Conditional functions

Overview

Using Conditional Results Directly

Conditionals always result to 0, 1 or NULL. So you can use conditional results directly like this:

SELECT left < right AS is_small
FROM LEFT_RIGHT

┌─is_small─┐
│     ᴺᵁᴸᴸ │
│        1 │
│        0 │
│        0 │
│     ᴺᵁᴸᴸ │
└──────────┘

NULL Values in Conditionals

When NULL values are involved in conditionals, the result will also be NULL.

SELECT
    NULL < 1,
    2 < NULL,
    NULL < NULL,
    NULL = NULL

┌─less(NULL, 1)─┬─less(2, NULL)─┬─less(NULL, NULL)─┬─equals(NULL, NULL)─┐
│ ᴺᵁᴸᴸ          │ ᴺᵁᴸᴸ          │ ᴺᵁᴸᴸ             │ ᴺᵁᴸᴸ               │
└───────────────┴───────────────┴──────────────────┴────────────────────┘

So you should construct your queries carefully if the types are Nullable.

The following example demonstrates this by failing to add equals condition to multiIf.

SELECT
    left,
    right,
    multiIf(left < right, 'left is smaller', left > right, 'right is smaller', 'Both equal') AS faulty_result
FROM LEFT_RIGHT

┌─left─┬─right─┬─faulty_result────┐
│ ᴺᵁᴸᴸ │     4 │ Both equal       │
│    1 │     3 │ left is smaller  │
│    2 │     2 │ Both equal       │
│    3 │     1 │ right is smaller │
│    4 │  ᴺᵁᴸᴸ │ Both equal       │
└──────┴───────┴──────────────────┘

CASE statement

The CASE expression in ClickHouse provides conditional logic similar to the SQL CASE operator. It evaluates conditions and returns values based on the first matching condition.

ClickHouse supports two forms of CASE:

1. CASE WHEN ... THEN ... ELSE ... END
   This form allows full flexibility and is internally implemented using the multiIf function. Each condition is evaluated independently, and expressions can include non-constant values.

SELECT
    number,
    CASE
        WHEN number % 2 = 0 THEN number + 1
        WHEN number % 2 = 1 THEN number * 10
        ELSE number
    END AS result
FROM system.numbers
WHERE number < 5;

-- is translated to
SELECT
    number,
    multiIf((number % 2) = 0, number + 1, (number % 2) = 1, number * 10, number) AS result
FROM system.numbers
WHERE number < 5

┌─number─┬─result─┐
│      0 │      1 │
│      1 │     10 │
│      2 │      3 │
│      3 │     30 │
│      4 │      5 │
└────────┴────────┘

5 rows in set. Elapsed: 0.002 sec.

2. CASE <expr> WHEN <val1> THEN ... WHEN <val2> THEN ... ELSE ... END
   This more compact form is optimized for constant value matching and internally uses caseWithExpression().

For example, the following is valid:

SELECT
    number,
    CASE number
        WHEN 0 THEN 100
        WHEN 1 THEN 200
        ELSE 0
    END AS result
FROM system.numbers
WHERE number < 3;

-- is translated to

SELECT
    number,
    caseWithExpression(number, 0, 100, 1, 200, 0) AS result
FROM system.numbers
WHERE number < 3

┌─number─┬─result─┐
│      0 │    100 │
│      1 │    200 │
│      2 │      0 │
└────────┴────────┘

3 rows in set. Elapsed: 0.002 sec.

This form also does not require return expressions to be constants.

SELECT
    number,
    CASE number
        WHEN 0 THEN number + 1
        WHEN 1 THEN number * 10
        ELSE number
    END
FROM system.numbers
WHERE number < 3;

-- is translated to

SELECT
    number,
    caseWithExpression(number, 0, number + 1, 1, number * 10, number)
FROM system.numbers
WHERE number < 3

┌─number─┬─caseWithExpr⋯0), number)─┐
│      0 │                        1 │
│      1 │                       10 │
│      2 │                        2 │
└────────┴──────────────────────────┘

3 rows in set. Elapsed: 0.001 sec.

Caveats

ClickHouse determines the result type of a CASE expression (or its internal equivalent, such as multiIf) before evaluating any conditions. This is important when the return expressions differ in type, such as different timezones or numeric types.

• The result type is selected based on the largest compatible type among all branches.
• Once this type is selected, all other branches are implicitly cast to it - even if their logic would never be executed at runtime.
• For types like DateTime64, where the timezone is part of the type signature, this can lead to surprising behavior: the first encountered timezone may be used for all branches, even when other branches specify different timezones.

