From 2ebf446cb4c8d8f0ff465f933eadd1c00e8bbbe0 Mon Sep 17 00:00:00 2001
From: Mathias Buus <mathiasbuus@gmail.com>
Date: Fri, 23 Apr 2021 13:50:32 +0200
Subject: [PATCH] add remoteAddress backed by the nat analyzer

---
 index.js            |  15 +++++-
 lib/nat-analyzer.js | 108 ++++++++++++++++++++++++++++++++++++++++++++
 test.js             |   6 +++
 3 files changed, 127 insertions(+), 2 deletions(-)
 create mode 100644 lib/nat-analyzer.js

diff --git a/index.js b/index.js
index ee30b60..5c56713 100644
--- a/index.js
+++ b/index.js
@@ -1,6 +1,7 @@
 const dns = require('dns')
 const RPC = require('./lib/rpc')
 const Query = require('./lib/query')
+const NatAnalyzer = require('./lib/nat-analyzer')
 const Table = require('kademlia-routing-table')
 const TOS = require('time-ordered-set')
 const FIFO = require('fast-fifo/fixed-size')
@@ -73,6 +74,7 @@ class DHT extends EventEmitter {
     this._refreshTick = this._tick + REFRESH_TICKS
     this._stableTick = this._tick + STABLE_TICKS
     this._tickInterval = setInterval(this._ontick.bind(this), TICK_INTERVAL)
+    this._nat = new NatAnalyzer(opts.natSampleSize || 16)
 
     this.table.on('row', (row) => row.on('full', (node) => this._onfullrow(node, row)))
   }
@@ -341,19 +343,20 @@ class DHT extends EventEmitter {
     if (oldNode) {
       if (oldNode.port === m.from.port && oldNode.host === m.from.host) {
         // refresh it
-        oldNode.to = m.to
         oldNode.seen = this._tick
         this.nodes.add(oldNode)
       }
       return
     }
 
+    // add a sample of our address from the remote nodes pov
+    this._nat.add(m.to)
+
     this._addNode({
       id: m.nodeId,
       token: null,
       port: m.from.port,
       host: m.from.host,
-      to: m.to,
       added: this._tick,
       seen: this._tick,
       prev: null,
@@ -399,6 +402,10 @@ class DHT extends EventEmitter {
     return this.rpc.address()
   }
 
+  remoteAddress () {
+    return this._nat.analyze()
+  }
+
   _reply (rpc, tid, target, status, value, token, to) {
     const closerNodes = target ? this.table.closest(target) : null
     const persistent = !this.ephemeral && rpc === this.rpc
@@ -422,6 +429,10 @@ class DHT extends EventEmitter {
 DHT.OK = 0
 DHT.UNKNOWN_COMMAND = 1
 DHT.BAD_TOKEN = 2
+DHT.NAT_UNKNOWN = NatAnalyzer.UNKNOWN
+DHT.NAT_PORT_CONSISTENT = NatAnalyzer.PORT_CONSISTENT
+DHT.NAT_PORT_INCREMENTING = NatAnalyzer.PORT_INCREMENTING
+DHT.NAT_PORT_RANDOMIZED = NatAnalyzer.PORT_RANDOMIZED
 
 module.exports = DHT
 
diff --git a/lib/nat-analyzer.js b/lib/nat-analyzer.js
new file mode 100644
index 0000000..66b1abc
--- /dev/null
+++ b/lib/nat-analyzer.js
@@ -0,0 +1,108 @@
+// how far can the port median distance be?
+const INCREMENTING_THRESHOLD = 200
+
+class NatAnalyzer {
+  constructor (sampleSize) {
+    // sampleSize must be 2^n
+    this.samples = new Array(sampleSize)
+    this.length = 0
+    this.top = 0
+  }
+
+  add (addr) {
+    if (this.length < this.samples.length) this.length++
+    this.samples[this.top] = { port: addr.port, host: addr.host, dist: 0 }
+    this.top = (this.top + 1) & (this.samples.length - 1)
+  }
+
+  analyze () {
+    if (this.length <= 2) return { type: NatAnalyzer.UNKNOWN, host: null, port: 0 }
+
+    const samples = this.samples.slice(0, this.length)
+    const hosts = new Map()
+
+    let bestHost = null
+    let bestHits = 0
+
+    for (let i = 0; i < samples.length; i++) {
+      const host = samples[i].host
+      const hits = (hosts.get(host) || 0) + 1
+
+      hosts.set(host, hits)
+
+      if (hits > bestHits) {
+        bestHits = hits
+        bestHost = host
+      }
+    }
+
+    if (bestHits < (samples.length >> 1)) {
+      return { type: NatAnalyzer.UNKNOWN, host: null, port: 0 }
+    }
+
+    samples.sort(cmpPort)
+
+    let start = 0
+    let end = samples.length
+    let mid = samples[samples.length >> 1].port
+
+    // remove the 3 biggest outliers from the median if we have more than 6 samples
+    if (samples.length >= 6) {
+      for (let i = 0; i < 3; i++) {
+        const s = samples[start]
+        const e = samples[end - 1]
+
+        if (Math.abs(mid - s.port) < Math.abs(mid - e.port)) end--
+        else start++
+      }
+    }
+
+    const len = end - start
+    mid = samples[len >> 1].port
+
+    for (let i = 0; i < samples.length; i++) {
+      samples[i].dist = Math.abs(mid - samples[i].port)
+    }
+
+    // note that still sorts with the outliers which is why we just start=0, end=len-1 below
+    samples.sort(cmpDist)
+    mid = samples[len >> 1].dist
+
+    if (samples[0].dist === 0 && samples[len - 1].dist === 0) {
+      return {
+        type: NatAnalyzer.PORT_CONSISTENT,
+        host: bestHost,
+        port: samples[0].port
+      }
+    }
+
+    if (mid < INCREMENTING_THRESHOLD) {
+      return {
+        type: NatAnalyzer.PORT_INCREMENTING,
+        host: bestHost,
+        port: 0
+      }
+    }
+
+    return {
+      type: NatAnalyzer.PORT_RANDOMIZED,
+      host: bestHost,
+      port: 0
+    }
+  }
+}
+
+NatAnalyzer.UNKNOWN = Symbol.for('NAT_UNKNOWN')
+NatAnalyzer.PORT_CONSISTENT = Symbol.for('NAT_PORT_CONSISTENT')
+NatAnalyzer.PORT_INCREMENTING = Symbol.for('NAT_PORT_INCREMENTING')
+NatAnalyzer.PORT_RANDOMIZED = Symbol.for('NAT_PORT_RANDOM')
+
+module.exports = NatAnalyzer
+
+function cmpDist (a, b) {
+  return a.dist - b.dist
+}
+
+function cmpPort (a, b) {
+  return a.port - b.port
+}
diff --git a/test.js b/test.js
index 8498467..aa926f1 100644
--- a/test.js
+++ b/test.js
@@ -40,6 +40,12 @@ tape('make bigger swarm', async function (t) {
 
   t.ok(found, 'found target again in ' + messages + ' message(s)')
 
+  const { type, host, port } = swarm[490].remoteAddress()
+
+  t.same(type, DHT.NAT_PORT_CONSISTENT)
+  t.same(port, swarm[490].address().port)
+  t.ok(host)
+
   destroy(swarm)
 })