[mysql] Include headers when using SELECT INTO OUTFILE?

Is it possible to include the headers somehow when using the MySQL INTO OUTFILE?

This question is related to mysql into-outfile

The answer is


Actually you can make it work even with an ORDER BY.

Just needs some trickery in the order by statement - we use a case statement and replace the header value with some other value that is guaranteed to sort first in the list (obviously this is dependant on the type of field and whether you are sorting ASC or DESC)

Let's say you have three fields, name (varchar), is_active (bool), date_something_happens (date), and you want to sort the second two descending:

select 
        'name'
      , 'is_active' as is_active
      , date_something_happens as 'date_something_happens'

 union all

 select name, is_active, date_something_happens

 from
    my_table

 order by
     (case is_active when 'is_active' then 0 else is_active end) desc
   , (case date when 'date' then '9999-12-30' else date end) desc

I faced similar problem while executing mysql query on large tables in NodeJS. The approach which I followed to include headers in my CSV file is as follows

  1. Use OUTFILE query to prepare file without headers

        SELECT * INTO OUTFILE [FILE_NAME] FIELDS TERMINATED BY ',' OPTIONALLY ENCLOSED 
        BY '\"' LINES TERMINATED BY '\n' FROM [TABLE_NAME]
    
  2. Fetch column headers for the table used in point 1

        select GROUP_CONCAT(CONCAT(\"\",COLUMN_NAME,\"\")) as col_names from 
        INFORMATION_SCHEMA.COLUMNS WHERE TABLE_NAME = [TABLE_NAME] AND TABLE_SCHEMA 
        = [DATABASE_NAME] ORDER BY ORDINAL_POSITION
    
  3. Append the column headers to the file created in step 1 using prepend-file npm package

Execution of each step was controlled using promises in NodeJS.


So, if all the columns in my_table are a character data type, we can combine the top answers (by Joe, matt and evilguc) together, to get the header added automatically in one 'simple' SQL query, e.g.

select * from (
  (select column_name
    from information_schema.columns
    where table_name = 'my_table'
    and table_schema = 'my_schema'
    order by ordinal_position)
  union all
  (select *  // potentially complex SELECT statement with WHERE, ORDER BY, GROUP BY etc.
  from my_table)) as tbl
into outfile '/path/outfile'
fields terminated by ',' optionally enclosed by '"' escaped by '\\'
lines terminated by '\n';

where the last couple of lines make the output csv.

Note that this may be slow if my_table is very large.


Here is a way to get the header titles from the column names dynamically.

/* Change table_name and database_name */
SET @table_name = 'table_name';
SET @table_schema = 'database_name';
SET @default_group_concat_max_len = (SELECT @@group_concat_max_len);

/* Sets Group Concat Max Limit larger for tables with a lot of columns */
SET SESSION group_concat_max_len = 1000000;

SET @col_names = (
  SELECT GROUP_CONCAT(QUOTE(`column_name`)) AS columns
  FROM information_schema.columns
  WHERE table_schema = @table_schema
  AND table_name = @table_name);

SET @cols = CONCAT('(SELECT ', @col_names, ')');

SET @query = CONCAT('(SELECT * FROM ', @table_schema, '.', @table_name,
  ' INTO OUTFILE \'/tmp/your_csv_file.csv\'
  FIELDS ENCLOSED BY \'\\\'\' TERMINATED BY \'\t\' ESCAPED BY \'\'
  LINES TERMINATED BY \'\n\')');

/* Concatenates column names to query */
SET @sql = CONCAT(@cols, ' UNION ALL ', @query);

/* Resets Group Contact Max Limit back to original value */
SET SESSION group_concat_max_len = @default_group_concat_max_len;

PREPARE stmt FROM @sql;
EXECUTE stmt;
DEALLOCATE PREPARE stmt;

This is an alternative cheat if you are familiar with Python or R, and your table can fit into memory.

Import the SQL table into Python or R and then export from there as a CSV and you'll get the column names as well as the data.

