Functions

VSCode

Malloy has a standard set of functions available at all times. These include scalar functions (e.g. concat), aggregate functions (e.g. stddev), and analytic (or "window") functions (e.g. lag).

These functions are defined in Malloy in order to provide basic typechecking of arguments and return types, as well as some compatibility transformations. Perfect compatibility across all dialects for all functions is not guaranteed, and specific differences are noted on a case-by-case basis.

For quick function lookup, see the alphebetized table of all available functions.

Note: the behavior of functions changed in v0.0.39. For more information, see a description of that change.

Syntax

Built-In Functions

Functions that are "built-in" to Malloy can be called the usual way with the name of the function, an open parenthesis, a comma separated list of Malloy expressions, and a close parenthesis, e.g. concat(upper(first), ' ', upper(last)).

Raw SQL Functions

Functions that are not built in may be called in the same way, but with an ! in between the function name and argument list: sinh!(x). In this case, Malloy assumes the return type of the function is the same as that of the first argument (or number if there are no arguments). When this is not correct, the return type can be specified after the !, e.g. hash!number(username).

Function Documentation

Section
String Functions
Numeric Functions
Interval Functions
Date and Timestamp Functions
Other Functions
Aggregate Functions
Ungrouped Aggregate Functions
Analytic Functions

String Functions

ascii	byte_length	chr	concat	ends_with
length	lower	ltrim	regexp_extract	repeat
replace	repeat	rtrim	starts_with	strpos
substr	trim	unicode	upper

concat

concat(value, ...)

Concatenates multiple values together, casting non-string values to string. The exact behavior of string casting may depend on the dialect.

If no values are given, concat returns the empty string.

Behavior for null depends on dialect: in BigQuery, if any argument is null, the result is null; in DuckDB and Postgres, null is treated as an empty string.

run: empty -> {
  group_by: 
    e1 is concat('foo', 'bar')
    e2 is concat(1, null)
    e4 is concat('Date: ', @2021-01-23)
    e3 is concat()
}

QUERY RESULTS

[
  {
    "e1": "foobar",
    "e2": "1",
    "e4": "Date: 2021-01-23",
    "e3": ""
  }
]

SELECT 
   CONCAT('foo','bar') as "e1",
   CONCAT(1,NULL) as "e2",
   CONCAT('Date: ',DATE '2021-01-23') as "e4",
   '' as "e3"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

lower

lower(value)

Returns a string like value but with all alphabetic characters in lowercase.

run: empty -> {
  group_by: e1 is lower('FOO')
}

QUERY RESULTS

[
  {
    "e1": "foo"
  }
]

SELECT 
   LOWER('FOO') as "e1"
FROM (
    SELECT 1
  ) as base
GROUP BY 1
ORDER BY 1 asc NULLS LAST

upper

upper(value)

Returns a string like value but with all alphabetic characters in uppercase.

run: empty -> {
  group_by: e1 is upper('foo')
}

QUERY RESULTS

[
  {
    "e1": "FOO"
  }
]

SELECT 
   UPPER('foo') as "e1"
FROM (
    SELECT 1
  ) as base
GROUP BY 1
ORDER BY 1 asc NULLS LAST

strpos

strpos(test_string, search_string)

Returns the 1-based position of the first occurrence of search_string in test_string, or 0 if search_string is not found.

run: empty -> {
  group_by: 
    e1 is strpos('abc', 'b')
    e2 is strpos('abc', 'd')
    e3 is strpos('abc', null)
    e4 is strpos(null, 'a')
}

QUERY RESULTS

[
  {
    "e1": 2,
    "e2": 0,
    "e3": null,
    "e4": null
  }
]

SELECT 
   STRPOS('abc','b') as "e1",
   STRPOS('abc','d') as "e2",
   STRPOS('abc',NULL) as "e3",
   STRPOS(NULL,'a') as "e4"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

starts_with

starts_with(value, prefix)

Returns true if value begins with prefix and false otherwise. If either value or prefix is null, the result is false (unlike in SQL, where it would be null).

run: empty -> {
  group_by: 
    e1 is starts_with('abc', 'a')
    e2 is starts_with('abc', 'b')
    e3 is starts_with('abc', null)
    e4 is starts_with(null, 'a')
}

QUERY RESULTS

[
  {
    "e1": true,
    "e2": false,
    "e3": false,
    "e4": false
  }
]

SELECT 
   COALESCE(STARTS_WITH('abc', 'a'), false) as "e1",
   COALESCE(STARTS_WITH('abc', 'b'), false) as "e2",
   COALESCE(STARTS_WITH('abc', NULL), false) as "e3",
   COALESCE(STARTS_WITH(NULL, 'a'), false) as "e4"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

ends_with

ends_with(value, suffix)

Returns true if value ends with prefix and false otherwise. If either value or suffix is null, the result is false (unlike in SQL, where it would be null).

run: empty -> {
  group_by: 
    e1 is ends_with('abc', 'c')
    e2 is ends_with('abc', 'b')
    e3 is ends_with('abc', null)
    e4 is ends_with(null, 'a')
}

QUERY RESULTS

[
  {
    "e1": true,
    "e2": false,
    "e3": false,
    "e4": false
  }
]

SELECT 
   COALESCE(SUFFIX('abc', 'c'), false) as "e1",
   COALESCE(SUFFIX('abc', 'b'), false) as "e2",
   COALESCE(SUFFIX('abc', NULL), false) as "e3",
   COALESCE(SUFFIX(NULL, 'a'), false) as "e4"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

trim

trim(value)
trim(value, trim_characters)

Returns a string with leading and trailing characters in trim_characters (or whitespace, if trim_characters is unspecified) removed.

run: empty -> {
  group_by: 
    e1 is trim('  keep this  ')
    e2 is trim('_ _keep_this_ _', '_ ')
    e3 is trim(' keep everything ', '')
    e4 is trim('null example', null)
    e5 is trim(null, ' _')
}

QUERY RESULTS

[
  {
    "e1": "keep this",
    "e2": "keep_this",
    "e3": " keep everything ",
    "e4": null,
    "e5": null
  }
]

SELECT 
   TRIM('  keep this  ') as "e1",
   TRIM('_ _keep_this_ _','_ ') as "e2",
   TRIM(' keep everything ','') as "e3",
   TRIM('null example',NULL) as "e4",
   TRIM(NULL,' _') as "e5"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5
ORDER BY 1 asc NULLS LAST

ltrim

ltrim(value)
ltrim(value, trim_characters)

Like trim(value, trim_characters) but only removes leading characters.

run: empty -> {
  group_by: 
    e1 is ltrim('  keep this ->  ')
    e2 is ltrim('_ _keep_this -> _ _', '_ ')
    e3 is ltrim(' keep everything ', '')
    e4 is ltrim('null example', null)
    e5 is ltrim(null, ' _')
}

QUERY RESULTS

[
  {
    "e1": "keep this ->  ",
    "e2": "keep_this -> _ _",
    "e3": " keep everything ",
    "e4": null,
    "e5": null
  }
]

SELECT 
   LTRIM('  keep this ->  ') as "e1",
   LTRIM('_ _keep_this -> _ _','_ ') as "e2",
   LTRIM(' keep everything ','') as "e3",
   LTRIM('null example',NULL) as "e4",
   LTRIM(NULL,' _') as "e5"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5
ORDER BY 1 asc NULLS LAST

rtrim

rtrim(value)
rtrim(value, trim_characters)

Like trim(value, trim_characters) but only removes trailing characters.

run: empty -> {
  group_by: 
    e1 is rtrim('  <- keep this  ')
    e2 is rtrim('_ _ <- keep_this _ _', '_ ')
    e3 is rtrim(' keep everything ', '')
    e4 is rtrim('null example', null)
    e5 is rtrim(null, ' _')
}

QUERY RESULTS

[
  {
    "e1": "  <- keep this",
    "e2": "_ _ <- keep_this",
    "e3": " keep everything ",
    "e4": null,
    "e5": null
  }
]

SELECT 
   RTRIM('  <- keep this  ') as "e1",
   RTRIM('_ _ <- keep_this _ _','_ ') as "e2",
   RTRIM(' keep everything ','') as "e3",
   RTRIM('null example',NULL) as "e4",
   RTRIM(NULL,' _') as "e5"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5
ORDER BY 1 asc NULLS LAST

substr

substr(value, position)
substr(value, position, length)

Returns a substring of value starting at the 1-based index position. If length is specified, the returned string will be at most length characters long; otherwise the returned string will extend to the end of value. Negative values of position index the starting value from the end of the string (with -1 for the last character of the string). A position of 0 is equivalent to a position of 1.

run: empty -> {
  group_by: 
    e1 is substr('123456789', 1)
    e2 is substr('123456789', 0)
    e3 is substr('123456789', 5)
    e4 is substr('123456789', -4)
    e5 is substr('123456789', 1, 3)
    e6 is substr('123456789', -5, 3)
    e7 is substr('123456789', null, 1)
    e8 is substr('123456789', 1, null)
    e9 is substr(null, 1, 1)
}

QUERY RESULTS

[
  {
    "e1": "123456789",
    "e2": "123456789",
    "e3": "56789",
    "e4": "6789",
    "e5": "123",
    "e6": "567",
    "e7": null,
    "e8": null,
    "e9": null
  }
]

SELECT 
   SUBSTR('123456789',1) as "e1",
   SUBSTR('123456789',0) as "e2",
   SUBSTR('123456789',5) as "e3",
   SUBSTR('123456789',-4) as "e4",
   SUBSTR('123456789',1,3) as "e5",
   SUBSTR('123456789',-5,3) as "e6",
   SUBSTR('123456789',NULL,1) as "e7",
   SUBSTR('123456789',1,NULL) as "e8",
   SUBSTR(NULL,1,1) as "e9"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5,6,7,8,9
ORDER BY 1 asc NULLS LAST

regexp_extract

regexp_extract(value, pattern)

Returns the first substring of value which matches pattern. Returns null if there is no matching substring.

run: empty -> {
  group_by: 
    e1 is regexp_extract('the Malloy language', r'M....y')
    e2 is regexp_extract('the SQL language', r'M....y')
    e3 is regexp_extract('the null language', null)
    e4 is regexp_extract(null, r'nothing')
}

QUERY RESULTS

[
  {
    "e1": "Malloy",
    "e2": "",
    "e3": null,
    "e4": null
  }
]

SELECT 
   REGEXP_EXTRACT('the Malloy language','M....y') as "e1",
   REGEXP_EXTRACT('the SQL language','M....y') as "e2",
   REGEXP_EXTRACT('the null language',NULL) as "e3",
   REGEXP_EXTRACT(NULL,'nothing') as "e4"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

replace

replace(value, pattern, replacement)

Returns a copy of value with all occurrences of pattern replaced with replacement.

