Creating a combined timeline based on several timelines in several tables

Question

Hi,
I need to extract a timeline for a customer based on valid_from and valid_to dates in several tables.
For example: I have a table named customers with an id, a valid_from and a valid_to date and a table named contracts with an contrat_name, customer_id and valid_from and valid_to:
CUSTOMERS:
ID | VALID_FROM | VALID_TO
1 | 01.03.2010 | 01.01.4000
CONTRACTS:
CONTRACT_NAME | CUSTOMER_ID | VALID_FROM | VALID_TO
ContractA | 1 | 01.03.2010 | 01.10.2010
ContractB | 1 | 01.10.2010 | 01.01.4000
The following statement would now give me the correct timeline:
select cus.id customer, con.contract_name contract, greatest(cus.valid_from,con.valid_from) valid_from, least(cus.valid_to,con.valid_to) valid_to
from customers cus
inner join contracts con on cus.id = con.customer_id;
CUSTOMER | CONTRACT | VALID_FROM | VALID_TO
1 | ContractA | 01.03.2010 | 01.10.2010
1 | ContractB | 01.10.2010 | 01.01.4000
That works, but I get a problem as soon as I have a point of time where there is no contract for a customer but I still would like to have these periods in my timeline:
Let's assume the following data and the same select statement:
CUSTOMERS:
ID | VALID_FROM | VALID_TO
1 | 01.03.2010 | 01.01.4000
CONTRACTS:
CONTRACT_NAME | CUSTOMER_ID | VALID_FROM | VALID_TO
ContractA | 1 | 01.05.2010 | 01.10.2010
ContractB | 1 | 01.12.2010 | 01.03.2011
What I would now get would be:
CUSTOMER | CONTRACT | VALID_FROM | VALID_TO
1 | ContractA | 01.05.2010 | 01.10.2010
1 | ContractB | 01.12.2010 | 01.03.2011
But what I would like to get is the following:
CUSTOMER | CONTRACT | VALID_FROM | VALID_TO
1 | null | 01.03.2010 | 01.05.2010
1 | ContractA | 01.05.2010 | 01.10.2010
1 | null | 01.10.2010 | 01.12.2010
1 | ContractB | 01.12.2010 | 01.03.2011
1 | null | 01.03.2011 | 01.01.4000
What I do not want to do is to generate a result with contract = null any time there is no contract since I actually want to join the timeline of several different tables into one and it would therefore become very complicated to assume things based on what data can or can not be found in one specific table.
Thanks for any help or ideas,
Regards,
Thomas

Answer 1

Hi, Thomas,
Thomas Schenkeli wrote:
... Is this the way you meant? Because I actually didn't have to change anything about part (b) of the statement since non-matching results were excluded by the where-clause "OR     valid_from     < valid_to" in the final select anyway.You're absolutely right. Sorry about the mistakes in my last message. I'm glad you solved the problem anyway.
Beware of SELECT DISTINCT . Adding DISTINCT causes the system to do extra work, often because the query was doing something wrong and generating too many rows, so you pay for it twice. In this case, the join conditions in (b) are different from (a) and (c), so b is generating too many rows. The DISTINCT in the main query corrects that mistake, but it would be more efficient just to avoid the mikstake in the first place, and use the same join conditions in all 3 branches of the UNION. (You could also factor out the join, doing it once in another sub-query, and then referencing that result set in each branch of the UNION.)
You can get the same results a little more efficiently, with a little less code, this way:
WITH     union_data     AS
     SELECT       MIN (ua.customer_id)     AS customer_id
     ,       NULL               AS contract_name
     ,       MIN (ua.valid_from)     AS valid_from
     ,       MIN (oa.valid_from)     AS valid_to
     ,       'A'                AS origin
     FROM       tmp_customers     ua
     JOIN       tmp_contracts     oa     ON     oa.customer_id     = ua.customer_id
                           AND     oa.valid_from     >= ua.valid_from
                         AND     oa.valid_to     <= ua.valid_to
     GROUP BY  ua.id
    UNION ALL
     SELECT       ub.customer_id
     ,       ob.contract_name
     ,       ob.valid_from
     ,       ob.valid_to
     ,       'B'                AS origin
     FROM       tmp_customers     ub
     JOIN       tmp_contracts     ob     ON     ob.customer_id     = ub.customer_id
                           AND     ob.valid_from     >= ub.valid_from
                         AND     ob.valid_to     <= ub.valid_to
    UNION ALL
     SELECT       uc.customer_id
     ,       NULL               AS contract_name
     ,       oc.valid_to          AS valid_from
     ,       LEAD ( oc.valid_from
                 , 1
                 , uc.valid_to
                 ) OVER ( PARTITION BY  uc.id
                    ORDER BY      oc.valid_from
                     )          AS valid_to
     ,       'C'                AS origin
     FROM       tmp_customers     uc
     JOIN       tmp_contracts     oc     ON     oc.customer_id     = uc.customer_id
                           AND     oc.valid_from     >= uc.valid_from
                         AND     oc.valid_to     <= uc.valid_to
SELECT       *
FROM       union_data
WHERE       contract_name     IS NOT NULL
OR       valid_from     < valid_to
ORDER BY  customer_id
,       valid_from
You may have noticed that this site normally doesn't display multiple spaces in a row.
Whenever you post formatted text on this site, type these 6 characters:
\(small letters only, inside curly brackets) before and after each section of formatted text, to preserve spacing.