How can I output MySQL query results in CSV format?

Question

Is there an easy way to run a MySQL query from the Linux command line and output the results in CSV format?

Here's what I'm doing now:

mysql -u uid -ppwd -D dbname << EOQ | sed -e 's/        /,/g' | tee list.csv
select id, concat("\"",name,"\"") as name
from students
EOQ

It gets messy when there are a lot of columns that need to be surrounded by quotes, or if there are quotes in the results that need to be escaped.

Answer 1

From Save MySQL query results into a text or CSV file:

SELECT order_id,product_name,qty
FROM orders
WHERE foo = 'bar'
INTO OUTFILE '/var/lib/mysql-files/orders.csv'
FIELDS TERMINATED BY ','
ENCLOSED BY '"'
LINES TERMINATED BY '\n';

Note: That syntax may need to be reordered to

SELECT order_id,product_name,qty
INTO OUTFILE '/var/lib/mysql-files/orders.csv'
FIELDS TERMINATED BY ','
ENCLOSED BY '"'
LINES TERMINATED BY '\n'
FROM orders
WHERE foo = 'bar';

in more recent versions of MySQL.

Using this command, columns names will not be exported.

Also note that /var/lib/mysql-files/orders.csv will be on the server that is running MySQL. The user that the MySQL process is running under must have permissions to write to the directory chosen, or the command will fail.

If you want to write output to your local machine from a remote server (especially a hosted or virtualize machine such as Heroku or Amazon RDS), this solution is not suitable.

Answer 2

mysql your_database --password=foo < my_requests.sql > out.tsv

This produces a tab-separated format. If you are certain that commas do not appear in any of the column data (and neither do tabs), you can use this pipe command to get a true CSV (thanks to user John Carter):

... .sql | sed 's/\t/,/g' > out.csv

Answer 3

mysql --batch, -B

Print results using tab as the column separator, with each row on a new line. With this option, mysql does not use the history file. Batch mode results in non-tabular output format and escaping of special characters. Escaping may be disabled by using raw mode; see the description for the --raw option.

This will give you a tab-separated file. Since commas (or strings containing comma) are not escaped, it is not straightforward to change the delimiter to comma.

Answer 4

Here's a fairly gnarly way of doing it^[1]:

mysql --user=wibble --password mydatabasename -B -e "select * from vehicle_categories;" | sed "s/'/\'/;s/\t/\",\"/g;s/^/\"/;s/$/\"/;s/\n//g" > vehicle_categories.csv

It works pretty well. Once again, though, a regular expression proves write-only.

---

Regex Explanation:

• s/// means substitute what's between the first // with what's between the second //
• the "g" at the end is a modifier that means "all instance, not just first"
• ^ (in this context) means beginning of line
• $ (in this context) means end of line

So, putting it all together:

s/'/\'/          Replace ' with \'
s/\t/\",\"/g     Replace all \t (tab) with ","
s/^/\"/          at the beginning of the line place a "
s/$/\"/          At the end of the line, place a "
s/\n//g          Replace all \n (newline) with nothing

^{[1] I found it somewhere and can't take any credit.}

Answer 5

Pipe it through 'tr' (Unix/Cygwin only):

mysql <database> -e "<query here>" | tr '\t' ',' > data.csv

N.B.: This handles neither embedded commas, nor embedded tabs.

Answer 6

This saved me a couple of times. It is fast and it works!

--batch Print results using tab as the column separator, with each row on a new line.

--raw disables character escaping (\n, \t, \0, and \)

Example:

mysql -udemo_user -p -h127.0.0.1 --port=3306 \
   --default-character-set=utf8mb4 --database=demo_database \
   --batch --raw < /tmp/demo_sql_query.sql > /tmp/demo_csv_export.tsv

For completeness you could convert to CSV (but be careful because tabs could be inside field values - e.g., text fields)

tr '\t' ',' < file.tsv > file.csv

Answer 7

The OUTFILE solution given by Paul Tomblin causes a file to be written on the MySQL server itself, so this will work only if you have FILE access, as well as login access or other means for retrieving the file from that box.

If you don't have such access, and tab-delimited output is a reasonable substitute for CSV (e.g., if your end goal is to import to Excel), then serbaut's solution (using mysql --batch and optionally --raw) is the way to go.

Answer 8

MySQL Workbench can export recordsets to CSV, and it seems to handle commas in fields very well. The CSV opens up in OpenOffice Calc fine.

Answer 9

All of the solutions here to date, except the MySQL Workbench one, are incorrect and quite possibly unsafe (i.e., security issues) for at least some possible content in the MySQL database.

MySQL Workbench (and similarly phpMyAdmin) provide a formally correct solution, but they are designed for downloading the output to a user's location. They're not so useful for things like automating data export.

It is not possible to generate reliably correct CSV content from the output of mysql -B -e 'SELECT ...' because that cannot encode carriage returns and white space in fields. The '-s' flag to mysql does do backslash escaping, and might lead to a correct solution. However, using a scripting language (one with decent internal data structures that is, not Bash), and libraries where the encoding issues have already been carefully worked out is far safer.