If pattern is empty, no replacement occurs and value is returned unchanged.

If pattern is a regular expression, parenthesized capturing groups can be included in the replacement string with '\\1' to '\\9'. In BigQuery and DuckDB, the full matched string can be referenced with '\\0'.

run: empty -> {
  group_by: 
    e1 is replace('the SQL language', 'SQL', 'Malloy')
    e2 is replace('the SQL language', r'S.L', 'Malloy')
    e3 is replace('SQL SQL SQL', 'SQL', 'Malloy')
    e4 is replace('SQL SQL SQL', r'S.L', 'Malloy')
    e5 is replace(null, 'SQL', 'Malloy')
    e6 is replace('the null language', null, 'Malloy')
    e7 is replace('the null language', 'SQL', null)
    e8 is replace('the language', '', 'Malloy')
    e9 is replace('axbxc', r'(a).(b).(c)', '\\0 - \\1 - \\2 - \\3')
}

QUERY RESULTS

[
  {
    "e1": "the Malloy language",
    "e2": "the Malloy language",
    "e3": "Malloy Malloy Malloy",
    "e4": "Malloy Malloy Malloy",
    "e5": null,
    "e6": null,
    "e7": null,
    "e8": "the language",
    "e9": "axbxc - a - b - c"
  }
]

SELECT 
   REPLACE('the SQL language','SQL','Malloy') as "e1",
   REGEXP_REPLACE('the SQL language', 'S.L', 'Malloy', 'g') as "e2",
   REPLACE('SQL SQL SQL','SQL','Malloy') as "e3",
   REGEXP_REPLACE('SQL SQL SQL', 'S.L', 'Malloy', 'g') as "e4",
   REPLACE(NULL,'SQL','Malloy') as "e5",
   REPLACE('the null language',NULL,'Malloy') as "e6",
   REPLACE('the null language','SQL',NULL) as "e7",
   REPLACE('the language','','Malloy') as "e8",
   REGEXP_REPLACE('axbxc', '(a).(b).(c)', '\0 - \1 - \2 - \3', 'g') as "e9"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5,6,7,8,9
ORDER BY 1 asc NULLS LAST

length

length(value)

Returns the number of characters in value.

run: empty -> {
  group_by: 
    e1 is length('Malloy')
    e2 is length('')
    e3 is length('𝔐𝔞𝔩𝔩𝔬𝔶')
    e4 is length(null)
}

QUERY RESULTS

[
  {
    "e1": 6,
    "e2": 0,
    "e3": 6,
    "e4": null
  }
]

SELECT 
   LENGTH('Malloy') as "e1",
   LENGTH('') as "e2",
   LENGTH('𝔐𝔞𝔩𝔩𝔬𝔶') as "e3",
   LENGTH(NULL) as "e4"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

byte_length

byte_length(value)

Returns the number of bytes in the unicode encoding of value.

run: empty -> {
  group_by: 
    e1 is byte_length('Malloy')
    e2 is byte_length('')
    e3 is byte_length('𝔐𝔞𝔩𝔩𝔬𝔶')
    e4 is byte_length(null)
}

QUERY RESULTS

[
  {
    "e1": 6,
    "e2": 0,
    "e3": 24,
    "e4": null
  }
]

SELECT 
   (BIT_LENGTH('Malloy') / 8) as "e1",
   (BIT_LENGTH('') / 8) as "e2",
   (BIT_LENGTH('𝔐𝔞𝔩𝔩𝔬𝔶') / 8) as "e3",
   (BIT_LENGTH(NULL) / 8) as "e4"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

chr

chr(value)

Returns a unicode code point value and returns a string containing the character with the matching code point. If value is 0, returns an empty string not a string containing the null character.

run: empty -> {
  group_by: 
    e1 is chr(65)
    e2 is chr(255)
    e3 is chr(0)
    e4 is chr(null)
}

QUERY RESULTS

[
  {
    "e1": "A",
    "e2": "ÿ",
    "e3": "",
    "e4": null
  }
]

SELECT 
   CASE WHEN 65 = 0 THEN '' ELSE CHR(65) END as "e1",
   CASE WHEN 255 = 0 THEN '' ELSE CHR(255) END as "e2",
   CASE WHEN 0 = 0 THEN '' ELSE CHR(0) END as "e3",
   CASE WHEN NULL = 0 THEN '' ELSE CHR(NULL) END as "e4"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

Returns an error if value is not a valid unicode code point.

ascii

ascii(value)

Returns the ASCII code point of the first character of value. If value is empty, returns 0.

In BigQuery, ascii errors if the first character of value is not representable in ASCII. In Postgres and DuckDB, ascii returns the Unicode code point of the first character.

run: empty -> {
  group_by: 
    e1 is ascii('Malloy')
    e2 is ascii('')
    e3 is ascii(null)
}

QUERY RESULTS

[
  {
    "e1": 77,
    "e2": 0,
    "e3": null
  }
]

SELECT 
   ASCII('Malloy') as "e1",
   ASCII('') as "e2",
   ASCII(NULL) as "e3"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3
ORDER BY 1 asc NULLS LAST

unicode

unicode(value)

Returns the Unicode code point of the first character of value. If value is empty, returns 0.

run: empty -> {
  group_by: 
    e1 is unicode('Malloy')
    e2 is unicode('𝔐𝔞𝔩𝔩𝔬𝔶')
    e3 is unicode('')
    e4 is unicode(null)
}

QUERY RESULTS

[
  {
    "e1": 77,
    "e2": 120080,
    "e3": 0,
    "e4": null
  }
]

SELECT 
   CASE WHEN 'Malloy' = '' THEN 0 ELSE UNICODE('Malloy') END as "e1",
   CASE WHEN '𝔐𝔞𝔩𝔩𝔬𝔶' = '' THEN 0 ELSE UNICODE('𝔐𝔞𝔩𝔩𝔬𝔶') END as "e2",
   CASE WHEN '' = '' THEN 0 ELSE UNICODE('') END as "e3",
   CASE WHEN NULL = '' THEN 0 ELSE UNICODE(NULL) END as "e4"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

repeat

repeat(value, num_repetitions)

Return a string consisting of value repeated num_repetitions times.

Undefined behavior if num_repetitions is negative or a non-integer.

run: empty -> {
  group_by: 
    e1 is repeat('(A)', 3)
    e2 is repeat('(A)', 1)
    e3 is repeat('(A)', 0)
    e4 is repeat('', 3)
    e5 is repeat(null, 1)
    e6 is repeat('(A)', null)
}

QUERY RESULTS

[
  {
    "e1": "(A)(A)(A)",
    "e2": "(A)",
    "e3": "",
    "e4": "",
    "e5": null,
    "e6": null
  }
]

SELECT 
   REPEAT('(A)',3) as "e1",
   REPEAT('(A)',1) as "e2",
   REPEAT('(A)',0) as "e3",
   REPEAT('',3) as "e4",
   REPEAT(NULL,1) as "e5",
   REPEAT('(A)',NULL) as "e6"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5,6
ORDER BY 1 asc NULLS LAST

reverse

reverse(value)

Return a copy of value with the characters in reverse order.

run: empty -> {
  group_by: 
    e1 is reverse('Malloy')
    e2 is reverse('𝔶𝔬𝔩𝔩𝔞𝔐')
    e3 is reverse('racecar')
    e4 is reverse('')
    e5 is reverse(null)
}

QUERY RESULTS

[
  {
    "e1": "yollaM",
    "e2": "𝔐𝔞𝔩𝔩𝔬𝔶",
    "e3": "racecar",
    "e4": "",
    "e5": null
  }
]

SELECT 
   REVERSE('Malloy') as "e1",
   REVERSE('𝔶𝔬𝔩𝔩𝔞𝔐') as "e2",
   REVERSE('racecar') as "e3",
   REVERSE('') as "e4",
   REVERSE(NULL) as "e5"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5
ORDER BY 1 asc NULLS LAST

Numeric Functions

abs	acos	asin	atan	atan2	ceil
cos	div	exp	floor	is_inf	is_nan
log	ln	pi	pow	rand	round
sign	sin	sqrt	tan	trunc

round

round(value)
round(value, precision)

Round value to the nearest integer, or if precision is specified, round to precision decimal places right of the decimal. If precision is negative, round to a precision that many places to the left of the decimal. Halfway cases are rounded away from zero.

run: empty -> {
  group_by: 
    e1 is round(1.49)
    e2 is round(1.51)
    e3 is round(1.5)
    e4 is round(-1.5)
    e5 is round(1.5, 0)
    e6 is round(1.551, 1)
    e7 is round(14.12, -1)
    e8 is round(1.4, null)
    e9 is round(null, 1)
}

QUERY RESULTS

[
  {
    "e1": 1,
    "e2": 2,
    "e3": 2,
    "e4": -2,
    "e5": 2,
    "e6": 1.6,
    "e7": 10,
    "e8": null,
    "e9": null
  }
]

