Analytics.js bridge updates

- Set network_userid to the same value set by the normal collector routes. - Set user_id to the value from the ajs_user_id cookie, fallback to the Segment payload -> userId value when the cookie is missing. - Set domain_userid to the value from the ajs_anonymous_id cookie.
opensnowcat · Aug 26, 2024 · dd54826 · dd54826
1 parent a601161
commit dd54826
Show file tree

Hide file tree

Showing 4 changed files with 83 additions and 61 deletions.
diff --git a/.../main/scala/com.snowplowanalytics.snowplow.collectors.scalastream/AnalyticsJsBridge.scala b/.../main/scala/com.snowplowanalytics.snowplow.collectors.scalastream/AnalyticsJsBridge.scala
@@ -14,6 +14,8 @@ object AnalyticsJsBridge {
   private val Vendor  = "com.segment"
   private val Version = "v1"
 
+  case class Event(eventType: EventType, anonymousUserId: Option[String], userId: Option[String])
+
   sealed trait EventType extends Product with Serializable
   object EventType {
     case object Page extends EventType
@@ -38,62 +40,70 @@ object AnalyticsJsBridge {
     collectorService: Service
   ) =
     path(Vendor / Version / Segment) { segment =>
-      // ideally, we should use /com.segment/v1 as the path but this requires a remote adapter on the enrich side
-      // instead, we are reusing the snowplow event type while attaching the segment payload
-      //
-      // consider using a path-mapping config instead.
-      val path = "/com.snowplowanalytics.snowplow/tp2"
-
-      // identify, track, page, screen, group, alias
-      val eventType = segment match {
-        case "i" => Some(AnalyticsJsBridge.EventType.Identify)
-        case "t" => Some(AnalyticsJsBridge.EventType.Track)
-        case "p" => Some(AnalyticsJsBridge.EventType.Page)
-        case "s" => Some(AnalyticsJsBridge.EventType.Screen)
-        case "g" => Some(AnalyticsJsBridge.EventType.Group)
-        case "a" => Some(AnalyticsJsBridge.EventType.Alias)
-        case _   => None
-      }
-      if (eventType.isDefined)
-        post {
-          extractContentType { ct =>
-            // analytics.js is sending "text/plain" content type which is not supported by the snowplow schema
-            val normalizedContentType =
-              ContentType.parse(ct.value.toLowerCase.replace("text/plain", "application/json")).toOption
-
-            entity(as[String]) { body =>
-              val r = collectorService.cookie(
-                queryString = queryString,
-                body = Some(body),
-                path = path,
-                cookie = cookie,
-                userAgent = userAgent,
-                refererUri = refererUri,
-                hostname = hostname,
-                ip = ip,
-                request = request,
-                pixelExpected = false,
-                doNotTrack = doNotTrack,
-                contentType = normalizedContentType,
-                spAnonymous = spAnonymous,
-                analyticsJsEvent = eventType
-              )
-              complete(r)
-            }
+      optionalCookie("ajs_anonymous_id") { ajsAnonymousUserIdCookie =>
+        optionalCookie("ajs_user_id") { ajsUserIdCookie =>
+          val anonymousUserId = ajsAnonymousUserIdCookie.map(_.toCookie().value())
+          val userId          = ajsUserIdCookie.map(_.toCookie().value())
+
+          // ideally, we should use /com.segment/v1 as the path but this requires a remote adapter on the enrich side
+          // instead, we are reusing the snowplow event type while attaching the segment payload
+          //
+          // consider using a path-mapping config instead.
+          val path = "/com.snowplowanalytics.snowplow/tp2"
+
+          // identify, track, page, screen, group, alias
+          val eventType = segment match {
+            case "i" => Some(AnalyticsJsBridge.EventType.Identify)
+            case "t" => Some(AnalyticsJsBridge.EventType.Track)
+            case "p" => Some(AnalyticsJsBridge.EventType.Page)
+            case "s" => Some(AnalyticsJsBridge.EventType.Screen)
+            case "g" => Some(AnalyticsJsBridge.EventType.Group)
+            case "a" => Some(AnalyticsJsBridge.EventType.Alias)
+            case _   => None
           }
+          if (eventType.isDefined)
+            post {
+              extractContentType { ct =>
+                // analytics.js is sending "text/plain" content type which is not supported by the snowplow schema
+                val normalizedContentType =
+                  ContentType.parse(ct.value.toLowerCase.replace("text/plain", "application/json")).toOption
+
+                entity(as[String]) { body =>
+                  val r = collectorService.cookie(
+                    queryString = queryString,
+                    body = Some(body),
+                    path = path,
+                    cookie = cookie,
+                    userAgent = userAgent,
+                    refererUri = refererUri,
+                    hostname = hostname,
+                    ip = ip,
+                    request = request,
+                    pixelExpected = false,
+                    doNotTrack = doNotTrack,
+                    contentType = normalizedContentType,
+                    spAnonymous = spAnonymous,
+                    analyticsJsEvent = eventType.map(t => Event(t, anonymousUserId = anonymousUserId, userId = userId))
+                  )
+                  complete(r)
+                }
+              }
+            }
+          else complete(HttpResponse(StatusCodes.BadRequest))
         }
-      else complete(HttpResponse(StatusCodes.BadRequest))
+      }
+
     }
 
