SQL with COUNTs and SUMs with cross-UNION GROUP BY

Question

SQL guru needed, this is killing my neurons. I have a db structure (simplified) like this:

DESC documents;
    id          INT PK
    bill_id     INT FK
    dtype       INT -- 1=receipts, 0=invoices
    total       DECIMAL

DESC bills;
    id          INT PK
    waiter_id   INT FK

DESC waiters;
    id          INT PK
    name        VARCHAR
    surname     VARCHAR

Quite self-explanatory, but I need to count and sum the totals for all the receipts (documents.dtype=1) and the invoices (documents.dtype=0), grouped by waiter. I did two SELECTs:

SELECT
        B.waiter_id,
        WA.name, 
        WA.surname, 
        COUNT(D.id) AS Receipts, 
        SUM(D.total) AS TotReceipts
FROM    
    documents D
    JOIN bills B ON (B.id = D.bill_id)
    JOIN waiters WA ON (WA.id = B.waiter_id)
WHERE 
    D.dtype = 1          
GROUP BY 
    waiter_id;

Fine, I get:

1, 'Mario',   'Rossi',   6, 485.20
2, 'Luigino', 'Bianchi', 1, 456.00

the did the second SELECT, just changing the documents.dtype to 0:

SELECT
        B.waiter_id,
        WA.name, 
        WA.surname, 
        COUNT(D.id) AS Invoices, 
        SUM(D.total) AS TotInvoices
FROM    
    documents D
    JOIN bills B ON (B.id = D.bill_id)
    JOIN waiters WA ON (WA.id = B.waiter_id)
WHERE 
    D.dtype = 0 
GROUP BY 
    waiter_id;

And now I get:

1, 'Mario', 'Rossi', 1, 38.00

Now I can UNION the two SELECTSs

SELECT
        B.waiter_id,
        WA.name, 
        WA.surname, 
        COUNT(D.id) AS Receipts, 
        SUM(D.total) AS TotReceipts
FROM    
    documents D
    JOIN bills B ON (B.id = D.bill_id)
    JOIN waiters WA ON (WA.id = B.waiter_id)
WHERE 
    D.dtype = 1      
GROUP BY 
    waiter_id
UNION SELECT
        B.waiter_id,
        WA.name, 
        WA.surname, 
        COUNT(D.id) AS Invoices, 
        SUM(D.total) AS TotInvoices
FROM    
    documents D
    JOIN bills B ON (B.id = D.bill_id)
    JOIN waiters WA ON (WA.id = B.waiter_id)
WHERE 
    D.dtype = 0 
GROUP BY 
    waiter_id;

And I get:

1, 'Mario',   'Rossi',   6, 485.20
2, 'Luigino', 'Bianchi', 1, 456.00
1, 'Mario',   'Rossi',   1, 38.00

mmm, correct but I need the rows grouped by waiter cross-union! That is I want a single line for the waiter Mario:

wid wname       wsurname    receipts    totreceipts     invoices        totinvoices
1,  'Mario',    'Rossi',    6,          485.20          1               38.0
2,  'Luigino',  'Bianchi',  1,          456.00          0               0.0

That would be just great, but I'd also like two more total columns to sum-up the numbers like:

wid wname       wsurname    receipts    totreceipts     invoices        totinvoices     docs    totdocs
1,  'Mario',    'Rossi',    6,          485.20          1               38.0            7       523.20
2,  'Luigino',  'Bianchi',  1,          456.00          0               0.0             1       456.00

And that would be super-ultra-cool.

Answer 1

You could move the conditon from the where clause to a case statement, like:

SELECT
        B.waiter_id,
        WA.name, 
        WA.surname, 
        SUM(case when d.dtype = 1 then 1 end) AS Receipts, 
        SUM(case when d.dtype = 1 then D.total end) AS TotReceipts,
        SUM(case when d.dtype = 0 then 1 end) AS Invoices, 
        SUM(case when d.dtype = 0 then D.total end) AS TotInvoices
FROM    
    documents D
    JOIN bills B ON (B.id = D.bill_id)
    JOIN waiters WA ON (WA.id = B.waiter_id)
GROUP BY 
    waiter_id

Answer 2

This query returns what you asked

 SELECT
    B.waiter_id,
    WA.name, 
    WA.surname,
    (select count(id)             from documents where dtype = 0 and bill_id = B.id) AS Invoices, 
    (select isnull(sum(total), 0) from documents where dtype = 0 and bill_id = B.id) AS TotInvoices, 
    (select count(id)             from documents where dtype = 1 and bill_id = B.id) AS Receipts, 
    (select isnull(sum(total), 0) from documents where dtype = 1 and bill_id = B.id) AS TotReceipts
  FROM bills B
 inner join waiters WA on WA.id = B.waiter_id

To summarize all the totals, you can make a inner select in order to sum up the totals as:

select  data.*, 
        data.Invoices + data.Receipts as docs,
        data.TotInvoices + data.TotReceipts as totaldocs
  from (
  select 
    B.waiter_id,
    WA.name, 
    WA.surname,
    (select count(id)             from documents where dtype = 0 and bill_id = B.id) AS Invoices, 
    (select isnull(sum(total), 0) from documents where dtype = 0 and bill_id = B.id) AS TotInvoices, 
    (select count(id)             from documents where dtype = 1 and bill_id = B.id) AS Receipts, 
    (select isnull(sum(total), 0) from documents where dtype = 1 and bill_id = B.id) AS TotReceipts
  from bills B
 inner join waiters WA on WA.id = B.waiter_id
 ) data