SELECT 
   ROUND(1.49::DOUBLE) as "e1",
   ROUND(1.51::DOUBLE) as "e2",
   ROUND(1.5::DOUBLE) as "e3",
   ROUND(-1.5::DOUBLE) as "e4",
   ROUND(1.5::DOUBLE,0) as "e5",
   ROUND(1.551::DOUBLE,1) as "e6",
   ROUND(14.12::DOUBLE,-1) as "e7",
   ROUND(1.4::DOUBLE,NULL) as "e8",
   ROUND(NULL,1) as "e9"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5,6,7,8,9
ORDER BY 1 asc NULLS LAST

trunc

trunc(value)
trunc(value, precision)

Truncate value to an integer, or if precision is specified, truncate after that many decimal places. If precision is negative, truncate to that many decimal places to the left of the decimal point. Similar to round(value, precision) but always rounds toward zero.

run: empty -> {
  group_by: 
    e1 is trunc(1.49)
    e2 is trunc(1.51)
    e3 is trunc(1.5)
    e4 is trunc(-1.5)
    e5 is trunc(1.5, 0)
    e6 is trunc(1.551, 1)
    e7 is trunc(19.999, -1)
    e8 is trunc(1.4, null)
    e9 is trunc(null, 1)
}

QUERY RESULTS

[
  {
    "e1": 1,
    "e2": 1,
    "e3": 1,
    "e4": -1,
    "e5": 1,
    "e6": 1.5,
    "e7": 10,
    "e8": null,
    "e9": null
  }
]

SELECT 
   CASE WHEN 1.49::DOUBLE < 0 THEN CEIL(1.49::DOUBLE) ELSE FLOOR(1.49::DOUBLE) END as "e1",
   CASE WHEN 1.51::DOUBLE < 0 THEN CEIL(1.51::DOUBLE) ELSE FLOOR(1.51::DOUBLE) END as "e2",
   CASE WHEN 1.5::DOUBLE < 0 THEN CEIL(1.5::DOUBLE) ELSE FLOOR(1.5::DOUBLE) END as "e3",
   CASE WHEN -1.5::DOUBLE < 0 THEN CEIL(-1.5::DOUBLE) ELSE FLOOR(-1.5::DOUBLE) END as "e4",
   CASE WHEN 1.5::DOUBLE < 0 THEN CEIL(1.5::DOUBLE * POW(10, 0)) / POW(10, 0) ELSE FLOOR(1.5::DOUBLE * POW(10, 0)) / POW(10, 0) END as "e5",
   CASE WHEN 1.551::DOUBLE < 0 THEN CEIL(1.551::DOUBLE * POW(10, 1)) / POW(10, 1) ELSE FLOOR(1.551::DOUBLE * POW(10, 1)) / POW(10, 1) END as "e6",
   CASE WHEN 19.999::DOUBLE < 0 THEN CEIL(19.999::DOUBLE * POW(10, -1)) / POW(10, -1) ELSE FLOOR(19.999::DOUBLE * POW(10, -1)) / POW(10, -1) END as "e7",
   CASE WHEN 1.4::DOUBLE < 0 THEN CEIL(1.4::DOUBLE * POW(10, NULL)) / POW(10, NULL) ELSE FLOOR(1.4::DOUBLE * POW(10, NULL)) / POW(10, NULL) END as "e8",
   CASE WHEN NULL < 0 THEN CEIL(NULL * POW(10, 1)) / POW(10, 1) ELSE FLOOR(NULL * POW(10, 1)) / POW(10, 1) END as "e9"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5,6,7,8,9
ORDER BY 1 asc NULLS LAST

floor

floor(value)

Round value down to the greatest integer not larger than value.

run: empty -> {
  group_by: 
    a is floor(1.0)
    b is floor(1.9)
    c is floor(-1.1)
    d is floor(null)
}

QUERY RESULTS

[
  {
    "a": 1,
    "b": 1,
    "c": -2,
    "d": null
  }
]

SELECT 
   FLOOR(1.0::DOUBLE) as "a",
   FLOOR(1.9::DOUBLE) as "b",
   FLOOR(-1.1::DOUBLE) as "c",
   FLOOR(NULL) as "d"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

ceil

ceil(value)

Round value up to the smallest integer not less than value.

run: empty -> {
  group_by: 
    a is ceil(1.0)
    b is ceil(1.1)
    c is ceil(-1.9)
    d is ceil(null)
}

QUERY RESULTS

[
  {
    "a": 1,
    "b": 2,
    "c": -1,
    "d": null
  }
]

SELECT 
   CEIL(1.0::DOUBLE) as "a",
   CEIL(1.1::DOUBLE) as "b",
   CEIL(-1.9::DOUBLE) as "c",
   CEIL(NULL) as "d"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

cos

cos(angle)

Computes the cosine of angle where angle is specified in radians.

run: empty -> {
  group_by: 
    a is cos(0)
    b is cos(pi())
    c is cos(pi() / 2)
    d is cos(null)
}

QUERY RESULTS

[
  {
    "a": 1,
    "b": -1,
    "c": 6.123233995736766e-17,
    "d": null
  }
]

SELECT 
   COS(0) as "a",
   COS(PI()) as "b",
   COS((PI())*1.0/2) as "c",
   COS(NULL) as "d"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

acos

acos(value)

Computes the principal value of the inverse cosine of value. The return value is in the range [0,π]. Generates an error if value is a value outside of the range [-1, 1].

run: empty -> {
  group_by: 
    a is acos(0)
    b is acos(1)
    c is acos(-1)
    d is acos(null)
}

QUERY RESULTS

[
  {
    "a": 1.5707963267948966,
    "b": 0,
    "c": 3.141592653589793,
    "d": null
  }
]

SELECT 
   ACOS(0) as "a",
   ACOS(1) as "b",
   ACOS(-1) as "c",
   ACOS(NULL) as "d"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

sin

sin(angle)

Computes the sine of angle where angle is specified in radians.

run: empty -> {
  group_by: 
    a is sin(0)
    b is sin(pi())
    c is sin(pi() / 2)
    d is sin(null)
}

QUERY RESULTS

[
  {
    "a": 0,
    "b": 1.2246467991473532e-16,
    "c": 1,
    "d": null
  }
]

SELECT 
   SIN(0) as "a",
   SIN(PI()) as "b",
   SIN((PI())*1.0/2) as "c",
   SIN(NULL) as "d"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

asin

asin(value)

Computes the principal value of the inverse sine of value. The return value is in the range [-π/2,π/2]. Generates an error if value is outside of the range [-1, 1].

run: empty -> {
  group_by: 
    a is asin(0)
    b is asin(1)
    c is asin(-1)
    d is asin(null)
}

QUERY RESULTS

[
  {
    "a": 0,
    "b": 1.5707963267948966,
    "c": -1.5707963267948966,
    "d": null
  }
]

SELECT 
   ASIN(0) as "a",
   ASIN(1) as "b",
   ASIN(-1) as "c",
   ASIN(NULL) as "d"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

tan

tan(angle)

Computes the tangent of angle where angle is specified in radians.

run: empty -> {
  group_by: 
    a is tan(0)
    b is tan(pi())
    c is tan(pi() / 4)
    d is tan(3 * pi() / 4)
    e is tan(null)
}

QUERY RESULTS

[
  {
    "a": 0,
    "b": -1.2246467991473532e-16,
    "c": 0.9999999999999999,
    "d": -1.0000000000000002,
    "e": null
  }
]

SELECT 
   TAN(0) as "a",
   TAN(PI()) as "b",
   TAN((PI())*1.0/4) as "c",
   TAN((3*(PI()))*1.0/4) as "d",
   TAN(NULL) as "e"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5
ORDER BY 1 asc NULLS LAST

atan

atan(value)

Computes the principal value of the inverse tangent of value. The return value is in the range [-π/2,π/2].

run: empty -> {
  group_by: 
    a is atan(0)
    b is atan(1)
    c is atan(-1)
    d is atan(null)
}

QUERY RESULTS

[
  {
    "a": 0,
    "b": 0.7853981633974483,
    "c": -0.7853981633974483,
    "d": null
  }
]

SELECT 
   ATAN(0) as "a",
   ATAN(1) as "b",
   ATAN(-1) as "c",
   ATAN(NULL) as "d"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

atan2

atan2(y, x)

Calculates the principal value of the inverse tangent of x / y using the signs of the two arguments to determine the quadrant. The return value is in the range [-π,π].

run: empty -> {
  group_by: 
    a is atan2(1, 0)
    b is atan2(0, 1)
    c is atan2(-1, 0)
    d is atan2(0, -1)
    e is atan2(null, 1)
    f is atan2(1, null)
}

QUERY RESULTS

[
  {
    "a": 1.5707963267948966,
    "b": 0,
    "c": -1.5707963267948966,
    "d": 3.141592653589793,
    "e": null,
    "f": null
  }
]

SELECT 
   ATAN2(1,0) as "a",
   ATAN2(0,1) as "b",
   ATAN2(-1,0) as "c",
   ATAN2(0,-1) as "d",
   ATAN2(NULL,1) as "e",
   ATAN2(1,NULL) as "f"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5,6
ORDER BY 1 asc NULLS LAST

sqrt

sqrt(value)

Computes the square root of value. Generates an error if value < 0.

run: empty -> {
  group_by: 
    a is sqrt(9)
    b is sqrt(0)
    c is sqrt(null)
}

QUERY RESULTS

[
  {
    "a": 3,
    "b": 0,
    "c": null
  }
]

SELECT 
   SQRT(9) as "a",
   SQRT(0) as "b",
   SQRT(NULL) as "c"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3
ORDER BY 1 asc NULLS LAST

pow

pow(base, exponent)

Returns base raised to the power of exponent.