I thought about writing a script for this, but as soon as I thought about what I'd call it, it occurred to me to search for preexisting work by the same name. While I haven't gone over it thoroughly, mysql2csv looks promising. Depending on your application, the YAML approach to specifying the SQL commands might or might not appeal though. I'm also not thrilled with the requirement for a more recent version of Ruby than comes as standard with my Ubuntu 12.04 (Precise Pangolin) laptop or Debian 6.0 (Squeeze) servers. Yes, I know I could use RVM, but I'd rather not maintain that for such a simple purpose.

Answer 10

Many of the answers on this page are weak, because they don't handle the general case of what can occur in CSV format. E.g., commas and quotes embedded in fields and other conditions that always come up eventually. We need a general solution that works for all valid CSV input data.

Here's a simple and strong solution in Python:

#!/usr/bin/env python

import csv
import sys

tab_in = csv.reader(sys.stdin, dialect=csv.excel_tab)
comma_out = csv.writer(sys.stdout, dialect=csv.excel)

for row in tab_in:
    comma_out.writerow(row)

Name that file tab2csv, put it on your path, give it execute permissions, then use it like this:

mysql OTHER_OPTIONS --batch --execute='select * from whatever;' | tab2csv > outfile.csv

The Python CSV-handling functions cover corner cases for CSV input format(s).

This could be improved to handle very large files via a streaming approach.

Answer 11

Use:

mysql your_database -p < my_requests.sql | awk '{print $1","$2}' > out.csv

Answer 12

From your command line, you can do this:

mysql -h *hostname* -P *port number* --database=*database_name* -u *username* -p -e *your SQL query* | sed 's/\t/","/g;s/^/"/;s/$/"/;s/\n//g' > *output_file_name.csv*

Credits: Exporting table from Amazon RDS into a CSV file

Answer 13

I encountered the same problem and Paul's Answer wasn't an option since it was Amazon RDS. Replacing the tab with the commas did not work as the data had embedded commas and tabs. I found that mycli, which is a drop-in alternative for the mysql-client, supports CSV output out of the box with the --csv flag:

mycli db_name --csv -e "select * from flowers" > flowers.csv

Answer 14

This answer uses Python and a popular third party library, PyMySQL. I'm adding it because Python's csv library is powerful enough to correctly handle many different flavors of .csv and no other answers are using Python code to interact with the database.

import contextlib
import csv
import datetime
import os

# https://github.com/PyMySQL/PyMySQL
import pymysql

SQL_QUERY = """
SELECT * FROM my_table WHERE my_attribute = 'my_attribute';
"""

# embedding passwords in code gets nasty when you use version control
# the environment is not much better, but this is an example
# https://stackoverflow.com/questions/12461484
SQL_USER = os.environ['SQL_USER']
SQL_PASS = os.environ['SQL_PASS']

connection = pymysql.connect(host='localhost',
                             user=SQL_USER,
                             password=SQL_PASS,
                             db='dbname')

with contextlib.closing(connection):
    with connection.cursor() as cursor:
        cursor.execute(SQL_QUERY)
        # Hope you have enough memory :)
        results = cursor.fetchall()

output_file = 'my_query-{}.csv'.format(datetime.datetime.today().strftime('%Y-%m-%d'))
with open(output_file, 'w', newline='') as csvfile:
    # http://stackoverflow.com/a/17725590/2958070 about lineterminator
    csv_writer = csv.writer(csvfile, lineterminator='\n')
    csv_writer.writerows(results)

Answer 15

Also, if you're performing the query on the Bash command line, I believe the tr command can be used to substitute the default tabs to arbitrary delimiters.

$ echo "SELECT * FROM Table123" | mysql Database456 | tr "\t" ,

Answer 16

This is simple, and it works on anything without needing batch mode or output files:

select concat_ws(',',
    concat('"', replace(field1, '"', '""'), '"'),
    concat('"', replace(field2, '"', '""'), '"'),
    concat('"', replace(field3, '"', '""'), '"'))

from your_table where etc;

Explanation:

1. Replace " with "" in each field --> replace(field1, '"', '""')
2. Surround each result in quotation marks --> concat('"', result1, '"')
3. Place a comma between each quoted result --> concat_ws(',', quoted1, quoted2, ...)

That's it!

Answer 17

You can have a MySQL table that uses the CSV engine.

Then you will have a file on your hard disk that will always be in a CSV format which you could just copy without processing it.

Answer 18

To expand on previous answers, the following one-liner exports a single table as a tab-separated file. It's suitable for automation, exporting the database every day or so.

mysql -B -D mydatabase -e 'select * from mytable'

Conveniently, we can use the same technique to list out MySQL's tables, and to describe the fields on a single table:

mysql -B -D mydatabase -e 'show tables'

mysql -B -D mydatabase -e 'desc users'

Field   Type    Null    Key Default Extra
id  int(11) NO  PRI NULL    auto_increment
email   varchar(128)    NO  UNI NULL    
lastName    varchar(100)    YES     NULL    
title   varchar(128)    YES UNI NULL    
userName    varchar(128)    YES UNI NULL    
firstName   varchar(100)    YES     NULL    

Answer 19

Building on user7610, here is the best way to do it. With mysql outfile there were 60 mins of file ownership and overwriting problems.

