Simple explanation of database table connection

The hardest part about relational databases , It's modeling （model）.

Intricate data , You need to model , To be stored in a database . So-called " Model " There are two things ： Entity （entity）+ Relationship （relationship）.

Entities refer to the actual objects , With its own properties , Containers that can be understood as a set of related properties . Relationship is the connection between entities , Usually it can be divided into " one-on-one "、" One to many " and " Many to many " Other types .

In a relational database , Each entity has its own table （table）, All properties are fields of this table （field）, Table to table according to the associated fields " Connect "（join） together . therefore , Table join is the core problem of relational database .

There are several types of table connections .

• Internal connection （inner join）
• External connection （outer join）
• Left connection （left join）
• The right connection （right join）
• Full connection （full join）

before , Many articles use Venntu （ The set operation of two circles ）, Explain the differences between the different connections .

Last week, , I read An article , Think there's a better way to explain it than Venn diagram . I found that it was , Explain it in a different way , Easier to understand .

So-called " Connect ", Two tables are based on the associated fields , Combine into a data set . The problem is , The values of the associated fields in the two tables are often inconsistent , If the associated fields do not match , How to deal with ？ such as , surface A Including Zhang San and Li Si , surface B Including Li Si and Wang Wu , The only match is the record of Li Si .

It's easy to see , There are four ways to deal with it .

• Only records matching two tables are returned , This is called internal connection （inner join）.
• Return matching records , And the watch A Redundant records , This is called left connection （left join）.
• Return matching records , And the watch B Redundant records , This is called right connection （right join）.
• Return matching records , And the watch A And table B Their own redundant records , This is called full connection （full join）.

The figure below shows the four connections . I think , This picture is easier to understand than Venn's , It's more accurate .

Above picture , surface A The record of is 123, surface B The record of is ABC, Color means match . Return the result , If the other table does not have a matching record , Then use null fill .

These four connections , It can be divided into two categories ： Internal connection （inner join） Indicates that only matching records are included , External connection （outer join） Indicates that there are also mismatched records . therefore , Left connection 、 The right connection 、 All connections belong to external connections .

These four connected SQL The statement is as follows .

SELECT * FROM A  
INNER JOIN B ON A.book_id=B.book_id;

SELECT * FROM A
LEFT JOIN B ON A.book_id=B.book_id;

SELECT * FROM A
RIGHT JOIN B ON A.book_id=B.book_id;

SELECT * FROM A
FULL JOIN B ON A.book_id=B.book_id;

above SQL You can also add where Conditional clause , Screening records , For example, just return to the table A It doesn't match the table B The record of .

SELECT * FROM A
LEFT JOIN B
ON A.book_id=B.book_id
WHERE B.id IS null;

Another example , Go back to the table A Or table B All mismatched records .

SELECT * FROM A
FULL JOIN B
ON A.book_id=B.book_id
WHERE A.id IS null OR B.id IS null;

Besides , There is also a special connection , be called " Cross connect "（cross join）, It means the watch A And table B There are no associated fields , At this time, the table A（ share n Bar record ） And watch B （ share m Bar record ） After connection , Will produce a sheet containing n x m New table of records （ See the picture below ）.