14.20.1 窗口函数说明
This section describes nonaggregate window functions that, for each row from a query, perform a calculation using rows related to that row. Most aggregate functions also can be used as window functions; see Section 14.19.1, “Aggregate Function Descriptions”.
For window function usage information and examples, and definitions of terms such as the OVER
clause, window, partition, frame, and peer, see Section 14.20.2, “Window Function Concepts and Syntax”.
Table 14.30 Window Functions
Name | Description |
---|---|
CUME_DIST() |
Cumulative distribution value |
DENSE_RANK() |
Rank of current row within its partition, without gaps |
FIRST_VALUE() |
Value of argument from first row of window frame |
LAG() |
Value of argument from row lagging current row within partition |
LAST_VALUE() |
Value of argument from last row of window frame |
LEAD() |
Value of argument from row leading current row within partition |
NTH_VALUE() |
Value of argument from N-th row of window frame |
NTILE() |
Bucket number of current row within its partition. |
PERCENT_RANK() |
Percentage rank value |
RANK() |
Rank of current row within its partition, with gaps |
ROW_NUMBER() |
Number of current row within its partition |
In the following function descriptions, over_clause
represents the OVER
clause, described in Section 14.20.2, “Window Function Concepts and Syntax”. Some window functions permit a null_treatment
clause that specifies how to handle NULL
values when calculating results. This clause is optional. It is part of the SQL standard, but the MySQL implementation permits only RESPECT NULLS
(which is also the default). This means that NULL
values are considered when calculating results. IGNORE NULLS
is parsed, but produces an error.
-
CUME_DIST()
over_clause
Returns the cumulative distribution of a value within a group of values; that is, the percentage of partition values less than or equal to the value in the current row. This represents the number of rows preceding or peer with the current row in the window ordering of the window partition divided by the total number of rows in the window partition. Return values range from 0 to 1.
This function should be used with
ORDER BY
to sort partition rows into the desired order. WithoutORDER BY
, all rows are peers and have valueN
/N
= 1, whereN
is the partition size.over_clause
is as described in Section 14.20.2, “Window Function Concepts and Syntax”.The following query shows, for the set of values in the
val
column, theCUME_DIST()
value for each row, as well as the percentage rank value returned by the similarPERCENT_RANK()
function. For reference, the query also displays row numbers usingROW_NUMBER()
:mysql> SELECT val, ROW_NUMBER() OVER w AS 'row_number', CUME_DIST() OVER w AS 'cume_dist', PERCENT_RANK() OVER w AS 'percent_rank' FROM numbers WINDOW w AS (ORDER BY val); +------+------------+--------------------+--------------+ | val | row_number | cume_dist | percent_rank | +------+------------+--------------------+--------------+ | 1 | 1 | 0.2222222222222222 | 0 | | 1 | 2 | 0.2222222222222222 | 0 | | 2 | 3 | 0.3333333333333333 | 0.25 | | 3 | 4 | 0.6666666666666666 | 0.375 | | 3 | 5 | 0.6666666666666666 | 0.375 | | 3 | 6 | 0.6666666666666666 | 0.375 | | 4 | 7 | 0.8888888888888888 | 0.75 | | 4 | 8 | 0.8888888888888888 | 0.75 | | 5 | 9 | 1 | 1 | +------+------------+--------------------+--------------+
-
DENSE_RANK()
over_clause
Returns the rank of the current row within its partition, without gaps. Peers are considered ties and receive the same rank. This function assigns consecutive ranks to peer groups; the result is that groups of size greater than one do not produce noncontiguous rank numbers. For an example, see the
RANK()
function description.This function should be used with
ORDER BY
to sort partition rows into the desired order. WithoutORDER BY
, all rows are peers.over_clause