It's not cool, but it worked in 5 mins.

php csvdump.php localhost root password database tablename > whatever-you-like.csv

<?php

$server = $argv[1];
$user = $argv[2];
$password = $argv[3];
$db = $argv[4];
$table = $argv[5];

mysql_connect($server, $user, $password) or die(mysql_error());
mysql_select_db($db) or die(mysql_error());

// fetch the data
$rows = mysql_query('SELECT * FROM ' . $table);
$rows || die(mysql_error());


// create a file pointer connected to the output stream
$output = fopen('php://output', 'w');

// output the column headings

$fields = [];
for($i = 0; $i < mysql_num_fields($rows); $i++) {
    $field_info = mysql_fetch_field($rows, $i);
    $fields[] = $field_info->name;
}
fputcsv($output, $fields);

// loop over the rows, outputting them
while ($row = mysql_fetch_assoc($rows)) fputcsv($output, $row);

?>

Answer 20

Here's what I do:

echo $QUERY | \
  mysql -B  $MYSQL_OPTS | \
  perl -F"\t" -lane 'print join ",", map {s/"/""/g; /^[\d.]+$/ ? $_ : qq("$_")} @F ' | \
  mail -s 'report' person@address

The Perl script (snipped from elsewhere) does a nice job of converting the tab spaced fields to CSV.

Answer 21

Not exactly as a CSV format, but the tee command from the MySQL client can be used to save the output into a local file:

tee foobar.txt
SELECT foo FROM bar;

You can disable it using notee.

The problem with SELECT … INTO OUTFILE …; is that it requires permission to write files at the server.

Answer 22

In my case from table_name ..... before INTO OUTFILE ..... gives an error:

Unexpected ordering of clauses. (near "FROM" at position 10)

What works for me:

SELECT *
INTO OUTFILE '/Volumes/Development/sql/sql/enabled_contacts.csv'
FIELDS TERMINATED BY ','
ENCLOSED BY '"'
LINES TERMINATED BY '\n'
FROM table_name
WHERE column_name = 'value'

Answer 23

What worked for me:

SELECT *
FROM students
WHERE foo = 'bar'
LIMIT 0,1200000
INTO OUTFILE './students-1200000.csv'
FIELDS TERMINATED BY ',' ESCAPED BY '"'
ENCLOSED BY '"'
LINES TERMINATED BY '\r\n';

None of the solutions on this thread worked for my particular case. I had pretty JSON data inside one of the columns, which would get messed up in my CSV output. For those with a similar problem, try lines terminated by \r\n instead.

Also another problem for those trying to open the CSV with Microsoft Excel, keep in mind there is a limit of 32,767 characters that a single cell can hold, above that it overflows to the rows below. To identify which records in a column have the issue, use the query below. You can then truncate those records or handle them as you'd like.

SELECT id,name,CHAR_LENGTH(json_student_description) AS 'character length'
FROM students
WHERE CHAR_LENGTH(json_student_description)>32767;

Answer 24

Using the solution posted by Tim Harding, I created this Bash script to facilitate the process (root password is requested, but you can modify the script easily to ask for any other user):

#!/bin/bash

if [ "$1" == "" ];then
    echo "Usage: $0 DATABASE TABLE [MYSQL EXTRA COMMANDS]"
    exit
fi

DBNAME=$1
TABLE=$2
FNAME=$1.$2.csv
MCOMM=$3

echo "MySQL password: "
stty -echo
read PASS
stty echo

mysql -uroot -p$PASS $MCOMM $DBNAME -B -e "SELECT * FROM $TABLE;" | sed "s/'/\'/;s/\t/\",\"/g;s/^/\"/;s/$/\"/;s/\n//g" > $FNAME

It will create a file named: database.table.csv

Answer 25

If you have PHP set up on the server, you can use mysql2csv to export an (actually valid) CSV file for an arbitrary MySQL query. See my answer at MySQL - SELECT * INTO OUTFILE LOCAL ? for a little more context/info.

I tried to maintain the option names from mysql so it should be sufficient to provide the --file and --query options:

./mysql2csv --file="/tmp/result.csv" --query='SELECT 1 as foo, 2 as bar;' --user="username" --password="password"

"Install" mysql2csv via

wget https://gist.githubusercontent.com/paslandau/37bf787eab1b84fc7ae679d1823cf401/raw/29a48bb0a43f6750858e1ddec054d3552f3cbc45/mysql2csv -O mysql2csv -q && (sha256sum mysql2csv | cmp <(echo "b109535b29733bd596ecc8608e008732e617e97906f119c66dd7cf6ab2865a65  mysql2csv") || (echo "ERROR comparing hash, Found:" ;sha256sum mysql2csv) ) && chmod +x mysql2csv

(Download content of the gist, check checksum and make it executable.)

Answer 26

The following produces tab-delimited and valid CSV output. Unlike most of the other answers, this technique correctly handles escaping of tabs, commas, quotes, and new lines without any stream filter like sed, AWK, or tr.

