Question

I have two tables:

+---------+-----------------+----------+----------+----------+---------+
| TrackId |       URI       | ArtistID |  Title   | AlbumID  | BitRate |
+---------+-----------------+----------+----------+----------+---------+
|  1      | /home/music/... |   234    | atune    |  8958223 |   192   |
|  2      | /music/uri1/... |   427    | goodsong |  222     |   192   |
|  3      | /music/uri2/... |   427    | goodsong |  222     |   128   |
|  4      | /music/uri3/... |   427    | goodsong |  222     |   160   |
|  5      | /home/music/... |   427    | goodsong |  333     |   128   |
|  6      | /home/music/... |   522    | another  |  3458859 |   128   |
+---------+-----------------+----------+----------+----------+---------+

and

+----------+------------+
| AlbumID  | AlbumTitle |
+----------+------------+
|  8958223 |   titleA   |
|  222     |   titleB   |
|  333     |   titleC   |
|  3458859 |   titleD   |
+----------+------------+

Simply put, what I want is this:

+---------+-----------------+----------+----------+----------+------------+---------+
| TrackId |       URI       | ArtistID |  Title   | AlbumId  | AlbumTitle | BitRate |
+---------+-----------------+----------+----------+----------+------------+---------+
|  3      | /music/uri2/... |   427    | goodsong |  222     |   titleB   |   128   |
|  4      | /music/uri3/... |   427    | goodsong |  222     |   titleB   |   160   |
+---------+-----------------+----------+----------+----------+------------+---------+

This is an attempt to remove duplicates that have the:

• same title
• same artist id
• different track id
• the non-highest bitrate duplicates
• same album name from the album table

while not returning the entry with the highest bitrate duplicate

I’ve asked a very similar question here: Select duplicates from a single row?

The solution to that was:

SELECT c1.*
  FROM CoreTracks c1
      ,(SELECT Title, ArtistID, MAX(FileSize) AS maxFileSize, MAX(BitRate) maxBitRate
          FROM CoreTracks
          GROUP BY Title, ArtistID) c2
  WHERE c1.Title = c2.Title
    AND c1.ArtistID = c2.ArtistID
    AND (c1.FileSize != c2.maxFileSize AND c1.BitRate != c2.maxBitRate)

…but I cannot seem to wrap my head around dealing with another table this time around.

Answer 1

Give this a try:

SELECT c1.*
FROM CoreTracks c1
LEFT JOIN 
     (SELECT Title, ArtistID, AlbumID, MAX(BitRate) maxBitRate
      FROM CoreTracks
      GROUP BY Title, ArtistID, AlbumID) c2
    ON c1.Title = c2.Title
    AND c1.ArtistID = c2.ArtistID
    AND c1.AlbumID = c2.AlbumID
    AND c1.BitRate = c2.maxBitRate
WHERE c2.Title IS NULL

Here is the SQL Fiddle.

–EDIT

To add the Album Title:

SELECT c1.*, a.AlbumTitle
FROM CoreTracks c1 
JOIN Albums a on c1.AlbumId = a.AlbumId
LEFT JOIN 
     (SELECT c.Title, c.ArtistID, c.AlbumID, MAX(c.BitRate) maxBitRate
      FROM CoreTracks c JOIN Albums A ON c.AlbumId = a.AlbumId
      GROUP BY Title, ArtistID, AlbumID) c2
    ON c1.Title = c2.Title
    AND c1.ArtistID = c2.ArtistID
    AND c1.AlbumID = c2.AlbumID
    AND c1.BitRate = c2.maxBitRate
WHERE c2.Title IS NULL

And the updated fiddle.

Good luck.