sql (for fields)

This page refers to the sql parameter that is part of a field.

sql can also be used as part of a derived table, as described on the sql (for derived tables) parameter documentation page.

Usage

view: view_name {
  dimension: field_name {
    sql: ${revenue_in_dollars} - ${inventory_item.cost_in_dollars} ;;
  }
}
Hierarchy
sql
Possible Field Types
Dimension, Dimension Group, Filter, Measure

Accepts
A SQL expression

Special Rules
A SQL expression that varies according to the type of the field (as described in detail on this documentation page)

Definition

The sql parameter takes several types of SQL expressions that will define a dimension, measure, or filter. The expression you need to write varies based on the type of field you are creating. More details about dimension and filter types can be found on the Dimension, filter, and parameter types documentation page, while more details about measure types can be found on the Measure types documentation page. See also the Incorporating SQL and referring to LookML objects documentation page.

sql for Dimensions

The sql block for dimensions can generally take any valid SQL that would go into a single column of a SELECT statement. These statements generally rely on Looker's substitution operator, which has several forms:

  • ${TABLE}.column_name references a column in the table that is connected to the view you're working on.
  • ${dimension_name} references a dimension within the view you're working on.
  • ${view_name.dimension_name} references a dimension from another view.
  • ${view_name.SQL_TABLE_NAME} references another view or derived table. (Note that SQL_TABLE_NAME in this reference is a literal string; you do not need to replace it with anything.)

If sql is left unspecified, then Looker assumes that there is a column in the underlying table with the same name as the field. For example, selecting a field called city without a sql parameter would be equivalent to specifying sql: ${TABLE}.city.

The sql parameter of a dimension cannot include any aggregations. This means it cannot contain SQL aggregations or references to LookML measures. If you want to create a field with sql that includes a SQL aggregation or that references a LookML measure, use a sql parameter in a measure, not in a dimension.

A very simple dimension that takes the value directly from a column called revenue could look like:

dimension: revenue_in_cents {
  sql: ${TABLE}.revenue ;;
  type: number
}

A dimension that relies on another dimension in the same view could look like this:

dimension: revenue_in_dollars {
  sql: ${revenue_in_cents} / 100 ;;
  type: number
}

A dimension that relies on another dimension in a different view could look like this:

dimension: profit_in_dollars {
  sql: ${revenue_in_dollars} - ${inventory_item.cost_in_dollars} ;;
  type: number
}

A dimension that relies on another dimension in a derived table could look like this:

dimension: average_margin {
  sql: (SELECT avg(${gross_margin} FROM ${order_facts.SQL_TABLE_NAME})) ;;
  type: number
}

More advanced SQL users can perform relatively advanced calculations, including correlated sub-queries (note: not all database dialect support correlated subqueries):

dimension: user_order_sequence_number {
  type: number
  sql:
    (
      SELECT COUNT(*)
      FROM orders AS o
      WHERE o.id <= ${TABLE}.id
        AND o.user_id = ${TABLE}.user_id
    ) ;;
}

For further details, refer to the documentation for a specific dimension type.

sql for Dimension Groups

The sql parameter for a dimension_group takes any valid SQL expression that contains data in a timestamp, datetime, date, epoch, or yyyymmdd format.

sql for Measures

The sql block for measures typically takes one of two forms:

  • The SQL over which an aggregate function (such as COUNT, SUM, AVG) will be performed, again using Looker's substitution operator as described in the SQL for Dimensions section
  • A value based on several other measures

For example, to calculate the total revenue in dollars, we might use:

measure: total_revenue_in_dollars {
  sql: ${revenue_in_dollars} ;;
  type: sum
}

To calculate our total profit, we might use:

measure: total_revenue_in_dollars {
  sql: ${total_revenue_in_dollars} - ${inventory_item.total_cost_in_dollars} ;;
  type: number
}

For further details, see the documentation for a specific measure type.

For a count measure type, you can leave off the sql parameter.

For other types of measures, if sql is left unspecified then Looker assumes that there is a column in the underlying table with the same name as the field. Since a measure should have a name indicating that it is an aggregate of an underlying set of values, in practice you should always include a sql parameter.

SQL math challenges

There are two frequent challenges that come up with division in the sql parameter.

First, if you are using division in your calculation, you want to protect against the possibility of dividing by zero, which will cause a SQL error. To do so, use the SQL NULLIF function. For example, this example means "if the denominator is zero, treat it like NULL instead":

measure: active_users_percent {
  sql: ${active_users} / NULLIF(${users}, 0) ;;
  type: number
}

Another issue is the way that SQL handles integer math. If you divide 5 by 2, most people expect the result to be 2.5. However, many SQL dialects will return the result as just 2, because when it divides two integers it also gives the result as an integer. To address this, you can multiply the numerator by a decimal number to force SQL into returning a decimal result. For example:

measure: active_users_percent {
  sql: 100.00 * ${active_users} / NULLIF(${users}, 0) ;;
  type: number
}

Liquid variables with sql

You can also use Liquid variables with the sql parameter. Liquid variables let you access data such as the values in a field, data about the field, and filters applied to the field.

For example, this dimension masks a customer password according to a Looker user attribute:

dimension: customer_password {
  sql:
    {% if _user_attributes['pw_access'] == 'yes' %}
      ${password}
    {% else %}
      "Password Hidden"
    {% endif %} ;;
}