run: empty -> {
  group_by: 
    a is pow(2, 3)
    b is pow(100, 0)
    c is pow(10, -1)
    d is pow(0, 10)
    e is pow(null, 1)
    f is pow(1, null)
}

QUERY RESULTS

[
  {
    "a": 8,
    "b": 1,
    "c": 0.1,
    "d": 0,
    "e": null,
    "f": null
  }
]

SELECT 
   POW(2,3) as "a",
   POW(100,0) as "b",
   POW(10,-1) as "c",
   POW(0,10) as "d",
   POW(NULL,1) as "e",
   POW(1,NULL) as "f"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5,6
ORDER BY 1 asc NULLS LAST

Generates an error if base is 0 and exponent < 0.

abs

abs(value)

Returns the absolute value of value.

run: empty -> {
  group_by: 
    a is abs(100)
    b is abs(-100)
    c is abs(0)
    d is abs(null)
}

QUERY RESULTS

[
  {
    "a": 100,
    "b": 100,
    "c": 0,
    "d": null
  }
]

SELECT 
   ABS(100) as "a",
   ABS(-100) as "b",
   ABS(0) as "c",
   ABS(NULL) as "d"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

sign

sign(value)

Returns -1, 0, or 1 for negative, zero and positive arguments respectively. For floating point arguments, this function does not distinguish between positive and negative zero.

run: empty -> {
  group_by: 
    a is sign(100)
    b is sign(-100)
    c is sign(0)
    d is sign(null)
}

QUERY RESULTS

[
  {
    "a": 1,
    "b": -1,
    "c": 0,
    "d": null
  }
]

SELECT 
   SIGN(100) as "a",
   SIGN(-100) as "b",
   SIGN(0) as "c",
   SIGN(NULL) as "d"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

is_inf

is_inf(value)

Returns true if value is infinite (positive or negative), false otherwise. Unlike in SQL, is_inf(null) = false (not null).

run: empty -> {
  group_by: 
    a is is_inf(100)
    b is is_inf('inf'::number)
    c is is_inf('-inf'::number)
    d is is_inf('NaN'::number)
    e is is_inf(null)
}

QUERY RESULTS

[
  {
    "a": false,
    "b": true,
    "c": true,
    "d": false,
    "e": false
  }
]

SELECT 
   COALESCE(ISINF(100), false) as "a",
   COALESCE(ISINF(CAST('inf' AS double precision)), false) as "b",
   COALESCE(ISINF(CAST('-inf' AS double precision)), false) as "c",
   COALESCE(ISINF(CAST('NaN' AS double precision)), false) as "d",
   COALESCE(ISINF(NULL), false) as "e"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5
ORDER BY 1 asc NULLS LAST

is_nan

is_nan(value)

Returns true if value is NaN, false otherwise. Unlike in SQL, is_nan(null) = false (not null).

run: empty -> {
  group_by: 
    a is is_nan(100)
    b is is_nan('NaN'::number)
    c is is_nan('inf'::number)
    d is is_nan(null)
}

QUERY RESULTS

[
  {
    "a": false,
    "b": true,
    "c": false,
    "d": false
  }
]

SELECT 
   COALESCE(ISNAN(100), false) as "a",
   COALESCE(ISNAN(CAST('NaN' AS double precision)), false) as "b",
   COALESCE(ISNAN(CAST('inf' AS double precision)), false) as "c",
   COALESCE(ISNAN(NULL), false) as "d"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

div

div(dividend, divisor)

Returns the (truncated) integer part of the division of dividend by divisor. Division by zero returns an error.

run: empty -> {
  group_by: 
    a is div(9, 2)
    a2 is trunc(9 / 2)
    b is div(-9, 2)
    c is div(10, 1)
    d is div(null, 1)
    e is div(1, null)
}

QUERY RESULTS

[
  {
    "a": 4,
    "a2": 4,
    "b": -4,
    "c": 10,
    "d": null,
    "e": null
  }
]

SELECT 
   CASE WHEN 9 / 2 < 0 THEN CEIL(9 / 2) ELSE FLOOR(9 / 2) END as "a",
   CASE WHEN 9*1.0/2 < 0 THEN CEIL(9*1.0/2) ELSE FLOOR(9*1.0/2) END as "a2",
   CASE WHEN -9 / 2 < 0 THEN CEIL(-9 / 2) ELSE FLOOR(-9 / 2) END as "b",
   CASE WHEN 10 / 1 < 0 THEN CEIL(10 / 1) ELSE FLOOR(10 / 1) END as "c",
   CASE WHEN NULL / 1 < 0 THEN CEIL(NULL / 1) ELSE FLOOR(NULL / 1) END as "d",
   CASE WHEN 1 / NULL < 0 THEN CEIL(1 / NULL) ELSE FLOOR(1 / NULL) END as "e"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5,6
ORDER BY 1 asc NULLS LAST

rand

rand()

Returns a random number in the range [0, 1).

run: empty -> {
  group_by: 
    a is rand()
    b is rand()
    c is rand()
}

QUERY RESULTS

[
  {
    "a": 0.7317041184287518,
    "b": 0.7137041476089507,
    "c": 0.49921956192702055
  }
]

SELECT 
   RANDOM() as "a",
   RANDOM() as "b",
   RANDOM() as "c"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3
ORDER BY 1 asc NULLS LAST

pi

pi()

Returns the value of π.

run: empty -> {
  group_by: pi is pi()
}

QUERY RESULTS

[
  {
    "pi": 3.141592653589793
  }
]

SELECT 
   PI() as "pi"
FROM (
    SELECT 1
  ) as base
GROUP BY 1
ORDER BY 1 asc NULLS LAST

log

log(value, base)

Returns the logarithm of value with the specified base. Note that base is required. Errors if value <= 0, base <= 0, or if base = 1.

run: empty -> {
  group_by: 
    a is log(10, 10)
    b is log(100, 10)
    c is log(32, 2)
    d is log(null, 2)
    e is log(10, null)
}

QUERY RESULTS

[
  {
    "a": 1,
    "b": 2,
    "c": 5,
    "d": null,
    "e": null
  }
]

SELECT 
   (LN(10) / LN(10)) as "a",
   (LN(100) / LN(10)) as "b",
   (LN(32) / LN(2)) as "c",
   (LN(NULL) / LN(2)) as "d",
   (LN(10) / LN(NULL)) as "e"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5
ORDER BY 1 asc NULLS LAST

ln

ln(value)

Returns the natural log (log base e) of value. Equivalent to log(value, exp(1)). Errors if value <= 0.

run: empty -> {
  group_by: 
    a is ln(exp(1))
    b is ln(exp(2))
    c is ln(100)
    d is ln(null)
}

QUERY RESULTS

[
  {
    "a": 1,
    "b": 2,
    "c": 4.605170185988092,
    "d": null
  }
]

SELECT 
   LN(EXP(1)) as "a",
   LN(EXP(2)) as "b",
   LN(100) as "c",
   LN(NULL) as "d"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

exp

exp(power)

Returns e (Euler's number) raised to the given power.

run: empty -> {
  group_by: 
    a is exp(1)
    b is exp(null)
}

QUERY RESULTS

[
  {
    "a": 2.718281828459045,
    "b": null
  }
]

SELECT 
   EXP(1) as "a",
   EXP(NULL) as "b"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2
ORDER BY 1 asc NULLS LAST

Interval Functions

days	hours
minutes	months
seconds	quarters
weeks	years

seconds

seconds(interval)

Calculate the number of seconds in interval.

minutes

minutes(interval)

Calculate the number of minutes in interval.

hours

hours(interval)

Calculate the number of hours in interval.

days

days(interval)

Calculate the number of days in interval.

weeks

weeks(interval)

Calculate the number of weeks in interval. Note: this function does not currently work.

months

months(interval)

Calculate the number of months in interval. Note: this function does not currently work.

quarters

quarters(interval)

Calculate the number of quarters in interval. Note: this function does not currently work.

years

years(interval)

Calculate the number of years in interval. Note: this function does not currently work.

run: empty -> {
  group_by: 
    a is seconds(now to now + 1 second)
    b is minutes(now to now + 1 minute)
    c is hours(now to now + 1 hour)
    d is days(now to now + 1 day)
    // e is weeks(now::date to now::date + 1 week)
    // f is months(now::date to now::date + 1 month)
    // g is quarters(now::date to now::date + 1 quarter)
    // h is years(now::date to now::date + 1 year)
}

QUERY RESULTS

[
  {
    "a": 1,
    "b": 1,
    "c": 1,
    "d": 1
  }
]

SELECT 
   DATE_SUB('second', LOCALTIMESTAMP, (LOCALTIMESTAMP + INTERVAL (1) second)) as "a",
   DATE_SUB('minute', LOCALTIMESTAMP, (LOCALTIMESTAMP + INTERVAL (1) minute)) as "b",
   DATE_SUB('hour', LOCALTIMESTAMP, (LOCALTIMESTAMP + INTERVAL (1) hour)) as "c",
   DATE_SUB('day', LOCALTIMESTAMP, (LOCALTIMESTAMP + INTERVAL (1) day)) as "d"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4
ORDER BY 1 asc NULLS LAST

Date and Timestamp Functions

day	day_of_week	day_of_year
hour	minute	month
quarter	second	week
year

day_of_year

day_of_year(moment)

Returns the day of the year (from 1 to 365) of moment.

day

day(moment)

Returns the day of the month (from 1 to 31) of moment.

day_of_week

day_of_week(moment)

Returns the day of the week (from 1 to 7) of moment, where 1 represents Sunday.

week

week(moment)

Returns the week of the year (from 1 to 53) of moment.

month

month(moment)

Returns the month of the year (from 1 to 12) of moment.

quarter

quarter(moment)

Returns the quarter of the year (from 1 to 53) of moment.

year

year(moment)

Returns the year of moment.

hour

hour(time)

