npm install [email protected]
Despite the alpha tag, this is the recommended version for new applications. For information about the previous 0.9.x releases, visit the v0.9 branch.
Sometimes I may also ask you to install the latest version from Github to check if a bugfix is working. In this case, please do:
npm install git://github.com/felixge/node-mysql.git
This is a node.js driver for mysql. It is written in JavaScript, does not require compiling, and is 100% MIT licensed.
Here is an example on how to use it:
var mysql = require('mysql');
var connection = mysql.createConnection({
host : 'localhost',
user : 'me',
password : 'secret',
});
connection.connect();
connection.query('SELECT 1', function(err, rows, fields) {
if (err) throw err;
console.log('Query result: ', rows);
});
connection.end();
From this example, you can learn the following:
- Every method you invoke on a connection is queued and executed in sequence.
- Closing the connection is done using
end()
which makes sure all remaining queries are executed before sending a quit packet to the mysql server.
Thanks goes to the people who have contributed code to this module, see the GitHub Contributors page.
Additionally I'd like to thank the following people:
- Andrey Hristov (Oracle) - for helping me with protocol questions.
- Ulf Wendel (Oracle) - for helping me with protocol questions.
The following companies have supported this project financially, allowing me to spend more time on it (ordered by time of contribution):
- Transloadit (my startup, we do file uploading & video encoding as a service, check it out)
- Joyent
- pinkbike.com
- Holiday Extras (they are hiring)
- Newscope (they are hiring)
If you are interested in sponsoring a day or more of my time, please get in touch.
If you'd like to discuss this module, or ask questions about it, please use one of the following:
- Mailing list: https://groups.google.com/forum/#!forum/node-mysql
- IRC Channel: #node.js (on freenode.net, I pay attention to any message
including the term
mysql
)
The recommended way to establish a connection is this:
var mysql = require('mysql');
var connection = mysql.createConnection({
host : 'example.org',
user : 'bob',
password : 'secret',
});
connection.connect(function(err) {
// connected! (unless `err` is set)
});
However, a connection can also be implicitly established by invoking a query:
var mysql = require('mysql');
var connection = mysql.createConnection(...);
connection.query('SELECT 1', function(err, rows) {
// connected! (unless `err` is set)
});
Depending on how you like to handle your errors, either method may be appropriate. Any type of connection error (handshake or network) is considered a fatal error, see the Error Handling section for more information.
When establishing a connection, you can set the following options:
host
: The hostname of the database you are connecting to. (Default:localhost
)port
: The port number to connect to. (Default:3306
)socketPath
: The path to a unix domain socket to connect to. When usedhost
andport
are ignored.user
: The MySQL user to authenticate as.password
: The passqword of that MySQL user.database
: Name of the database to use for this connection (Optional).charset
: The charset for the connection. (Default:'UTF8_GENERAL_CI'
)insecureAuth
: Allow connecting to MySQL instances that ask for the old (insecure) authentication method. (Default:false
)typeCast
: Determines if column values should be converted to native JavaScript types. (Default:true
)debug
: Prints protocol details to stdout. (Default:false
)multipleStatements
: Allow multiple mysql statements per query. Be careful with this, it exposes you to SQL injection attacks. (Default: `false)
In addition to passing these options as an object, you can also use a url string. For example:
var connection = mysql.createConnection('mysql://user:pass@host/db?debug=true&charset=BIG5_CHINESE_CI');
Note: The query values are first attempted to be parsed as JSON, and if that fails assumed to be plaintext strings.
There are two ways to end a connection. Terminating a connection gracefully is
done by calling the end()
method:
connection.end(function(err) {
// The connection is terminated now
});
This will make sure all previously enqueued queries are still before sending a
COM_QUIT
packet to the MySQL server. If a fatal error occurs before the
COM_QUIT
packet can be sent, an err
argument will be provided to the
callback, but the connection will be terminated regardless of that.
An alternative way to end the connection is to call the destroy()
method.
This will cause an immediate termination of the underlaying socket.
Additionally destroy()
guarantees that no more events or callbacks will be
triggered for the connection.
connection.destroy();
Unlike end()
the destroy()
method does not take a callback argument.
MySQL offers a changeUser command that allows you to alter the current user and other aspects of the connection without shutting down the underlaying socket:
connection.changeUser({user : 'john'}, function(err) {
if (err) throw err;
});
The available options for this feature are:
user
: The name of the new user (defaults to the previous one).password
: The password of the new user (defaults to the previous one).charset
: The new charset (defaults to the previous one).database
: The new database (defaults to the previous one).
A sometimes useful side effect of this functionality is that this function also resets any connection state (variables, transactions, etc.).
Errors encountered during this operation are treated as fatal connection errors by this module.
