子查询的目的

Question

I am working on one maintenance project where I need to optimize the query. It is very coplex query so I have divided into 5 parts.

There in one table "document" where fields are

document_id(PK)
doc_name
fk_product_id
issue_date
date_expiring
status

Query is

SELECT SQL_CALC_FOUND_ROWS this.* , COUNT(this.document_id) as count_document_id,
FROM ( SELECT * from documents where status != 'D' order by date_expiring desc ) this 

There are millions of records but above query return only record. I don't understand what this query means/ what records will it return.

Answer 1

Let's break it down.

Starting with the subquery :

SELECT * FROM documents WHERE status != 'D' ORDER BY date_expiring desc

The resultset is a subset of your documents table.

Next, the main query's FROM :

FROM ( /* that subquery up there ^^^ */ ) this

This treats the resultset of the subquery as a virtual table and gives it this for a name. MySQL's query optimizer is usually smart enough to avoid extra overhead with this sort of subquery pattern.

Next, we'll look at the SELECT :

SELECT SQL_CALC_FOUND_ROWS 
       this.* , 
       COUNT(this.document_id) as count_document_id
 FROM ...

SQL_CALC_FOUND_ROWS is deprecated and pointless here, but it does no harm. It's deprecated because there are better ways to count rows. It's pointless because this query necessarily must return only one row; read on.

COUNT(this.document_id) as count_document_id counts all the rows coming from the subquery that contain a non-NULL document_id. But document_id is the primary key of the underlying table so it will never be null. Therefore COUNT(*) would work just as well. And, by the way, it makes the subquery's ORDER BY clause entirely pointless.

Because there's an aggregate function on the SELECT line and no GROUP BY clause, the query aggregates all the rows in the subquery and yields just one row.

this.* is just plain wrong in standard SQL. And it's meaningless in MySQL: the single row you get back from this query contains the count: that's fine. But by specifying this.* you also get back the contents of just one randomly chosen row from the subquery. This will break when you move to MySQL 8.0. MySQL has a notorious nonstandard extension to GROUP BY that enables this sort of random thing.

I hope this helps. Taking over somebody's code can be loads of fun, eh?