CODEWITHSUNDEEP
sqlleft-join
Sandra
Sandra

Reputation: 721

Select first record in a One-to-Many relation using left join

I'm trying to join two tables using a left-join. And the result set has to include only the first record from the "right" joined table.

Lets say I have two tables A and B as below;

Table "A"

code | emp_no

101  | 12222
102  | 23333
103  | 34444
104  | 45555
105  | 56666

Table "B"

code | city       | county
101  | Glen Oaks  | Queens
101  | Astoria    | Queens
101  | Flushing   | Queens
102  | Ridgewood  | Brooklyn
103  | Bayside    | New York

Expected Output:

code | emp_no | city      | county
101  | 12222  | Glen Oaks | Queens
102  | 23333  | Ridgewood | Brooklyn
103  | 34444  | Bayside   | New York
104  | 45555  | NULL      | NULL
105  | 56666  | NULL      | NULL

If you notice my result has only the one matched record from table "B"(doesn't matter what record is matched) after left join (and it is a one to many mapping)

I need to pick the first matched record from table B and ignore all other rows.

Please help!

Thanks

Upvotes: 64

Views: 100704

Answers (7)

QuintinDB
QuintinDB

Reputation: 153

Another option: OUTER APPLY

If supported by the database, OUTER APPLY is an efficient and terse option.

SELECT *
FROM 
    Table_A a
OUTER APPLY
    (SELECT TOP 1 * 
    FROM Table_B b_1
    WHERE b_1.code = a.code
    ) b
;

This results in a left join to the indeterminate first matched record. My tests show it to be quicker than any other posted solution (on MS SQL Server 2012).

Upvotes: 15

Gayan
Gayan

Reputation: 41

I modified the answer from ruakh and this seem to work perfectly with mysql.

SELECT 
   table_a.code,
   table_a.emp_no,
   table_b.city,
   table_b.county
FROM table_a a
LEFT JOIN table_b b
ON b.code = a.code 
AND b.id = (  SELECT id FROM table_b 
              WHERE table_b.code = table_a.code 
              LIMIT 1
           )
;

Upvotes: 4

CShark
CShark

Reputation: 2221

The highest voted answer does not seem correct to me, and seems overcomplicated. Just group by the code field on table B in your subquery and select the maximum Id per grouping.

SELECT 
    table_a.code,
    table_a.emp_no,
    table_b.city,
    table_b.county
FROM 
    table_a
    LEFT JOIN 
        table_b
        ON table_b.code = table_a.code
        AND table_b.field_that_is_unique IN
            (SELECT MAX(field_that_is_unique)
             FROM table_b
             GROUP BY table_b.code)

Upvotes: 10

Andriy M
Andriy M

Reputation: 77667

If you are on SQL Server 2005 or later version, you could use ranking to achieve what you want. In particular, ROW_NUMBER() seems to suit your needs nicely:

WITH B_ranked AS (
  SELECT
    *,
    rnk = ROW_NUMBER() OVER (PARTITION BY code ORDER BY city)
  FROM B
)
SELECT
  A.code,
  A.emp_no,
  B.city,
  B.county
FROM A
  LEFT JOIN B_ranked AS B ON A.code = B.code AND b.rnk = 1

OR

WITH B_unique_code AS (
  select * from(
     SELECT
      *,
      rnk = ROW_NUMBER() OVER (PARTITION BY code ORDER BY city)
      FROM B
     ) AS s
  where rnk = 1
)
SELECT
  A.code,
  A.emp_no,
  B.city,
  B.county
FROM A
  LEFT JOIN B_unique_code AS B ON A.code = B.code

Upvotes: 5

kevin cline
kevin cline

Reputation: 2736

In Oracle you can do:

WITH first_b AS (SELECT code, min(rowid) AS rid FROM b GROUP BY code)) 
SELECT a.code, a.emp_no, b.city, b.county
FROM a 
INNER JOIN first_b 
 ON first_b.code = a.code
INNER JOIN b
 ON b.rowid = first_b.rid

Upvotes: 2

Charles Bretana
Charles Bretana

Reputation: 146469

this is how:

 Select * From TableA a
     Left Join TableB b
         On b.Code = a.Code 
             And [Here put criteria predicate that 'defines' what the first record is]

Hey, if the city and county are unique, then use them

   Select * From TableA a
     Left Join TableB b
         On b.Code = a.Code 
             And b.City + b.county =
                  (Select Min(city + county)
                   From TableB 
                   Where Code = b.Code)

But the point is you have to put some expression in there to tell the query processor what it means to be first.

Upvotes: 1

ruakh
ruakh

Reputation: 183270

After playing around a bit, this turns out to be trickier than I'd expected! Assuming that table_b has some single column that is unique (say, a single-field primary key), it looks like you can do this:

SELECT table_a.code,
       table_a.emp_no,
       table_b.city,
       table_b.county
  FROM table_a
  LEFT
  JOIN table_b
    ON table_b.code = table_a.code
   AND table_b.field_that_is_unique =
        ( SELECT TOP 1
                 field_that_is_unique
            FROM table_b
           WHERE table_b.code = table_a.code
       )
;

Upvotes: 43

Related Questions