Bereiken selecteren in een recordbereik in orakel

Dit is een leuke manier, mooie naam "Tabibitosan-methode " gegeven door Aketi Jyuuzou.

SQL> WITH data AS
  2    (SELECT num - DENSE_RANK() OVER(PARTITION BY status ORDER BY num) grp,
  3      status,
  4      num
  5    FROM t
  6    )
  7  SELECT MIN(num)
  8    ||' - '
  9    || MAX(num) range,
 10    COUNT(*) cnt
 11  FROM data
 12  WHERE status='A'
 13  GROUP BY grp
 14  ORDER BY grp
 15  /

RANGE         CNT
------ ----------
1 - 3           3
6 - 6           1
9 - 10          2

SQL>

Opmerking Het is beter om DENSE_RANK . te gebruiken om duplicaten te voorkomen.

Tafel

SQL> SELECT * FROM t ORDER BY num;

       NUM S
---------- -
         1 A
         1 A
         2 A
         2 A
         3 A
         4 U
         5 U
         6 A
         7 U
         8 U
         9 A

       NUM S
---------- -
        10 A

12 rows selected.

Er zijn duplicaten voor num =1.

DENSE_RANK gebruiken :

SQL> WITH data AS
  2    (SELECT num - DENSE_RANK() OVER(PARTITION BY status ORDER BY num) grp,
  3      status,
  4      num
  5    FROM t
  6    )
  7  SELECT MIN(num)
  8    ||' - '
  9    || MAX(num) range,
 10    COUNT(*) cnt
 11  FROM data
 12  WHERE status='A'
 13  GROUP BY grp
 14  ORDER BY grp
 15  /

RANGE         CNT
------ ----------
1 - 3           5
6 - 6           1
9 - 10          2

SQL>

ROW_NUMBER . gebruiken :

SQL> WITH DATA AS
  2    (SELECT num - ROW_NUMBER() OVER(PARTITION BY status ORDER BY num) grp,
  3      status,
  4      num
  5    FROM t
  6    )
  7  SELECT MIN(num)
  8    ||' - '
  9    || MAX(num) range,
 10    COUNT(*) cnt
 11  FROM data
 12  WHERE status='A'
 13  GROUP BY grp
 14  ORDER BY grp
 15  /

RANGE         CNT
------ ----------
2 - 3           2
1 - 2           2
1 - 6           2
9 - 10          2

SQL>

Dus, in het geval van duplicaten, de ROW_NUMBER query zou onjuiste resultaten opleveren. Gebruik DENSE_RANK .