The example shows how to pipe a remote MySQL table directly into a local SQLite database using streams. This works without FILE permission or SELECT INTO OUTFILE permission. I have added new lines for readability.

mysql -B -C --raw -u 'username' --password='password' --host='hostname' 'databasename'
-e 'SELECT
    CONCAT('\''"'\'',REPLACE(`id`,'\''"'\'', '\''""'\''),'\''"'\'') AS '\''id'\'',
    CONCAT('\''"'\'',REPLACE(`value`,'\''"'\'', '\''""'\''),'\''"'\'') AS '\''value'\''
    FROM sampledata'
2>/dev/null | sqlite3 -csv -separator $'\t' mydb.db '.import /dev/stdin mycsvtable'

The 2>/dev/null is needed to suppress the warning about the password on the command line.

If your data has NULLs, you can use the IFNULL() function in the query.

Answer 27

A simple solution in Python that writes a standard-format CSV file with headers and writes data as a stream (low memory use):

import csv

def export_table(connection, table_name, output_filename):
    cursor = connection.cursor()
    cursor.execute("SELECT * FROM " + table_name)

    # thanks to https://gist.github.com/madan712/f27ac3b703a541abbcd63871a4a56636 for this hint
    header = [descriptor[0] for descriptor in cursor.description]

    with open(output_filename, 'w') as csvfile:
        csv_writer = csv.writer(csvfile, dialect='excel')
        csv_writer.writerow(header)
        for row in cursor:
            csv_writer.writerow(row)

You could use it like:

import mysql.connector as mysql
# (or https://github.com/PyMySQL/PyMySQL should work but I haven't tested it)

db = mysql.connect(
    host="localhost",
    user="USERNAME",
    db="DATABASE_NAME",
    port=9999)

for table_name in ['table1', 'table2']:
    export_table(db, table_name, table_name + '.csv')

db.close()

For simplicity, this intentionally doesn't include some fancier stuff from another answer like using an environment variable for credentials, contextlib, etc. There is a subtlety mentioned there about line endings that I haven't evaluated.

Answer 28

Tiny Bash script for doing simple query to CSV dumps, inspired by Tim Harding's answer.

#!/bin/bash

# $1 = query to execute
# $2 = outfile
# $3 = mysql database name
# $4 = mysql username

if [ -z "$1" ]; then
    echo "Query not given"
    exit 1
fi

if [ -z "$2" ]; then
    echo "Outfile not given"
    exit 1
fi

MYSQL_DB=""
MYSQL_USER="root"

if [ ! -z "$3" ]; then
    MYSQL_DB=$3
fi

if [ ! -z "$4" ]; then
    MYSQL_USER=$4
fi

if [ -z "$MYSQL_DB" ]; then
    echo "Database name not given"
    exit 1
fi

if [ -z "$MYSQL_USER" ]; then
    echo "Database user not given"
    exit 1
fi

mysql -u $MYSQL_USER -p -D $MYSQL_DB -B -s -e "$1" | sed "s/'/\'/;s/\t/\",\"/g;s/^/\"/;s/$/\"/;s/\n//g" > $2
echo "Written to $2"

Answer 29

If you are getting an error of secure-file-priv then, also after shifting your destination file location inside the C:\ProgramData\MySQL\MySQL Server 8.0\Uploads and also after then the query -

SELECT * FROM attendance INTO OUTFILE 'C:\ProgramData\MySQL\MySQL Server 8.0\Uploads\FileName.csv' FIELDS TERMINATED BY ',' ENCLOSED BY '"' LINES TERMINATED BY '\n';

is not working, you have to just change \(backsplash) from the query to / (forwardsplash)

And that works!!

Example:

SELECT * FROM attendance INTO OUTFILE 'C:/ProgramData/MySQL/MySQL Server 8.0/Uploads/FileName.csv' FIELDS TERMINATED BY ',' ENCLOSED BY '"' LINES TERMINATED BY '\n';

Each time when you run the successful query, it will generate the new CSV file each time! Cool, right?

Answer 30

The following Bash script works for me. It optionally also gets the schema for the requested tables.

#!/bin/bash
#
# Export MySQL data to CSV
#https://stackoverflow.com/questions/356578/how-to-output-mysql-query-results-in-csv-format
#

# ANSI colors
#http://www.csc.uvic.ca/~sae/seng265/fall04/tips/s265s047-tips/bash-using-colors.html
blue='\033[0;34m'
red='\033[0;31m'
green='\033[0;32m' # '\e[1;32m' is too bright for white bg.
endColor='\033[0m'

#
# A colored message
#   params:
#     1: l_color - the color of the message
#     2: l_msg - the message to display
#
color_msg() {
  local l_color="$1"
  local l_msg="$2"
  echo -e "${l_color}$l_msg${endColor}"
}


#
# Error
#
# Show the given error message on standard error and exit
#
#   Parameters:
#     1: l_msg - the error message to display
#
error() {
  local l_msg="$1"
  # Use ANSI red for error
  color_msg $red "Error:" 1>&2
  color_msg $red "\t$l_msg" 1>&2
  usage
}