-  def createSnowplowPayload(body: Json, eventType: EventType): Json = {
+  def createSnowplowPayload(body: Json, event: Event, networkUserId: String): Json = {
     import io.circe._
 
     import java.nio.charset.StandardCharsets
     import java.util.Base64
 
     val appId = "ajs_bridge"
 
-    val eventSchema = eventType match {
+    val eventSchema = event.eventType match {
       case EventType.Page     => "iglu:com.segment/page/jsonschema/2-0-0"
       case EventType.Identify => "iglu:com.segment/identify/jsonschema/1-0-0"
       case EventType.Track    => "iglu:com.segment/track/jsonschema/1-0-0"
@@ -118,11 +128,16 @@ object AnalyticsJsBridge {
     val properties = body.hcursor.downField("properties")
     val context    = body.hcursor.downField("context")
 
-    val url      = "url"  -> properties.get[String]("url")
-    val page     = "page" -> properties.get[String]("page")
-    val locale   = "lang" -> context.get[String]("locale")
-    val timezone = "tz"   -> context.get[String]("timezone")
-    val userId   = "uid"  -> body.hcursor.get[String]("userId")
+    val url      = "url"  -> properties.get[String]("url").toOption
+    val page     = "page" -> properties.get[String]("page").toOption
+    val locale   = "lang" -> context.get[String]("locale").toOption
+    val timezone = "tz"   -> context.get[String]("timezone").toOption
+    // user_id
+    val userId = "uid" -> event.userId.orElse {
+      body.hcursor.get[String]("userId").toOption
+    }
+    // domain_userid
+    val domainUserId = "duid" -> event.anonymousUserId
 
     val trackerVersion = context
       .downField("library")
@@ -141,11 +156,14 @@ object AnalyticsJsBridge {
       // base64-encoded event
       "ue_px" -> Json.fromString(
         Base64.getEncoder.encodeToString(eventPayload.noSpaces.getBytes(StandardCharsets.UTF_8))
-      )
+      ),
+      // network_userid
+      "tnuid" -> Json.fromString(networkUserId)
     )
 
     // merge optional arguments
-    val data = List(url, page, locale, timezone, userId).map(x => x._1 -> x._2.toOption).foldLeft(initialData) {
+    val optionalEntries = List(url, page, locale, timezone, userId, domainUserId)
+    val data = optionalEntries.foldLeft(initialData) {
       case (acc, (key, Some(value))) => acc.add(key, Json.fromString(value))
       case (acc, _)                  => acc
     }

diff --git a/...c/main/scala/com.snowplowanalytics.snowplow.collectors.scalastream/CollectorService.scala b/...c/main/scala/com.snowplowanalytics.snowplow.collectors.scalastream/CollectorService.scala
@@ -55,7 +55,7 @@ trait Service {
     doNotTrack: Boolean,
     contentType: Option[ContentType] = None,
     spAnonymous: Option[String] = None,
-    analyticsJsEvent: Option[AnalyticsJsBridge.EventType] = None
+    analyticsJsEvent: Option[AnalyticsJsBridge.Event] = None
   ): HttpResponse
   def cookieName: Option[String]
   def doNotTrackCookie: Option[DntCookieMatcher]
@@ -113,7 +113,7 @@ class CollectorService(
     doNotTrack: Boolean,
     contentType: Option[ContentType] = None,
     spAnonymous: Option[String],
-    analyticsJsEvent: Option[AnalyticsJsBridge.EventType] = None
+    analyticsJsEvent: Option[AnalyticsJsBridge.Event] = None
   ): HttpResponse = {
     val (ipAddress, partitionKey) = ipAndPartitionKey(ip, config.streams.useIpAddressAsPartitionKey)
 
