Window function over

What are window functions and windows ？

Window function : window 、 function ( Functions applied in the window )— Windows are similar to windows , Limit a space

What is a window ?

The concept of window is very important , It can be understood as a collection of records , Window function is a special function executed on the record set that meets certain conditions . For each record, execute the function... In this window , The window size is fixed , This is a static window ; Different records correspond to different windows , This dynamic window is called sliding window .

structure

function () over（）

among ,over Is the key word , Used to specify the window range of function execution , Contains three analysis clauses : grouping (partition by) Clause , Sort (orderby) Clause , window (rows) Clause , If you don't write anything in parentheses , It means that the window contains where All the lines of the condition , The window function calculates based on all rows ; If it's not empty , The following syntax is supported to set the window :

Function name （[expr]) over(partition by < Groups to break down >order by < Columns to sort >rows between< Data range >)

Be careful

rows between 2 preceding and current row# Take the current line and the first two lines

rows between unbounded preceding and current row# Including this line and all previous lines

rows between current row and unbounded fo11owing# Including this line and all subsequent lines

rows between 3 preceding and current row# Including this line and the first three lines
I
rows between 3 preceding and 1 fo7lowing # From the front three lines and the bottom one , Five lines in all

# When order by There is no window clause condition behind , The default window specification is rows between unbounded preceding and currentrow.

# When order by And window clauses are missing , The default window specification is rows between unbounded preceding and unboundedfo11owing

---

commonly , We can divide window functions into two types :

Proprietary window functions :

rank()

dense_rank()

row_number()·

Aggregate class window function :

sum(),count(),avg(),max(),min()

Example

One 、ROW_NUMBER()
Row_number() The serial number is continuity No repetition , This is true even if you encounter two of the same values in the table

select *,row_number()
OVER(order by number )
as row_num from num 

Be careful ： In the use of row_number() One thing to pay special attention to when implementing paging ,over In Clause order by To work with SQL Sort... In records order by bring into correspondence with , Otherwise, the sequence number obtained may not be continuous

select *,row_number() 
OVER(order by number ) 
as row_num from num
ORDER BY id

Two 、rank()
Rank() The function will group the same values required to be sorted into a group with the same sequence number of each group , Sorting is not performed consecutively

select *,rank()
OVER(order by number ) 
as row_num from num

3、 ... and 、dense_rank()
Dense_rank() Sorting is continuous , The same values will also be divided into a group with the same sort number in each group

select *,dense_rank() 
OVER(order by number ) 
as row_num from num 

Four 、ntile()
Ntile(group_num) Divide all records into group_num A set of , The serial number of each group is the same

select *,ntile(2) 
OVER(order by number ) 
as row_num from num