Here's how I do it using R, requires the RMySQL library:

db <- dbConnect(MySQL(), user='user', password='password', dbname='myschema', host='localhost')

query <- dbSendQuery(db, "select * from mytable")
dataset <- fetch(query, n=-1)

write.csv(dataset, 'mytable_backup.csv')

It's a bit of a cheat but I found this was a quick workaround when my number of columns was too long to use the concat method above. Note: R will add a 'row.names' column at the start of the CSV so you'll want to drop that if you do need to rely on the CSV to recreate the table.


an example from my database table name sensor with colums (id,time,unit)

select ('id') as id, ('time') as time, ('unit') as unit
UNION ALL
SELECT * INTO OUTFILE 'C:/Users/User/Downloads/data.csv'
  FIELDS TERMINATED BY ',' OPTIONALLY ENCLOSED BY '"'
  LINES TERMINATED BY '\n'
  FROM sensor

SELECT 'ColName1', 'ColName2', 'ColName3'
UNION ALL
SELECT ColName1, ColName2, ColName3
    FROM YourTable
    INTO OUTFILE 'c:\\datasheet.csv' FIELDS TERMINATED BY ',' OPTIONALLY ENCLOSED BY '"' LINES TERMINATED BY '\n' 

The easiest way is to hard code the columns yourself to better control the output file:

SELECT 'ColName1', 'ColName2', 'ColName3'
UNION ALL
SELECT ColName1, ColName2, ColName3
    FROM YourTable
    INTO OUTFILE '/path/outfile'

I think if you use a UNION it will work:

select 'header 1', 'header 2', ...
union
select col1, col2, ... from ...

I don't know of a way to specify the headers with the INTO OUTFILE syntax directly.


Since the 'include-headers' functionality doesn't seem to be build-in yet, and most "solutions" here need to type the columns names manually, and/or don't even take joins into account, I'd recommand to get around the problem.

  • The best alternative I found so far is using a decent tool (I use HeidiSQL).
    Put your request, select the grid, just right click and export to a file. It got all necessary options for a clean export, ans should handle most needs.

  • In the same idea, user3037511's approach works fine, and can be automated easily.
    Just launch your request with some command line to get your headers. You may get the data with a SELECT INTO OUTFILE... or by running your query without the limit, yours to choose.

    Note that output redirect to a file works like a charm on both Linux AND Windows.


This makes me want to highlight that 80% of the time, when I want to use SELECT FROM INFILE or SELECT INTO OUTFILE, I end-up using something else due to some limitations (here, the absence of a 'headers options', on an AWS-RDS, the missing rights, and so on.)

Hence, I don't exactly answer to the op's question... but it should answer his needs :)
EDIT : and to actually answer his question : no
As of 2017-09-07, you just can't include headers if you stick with the SELECT INTO OUTFILE command
:|


I simply make 2 queries, first to get query output (limit 1) with column names (no hardcode, no problems with Joins, Order by, custom column names, etc), and second to make query itself, and combine files into one CSV file:

CSVHEAD=`/usr/bin/mysql $CONNECTION_STRING -e "$QUERY limit 1;"|head -n1|xargs|sed -e "s/ /'\;'/g"`
echo "\'$CSVHEAD\'" > $TMP/head.txt
/usr/bin/mysql $CONNECTION_STRING -e "$QUERY into outfile '${TMP}/data.txt' fields terminated by ';' optionally enclosed by '\"' escaped by '' lines terminated by '\r\n';"
cat $TMP/head.txt $TMP/data.txt > $TMP/data.csv

I had no luck with any of these, so after finding a solution, I wanted to add it to the prior answers. Python==3.8.6 MySQL==8.0.19

(Forgive my lack of SO formatting foo. Somebody please clean up.)

Note a couple of things:

First, the query to return column names is unforgiving of punctuation. Using ` backticks or leaving out ' quote around the 'schema_name' and 'table_name' will throw an "unknown column" error.