#
# Display usage
#
usage() {
  echo "usage: $0 [-h|--help]" 1>&2
  echo "               -o  | --output      csvdirectory"    1>&2
  echo "               -d  | --database    database"   1>&2
  echo "               -t  | --tables      tables"     1>&2
  echo "               -p  | --password    password"   1>&2
  echo "               -u  | --user        user"       1>&2
  echo "               -hs | --host        host"       1>&2
  echo "               -gs | --get-schema"             1>&2
  echo "" 1>&2
  echo "     output: output CSV directory to export MySQL data into" 1>&2
  echo "" 1>&2
  echo "         user: MySQL user" 1>&2
  echo "     password: MySQL password" 1>&2
  echo "" 1>&2
  echo "     database: target database" 1>&2
  echo "       tables: tables to export" 1>&2
  echo "         host: host of target database" 1>&2
  echo "" 1>&2
  echo "  -h|--help: show help" 1>&2
  exit 1
}

#
# show help
#
help() {
  echo "$0 Help" 1>&2
  echo "===========" 1>&2
  echo "$0 exports a CSV file from a MySQL database optionally limiting to a list of tables" 1>&2
  echo "   example: $0 --database=cms --user=scott --password=tiger  --tables=person --output person.csv" 1>&2
  echo "" 1>&2
  usage
}

domysql() {
  mysql --host $host -u$user --password=$password $database
}

getcolumns() {
  local l_table="$1"
  echo "describe $l_table" | domysql | cut -f1 | grep -v "Field" | grep -v "Warning" | paste -sd "," - 2>/dev/null
}

host="localhost"
mysqlfiles="/var/lib/mysql-files/"

# Parse command line options
while true; do
  #echo "option $1"
  case "$1" in
    # Options without arguments
    -h|--help) usage;;
    -d|--database)     database="$2" ; shift ;;
    -t|--tables)       tables="$2" ; shift ;;
    -o|--output)       csvoutput="$2" ; shift ;;
    -u|--user)         user="$2" ; shift ;;
    -hs|--host)        host="$2" ; shift ;;
    -p|--password)     password="$2" ; shift ;;
    -gs|--get-schema)  option="getschema";;
    (--) shift; break;;
    (-*) echo "$0: error - unrecognized option $1" 1>&2; usage;;
    (*) break;;
  esac
  shift
done

# Checks
if [ "$csvoutput" == "" ]
then
  error "output CSV directory is not set"
fi
if [ "$database" == "" ]
then
  error "MySQL database is not set"
fi
if [ "$user" == "" ]
then
  error "MySQL user is not set"
fi
if [ "$password" == "" ]
then
  error "MySQL password is not set"
fi

color_msg $blue "exporting tables of database $database"
if [ "$tables" = "" ]
then
tables=$(echo "show tables" | domysql)
fi

case $option in
  getschema)
   rm $csvoutput$database.schema
   for table in $tables
   do
     color_msg $blue "getting schema for $table"
     echo -n "$table:" >> $csvoutput$database.schema
     getcolumns $table >> $csvoutput$database.schema
   done
   ;;
  *)
for table in $tables
do
  color_msg $blue "exporting table $table"
  cols=$(grep "$table:" $csvoutput$database.schema | cut -f2 -d:)
  if [  "$cols" = "" ]
  then
    cols=$(getcolumns $table)
  fi
  ssh $host rm $mysqlfiles/$table.csv
cat <<EOF | mysql --host $host -u$user --password=$password $database
SELECT $cols FROM $table INTO OUTFILE '$mysqlfiles$table.csv'
FIELDS TERMINATED BY ','
ENCLOSED BY '"'
LINES TERMINATED BY '\n';
EOF
  scp $host:$mysqlfiles/$table.csv $csvoutput$table.csv.raw
  (echo "$cols"; cat $csvoutput$table.csv.raw) > $csvoutput$table.csv
  rm $csvoutput$table.csv.raw
done
  ;;
esac

Answer 31

Standing on the shoulders of Chris Johnson, I extended the answer from Feb 2016 with a custom dialect for reading.

This shell pipeline tool does not need to connect to your database, handles random commas and quotes in the input, and works nicely in Python 2 and Python 3!

#!/usr/bin/env python
import csv
import sys

# fields are separated by tabs; double-quotes may occur anywhere
csv.register_dialect("mysql", delimiter="\t", quoting=csv.QUOTE_NONE)
tab_in = csv.reader(sys.stdin, dialect="mysql")
comma_out = csv.writer(sys.stdout, dialect=csv.excel)
for row in tab_in:
    # print("row: {}".format(row))
    comma_out.writerow(row)

Use that print statement to convince yourself it's parsing your input correctly :)

A major caveat: treatment of carriage return characters, ^M aka control-M, \r in Linux terms. Although batch-mode MySQL output correctly escapes embedded newline characters, so there is truly one row per line (defined by the Linux newline character \n), MySQL puts no quotes around column data. If a data item has an embedded carriage-return character, csv.reader rejects that input with this exception:

new-line character seen in unquoted field -
do you need to open the file in universal-newline mode?

Please don't @ me saying I should use universal file mode by re-opening sys.stdin.fileno with mode 'rU'. I tried that, and it causes the embedded \r characters to be treated as end-of-record markers, so a single input record is incorrectly transformed into many incomplete output records.

