Skip to content

Dynamic queries

This page explains how to write dynamic SQL queries using ts-sql-query while preserving type safety, composability, and the declarative style of SQL. Instead of using if statements or imperative logic to conditionally add filters, projections, or joins, ts-sql-query provides a rich set of methods — such as .equalsIfValue() and .onlyWhen() — that allow you to express optional logic directly in the query definition. You can also define optional joins that are included only if their columns are used, and safely handle null values in projections.

Introduction

ts-sql-query provides several convenience methods ending in IfValue to help construct dynamic queries. These methods are automatically ignored when the provided values are null, undefined, or an empty string (unless the allowEmptyString flag in the connection is set to true — by default, it is false). When these methods are used in operations that return boolean values, ts-sql-query is smart enough to omit the operation when it is required, even when the operation is part of complex composition with ands and ors.

When you realize an insert or update, you can:

  • set a column value conditionally using the method setIfValue
  • replace a previously set value during the construction of the query using the method setIfSet or the method setIfSetIfValue
  • set a value if it was not previously set during the construction of the query using the method setIfNotSet or the method setIfNotSetIfValue
  • ignore a previously set value using the method ignoreIfSet
  • Don't worry if you end up with an UPDATE or DELETE without a WHERE clause — ts-sql-query will throw an error to prevent affecting all rows. You can allow explicitly having an update or delete with no where if you create it using the method updateAllowingNoWhere or deleteAllowingNoWhereFrom respectively

When you realize a select, you can:

  • specify in your order by clause that the order must be case insensitive when the column type is string (ignored otherwise). To do it, add insensitive at the end of the ordering criteria/mode
  • add a dynamic order by provided by the user without risk of SQL injection and without exposing the internal structure of the database. To build a dynamic order by use the method orderByFromString with the usual order by syntax (and with the possibility to use the insensitive extension), but using as column's name the name of the property in the resulting object
  • You can apply order by, limit and offset optionally calling orderByFromStringIfValue, limitIfValue and offsetIfValue

Additionally, you can:

  • create a boolean expression that only applies if a certain condition is met, calling the onlyWhen method in the boolean expression. The ignoreWhen method does the opposite.
  • create an expression that only applies if a certain condition is met; otherwise, the value will be null, calling the onlyWhenOrNull method in the expression. The ignoreWhenAsNull method does the opposite.
  • define an optional join in a select query. That join only must be included in the final query if the table involved in the join is used in the final query. For example, a column of the joined table was picked or used in a dynamic where.

Declarative dynamic queries

ts-sql-query embraces a declarative approach, in the spirit of SQL itself, when building dynamic queries.

As a result, writing conditions imperatively (e.g., using if statements to push filters into an array) won’t work as expected, and will bypass type safety and composability.

Instead, use the dynamic variants of comparison functions — those ending in IfValue, such as equalsIfValue, containsIfValue, etc. — to express optional filters directly and declaratively.

Avoid imperative control flow when building dynamic queries

When building dynamic queries, it's common in many ORMs or query builders to rely on imperative control flow — using if statements, mutating arrays, or combining query parts conditionally.

However, ts-sql-query embraces a declarative approach, inspired by the spirit of SQL itself. Instead of relying on control structures (if, for, etc.), you should express dynamic logic using built-in methods like equalsIfValue, onlyWhen, and ignoreWhen, which make your intent explicit, type-safe, and composable.

Easy dynamic queries

Methods ending in IfValue allow you to write dynamic queries in the easiest way; these methods work by ignoring the expression if the value provided is null, undefined, or an empty string (only when the allowEmptyString flag in the connection is not set to true, that is the default behaviour) return a special neutral boolean (ignoring the expression) that is ignored when it is used in ands, ors, ons or wheres.

const firstNameContains = 'ohn';
const lastNameContains = null;
const birthdayIs = null;
const searchOrderBy = 'name insensitive, birthday asc nulls last';

const searchedCustomers = connection.selectFrom(tCustomer)
    .where(
                tCustomer.firstName.containsIfValue(firstNameContains)
            .or(tCustomer.lastName.containsIfValue(lastNameContains))
        ).and(
            tCustomer.birthday.equalsIfValue(birthdayIs)
        )
    .select({
        id: tCustomer.id,
        name: tCustomer.firstName.concat(' ').concat(tCustomer.lastName),
        birthday: tCustomer.birthday
    })
    .orderByFromString(searchOrderBy)
    .executeSelectMany();