@@ -249,7 +249,7 @@ class CollectorService(
     networkUserId: String,
     contentType: Option[String],
     spAnonymous: Option[String],
-    analyticsJsEvent: Option[AnalyticsJsBridge.EventType] = None
+    analyticsJsEvent: Option[AnalyticsJsBridge.Event] = None
   ): CollectorPayload = {
     val customBody = analyticsJsEvent match {
       case Some(eventType) =>
@@ -258,7 +258,7 @@ class CollectorService(
           .parser
           .parse(body.getOrElse("{}"))
           .getOrElse(throw new RuntimeException("The request body must be a JSON-encoded Analytic.js payload"))
-        val payload = AnalyticsJsBridge.createSnowplowPayload(jsonBody, eventType)
+        val payload = AnalyticsJsBridge.createSnowplowPayload(jsonBody, eventType, networkUserId)
         Some(payload.noSpaces)
 
       case None => body
@@ -310,7 +310,7 @@ class CollectorService(
     pixelExpected: Boolean,
     bounce: Boolean,
     redirectMacroConfig: RedirectMacroConfig,
-    analyticsJsEvent: Option[AnalyticsJsBridge.EventType] = None
+    analyticsJsEvent: Option[AnalyticsJsBridge.Event] = None
   ): HttpResponse =
     if (redirect) {
       val r = buildRedirectHttpResponse(event, queryParams, redirectMacroConfig)
@@ -324,7 +324,7 @@ class CollectorService(
   def buildUsualHttpResponse(
     pixelExpected: Boolean,
     bounce: Boolean,
-    analyticsJsEvent: Option[AnalyticsJsBridge.EventType] = None
+    analyticsJsEvent: Option[AnalyticsJsBridge.Event] = None
   ): HttpResponse =
     (pixelExpected, bounce, analyticsJsEvent) match {
       case (true, true, _) => HttpResponse(StatusCodes.Found)

diff --git a/...test/scala/com.snowplowanalytics.snowplow.collectors.scalastream/CollectorRouteSpec.scala b/...test/scala/com.snowplowanalytics.snowplow.collectors.scalastream/CollectorRouteSpec.scala
@@ -48,7 +48,7 @@ class CollectorRouteSpec extends Specification with Specs2RouteTest {
           doNotTrack: Boolean,
           contentType: Option[ContentType] = None,
           spAnonymous: Option[String] = spAnonymous,
-          analyticsJsEvent: Option[AnalyticsJsBridge.EventType] = None
+          analyticsJsEvent: Option[AnalyticsJsBridge.Event] = None
         ): HttpResponse =
           if (analyticsJsEvent.isDefined) {
             HttpResponse(200, entity = AnalyticsJsBridge.jsonResponse.noSpaces)

diff --git a/...st/scala/com.snowplowanalytics.snowplow.collectors.scalastream/CollectorServiceSpec.scala b/...st/scala/com.snowplowanalytics.snowplow.collectors.scalastream/CollectorServiceSpec.scala
@@ -535,7 +535,11 @@ class CollectorServiceSpec extends Specification {
         service.buildUsualHttpResponse(false, true) shouldEqual HttpResponse(200, entity = "ok")
       }
       "send back the analytics.js supported response" in {
-        service.buildUsualHttpResponse(false, true, Some(AnalyticsJsBridge.EventType.Page)) shouldEqual HttpResponse(
+        service.buildUsualHttpResponse(
+          false,
+          true,
+          Some(AnalyticsJsBridge.Event(AnalyticsJsBridge.EventType.Page, None, None))
+        ) shouldEqual HttpResponse(
           200,
           entity = """{"success":true}"""
         )
@@ -970,7 +974,7 @@ class CollectorServiceSpec extends Specification {
         request = HttpRequest(),
         pixelExpected = false,
         doNotTrack = false,
-        analyticsJsEvent = Some(eventType)
+        analyticsJsEvent = Some(AnalyticsJsBridge.Event(eventType, None, None))
       )
 
       good.storedRawEvents must have size 1