Reputation: 23
I come here because of a problem that I am not experienced enough to solve, in an Oracle Database. Let me explain:
I have a table that we'll call Attributes, containing 3 columns : ID, the id of the attribute, -EDIT: Entity_ID as well, the entity it refers to /EDIT-, Table_name, containing the name of the table in which the value of the attribute is stored, and Column_name, containing the name of the column in that table in which is the value is stored.
Every tables referenced in the column Table_name, contains the same column names (such as Value1,Value2, etc..) except for the first one that references another entity (Entity_id), to which the attribute is linked to.
My goal is to build a query that selects every attribute (based on its id) and its value. But what I don't know is how to query that since the name of table and of the column changes. Is there a way to use variables? But if so, how can I put them in the query so that it automatically changes for each row?
EDIT
ATTRIBUTES table
ID ENTITY_ID TABLE_NAME COLUMN_NAME
---------- -------------- ------------ -----------
1 3 Values_A Value_1
2 2 Values_B Value_3
3 2 Values_A Value_2
VALUES_A table
ENTITY_ID Value_1 Value_2 Value_3
---------- -------------- ------------ -----------
1 Monday 42 Green
2 Sunday 3000 Blue
3 Wednesday 1 Black
VALUES_B table
ENTITY_ID Value_1 Value_2 Value_3
---------- -------------- ------------ ------------
1 Tuesday 26 Green
2 Saturday 3 Red
3 Wednesday 15 White
So the result I'm looking for would be:
RESULT:
ID Value
--------- -----------
1 Wednesday
2 Red
3 3000
Sorry if this is painful to watch, it was as painful to make (didn't find how to format it better)
Upvotes: 2
Views: 1965
Reputation: 16001
Using Peter M's query to build the SQL text, and then harnessing the dark power of XML:
create table attributes (id, entity_id, table_name, column_name)
as
select 1, 3, 'VALUES_A', 'VALUE_1' from dual union all
select 2, 2, 'VALUES_B', 'VALUE_3' from dual union all
select 3, 2, 'VALUES_A', 'VALUE_2' from dual;
create table values_a (entity_id, value_1, value_2, value_3)
as
select 1, 'Monday', 42, 'Green' from dual union all
select 2, 'Sunday', 3000, 'Blue' from dual union all
select 3, 'Wednesday', 1, 'Black' from dual;
create table values_b (entity_id, value_1, value_2, value_3)
as
select 1, 'Tuesday', 26, 'Green' from dual union all
select 2, 'Saturday', 3, 'Red' from dual union all
select 3, 'Wednesday', 15, 'White' from dual;
Query:
with queries as
( select table_name, column_name, entity_id
, 'select '|| column_name || ' as c from ' || table_name ||
' where entity_id = ' || entity_id ||
case
when id = max_id then ''
else ' union all '
end as sqltext
from
( select a.*, max(a.id) over (order by id) max_id from attributes a ) )
select table_name, column_name, entity_id
, extractvalue(xmltype(dbms_xmlgen.getxml(sqltext)),'/ROWSET/ROW/C') as sql_result
from queries;
Results:
TABLE_NAME COLUMN_NAME ENTITY_ID SQL_RESULT
---------- ----------- ---------- ---------------------------------------------------
VALUES_A VALUE_1 3 Wednesday
VALUES_B VALUE_3 2 Red
VALUES_A VALUE_2 2 3000
Upvotes: 2
Reputation: 192
this is ugly
but it would generate a sql statement that you could run, maybe it is not what you want
select 'select '|| column_name || ' from ' || table_name || ' where entity_id = ' || entity_id || case when id = max_id then '' else ' union all ' end
from
(select a.*, max(a.id) over (order by id) max_id
from attributes a)
Upvotes: 2