Operators

Expressions can be composed from function calls and operations, such as 2 + 3 or ((1 + x) * -y). In the example below, note the use of expressions to calculate the alias circumference and in the filter transform.

PRQL

  1. from foo
  2. select {
  3. circumference = diameter * 3.14159,
  4. area = (diameter / 2) ** 2,
  5. color,
  6. }
  7. filter circumference > 10 && color != "red"

SQL

  1. WITH table_0 AS (
  2. SELECT
  3. diameter * 3.14159 AS circumference,
  4. POW(diameter / 2, 2) AS area,
  5. color
  6. FROM
  7. foo
  8. )
  9. SELECT
  10. circumference,
  11. area,
  12. color
  13. FROM
  14. table_0
  15. WHERE
  16. circumference > 10
  17. AND color <> 'red'

Operator precedence

This table shows operator precedence. Use parentheses () to prioritize operations and for function calls (see the discussion below.)

GroupOperatorsPrecedenceAssociativity
parentheses()0see below
identifier dot.1
unary- + ! ==2
range..3
pow*4right-to-left
mul / // %5left-to-right
add+ -6left-to-right
compare== != <= >= < >7left-to-right
coalesce??8left-to-right
and&&9left-to-right
or||10left-to-right
function call11

Division and integer division

The / operator performs division that always returns a float value, while the // operator does integer division (truncated division) that always returns an integer value.

PRQL

  1. prql target:sql.sqlite
  2. from [
  3. {a = 5, b = 2},
  4. {a = 5, b = -2},
  5. ]
  6. select {
  7. div_out = a / b,
  8. int_div_out = a // b,
  9. }

SQL

  1. WITH table_0 AS (
  2. SELECT
  3. 5 AS a,
  4. 2 AS b
  5. UNION
  6. ALL
  7. SELECT
  8. 5 AS a,
  9. -2 AS b
  10. )
  11. SELECT
  12. (a * 1.0 / b) AS div_out,
  13. ROUND(ABS(a / b) - 0.5) * SIGN(a) * SIGN(b) AS int_div_out
  14. FROM
  15. table_0

Coalesce

We can coalesce values with an ?? operator. Coalescing takes either the first value or, if that value is null, the second value.

PRQL

  1. from orders
  2. derive amount ?? 0

SQL

  1. SELECT
  2. *,
  3. COALESCE(amount, 0)
  4. FROM
  5. orders

Regex expressions

Note

This is currently experimental

To perform a case-sensitive regex search, use the ~= operator. This generally compiles to REGEXP, though differs by dialect. A regex search means that to match an exact value, the start and end need to be anchored with ^foo$.

PRQL

  1. from tracks
  2. filter (name ~= "Love")

SQL

  1. SELECT
  2. *
  3. FROM
  4. tracks
  5. WHERE
  6. REGEXP(name, 'Love')

PRQL

  1. prql target:sql.duckdb
  2. from artists
  3. filter (name ~= "Love.*You")

SQL

  1. SELECT
  2. *
  3. FROM
  4. artists
  5. WHERE
  6. REGEXP_MATCHES(name, 'Love.*You')

PRQL

  1. prql target:sql.bigquery
  2. from tracks
  3. filter (name ~= "\\bLove\\b")

SQL

  1. SELECT
  2. *
  3. FROM
  4. tracks
  5. WHERE
  6. REGEXP_CONTAINS(name, '\bLove\b')

PRQL

  1. prql target:sql.postgres
  2. from tracks
  3. filter (name ~= "\\(I Can't Help\\) Falling")

SQL

  1. SELECT
  2. *
  3. FROM
  4. tracks
  5. WHERE
  6. name ~ '\(I Can''t Help\) Falling'

PRQL

  1. prql target:sql.mysql
  2. from tracks
  3. filter (name ~= "With You")

SQL

  1. SELECT
  2. *
  3. FROM
  4. tracks
  5. WHERE
  6. REGEXP_LIKE(name, 'With You', 'c')

PRQL

  1. prql target:sql.sqlite
  2. from tracks
  3. filter (name ~= "But Why Isn't Your Syntax More Similar\\?")

