The code to create a new server looks like:
const server = ldap.createServer();
The full list of options is:
||log||You can optionally pass in a Bunyan compatible logger instance the client will use to acquire a child logger.|| ||certificate||A PEM-encoded X.509 certificate; will cause this server to run in TLS mode.|| ||key||A PEM-encoded private key that corresponds to certificate for SSL.||
The passed in logger is expected to conform to the Log4j standard API. Internally, abstract-logging is used to implement the interface. As a result, no log messages will be generated unless an external logger is supplied.
Known compatible loggers are:
Set this property to reject connections when the server's connection count gets high.
The number of concurrent connections on the server. This property is deprecated, please use server.getConnections() instead.
Returns the fully qualified URL this server is listening on. For example:
ldaps://10.1.2.3:1636
. If you haven't yet called listen
, it will always
return ldap://localhost:389
.
function() {}
Emitted when the server closes.
The LDAP server API wraps up and mirrors the node.js server.listen
family of
APIs.
After calling listen
, the property url
on the server object itself will be
available.
Example:
server.listen(389, '127.0.0.1', function() {
console.log('LDAP server listening at: ' + server.url);
});
listen(port, [host], [callback])
Begin accepting connections on the specified port and host. If the host is
omitted, the server will accept connections directed to the IPv4 address
127.0.0.1
. To listen on any other address, supply said address as the host
parameter. For example, to listen on all available IPv6 addresses supply
::
as the host
(note, this may also result in listening on all
available IPv4 addresses, depending on operating system behavior).
We highly recommend being as explicit as possible with the host
parameter.
Listening on all available addresses (through ::
or 0.0.0.0
) can lead
to potential security issues.
This function is asynchronous. The last parameter callback will be called when the server has been bound.
listen(path, [callback])
Start a UNIX socket server listening for connections on the given path.
This function is asynchronous. The last parameter callback will be called when the server has been bound.
listenFD(fd)
Start a server listening for connections on the given file descriptor.
This file descriptor must have already had the bind(2)
and listen(2)
system
calls invoked on it. Additionally, it must be set non-blocking; try
fcntl(fd, F_SETFL, O_NONBLOCK)
.
The LDAP server API mirrors the Node.js server.getConnections
API. Callback
should take two arguments err and count.
The LDAP server API is meant to be the LDAP-equivalent of the express/restify
paradigm of programming. Essentially every method is of the form
OP(req, res, next)
where OP is one of bind, add, del, etc. You can chain
handlers together by calling next()
and ordering your functions in the
definition of the route. For example:
function authorize(req, res, next) {
if (!req.connection.ldap.bindDN.equals('cn=root'))
return next(new ldap.InsufficientAccessRightsError());
return next();
}
server.search('o=example', authorize, function(req, res, next) { ... });
Note that ldapjs is also slightly different, since it's often going to be backed to a DB-like entity, in that it also has an API where you can pass in a 'backend' object. This is necessary if there are persistent connection pools, caching, etc. that need to be placed in an object.
For example ldapjs-riak is a complete implementation of the LDAP protocol over Riak. Getting an LDAP server up with riak looks like:
const ldap = require('ldapjs');
const ldapRiak = require('ldapjs-riak');
const server = ldap.createServer();
const backend = ldapRiak.createBackend({
"host": "localhost",
"port": 8098,
"bucket": "example",
"indexes": ["l", "cn"],
"uniqueIndexes": ["uid"],
"numConnections": 5
});
server.add("o=example",
backend,
backend.add());
...
The first parameter to an ldapjs route is always the point in the
tree to mount the handler chain at. The second argument is optionally a
backend object. After that you can pass in an arbitrary combination of
functions in the form f(req, res, next)
or arrays of functions of the same
signature (ldapjs will unroll them).
Unlike HTTP, LDAP operations do not have a heterogeneous wire format, so each
operation requires specific methods/fields on the request/response
objects. However, there is a .use()
method availabe, similar to
that on express/connect, allowing you to chain up "middleware":
server.use(function(req, res, next) {
console.log('hello world');
return next();
});
All request objects have the dn
getter on it, which is "context-sensitive"
and returns the point in the tree that the operation wants to operate on. The
LDAP protocol itself sadly doesn't define operations this way, and has a unique
name for just about every op. So, ldapjs calls it dn
. The DN object itself
is documented at DN.
