Sql inner join only with last row in second table - sql

Question

I have two tables: leads and tracking_leads.
Table structure is as below,
---------------------------- ----------------------
| leads | | tracking_leads |
---------------------------- ----------------------
| id | | tracking_id |
| lead_id | | lead_id |
| anzahl_tickets | | field_name |
| bearbeitungs_id_einkauf | | date |
---------------------------- -----------------------
I need sql for join table lead with tracking_leads table but get only LAST match row in table tracking_leads .
Sql example:
SELECT DATE_FORMAT(tracking_leads.date, "%d.%m.%Y") as trackDate, SUM(l.anzahl_tickets)
as sumValue FROM leads as l INNER JOIN tracking_leads ON l.lead_id=tracking_leads.lead_id
WHERE bearbeitungs_id_einkauf <> '' AND tracking_leads.field_name='bearbeitungs_id_einkauf'
GROUP BY DATE_FORMAT(tracking_leads.date, "%d.%m.%Y")
In this part : INNER JOIN tracking_leads ON l.lead_id=tracking_leads.lead_id need only last record from tracking_leads table.
For example, leads data:
id lead_id anzahl_tickets bearbeitungs_id_einkauf
1 20 2 100
tracking_leads data:
tracking_id lead_id field_name date
1 20 bearbeitungs_id_einkauf 2019-05-31 13:55
2 20 bearbeitungs_id_einkauf 2019-05-31 15:00
In result i need get :
2019-05-31 2
But now i get
2019-05-31 4
Because there are duplicated of lead_id (need only last record).
How can i solve this problem?
Thanks!

Answer 1

My preference would be to use an inline view to get the max dates.
A correlated subquery would be executed once for each row, while the inline view would only need to be executed once.
This should work:
SELECT DATE_FORMAT(tl.date, "%d.%m.%Y") as trackDate,
SUM(l.anzahl_tickets) as sumValue
FROM leads as l
INNER JOIN (
select x.lead_id, max(x.date) date from tracking_leads x where x.field_name = 'bearbeitungs_id_einkauf' group by x.lead_id
) tl ON l.lead_id=tl.lead_id
WHERE bearbeitungs_id_einkauf <> ''
GROUP BY DATE_FORMAT(tl.date, "%d.%m.%Y")
Side node: the test for empty value of bearbeitungs_id_einkauf in the WHERE clause is database-specific, so watch out for issues there. In Oracle, for example, there is no such thing as an empty string, so you would have to test it for NOT NULL. I'm assuming this is not Oracle.

Answer 2

First, I don't like the date format DD-MM-YYYY, because you cannot sort by it. Just use YYYY-MM-DD.
Second, you can use a correlated subquery to get the most recent date:
SELECT DATE(tl.date) as trackDate, SUM(l.anzahl_tickets) as sumValue
FROM leads l INNER JOIN
tracking_leads tl
ON l.lead_id = tl.lead_id
WHERE l.bearbeitungs_id_einkauf <> '' AND
tl.field_name = 'bearbeitungs_id_einkauf' AND
tl.date = (SELECT MAX(tl2.date)
FROM tracking_leads tl2
WHERE tl2.lead_id = tl.lead_id AND
tl2.field_name = tl.field_name
)
GROUP BY DATE(tl.date);
Of course, you can leave your original date format if you prefer. If you do, you can use:
ORDER BY MIN(tl.date)
so the results are order by the date.