fix(http): data urls are now shortened

Author: gergelyke

lib/instrumentations/core/http/request.js

@@ -2,6 +2,8 @@ var debug = require('debug')('risingstack/trace')
 var url = require('url')
 var microtime = require('microtime')
 
+var util = require('./util')
+
 function wrapRequest (originalHttpRequest, agent, mustCollectStore) {
   var whiteListHosts = agent.getConfig().whiteListHosts
 
@@ -54,7 +56,7 @@ function wrapRequest (originalHttpRequest, agent, mustCollectStore) {
       id: requestId,
       spanId: spanId,
       host: requestParams.host,
-      url: requestParams.path,
+      url: util.formatDataUrl(requestParams.path),
       time: clientSendTime,
       method: requestParams.method,
       mustCollect: !!mustCollectStore[requestId]
@@ -76,7 +78,7 @@ function wrapRequest (originalHttpRequest, agent, mustCollectStore) {
         id: requestId,
         spanId: spanId,
         host: requestParams.host,
-        url: requestParams.path,
+        url: util.formatDataUrl(requestParams.path),
         mustCollect: true,
         err: {
           type: 'network-error',
@@ -108,7 +110,7 @@ function wrapRequest (originalHttpRequest, agent, mustCollectStore) {
         id: requestId,
         spanId: spanId,
         host: requestParams.host,
-        url: requestParams.path,
+        url: util.formatDataUrl(requestParams.path),
         statusCode: incomingMessage.statusCode,
         mustCollect: mustCollectStore[requestId]
       }

lib/instrumentations/core/http/server.js

@@ -4,6 +4,8 @@ var qs = require('qs')
 var microtime = require('microtime')
 var debug = require('debug')('risingstack/trace')
 
+var util = require('./util')
+
 function wrapListener (listener, agent, mustCollectStore) {
   var ignoreHeaders = agent.getConfig().ignoreHeaders
 
@@ -18,12 +20,13 @@ function wrapListener (listener, agent, mustCollectStore) {
     })
 
     if (skipped) {
-      debug('trace event (sr); request skipped', headers)
+      debug('trace event (sr); request skipped because of ignoreHeaders', headers)
       return listener.apply(this, arguments)
     }
 
     var requestUrl = url.parse(request.url)
     var requestQuery = qs.parse(requestUrl.query).requestId
+
     var originalWriteHead = response.writeHead
 
     var requestId = headers['request-id'] || requestQuery || agent.generateId()
@@ -48,7 +51,7 @@ function wrapListener (listener, agent, mustCollectStore) {
       id: requestId,
       spanId: spanId,
       host: headers.host,
-      url: requestUrl.pathname,
+      url: util.formatDataUrl(requestUrl.pathname),
       time: serverRecieveTime,
       method: method,
       parentId: headers['x-parent'],
@@ -70,7 +73,7 @@ function wrapListener (listener, agent, mustCollectStore) {
         id: requestId,
         spanId: headers['x-span-id'],
         host: headers.host,
-        url: requestUrl.pathname,
+        url: util.formatDataUrl(requestUrl.pathname),
         time: serverSendTime,
         statusCode: response.statusCode,
         responseTime: responseTime

lib/instrumentations/core/http/util.js

@@ -0,0 +1,13 @@
+var DATA_URI_REGEX = /(data:([a-z]+\/[a-z0-9\-\+]+(;[a-z\-]+\=[a-z0-9\-]+)?)?(;base64)?,)([a-z0-9\!\$\&\'\,\(\)\*\+\,\;\=\-\.\_\~\:\@\/\?\%\s]*\s*)$/i
+
+function formatDataUrl (text) {
+  if (!text) {
+    return
+  }
+
+  return text.replace(DATA_URI_REGEX, function replacer (match, p1) {
+    return [p1, '{data}'].join('')
+  })
+}
+
+module.exports.formatDataUrl = formatDataUrl

lib/instrumentations/core/http/util.spec.js

@@ -0,0 +1,17 @@
+var expect = require('chai').expect
+var util = require('./util')
+
+describe('The http util module', function () {
+  describe('#formatDataUrl', function () {
+    it('substitutes the data part', function () {
+      var url = '/image?data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAABAAAAAQAQMAAAAlPW0iAAAABlBMVEUAAAD///+l2Z/dAAAAM0lEQVR4nGP4/5/h/1+G/58ZDrAz3D/McH8yw83NDDeNGe4Ug9C9zwz3gVLMDA/A6P9/AFGGFyjOXZtQAAAAAElFTkSuQmCC'
+      var formatted = util.formatDataUrl(url)
+      expect(formatted).to.eql('/image?data:image/png;base64,{data}')
+    })
+    it('doesn\'t touch non-data urls', function () {
+      var url = '/image?data=4'
+      var formatted = util.formatDataUrl(url)
+      expect(formatted).to.eql(url)
+    })
+  })
+})