is as described in Section 14.20.2, “Window Function Concepts and Syntax”. -
FIRST_VALUE(
[expr
)null_treatment
]over_clause
Returns the value of
expr
from the first row of the window frame.over_clause
is as described in Section 14.20.2, “Window Function Concepts and Syntax”.null_treatment
is as described in the section introduction.The following query demonstrates
FIRST_VALUE()
,LAST_VALUE()
, and two instances ofNTH_VALUE()
:mysql> SELECT time, subject, val, FIRST_VALUE(val) OVER w AS 'first', LAST_VALUE(val) OVER w AS 'last', NTH_VALUE(val, 2) OVER w AS 'second', NTH_VALUE(val, 4) OVER w AS 'fourth' FROM observations WINDOW w AS (PARTITION BY subject ORDER BY time ROWS UNBOUNDED PRECEDING); +----------+---------+------+-------+------+--------+--------+ | time | subject | val | first | last | second | fourth | +----------+---------+------+-------+------+--------+--------+ | 07:00:00 | st113 | 10 | 10 | 10 | NULL | NULL | | 07:15:00 | st113 | 9 | 10 | 9 | 9 | NULL | | 07:30:00 | st113 | 25 | 10 | 25 | 9 | NULL | | 07:45:00 | st113 | 20 | 10 | 20 | 9 | 20 | | 07:00:00 | xh458 | 0 | 0 | 0 | NULL | NULL | | 07:15:00 | xh458 | 10 | 0 | 10 | 10 | NULL | | 07:30:00 | xh458 | 5 | 0 | 5 | 10 | NULL | | 07:45:00 | xh458 | 30 | 0 | 30 | 10 | 30 | | 08:00:00 | xh458 | 25 | 0 | 25 | 10 | 30 | +----------+---------+------+-------+------+--------+--------+
Each function uses the rows in the current frame, which, per the window definition shown, extends from the first partition row to the current row. For the
NTH_VALUE()
calls, the current frame does not always include the requested row; in such cases, the return value isNULL
. -
LAG(
[expr
[,N
[,default
]])null_treatment
]over_clause
Returns the value of
expr
from the row that lags (precedes) the current row byN
rows within its partition. If there is no such row, the return value isdefault
. For example, ifN
is 3, the return value isdefault
for the first three rows. IfN
ordefault
are missing, the defaults are 1 andNULL
, respectively.N
must be a literal nonnegative integer. IfN
is 0,expr
is evaluated for the current row.N
cannot beNULL
, and must be an integer in the range0
to263
, inclusive, in any of the following forms:-
an unsigned integer constant literal
-
a positional parameter marker (
?
) -
a user-defined variable
-
a local variable in a stored routine
over_clause
is as described in Section 14.20.2, “Window Function Concepts and Syntax”.null_treatment
is as described in the section introduction.LAG()
(and the similarLEAD()
function) are often used to compute differences between rows. The following query shows a set of time-ordered observations and, for each one, theLAG()
andLEAD()
values from the adjoining rows, as well as the differences between the current and adjoining rows:mysql> SELECT t, val, LAG(val) OVER w AS 'lag', LEAD(val) OVER w AS 'lead', val - LAG(val) OVER w AS 'lag diff', val - LEAD(val) OVER w AS 'lead diff' FROM series WINDOW w AS (ORDER BY t); +----------+------+------+------+----------+-----------+ | t | val | lag | lead | lag diff | lead diff | +----------+------+------+------+----------+-----------+ | 12:00:00 | 100 | NULL | 125 | NULL | -25 | | 13:00:00 | 125 | 100 | 132 | 25 | -7 | | 14:00:00 | 132 | 125 | 145 | 7 | -13 | | 15:00:00 | 145 | 132 | 140 | 13 | 5 | | 16:00:00 | 140 | 145 | 150 | -5 | -10 | | 17:00:00 | 150 | 140 | 200 | 10 | -50 | | 18:00:00 | 200 | 150 | NULL | 50 | NULL | +----------+------+------+------+----------+-----------+
In the example, the
LAG()
andLEAD()
calls use the defaultN
anddefault
values of 1 andNULL
, respectively.The first row shows what happens when there is no previous row for
LAG()
: The function returns thedefault
value (in this case,NULL
). The last row shows the same thing when there is no next row forLEAD()
.LAG()
andLEAD()
