SQL连接两个表并根据另一列及其累积计数映射公用值

Question

I have the following two tables,

Table 1 :

id    document
--------------
    A2      B200
    A2      B6
    A2      B2
    A2      B3
    A3      B2
    A3      B400
    A5      B100
    A5      B500 
    A6      B6
    A7      B200
    A8      B6
    A8      B2
    A8      B3    
    A8      C1

table 2:

id    name
--------------
A1      Jack
A2      Martin
A3      Jack
A4      Thomas
A5      Jack
A6      Thomas
A7      Thomas
A8      John
A9      John
A10     Kate

My filter is the following that should compare the document column and filter based on this list:

WHERE table1.document IN (B2,B400,B100,B500,B200,B6,B2,B3)

The result should be like:

name1   name2   freq
--------------------
Jack    John    1
Martin  Jack    1
Martin  Thomas  2
Martin  John    3
Thomas  John    1

Some explanations:

We need to build a results table that maps the name s that have document s in common and its frequency. First we need to filter the document list using the WHERE IN statement to get the list of documents we want to map.

Then we get the list of documents that the count is more than one, because that document is shared between at least two ids.

Then we look up the name s of those ids in table2 and put them in results table and the count of the documents they had in common.Some names have multiple ids, so if we hit on those, then we add to the count.

For example, document B6 is assigned to ids of A2 and A6, so they have this document in common, we create an entry in the results table, putting their corresponding names as name1 and name2 (order doesn't matter) and give it frequency of 1. But if we look further, we see that document B200 is shared by A2 and A7, when we look up the names of these two ids, we see that we already have an entry with those names, because they corresond to Martin and Thomas as in the previous one, so we add to their count, so it becomes 2.

Another example is that of the documents B6, B2, B3 that are shared by A2 and A8, Martin and John, so we create an entry for these two, and the count will be 3.

Here is a demo data.

Answer 1

You have to do the joins two times and then group by both names:

SELECT t2a.name, t2b.name, COUNT(*)
FROM Table1 t1a
INNER JOIN Table2 t2a ON t2a.id = t1a.id
INNER JOIN Table1 t1b ON t1b.document = t1a.document
INNER JOIN Table2 t2b ON t2b.id = t1b.id AND t2a.name < t2b.name
GROUP BY t2a.name, t2b.name

Answer 2

Do like this

CREATE TABLE Table1
    (id varchar(10), document varchar(10))

INSERT INTO Table1
VALUES
    ('A2', 'B200'),
    ('A2', 'B6'),
    ('A2', 'B2'),
    ('A2', 'B3'),
    ('A3', 'B2'),
    ('A3', 'B400'),
    ('A5', 'B100'),
    ('A5', 'B500'), 
    ('A6', 'B6'),
    ('A7', 'B200'),
    ('A8', 'B6'),
    ('A8', 'B2'),
    ('A8', 'B3'),    
    ('A8', 'C1')

CREATE TABLE Table2
    (id varchar(3), name varchar(10))

INSERT INTO Table2
VALUES
    ('A1', 'Jack'),
    ('A2', 'Martin'),
    ('A3', 'Jack'),
    ('A4', 'Thomas'),
    ('A5', 'Jack'),
    ('A6', 'Thomas'),
    ('A7', 'Thomas'),
    ('A8', 'John'),
    ('A9', 'John'),
    ('A10','Kate')


;WITH docs AS (

    SELECT id, document FROM Table1
    WHERE table1.document IN ('B2','B400','B100','B500','B200','B6','B2','B3')
)
, user_docs AS (
    SELECT t2.id, t2.name, docs.document FROM docs
    INNER JOIN Table2 t2 ON t2.id = docs.id
)
, freq AS (
    SELECT ud.name, ud1.name name1, count(*) freq FROM user_docs ud
    INNER JOIN user_docs ud1 ON ud1.document = ud.document AND ud1.name != ud.name AND ud1.id > ud.id
    GROUP BY ud.name, ud1.name
)
SELECT * FROM freq


DROP TABLE Table1, Table2