[sql] usage of self connection

《SQL Advanced tutorial 》 MICK 1-2  Usage of self connection

About... For creating tables for examples SQL Statements and sample code , Please refer to the following website .
http://www.ituring.com.cn/book/1813（ Please click on “ Download with book ” Download relevant materials in Chinese ）

---------------------------------------------------------------

explain ： The tests in the following notes are based on postgresql14

Command line connection local PSQL：  psql -U <username> -d <dbname> -h 127.0.0.1 -W

Catalog

Rearrangeable 、 array 、 Combine

Delete duplicate lines

Method 1 ： Use the extremum function

Method 2 ： Use non equivalent connection

Find locally inconsistent columns

Sort

---

Rearrangeable 、 array 、 Combine

Products Create table statement

CREATE TABLE Products
(name VARCHAR(16) PRIMARY KEY,
 price INTEGER NOT NULL);

-- Rearrangeable · array · Combine 
INSERT INTO Products VALUES(' Apple ',	50);
INSERT INTO Products VALUES(' a mandarin orange ',	100);
INSERT INTO Products VALUES(' Banana ',	80);

==========================================
test=> select * from Products
test-> ;
 name | price
------+-------
  Apple  |    50
  a mandarin orange  |   100
  Banana  |    80
(3 rows)

test=>

A join to the same table is called “ Self join ”（self join）

（1） Rearrangeable

test=> SELECT P1.name AS name_1, P2.name AS name_2
test-> FROM Products P1, Products P2;
 name_1 | name_2
--------+--------
  Apple    |  Apple 
  Apple    |  a mandarin orange 
  Apple    |  Banana 
  a mandarin orange    |  Apple 
  a mandarin orange    |  a mandarin orange 
  a mandarin orange    |  Banana 
  Banana    |  Apple 
  Banana    |  a mandarin orange 
  Banana    |  Banana 
(9 rows)

test=>

（2） array

Remove pairs made of the same elements , for example （ Apple , Apple ）

test=> SELECT P1.name AS name_1, P2.name AS name_2
test-> FROM Products P1, Products P2
test-> WHERE P1.name <> P2.name;
 name_1 | name_2
--------+--------
  Apple    |  a mandarin orange 
  Apple    |  Banana 
  a mandarin orange    |  Apple 
  a mandarin orange    |  Banana 
  Banana    |  Apple 
  Banana    |  a mandarin orange 
(6 rows)

test=>

（3） Combine

test=> SELECT P1.name AS name_1, P2.name AS name_2
test-> FROM Products P1, Products P2
test-> WHERE P1.name > P2.name;
 name_1 | name_2
--------+--------
  Apple    |  a mandarin orange 
  Banana    |  a mandarin orange 
  Banana    |  Apple 
(3 rows)

test=>

Delete duplicate lines

The reconstruction Products

test=> DROP TABLE Products;
DROP TABLE
test=>
test=> CREATE TABLE Products
test-> (rowid INTEGER PRIMARY KEY,
test(>  name VARCHAR(16),
test(>  price INTEGER NOT NULL);
CREATE TABLE
test=>
test=> INSERT INTO Products VALUES(1, ' Apple ',50);
INSERT 0 1
test=> INSERT INTO Products VALUES(2, ' a mandarin orange ',100);
INSERT 0 1
test=> INSERT INTO Products VALUES(3, ' a mandarin orange ',100);
INSERT 0 1
test=> INSERT INTO Products VALUES(4, ' a mandarin orange ',100);
INSERT 0 1
test=> INSERT INTO Products VALUES(5, ' Banana ',80);
INSERT 0 1
test=>
test=> SELECT * FROM Products;
 rowid | name | price
-------+------+-------
     1 |  Apple  |    50
     2 |  a mandarin orange  |   100
     3 |  a mandarin orange  |   100
     4 |  a mandarin orange  |   100
     5 |  Banana  |    80
(5 rows)

test=>

Use the associated subquery to delete duplicate rows

Method 1 ： Use the extremum function

test=> DELETE FROM Products P1
test->  WHERE rowid < ( SELECT MAX(P2.rowid)
test(>                    FROM Products P2
test(>                   WHERE P1.name = P2. name
test(>                     AND P1.price = P2.price ) ;
DELETE 2
test=>
test=> SELECT * FROM Products;
 rowid | name | price
-------+------+-------
     1 |  Apple  |    50
     4 |  a mandarin orange  |   100
     5 |  Banana  |    80
(3 rows)

test=>

Method 2 ： Use non equivalent connection

test=> DELETE FROM Products P1
test->  WHERE EXISTS ( SELECT *
test(>                   FROM Products P2
test(>                  WHERE P1.name = P2.name
test(>                    AND P1.price = P2.price
test(>                    AND P1.rowid < P2.rowid );
DELETE 2
test=> SELECT * FROM Products;
 rowid | name | price
-------+------+-------
     1 |  Apple  |    50
     4 |  a mandarin orange  |   100
     5 |  Banana  |    80
(3 rows)

test=>

Find locally inconsistent columns

【 example 】 Suppose there is such an address table , The primary key is the person's name , The same family ID equally

