Oracle 11g：如何将oracle表与自身结合起来？

Question

I have a table which looks like the following:

date          code                name    score         set
09/09/12     967873         Team A         24            1
09/09/12     967873         Team B         22            1
09/09/12     967873         Team A         21            2 
09/09/12     967873         Team B         16            2
02/04/12     965454         Team X         21            1
02/04/12     965454         Team Y         19            1
02/04/12     965454         Team X         21            2
02/04/12     965454         Team Y         19            2

you guessed it right! it's a volleyball match! however, I would like my output to be in a single line. For example:

date           code               Teams                 Set-1    Set-2     Set-3
09/09/12      967873             Team A VS.Team B       24-22    21-16       -
and so on.... 

**Notice that the game could have a third set as well

I will need some kind of self join to refine the above format to the format that is easier for Users view...Let me know if you need more details.

Thanks,

Answer 1

The query could look like this:

with matches as (
   select "DATE", code, name,
      max(case when "SET" = 1 then score end) score_1,
      max(case when "SET" = 2 then score end) score_2,
      max(case when "SET" = 3 then score end) score_3,
      row_number() over(partition by "DATE", code order by name) team_no
    from games
    group by "DATE", code, name
)
select a."DATE", a.code, a.name || ' vs. ' || b.name teams,
  a.score_1 || '-' || b.score_1 set_1,
  a.score_2 || '-' || b.score_2 set_2,
  a.score_3 || '-' || b.score_3 set_3
from matches a
join matches b on a."DATE" = b."DATE" and a.code = b.code
where a.team_no = 1 and b.team_no = 2;

date and set are rather unfortunate column names.

The query works in 3 steps:

1. The records are aggregated to create a single row per team and match. In that process, the score is assigned to one of the three columns set_1, set_2, set_3.
2. Row numbers are assigned to each row, starting at 1 for each match. The result is that one team is assigned 1 and the other one is assigned 2 (column team_no).
3. The resulting table is joined to itself, the left side for teams with no. 1 and the right side for teams with no. 2 using the match (date and code) as the join condition. The result is one row per match with the names and scores of both teams.

Answer 2

First, group the data by "date", code, "set" to LISTAGG the teams and the scores. Then pivot the results on the scores column. Here's the SQL for it:

WITH grouped AS (
  SELECT
    "date", code, "set",
    LISTAGG(name,  ' VS. ') WITHIN GROUP (ORDER BY name) AS teams,
    LISTAGG(score, '-'    ) WITHIN GROUP (ORDER BY name) AS score
  FROM matches
  GROUP BY
    "date", code, "set"
)
,    pivoted AS (
  SELECT
    "date", code, teams,
    nvl("1", '-') AS set1,
    nvl("2", '-') AS set2,
    nvl("3", '-') AS set3
  FROM grouped
  PIVOT (
    MAX(score) FOR "set" IN (1, 2, 3)
  ) p
)
SELECT * FROM pivoted
;

Please have a look at this query at SQL Fiddle too.

Answer 3

I'd do sg like this, if the name of the table is let's say VOLLEYBALL:

SELECT temp.date, temp.code, 
    temp.team1 || ' vs. ' || temp.team2 AS teams, 
    (SELECT v.score FROM volleyball v WHERE v.code = temp.code AND v.name = team1 AND v.set = 1) || '-' || 
        (SELECT v.score FROM volleyball v WHERE v.code = temp.code AND v.name = team2 AND v.set = 1) AS set1, 
    (SELECT v.score FROM volleyball v WHERE v.code = temp.code AND v.name = team1 AND v.set = 2) || '-' || 
        (SELECT v.score FROM volleyball v WHERE v.code = temp.code AND v.name = team2 AND v.set = 2) AS set2, 
    nvl((SELECT v.score FROM volleyball v WHERE v.code = temp.code AND v.name = team1 AND v.set = 3) || '-' || 
        (SELECT v.score FROM volleyball v WHERE v.code = temp.code AND v.name = team2 AND v.set = 3)
        , '-') AS set3 -- optional, if no results, then it will be a '-' 
FROM
    (SELECT v.date, v.code, 
        min(v.name) AS team1,  max(v.name) AS team2 
    FROM volleyball v 
    GROUP BY v.date, v.code) temp; 

This will result a one row summary.

Answer 4

To return the required results without needing a join, try:

select "date", 
       code, 
       min_name || ' VS. ' || max_name teams,
       sum(case when "set" = 1 and name = min_name then score end) || '-' || 
          sum(case when "set" = 1 and name = max_name then score end) "Set-1",
       sum(case when "set" = 2 and name = min_name then score end) || '-' || 
          sum(case when "set" = 2 and name = max_name then score end) "Set-2",
       sum(case when "set" = 3 and name = min_name then score end) || '-' || 
          sum(case when "set" = 3 and name = max_name then score end) "Set-3"
from (select g.*,
             min(name) over (partition by "date", code) min_name,
             max(name) over (partition by "date", code) max_name
      from games)
group by "date", code