For #10300 - Fix Baidu search telemetry (#19127)

app/src/main/assets/extensions/ads/manifest.template.json

@@ -13,6 +13,8 @@
       "include_globs": [
         "https://www.google.*/search*",
         "https://www.bing.com/search*",
+        "https://www.baidu.com/*",
+        "https://m.baidu.com/*",
         "https://duckduckgo.com/*"
       ],
       "js": ["ads.js"],

app/src/main/assets/extensions/cookies/manifest.template.json

@@ -12,8 +12,8 @@
       "matches": ["https://*/*"],
       "include_globs": [
         "https://www.google.*/search*",
-        "https://www.baidu.com/from=844b/s*",
-        "https://www.baidu.com/from=844b/baidu*",
+        "https://www.baidu.com/*",
+        "https://m.baidu.com/*",
         "https://*search.yahoo.com/search*",
         "https://www.bing.com/search*",
         "https://duckduckgo.com/*"

app/src/main/java/org/mozilla/fenix/search/telemetry/BaseSearchTelemetry.kt

@@ -52,10 +52,10 @@ abstract class BaseSearchTelemetry {
         ),
         SearchProviderModel(
             name = "baidu",
-            regexp = "^https:\\/\\/www\\.baidu\\.com\\/from=844b\\/(?:s|baidu)",
-            queryParam = "wd",
-            codeParam = "tn",
-            codePrefixes = listOf("34046034_", "monline_"),
+            regexp = "^https:\\/\\/m\\.baidu\\.com(?:.*)\\/s",
+            queryParam = "word",
+            codeParam = "from",
+            codePrefixes = listOf("1000969a"),
             followOnParams = listOf("oq")
         ),
         SearchProviderModel(

app/src/main/java/org/mozilla/fenix/search/telemetry/Utils.kt

@@ -22,6 +22,20 @@ internal fun getTrackKey(
 
     if (provider.codeParam.isNotEmpty()) {
         code = uri.getQueryParameter(provider.codeParam)
+        if (code.isNullOrEmpty() &&
+            provider.name == "baidu" &&
+            uri.toString().contains("from=")) {
+            code = uri.toString().substringAfter("from=", "")
+                    .substringBefore("/", "")
+        }
+
+        // Glean doesn't allow code starting with a figure
+        if (code != null && code.isNotEmpty()) {
+            val codeStart = code.first()
+            if (codeStart.isDigit()) {
+                code = "_$code"
+            }
+        }
 
         // Try cookies first because Bing has followOnCookies and valid code, but no
         // followOnParams => would tracks organic instead of sap-follow-on
@@ -33,7 +47,7 @@ internal fun getTrackKey(
         }
 
         // For Bing if it didn't have a valid cookie and for all the other search engines
-        if (hasValidCode(code, provider)) {
+        if (hasValidCode(uri.getQueryParameter(provider.codeParam), provider)) {
             val channel = uri.getQueryParameter(CHANNEL_KEY)
             val type = getSapType(provider.followOnParams, paramSet)
             return TrackKeyInfo(provider.name, type, code, channel).createTrackKey()

app/src/test/java/org/mozilla/fenix/search/telemetry/incontent/InContentTelemetryTest.kt

@@ -84,11 +84,11 @@ class InContentTelemetryTest {
 
     @Test
     fun `track baidu sap metric`() {
-        val url = "https://www.baidu.com/from=844b/s?wd=aaa&tn=34046034_firefox"
+        val url = "https://m.baidu.com/s?from=1000969a&word=aaa"
 
         telemetry.trackPartnerUrlTypeMetric(url, listOf())
 
-        verify { metrics.track(Event.SearchInContent("baidu.in-content.sap.34046034_firefox")) }
+        verify { metrics.track(Event.SearchInContent("baidu.in-content.sap._1000969a")) }
     }
 
     @Test
@@ -120,11 +120,11 @@ class InContentTelemetryTest {
 
     @Test
     fun `track baidu sap-follow-on metric`() {
-        val url = "https://www.baidu.com/from=844b/s?wd=aaa&tn=34046034_firefox&oq=random"
+        val url = "https://m.baidu.com/s?from=1000969a&word=aaa&oq=random"
 
         telemetry.trackPartnerUrlTypeMetric(url, listOf())
 
-        verify { metrics.track(Event.SearchInContent("baidu.in-content.sap-follow-on.34046034_firefox")) }
+        verify { metrics.track(Event.SearchInContent("baidu.in-content.sap-follow-on._1000969a")) }
     }
 
     @Test
@@ -165,7 +165,7 @@ class InContentTelemetryTest {
 
     @Test
     fun `track baidu organic metric`() {
-        val url = "https://www.baidu.com/from=844b/s?wd=aaa"
+        val url = "https://m.baidu.com/s?word=aaa"
 
         telemetry.trackPartnerUrlTypeMetric(url, listOf())