How can I clean up this query?
I am generating queries which carry out tag searches on the following tables:
entry : id, name, desc tag : id, name entry_tag : entry, tag, val
The query is generated from a user-entered predicate such as:
(year >= 1990 and year < 2000) and (status = 'complete' or status = 'pending')
It resembles the following:
select * from entry where id in
(select entry
from
(select t0.entry, t0.val as c0, t1.val as c1, …[1]
from
(select entry, val from entry_tag where tag in
(select id from tag where name = 'year')) as t0,
(select entry, val from entry_tag where tag in
(select id from tag where name = 'status')) as t1,
…[2]
where t0.entry = t1.entry and …[3]) as t
where
…[4]);
(The deep nesting is my naïve attempt to minimise the number of roundtrips, assuming of course that 开发者_JAVA技巧that’s a Good Thing™ to do.)
Find tags by
name
and alias eachvalue
toc0
…cn
in order of input appearance.Alias each
tag_entry
query to the correspondingt0
…tn
.Group tag mappings by
entry
.Insert the transformed Boolean expression directly into the
where
clause, e.g.:(cast(c0 as signed) >= 1990 and cast(c0 as signed) < 2000) and (c1 = 'complete' or c1 = 'pending')
Is there a better way to go about this, or am I on the right track?
Having to produce this seems wrong:
t0.entry = t1.entry and … and t0.entry = tn.entry
I’m just not sure how else to go about this in a way that preserves the structure of the predicate entered by the user. I don’t want to generate a mangled series of joins when I can do a straightforward mechanical transformation.
精彩评论