ldapjs-promise Client API
This document covers the ldapjs-promise client API and assumes that you are familiar with LDAP. If you are not, read the guide first.
Create a client
The code to create a new client looks like:
import { createClient } from 'ldapjs-promise';
const client = createClient({
url: ['ldap://127.0.0.1:1389', 'ldap://127.0.0.2:1389']
});
client.on('connectError', (err) => {
// handle connection error
})
You can use ldap:// or ldaps://; the latter would connect over SSL (note
that this will not use the LDAP TLS extended operation, but literally an SSL
connection to port 636, as in LDAP v2). The full set of options to create a
client is:
| Attribute | Description |
|---|---|
| url | A string or array of valid LDAP URL(s) (proto/host/port) |
| socketPath | Socket path if using AF_UNIX sockets |
| log | A compatible logger instance (Default: no-op logger) |
| timeout | Milliseconds client should let operations live for before timing out (Default: Infinity) |
| connectTimeout | Milliseconds client should wait before timing out on TCP connections (Default: OS default) |
| tlsOptions | Additional options passed to TLS connection layer when connecting via ldaps:// (See: The TLS docs for node.js) |
| idleTimeout | Milliseconds after last activity before client emits idle event |
| reconnect | Try to reconnect when the connection gets lost (Default is false) |
url
This parameter takes a single connection string or an array of connection strings as an input. In case an array is provided, the client tries to connect to the servers in given order. To achieve random server strategy (e.g. to distribute the load among the servers), please shuffle the array before passing it as an argument.
Note On Logger
A passed in logger is expected to conform to the Bunyan
API. Specifically, the logger is expected to have a child() method. If a logger
is supplied that does not have such a method, then a shim version is added
that merely returns the passed in logger.
Known compatible loggers are:
Note On Error Handling
The client is an EventEmitter. If you do not register an error handler and
e.g. a connection error occurs, Node.js will print a stack trace and exit the
process (reference).
Connection management
As LDAP is a stateful protocol (as opposed to HTTP), having connections torn down from underneath you can be difficult to deal with. Several mechanisms have been provided to mitigate this trouble.
Reconnect
You can provide a Boolean option indicating if a reconnect should be tried. For
more sophisticated control, you can provide an Object with the properties
initialDelay (default: 100), maxDelay (default: 10000) and
failAfter (default: Infinity).
After the reconnect you maybe need to bind again.
Client events
The client is an EventEmitter and can emit the following events:
| Event | Description |
|---|---|
| error | General error |
| connectRefused | Server refused connection. Most likely bad authentication |
| connectTimeout | Server timeout |
| connectError | Socket connection error |
| setupError | Setup error after successful connection |
| socketTimeout | Socket timeout |
| resultError | Search result error |
| timeout | Search result timeout |
| destroy | After client is disconnected |
| end | Socket end event |
| close | Socket closed |
| connect | Client connected |
| idle | Idle timeout reached |
Common patterns
The last parameter in every API is controls. controls can be either a
single instance of a Control or an array of Control objects.
You can, and probably will, omit this option.
Almost every operation returns a promise that either resolves a res or throws
an err where err will be an instance of an LDAPError (you can use
instanceof to switch).
You probably will not need to check the res parameter, but it is there if you do.
bind
bind(dn, password, controls)
Performs a bind operation against the LDAP server.
The bind API only allows LDAP 'simple' binds (equivalent to HTTP Basic
Authentication) for now. Note that all client APIs can optionally take an array
of Control objects. You probably do not need them though...
Example:
await client.bind('cn=root', 'secret');
add
add(dn, entry, controls)
Performs an add operation against the LDAP server.
Allows you to add an entry (which is just a plain JS object), and as always, controls are optional.
Example:
const entry = {
cn: 'foo',
sn: 'bar',
email: ['foo@bar.com', 'foo1@bar.com'],
objectclass: 'fooPerson'
};
await client.add('cn=foo, o=example', entry);
compare
compare(dn, attribute, value, controls)
Performs an LDAP compare operation with the given attribute and value against the entry referenced by dn.
Example:
const res = await client.compare('cn=foo, o=example', 'sn', 'bar');
console.log('matched: ' + res.matched);
del
del(dn, controls)
Deletes an entry from the LDAP server.
Example:
await client.del('cn=foo, o=example');
exop
exop(name, value, controls)
Performs an LDAP extended operation against an LDAP server. name is typically
going to be an OID (well, the RFC says it must be; however, ldapjs-promise has
no such restriction). value is completely arbitrary, and is whatever the
exop says it should be.