SQL

  1. SELECT
  2. *
  3. FROM
  4. tracks
  5. WHERE
  6. name REGEXP 'But Why Isn''t Your Syntax More Similar\?'

Parentheses

PRQL uses parentheses () for several purposes:

  • Parentheses group operands to control the order of evaluation, for example: ((1 + x) * y)

  • Parentheses delimit a minus sign of a function argument, for example: add (-1) (-3)

  • Parentheses delimit nested function calls that contain a pipe, either the | symbol or a new line. “Nested” means within a transform; i.e. not just the main pipeline, for example: (column-name | in 0..20)

  • Parentheses wrap a function call that is part of a larger expression, for example: math.round 0 (sum distance)

Parentheses are not required for expressions that do not contain function calls, for example: foo + bar.

Here’s a set of examples of these rules:

PRQL

  1. from employees
  2. # Requires parentheses, because it contains a pipe
  3. derive is_proximate = (distance | in 0..20)
  4. # Requires parentheses, because it's a function call
  5. derive total_distance = (sum distance)
  6. # `??` doesn't require parentheses, as it's not a function call
  7. derive min_capped_distance = (min distance ?? 5)
  8. # No parentheses needed, because no function call
  9. derive travel_time = distance / 40
  10. # No inner parentheses needed around `1+1` because no function call
  11. derive distance_rounded_2_dp = (math.round 1+1 distance)
  12. derive {
  13. # Requires parentheses, because it contains a pipe
  14. is_far = (distance | in 100..),
  15. # The left value of the range requires parentheses,
  16. # because of the minus sign
  17. is_negative = (distance | in (-100..0)),
  18. # ...this is equivalent
  19. is_negative = (distance | in (-100)..0),
  20. # _Technically_, this doesn't require parentheses, because it's
  21. # the RHS of an assignment in a tuple
  22. # (this is especially confusing)
  23. average_distance = average distance,
  24. }
  25. # Requires parentheses because of the minus sign
  26. sort (-distance)
  27. # A tuple is fine too
  28. sort {-distance}

SQL

  1. SELECT
  2. *,
  3. distance BETWEEN 0 AND 20 AS is_proximate,
  4. SUM(distance) OVER () AS total_distance,
  5. MIN(COALESCE(distance, 5)) OVER () AS min_capped_distance,
  6. distance / 40 AS travel_time,
  7. ROUND(distance, 1 + 1) AS distance_rounded_2_dp,
  8. distance >= 100 AS is_far,
  9. distance BETWEEN -100 AND 0,
  10. distance BETWEEN -100 AND 0 AS is_negative,
  11. AVG(distance) OVER () AS average_distance
  12. FROM
  13. employees
  14. ORDER BY
  15. distance DESC

For example, the snippet below produces an error because the sum function call is not in a tuple.

PRQL

  1. from employees
  2. derive total_distance = sum distance

Error

  1. Error:
  2. ╭─[:2:29]
  3. 2 derive total_distance = sum distance
  4. ────┬───
  5. ╰───── Unknown name `distance`
  6. ───╯

…while with parentheses, it works at expected:

PRQL

  1. from employees
  2. derive other_distance = (sum distance)

SQL

  1. SELECT
  2. *,
  3. SUM(distance) OVER () AS other_distance
  4. FROM
  5. employees

Note

We’re continuing to think whether these rules can be more intuitive. We’re also planning to make the error messages much better, so the compiler can help out.

Wrapping lines

Line breaks in PRQL have semantic meaning, so to wrap a single logical line into multiple physical lines, we can use \ at the beginning of subsequent physical lines:

PRQL

  1. from artists
  2. select is_europe =
  3. \ country == "DE"
  4. \ || country == "FR"
  5. \ || country == "ES"

SQL

  1. SELECT
  2. country = 'DE'
  3. OR country = 'FR'
  4. OR country = 'ES' AS is_europe
  5. FROM
  6. artists

Wrapping will “jump over” empty lines or lines with comments. For example, the select here is only one logical line:

PRQL

  1. from tracks
  2. # This would be a really long line without being able to split it:
  3. select listening_time_years = (spotify_plays + apple_music_plays + pandora_plays)
  4. # We can toggle between lines when developing:
  5. # \ * length_seconds
  6. \ * length_s
  7. # min hour day year
  8. \ / 60 / 60 / 24 / 365

SQL

  1. SELECT
  2. (
  3. spotify_plays + apple_music_plays + pandora_plays
  4. ) * length_s / 60 / 60 / 24 / 365 AS listening_time_years
  5. FROM
  6. tracks

Info

Note that PRQL differs from most languages, which use a \ at the end of the preceding line. Because PRQL aims to be friendly for data exploration, we want to make it possible to comment out any line, including the final line, without breaking the query. This requires all lines after the first to be structured similarly, and for the character to be at the start of each following line.

See Pipes for more details on line breaks.