WHERE TABLE_SCHEMA = 'schema' AND TABLE_NAME = 'table'

Second, the column header names return as a single-entity tuple with all the column names concatenated in one quoted string. Convert to quoted list was easy, but not intuitive (for me at least).

headers_list = headers_result[0].split(",")

Third, cursor must be buffered or the "lazy" thing will not fetch your results as you need them. For very large tables, memory could be an issue. Perhaps chunking would solve that problem.

cur = db.cursor(buffered=True)

Last, all types of UNION attempts yielded errors for me. By zipping the whole mess into a list of dicts, it became trivial to write to a csv, using csv.DictWriter.

headers_sql = """   
SELECT 
GROUP_CONCAT(CONCAT(COLUMN_NAME) ORDER BY ORDINAL_POSITION)   
FROM INFORMATION_SCHEMA.COLUMNS   
WHERE TABLE_SCHEMA = 'schema' AND TABLE_NAME = 'table';   
"""" 

cur = db.cursor(buffered=True) 
cur.execute(header_sql) 
headers_result = cur.fetchone() 
headers_list = headers_result[0].split(",")

rows_sql = """   SELECT * FROM schema.table;   """" 

data = cur.execute(rows_sql) 
data_rows = cur.fetchall() 
data_as_list_of_dicts = [dict(zip(headers_list, row)) for row in data_rows]

with open(csv_destination_file, 'w', encoding='utf-8') as destination_file_opened: 
    dict_writer = csv.DictWriter(destination_file_opened, fieldnames=headers_list) 
    dict_writer.writeheader()   for dict in dict_list:
        dict_writer.writerow(dict)

For complex select with ORDER BY I use the following:

SELECT * FROM (
    SELECT 'Column name #1', 'Column name #2', 'Column name ##'
    UNION ALL
    (
        // complex SELECT statement with WHERE, ORDER BY, GROUP BY etc.
    )
) resulting_set
INTO OUTFILE '/path/to/file';

I would like to add to the answer provided by Sangam Belose. Here's his code:

select ('id') as id, ('time') as time, ('unit') as unit
UNION ALL
SELECT * INTO OUTFILE 'C:/Users/User/Downloads/data.csv'
  FIELDS TERMINATED BY ',' OPTIONALLY ENCLOSED BY '"'
  LINES TERMINATED BY '\n'
  FROM sensor

However, if you have not set up your "secure_file_priv" within the variables, it may not work. For that, check the folder set on that variable by:

SHOW VARIABLES LIKE "secure_file_priv"

The output should look like this:

mysql> show variables like "%secure_file_priv%";
+------------------+------------------------------------------------+
| Variable_name    | Value                                          |
+------------------+------------------------------------------------+
| secure_file_priv | C:\ProgramData\MySQL\MySQL Server 8.0\Uploads\ |
+------------------+------------------------------------------------+
1 row in set, 1 warning (0.00 sec)

You can either change this variable or change the query to output the file to the default path showing.


The solution provided by Joe Steanelli works, but making a list of columns is inconvenient when dozens or hundreds of columns are involved. Here's how to get column list of table my_table in my_schema.

-- override GROUP_CONCAT limit of 1024 characters to avoid a truncated result
set session group_concat_max_len = 1000000;

select GROUP_CONCAT(CONCAT("'",COLUMN_NAME,"'"))
from INFORMATION_SCHEMA.COLUMNS
WHERE TABLE_NAME = 'my_table'
AND TABLE_SCHEMA = 'my_schema'
order BY ORDINAL_POSITION

Now you can copy & paste the resulting row as first statement in Joe's method.