For example, below all rows return the timestamp in the timezone of the first matched branch i.e. Asia/Kolkata

SELECT
    number,
    CASE
        WHEN number = 0 THEN fromUnixTimestamp64Milli(0, 'Asia/Kolkata')
        WHEN number = 1 THEN fromUnixTimestamp64Milli(0, 'America/Los_Angeles')
        ELSE fromUnixTimestamp64Milli(0, 'UTC')
    END AS tz
FROM system.numbers
WHERE number < 3;

-- is translated to

SELECT
    number,
    multiIf(number = 0, fromUnixTimestamp64Milli(0, 'Asia/Kolkata'), number = 1, fromUnixTimestamp64Milli(0, 'America/Los_Angeles'), fromUnixTimestamp64Milli(0, 'UTC')) AS tz
FROM system.numbers
WHERE number < 3

┌─number─┬──────────────────────tz─┐
│      0 │ 1970-01-01 05:30:00.000 │
│      1 │ 1970-01-01 05:30:00.000 │
│      2 │ 1970-01-01 05:30:00.000 │
└────────┴─────────────────────────┘

3 rows in set. Elapsed: 0.011 sec.

Here, ClickHouse sees multiple DateTime64(3, <timezone>) return types. It infers the common type as DateTime64(3, 'Asia/Kolkata' as the first one it sees, implicitly casting other branches to this type.

This can be addressed by converting to a string to preserve intended timezone formatting:

SELECT
    number,
    multiIf(
        number = 0, formatDateTime(fromUnixTimestamp64Milli(0), '%F %T', 'Asia/Kolkata'),
        number = 1, formatDateTime(fromUnixTimestamp64Milli(0), '%F %T', 'America/Los_Angeles'),
        formatDateTime(fromUnixTimestamp64Milli(0), '%F %T', 'UTC')
    ) AS tz
FROM system.numbers
WHERE number < 3;

-- is translated to

SELECT
    number,
    multiIf(number = 0, formatDateTime(fromUnixTimestamp64Milli(0), '%F %T', 'Asia/Kolkata'), number = 1, formatDateTime(fromUnixTimestamp64Milli(0), '%F %T', 'America/Los_Angeles'), formatDateTime(fromUnixTimestamp64Milli(0), '%F %T', 'UTC')) AS tz
FROM system.numbers
WHERE number < 3

┌─number─┬─tz──────────────────┐
│      0 │ 1970-01-01 05:30:00 │
│      1 │ 1969-12-31 16:00:00 │
│      2 │ 1970-01-01 00:00:00 │
└────────┴─────────────────────┘

3 rows in set. Elapsed: 0.002 sec.

clamp

Introduced in: v24.5

Restricts a value to be within the specified minimum and maximum bounds.

If the value is less than the minimum, returns the minimum. If the value is greater than the maximum, returns the maximum. Otherwise, returns the value itself.

All arguments must be of comparable types. The result type is the largest compatible type among all arguments.

Syntax

clamp(value, min, max)

Arguments

• value — The value to clamp. - min — The minimum bound. - max — The maximum bound.

Returned value

Returns the value, restricted to the [min, max] range.

Examples

Basic usage

SELECT clamp(5, 1, 10) AS result;

┌─result─┐
│      5 │
└────────┘

Value below minimum

SELECT clamp(-3, 0, 7) AS result;

┌─result─┐
│      0 │
└────────┘

Value above maximum

SELECT clamp(15, 0, 7) AS result;

┌─result─┐
│      7 │
└────────┘

greatest

Introduced in: v1.1

Returns the greatest value among the arguments.

• For arrays, returns the lexicographically greatest array.
• For DateTime types, the result type is promoted to the largest type (e.g., DateTime64 if mixed with DateTime32).

Syntax

greatest(x1[, x2, ..., xN])

Arguments

• x1[, x2, ..., xN] — One or multiple values to compare. All arguments must be of comparable types.

Returned value

The greatest value among the arguments, promoted to the largest compatible type.

Examples

Numeric types

SELECT greatest(1, 2, toUInt8(3), 3.) AS result, toTypeName(result) AS type;
-- The type returned is a Float64 as the UInt8 must be promoted to 64 bit for the comparison.

┌─result─┬─type────┐
│      3 │ Float64 │
└────────┴─────────┘

Arrays

SELECT greatest(['hello'], ['there'], ['world']);

┌─greatest(['hello'], ['there'], ['world'])─┐
│ ['world']                                 │
└───────────────────────────────────────────┘

DateTime types

SELECT greatest(toDateTime32(now() + toIntervalDay(1)), toDateTime64(now(), 3));
-- The type returned is a DateTime64 as the DateTime32 must be promoted to 64 bit for the comparison.