If the family ID equally , The address must also be the same , It can be said that Mr. and Mrs. Maeda are like this “ Records of the same family with different addresses ” It's a hand error , Non equivalent self connection can be used to find such records

Addresses Create table statement

-- Find locally inconsistent columns 
CREATE TABLE Addresses
(name VARCHAR(32),
 family_id INTEGER,
 address VARCHAR(32),
 PRIMARY KEY(name, family_id));

INSERT INTO Addresses VALUES(' Maeda Yiming ', '100', ' Tiger gate, Tokyo Metropolitan port area 3-2-29');
INSERT INTO Addresses VALUES(' Maeda Youmei ', '100', ' Tiger gate, Tokyo Metropolitan port area 3-2-92');
INSERT INTO Addresses VALUES(' Kato tea ',   '200', ' West Shinjuku, Shinjuku District, Tokyo 2-8-1');
INSERT INTO Addresses VALUES(' Kato Sheng ',   '200', ' West Shinjuku, Shinjuku District, Tokyo 2-8-1');
INSERT INTO Addresses VALUES(' Sherlock Holmes ',  '300', ' Baker Street 221B');
INSERT INTO Addresses VALUES(' Watson ',  '400', ' Baker Street 221B');

Find records with the same family but different addresses

test=> select * from Addresses;
   name   | family_id |         address
----------+-----------+-------------------------
  Maeda Yiming  |       100 |  Tiger gate, Tokyo Metropolitan port area 3-2-29
  Maeda Youmei  |       100 |  Tiger gate, Tokyo Metropolitan port area 3-2-92
  Kato tea    |       200 |  West Shinjuku, Shinjuku District, Tokyo 2-8-1
  Kato Sheng    |       200 |  West Shinjuku, Shinjuku District, Tokyo 2-8-1
  Sherlock Holmes  |       300 |  Baker Street 221B
  Watson      |       400 |  Baker Street 221B
(6 rows)

test=>
test=> SELECT DISTINCT A1.name, A1.address
test-> FROM Addresses A1, Addresses A2
test-> WHERE A1.family_id = A2.family_id
test-> AND A1.address <> A2.address ;
   name   |        address
----------+------------------------
  Maeda Yiming  |  Tiger gate, Tokyo Metropolitan port area 3-2-29
  Maeda Youmei  |  Tiger gate, Tokyo Metropolitan port area 3-2-92
(2 rows)

test=>

【 example 】 Use non equivalent self connection to find Products Goods with the same price in

test=> DROP TABLE Products;
DROP TABLE
test=> CREATE TABLE Products
test-> (name VARCHAR(16) PRIMARY KEY,
test(> price INTEGER NOT NULL);
CREATE TABLE
test=> DELETE FROM Products;
DELETE 0
test=> INSERT INTO Products VALUES(' Apple ',50);
INSERT 0 1
test=> INSERT INTO Products VALUES(' a mandarin orange ',100);
INSERT 0 1
test=> INSERT INTO Products VALUES(' grapes ',50);
INSERT 0 1
test=> INSERT INTO Products VALUES(' watermelon ',80);
INSERT 0 1
test=> INSERT INTO Products VALUES(' lemon ',30);
INSERT 0 1
test=> INSERT INTO Products VALUES(' Banana ',50);
INSERT 0 1
test=>
test=> SELECT * FROM Products
test-> ;
 name | price
------+-------
  Apple  |    50
  a mandarin orange  |   100
  grapes  |    50
  watermelon  |    80
  lemon  |    30
  Banana  |    50
(6 rows)

test=> SELECT DISTINCT P1.name, P1.price
test-> FROM Products P1, Products P2
test-> WHERE P1.price = P2.price
test-> AND P1.name <> P2.name;
 name | price
------+-------
  Banana  |    50
  Apple  |    50
  grapes  |    50
(3 rows)

test=>

Sort

【 example 】 Set the table according to the price from high to low Products Sort the goods in . We make the same order of goods with the same price , And then their products have two sorting methods , One is to skip the next position , The other is not to skip the next position .

test=> SELECT name, price,
test->     RANK() OVER (ORDER BY price DESC) AS rank_1,
test->     DENSE_RANK() OVER (ORDER BY price DESC) AS rank_2
test->   FROM Products;
 name | price | rank_1 | rank_2
------+-------+--------+--------
  a mandarin orange  |   100 |      1 |      1
  watermelon  |    80 |      2 |      2
  Apple  |    50 |      3 |      3
  grapes  |    50 |      3 |      3
  Banana  |    50 |      3 |      3
  lemon  |    30 |      6 |      4
(6 rows)

test=>

【 example 】 Do not rely on sorting function implementation , If the same order has appeared , Then skip the next position

test=> SELECT P1.name,
test->        P1.price,
test->       (SELECT COUNT(P2.price)
test(>          FROM Products P2
test(>         WHERE P2.price > P1.price) + 1 AS rank_1
test->  FROM Products P1
test->  ORDER BY rank_1;
 name | price | rank_1
------+-------+--------
  a mandarin orange  |   100 |      1
  watermelon  |    80 |      2
  Apple  |    50 |      3
  grapes  |    50 |      3
  Banana  |    50 |      3
  lemon  |    30 |      6
(6 rows)

test=>