The executed query is:

select 
    id as id, 
    concat(first_name, ?, last_name) as name, 
    birthday as birthday 
from customer 
where first_name like concat('%', ?, '%') 
order by 
    lower(name), 
    birthday is null, 
    birthday asc

select 
    id as id, 
    concat(first_name, ?, last_name) as `name`, 
    birthday as birthday 
from customer 
where first_name like concat('%', ?, '%') 
order by 
    lower(`name`), 
    birthday is null, 
    birthday asc

select 
    id as "id", 
    first_name || :0 || last_name as "name", 
    birthday as "birthday" 
from customer 
where first_name like ('%' || :1 || '%') escape '\\' 
order by 
    lower("name"), 
    "birthday" asc nulls last

select 
    id as id, 
    first_name || $1 || last_name as name, 
    birthday as birthday 
from customer 
where first_name like ('%' || $2 || '%') 
order by 
    lower(name), 
    birthday asc nulls last

select 
    id as id, 
    first_name || ? || last_name as name, 
    birthday as birthday 
from customer 
where first_name like ('%' || ? || '%') escape '\\' 
order by 
    lower(name), 
    birthday asc nulls last

select 
    id as id, 
    first_name + @0 + last_name as name, 
    birthday as birthday 
from customer 
where first_name like ('%' + @1 + '%') 
order by 
    lower(name), 
    iif(birthday is null, 1, 0), 
    birthday asc

The parameters are: [ ' ', 'ohn' ]

The result type is: 

const customerWithId: Promise<{
    id: number;
    name: string;
    birthday?: Date;
}[]>

Ignorable boolean expression

You can create a boolean expression that only applies if a certain condition is met, calling the onlyWhen method at the end of the boolean expression; in case the condition is false it returns a special neutral boolean (ignoring the expression) that is ignored when it is used in ands, ors, ons or wheres. You can also use the ignoreWhen method at the end of the boolean expression to do the opposite; in case the condition is true it returns a special neutral boolean that is ignored. The onlyWhen and ignoreWhen methods can be useful to apply restrictions in the query, by example, when the user have some roles.

const userCompanyId = 16
const onlyCustomersOfUserCompany = true

const customers = await connection.selectFrom(tCustomer)
    .where(tCustomer.companyId.equals(userCompanyId).onlyWhen(onlyCustomersOfUserCompany))
    .select({
        firstName: tCustomer.firstName,
        lastName: tCustomer.lastName,
        birthday: tCustomer.birthday
    })
    .executeSelectMany()

The executed query is:

select 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday 
from customer 
where company_id = ?

select 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday 
from customer 
where company_id = ?

select 
    first_name as "firstName", 
    last_name as "lastName", 
    birthday as "birthday" 
from customer 
where company_id = :0

select 
    first_name as "firstName", 
    last_name as "lastName", 
    birthday as birthday 
from customer 
where company_id = $1

select 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday 
from customer 
where company_id = ?

select 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday 
from customer 
where company_id = @0

The parameters are: [ 16 ]

The result type is: 

const customers: Promise<{
    firstName: string;
    lastName: string;
    birthday?: Date;
}[]>

But in the case of onlyCustomersOfUserCompany is false, the condition in the where is omitted: 

const onlyCustomersOfUserCompany = false

The executed query is:

select 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday 
from customer

select 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday 
from customer

select 
    first_name as "firstName", 
    last_name as "lastName", 
    birthday as "birthday" 
from customer

select 
    first_name as "firstName", 
    last_name as "lastName", 
    birthday as birthday 
from customer

select 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday 
from customer

select 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday 
from customer

The parameters are: [ ]

Ignorable expression as null

You can create an expression that only applies if a certain condition is met, calling the onlyWhenOrNull method at the end of the expression; in case the condition is false it returns a null constant (ignoring the expression). You can also use the ignoreWhenAsNull method at the end of the expression to do the opposite; in case the condition is true it returns a null constant. The onlyWhenOrNull and ignoreWhenAsNull methods can be useful to apply restrictions in the query, by example, when the user have some roles.

