从多个表中选择超出范围的日期，列名和时间戳

Question

I have multiple tables (10+) in which I am trying to pull any dates outside a specified date range, the column name (which holds the out-of-range date), and the MODIFIED_BY of that date's row. Below are two example tables and the required output. The date range I'm checking for in table A is a birthdate greater than 1990. In table B, an effective date greater than 2017

TABLE_A

|---------------------|------------------|------------------|
|      ACCT_ID        |    BIRTHDATE     |    MODIFIED_BY   |
|---------------------|------------------|------------------|
|          12         |     04-MAR-96    |     user1234     |
|---------------------|------------------|------------------|
|          13         |     10-MAY-79    |     user9999     |
|---------------------|------------------|------------------|

TABLE_B

|---------------------|------------------|------------------|
|      ACCT_ID        |   EFFECTIVE_DT   |    MODIFIED_BY   |
|---------------------|------------------|------------------|
|          12         |     01-JAN-15    |     user9876     |
|---------------------|------------------|------------------|
|          13         |     01-APR-17    |     user1111     |
|---------------------|------------------|------------------|

OUTPUT

|---------------------|------------------|------------------|
|      ACCT_ID        |     BAD_DATE     |    MODIFIED_BY   |
|---------------------|------------------|------------------|
|          12         |     04-MAR-96    |     user1234     |
|---------------------|------------------|------------------|
|          13         |     01-APR-17    |     user1111     |
|---------------------|------------------|------------------|

I'm not looking for the SQL code in this instance, but rather a solution. I could probably accomplish this using UNION, but since not all joins are on the acct_id this could be tedious and slow when doing it for many tables. I was able to accomplish this using two columns, one being the account ID and another called BAD_DATE in which I appended the column name and modified_by to the case return.

Any help is appreciated.

Thanks!

Answer 1

I don't know how to do that without UNIONing those tables; might be a tedious job, but hey - you'll do it only once. Here's an example:

SQL> create table table_a (acct_id number, birthdate date, modified_by varchar2(20));

Table created.

SQL> create table table_b (some_id number, effective_dt date, username varchar2(20));

Table created.

SQL> insert all
  2    into table_a values (12, date '1996-03-04', 'user1234')
  3    into table_a values (13, date '1979-05-10', 'user9999')
  4    into table_b values (12, date '2015-01-01', 'user9876')
  5    into table_b values (12, date '2017-04-01', 'user1111')
  6  select * From dual;

4 rows created.

SQL> create or replace view wc as
  2  select 'table_a' source, acct_id id, birthdate bad_date, modified_by from table_a union all
  3  select 'table_b' source, some_id, effective_dt, username from table_b;

View created.

SQL>
SQL> select *
  2  from wc
  3  where (source = 'table_a' and bad_date > date '1990-01-01')
  4     or (source = 'table_b' and bad_date > date '2017-01-01');

SOURCE          ID BAD_DATE   MODIFIED_BY
------- ---------- ---------- --------------------
table_a         12 04.03.1996 user1234
table_b         12 01.04.2017 user1111

SQL>

Answer 2

Found a solution!

I am using a case statement to determine the bad dates. If I add two additional case statements that have the same criteria, but instead have one return a string of the column name and the other return the modified_by ID, it works!