select columns if happened on same date

问题

I am trying to craft a query that will select multiple columns (uniqueID, IPaddress, date, country) but I would like the query only to select those records where the same IPaddress and UniqueID showed up on a given day. I have been trying ...

SELECT uniqueID AS uniqueID, 
IPAddress AS IPAddr,
Date AS DDate, 
Location AS Countries,
COUNT(IPAddress) AS IPCount
FROM History WHERE (uniqueID= 20677) 
GROUP BY Date
ORDER BY DDate DESC, IPAddr DESC

But it does eliminate records where only one Ip/UniqueID showed up on one day. I really only want to find records where the uniqueID showed up on the same day with different IPaddresses.

回答1:

Not sure if I follow exactly, but sounds like you want to join a subquery:

SELECT a.*
FROM History a
JOIN (  SELECT uniqueID, Date, COUNT(DISTINCT IPAddress)
        FROM History 
        GROUP BY uniqueID,Date
        HAVING COUNT(DISTINCT IPAddress) > 1
     )b
ON a.uniqueID = b.uniqueID
AND a.date = b.date

EDIT: misread your with different IP addresses part, updated.

The subquery will show you all the records that have multiple ip addresses for the same uniqueID/date, then joining that to your history table shows you all the fields for the records you're after.

回答2:

You need to group by three fields, if you are looking for duplicates amongst them:

SELECT uniqueID, IPAddress, "Date" AS DDate, Location AS Countries,
       COUNT(IPAddress) AS IPCount
FROM History
WHERE uniqueID = 20677
GROUP BY uniqueId, IPAddress, "Date"
having IPCount > 1;

You dn't really need the uniqueid in the group by clause because you are selecting only one. However, I think it makes the intention of the query clearer.

As written, this will either generate an error (because location is not in the aggregation clause) or return a random value for location, if you are using MySQL). If the latter, try:

SELECT uniqueID, IPAddress, "Date" AS DDate,
       group_concat(Location) AS Countries,
       COUNT(IPAddress) AS IPCount
FROM History
WHERE uniqueID = 20677
GROUP BY uniqueId, IPAddress, "Date"
having IPCount > 1;

Also, naming columns after data types and functions is not good idea. In many databases, date is both a data type and a function. Some other name, like DateAdded avoids this conflict.

来源：https://stackoverflow.com/questions/16903966/select-columns-if-happened-on-same-date

标签

sql

date

unique