Returns the hour of the day (from 0 to 23) of time.

minute

minute(time)

Returns the minute of the hour (from 0 to 59) of time.

second

second(time)

Returns the second of the minute (from 0 to 59) of time.

run: empty extend { dimension: t is @2023-12-31 23:59:59 } -> {
  group_by: 
    doy is day_of_year(t)
    dom is day(t)
    dow is day_of_week(t)
    woy is week(t)
    moy is month(t)
    qoy is quarter(t)
    yyy is year(t)
    hod is hour(t)
    moh is minute(t)
    som is second(t)
}

QUERY RESULTS

[
  {
    "doy": 365,
    "dom": 31,
    "dow": 1,
    "woy": 52,
    "moy": 12,
    "qoy": 4,
    "yyy": 2023,
    "hod": 23,
    "moh": 59,
    "som": 59
  }
]

SELECT 
   EXTRACT(doy FROM (TIMESTAMP '2023-12-31 23:59:59')) as "doy",
   EXTRACT(day FROM (TIMESTAMP '2023-12-31 23:59:59')) as "dom",
   mod(EXTRACT(dow FROM (TIMESTAMP '2023-12-31 23:59:59'))+1,7) as "dow",
   EXTRACT(week FROM (TIMESTAMP '2023-12-31 23:59:59')) as "woy",
   EXTRACT(month FROM (TIMESTAMP '2023-12-31 23:59:59')) as "moy",
   EXTRACT(quarter FROM (TIMESTAMP '2023-12-31 23:59:59')) as "qoy",
   EXTRACT(year FROM (TIMESTAMP '2023-12-31 23:59:59')) as "yyy",
   EXTRACT(hour FROM (TIMESTAMP '2023-12-31 23:59:59')) as "hod",
   EXTRACT(minute FROM (TIMESTAMP '2023-12-31 23:59:59')) as "moh",
   EXTRACT(second FROM (TIMESTAMP '2023-12-31 23:59:59')) as "som"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5,6,7,8,9,10
ORDER BY 1 asc NULLS LAST

Other Functions

greatest	least
ifnull
nullif
coalesce

greatest

greatest(value, ...)

Returns the greatest value of all arguments, supporting, number, string, date, and timestamp. Returns null if any argument is null.

run: empty -> {
  group_by: 
    a is greatest(3, 1, 8)
    b is greatest('z', 'b', 'f')
    c is greatest(@2003-11-03, @2001-10-21)
    d is greatest(@2003-11-03 11:25, @2003-11-03 11:24)
    e is greatest(100, 99, null)
}

QUERY RESULTS

[
  {
    "a": 8,
    "b": "z",
    "c": "2003-11-03T00:00:00.000Z",
    "d": "2003-11-03T11:25:00.000Z",
    "e": null
  }
]

SELECT 
   CASE WHEN LEN(LIST_FILTER([3,1,8], x -> x is null)) > 0 THEN NULL ELSE GREATEST(3,1,8) END as "a",
   CASE WHEN LEN(LIST_FILTER(['z','b','f'], x -> x is null)) > 0 THEN NULL ELSE GREATEST('z','b','f') END as "b",
   CASE WHEN LEN(LIST_FILTER([DATE '2003-11-03',DATE '2001-10-21'], x -> x is null)) > 0 THEN NULL ELSE GREATEST(DATE '2003-11-03',DATE '2001-10-21') END as "c",
   CASE WHEN LEN(LIST_FILTER([TIMESTAMP '2003-11-03 11:25:00',TIMESTAMP '2003-11-03 11:24:00'], x -> x is null)) > 0 THEN NULL ELSE GREATEST(TIMESTAMP '2003-11-03 11:25:00',TIMESTAMP '2003-11-03 11:24:00') END as "d",
   CASE WHEN LEN(LIST_FILTER([100,99,NULL], x -> x is null)) > 0 THEN NULL ELSE GREATEST(100,99,NULL) END as "e"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5
ORDER BY 3 desc NULLS LAST

least

least(value, ...)

Returns the least value of all arguments, supporting, number, string, date, and timestamp. Returns null if any argument is null.

run: empty -> {
  group_by: 
    a is least(3, 1, 8)
    b is least('z', 'b', 'f')
    c is least(@2003-11-03, @2001-10-21)
    d is least(@2003-11-03 11:25, @2003-11-03 11:24)
    e is least(100, 99, null)
}

QUERY RESULTS

[
  {
    "a": 1,
    "b": "b",
    "c": "2001-10-21T00:00:00.000Z",
    "d": "2003-11-03T11:24:00.000Z",
    "e": null
  }
]

SELECT 
   CASE WHEN LEN(LIST_FILTER([3,1,8], x -> x is null)) > 0 THEN NULL ELSE LEAST(3,1,8) END as "a",
   CASE WHEN LEN(LIST_FILTER(['z','b','f'], x -> x is null)) > 0 THEN NULL ELSE LEAST('z','b','f') END as "b",
   CASE WHEN LEN(LIST_FILTER([DATE '2003-11-03',DATE '2001-10-21'], x -> x is null)) > 0 THEN NULL ELSE LEAST(DATE '2003-11-03',DATE '2001-10-21') END as "c",
   CASE WHEN LEN(LIST_FILTER([TIMESTAMP '2003-11-03 11:25:00',TIMESTAMP '2003-11-03 11:24:00'], x -> x is null)) > 0 THEN NULL ELSE LEAST(TIMESTAMP '2003-11-03 11:25:00',TIMESTAMP '2003-11-03 11:24:00') END as "d",
   CASE WHEN LEN(LIST_FILTER([100,99,NULL], x -> x is null)) > 0 THEN NULL ELSE LEAST(100,99,NULL) END as "e"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3,4,5
ORDER BY 3 desc NULLS LAST

ifnull

ifnull(value, default)

Return value unless it is null, or default otherwise.

Note: it is more idiomatic in Malloy to use ??, the null-coalescing operator.

run: empty -> {
  group_by: 
    a_sql is ifnull(null, 1)
    a_malloy is null ?? 1     
}

QUERY RESULTS

[
  {
    "a_sql": 1,
    "a_malloy": 1
  }
]

SELECT 
   IFNULL(NULL,1) as "a_sql",
   1 as "a_malloy"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2
ORDER BY 1 asc NULLS LAST

coalesce

ifnull(value, ...)

Return the first value which is not null, or null if all values are null. Essentially the same as ifnull but allowing more than two arguments.

Note: it is more idiomatic in Malloy to use ??, the null-coalescing operator.

run: empty -> {
  group_by: 
    a_sql is coalesce(null, 1, 2)
    a_malloy is null ?? 1 ?? 2   
}

QUERY RESULTS

[
  {
    "a_sql": 1,
    "a_malloy": 1
  }
]

SELECT 
   COALESCE(NULL,1,2) as "a_sql",
   COALESCE(1,2) as "a_malloy"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2
ORDER BY 1 asc NULLS LAST

nullif

nullif(value, condition)

Equivalent to the SQL NULLIF function: returns value unless value = condition, in which case it returns null.

Note: the use of nullif is not idiomatic to Malloy; use pick statements instead, as they are more flexible.

run: empty extend { dimension: value is 1 } -> {
  group_by: 
    a_sql is nullif(value, 2)
    a_malloy is value ? pick null when 2
    b_malloy is value ? pick null when < 2
}

QUERY RESULTS

[
  {
    "a_sql": 1,
    "a_malloy": 1,
    "b_malloy": null
  }
]

SELECT 
   NULLIF((1),2) as "a_sql",
   CASE WHEN ((1)=2) THEN NULL ELSE (1) END as "a_malloy",
   CASE WHEN ((1)<2) THEN NULL ELSE (1) END as "b_malloy"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2,3
ORDER BY 1 asc NULLS LAST

There are many functions which are useful in SQL which are not part of Malloy for a variety of reasons including a) there is a different way to do it in Malloy, b) it is not generalizable to other dialects, c) it just hasn't been added yet. Some of the more common ones are listed here.

IFNULL, COALESCE

Neither IFNULL nor COALESCE is available in Malloy, because Malloy has a specific coalescing operator: ??.

run: empty -> {
  group_by: 
    a is null ?? 1      // same as IFNULL(NULL, 1)
    b is null ?? 1 ?? 2 // same as COALESCE(NULL, 1, 2)
}

QUERY RESULTS

[
  {
    "a": 1,
    "b": 1
  }
]

SELECT 
   1 as "a",
   COALESCE(1,2) as "b"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2
ORDER BY 1 asc NULLS LAST

NULLIF

The NULLIF SQL function is not available in Malloy, because the same thing is expressible in a more Malloy-like way, using a pick statement. This is also more flexible, as the pick statement allows for partial comparisons.

run: empty extend { dimension: value is 1 } -> {
  group_by: 
    a is value ? pick null when 2   // same as NULLIF(value, 2)
    b is value ? pick null when < 2 // same as CASE NULL WHEN value < 2 ELSE value END
}

QUERY RESULTS

[
  {
    "a": 1,
    "b": null
  }
]

SELECT 
   CASE WHEN ((1)=2) THEN NULL ELSE (1) END as "a",
   CASE WHEN ((1)<2) THEN NULL ELSE (1) END as "b"
FROM (
    SELECT 1
  ) as base
GROUP BY 1,2
ORDER BY 1 asc NULLS LAST

Aggregate Functions

Malloy has several "native" aggregate functions, sum, avg, max, min, and count, as well as "non-native" aggregate functions, which currently only includes stddev. All of these support aggregate locality and symmetric aggregate handling.