┌─greatest(toD⋯(now(), 3))─┐
│  2025-05-28 15:50:53.000 │
└──────────────────────────┘

if

Introduced in: v1.1

Performs conditional branching.

• If the condition cond evaluates to a non-zero value, the function returns the result of the expression then.
• If cond evaluates to zero or NULL, the result of the else expression is returned.

The setting short_circuit_function_evaluation controls whether short-circuit evaluation is used.

If enabled, the then expression is evaluated only on rows where cond is true and the else expression where cond is false.

For example, with short-circuit evaluation, no division-by-zero exception is thrown when executing the following query:

SELECT if(number = 0, 0, intDiv(42, number)) FROM numbers(10)

then and else must be of a similar type.

Syntax

if(cond, then, else)

Arguments

• cond — The evaluated condition. UInt8 or Nullable(UInt8) or NULL
• then — The expression returned if cond is true. - else — The expression returned if cond is false or NULL.

Returned value

The result of either the then or else expressions, depending on condition cond.

Examples

Example usage

SELECT if(1, 2 + 2, 2 + 6) AS res;

┌─res─┐
│   4 │
└─────┘

least

Introduced in: v1.1

Returns the smallest value among the arguments.

• For arrays, returns the lexicographically least array.
• For DateTime types, the result type is promoted to the largest type (e.g., DateTime64 if mixed with DateTime32).

Syntax

least(x1[, x2, ..., xN])

Arguments

• x1[, x2, ..., xN] — One or multiple values to compare. All arguments must be of comparable types.

Returned value

Returns the least value among the arguments, promoted to the largest compatible type.

Examples

Numeric types

SELECT least(1, 2, toUInt8(3), 3.) AS result, toTypeName(result) AS type;
-- The type returned is a Float64 as the UInt8 must be promoted to 64 bit for the comparison.

┌─result─┬─type────┐
│      1 │ Float64 │
└────────┴─────────┘

Arrays

SELECT least(['hello'], ['there'], ['world']);

┌─least(['hell⋯ ['world'])─┐
│ ['hello']                │
└──────────────────────────┘

DateTime types

SELECT least(toDateTime32(now() + toIntervalDay(1)), toDateTime64(now(), 3));
-- The type returned is a DateTime64 as the DateTime32 must be promoted to 64 bit for the comparison.

┌─least(toDate⋯(now(), 3))─┐
│  2025-05-27 15:55:20.000 │
└──────────────────────────┘

multiIf

Introduced in: v1.1

Allows writing the CASE operator more compactly in the query. Evaluates each condition in order. For the first condition that is true (non-zero and not NULL), returns the corresponding branch value. If none of the conditions are true, returns the else value.

Setting short_circuit_function_evaluation controls whether short-circuit evaluation is used. If enabled, the then_i expression is evaluated only on rows where ((NOT cond_1) AND ... AND (NOT cond_{i-1}) AND cond_i) is true.

For example, with short-circuit evaluation, no division-by-zero exception is thrown when executing the following query:

SELECT multiIf(number = 2, intDiv(1, number), number = 5) FROM numbers(10)

All branch and else expressions must have a common supertype. NULL conditions are treated as false.

Syntax

multiIf(cond_1, then_1, cond_2, then_2, ..., else)

Arguments

• cond_N — The N-th evaluated condition which controls if then_N is returned. UInt8 or Nullable(UInt8) or NULL
• then_N — The result of the function when cond_N is true. - else — The result of the function if none of the conditions is true.

Returned value

Returns the result of then_N for matching cond_N, otherwise returns the else condition.

Examples

Example usage

CREATE TABLE LEFT_RIGHT (left Nullable(UInt8), right Nullable(UInt8)) ENGINE = Memory;
INSERT INTO LEFT_RIGHT VALUES (NULL, 4), (1, 3), (2, 2), (3, 1), (4, NULL);

SELECT
    left,
    right,
    multiIf(left < right, 'left is smaller', left > right, 'left is greater', left = right, 'Both equal', 'Null value') AS result
FROM LEFT_RIGHT;

┌─left─┬─right─┬─result──────────┐
│ ᴺᵁᴸᴸ │     4 │ Null value      │
│    1 │     3 │ left is smaller │
│    2 │     2 │ Both equal      │
│    3 │     1 │ left is greater │
│    4 │  ᴺᵁᴸᴸ │ Null value      │
└──────┴───────┴─────────────────┘