All requests have an optional array of Control
objects. Control
will have
the properties type
(string), criticality
(boolean), and optionally, a
string value
.
All request objects will have a connection
object, which is the net.Socket
associated to this request. Off the connection
object is an ldap
object.
The most important property to pay attention to is the bindDN
property
which will be an instance of an ldap.DN
object. This is what the client
authenticated as on this connection. If the client didn't bind, then a DN object
will be there defaulted to cn=anonymous
.
Additionally, request will have a logId
parameter you can use to uniquely
identify the request/connection pair in logs (includes the LDAP messageId).
All response objects will have an end
method on them. By default, calling
res.end()
with no arguments will return SUCCESS (0x00) to the client
(with the exception of compare
which will return COMPARE_TRUE (0x06)). You
can pass in a status code to the end()
method to return an alternate status
code.
However, it's more common/easier to use the return next(new LDAPError())
pattern, since ldapjs will fill in the extra LDAPResult fields like matchedDN
and error message for you.
ldapjs includes an exception hierarchy that directly corresponds to the RFC list
of error codes. The complete list is documented in errors. But
the paradigm is something defined like CONSTRAINT_VIOLATION in the RFC would be
ConstraintViolationError
in ldapjs. Upon calling next(new LDAPError())
,
ldapjs will stop calling your handler chain. For example:
server.search('o=example',
(req, res, next) => { return next(); },
(req, res, next) => { return next(new ldap.OperationsError()); },
(req, res, next) => { res.end(); }
);
In the code snipped above, the third handler would never get invoked.
Adds a mount in the tree to perform LDAP binds with. Example:
server.bind('ou=people, o=example', (req, res, next) => {
console.log('bind DN: ' + req.dn.toString());
console.log('bind PW: ' + req.credentials);
res.end();
});
BindRequest objects have the following properties:
The LDAP protocol version the client is requesting to run this connection on. Note that ldapjs only supports LDAP version 3.
The DN the client is attempting to bind as (note this is the same as the dn
property).
The method of authentication. Right now only simple
is supported.
The credentials to go with the name/authentication
pair. For simple
, this
will be the plain-text password.
No extra methods above an LDAPResult
API call.
Adds a mount in the tree to perform LDAP adds with.
server.add('ou=people, o=example', (req, res, next) => {
console.log('DN: ' + req.dn.toString());
console.log('Entry attributes: ' + req.toObject().attributes);
res.end();
});
AddRequest objects have the following properties:
The DN the client is attempting to add (this is the same as the dn
property).
The set of attributes in this entry. This will be an array of
Attribute
objects (which have a type and an array of values). This directly
maps to how the request came in off the wire. It's likely you'll want to use
toObject()
and iterate that way, since that will transform an AddRequest into
a standard JavaScript object.
This operation will return a plain JavaScript object from the request that looks like:
{
dn: 'cn=foo, o=example', // string, not DN object
attributes: {
cn: ['foo'],
sn: ['bar'],
objectclass: ['person', 'top']
}
}
No extra methods above an LDAPResult
API call.
Adds a handler for the LDAP search operation.
server.search('o=example', (req, res, next) => {
console.log('base object: ' + req.dn.toString());
console.log('scope: ' + req.scope);
console.log('filter: ' + req.filter.toString());
res.end();
});
SearchRequest objects have the following properties:
The DN the client is attempting to start the search at (equivalent to dn
).
(string) one of:
An integer (defined in the LDAP protocol). Defaults to '0' (meaning never deref).
The number of entries to return. Defaults to '0' (unlimited). ldapjs doesn't currently automatically enforce this, but probably will at some point.
Maximum amount of time the server should take in sending search entries. Defaults to '0' (unlimited).
Whether to return only the names of attributes, and not the values. Defaults to 'false'. ldapjs will take care of this for you.
The filter object that the client requested. Notably this has
a matches()
method on it that you can leverage. For an example of
introspecting a filter, take a look at the ldapjs-riak source.
An optional list of attributes to restrict the returned result sets to. ldapjs will automatically handle this for you.
