Run query for multiple tables and columns - sql

Question

I have the following code, I need to run this code through over 200 tables, can I create a list and a for loop in the variable to go through all the tables and fetch the results for each? Also, how can I include the table name for each row in this query? I also need to go through a list on the column but I can't figure that out from the table list.
DECLARE #table AS VARCHAR(100)
DECLARE #column as varchar(50)
set #table = 'x'
set #column = 'y'
declare #query as varchar(max)
set #query = ' SELECT CAST(MONTH(' + #column + ') AS VARCHAR(2)) + ''-'' + CAST(YEAR(' + #column + ') AS VARCHAR(4)) as date, count(*) as SRC_CNT
FROM ' + #table +
' WHERE ' + #column + ' >= ''2018-01-01'' AND ' + #column + '< ''2021-12-01''
group BY CAST(MONTH(' + #column + ') AS VARCHAR(2)) + ''-'' + CAST(YEAR(' + #column + ') AS VARCHAR(4))
order by date;'
exec(#query)

Answer 1

Rather than loop, I would create a dynamic batch, with all the statements you need, which you can then execute. You can use your best friend to debug the statement(s) if needed:
DECLARE #Column sysname = N'y',
#DateFrom date = '20180101',
#DateTo date = '20211201';
DECLARE #SQL nvarchar(MAX),
#CRLF nchar(2) = NCHAR(13) + NCHAR(10);
SELECT #SQL = STRING_AGG(N'SELECT RIGHT(CONVERT(varchar(10),' + QUOTENAME(#Column) + N',105),7) AS [Date],' + #CRLF +
--N' N' + QUOTENAME(t.[name],'''') + N' AS TableName,' + #CRLF + --Uncomment this line if you need it.
N' COUNT(*) AS SRC_CNT' + #CRLF +
N'FROM ' + QUOTENAME(s.[name]) + N'.' + QUOTENAME(t.[name]) + #CRLF +
N'WHERE ' + QUOTENAME(#Column) + N' >= #DateFrom AND ' + QUOTENAME(#Column) + N' < #DateTo' + #CRLF +
N'GROUP BY RIGHT(CONVERT(varchar(10),' + QUOTENAME(#Column) + N',105),7)' + #CRLF +
N'ORDER BY [date];',#CRLF)
FROM sys.schemas s
JOIN sys.tables t ON s.schema_id = t.schema_id
WHERE EXISTS (SELECT 1
FROM sys.columns c
WHERE c.object_id = t.object_id
AND c.[name] = #Column);
--PRINT #SQL; --Your best friend
EXEC sys.sp_executesql #SQL, N'#DateFrom date, #DateTo date', #DateFrom, #DateTo;

Answer 2

Given tables and sample data:
CREATE TABLE dbo.x(y date, r int);
CREATE TABLE dbo.y(y date, r int);
CREATE TABLE dbo.z(y date, r int);
INSERT dbo.x(y,r) VALUES('20180105',5),
('20180107',6),('20180509',7);
INSERT dbo.y(y,r) VALUES('20180905',5),
('20181007',6),('20181009',7);
INSERT dbo.z(y,r) VALUES('20180605',5),
('20180607',6),('20180609',7);
This dynamic SQL can be generated:
DECLARE #col sysname = N'y',
#StartDate date = '20180101',
#EndDate date = '20211201';
DECLARE #sql nvarchar(max) = N'',
#base nvarchar(max) = N'SELECT [Table Name] = $tblQ$,
[date] = DATEFROMPARTS(YEAR($col$), MONTH($col$), 1),
SRC_CNT = COUNT(*)
FROM dbo.$tbl$ WHERE $col$ >= #s AND y < #e
GROUP BY YEAR($col$), MONTH($col$)';
SELECT #sql = STRING_AGG(REPLACE(REPLACE(REPLACE
(#base, N'$tblQ$', QUOTENAME(t.name, char(39))
),N'$tbl$',QUOTENAME(t.name)), N'$col$', #col), N'
UNION ALL
') + ' ORDER BY [date];'
FROM sys.tables AS t
WHERE EXISTS (SELECT 1 FROM sys.columns
WHERE [object_id] = t.[object_id]
AND name = #col);
PRINT #sql;
EXEC sys.sp_executesql #sql,
N'#s date, #e date',
#StartDate, #EndDate;
Which produces a query like this (only tables that actually have that column name, could be made even safer by making sure they're using a date/time type):
SELECT [Table Name] = 'x',
[date] = DATEFROMPARTS(YEAR(y), MONTH(y), 1),
SRC_CNT = COUNT(*)
FROM dbo.[x] WHERE y >= #s AND y < #e
GROUP BY YEAR(y), MONTH(y)
UNION ALL
SELECT [Table Name] = 'y',
[date] = DATEFROMPARTS(YEAR(y), MONTH(y), 1),
SRC_CNT = COUNT(*)
FROM dbo.[y] WHERE y >= #s AND y < #e
GROUP BY YEAR(y), MONTH(y)
UNION ALL
SELECT [Table Name] = 'z',
[date] = DATEFROMPARTS(YEAR(y), MONTH(y), 1),
SRC_CNT = COUNT(*)
FROM dbo.[z] WHERE y >= #s AND y < #e
GROUP BY YEAR(y), MONTH(y) ORDER BY [date];
That generates output like this:
Table Name
date
SRC_CNT
x
2018-01-01
2
x
2018-05-01
1
z
2018-06-01
3
y
2018-09-01
1
y
2018-10-01
2
Example db<>fiddle
If you really want yyyy-MM instead of yyyy-MM-dd on the output, you can just change this line in the declaration of #base:
[date] = CONVERT(char(7), DATEFROMPARTS(YEAR($col$), MONTH($col$), 1), 120),
And I realized the requirement was MM-yyyy, in which case:
[date] = RIGHT(CONVERT(char(10),
DATEFROMPARTS(YEAR($col$), MONTH($col$), 1), 105), 7),

Answer 3

I usually do the following for tables:
SELECT TABLE_NAME
FROM [<DATABASE_NAME>].INFORMATION_SCHEMA.TABLES
WHERE TABLE_TYPE = 'BASE TABLE'
You said you could handle columns from there.