Question

I have a table with the format:

Id     | Loc |
-------|-----|
789-A  | 4   |
123    | 1   |
123-BZ | 1   |
123-CG | 2   |
456    | 2   |
456    | 3   |
789    | 4   |

I want to exclude certain rows from the result of query based on whether a duplicate exists. In this case, though, the definition of a duplicate row is pretty complex:

If any row returned by the query (let’s refer to this hypothetical row as ThisRow) has a counterpart row also contained within the query results where Loc is identical to ThisRow.Loc AND Id is of the form <ThisRow.Id>-<an alphanumeric suffix> then ThisRow should be considered a duplicate and excluded from the query results.

For example, using the table above, SELECT * FROM table should return the results set below:

Id     | Loc |
-------|-----|
789-A  | 4   |
123-BZ | 1   |
123-CG | 2   |
456    | 2   |
456    | 3   |

I understand how to write the string matching conditional:

ThisRow.Id REGEXP '^PossibleDuplicateRow.Id-[A-Za-z0-9]*'

and the straight comparison of Loc:

ThisRow.Loc = PossibleDuplicateRow.Loc

What I don’t understand is how to form these conditionals into a (self-referential?) query.

Answer 1

Here’s one way:

SELECT *
FROM myTable t1
WHERE NOT EXISTS
(
    SELECT 1
    FROM myTable t2
    WHERE t2.Loc = t1.Loc
    AND t2.Id LIKE CONCAT(t1.Id, '-%')
)

SQL Fiddle example

Or, the same query using an anti-join (which should be a little faster):

SELECT *
FROM myTable t1
LEFT OUTER JOIN myTable t2 
    ON t2.Loc = t1.Loc
    AND t2.Id LIKE CONCAT(t1.Id, '-%')
WHERE t2.Id IS NULL

SQL Fiddle example