also serve to compute sums rather than differences. Consider this data set, which contains the first few numbers of the Fibonacci series:mysql> SELECT n FROM fib ORDER BY n; +------+ | n | +------+ | 1 | | 1 | | 2 | | 3 | | 5 | | 8 | +------+
The following query shows the
LAG()
andLEAD()
values for the rows adjacent to the current row. It also uses those functions to add to the current row value the values from the preceding and following rows. The effect is to generate the next number in the Fibonacci series, and the next number after that:mysql> SELECT n, LAG(n, 1, 0) OVER w AS 'lag', LEAD(n, 1, 0) OVER w AS 'lead', n + LAG(n, 1, 0) OVER w AS 'next_n', n + LEAD(n, 1, 0) OVER w AS 'next_next_n' FROM fib WINDOW w AS (ORDER BY n); +------+------+------+--------+-------------+ | n | lag | lead | next_n | next_next_n | +------+------+------+--------+-------------+ | 1 | 0 | 1 | 1 | 2 | | 1 | 1 | 2 | 2 | 3 | | 2 | 1 | 3 | 3 | 5 | | 3 | 2 | 5 | 5 | 8 | | 5 | 3 | 8 | 8 | 13 | | 8 | 5 | 0 | 13 | 8 | +------+------+------+--------+-------------+
One way to generate the initial set of Fibonacci numbers is to use a recursive common table expression. For an example, see Fibonacci Series Generation.
You cannot use a negative value for the rows argument of this function.
-
-
LAST_VALUE(
[expr
)null_treatment
]over_clause
Returns the value of
expr
from the last row of the window frame.over_clause
is as described in Section 14.20.2, “Window Function Concepts and Syntax”.null_treatment
is as described in the section introduction.For an example, see the
FIRST_VALUE()
function description. -
LEAD(
[expr
[,N
[,default
]])null_treatment
]over_clause
Returns the value of
expr
from the row that leads (follows) the current row byN
rows within its partition. If there is no such row, the return value isdefault
. For example, ifN
is 3, the return value isdefault
for the last three rows. IfN
ordefault
are missing, the defaults are 1 andNULL
, respectively.N
must be a literal nonnegative integer. IfN
is 0,expr
is evaluated for the current row.N
cannot beNULL
, and must be an integer in the range0
to263
, inclusive, in any of the following forms:-
an unsigned integer constant literal
-
a positional parameter marker (
?
) -
a user-defined variable
-
a local variable in a stored routine
over_clause
is as described in Section 14.20.2, “Window Function Concepts and Syntax”.null_treatment
is as described in the section introduction.For an example, see the
LAG()
function description.Use of a negative value for the rows argument of this function is not permitted.
-
-
NTH_VALUE(
[expr
,N
)from_first_last
] [null_treatment
]over_clause
Returns the value of
expr
from theN
-th row of the window frame. If there is no such row, the return value isNULL
.N
must be a literal positive integer.from_first_last
is part of the SQL standard, but the MySQL implementation permits onlyFROM FIRST
(which is also the default). This means that calculations begin at the first row of the window.FROM LAST
is parsed, but produces an error. To obtain the same effect asFROM LAST
(begin calculations at the last row of the window), useORDER BY
to sort in reverse order.over_clause
is as described in Section 14.20.2, “Window Function Concepts and Syntax”.null_treatment
is as described in the section introduction.For an example, see the
FIRST_VALUE()
function description.You cannot use
NULL
for the row argument of this function. -
NTILE(
N
)over_clause
Divides a partition into
N
groups (buckets), assigns each row in the partition its bucket number, and returns the bucket number of the current row within its partition. For example, ifN
is 4,NTILE()
divides rows into four buckets. IfN
is 100,NTILE()
divides rows into 100 buckets.N
must be a literal positive integer. Bucket number return values range from 1 toN
.N
cannot beNULL
, and must be an integer in the range0
to263
, inclusive, in any of the following forms:-
an unsigned integer constant literal
-
a positional parameter marker (
?