You may loose the connection to a MySQL server due to network problems, the
server timing you out, or the server crashing. All of these events are
considered fatal errors, and will have the err.code = 'PROTOCOL_CONNECTION_LOST'
. See the Error Handling section
for more information.
The best way to handle such unexpected disconnects is shown below:
function handleDisconnect(connection) {
connection.on('error', function(err) {
if (!err.fatal) {
return;
}
if (err.code !== 'PROTOCOL_CONNECTION_LOST') {
throw err;
}
console.log('Re-connecting lost connection: ' + err.stack);
connection = mysql.createConnection(connection.config);
handleDisconnect(connection);
connection.connect();
});
}
handleDisconnect(connection);
As you can see in the example above, re-connecting a connection is done by establishing a new connection. Once terminated, an existing connection object cannot be re-connected by design.
This logic will also be part of connection pool support once I add that to this library.
In order to avoid SQL Injection attacks, you should always escape any user
provided data before using it inside a SQL query. You can do so using the
connection.escape()
method:
var userId = 'some user provided value';
var sql = 'SELECT * FROM users WHERE id = ' + connection.escape(userId);
connection.query(sql, function(err, results) {
// ...
});
Alternatively, you can use ?
characters as placeholders for values you would
like to have escaped like this:
connection.query('SELECT * FROM users WHERE id = ?', [userId], function(err, results) {
// ...
});
This looks similar to prepared statements in MySQL, however it really just uses
the same connection.escape()
method internally.
Different value types are escaped differently, here is how:
- Numbers are left untouched
- Booleans are converted to
true
/false
strings - Date objects are converted to
'YYYY-mm-dd HH:ii:ss'
strings - Buffers are converted to hex strings, e.g.
X'0fa5'
- Strings are safely escaped
- Arrays are turned into list, e.g. ['a', 'b'] turns into
'a', 'b'
- Objects are turned into
key = 'val'
pairs. Nested objects are cast to strings. undefined
/null
are converted toNULL
NaN
/Infinity
are left as-is. MySQL does not support these, and trying to insert them as values will trigger MySQL errors until they implement support.
If you paid attention, you may have noticed that this escaping allows you to do neat things like this:
var post = {id: 1, title: 'Hello MySQL'};
var query = connection.query('INSERT INTO posts SET ?', post, function(err, result) {
// Neat!
});
console.log(query.sql); // INSERT INTO posts SET `id` = 1, `title` = 'Hello MySQL'
If you are inserting a row into a table with an auto increment primary key, you can retrieve the insert id like this:
connection.query('INSERT INTO posts SET ?', {title: 'test'}, function(err, result) {
if (err) throw err;
console.log(result.insertId);
});
The MySQL protocol is sequential, this means that you need multiple connections to execute queries in parallel. Future version of this module may ship with a connection pool implementation, but for now you have to figure out how to manage multiple connections yourself if you want to execute queries in parallel.
One simple approach is to create one connection per incoming http request.
Sometimes you may want to select large quantities of rows and process each of them as they are received. This can be done like this:
var query = connection.query('SELECT * FROM posts');
query
.on('error', function(err) {
// Handle error, an 'end' event will be emitted after this as well
})
.on('fields', function(fields) {
// the field packets for the rows to follow
})
.on('result', function(row) {
// Pausing the connnection is useful if your processing involves I/O
connection.pause();
processRow(row, function() {
connection.resume();
});
})
.on('end', function() {
// all rows have been received
});
Please note a few things about the example above:
- Usually you will want to receive a certain amount of rows before starting to
throttle the connection using
pause()
. This number will depend on the amount and size of your rows. pause()
/resume()
operate on the underlaying socket and parser. You are guaranteed that no more'result'
events will fire after callingpause()
.- You MUST NOT provide a callback to the
query()
method when streaming rows. - The
'result'
event will fire for both rows as well as OK packets confirming the success of a INSERT/UPDATE query.
Additionally you may be interested to know that it is currently not possible to stream individual row columns, they will always be buffered up entirely. If you have a good use case for streaming large fields to and from MySQL, I'd love to get your thoughts and conributions on this.
Support for multiple statements is disabled for security reasons (it allows for SQL injection attacks if values are not properly escaped). To use this feature you have to enable it for your connection:
var connection = mysql.createConnection({multipleStatements: true});
Once enabled, you can execute multiple statement queries like any other query:
connection.query('SELECT 1; SELECT 2', function(err, results) {
if (err) throw err;
// `results` is an array with one element for every statement in the query:
console.log(results[0]); // [{1: 1}]
console.log(results[1]); // [{2: 2}]
});
Additionally you can also stream the results of multiple statement queries:
var query = connection.query('SELECT 1; SELECT 2');
query
.on('fields', function(fields, index) {
// the fields for the result rows that follow
})
.on('result', function(row, index) {
// index refers to the statement this result belongs to (starts at 0)
});
If one of the statements in your query causes an error, the resulting Error
object contains a err.index
property which tells you which statement caused
it. MySQL will also stop executing any remaining statements when an error
occurs.
