I want to QUERY to generate a table like below:
Tablename||noofrows||noofcolumns||PRIMARKEYCOL(IF ANY for the table)
xyz. 590. 11. xyz_id
bcd. 934 15 null
...
...
So far I was able to do this until now in 2 query:
Query 1:
select a.table_name,count_rows(a.table_name) total_rows,count(b.column_name) total_cols from user_tables a,
,user_tab_columns b
where a.table_name =b.table_name
and a.table_name not like('amp%')
group by a.table_name;
note:Count_rows() is function to calculate rows as stats are not up to date
query 2:
select b.table_name b.column_name PRIMKEY_COL FROM user_constraints a,user_cons_columns b
where
a.constraint_type = 'P'
and a.constraint_name=b.constraint_name
and a.table_name=b.table_name
and b.table_name not like ('amp%');
Problem
I need to merge this table to one query (as shown in example above) so that I can represent the data in one table. My issue in clubbing the table is, with joins and how to make sure table without any primary keys are represent because if I just directly give constraint type ='p' in the where clause of the join I see that it only shows table with Primarykeys I am not able to figure this out.
Primary key can have more than just a single column, so - if you'd want to return them all, you'd either have to "aggregate" them, somehow (listagg looks like a natural choice, but - in 12c - you can't fetch distinct list of columns which you might need because of joining those tables duplicates appear).
As you already use a function to return number of rows (you'd probably rather regularly collect statistics, though, but yes - if tables are frequently modified (inserts and deletes), counting on-the-fly is the way to go.
So, alternative approach with 2 functions and 1 simple query. Yes, I know - context switching and stuff, but - see if it does any good.
This is what you already have (see if your and my code look similar):
SQL> create or replace function count_rows (par_table_name in varchar2)
2 return number
3 is
4 -- return number of rows in PAR_TABLE_NAME
5 retval number;
6 begin
7 execute immediate 'select count(*) from ' ||
8 dbms_assert.sql_object_name(par_table_name) into retval;
9 return retval;
10 end;
11 /
Function created.
This is new (to simplify fetching distinct list of primary key columns):
SQL> create or replace function pk_cols (par_table_name in varchar2)
2 return varchar2
3 is
4 -- return list of primary key columns for PAR_TABLE_NAME, sorted by column's
5 -- position within the primary key constraint
6 retval varchar2(100);
7 begin
8 select listagg(b.column_name, ', ') within group (order by b.position)
9 into retval
10 from user_constraints a join user_cons_columns b on b.constraint_name = a.constraint_name
11 where upper(a.table_name) = dbms_assert.sql_object_name(upper(par_table_name))
12 and a.constraint_type = 'P';
13 return retval;
14 end;
15 /
Function created.
Finally, that simple query I mentioned:
SQL> select a.table_name,
2 count_rows(a.table_name) total_rows,
3 max(a.column_id) total_cols,
4 pk_cols(a.table_name) primkey_cols
5 from user_tab_columns a
6 group by a.table_name;
TABLE_NAME TOTAL_ROWS TOTAL_COLS PRIMKEY_COLS
--------------- ---------- ---------- -------------------------
DEPT 4 3 DEPTNO
SO_TEST 0 1
TEST 7 3 ORD, TERMREGIONAL
EMP 14 8 EMPNO
TRAINING 0 5 TRAINING_PLACE_ID
TEMP_SE 10 3
<snip>