) -
a user-defined variable
-
a local variable in a stored routine
This function should be used with
ORDER BY
to sort partition rows into the desired order.over_clause
is as described in Section 14.20.2, “Window Function Concepts and Syntax”.The following query shows, for the set of values in the
val
column, the percentile values resulting from dividing the rows into two or four groups. For reference, the query also displays row numbers usingROW_NUMBER()
:mysql> SELECT val, ROW_NUMBER() OVER w AS 'row_number', NTILE(2) OVER w AS 'ntile2', NTILE(4) OVER w AS 'ntile4' FROM numbers WINDOW w AS (ORDER BY val); +------+------------+--------+--------+ | val | row_number | ntile2 | ntile4 | +------+------------+--------+--------+ | 1 | 1 | 1 | 1 | | 1 | 2 | 1 | 1 | | 2 | 3 | 1 | 1 | | 3 | 4 | 1 | 2 | | 3 | 5 | 1 | 2 | | 3 | 6 | 2 | 3 | | 4 | 7 | 2 | 3 | | 4 | 8 | 2 | 4 | | 5 | 9 | 2 | 4 | +------+------------+--------+--------+
The construct
NTILE(NULL)
is not permitted. -
-
PERCENT_RANK()
over_clause
Returns the percentage of partition values less than the value in the current row, excluding the highest value. Return values range from 0 to 1 and represent the row relative rank, calculated as the result of this formula, where
rank
is the row rank androws
is the number of partition rows:(rank - 1) / (rows - 1)
This function should be used with
ORDER BY
to sort partition rows into the desired order. WithoutORDER BY
, all rows are peers.over_clause
is as described in Section 14.20.2, “Window Function Concepts and Syntax”.For an example, see the
CUME_DIST()
function description. -
RANK()
over_clause
Returns the rank of the current row within its partition, with gaps. Peers are considered ties and receive the same rank. This function does not assign consecutive ranks to peer groups if groups of size greater than one exist; the result is noncontiguous rank numbers.
This function should be used with
ORDER BY
to sort partition rows into the desired order. WithoutORDER BY
, all rows are peers.over_clause
is as described in Section 14.20.2, “Window Function Concepts and Syntax”.The following query shows the difference between
RANK()
, which produces ranks with gaps, andDENSE_RANK()
, which produces ranks without gaps. The query shows rank values for each member of a set of values in theval
column, which contains some duplicates.RANK()
assigns peers (the duplicates) the same rank value, and the next greater value has a rank higher by the number of peers minus one.DENSE_RANK()
also assigns peers the same rank value, but the next higher value has a rank one greater. For reference, the query also displays row numbers usingROW_NUMBER()
:mysql> SELECT val, ROW_NUMBER() OVER w AS 'row_number', RANK() OVER w AS 'rank', DENSE_RANK() OVER w AS 'dense_rank' FROM numbers WINDOW w AS (ORDER BY val); +------+------------+------+------------+ | val | row_number | rank | dense_rank | +------+------------+------+------------+ | 1 | 1 | 1 | 1 | | 1 | 2 | 1 | 1 | | 2 | 3 | 3 | 2 | | 3 | 4 | 4 | 3 | | 3 | 5 | 4 | 3 | | 3 | 6 | 4 | 3 | | 4 | 7 | 7 | 4 | | 4 | 8 | 7 | 4 | | 5 | 9 | 9 | 5 | +------+------------+------+------------+
-
ROW_NUMBER()
over_clause
Returns the number of the current row within its partition. Rows numbers range from 1 to the number of partition rows.
ORDER BY
affects the order in which rows are numbered. WithoutORDER BY
, row numbering is nondeterministic.ROW_NUMBER()
assigns peers different row numbers. To assign peers the same value, useRANK()
orDENSE_RANK()
. For an example, see theRANK()
function description.over_clause
is as described in Section 14.20.2, “Window Function Concepts and Syntax”.