Example (performs an LDAP 'whois' extended op):
const res = await client.exop('1.3.6.1.4.1.4203.1.11.3');
console.log('whois: ' + res.value);
modify
modify(name, changes, controls)
Performs an LDAP modify operation against the LDAP server. This API requires
you to pass in a Change object, which is described below. Note that you can
pass in a single Change or an array of Change objects.
Example:
const change = new ldap.Change({
operation: 'add',
modification: {
type: 'pets',
values: ['cat', 'dog']
}
});
await client.modify('cn=foo, o=example', change);
Change
A Change object maps to the LDAP protocol of a modify change, and requires
you to set the operation and modification. The operation is a string,
and must be one of:
| Operation | Description |
|---|---|
| replace | Replaces the attribute referenced in modification. If the modification has no values, it is equivalent to a delete. |
| add | Adds the attribute value(s) referenced in modification. The attribute may or may not already exist. |
| delete | Deletes the attribute (and all values) referenced in modification. |
modification is just a plain old JS object with the required type and values you want.
| Operation | Description |
|---|---|
| type | String that defines the attribute type for the modification. |
| values | Defines the values for modification. |
modifyDN
modifyDN(dn, newDN, controls)
Performs an LDAP modifyDN (rename) operation against an entry in the LDAP server. A couple points with this client API:
- There is no ability to set "keep old dn." It is always going to flag the old dn to be purged.
- The client code will automatically figure out if the request is a "new superior" request ("new superior" means move to a different part of the tree, as opposed to just renaming the leaf).
Example:
await client.modifyDN('cn=foo, o=example', 'cn=bar');
search
search(base, options, controls)
Performs a search operation against the LDAP server.
The search operation is more complex than the other operations, so this one
takes an options object for all the parameters. However, ldapjs-promise
makes some defaults for you so that if you pass nothing in, it is pretty much
equivalent to an HTTP GET operation (i.e., base search against the DN, filter
set to always match).
Like every other operation, base is a DN string.
Options can be a string representing a valid LDAP filter or an object containing the following fields:
| Attribute | Description |
|---|---|
| scope | One of base, one, or sub. Defaults to base. |
| filter | A string version of an LDAP filter (see below), or a programatically constructed Filter object. Defaults to (objectclass=*). |
| attributes | attributes to select and return (if these are set, the server will return only these attributes). Defaults to the empty set, which means all attributes. You can provide a string if you want a single attribute or an array of string for one or many. |
| attrsOnly | boolean on whether you want the server to only return the names of the attributes, and not their values. Borderline useless. Defaults to false. |
| sizeLimit | the maximum number of entries to return. Defaults to 0 (unlimited). |
| timeLimit | the maximum amount of time the server should take in responding, in seconds. Defaults to 10. Lots of servers will ignore this. |
| paged | enable and/or configure automatic result paging |
Responses returned by the search method are an EventEmitter where you will
get a notification for each searchEntry that comes back from the server. You
will additionally be able to listen for a searchRequest, searchReference,
error and end event.
searchRequest is emitted immediately after every SearchRequest is sent with
a SearchRequest parameter. You can do operations like client.abandon with
searchRequest.messageId to abandon this search request. Note that the error
event will only be for client/TCP errors, not LDAP error codes like the other
APIs. You will want to check the LDAP status code (likely for 0) on the end
event to assert success. LDAP search results can give you a lot of status
codes, such as time or size exceeded, busy, inappropriate matching, etc., which
is why this method does not try to wrap up the code matching.
Example:
const opts = {
filter: '(&(l=Seattle)(email=*@foo.com))',
scope: 'sub',
attributes: ['dn', 'sn', 'cn']
};
const res = await client.search('o=example', opts);
res.on('searchRequest', (searchRequest) => {
console.log('searchRequest: ', searchRequest.messageId);
});
res.on('searchEntry', (entry) => {
console.log('entry: ' + JSON.stringify(entry.pojo));
});
res.on('searchReference', (referral) => {
console.log('referral: ' + referral.uris.join());
});
res.on('error', (err) => {
console.error('error: ' + err.message);
});
res.on('end', (result) => {
console.log('status: ' + result.status);
});
Filter Strings
The easiest way to write search filters is to write them compliant with RFC2254, which is "The string representation of LDAP search filters." Note that ldapjs-promise does not support extensible matching, since it is one of those features that almost nobody actually uses in practice.