Note: Aggregate locality is currently not supported in BigQuery for stddev.

avg	count
max	min
stddev	sum

stddev

stddev(value)

Returns the standard deviation of values of value across all rows.

run: flights -> {
  group_by: carrier
  aggregate: dep_delay_avg is avg(dep_delay)
  aggregate: dep_delay_stddev is stddev(dep_delay)
  order_by: dep_delay_avg asc
}

QUERY RESULTS

[
  {
    "carrier": "TZ",
    "dep_delay_avg": 4.366963402571711,
    "dep_delay_stddev": 31.293795536719024
  },
  {
    "carrier": "RU",
    "dep_delay_avg": 5.342229687694413,
    "dep_delay_stddev": 30.35020135821442
  },
  {
    "carrier": "MQ",
    "dep_delay_avg": 5.438716995399836,
    "dep_delay_stddev": 28.64584040820042
  },
  {
    "carrier": "CO",
    "dep_delay_avg": 6.0128869589578375,
    "dep_delay_stddev": 30.242701868501392
  },
  {
    "carrier": "NW",
    "dep_delay_avg": 6.116289458010721,
    "dep_delay_stddev": 30.586134538485435
  }
]

SELECT 
   base."carrier" as "carrier",
   AVG(base."dep_delay") as "dep_delay_avg",
   STDDEV(base."dep_delay") as "dep_delay_stddev"
FROM '../data/flights.parquet' as base
GROUP BY 1
ORDER BY 2 asc NULLS LAST

Ungrouped Aggregate Functions

Malloy has two ungrouped aggregate functions, all(), and exclude(), which allow you to control which dimensional values are included in the grouping while calculating a particular aggregate expression.

all

exclude

Window Functions

Syntax

Window functions can only appear in a calculate: statement:

run: flights -> {
  group_by: carrier
  calculate: prev_carrier is lag(carrier)
}

QUERY RESULTS

[
  {
    "carrier": "AA",
    "prev_carrier": null
  },
  {
    "carrier": "AS",
    "prev_carrier": "AA"
  },
  {
    "carrier": "B6",
    "prev_carrier": "AS"
  },
  {
    "carrier": "CO",
    "prev_carrier": "B6"
  },
  {
    "carrier": "DL",
    "prev_carrier": "CO"
  }
]

SELECT 
   base."carrier" as "carrier",
   LAG((base."carrier")) OVER(  ORDER BY  base."carrier" asc NULLS LAST ) as "prev_carrier"
FROM '../data/flights.parquet' as base
GROUP BY 1
ORDER BY 1 asc NULLS LAST

For more detailed information, see the Calculations and Window Functions section.

run: flights -> {
  group_by: 
    carrier is 'CA'
    carrier2 is carrier
  calculate: prev_carrier is lag(carrier)
}

QUERY RESULTS

[
  {
    "carrier": "CA",
    "carrier2": "RU",
    "prev_carrier": null
  },
  {
    "carrier": "CA",
    "carrier2": "DL",
    "prev_carrier": "CA"
  },
  {
    "carrier": "CA",
    "carrier2": "WN",
    "prev_carrier": "CA"
  },
  {
    "carrier": "CA",
    "carrier2": "AA",
    "prev_carrier": "CA"
  },
  {
    "carrier": "CA",
    "carrier2": "NW",
    "prev_carrier": "CA"
  }
]

SELECT 
   'CA' as "carrier",
   base."carrier" as "carrier2",
   LAG(('CA')) OVER(  ORDER BY  'CA' asc NULLS LAST ) as "prev_carrier"
FROM '../data/flights.parquet' as base
GROUP BY 1,2
ORDER BY 1 asc NULLS LAST

In the above example, carrier in the output refers to the carrier is 'CA', whereas carrier in the input refers to the column in flights. Inside a group_by it is not possible to reference an output field, so carrier2 is carrier refers to the column carrier. In the calculate: prev_carrier is lag(carrier), however, carrier refers to the output field, i.e. carrier is 'CA'.

Most arguments to analytic functions are required to be output fields or aggregate fields (see specific requirements below).

run: flights -> {
  // ERROR: Parameter 1 ('value') of lag must be literal, constant, or output, but received input
  calculate: prev_carrier is lag(carrier)
}

avg_moving	first_value	lag	last_value
lead	max_cumulative	max_window	min_cumulative
min_window	rank	row_number	sum_cumulative
sum_window

row_number

row_number()

Returns the row number of the current result row after grouping and aggregating.

run: flights -> {
  group_by: carrier
  calculate: row is row_number()
}

QUERY RESULTS

[
  {
    "carrier": "AA",
    "row": 1
  },
  {
    "carrier": "AS",
    "row": 2
  },
  {
    "carrier": "B6",
    "row": 3
  },
  {
    "carrier": "CO",
    "row": 4
  },
  {
    "carrier": "DL",
    "row": 5
  }
]

SELECT 
   base."carrier" as "carrier",
   ROW_NUMBER() OVER(  ORDER BY  base."carrier" asc NULLS LAST ) as "row"
FROM '../data/flights.parquet' as base
GROUP BY 1
ORDER BY 1 asc NULLS LAST

rank

rank()

Returns the rank according to the query ordering, with values having equal ordering value getting equal rank.

run: flights -> {
  group_by: carrier
  aggregate: flight_count_bucket is round(flight_count, -6)
  order_by: flight_count_bucket desc
  calculate: flight_count_rank is rank()
}

QUERY RESULTS

[
  {
    "carrier": "DL",
    "flight_count_bucket": 0,
    "flight_count_rank": 1
  },
  {
    "carrier": "RU",
    "flight_count_bucket": 0,
    "flight_count_rank": 1
  },
  {
    "carrier": "OH",
    "flight_count_bucket": 0,
    "flight_count_rank": 1
  },
  {
    "carrier": "CO",
    "flight_count_bucket": 0,
    "flight_count_rank": 1
  },
  {
    "carrier": "US",
    "flight_count_bucket": 0,
    "flight_count_rank": 1
  }
]

SELECT 
   base."carrier" as "carrier",
   ROUND((COUNT(1)),-6) as "flight_count_bucket",
   RANK() OVER(  ORDER BY  ROUND((COUNT(1)),-6) desc NULLS LAST ) as "flight_count_rank"
FROM '../data/flights.parquet' as base
GROUP BY 1
ORDER BY 2 desc NULLS LAST

lag

lag(expr)
lag(expr, offset)
lag(expr, offset, default)

Returns the value of expr for the previous row. If offset is specified, returns the value of expr for the row offset rows before the current row. Returns default (or null if unspecified) when the referenced row doesn't exist.

Generates an error if offset is null, negative, or not an integer.

The value of offset must be a constant.

run: flights -> {
  group_by: carrier
  calculate: prev_carrier is lag(carrier)
  calculate: prev_prev_carrier is lag(carrier, 2)
  calculate: prev_carrier_or_none is lag(carrier, 1, 'NONE')
}

QUERY RESULTS

[
  {
    "carrier": "AA",
    "prev_carrier": null,
    "prev_prev_carrier": null,
    "prev_carrier_or_none": "NONE"
  },
  {
    "carrier": "AS",
    "prev_carrier": "AA",
    "prev_prev_carrier": null,
    "prev_carrier_or_none": "AA"
  },
  {
    "carrier": "B6",
    "prev_carrier": "AS",
    "prev_prev_carrier": "AA",
    "prev_carrier_or_none": "AS"
  },
  {
    "carrier": "CO",
    "prev_carrier": "B6",
    "prev_prev_carrier": "AS",
    "prev_carrier_or_none": "B6"
  },
  {
    "carrier": "DL",
    "prev_carrier": "CO",
    "prev_prev_carrier": "B6",
    "prev_carrier_or_none": "CO"
  }
]

SELECT 
   base."carrier" as "carrier",
   LAG((base."carrier")) OVER(  ORDER BY  base."carrier" asc NULLS LAST ) as "prev_carrier",
   LAG((base."carrier"),2) OVER(  ORDER BY  base."carrier" asc NULLS LAST ) as "prev_prev_carrier",
   LAG((base."carrier"),1,'NONE') OVER(  ORDER BY  base."carrier" asc NULLS LAST ) as "prev_carrier_or_none"
FROM '../data/flights.parquet' as base
GROUP BY 1
ORDER BY 1 asc NULLS LAST

lead

lead(expr)
lead(expr, offset)
lead(expr, offset, default)

Returns the value of expr for the next row. If offset is specified, returns the value of expr for the row offset rows after the current row. Returns default (or null if unspecified) when the referenced row doesn't exist.

Generates an error if offset is null, negative, or not an integer.

The value of offset must be a constant.

run: flights -> { 
  group_by: carrier
  limit: 5 
} -> {
  group_by: carrier
  calculate: next_carrier is lead(carrier)
  calculate: next_next_carrier is lead(carrier, 2)
  calculate: next_carrier_or_none is lead(carrier, 1, 'NONE')
}

QUERY RESULTS

[
  {
    "carrier": "AA",
    "next_carrier": "AS",
    "next_next_carrier": "B6",
    "next_carrier_or_none": "AS"
  },
  {
    "carrier": "AS",
    "next_carrier": "B6",
    "next_next_carrier": "CO",
    "next_carrier_or_none": "B6"
  },
  {
    "carrier": "B6",
    "next_carrier": "CO",
    "next_next_carrier": "DL",
    "next_carrier_or_none": "CO"
  },
  {
    "carrier": "CO",
    "next_carrier": "DL",
    "next_next_carrier": null,
    "next_carrier_or_none": "DL"
  },
  {
    "carrier": "DL",
    "next_carrier": null,
    "next_next_carrier": null,
    "next_carrier_or_none": "NONE"
  }
]

WITH __stage0 AS (
  SELECT 
     base."carrier" as "carrier"
  FROM '../data/flights.parquet' as base
  GROUP BY 1
  ORDER BY 1 asc NULLS LAST
  LIMIT 5
)
SELECT 
   base."carrier" as "carrier",
   LEAD((base."carrier")) OVER(  ORDER BY  base."carrier" asc NULLS LAST ) as "next_carrier",
   LEAD((base."carrier"),2) OVER(  ORDER BY  base."carrier" asc NULLS LAST ) as "next_next_carrier",
   LEAD((base."carrier"),1,'NONE') OVER(  ORDER BY  base."carrier" asc NULLS LAST ) as "next_carrier_or_none"
FROM __stage0 as base
GROUP BY 1
ORDER BY 1 asc NULLS LAST

first_value

first_value(expr)