I have not found a Python solution to this limitation of Python's csv.reader module. I think the root cause is the csv.reader implementation/limitation noted in their documentation, csv.reader:

The reader is hard-coded to recognise either '\r' or '\n' as end-of-line,
and ignores lineterminator.

The weak and unsatisfying solution I can offer is to change each \r character to the two-character sequence '\n' before Python's csv.reader sees the data. I used the sed command. Here's an example of a pipeline with a MySQL select and the Python script from above:

mysql -u user db --execute="select * from table where id=12345" \
  | sed -e 's/\r/\\n/g' \
  | mysqlTsvToCsv.py

After fighting this for some time I think Python is not the right solution. If you can live with Perl, I think the one-liner script offered by artfulrobot may be the most-effective and simplest solution.

Answer 32

This one avoids having to write output to a file, only requires expat to be installed, properly escapes values, and outputs empty string (instead of a literal NULL) for null values.

You tell MySQL to output the results in XML format (using the --xml flag), and then pipe the results through the C program below.

This should also be pretty close to the fastest possible way to do this.

// mysql-xml-to-csv.c

#include <assert.h>
#include <ctype.h>
#include <err.h>
#include <expat.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>

/*
    Example of MySQL XML output:

    <?xml version="1.0"?>

    <resultset xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" statement="SELECT id as IdNum, lastName, firstName FROM User">
        <row>
            <field name="IdNum">100040</field>
            <field name="lastName" xsi:nil="true"/>
            <field name="firsttName">Cher</field>
        </row>
    </resultset>
*/

#define BUFFER_SIZE     (1 << 16)

// These accumulate the first row column names and values until first row is entirely read (unless the "-N" flag is given)
static XML_Char **column_names;
static size_t num_column_names;
static XML_Char **first_row_values;
static size_t num_first_row_values;

// This accumulates one column's value
static XML_Char *elem_text;                     // note: not nul-terminated
static size_t elem_text_len;

// Flags
static int first_column;
static int reading_value;

// Expat callback functions
static void handle_elem_start(void *data, const XML_Char *el, const XML_Char **attr);
static void handle_elem_text(void *userData, const XML_Char *s, int len);
static void handle_elem_end(void *data, const XML_Char *el);

// Helper functions
static void output_csv_row(XML_Char **values, size_t num);
static void output_csv_text(const char *s, size_t len);
static void add_string(XML_Char ***arrayp, size_t *lengthp, const XML_Char *string, size_t len);
static void add_chars(XML_Char **strp, size_t *lenp, const XML_Char *string, size_t nchars);
static size_t xml_strlen(const XML_Char *string);
static void free_strings(XML_Char ***arrayp, size_t *lengthp);
static void usage(void);

int
main(int argc, char **argv)
{
    char buf[BUFFER_SIZE];
    int want_column_names = 1;
    XML_Parser p;
    FILE *fp;
    size_t r;
    int i;

    // Parse command line
    while ((i = getopt(argc, argv, "hN")) != -1) {
        switch (i) {
        case 'N':
            want_column_names = 0;
            break;
        case 'h':
            usage();
            exit(0);
        case '?':
        default:
            usage();
            exit(1);
        }
    }
    argv += optind;
    argc -= optind;
    switch (argc) {
    case 0:
        fp = stdin;
        break;
    case 1:
        if ((fp = fopen(argv[0], "r")) == NULL)
            err(1, "%s", argv[0]);
        break;
    default:
        usage();
        exit(1);
    }

    // Initialize arrays for column names and first row values
    if (want_column_names) {
        if ((column_names = malloc(10 * sizeof(*column_names))) == NULL)
            err(1, "malloc");
        if ((first_row_values = malloc(10 * sizeof(*first_row_values))) == NULL)
            err(1, "malloc");
    }

    // Initialize parser
    if ((p = XML_ParserCreate(NULL)) == NULL)
        errx(1, "can't initialize parser");
    XML_SetElementHandler(p, handle_elem_start, handle_elem_end);
    XML_SetCharacterDataHandler(p, handle_elem_text);

    // Process file
    while (1) {
        if ((r = fread(buf, 1, sizeof(buf), fp)) == 0 && ferror(fp))
            errx(1, "error reading input");
        if (XML_Parse(p, buf, r, r == 0) == XML_STATUS_ERROR)
            errx(1, "line %u: %s", (unsigned int)XML_GetCurrentLineNumber(p), XML_ErrorString(XML_GetErrorCode(p)));
        if (r == 0)
            break;
    }

    // Clean up
    XML_ParserFree(p);
    fclose(fp);