Assuming you do not really want to read the RFC, search filters in LDAP are
basically a "tree" of attribute/value assertions, with the tree specified in
prefix notation. For example, let us start simple, and build up a complicated
filter. The most basic filter is equality, so let us assume you want to search
for an attribute email with a value of foo@bar.com. The syntax would be:
(email=foo@bar.com)
ldapjs-promise requires all filters to be surrounded by '()' blocks. Ok, that was easy. Let us now assume that you want to find all records where the email is actually just anything in the "@bar.com" domain and the location attribute is set to Seattle:
(&(email=*@bar.com)(l=Seattle))
Now our filter is actually three LDAP filters. We have an and filter (single
amp &), an equality filter (the l=Seattle), and a substring filter.
Substrings are wildcard filters. They use * as the wildcard. You can put more
than one wildcard for a given string. For example you could do
(email=*@*bar.com) to match any email of @bar.com or its subdomains like
"example@foo.bar.com".
Now, let us say we also want to set our filter to include a specification that either the employeeType not be a manager nor a secretary:
(&(email=*@bar.com)(l=Seattle)(!(|(employeeType=manager)(employeeType=secretary))))
The not character is represented as a !, the or as a single pipe |.
It gets a little bit complicated, but it is actually quite powerful, and lets
you find almost anything you are looking for.
Paging
Many LDAP server enforce size limits upon the returned result set (commonly
1000). In order to retrieve results beyond this limit, a PagedResultControl
is passed between the client and server to iterate through the entire dataset.
While callers could choose to do this manually via the controls parameter to
search(), ldapjs-promise has internal mechanisms to easily automate the process. The
most simple way to use the paging automation is to set the paged option to
true when performing a search:
const opts = {
filter: '(objectclass=commonobject)',
scope: 'sub',
paged: true,
sizeLimit: 200
};
const res = await client.search('o=largedir', opts);
res.on('searchEntry', (entry) => {
// do per-entry processing
});
res.on('page', (result) => {
console.log('page end');
});
res.on('error', (resErr) => {
assert.ifError(resErr);
});
res.on('end', (result) => {
console.log('done ');
});
This will enable paging with a default page size of 199 (sizeLimit - 1) and
will output all of the resulting objects via the searchEntry event. At the
end of each result during the operation, a page event will be emitted as
well (which includes the intermediate searchResult object).
For those wanting more precise control over the process, an object with several
parameters can be provided for the paged option. The pageSize parameter
sets the size of result pages requested from the server. If no value is
specified, it will fall back to the default (100 or sizeLimit - 1, to obey
the RFC). The pagePause parameter allows back-pressure to be exerted on the
paged search operation by pausing at the end of each page. When enabled, a
callback function is passed as an additional parameter to page events. The
client will wait to request the next page until that callback is executed.
Here is an example where both of those parameters are used:
const queue = new MyWorkQueue(someSlowWorkFunction);
const opts = {
filter: '(objectclass=commonobject)',
scope: 'sub',
paged: {
pageSize: 250,
pagePause: true
},
};
const res = await client.search('o=largerdir', opts);
res.on('searchEntry', (entry) => {
// Submit incoming objects to queue
queue.push(entry);
});
res.on('page', (result, cb) => {
// Allow the queue to flush before fetching next page
queue.cbWhenFlushed(cb);
});
res.on('error', (resErr) => {
assert.ifError(resErr);
});
res.on('end', (result) => {
console.log('done');
});
searchReturnAll
searchReturnAll(base, options, controls)
Performs a search operation against the LDAP server.
A wrapper method around the search method that collects all of the entries and referrals from the search and returns it as an object. An opinionated function that simply returns the results but lacks the control of the search method.
Example:
const opts = {
filter: '(&(l=Seattle)(email=*@foo.com))',
scope: 'sub',
attributes: ['dn', 'sn', 'cn']
};
const res = await client.searchReturnAll('o=example', opts);
console.log(res.entries);
starttls
starttls(options, controls)
Attempt to secure existing LDAP connection via STARTTLS.
Example:
const opts = {
ca: [fs.readFileSync('mycacert.pem')]
};
await client.starttls(opts);
// Client communication now TLS protected
unbind
unbind()
Performs an unbind operation against the LDAP server.
Note that unbind operation is not an opposite operation for bind. Unbinding results in disconnecting the client regardless of whether a bind operation was performed.
Example:
await client.unbind();