Returns the first value of expr across all rows (i.e. the value for the first row).

run: flights -> {
  group_by: carrier
  calculate: first_carrier is first_value(carrier)
}

QUERY RESULTS

[
  {
    "carrier": "AA",
    "first_carrier": "AA"
  },
  {
    "carrier": "AS",
    "first_carrier": "AA"
  },
  {
    "carrier": "B6",
    "first_carrier": "AA"
  },
  {
    "carrier": "CO",
    "first_carrier": "AA"
  },
  {
    "carrier": "DL",
    "first_carrier": "AA"
  }
]

SELECT 
   base."carrier" as "carrier",
   FIRST_VALUE((base."carrier")) OVER(  ORDER BY  base."carrier" asc NULLS LAST ) as "first_carrier"
FROM '../data/flights.parquet' as base
GROUP BY 1
ORDER BY 1 asc NULLS LAST

last_value

last_value(expr)

Returns the last value of expr across all rows (i.e. the value for the last row).

run: flights -> { 
  group_by: carrier
  limit: 5 
} -> {
  group_by: carrier
  calculate: last_carrier is last_value(carrier)
}

QUERY RESULTS

[
  {
    "carrier": "AA",
    "last_carrier": "DL"
  },
  {
    "carrier": "AS",
    "last_carrier": "DL"
  },
  {
    "carrier": "B6",
    "last_carrier": "DL"
  },
  {
    "carrier": "CO",
    "last_carrier": "DL"
  },
  {
    "carrier": "DL",
    "last_carrier": "DL"
  }
]

WITH __stage0 AS (
  SELECT 
     base."carrier" as "carrier"
  FROM '../data/flights.parquet' as base
  GROUP BY 1
  ORDER BY 1 asc NULLS LAST
  LIMIT 5
)
SELECT 
   base."carrier" as "carrier",
   LAST_VALUE((base."carrier")) OVER(  ORDER BY  base."carrier" asc NULLS LAST ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) as "last_carrier"
FROM __stage0 as base
GROUP BY 1
ORDER BY 1 asc NULLS LAST

min_cumulative

min_cumulative(expr)

Returns the minimum value of expr among rows from the first row to the current row.

run: flights -> {
  group_by: carrier
  order_by: carrier asc
  aggregate: flight_count
  calculate: min_cumulative_flight_count is min_cumulative(flight_count)
}

QUERY RESULTS

[
  {
    "carrier": "AA",
    "flight_count": 34577,
    "min_cumulative_flight_count": 34577
  },
  {
    "carrier": "AS",
    "flight_count": 8453,
    "min_cumulative_flight_count": 8453
  },
  {
    "carrier": "B6",
    "flight_count": 4842,
    "min_cumulative_flight_count": 4842
  },
  {
    "carrier": "CO",
    "flight_count": 7139,
    "min_cumulative_flight_count": 4842
  },
  {
    "carrier": "DL",
    "flight_count": 32130,
    "min_cumulative_flight_count": 4842
  }
]

SELECT 
   base."carrier" as "carrier",
   COUNT(1) as "flight_count",
   MIN((COUNT(1))) OVER(  ORDER BY  base."carrier" asc NULLS LAST ) as "min_cumulative_flight_count"
FROM '../data/flights.parquet' as base
GROUP BY 1
ORDER BY 1 asc NULLS LAST

max_cumulative

max_cumulative(expr)

Returns the maximum value of expr among rows from the first row to the current row.

run: flights -> {
  group_by: carrier
  order_by: carrier asc
  aggregate: flight_count
  calculate: max_cumulative_flight_count is max_cumulative(flight_count)
}

QUERY RESULTS

[
  {
    "carrier": "AA",
    "flight_count": 34577,
    "max_cumulative_flight_count": 34577
  },
  {
    "carrier": "AS",
    "flight_count": 8453,
    "max_cumulative_flight_count": 34577
  },
  {
    "carrier": "B6",
    "flight_count": 4842,
    "max_cumulative_flight_count": 34577
  },
  {
    "carrier": "CO",
    "flight_count": 7139,
    "max_cumulative_flight_count": 34577
  },
  {
    "carrier": "DL",
    "flight_count": 32130,
    "max_cumulative_flight_count": 34577
  }
]

SELECT 
   base."carrier" as "carrier",
   COUNT(1) as "flight_count",
   MAX((COUNT(1))) OVER(  ORDER BY  base."carrier" asc NULLS LAST ) as "max_cumulative_flight_count"
FROM '../data/flights.parquet' as base
GROUP BY 1
ORDER BY 1 asc NULLS LAST

sum_cumulative

sum_cumulative(expr)

Returns the cumulative sum of values of expr from the first row to the current row.

run: flights -> {
  group_by: carrier
  order_by: carrier asc
  aggregate: flight_count
  calculate: sum_cumulative_flight_count is sum_cumulative(flight_count)
}

QUERY RESULTS

[
  {
    "carrier": "AA",
    "flight_count": 34577,
    "sum_cumulative_flight_count": 34577
  },
  {
    "carrier": "AS",
    "flight_count": 8453,
    "sum_cumulative_flight_count": 43030
  },
  {
    "carrier": "B6",
    "flight_count": 4842,
    "sum_cumulative_flight_count": 47872
  },
  {
    "carrier": "CO",
    "flight_count": 7139,
    "sum_cumulative_flight_count": 55011
  },
  {
    "carrier": "DL",
    "flight_count": 32130,
    "sum_cumulative_flight_count": 87141
  }
]

SELECT 
   base."carrier" as "carrier",
   COUNT(1) as "flight_count",
   SUM((COUNT(1))) OVER(  ORDER BY  base."carrier" asc NULLS LAST ) as "sum_cumulative_flight_count"
FROM '../data/flights.parquet' as base
GROUP BY 1
ORDER BY 1 asc NULLS LAST

min_window

min_window(expr)

Returns the minimum of all values of expr across all rows.

run: flights -> {
  group_by: carrier
  aggregate: flight_count
  calculate: min_flight_count is min_window(flight_count)
}

QUERY RESULTS

[
  {
    "carrier": "WN",
    "flight_count": 88751,
    "min_flight_count": 3033
  },
  {
    "carrier": "US",
    "flight_count": 37683,
    "min_flight_count": 3033
  },
  {
    "carrier": "AA",
    "flight_count": 34577,
    "min_flight_count": 3033
  },
  {
    "carrier": "NW",
    "flight_count": 33580,
    "min_flight_count": 3033
  },
  {
    "carrier": "UA",
    "flight_count": 32757,
    "min_flight_count": 3033
  }
]

SELECT 
   base."carrier" as "carrier",
   COUNT(1) as "flight_count",
   MIN((COUNT(1))) OVER(  ) as "min_flight_count"
FROM '../data/flights.parquet' as base
GROUP BY 1
ORDER BY 2 desc NULLS LAST

max_window

max_window(expr)

Returns the maximum of all values of expr across all rows.

run: flights -> {
  group_by: carrier
  aggregate: flight_count
  calculate: max_flight_count is max_window(flight_count)
}

QUERY RESULTS

[
  {
    "carrier": "WN",
    "flight_count": 88751,
    "max_flight_count": 88751
  },
  {
    "carrier": "US",
    "flight_count": 37683,
    "max_flight_count": 88751
  },
  {
    "carrier": "AA",
    "flight_count": 34577,
    "max_flight_count": 88751
  },
  {
    "carrier": "NW",
    "flight_count": 33580,
    "max_flight_count": 88751
  },
  {
    "carrier": "UA",
    "flight_count": 32757,
    "max_flight_count": 88751
  }
]

SELECT 
   base."carrier" as "carrier",
   COUNT(1) as "flight_count",
   MAX((COUNT(1))) OVER(  ) as "max_flight_count"
FROM '../data/flights.parquet' as base
GROUP BY 1
ORDER BY 2 desc NULLS LAST

sum_window

sum_window(expr)

Returns the sum of all values of expr across all rows.

run: flights -> {
  group_by: carrier
  aggregate: flight_count
  calculate: total_flight_count1 is sum_window(flight_count)
  aggregate: total_flight_count2 is all(flight_count)
}

QUERY RESULTS

[
  {
    "carrier": "WN",
    "flight_count": 88751,
    "total_flight_count1": 344827,
    "total_flight_count2": 344827
  },
  {
    "carrier": "US",
    "flight_count": 37683,
    "total_flight_count1": 344827,
    "total_flight_count2": 344827
  },
  {
    "carrier": "AA",
    "flight_count": 34577,
    "total_flight_count1": 344827,
    "total_flight_count2": 344827
  },
  {
    "carrier": "NW",
    "flight_count": 33580,
    "total_flight_count1": 344827,
    "total_flight_count2": 344827
  },
  {
    "carrier": "UA",
    "flight_count": 32757,
    "total_flight_count1": 344827,
    "total_flight_count2": 344827
  }
]

WITH __stage0 AS (
  SELECT
    group_set,
    CASE WHEN group_set=1 THEN
      base."carrier"
      END as "carrier__1",
    CASE WHEN group_set=1 THEN
      COUNT(1)
      END as "flight_count__1",
    CASE WHEN group_set=1 THEN SUM((CASE WHEN group_set=1 THEN
      COUNT(1)
      END)) OVER(PARTITION BY group_set  ) END as "total_flight_count1__1",
    MAX((CASE WHEN group_set=0 THEN
      COUNT(1)
      END)) OVER () as "total_flight_count2__1"
  FROM '../data/flights.parquet' as base
  CROSS JOIN (SELECT UNNEST(GENERATE_SERIES(0,1,1)) as group_set  ) as group_set
  GROUP BY 1,2
)
SELECT
  "carrier__1" as "carrier",
  MAX(CASE WHEN group_set=1 THEN "flight_count__1" END) as "flight_count",
  MAX(CASE WHEN group_set=1 THEN "total_flight_count1__1" END) as "total_flight_count1",
  MAX(CASE WHEN group_set=1 THEN "total_flight_count2__1" END) as "total_flight_count2"
FROM __stage0
WHERE group_set NOT IN (0)
GROUP BY 1
ORDER BY 2 desc NULLS LAST

avg_moving

avg_moving(expr, preceding)
avg_moving(expr, preceding, following)

Produces a moving average (or 'rolling average') of values of expr among rows between preceding rows before the current row and following rows after the current row (or the current row if following is not specified).

