T-SQL (Azure Synapse) 基于未知数量主键的动态分组

Question

I'm tasked with creating a sp that will loop through every table in a supplied schema and find any that do not have unique primary keys (duplicate primary keys). Some tables may have only one column creating the primary key while others may have composite keys with an unknown number of columns.

For example:

CREATE DATABASE test
USE TEST
CREATE PROC duplicateCheck @schemaChoice nvarchar(60)= 'test'
CREATE TABLE onePrimary(

      id int,
      orders int,
      orderCustomer varchar(20)
      PRIMARY KEY (id)
    )
CREATE TABLE twoPrimary(
      id int,
      item int,
      color varchar(20),
      size varchar(20),
      PRIMARY KEY(id,item)
    )
INSERT INTO onePrimary
VALUES(1,2,'Brad'),
(2,1,'Lenny'),
(3,1,'Rachel')

INSERT INTO twoPrimary
VALUES(1,1,'yellow','large'),
(2,1,'blue','small')


create table #primaryKey(
       TABLE_NAME varchar(20),
       COLUMN_NAME varchar(20)
    )
SELECT @schemaChoice,
       x.TABLE_NAME,
       x.COLUMN_NAME,
    FROM INFORMATION_SCHEMA.CONSTRAINT_COLUMN_USAGE x, INFORMATION_SCHEMA.TABLE_CONSTRAINTS y
    WHERE x.TABLE_SCHEMA = @schemaChoice
    and y.CONSTRAINT_TYPE like 'PRIMARY KEY'

That's where I get stuck.

Answer 1

You can find this information in Sql Server (and Azure Synapse) using the system views.

The following will list all user tables, the name of the primary key and how many key columns are defined.

select Schema_Name(t.schema_id) SchemaName, t.name TableName, i.name PrimaryKey, Count(*) KeyColumnCount
from sys.tables t
join sys.indexes i on i.object_id = t.object_id and i.is_primary_key = 1
join sys.index_columns ic on ic.object_id = t.object_id and ic.index_id = i.index_id
group by t.name, i.name, Schema_Name(t.schema_id)
order by KeyColumnCount

You can add having count(*)>1 after the group by to only list tables with more than 1 key column, and of course any other filtering you need.