dns-packet
v5.6.1
Published
An abstract-encoding compliant module for encoding / decoding DNS packets
Downloads
51,507,614
Readme
dns-packet
An abstract-encoding compliant module for encoding / decoding DNS packets. Lifted out of multicast-dns as a separate module.
npm install dns-packet
UDP Usage
const dnsPacket = require('dns-packet')
const dgram = require('dgram')
const socket = dgram.createSocket('udp4')
const buf = dnsPacket.encode({
type: 'query',
id: 1,
flags: dnsPacket.RECURSION_DESIRED,
questions: [{
type: 'A',
name: 'google.com'
}]
})
socket.on('message', message => {
console.log(dnsPacket.decode(message)) // prints out a response from google dns
})
socket.send(buf, 0, buf.length, 53, '8.8.8.8')
Also see the UDP example.
TCP, TLS, HTTPS
While DNS has traditionally been used over a datagram transport, it is increasingly being carried over TCP for larger responses commonly including DNSSEC responses and TLS or HTTPS for enhanced security. See below examples on how to use dns-packet
to wrap DNS packets in these protocols:
API
var buf = packets.encode(packet, [buf], [offset])
Encodes a DNS packet into a buffer containing a UDP payload.
var packet = packets.decode(buf, [offset])
Decode a DNS packet from a buffer containing a UDP payload.
var buf = packets.streamEncode(packet, [buf], [offset])
Encodes a DNS packet into a buffer containing a TCP payload.
var packet = packets.streamDecode(buf, [offset])
Decode a DNS packet from a buffer containing a TCP payload.
var len = packets.encodingLength(packet)
Returns how many bytes are needed to encode the DNS packet
Packets
Packets look like this
{
type: 'query|response',
id: optionalIdNumber,
flags: optionalBitFlags,
questions: [...],
answers: [...],
additionals: [...],
authorities: [...]
}
The bit flags available are
packet.RECURSION_DESIRED
packet.RECURSION_AVAILABLE
packet.TRUNCATED_RESPONSE
packet.AUTHORITATIVE_ANSWER
packet.AUTHENTIC_DATA
packet.CHECKING_DISABLED
To use more than one flag bitwise-or them together
var flags = packet.RECURSION_DESIRED | packet.RECURSION_AVAILABLE
And to check for a flag use bitwise-and
var isRecursive = message.flags & packet.RECURSION_DESIRED
A question looks like this
{
type: 'A', // or SRV, AAAA, etc
class: 'IN', // one of IN, CS, CH, HS, ANY. Default: IN
name: 'google.com' // which record are you looking for
}
And an answer, additional, or authority looks like this
{
type: 'A', // or SRV, AAAA, etc
class: 'IN', // one of IN, CS, CH, HS
name: 'google.com', // which name is this record for
ttl: optionalTimeToLiveInSeconds,
(record specific data, see below)
}
Supported record types
A
{
data: 'IPv4 address' // fx 127.0.0.1
}
AAAA
{
data: 'IPv6 address' // fx fe80::1
}
CAA
{
flags: 128, // octet
tag: 'issue|issuewild|iodef',
value: 'ca.example.net',
issuerCritical: false
}
CNAME
{
data: 'cname.to.another.record'
}
DNAME
{
data: 'dname.to.another.record'
}
DNSKEY
{
flags: 257, // 16 bits
algorithm: 1, // octet
key: Buffer
}
DS
{
keyTag: 12345,
algorithm: 8,
digestType: 1,
digest: Buffer
}
HINFO
{
data: {
cpu: 'cpu info',
os: 'os info'
}
}
MX
{
preference: 10,
exchange: 'mail.example.net'
}
NAPTR
{
data:
{
order: 100,
preference: 10,
flags: 's',
services: 'SIP+D2U',
regexp: '!^.*$!sip:[email protected]!',
replacement: '_sip._udp.example.com'
}
}
NS
{
data: nameServer
}
NSEC
{
nextDomain: 'a.domain',
rrtypes: ['A', 'TXT', 'RRSIG']
}
NSEC3
{
algorithm: 1,
flags: 0,
iterations: 2,
salt: Buffer,
nextDomain: Buffer, // Hashed per RFC5155
rrtypes: ['A', 'TXT', 'RRSIG']
}
NULL
{
data: Buffer('any binary data')
}
OPT
EDNS0 options.
{
type: 'OPT',
name: '.',
udpPayloadSize: 4096,
flags: packet.DNSSEC_OK,
options: [{
// pass in any code/data for generic EDNS0 options
code: 12,
data: Buffer.alloc(31)
}, {
// Several EDNS0 options have enhanced support
code: 'PADDING',
length: 31,
}, {
code: 'CLIENT_SUBNET',
family: 2, // 1 for IPv4, 2 for IPv6
sourcePrefixLength: 64, // used to truncate IP address
scopePrefixLength: 0,
ip: 'fe80::',
}, {
code: 'TCP_KEEPALIVE',
timeout: 150 // increments of 100ms. This means 15s.
}, {
code: 'KEY_TAG',
tags: [1, 2, 3],
}]
}
The options PADDING
, CLIENT_SUBNET
, TCP_KEEPALIVE
and KEY_TAG
support enhanced de/encoding. See optionscodes.js for all supported option codes. If the data
property is present on a option, it takes precedence. On decoding, data
will always be defined.
PTR
{
data: 'points.to.another.record'
}
RP
{
mbox: 'admin.example.com',
txt: 'txt.example.com'
}
SSHFP
{
algorithm: 1,
hash: 1,
fingerprint: 'A108C9F834354D5B37AF988141C9294822F5BC00'
}
RRSIG
{
typeCovered: 'A',
algorithm: 8,
labels: 1,
originalTTL: 3600,
expiration: timestamp,
inception: timestamp,
keyTag: 12345,
signersName: 'a.name',
signature: Buffer
}
SOA
{
data:
{
mname: domainName,
rname: mailbox,
serial: zoneSerial,
refresh: refreshInterval,
retry: retryInterval,
expire: expireInterval,
minimum: minimumTTL
}
}
SRV
{
data: {
port: servicePort,
target: serviceHostName,
priority: optionalServicePriority,
weight: optionalServiceWeight
}
}
TLSA
{
usage: 3,
selector: 1,
matchingType: 1,
certificate: Buffer
}
TXT
{
data: 'text' || Buffer || [ Buffer || 'text' ]
}
When encoding, scalar values are converted to an array and strings are converted to UTF-8 encoded Buffers. When decoding, the return value will always be an array of Buffer.
If you need another record type, open an issue and we'll try to add it.
License
MIT