Both preceding and following must be literals.

Note: avg_moving(value, 3) means that the average is computed over 4 rows (the current row and 3 preceding).

run: flights -> {
  group_by: carrier
  aggregate: flight_count
  nest: carrier_analysis is {
    group_by: yr is dep_time.year
    aggregate: flight_count
    order_by: yr asc
    calculate: three_year_avg is avg_moving(flight_count, 2)
    calculate: three_year_avg_centered is avg_moving(flight_count, 1, 1)
  }
}

QUERY RESULTS

[
  {
    "carrier": "WN",
    "flight_count": 88751,
    "carrier_analysis": [
      {
        "yr": "2000-01-01T00:00:00.000Z",
        "flight_count": 13133,
        "three_year_avg": 13133,
        "three_year_avg_centered": 13777
      },
      {
        "yr": "2001-01-01T00:00:00.000Z",
        "flight_count": 14421,
        "three_year_avg": 13777,
        "three_year_avg_centered": 14087.333333333334
      },
      {
        "yr": "2002-01-01T00:00:00.000Z",
        "flight_count": 14708,
        "three_year_avg": 14087.333333333334,
        "three_year_avg_centered": 14476.333333333334
      },
      {
        "yr": "2003-01-01T00:00:00.000Z",
        "flight_count": 14300,
        "three_year_avg": 14476.333333333334,
        "three_year_avg_centered": 14549.333333333334
      },
      {
        "yr": "2004-01-01T00:00:00.000Z",
        "flight_count": 14640,
        "three_year_avg": 14549.333333333334,
        "three_year_avg_centered": 15496.333333333334
      },
      {
        "yr": "2005-01-01T00:00:00.000Z",
        "flight_count": 17549,
        "three_year_avg": 15496.333333333334,
        "three_year_avg_centered": 16094.5
      }
    ]
  },
  {
    "carrier": "US",
    "flight_count": 37683,
    "carrier_analysis": [
      {
        "yr": "2000-01-01T00:00:00.000Z",
        "flight_count": 7161,
        "three_year_avg": 7161,
        "three_year_avg_centered": 7013.5
      },
      {
        "yr": "2001-01-01T00:00:00.000Z",
        "flight_count": 6866,
        "three_year_avg": 7013.5,
        "three_year_avg_centered": 6842.666666666667
      },
      {
        "yr": "2002-01-01T00:00:00.000Z",
        "flight_count": 6501,
        "three_year_avg": 6842.666666666667,
        "three_year_avg_centered": 5896
      },
      {
        "yr": "2003-01-01T00:00:00.000Z",
        "flight_count": 4321,
        "three_year_avg": 5896,
        "three_year_avg_centered": 5403.333333333333
      },
      {
        "yr": "2004-01-01T00:00:00.000Z",
        "flight_count": 5388,
        "three_year_avg": 5403.333333333333,
        "three_year_avg_centered": 5718.333333333333
      },
      {
        "yr": "2005-01-01T00:00:00.000Z",
        "flight_count": 7446,
        "three_year_avg": 5718.333333333333,
        "three_year_avg_centered": 6417
      }
    ]
  },
  {
    "carrier": "AA",
    "flight_count": 34577,
    "carrier_analysis": [
      {
        "yr": "2000-01-01T00:00:00.000Z",
        "flight_count": 6050,
        "three_year_avg": 6050,
        "three_year_avg_centered": 5950.5
      },
      {
        "yr": "2001-01-01T00:00:00.000Z",
        "flight_count": 5851,
        "three_year_avg": 5950.5,
        "three_year_avg_centered": 5954.666666666667
      },
      {
        "yr": "2002-01-01T00:00:00.000Z",
        "flight_count": 5963,
        "three_year_avg": 5954.666666666667,
        "three_year_avg_centered": 5793.333333333333
      },
      {
        "yr": "2003-01-01T00:00:00.000Z",
        "flight_count": 5566,
        "three_year_avg": 5793.333333333333,
        "three_year_avg_centered": 5711
      },
      {
        "yr": "2004-01-01T00:00:00.000Z",
        "flight_count": 5604,
        "three_year_avg": 5711,
        "three_year_avg_centered": 5571
      },
      {
        "yr": "2005-01-01T00:00:00.000Z",
        "flight_count": 5543,
        "three_year_avg": 5571,
        "three_year_avg_centered": 5573.5
      }
    ]
  },
  {
    "carrier": "NW",
    "flight_count": 33580,
    "carrier_analysis": [
      {
        "yr": "2000-01-01T00:00:00.000Z",
        "flight_count": 5756,
        "three_year_avg": 5756,
        "three_year_avg_centered": 5568.5
      },
      {
        "yr": "2001-01-01T00:00:00.000Z",
        "flight_count": 5381,
        "three_year_avg": 5568.5,
        "three_year_avg_centered": 5406
      },
      {
        "yr": "2002-01-01T00:00:00.000Z",
        "flight_count": 5081,
        "three_year_avg": 5406,
        "three_year_avg_centered": 5216.333333333333
      },
      {
        "yr": "2003-01-01T00:00:00.000Z",
        "flight_count": 5187,
        "three_year_avg": 5216.333333333333,
        "three_year_avg_centered": 5524.666666666667
      },
      {
        "yr": "2004-01-01T00:00:00.000Z",
        "flight_count": 6306,
        "three_year_avg": 5524.666666666667,
        "three_year_avg_centered": 5787.333333333333
      },
      {
        "yr": "2005-01-01T00:00:00.000Z",
        "flight_count": 5869,
        "three_year_avg": 5787.333333333333,
        "three_year_avg_centered": 6087.5
      }
    ]
  },
  {
    "carrier": "UA",
    "flight_count": 32757,
    "carrier_analysis": [
      {
        "yr": "2000-01-01T00:00:00.000Z",
        "flight_count": 3621,
        "three_year_avg": 3621,
        "three_year_avg_centered": 3703
      },
      {
        "yr": "2001-01-01T00:00:00.000Z",
        "flight_count": 3785,
        "three_year_avg": 3703,
        "three_year_avg_centered": 4332
      },
      {
        "yr": "2002-01-01T00:00:00.000Z",
        "flight_count": 5590,
        "three_year_avg": 4332,
        "three_year_avg_centered": 5147.666666666667
      },
      {
        "yr": "2003-01-01T00:00:00.000Z",
        "flight_count": 6068,
        "three_year_avg": 5147.666666666667,
        "three_year_avg_centered": 6158.333333333333
      },
      {
        "yr": "2004-01-01T00:00:00.000Z",
        "flight_count": 6817,
        "three_year_avg": 6158.333333333333,
        "three_year_avg_centered": 6587
      },
      {
        "yr": "2005-01-01T00:00:00.000Z",
        "flight_count": 6876,
        "three_year_avg": 6587,
        "three_year_avg_centered": 6846.5
      }
    ]
  }
]

WITH __stage0 AS (
  SELECT
    group_set,
    base."carrier" as "carrier__0",
    CASE WHEN group_set=0 THEN
      COUNT(1)
      END as "flight_count__0",
    CASE WHEN group_set=1 THEN
      DATE_TRUNC('year', base."dep_time")
      END as "yr__1",
    CASE WHEN group_set=1 THEN
      COUNT(1)
      END as "flight_count__1",
    CASE WHEN group_set=1 THEN AVG((CASE WHEN group_set=1 THEN
      COUNT(1)
      END)) OVER(PARTITION BY group_set, base."carrier"  ORDER BY  CASE WHEN group_set=1 THEN
      DATE_TRUNC('year', base."dep_time")
      END asc NULLS LAST ROWS BETWEEN 2 PRECEDING AND 0 FOLLOWING) END as "three_year_avg__1",
    CASE WHEN group_set=1 THEN AVG((CASE WHEN group_set=1 THEN
      COUNT(1)
      END)) OVER(PARTITION BY group_set, base."carrier"  ORDER BY  CASE WHEN group_set=1 THEN
      DATE_TRUNC('year', base."dep_time")
      END asc NULLS LAST ROWS BETWEEN 1 PRECEDING AND 1 FOLLOWING) END as "three_year_avg_centered__1"
  FROM '../data/flights.parquet' as base
  CROSS JOIN (SELECT UNNEST(GENERATE_SERIES(0,1,1)) as group_set  ) as group_set
  GROUP BY 1,2,4
)
SELECT
  "carrier__0" as "carrier",
  MAX(CASE WHEN group_set=0 THEN "flight_count__0" END) as "flight_count",
  COALESCE(LIST({
    "yr": "yr__1", 
    "flight_count": "flight_count__1", 
    "three_year_avg": "three_year_avg__1", 
    "three_year_avg_centered": "three_year_avg_centered__1"}  ORDER BY  "yr__1" asc NULLS LAST) FILTER (WHERE group_set=1),[]) as "carrier_analysis"
FROM __stage0
GROUP BY 1
ORDER BY 2 desc NULLS LAST