En caso de que sus intervalos de fechas no estén anidados o intersectados, puede usar el truco con JOIN para producir un conjunto de resultados. Entonces, para seleccionar el conjunto de registros deseado, necesitará:
SELECT
*
FROM
(SELECT
ends.point AS date_start,
starts.point AS date_end
FROM
(SELECT
d.date_end+INTERVAL 1 DAY AS point,
@n:[email protected]+1 AS num
FROM
d
CROSS JOIN (SELECT @n:=1) AS init0
) AS ends
INNER JOIN
(SELECT
d.date_start-INTERVAL 1 DAY AS point,
@m:[email protected]+1 AS num
FROM
d
CROSS JOIN (SELECT @m:=0) AS init1
) AS starts
ON ends.num=starts.num
UNION ALL
SELECT
'2014-01-01',
MIN(date_start) - INTERVAL 1 DAY AS date_end
FROM
d
HAVING
date_end>'2014-01-01'
UNION ALL
SELECT
MAX(date_end)+INTERVAL 1 DAY AS date_start,
'2014-12-31'
FROM
d
HAVING
date_start<'2014-12-31'
) as dates
WHERE
date_start<=date_end
ORDER BY
date_start;
eso resultará en
+------------+------------+ | date_start | date_end | +------------+------------+ | 2014-01-01 | 2014-03-04 | | 2014-04-11 | 2014-04-30 | | 2014-06-06 | 2014-07-09 | | 2014-08-16 | 2014-12-31 | +------------+------------+
(violín porque esto está aquí)
Para "completar" su tabla, puede usar INSERT..SELECT
sintaxis con SELECT consulta anterior.
