在我的 SQLite 语句中,如果count(t2.ref_id)
为零,则会出现除零的问题。
我可以调整 SQLite 语句,这样如果count(t2.ref_id)
为零,那么稀释性(稀释因子)将高于最高的非零稀释度吗?
select t1.id, cast(:totalItems as float) / count(t2.ref_id) as scarsity
from t1 left join t2 on t1.id = t2.ref_id
group by t1.id
order by scarsity
答案 0 :(得分:2)
您需要检查值0并执行不同的操作。一种常见的方法是将NULL
除以NULLIF()
个NULL
来将值转换为select t1.id, cast(:totalItems as float) / NULLIF(count(t2.ref_id), 0) as scarsity
from t1 left join t2 on t1.id = t2.ref_id
group by t1.id
order by scarsity
。
NULLIF()
但我不确定SQLite是否有CASE
,所以你可以使用更长的select t1.id, cast(:totalItems as float) / CASE WHEN count(t2.ref_id) = 0 THEN NULL ELSE count(t2.ref_id) END as scarsity
from t1 left join t2 on t1.id = t2.ref_id
group by t1.id
order by scarsity
版本来代替......
COUNT()
即使您输入两次,select t1.id, CASE WHEN count(t2.ref_id) = 0 THEN 9999999 ELSE cast(:totalItems as float) / count(t2.ref_id) END as scarsity
from t1 left join t2 on t1.id = t2.ref_id
group by t1.id
order by scarsity
也不会被计算两次。该值将被重用:)
或者,如果你得到0行,做一些完全不同的事情......
{{1}}
然后你可以检查NULL是最高值,还是只检查9999999等等。