Rename duplicate rows in MySQL, WHERE IN () - dosen't works fine - mysql

Question

I started from here:
Rename duplicate rows in MySQL
SELECT mid(s,instr(s,',')+1) as dubid
FROM (
SELECT group_concat(id ORDER BY id ASC) s, count(*) c
FROM `...`
GROUP BY field
HAVING c > 1) as z
Seems to work fine and returns dubid rows like
1
3,5,6
8
10,15
So I made this:
set #suf:=0;
UPDATE `table_cf_vendlist`
SET `cf_vendlist` = concat(`cf_vendlist`, ' \(', #suf:=#suf+1,'\)')
WHERE `cf_vendlistid` IN ({SELECT mid...})
But it renames only first ID from each row. So how I can make it work? Also how to nullyfy #suf counter for each row?
And no PHP. MySQL only!
set #suf:=0;
set #last := 0;
UPDATE `vtiger_cf_vendlist` t1
INNER JOIN (
SELECT MIN(cf_vendlistid) as id, cf_vendlist
FROM vtiger_cf_vendlist
GROUP BY cf_vendlist) AS unique1
ON (t1.cf_vendlist = unique1.cf_vendlist)
SET t1.`cf_vendlist` = concat(t1.`cf_vendlist`, ' \(',
#suf := if (#last = t1.cf_vendlistid, #suf+1, 1 and #last := t1.cf_vendlist), '\)')
WHERE t1.cf_vendlistid <> unique1.id
But I am not shure where is t1.cf_vendlist, and where is unique1.cf_vendlist. Also result is 0 and 1 in names, no increments and zeroing for each id set. But still, loks like it renamed everyting is needed.
Here is some data to run with sqlfiddle.com:
CREATE TABLE IF NOT EXISTS `table_cf_vendlist` (
`cf_vendlistid` int(11) NOT NULL AUTO_INCREMENT,
`cf_vendlist` varchar(200) NOT NULL,
PRIMARY KEY (`cf_vendlistid`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 AUTO_INCREMENT=78 ;
INSERT INTO `table_cf_vendlist` (`cf_vendlistid`, `cf_vendlist`) VALUES
(2, 'Поставщик 2'),
(3, 'Поставщик 3'),
(4, 'Поставщик 4'),
(5, 'del 1'),
(6, 'Поставщик 1 (1)'),
(7, 'Поставщик 1 (2)'),
(8, 'пост 66'),
(9, 'пост 77'),
(10, 'пост 67'),
(11, 'пост 5'),
(12, '123'),
(13, '102'),
(14, 'пост 5 (1)'),
(15, 'пост 70'),
(47, 'Поставщик 1'),
(48, 'Поставщик 2'),
(49, 'Поставщик 3'),
(50, 'Поставщик 4'),
(51, 'del 1'),
(52, 'Поставщик 1'),
(53, 'Поставщик 1'),
(54, 'пост 5'),
(55, '124523452346256456'),
(56, 'пост 66'),
(57, 'пост 77'),
(58, 'пост 5'),
(59, 'пост 5'),
(60, 'пост 5'),
(61, 'пост 67'),
(62, '123'),
(63, '123'),
(64, '123'),
(65, 'пост 5'),
(66, 'пост 68'),
(67, '123'),
(68, '123'),
(69, '123'),
(70, '123'),
(71, '102'),
(72, 'пост 5'),
(73, 'пост 70');
UPD
We got
52 Поставщик 1 (11)
53 Поставщик 1 (12)
Instead of
52 Поставщик 1 (3)
53 Поставщик 1 (4)
and
48 Поставщик 2 (13)
49 Поставщик 3 (14)
50 Поставщик 4 (15)
Instead of
48 Поставщик 2 (2)
49 Поставщик 3 (2)
50 Поставщик 4 (2)
UPD 2:
set #suf:=0;
set #last := 0;
UPDATE table_cf_vendlist t
INNER JOIN (SELECT cf_vendlistid, concat(cf_vendlist, ' (',
#suf := if (STRCMP(#last1, cf_vendlist) = 0,
#suf+1,
1 or #last := cf_vendlist), ')') as new_label
FROM (
SELECT t1.cf_vendlistid, t1.cf_vendlist, unique1.id as reference_id
FROM table_cf_vendlist t1
INNER JOIN (SELECT MIN(cf_vendlistid) as id, cf_vendlist
FROM table_cf_vendlist
GROUP BY cf_vendlist) AS unique1
ON t1.cf_vendlist = unique1.cf_vendlist
ORDER BY cf_vendlist, cf_vendlistid
) AS t1_sorted
WHERE t1_sorted.cf_vendlistid <> t1_sorted.reference_id)
AS t1_sorted_labeled
ON t.cf_vendlistid = t1_sorted_labeled.cf_vendlistid
SET t.cf_vendlist = t1_sorted_labeled.new_label;
works. But 1 or #last := cf_vendlist is a hack that allows to return 1 or 0(and) instead of cf_vendlist string (see my comment below)

Answer 1

I have a solution in sqlfiddle. To follow how it works,
1) look at the subquery t1_sorted
SELECT t1.cf_vendlistid, t1.cf_vendlist, unique1.id as reference_id
FROM table_cf_vendlist t1
INNER JOIN (SELECT MIN(cf_vendlistid) as id, cf_vendlist
FROM table_cf_vendlist
GROUP BY cf_vendlist) AS unique1
ON t1.cf_vendlist = unique1.cf_vendlist
WHERE t1.cf_vendlistid <> unique1.id
ORDER BY cf_vendlist, cf_vendlistid
which returns results like in fig. 1.
2a) In subquery t1_sorted_labeled, the results are projected to remove the first occurrence of the cf_vendlist in each group and
2b) an extra field new_label is created with the cf_vendlist and (copy number)
3) Finally, the original query is joined with this subquery in the UPDATE statement, which merely copies new_label to cf_vendlist
Thus, complete answer:
set #suf:=0;
set #last := 0;
UPDATE table_cf_vendlist t
INNER JOIN (SELECT cf_vendlistid, concat(cf_vendlist, ' (',
#suf := if (STRCMP(#last, cf_vendlist) = 0,
#suf+1,
1 and length(#last := cf_vendlist)), ')') as new_label
FROM (
SELECT t1.cf_vendlistid, t1.cf_vendlist, unique1.id as reference_id
FROM table_cf_vendlist t1
INNER JOIN (SELECT MIN(cf_vendlistid) as id, cf_vendlist
FROM table_cf_vendlist
GROUP BY cf_vendlist) AS unique1
ON t1.cf_vendlist = unique1.cf_vendlist
ORDER BY cf_vendlist, cf_vendlistid
) AS t1_sorted
WHERE t1_sorted.cf_vendlistid <> t1_sorted.reference_id)
AS t1_sorted_labeled
ON t.cf_vendlistid = t1_sorted_labeled.cf_vendlistid
SET t.cf_vendlist = t1_sorted_labeled.new_label;
BTW, you should really have an index on cf_vendlist.