const customerId = 10
const displayNames = true

const customerWithIdWithRules = connection.selectFrom(tCustomer)
    .where(tCustomer.id.equals(customerId))
    .select({
        id: tCustomer.id,
        firstName: tCustomer.firstName.onlyWhenOrNull(displayNames),
        lastName: tCustomer.lastName.onlyWhenOrNull(displayNames),
        birthday: tCustomer.birthday
    })
    .executeSelectOne()

The executed query is:

select 
    id as id, 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday 
from customer 
where id = ?

select 
    id as id, 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday 
from customer 
where id = ?

select 
    id as "id", 
    first_name as "firstName", 
    last_name as "lastName", 
    birthday as "birthday" 
from customer 
where id = :0

select 
    id as id, 
    first_name as "firstName", 
    last_name as "lastName", 
    birthday as birthday 
from customer 
where id = $1

select 
    id as id, 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday 
from customer 
where id = ?

select 
    id as id, 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday 
from customer 
where id = @0

The parameters are: [ 10 ]

The result type is: 

const customerWithIdWithRules: Promise<{
    id: number;
    firstName?: string;
    lastName?: string;
    birthday?: Date;
}>

But in the case of displayNames is false, the omitted expressions are replaced by null: 

const displayNames = false

The executed query is:

select 
    id as id, 
    null as firstName, 
    null as lastName, 
    birthday as birthday 
from customer 
where id = ?

select 
    id as id, 
    null as firstName, 
    null as lastName, 
    birthday as birthday 
from customer 
where id = ?

select 
    id as "id", 
    null as "firstName", 
    null as "lastName", 
    birthday as "birthday" 
from customer 
where id = :0

select 
    id as id, 
    null::text as "firstName", 
    null::text as "lastName", 
    birthday as birthday 
from customer 
where id = $1

select 
    id as id, 
    null as firstName, 
    null as lastName, 
    birthday as birthday 
from customer 
where id = ?

select 
    id as id, 
    null as firstName, 
    null as lastName, 
    birthday as birthday 
from customer 
where id = @0

The parameters are: [ 10 ]

Optional joins

You can write selects where the columns are picked dynamically, but maybe a join is required depending on the picked columns. ts-sql-query allows you to specify that a join should only be included in the final query if the table involved in the join is used in the final query (by example, a column from that table was selected or used in a dynamic WHERE clause).

To indicate the join can be optionally included in the query, you must create the join using one of the following methods:

  • optionalJoin
  • optionalInnerJoin
  • optionalLeftJoin
  • optionalLeftOuterJoin
const companyName = 'My company name'

const customers = connection.selectFrom(tCustomer)
    .optionalJoin(tCompany).on(tCompany.id.equals(tCustomer.companyId))
    .where(tCompany.name.equalsIfValue(companyName))
    .select({
        firstName: tCustomer.firstName,
        lastName: tCustomer.lastName,
        birthday: tCustomer.birthday
    })
    .executeSelectMany()

The executed query is:

select 
    customer.first_name as firstName, 
    customer.last_name as lastName, 
    customer.birthday as birthday 
from customer 
join company on company.id = customer.company_id 
where company.name = ?

select 
    customer.first_name as firstName, 
    customer.last_name as lastName, 
    customer.birthday as birthday 
from customer 
join company on company.id = customer.company_id 
where company.`name` = ?

select 
    customer.first_name as "firstName", 
    customer.last_name as "lastName", 
    customer.birthday as "birthday" 
from customer 
join company on company.id = customer.company_id 
where company.name = :0

select 
    customer.first_name as "firstName", 
    customer.last_name as "lastName", 
    customer.birthday as birthday 
from customer 
join company on company.id = customer.company_id 
where company.name = $1

select 
    customer.first_name as firstName, 
    customer.last_name as lastName, 
    customer.birthday as birthday 
from customer 
join company on company.id = customer.company_id 
where company.name = ?

select 
    customer.first_name as firstName, 
    customer.last_name as lastName, 
    customer.birthday as birthday 
from customer 
join company on company.id = customer.company_id 
where company.name = @0

The parameters are: [ "My company name" ]

The result type is: 

const customers: Promise<{
    firstName: string;
    lastName: string;
    birthday?: Date;
}[]>

But in the case of companyName is null or undefined, the condition in the where is omitted; in consequence, the company table is not used; thus the join is omitted: 

const companyName = null

The executed query is:

select 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday 
from customer

select 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday 
from customer

select 
    first_name as "firstName", 
    last_name as "lastName", 
    birthday as "birthday" 
from customer

select 
    first_name as "firstName", 
    last_name as "lastName", 
    birthday as birthday 
from customer

select 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday 
from customer

select 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday 
from customer

The parameters are: [ ]

Warning

An omitted join can change the number of returned rows depending on your data structure. This behaviour doesn't happen when all rows of the initial table have one row in the joined table (or none if you use a left join), but not many rows.