This will alow you to have ordered columns and/or a limit

SELECT 'ColName1', 'ColName2', 'ColName3'
UNION ALL
SELECT * from (SELECT ColName1, ColName2, ColName3
    FROM YourTable order by ColName1 limit 3) a
    INTO OUTFILE '/path/outfile';

I was writing my code in PHP, and I had a bit of trouble using concat and union functions, and also did not use SQL variables, any ways I got it to work, here is my code:

//first I connected to the information_scheme DB

$headercon=mysqli_connect("localhost", "USERNAME", "PASSWORD", "information_schema");

//took the healders out in a string (I could not get the concat function to work, so I wrote a loop for it)

    $headers = '';
    $sql = "SELECT column_name AS columns FROM `COLUMNS` WHERE table_schema = 'YOUR_DB_NAME' AND table_name = 'YOUR_TABLE_NAME'";
    $result = $headercon->query($sql);
    while($row = $result->fetch_row())
    {
        $headers = $headers . "'" . $row[0] . "', ";
    }
$headers = substr("$headers", 0, -2);

// connect to the DB of interest

$con=mysqli_connect("localhost", "USERNAME", "PASSWORD", "YOUR_DB_NAME");

// export the results to csv
$sql4 = "SELECT $headers UNION SELECT * FROM YOUR_TABLE_NAME WHERE ... INTO OUTFILE '/output.csv' FIELDS TERMINATED BY ','";
$result4 = $con->query($sql4);

You can use prepared statement with lucek's answer and export dynamically table with columns name in CSV :

--If your table has too many columns
SET GLOBAL group_concat_max_len = 100000000;
--Prepared statement
SET @SQL = ( select CONCAT('SELECT * INTO OUTFILE \'YOUR_PATH\' FIELDS TERMINATED BY \',\' OPTIONALLY ENCLOSED BY \'"\' ESCAPED BY \'\' LINES TERMINATED BY \'\\n\' FROM (SELECT ', GROUP_CONCAT(CONCAT("'",COLUMN_NAME,"'")),' UNION select * from YOUR_TABLE) as tmp') from INFORMATION_SCHEMA.COLUMNS WHERE TABLE_NAME = 'YOUR_TABLE' AND TABLE_SCHEMA = 'YOUR_SCHEMA' order BY ORDINAL_POSITION );
--Execute it
PREPARE stmt FROM @SQL;
EXECUTE stmt;

Thank lucek.


MySQL alone isn't enough to do this simply. Below is a PHP script that will output columns and data to CSV.

Enter your database name and tables near the top.

<?php

set_time_limit( 24192000 );
ini_set( 'memory_limit', '-1' );
setlocale( LC_CTYPE, 'en_US.UTF-8' );
mb_regex_encoding( 'UTF-8' );

$dbn = 'DB_NAME';
$tbls = array(
'TABLE1',
'TABLE2',
'TABLE3'
);

$db = new PDO( 'mysql:host=localhost;dbname=' . $dbn . ';charset=UTF8', 'root', 'pass' );

foreach( $tbls as $tbl )
{
    echo $tbl . "\n";
    $path = '/var/lib/mysql/' . $tbl . '.csv';

    $colStr = '';
    $cols = $db->query( 'SELECT COLUMN_NAME AS `column` FROM INFORMATION_SCHEMA.COLUMNS WHERE TABLE_NAME = "' . $tbl . '" AND TABLE_SCHEMA = "' . $dbn . '"' )->fetchAll( PDO::FETCH_COLUMN );
    foreach( $cols as $col )
    {
        if( $colStr ) $colStr .= ', ';
        $colStr .= '"' . $col . '"';
    }

    $db->query(
    'SELECT *
    FROM
    (
        SELECT ' . $colStr . '
        UNION ALL
        SELECT * FROM ' . $tbl . '
    ) AS sub
    INTO OUTFILE "' . $path . '"
    FIELDS TERMINATED BY ","
    ENCLOSED BY "\""
    LINES TERMINATED BY "\n"'
    );

    exec( 'gzip ' . $path );

    print_r( $db->errorInfo() );
}

?>

You'll need this to be the directory you'd like to output to. MySQL needs to have the ability to write to the directory.

$path = '/var/lib/mysql/' . $tbl . '.csv';

You can edit the CSV export options in the query:

INTO OUTFILE "' . $path . '"
FIELDS TERMINATED BY ","
ENCLOSED BY "\""
LINES TERMINATED BY "\n"'

At the end there is an exec call to GZip the CSV.