feat: Enhance Proxy Request Handling to Display Actual External URLs (#…

…2698)

Thanks for the contribution @mstarzec386

lib/instrumentation/core/http-outbound.js

@@ -15,6 +15,7 @@ const copy = require('../../util/copy')
 const symbols = require('../../symbols')
 const http = require('http')
 const synthetics = require('../../synthetics')
+const { URL } = require('node:url')
 
 const NAMES = require('../../metrics/names')
 const DEFAULT_HOST = 'localhost'
@@ -93,6 +94,27 @@ function extractHostPort(opts) {
   return { host, hostname, port }
 }
 
+/**
+ * Extracts the host, hostname, and port from HTTP request options when using a proxy
+ *
+ * @param {object} opts HTTP request options
+ * @returns {object|null} { host, hostname, port } if proxy request is detected, or null otherwise.
+ */
+function extractHostPortViaProxy(opts) {
+  const pathname = opts.pathname || opts.path
+
+  if (pathname && (pathname.startsWith('https://') || pathname.startsWith('http://'))) {
+    const url = new URL(pathname)
+    return {
+      host: url.host,
+      hostname: url.hostname,
+      port: url.port || url.protocol === 'https:' ? '443' : '80'
+    }
+  }
+
+  return null
+}
+
 /**
  * Instruments an outbound HTTP request.
  *
@@ -103,7 +125,9 @@ function extractHostPort(opts) {
  */
 module.exports = function instrumentOutbound(agent, opts, makeRequest) {
   opts = parseOpts(opts)
-  const { host, hostname, port } = extractHostPort(opts)
+
+  const viaProxy = extractHostPortViaProxy(opts)
+  const { host, hostname, port } = viaProxy ? viaProxy : extractHostPort(opts)
 
   if (!hostname || port < 1) {
     logger.warn('Invalid host name (%s) or port (%s) for outbound request.', hostname, port)

test/integration/instrumentation/http-outbound.tap.js

@@ -111,6 +111,45 @@ tap.test('external requests', function (t) {
     }
   })
 
+  t.test('should recognize requests via proxy correctly', function (t) {
+    const proxyUrl = 'https://www.google.com/proxy/path'
+    const proxyServer = http.createServer(function onRequest(req, res) {
+      t.equal(req.url, proxyUrl)
+      req.resume()
+      res.end('ok')
+    })
+    t.teardown(() => proxyServer.close())
+
+    proxyServer.listen(0)
+
+    helper.runInTransaction(agent, function inTransaction() {
+      const opts = {
+        host: 'localhost',
+        port: proxyServer.address().port,
+        path: proxyUrl,
+        protocol: 'http:'
+      }
+
+      const req = http.get(opts, function onResponse(res) {
+        res.resume()
+        res.once('end', function () {
+          const segment = agent.tracer.getTransaction().trace.root.children[0]
+          t.equal(
+            segment.name,
+            `External/www.google.com/proxy/path`,
+            'should name segment as an external service'
+          )
+          t.end()
+        })
+      })
+
+      req.on('error', function onError(err) {
+        t.fail('Request should not error: ' + err.message)
+        t.end()
+      })
+    })
+  })
+
   t.test('should not duplicate the external segment', function (t) {
     const https = require('https')