Query to return unique values when one of column value is repeating

问题

I have a sql query like the following

SELECT Name.Nameguid,Name.Name,relation.ancestorguid,relation.productguid,relation.pathLength
     FROM Name Name 
     JOIN ProductRelationship relation 
       ON Name.productGuid = relation.ancestorGuid 
      AND relation.productGuid = '6075D04A-E74A-464B-94E7-25374F0B9833' 
     ORDER BY relation.pathLength DESC

Which returns the following

NameGuid                                Name                AncestorGuid                            ProductGuid                             PathLength
6D39CB04-88D9-4125-A052-8DF608AAD29C    NameParentOnly  8E07F824-763C-434F-926C-80FCC8690243    6075D04A-E74A-464B-94E7-25374F0B9833    2
186E1DF3-4D1A-4020-B845-1280CF1092EA    NameParentChild 8E07F824-763C-434F-926C-80FCC8690243    6075D04A-E74A-464B-94E7-25374F0B9833    2
CA60E542-4047-4B4D-AA22-0744A1E5F2E0    childlevelName  D9833FCA-93A7-42F5-AFC3-5544F7A4425D    6075D04A-E74A-464B-94E7-25374F0B9833    1
A09D01FC-D69D-4AFA-B4D0-C804F030A281    NameParentChild D9833FCA-93A7-42F5-AFC3-5544F7A4425D    6075D04A-E74A-464B-94E7-25374F0B9833    1
63148C89-59FD-4C96-883F-60E8446A6BC4    NameParentChild 6075D04A-E74A-464B-94E7-25374F0B9833    6075D04A-E74A-464B-94E7-25374F0B9833    0

My scenario is such that when ever name gets repeated , I want the query to return only the row with least value for Path Length .

In the above example only NameParentChild gets repeated three times i want it to return the one with least path length(In this case Zero)

This following is the result i am expecting

NameGuid                                Name                AncestorGuid                            ProductGuid                             PathLength
6D39CB04-88D9-4125-A052-8DF608AAD29C    NameParentOnly  8E07F824-763C-434F-926C-80FCC8690243    6075D04A-E74A-464B-94E7-25374F0B9833    2
CA60E542-4047-4B4D-AA22-0744A1E5F2E0    childlevelName  D9833FCA-93A7-42F5-AFC3-5544F7A4425D    6075D04A-E74A-464B-94E7-25374F0B9833    1
63148C89-59FD-4C96-883F-60E8446A6BC4    NameParentChild 6075D04A-E74A-464B-94E7-25374F0B9833    6075D04A-E74A-464B-94E7-25374F0B9833    0

How can i tweak the query to do this ? I want the query to work on Oracle , DB2 and SQL server as well.

Thanks

回答1:

You should be able to do this with analytic (or window) functions. Whether analytic functions are available in your RDBMS or not, and which syntax you have to use is another matter. In Oracle, assumming that NameGuid is unique for your original query, you could do:

select NameGuid, Name, AncestorGuid, ProductGuid, PathLength
from (
    select 
    NameGuid, 
    Name, 
    AncestorGuid, 
    ProductGuid, 
    PathLength, 
    -- take every row from original query with the same Name as this,
    -- order those rows by PathLength (and NameGuid to disambiguate)
    -- and return the NameGuid of the first row in that "partition"
    first_value(NameGuid) over (partition by Name order by PathLength asc, NameGuid asc) MinNameGuid
    from ( 
        ... your original query ...
    )
)
where 
-- return rows whose NameGuid is the same as the NameGuid calculated by first_value(...)
NameGuid = MinNameGuid

来源：https://stackoverflow.com/questions/20373082/query-to-return-unique-values-when-one-of-column-value-is-repeating

标签

sql

sql-server

Oracle

oracle11g

db2