    // Done
    return 0;
}

static void
handle_elem_start(void *data, const XML_Char *name, const XML_Char **attr)
{
    if (strcmp(name, "row") == 0)
        first_column = 1;
    else if (strcmp(name, "field") == 0) {
        if (column_names != NULL) {
            while (*attr != NULL && strcmp(*attr, "name") != 0)
                attr += 2;
            if (*attr == NULL)
                errx(1, "\"field\" element is missing \"name\" attribute");
            add_string(&column_names, &num_column_names, attr[1], xml_strlen(attr[1]));
        } else {
            if (!first_column)
                putchar(',');
            putchar('"');
        }
        reading_value = 1;
    }
}

static void
handle_elem_text(void *userData, const XML_Char *s, int len)
{
    if (!reading_value)
        return;
    if (column_names != NULL)
        add_chars(&elem_text, &elem_text_len, s, len);
    else
        output_csv_text(s, len);
}

static void
handle_elem_end(void *data, const XML_Char *name)
{
    if (strcmp(name, "row") == 0) {
        if (column_names != NULL) {
            output_csv_row(column_names, num_column_names);
            output_csv_row(first_row_values, num_first_row_values);
            free_strings(&column_names, &num_column_names);
            free_strings(&first_row_values, &num_first_row_values);
        } else
            putchar('\n');
    } else if (strcmp(name, "field") == 0) {
        if (column_names != NULL) {
            add_string(&first_row_values, &num_first_row_values, elem_text, elem_text_len);
            free(elem_text);
            elem_text = NULL;
            elem_text_len = 0;
        } else
            putchar('"');
        first_column = 0;
        reading_value = 0;
    }
}

static void
output_csv_row(XML_Char **values, size_t num_columns)
{
    int i;

    for (i = 0; i < num_columns; i++) {
        if (i > 0)
            putchar(',');
        putchar('"');
        output_csv_text(values[i], xml_strlen(values[i]));
        putchar('"');
    }
    putchar('\n');
}

static void
output_csv_text(const XML_Char *s, size_t len)
{
    while (len-- > 0) {
        if (*s == '"')
            putchar('"');
        putchar(*s);
        s++;
    }
}

static void
add_string(XML_Char ***arrayp, size_t *lengthp, const XML_Char *string, size_t nchars)
{
    char **new_array;

    if ((new_array = realloc(*arrayp, (*lengthp + 1) * sizeof(**arrayp))) == NULL)
        err(1, "malloc");
    *arrayp = new_array;
    if (((*arrayp)[*lengthp] = malloc((nchars + 1) * sizeof(XML_Char))) == NULL)
        err(1, "malloc");
    memcpy((*arrayp)[*lengthp], string, nchars * sizeof(XML_Char));
    (*arrayp)[*lengthp][nchars] = (XML_Char)0;
    (*lengthp)++;
}

static void
add_chars(XML_Char **strp, size_t *lenp, const XML_Char *string, size_t nchars)
{
    XML_Char *new_array;

    if ((new_array = realloc(*strp, (*lenp + nchars) * sizeof(XML_Char))) == NULL)
        err(1, "malloc");
    *strp = new_array;
    memcpy(*strp + *lenp, string, nchars * sizeof(XML_Char));
    *lenp += nchars;
}

static size_t
xml_strlen(const XML_Char *string)
{
    size_t len;

    len = 0;
    while (string[len] != (XML_Char)0)
        len++;
    return len;
}

static void
free_strings(char ***arrayp, size_t *lengthp)
{
    while (*lengthp > 0)
        free((*arrayp)[--*lengthp]);
    free(*arrayp);
    *arrayp = NULL;
}

static void
usage(void)
{
    fprintf(stderr, "Usage: mysql-xml-to-csv [options] [file.xml]\n");
    fprintf(stderr, "Options:\n");
    fprintf(stderr, "  -N\tDo not output column names as the first row\n");
    fprintf(stderr, "  -h\tShow this usage info\n");
}

For those who don't work with C very often, you can build this code by running the following (assuming you have the expat library installed):

gcc mysql-xml-to-csv.c -lexpat -o mysql-xml-to-csv

Tested with openSUSE 15.2 and gcc 7.5.0.

Update: Now available as an open source project on github.

Answer 33

This is dirty and ugly. It is only suitable for the particular situation where all you have is a PHP-*-admin and the server is running with the --secure-file-priv option, so then you cannot use the INTO OUTFILE '/path/to/export.csv' clause in your query.

What you can do is parse lines of CSV with ... wait for it!, CONCAT, then copy the results and paste them into a file.

Here's an example, where I needed SQL format (it'd be trivial to adapt it for CSV):

SELECT CONCAT(
"('",
`username`, "', '",
`password`, "', '",
`first_name`, "', '",
`last_name`, "', '",
`gender`, "'),"
) AS `row` 
FROM `users`
WHERE `role` != 'Not set'
AND `user_status` = 'Active'
ORDER BY `role`, `gender`, `user_id`
LIMIT 200

That gives nice, ready for import output similar to this:

('jane', '3d7ff...', 'Jane', 'Doe', 'Female'),  
('john', 'd2a33...', 'John', 'Doe', 'Male'),
...

