Dynamic SQL Server Query loop through schema find primary key duplicate - sql

Question

EDIT: There seems to be some confusion around the create table statements. These are there solely as a demonstration of what tables *might come in to our synapse instance, not as actual code that will run. The important part of the question is contained in the latter half.
I am trying to create a stored procedure that loops through every table in a supplied schema and outputs the count of duplicate primary key rows for each table. Assume that the data is being supplied from elsewhere and the primary keys are not being enforced. For example I may have three tables in the stack schema:
CREATE TABLE stack.table1(
id int,
name NVARCHAR(MAX),
color NVARCHAR(20)
PRIMARY KEY (id))
INSERT INTO stack.table1 VALUES(1,'item1','yellow')
(2,'item2','blue')
(2,'item2','blue')
CREATE TABLE stack.table2(
id int,
name NVARCHAR(MAX),
size NVARCHAR(1)
PRIMARY KEY (id,size))
INSERT INTO stack.table2 VALUES(1,'item1','L')
(2,'item2','M')
(3,'item2','S')
CREATE TABLE stack.table3(
id int,
name NVARCHAR(MAX),
weight NVARCHAR(20)
PRIMARY KEY (id))
INSERT INTO stack.table1 VALUES(1,'item1','200lb')
(2,'item2','150lb')
(3,'item2','125lb')
I want to supply a variable to a stored procedure to indicate the schema (in this case 'stack') and have that procedure spit out a table with the names of the tables in the schema and the counts of duplicate primary key rows. So in this example a stored procedure called 'loopcheck' would look like this:
Query:
EXEC loopcheck #schema = 'stack'
Output:
table
duplicate_count
table1
1
table2
0
table3
0
I am using an Azure Synapse instance so there are several functions that are not available (such as FOR XML PATH and others.) Since each table may have a single column primary key or a composite primary key I need to join to the system provided tables to get primary key info. My general idea was like so:
CREATE procedure loopcheck #schema= NVARCHAR(MAX)
AS
BEGIN
create table #primarykey(
SCHEMA_NAME nvarchar(400),
TABLE_NAME nvarchar(500),
COLUMN_NAME nvarchar(500)
)
insert into #primarykey
select l.TABLE_SCHEMA,
l.TABLE_NAME,
l.COLUMN_NAME
from INFORMATION_SCHEMA.CONSTRAINT_COLUMN_USAGE l
inner join INFORMATION_SCHEMA.TABLE_CONSTRAINTS t on l.constraint_Name = t.CONSTRAINT_NAME
where
l.table_schema = #schema
CREATE TABLE #groupBy2(
TABLE_NAME nvarchar(50),
groupby nvarchar(200)
)
INSERT INTO #groupBy2
SELECT TABLE_NAME, STRING_AGG(CONVERT(NVARCHAR(max), COLUMN_NAME), ',') as groupby
FROM #primarykey
GROUP BY TABLE_NAME
DECLARE #currentTable NVARCHAR(MAX)=''
DECLARE #currentGroup NVARCHAR(MAX)=''
create table #work4(
TABLE_NAME nvarchar(400),
COUNT int)
DECLARE #final NVARCHAR(MAX)=
'INSERT INTO #work4
SELECT '+#currentTable+', COUNT(*) FROM '+#currentTable+'GROUP BY'+#currentGroup
WHILE (SELECT COUNT(*) FROM #groupby2)>0
BEGIN
SET #currentTable =(SELECT TOP 1 TABLE_NAME FROM #groupby2 ORDER BY TABLE_NAME)
SET #currentGroup =(SELECT TOP 1 groupby FROM #groupby2 ORDER BY TABLE_NAME)
exec #final
DELETE #groupby2 where TABLE_NAME =#currentTable
END
END
This code gives me an error:
Incorrect syntax near 'SELECT'
but doesn't give me the actual line it has the error on.

Answer 1

Your primary issue was syntax errors: parameter declaration should not have = between name and type name, and missing spaces in the dynamic SQL.
Also
A schema name (or any object) can be up to nvarchar(128), you can use the alias sysname
You don't need to do any loops or use temp tables, you can build one big dynamic statement to execute
CREATE procedure loopcheck
#schema sysname
AS
DECLARE #sql nvarchar(max) = (
SELECT STRING_AGG(CAST('
SELECT
TableName = ' + QUOTENAME(t.name, '''') + ',
IndexName = ' + QUOTENAME(i.name, '''') + ',
Duplicates = COUNT(*)
FROM (
SELECT 1 n
FROM ' + QUOTENAME(s.name) + '.' + QUOTENAME(t.name) + ' t
GROUP BY ' + cols.agg + '
HAVING COUNT(*) > 1
) t
'
AS nvarchar(max)), 'UNION ALL')
FROM sys.tables t
JOIN sys.schemas s ON s.schema_id = t.schema_id
AND s.name = #schema
JOIN sys.indexes i ON i.object_id = t.object_id
AND i.is_unique = 1
CROSS APPLY (
SELECT STRING_AGG('t.' + QUOTENAME(c.name), ', ')
FROM sys.index_columns ic
JOIN sys.columns c ON c.column_id = ic.column_id AND c.object_id = t.object_id
WHERE ic.index_id = i.index_id
AND ic.object_id = i.object_id
AND ic.is_included_column = 0
) cols(agg)
);
PRINT #sql; -- for testing
EXEC sp_executesql #sql;
db<>fiddle
I feel like there might be a slightly more efficient method using GROUPING SETS in cases where there are multiple unique indexes/constraints on a single table, but I'll leave that to you.