MS Access SQL 查询表并删除重复项

Question

I have reviewed several posts here and can't find the answer, but it may be that the problem and answer are a bit above my pay grade.

I have an MSAccess Table with two ID fields (ID1 and ID2), and many other columns.

I want to query (SELECT statement) the table and need two things to happen. Not sure if the order matters. I want the query to return all columns from the table.

• No duplicates for ID1, and I simply don't care which record(s) gets thrown away.
• No duplicates for ID2, and I want to keep the record where the Date field in the table is earlier than the other values for records with the same ID2 value.

Just can't figure out how to do this with an SQL Select statement.

Example Data - assume DateFld is ASC from top to bottom

ID1    ID2    DateFld    ...
1      24
1      24
2      23
3      98
4      23
4      23
5      98
6      72

Keep rows 1, 3, 4, 8

No indices, as this table was created from a make table query; but I can add them. Let me know if you need more.

Answer 1

For this sample data you can use NOT EXISTS :

SELECT t.*
FROM tablename AS t
WHERE NOT EXISTS (
  SELECT 1 FROM tablename
  WHERE ID2 = t.ID2 AND DateFld < t.DateFld
)

Results:

ID1 ID2 DateFld
1   24  ...
2   23  ...
3   98  ...
6   72  ...

Answer 2

SELECT
  ID1, 
  ID2, 
  [allOtherColumns]
FROM(
  SELECT 
      ID1, 
      ID2, 
      ROW_NUMBER() OVER (PARTITION BY ID1 ORDER BY ID1) AS ID1Selector, 
      ROW_NUMBER() OVER (PARTITION BY ID2 ORDER BY DateValue ASC) AS ID2Selector,
      [allOtherColumns)
  FROM tbl) AS InnerQry
WHERE ID1Selector=1 AND ID2Selector=1

Note, this is SQL Server code that might work in Access. I don't have a lot of experience with Access, but I know SQL. In SQL, the ROW_NUMBER() window function will split ID1 by unique ID1s so the first ID1 it comes across will be the only one it selects, and the ID2Selector will give the Row Number 1 to the earliest date value it comes across within each individual ID2.

There might be a more efficient way to do it, but that is how I always do it in SQL.