You can also use optional joins with ignorable expression as null

const canSeeCompanyInfo = false

const customerWithOptionalCompany = connection.selectFrom(tCustomer)
    .optionalInnerJoin(tCompany).on(tCompany.id.equals(tCustomer.companyId))
    .select({
        id: tCustomer.id,
        firstName: tCustomer.firstName,
        lastName: tCustomer.lastName,
        birthday: tCustomer.birthday,
        companyId: tCompany.id.onlyWhenOrNull(canSeeCompanyInfo),
        companyName: tCompany.name.onlyWhenOrNull(canSeeCompanyInfo)
    })
    .where(tCustomer.id.equals(12))
    .executeSelectMany()

The executed query is:

select 
    id as id, 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday, 
    null as companyId, 
    null as companyName 
from customer 
where id = ?

select 
    id as id, 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday, 
    null as companyId, 
    null as companyName 
from customer 
where id = ?

select 
    id as "id", 
    first_name as "firstName", 
    last_name as "lastName", 
    birthday as "birthday", 
    null as "companyId", 
    null as "companyName" 
from customer 
where id = :0

select 
    id as id, 
    first_name as "firstName", 
    last_name as "lastName", 
    birthday as birthday, 
    null::int4 as "companyId", 
    null::text as "companyName" 
from customer 
where id = $1

select 
    id as id, 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday, 
    null as companyId, 
    null as companyName 
from customer 
where id = ?

select 
    id as id, 
    first_name as firstName, 
    last_name as lastName, 
    birthday as birthday, 
    null as companyId, 
    null as companyName 
from customer 
where id = @0

The parameters are: [ 12 ]

The result type is: 

const customerWithOptionalCompany: Promise<{
    id: number;
    firstName: string;
    lastName: string;
    birthday?: Date;
    companyName?: string;
    companyId?: number;
}[]>

But in the case of a column provided by the join is required, like when canSeeCompanyInfo is: 

const canSeeCompanyInfo = true

The executed query is:

select 
    customer.id as id, 
    customer.first_name as firstName, 
    customer.last_name as lastName, 
    customer.birthday as birthday, 
    company.id as companyId, 
    company.name as companyName 
from customer 
inner join company on company.id = customer.company_id 
where customer.id = ?

select 
    customer.id as id, 
    customer.first_name as firstName, 
    customer.last_name as lastName, 
    customer.birthday as birthday, 
    company.id as companyId, 
    company.`name` as companyName 
from customer 
inner join company on company.id = customer.company_id 
where customer.id = ?

select 
    customer.id as "id", 
    customer.first_name as "firstName", 
    customer.last_name as "lastName", 
    customer.birthday as "birthday", 
    company.id as "companyId", 
    company.name as "companyName" 
from customer 
inner join company on company.id = customer.company_id 
where customer.id = :0

select 
    customer.id as id, 
    customer.first_name as "firstName", 
    customer.last_name as "lastName", 
    customer.birthday as birthday, 
    company.id as "companyId", 
    company.name as "companyName" 
from customer 
inner join company on company.id = customer.company_id 
where customer.id = $1

select 
    customer.id as id, 
    customer.first_name as firstName, 
    customer.last_name as lastName, 
    customer.birthday as birthday, 
    company.id as companyId, 
    company.name as companyName 
from customer 
inner join company on company.id = customer.company_id 
where customer.id = ?

select 
    customer.id as id, 
    customer.first_name as firstName, 
    customer.last_name as lastName, 
    customer.birthday as birthday, 
    company.id as companyId, 
    company.name as companyName 
from customer 
inner join company on company.id = customer.company_id 
where customer.id = @0

The parameters are: [ 12 ]

Warning

An omitted join can change the number of returned rows depending on your data structure. This behaviour doesn't happen when all rows of the initial table have one row in the joined table (or none if you use a left join), but not many rows.