Allows you to send a SearchEntry
object. You do not need to
explicitly pass in a SearchEntry
object, and can instead just send a plain
JavaScript object that matches the format used from AddRequest.toObject()
.
server.search('o=example', (req, res, next) => {
const obj = {
dn: 'o=example',
attributes: {
objectclass: ['top', 'organization'],
o: ['example']
}
};
if (req.filter.matches(obj))
res.send(obj)
res.end();
});
Allows you to handle an LDAP modify operation.
server.modify('o=example', (req, res, next) => {
console.log('DN: ' + req.dn.toString());
console.log('changes:');
for (const c of req.changes) {
console.log(' operation: ' + c.operation);
console.log(' modification: ' + c.modification.toString());
}
res.end();
});
ModifyRequest objects have the following properties:
The DN the client is attempting to update (this is the same as the dn
property).
An array of Change
objects the client is attempting to perform. See below for
details on the Change
object.
The Change
object will have the following properties:
A string, and will be one of: 'add', 'delete', or 'replace'.
Will be an Attribute
object, which will have a 'type' (string) field, and
'vals', which will be an array of string values.
No extra methods above an LDAPResult
API call.
Allows you to handle an LDAP delete operation.
server.del('o=example', (req, res, next) => {
console.log('DN: ' + req.dn.toString());
res.end();
});
The DN the client is attempting to delete (this is the same as the dn
property).
No extra methods above an LDAPResult
API call.
Allows you to handle an LDAP compare operation.
server.compare('o=example', (req, res, next) => {
console.log('DN: ' + req.dn.toString());
console.log('attribute name: ' + req.attribute);
console.log('attribute value: ' + req.value);
res.end(req.value === 'foo');
});
The DN the client is attempting to compare (this is the same as the dn
property).
The string name of the attribute to compare values of.
The string value of the attribute to compare.
The end()
method for compare takes a boolean, as opposed to a numeric code
(you can still pass in a numeric LDAP status code if you want). Beyond
that, there are no extra methods above an LDAPResult
API call.
Allows you to handle an LDAP modifyDN operation.
server.modifyDN('o=example', (req, res, next) => {
console.log('DN: ' + req.dn.toString());
console.log('new RDN: ' + req.newRdn.toString());
console.log('deleteOldRDN: ' + req.deleteOldRdn);
console.log('new superior: ' +
(req.newSuperior ? req.newSuperior.toString() : ''));
res.end();
});
The DN the client is attempting to rename (this is the same as the dn
property).
The leaf RDN the client wants to rename this entry to. This will be a DN object.
Whether or not to delete the old RDN (i.e., rename vs copy). Defaults to 'true'.
Optional (DN). If the modifyDN operation wishes to relocate the entry in the
tree, the newSuperior
field will contain the new parent.
No extra methods above an LDAPResult
API call.
Allows you to handle an LDAP extended operation. Extended operations are pretty much arbitrary extensions, by definition. Typically the extended 'name' is an OID, but ldapjs makes no such restrictions; it just needs to be a string. Unlike the other operations, extended operations don't map to any location in the tree, so routing here will be exact match, as opposed to subtree.
// LDAP whoami
server.exop('1.3.6.1.4.1.4203.1.11.3', (req, res, next) => {
console.log('name: ' + req.name);
console.log('value: ' + req.value);
res.value = 'u:xxyyz@EXAMPLE.NET';
res.end();
return next();
});
Will always be a match to the route-defined name. Clients must include this in their requests.
Optional string. The arbitrary blob the client sends for this extended operation.
The name of the extended operation. ldapjs will automatically set this.
The arbitrary (string) value to send back as part of the response.
ldapjs by default provides an unbind handler that just disconnects the client and cleans up any internals (in ldapjs core). You can override this handler if you need to clean up any items in your backend, or perform any other cleanup tasks you need to.
server.unbind((req, res, next) => {
res.end();
});
Note that the LDAP unbind operation actually doesn't send any response (by
definition in the RFC), so the UnbindResponse is really just a stub that
ultimately calls net.Socket.end()
for you. There are no properties available
on either the request or response objects, except, of course, for end()
on the
response.