Group output of SHOW COLUMNS into comma-delimited list

问题

I am comparing database tables on a development server against a live server, looking for column name changes, new columns, and columns that have been dropped. I'd like to do something like this:

SELECT GROUP_CONCAT(Field) FROM (SHOW COLUMNS ON table_name) GROUP BY Field

What I am after is a comma-delimited list that I can then take to the live server and do:

SHOW COLUMNS FROM table_name WHERE NOT IN ([comma-delimited list from above query])

Any thoughts on how best to do this - either by correcting me in my own approach, or by another means all together? Obviously, the above SQL does not work.

A note: The servers are entirely separate and may not communicate with each other, so no direct comparison is possible.

EDIT

Thanks for the answers, guys! Applying your answers to the question, this is my final SQL to get the column names:

SELECT CONCAT("'", GROUP_CONCAT(column_name ORDER BY ordinal_position SEPARATOR "', '"), "'") AS columns
FROM information_schema.columns
WHERE table_schema = 'db_name' AND table_name = 'tbl_name'

That gives me a list that looks like this:

'id', 'name', 'field1', 'field2'

Then I can use this query to compare:

SELECT GROUP_CONCAT(column_name ORDER BY ordinal_position)
FROM information_schema.columns
WHERE table_schema = 'db_name' AND table_name = 'tbl_name' AND column_name NOT IN ('id', 'name', 'field1', 'field2')

The results are a list of any columns that exist in the first database and not in the second. Perfect!

回答1:

Take a look at the information_schema.columns table

select group_concat(column_name order by ordinal_position)
from information_schema.columns
where table_schema = 'database_name' and table_name = 'table_name'

edit. Information schema allows you to make queries on metadata. So, you can even compare fields between tables with a left join for example.

edit. Hi Chris. Glad you've solved. As you said your problem was quite different because it concerned with different servers. I add an example of two different databases on the same server.

create database db1;
use db1;
create table table1(
id int not null auto_increment primary key,
name varchar(50),
surname varchar(50),
dob date)
engine = myisam;

create database db2;
create table db2.table2 like db1.table1;
alter table db2.table2 drop column dob;

select i1.column_name from (
select column_name
from information_schema.columns 
where table_schema = 'db1' and table_name = 'table1' ) as i1
left join (
select column_name
from information_schema.columns 
where table_schema = 'db2' and table_name = 'table2' ) as i2
on i1.column_name = i2.column_name
where i2.column_name is null

and the obvious result is dob that is present in table1 and not in table2.

Hope that it helps someone else. Regards guys. :)

回答2:

You should use INFORMATION_SCHEMA.

You can copy the information_schema.columns table from each database into a shared schema and then run SQL queries to compare them.

来源：https://stackoverflow.com/questions/5186525/group-output-of-show-columns-into-comma-delimited-list

标签

mysql

database-schema