How do I delete all the duplicate records in a MySQL table without temp tables

问题

I've seen a number of variations on this but nothing quite matches what I'm trying to accomplish.

I have a table, TableA, which contain the answers given by users to configurable questionnaires. The columns are member_id, quiz_num, question_num, answer_num.

Somehow a few members got their answers submitted twice. So I need to remove the duplicated records, but make sure that one row is left behind.

There is no primary column so there could be two or three rows all with the exact same data.

Is there a query to remove all the duplicates?

回答1:

Add Unique Index on your table:

ALTER IGNORE TABLE `TableA`   
ADD UNIQUE INDEX (`member_id`, `quiz_num`, `question_num`, `answer_num`);

Another way to do this would be:

Add primary key in your table then you can easily remove duplicates from your table using the following query:

DELETE FROM member  
WHERE id IN (SELECT * 
             FROM (SELECT id FROM member 
                   GROUP BY member_id, quiz_num, question_num, answer_num HAVING (COUNT(*) > 1)
                  ) AS A
            );

回答2:

This doesn't use TEMP Tables, but real tables instead. If the problem is just about temp tables and not about table creation or dropping tables, this will work:

SELECT DISTINCT * INTO TableA_Verify FROM TableA;

DROP TABLE TableA;

RENAME TABLE TableA_Verify TO TableA;

回答3:

Instead of drop table TableA, you could delete all registers (delete from TableA;) and then populate original table with registers coming from TableA_Verify (insert into TAbleA select * from TAbleA_Verify). In this way you won't lost all references to original table (indexes,... )

CREATE TABLE TableA_Verify AS SELECT DISTINCT * FROM TableA;

DELETE FROM TableA;

INSERT INTO TableA SELECT * FROM TAbleA_Verify;

DROP TABLE TableA_Verify;

回答4:

Thanks to jveirasv for the answer above.

If you need to remove duplicates of a specific sets of column, you can use this (if you have a timestamp in the table that vary for example)

CREATE TABLE TableA_Verify AS SELECT * FROM TableA WHERE 1 GROUP BY [COLUMN TO remove duplicates BY];

DELETE FROM TableA;

INSERT INTO TableA SELECT * FROM TAbleA_Verify;

DROP TABLE TableA_Verify;

回答5:

Add Unique Index on your table:

ALTER IGNORE TABLE TableA   
ADD UNIQUE INDEX (member_id, quiz_num, question_num, answer_num);

is work very well

回答6:

If you are not using any primary key, then execute following queries at one single stroke. By replacing values:

# table_name - Your Table Name
# column_name_of_duplicates - Name of column where duplicate entries are found

create table table_name_temp like table_name;
insert into table_name_temp select distinct(column_name_of_duplicates),value,type from table_name group by column_name_of_duplicates;
delete from table_name;
insert into table_name select * from table_name_temp;
drop table table_name_temp

create temporary table and store distinct(non duplicate) values
make empty original table
insert values to original table from temp table
delete temp table

It is always advisable to take backup of database before you play with it.

回答7:

As noted in the comments, the query in Saharsh Shah's answer must be run multiple times if items are duplicated more than once.

Here's a solution that doesn't delete any data, and keeps the data in the original table the entire time, allowing for duplicates to be deleted while keeping the table 'live':

alter table tableA add column duplicate tinyint(1) not null default '0';

update tableA set
duplicate=if(@member_id=member_id
             and @quiz_num=quiz_num
             and @question_num=question_num
             and @answer_num=answer_num,1,0),
member_id=(@member_id:=member_id),
quiz_num=(@quiz_num:=quiz_num),
question_num=(@question_num:=question_num),
answer_num=(@answer_num:=answer_num)
order by member_id, quiz_num, question_num, answer_num;

delete from tableA where duplicate=1;

alter table tableA drop column duplicate;

This basically checks to see if the current row is the same as the last row, and if it is, marks it as duplicate (the order statement ensures that duplicates will show up next to each other). Then you delete the duplicate records. I remove the duplicate column at the end to bring it back to its original state.

It looks like alter table ignore also might go away soon: http://dev.mysql.com/worklog/task/?id=7395

回答8:

An alternative way would be to create a new temporary table with same structure.

CREATE TABLE temp_table AS SELECT * FROM original_table LIMIT 0

Then create the primary key in the table.

ALTER TABLE temp_table ADD PRIMARY KEY (primary-key-field)

Finally copy all records from the original table while ignoring the duplicate records.

INSERT IGNORE INTO temp_table AS SELECT * FROM original_table

Now you can delete the original table and rename the new table.

DROP TABLE original_table
RENAME TABLE temp_table TO original_table

来源：https://stackoverflow.com/questions/14046355/how-do-i-delete-all-the-duplicate-records-in-a-mysql-table-without-temp-tables

标签

mysql

sql

duplicates

sql-delete

unique-index