MS Access 2007更新/插入或合并或合并

Question

I have been searching for quite a while for how to do something that should be pretty easy to do in MS Access 2007.

I have a main table, items_moved, that tracks the number of files received, by type, by day. The basic structure is:

ID (autonumber), drive (text), type (text), date (DateTime), file_count (Number), file_size (Number)

Typical data looks like:

1777, F:\snaps, pics, 6/09/2010, 66, 151616131
1778, F:\snaps, pics, 6/10/2010, 5, 464864
1779, G:\pics, pics, 6/09/2010, 58, 45645646
1780, G:\pics, pics, 6/10/2010, 70, 123456667

I have a temp table with the exact same data structure. The temp table is generated by starting with (and including) the last day of the items_moved table and finding what's new.

Typical data looks like:

1, F:\snaps, pics, 6/10/2010, 366, 6531616131
2, F:\snaps, pics, 6/11/2010, 5, 464864
3, G:\pics, pics, 6/10/2010, 70, 123456667
4, G:\pics, pics, 6/11/2010, 56, 123645964

All I'm trying to do is append the temp table on to the main table so that:

1. There are no duplicates (based on drive and date)
2. If the temp table has matching drive and date and larger values it overwrites the main table
3. If the row exists in the temp table but not the main table, it gets appended to the end of the main .

The result would be in the items_moved table which would records 1-1776 unchanged, but now end like this:

1777, F:\snaps, pics, 6/09/2010, 66, 151616131         (Unchanged)
1778, F:\snaps, pics, 6/10/2010, 366, 6531616131       (Updated temp was larger)
1779, G:\pics, pics, 6/09/2010, 58, 45645646           (unchanged)
1780, G:\pics, pics, 6/10/2010, 70, 123456667          (unchanged)
1781, F:\snaps, pics, 6/11/2010, 5, 464864             (added)
1782, G:\pics, pics, 6/11/2010, 56, 123645964          (added)

I've tried every variation on join that I can think of.

I can get the rows with matching dates, so I can use that for an update statement. I can't get the rows from the temp that don't have matching dates in the items_moved table, so I can't run the insert.

The final point is to execute this as part of a VB action when the user requests a report from the access db.

Here's a snippet

INSERT INTO items_moved (drive, type, file_date, file_count, file_size) 
 SELECT 'F:\snaps','pics', temp_table.created, Count(temp_table.created),
  Sum(temp_table.size) FROM temp_table where temp_table.drive = 'F:\snaps' 
  GROUP BY temp_table.created

This works, but just adds to the end of the table. I'm left with the problem of duplicates. I'm sure there's a way to de-duplicate the table on the fly, but that seems like a waste. I've found some hints on using union , but nothing on how to use union to update one of the tables in the union.

Any input would be appreciated. Thanks.

Answer 1

It seems that the match is on drive, type and date, so for new records, perhaps:

INSERT INTO items_moved (drive, type, file_date, file_count, file_size) 
SELECT t.drive, type, t.file_date, t.file_count, t.file_size 
FROM temp_table t
LEFT JOIN items_moved m
ON t.Drive = m.Drive AND t.Type = m.Type and t.file_date = m.file_date
WHERE m.ID Is Null

A second query will be needed for updates:

UPDATE items_moved  m
INNER JOIN temp_table t
ON t.Drive = m.Drive AND t.Type = m.Type and t.file_date = m.file_date
SET m.file_count = t.file_count, m.file_size = t.file_size
WHERE m.file_size < t.file_size

Or there abouts. Is this what you mean?