I have these two tables. Subscriber table contains SubscriberID and his Phone number. NetworkP2P contents SubscriberID, to who he has called AddresseeNumber and when the call started, ended.

TABLE Subscriber 
(
    SubscriberID int,
    PhoneNumber char(10)
)

TABLE NetworkP2P 
(
    SubscriberID int,
    AddresseeNumber char(10),
    CallStart datetime,
    CallEnd datetime
)

Example data:

Subscriber

SubscriberID | PhoneNumber
1              613-555-0156
2              613-555-0112
3              613-555-0119
4              613-555-0182

NetworkP2P

SubscriberID | AddresseeNumber | CallStart               | CallEnd
1              613-555-0182      2013-02-03 08:27:00.000   2013-02-03 08:30:54.000
1              613-555-0119      2013-02-03 10:11:26.000   2013-02-03 10:14:26.000
1              613-555-3333      2013-02-03 15:13:58.000   2013-02-03 15:18:13.000
2              613-555-0156      2013-02-05 23:21:50.000   2013-02-05 23:24:50.000
2              613-555-0119      2013-04-05 23:21:50.000   2013-04-05 23:26:50.000
3              613-555-0112      2013-06-05 23:21:50.000   2013-06-05 23:26:50.000

Not all Phone numbers match the ones that Subscribers have.

How would I approach this, if I wish to have which outputs subscribers who have been in contact with each other for a given month, i.e. subscriber1 called subscriber2 and vice versa?

Example of the desired output:

 Caller | Reciever | CallerPhone | RecieverPhone
 1        4          613-555-0156  613-555-0182
 1        3          613-555-0156  613-555-0119
 2        1          613-555-0112  613-555-0156
 2        3          613-555-0112  613-555-0119
 3        2          613-555-0119  613-555-0112

CodePudding user response：

Your model is not wrong as suggested in comments - your NetworkP2P might be missing only one column (Reciever), which could work as optional foreign key (to satisfy your sentence "Not all Phone numbers match the ones that Subscribers have."), but it looks more log than relational table.

Query for your desired output:

select
    caller.SubscriberID as Caller,
    reciever.SubscriberID as Reciever,
    caller.PhoneNumber as CallerPhone,
    nt.AddresseeNumber as RecieverPhone, -- better than reciever.PhoneNumber because you can use left join recievers to get missing substribers
    nt.CallStart,
    nt.CallEnd
from Subscriber as caller
inner join NetworkP2P as nt
    on nt.SubscriberID = caller.SubscriberID
inner join Subscriber as reciever
    on nt.AddresseeNumber = reciever.PhoneNumber
-- where '2021-02-03' between nt.CallStart and nt.CallEnd -- calls in 2021-02-03
-- where nt.CallStart between '2021-02-01' and '2021-02-28' -- calls in specific interval
-- where eomonth(nt.CallStart) = eomonth('2021-02-01') -- calls in specific month

You should always use database schema for all your tables (for example dbo.Subscriber).

CodePudding user response：

Two joins to Subscriber should do it.

SELECT 
  nw.SubscriberID AS Caller
, rcvr.SubscriberID AS Receiver
, callr.PhoneNumber AS CallerPhone
, nw.AddresseeNumber AS ReceiverPhone
FROM NetworkP2P nw
JOIN Subscriber callr ON callr.SubscriberID = nw.SubscriberID
LEFT JOIN Subscriber rcvr ON rcvr.PhoneNumber = nw.AddresseeNumber
WHERE rcvr.SubscriberID IS NOT NULL
ORDER BY Caller, Receiver

Demo on db<>fiddle here