Please note that the interface for streaming multiple statement queries is experimental and I am looking forward to feedback on it.
You can call stored procedures from your queries as with any other mysql driver. If the stored procedure produces several result sets, they are exposed to you the same way as the results for multiple statement queries.
When executing joins, you are likely to get result sets with overlapping column names.
By default, node-mysql will overwrite colliding column names in the order the columns are received from MySQL, causing some of the received values to be unavailable.
However, you can also specify that you want your columns to be nested below the table name like this:
var options = {sql: '...', nestTables: true};
connection.query(options, function(err, results) {
/* results will be an array like this now:
[{
table1: {
fieldA: '...',
fieldB: '...',
},
table2: {
fieldA: '...',
fieldB: '...',
},
}, ...]
*/
});
This module comes with a consistent approach to error handling that you should review carefully in order to write solid applications.
All errors created by this module are instances of the JavaScript Error object. Additionally they come with two properties:
err.code
: Either a MySQL server error (e.g.'ER_ACCESS_DENIED_ERROR'
), a node.js error (e.g.'ECONNREFUSED'
) or an internal error (e.g.'PROTOCOL_CONNECTION_LOST'
).err.fatal
: Boolean, indicating if this error is terminal to the connection object.
Fatal errors are propagated to all pending callbacks. In the example below, a fatal error is triggered by trying to connect to an invalid port. Therefore the error object is propagated to both pending callbacks:
var connection = require('mysql').createConnection({
port: 84943, // WRONG PORT
});
connection.connect(function(err) {
console.log(err.code); // 'ECONNREFUSED'
console.log(err.fatal); // true
});
connection.query('SELECT 1', function(err) {
console.log(err.code); // 'ECONNREFUSED'
console.log(err.fatal); // true
});
Normal errors however are only delegated to the callback they belong to. So in the example below, only the first callback receives an error, the second query works as expected:
connection.query('USE name_of_db_that_does_not_exist', function(err, rows) {
console.log(err.code); // 'ER_BAD_DB_ERROR'
});
connection.query('SELECT 1', function(err, rows) {
console.log(err); // null
console.log(rows.length); // 1
});
Last but not least: If a fatal errors occurs and there are no pending
callbacks, or a normal error occurs which has no callback belonging to it, the
error is emitted as an 'error'
event on the connection object. This is
demonstrated in the example below:
connection.on('error', function(err) {
console.log(err.code); // 'ER_BAD_DB_ERROR'
});
connection.query('USE name_of_db_that_does_not_exist');
Note: 'error'
are special in node. If they occur without an attached
listener, a stack trace is printed and your process is killed.
tl;dr: This module does not want you to to deal with silent failures. You should always provide callbacks to your method calls. If you want to ignore this advice and suppress unhandled errors, you can do this:
// I am Chuck Norris:
connection.on('error', function() {});
This module is exception safe. That means you can continue to use it, even if one of your callback functions throws an error which you're catching using 'uncaughtException' or a domain.
For your convenience, this driver will cast mysql types into native JavaScript types by default. The following mappings exist:
- TINYINT
- SMALLINT
- INT
- MEDIUMINT
- YEAR
- FLOAT
- DOUBLE
- TIMESTAMP
- DATE
- DATETIME
- TINYBLOB
- MEDIUMBLOB
- LONGBLOB
- BLOB
- BINARY
- VARBINARY
- BIT (last byte will be filled with 0 bits as neccessary)
- CHAR
- VARCHAR
- TINYTEXT
- MEDIUMTEXT
- LONGTEXT
- TEXT
- ENUM
- SET
- DECIMAL (may exceed float precision)
- BIGINT (may exceed float precision)
- TIME (could be mapped to Date, but what date would be set?)
- GEOMETRY (never used those, get in touch if you do)
It is not recommended (and may go away / change in the future) to disable type casting, but you can currently do so on either the connection:
var connection = require('mysql').createConnection({typeCast: false});
Or on the query level:
var options = {sql: '...', typeCast: false};
var query = connection.query(options, function(err, results) {
}):
If you are running into problems, one thing that may help is enabling the
debug
mode for the connection:
var connection = mysql.createConnection({debug: true});
This will print all incoming and outgoing packets on stdout.
If that does not help, feel free to open a GitHub issue. A good GitHub issue will have:
- The minimal amount of code required to reproduce the problem (if possible)
- As much debugging output and information about your environment (mysql version, node version, os, etc.) as you can gather.
- Prepared statements
- setTimeout() for Connection / Query
- connection pooling
- Support for encodings other than UTF-8 / ASCII
- API support for transactions, similar to php