Answer 34

Try this code:

SELECT 'Column1', 'Column2', 'Column3', 'Column4', 'Column5'
UNION ALL
SELECT column1, column2,
column3 , column4, column5 FROM demo
INTO OUTFILE '/tmp/demo.csv'
FIELDS TERMINATED BY ','
ENCLOSED BY '"'
LINES TERMINATED BY '\n';

For more information: http://dev.mysql.com/doc/refman/5.1/en/select-into.html

Answer 35

This solution places the SQL query in a heredoc and pipes the output though a filter:

File query.sh

#!/bin/bash

mysql --defaults-group-suffix=[DATABASE_NAME] --batch << EOF | python query.py
SELECT [FIELDS]
FROM [TABLE]
EOF

This version of the Python filter works without requiring the use of the csv module:

File query.sh

import sys

for line in sys.stdin:
    print(','.join(["\"" + str(element) + "\"" for element in line.rstrip('\n').split('\t')]))

This version of the Python filter uses the CSV module and involves slightly more code but is arguably a little bit more clear:

File query.sh

import csv, sys

csv_reader = csv.reader(sys.stdin, delimiter='\t')
csv_writer = csv.writer(sys.stdout, quoting=csv.QUOTE_NONNUMERIC)

for line in csv_reader:
    csv_writer.writerow(line)

Or you could use Pandas:

File query.py

import csv, sys
import pandas as pd

df = pd.read_csv(sys.stdin, sep='\t')
df.to_csv(sys.stdout, index=False, quoting=csv.QUOTE_NONNUMERIC)

Answer 36

For those, who may want to download a query result in CSV format, but doesn't have access the server file but the database.

First of all, it's not a Linux command. Steps are bellow:

1. Create a view with the query. For example: (Create VIEW v as (Select * from user where status = 0))
2. The view will be created under the view section of your database.
3. Now export the view as CSV.
4. If you need the table column as header of CSV, set Export method: to Custom - display all possible options and check Put columns names in the first row.

Answer 37

If you are getting this error while you try to export your file

ERROR 1290 (HY000): The MySQL server is running with the --secure-file-priv option so it cannot execute this statement

and you are not able to solve this error, you can do one thing by simply running this Python script

import mysql.connector
import csv

con = mysql.connector.connect(
    host="localhost",
    user="root",
    passwd="Your Password"
)

cur = con.cursor()

cur.execute("USE DbName")
cur.execute("""
select col1,col2 from table
where <cond>
""")

with open('Filename.csv',mode='w') as data:
    fieldnames=["Field1","Field2"]
    writer=csv.DictWriter(data,fieldnames=fieldnames)
    writer.writeheader()
    for i in cur:
        writer.writerow({'Field1':i[0],'Field2':i[1]})

Answer 38

If there is PHP installed on the machine you are using, you can write a PHP script to do that. It requires the PHP installation has the MySQL extension installed.

You can call the PHP interpreter from the command line like so:

php --php-ini path/to/php.ini your-script.php

I am including the --php-ini switch, because you may need to use your own PHP configuration that enables the MySQL extension. On PHP 5.3.0+ that extension is enabled by default, so that is no longer necessary to use the configuration to enable it.

Then you can write your export script like any normal PHP script:

<?php
    #mysql_connect("localhost", "username", "password") or die(mysql_error());
    mysql_select_db("mydb") or die(mysql_error());

    $result = mysql_query("SELECT * FROM table_with_the_data p WHERE p.type = $typeiwant");

    $result || die(mysql_error());

    while($row = mysql_fetch_row($result)) {
      $comma = false;
      foreach ($row as $item) {

        # Make it comma separated
        if ($comma) {
          echo ',';
        } else {
          $comma = true;
        }

        # Quote the quotes
        $quoted = str_replace("\"", "\"\"", $item);

        # Quote the string
        echo "\"$quoted\"";
      }
        echo "\n";
    }
?>

The advantage of this method is, that it has no problems with varchar and text fields, that have text containing newlines. Those fields are correctly quoted and those newlines in them will be interpreted by the CSV reader as a part of the text, not record separators. That is something that is hard to correct afterwards with sed or so.

Answer 39

If you are on production or any other server with no access to a file system, you can use this simple trick and a little bit of manual effort to get what you want.

Step 1. Just wrap all the columns under CONCAT and use as CSVFormat option provided by MySQL to get comma-delimited results (or use any delimiter you want). Here is an example:

SELECT
    CONCAT(u.id,
            ',',
            given,
            ',',
            family,
            ',',
            email,
            ',',
            phone,
            ',',
            ua.street_number,
            ',',
            ua.route,
            ',',
            ua.locality,
            ',',
            ua.state,
            ',',
            ua.country,
            ',',
            ua.latitude,
            ',',
            ua.longitude) AS CSVFormat
FROM
    table1 u
        LEFT JOIN
    table2 ua ON u.address_id = ua.id
WHERE
    role_policy = 31 and is_active = 1;

Step 2. Copy results from your terminal to a file and clean up all the pipe characters (that forms the layout of your results) using any text editor.

Step 3. Save as .csv file and that's it.

Answer 40

You can use the below command from your SQL editor/Terminal:

mysql -h(hostname/IP>) -u(username) -p(password) databasename <(query.sql) > outputFILE(.txt/.xls)

E.g.,

• hostname -x.x.x.x

• uname - username

• password - password

• DBName - employeeDB

• queryFile - employee.sql

• outputFile - outputFile.xls

mysql -hx.x.x.x -uusername -ppassword employeeDB< employee.sql> outputFile.xls

Make sure you are executing the command from the directory where SQL query is located or mention the full path of the SQL query location in the above command.