Entity Query

Ktorm provides a set of APIs named Entity Sequence, which can be used to obtain entity objects from databases. As the name implies, its style and use pattern are highly similar to the sequence APIs in Kotlin standard lib, as it provides many extension functions with the same names, such as filter, map, reduce, etc.

Get Entities by Sequences

To use sequence APIs, we need to create sequence objects first. In general, we’d like to define some extension properties for Database. These properties return new created sequence objects via sequenceOf and they can help us improve the readability of the code:

1 2	val Database.departments get() = this.sequenceOf(Departments) val Database.employees get() = this.sequenceOf(Employees)

The following code uses the find function to obtain an employee by its name:

1	val employee = database.employees.find { it.name eq "vince" }

This is natural, just like finding from a collection via Kotlin’s built-in extension functions, the only difference is the == in the lambda is replace by the eq function.

The find function accepts a closure typed of (T) -> ColumnDeclaring<Boolean>, executes a query with the filter condition returned by the closure, then returns the first entity object obtained from the result set. The closure function itself as the parameter also accepts a parameter typed of T, which is the current table object, so we can use it to access the table in the closure.

Generated SQL:

select t_employee.id as t_employee_id, t_employee.name as t_employee_name, t_employee.job as t_employee_job, t_employee.manager_id as t_employee_manager_id, t_employee.hire_date as t_employee_hire_date, t_employee.salary as t_employee_salary, t_employee.department_id as t_employee_department_id, _ref0.id as _ref0_id, _ref0.name as _ref0_name, _ref0.location as _ref0_location 
from t_employee 
left join t_department _ref0 on t_employee.department_id = _ref0.id 
where t_employee.id = ?

The generated SQL contains a very long field list, that’s necessary, Ktorm tries its best to avoid using select *. But for the sake of presentation, in later documents, we will still replace those field lists with select *.

Reading the generated SQL, we can find that Ktorm auto left joins t_employee’s reference table t_department using a foreign key. That’s because we bind the departmentId column to Departments table by a reference binding in the table object. By using the reference binding, when we obtain employees via sequence APIs, Ktorm will auto left join the referenced table, obtaining the departments at the same time, and filling them into property Employee.department.

Note: please avoid circular references while using reference bindings. For instance, now that Employees references Departments, then Departments cannot reference Employees directly or indirectly, otherwise a stack overflow will occur when Ktorm tries to left join Departments.

Now that referenced tables are auto left joined, we can also use their columns in our filter conditions. The code below uses Column.referenceTable to access departmentId’s referenced table and obtains an employee who works at Guangzhou:

val employee = database.employees.find {
    val dept = it.departmentId.referenceTable as Departments
    dept.location eq "Guangzhou"
}

To make it more elegant, we can add a get property to Employees table. The following code is completely equivalent to the above’s, but it reads more natural:

open class Employees(alias: String?) : Table<Employee>("t_employee", alias) {
    // Omit columns definitions here...
    val department get() = departmentId.referenceTable as Departments
}

val employee = database.employees.find { it.department.location eq "Guangzhou" }

Generated SQL:

select * 
from t_employee 
left join t_department _ref0 on t_employee.department_id = _ref0.id 
where _ref0.location = ?

Note: here we get the referenced table object via it.departmentId.referenceTable and cast it as Departments, which requires us to define tables as classes instead of singleton objects and to override the aliased function. More details can be found in the documentation of table aliases.

Besides the find function, Entity Sequence also provides many convenient functions for us. For example, using filter to find elements that matches the given condition, using groupingBy to group elements and do some aggregation, etc… Comparing with SQL DSL, sequence APIs are more functional, and can be used just like operating a collection in memory, so we recommend it as your first choice. For more documents, see Entity Sequence and Sequence Aggregation.

Get Entities by Query DSL

Sequence APIs will auto left join reference tables, that may be unnecessary in some cases. If you want more fine-grained control over the queries, you can use the query DSL introduced in the former sections. Ktorm provides a way to create entity objects from query DSL.

The example below uses createEntity function to obtain a list of entities from a query:

val employees = database
    .from(Employees)
    .select()
    .orderBy(Employees.id.asc())
    .map { row -> Employees.createEntity(row) }

employees.forEach { println(it) }

Here, we use the map function to iterate the query and create an entity object from the result set via createEntity for each row. createEntity is a function of Table class, it will create an entity object from the result set, using the binding configurations of the table, filling columns’ values into corresponding entities’ properties. And if there are any reference bindings to other tables, it will also create the referenced entity objects recursively.

However, the selected columns in query DSL are customizable, and there may be no columns from referenced tables. In this case, the function provides a parameter named withReferences, which is defaultly true. But if we set it to false, it will not obtain referenced entities’ data anymore, it will treat all reference bindings as nested bindings to the referenced entities’ primary keys. For example the binding c.references(Departments) { it.department }, it is equivalent to c.bindTo { it.department.id } for it, that avoids some unnecessary object creations.

1	Employees.createEntity(row, withReferences = false)

Get back the above example that we didn’t join any tables, so no matter we set the parameter to true or false, Ktorm will generate a simple SQL select * from t_employee order by t_employee.id and print the same results:

Employee{id=1, name=vince, job=engineer, hireDate=2018-01-01, salary=100, department=Department{id=1}}
Employee{id=2, name=marry, job=trainee, manager=Employee{id=1}, hireDate=2019-01-01, salary=50, department=Department{id=1}}
Employee{id=3, name=tom, job=director, hireDate=2018-01-01, salary=200, department=Department{id=2}}
Employee{id=4, name=penny, job=assistant, manager=Employee{id=3}, hireDate=2019-01-01, salary=100, department=Department{id=2}}

joinReferencesAndSelect

joinReferencesAndSelect is an extension function of QuerySource, it returns a new-created Query object, left joining all the reference tables recursively, and selecting all columns of them. Not only we can use the returned query to obtain all entity objects, but also we can call any other extension functions of Query to modify it. Actually, sequence APIs are based on this function to implement the auto joining of reference tables.

The example below queries all the employees along with their departments, sorting them by their IDs ascending:

val employees = database
    .from(Employees)
    .joinReferencesAndSelect()
    .orderBy(Employees.id.asc())
    .map { row -> Employees.createEntity(row) }

Generated SQL:

select * 
from t_employee 
left join t_department _ref0 on t_employee.department_id = _ref0.id 
order by t_employee.id

We can see in the SQL that the query above is equivalent to calling the leftJoin function manually, the following query is completely equal to the example above. Using joinReferencesAndSelect can help us to reduce some boilerplate code.

val emp = Employees
val dept = emp.departmentId.referenceTable as Departments

val employees = database
    .from(emp)
    .leftJoin(dept, on = emp.departmentId eq dept.id)
    .select(emp.columns + dept.columns)
    .orderBy(emp.id.asc())
    .map { row -> emp.createEntity(row) }