected any value, where clause can't just be in query).
To make it simple. I need something like findAll(Predicate, Pageable), but I need to force that query to order itself by "amount * unitPrice", but also have my Preditate (filter) and Pageable (offset, limit, other sortings) untouched.
Spring Sort can be used only for sorting by properties, not by expressions.
But you can create a unique sort in a Predicate, so you can add this sort-predicate to your other one before you call the findAll method.
I've got a bit of a bizarre problem that I can't figure out why it's happening. I'm sure I did something wrong, because this is my first time using a data projection and I've never had such problems using DTOs.
Pretty much I have a SELECT statemen that is returning certain columns of various data types. And I have an interface that I'm passing to the JPA Repository so it can do the interface mapping. But instead of mapping the results based on the column name (eg. 'accountnum' -> getAccountnumber()), it's mapping the columns in alphabetical order. So if 'date_of_order' is the first in the SELECT statement, its value will be returned by getAccountnumber().
I have a projected interface that looks something like this:
public interface FlatSearchResult {
String getAccountnumber();
UUID getTrackingId;
Date getDateOfOrder;
}
My model has three tables something like this:
ACCOUNT
- account_id : uuid (pkey)
- accountnumber : string
ORDERS
- order_id : uuid (pkey)
- date_of_order : timestamp
- account_id : uuid (fkey)
TRACKING
- tracking_id : uuid (pkey)
- order_id : uuid (fkey)
There's other columns in each of those tables, but they're not relevant.
I have a repository defined with a simple query:
public interface OrderTrackingRepository extends JpaRepository<Account, UUID> {
#Query( nativeQuery = true,
value = "SELECT o.date_of_order, a.accountnumber, t.tracking_id " +
"FROM account as a " +
"INNER JOIN orders as o USING (account_id) " +
"INNER JOIN tracking as t USING (tracking_id) " +
"WHERE a.accountnumber = :acctnum")
<T> Collection<T> findOrderInfoForAccount(#Param("acctnum") acctNumber, Class<T> type);
}
When I call this method, the correct rows are returned by the query. But instead of mapping using the column name (eg. date_of_order to getDateOfOrder()), it is mapping based on the order of the columns in the SELECT statement to the alphabetically-ordered methods in the interface.
So:
SELECT date_of_order, accountnumber, tracking_id
Results in:
getAccountNumber() -> date_of_order
getDateOfOrder() -> accountnumber
getTrackingId() -> tracking_id
It will consistently return in this fashion, so it's not a transient issue.
As a temporary workaround, I've reordered the columns in my SELECT statement. But I would rather not have to do this since it's like iterating through a result set and relying on column position, which just makes me twitchy....
How can I get Spring JPA to map from the result set to my interface? Do I need to annotate my projection interface's methods with something to tell Spring what column name it's referring to?
My database is Postgres. I'm using Spring 5.0.2.RELEASE and Spring-Boot 2.0.0.M7. I can adjust either of those to newer versions if needed, but nothing older. I'm using C3P0 0.9.5.2 for my connection pooling, and postgres-9.2-1002.jdbc4. All my other dependencies (hibernate, etc) are what is pulled in by this version of Spring-Boot.
Not sure if this is the correct solution because it only fits 80% of the description. But it is too long for a comment. So here we go.
I think you misunderstood #osamayaccoub or the documentation. Your property name is fine. But the columns in your select should match the java convention.
So the first attempt to fix that would be
value = "SELECT o.date_of_order as dateOfOrder, a.accountnumber as accountNumber, t.tracking_id as trackingId "
Note: This might actually work, but might break later, so read on, even if it does work
But Postgres converts everything that isn't double quoted into lower case (Oracle and MySql do similar stuff though details vary, don't know about other DBs yet). So you really should use:
value = "SELECT o.date_of_order as \"dateOfOrder\", a.accountnumber as \"accountNumber\", t.tracking_id as \"trackingId\" "
This probably doesn't work, because the Hibernate version you are using has a bug in that it converted everything to lower case.
So you should upgrade to the latest Hibernate version 5.3.13 which has the issue fixed.
This bug fix interestingly might break the version without the double quotes.
But it should work again with this PR for this Spring Data JPA issue.
The part I don't understand is, why stuff gets assigned using the column order.
I had the same problem and i solved by odering the query columns alphabetically.
In you case:
public interface OrderTrackingRepository extends JpaRepository<Account, UUID> {
#Query( nativeQuery = true,
value = "SELECT a.accountnumber, o.date_of_order, t.tracking_id " +
"FROM account as a " +
"INNER JOIN orders as o USING (account_id) " +
"INNER JOIN tracking as t USING (tracking_id) " +
"WHERE a.accountnumber = :acctnum")
<T> Collection<T> findOrderInfoForAccount(#Param("acctnum") acctNumber, Class<T> type);
}
So you will get:
getAccountNumber() -> accountnumber
getDateOfOrder() -> date_of_order
getTrackingId() -> tracking_id
Hibernate sorts the query in alphabetic order so you have to change the select to: "SELECT a.accountnumber, o.date_of_order, t.tracking_id ..." and the interface's getters should follow the same alphabetic order.