Merge tag 'adwords-npe' into develop

Fix for location's parameter '&gclid' or '&gclsrc' without '?'.
Hurence · Oct 16, 2023 · fa589fd · fa589fd
2 parents 6346bff + a20186c
commit fa589fd
Show file tree

Hide file tree

Showing 2 changed files with 42 additions and 4 deletions.
diff --git a/...lytics/src/main/java/com/hurence/logisland/processor/webanalytics/SetSourceOfTraffic.java b/...lytics/src/main/java/com/hurence/logisland/processor/webanalytics/SetSourceOfTraffic.java
@@ -336,8 +336,9 @@ public void init(final ProcessContext context) throws InitializationException {
     private boolean adwords(final String location, final SourceOfTrafficMap sourceOfTraffic) {
         boolean processed = false;
 
-        final boolean adwords = location!=null && ADWORDS_RE.matcher(location).matches();
-        if (adwords || (location!=null && DOUBLECLICK_RE.matcher(location).matches())) {
+        final boolean hasQuery = location!=null && location.contains("?");
+        final boolean adwords = hasQuery && ADWORDS_RE.matcher(location).matches();
+        if (adwords || (hasQuery && DOUBLECLICK_RE.matcher(location).matches())) {
             processed = true;
             sourceOfTraffic.setSource(GOOGLE);
             sourceOfTraffic.setMedium(CPC);
@@ -346,11 +347,17 @@ private boolean adwords(final String location, final SourceOfTrafficMap sourceOf
             sourceOfTraffic.setKeyword(adwords ? ADWORDS : DOUBLECLICK);
 
             try {
-                setAdwordsCampaign(new URI(location).getQuery().split("&"), sourceOfTraffic);
+                final String query = new URI(location).getQuery();
+                if ( query!=null ) {
+                    setAdwordsCampaign(query.split("&"), sourceOfTraffic);
+                }
             } catch (URISyntaxException e) {
                 // Symbol `|` generates exception in URI but not in URL.
                 try {
-                    setAdwordsCampaign(new URL(location).getQuery().split("&"), sourceOfTraffic);
+                    final String query = new URL(location).getQuery();
+                    if ( query!=null ) {
+                        setAdwordsCampaign(query.split("&"), sourceOfTraffic);
+                    }
                 } catch (final MalformedURLException e2) {
                     getLogger().error("URISyntaxException", e);
                 }

diff --git a/...cs/src/test/java/com/hurence/logisland/processor/webanalytics/setSourceOfTrafficTest.java b/...cs/src/test/java/com/hurence/logisland/processor/webanalytics/setSourceOfTrafficTest.java
@@ -238,6 +238,37 @@ public void testAdwords(final String url) throws InitializationException {
         out.assertFieldEquals("source_of_traffic_keyword", "adwords");
     }
 
+    @Test
+    public void testGoogleAdNPEFix() throws InitializationException {
+        // The '?' is missing in the Url before the parameters. This was raising NPE.
+        final Record record1 = new StandardRecord();
+        record1.setField("firstVisitedPage", FieldType.STRING, "https://www.xyz_website.com/fr/index.htmlfake&gclid=XXX");
+        final Record record2 = new StandardRecord();
+        record2.setField("firstVisitedPage", FieldType.STRING, "https://www.xyz_website.com/fr/index.htmlfake&gclsrc=XXX");
+
+        final TestRunner testRunner = getTestRunner();
+
+        testRunner.assertValid();
+        testRunner.setProperty("cache.size", "5");
+        testRunner.setProperty("debug", "true");
+        testRunner.setProperty("source.out.field", "source_of_traffic");
+        testRunner.setProperty("source_of_traffic.hierarchical", "false");
+        testRunner.enqueue(record1, record2);
+        testRunner.run();
+        testRunner.assertAllInputRecordsProcessed();
+        testRunner.assertOutputRecordsCount(2);
+
+        final MockRecord out1 = testRunner.getOutputRecords().get(0);
+        out1.assertFieldEquals("source_of_traffic_source", "direct");
+        out1.assertFieldEquals("source_of_traffic_medium", "");
+        out1.assertFieldEquals("source_of_traffic_campaign", "direct");
+
+        final MockRecord out2 = testRunner.getOutputRecords().get(0);
+        out2.assertFieldEquals("source_of_traffic_source", "direct");
+        out2.assertFieldEquals("source_of_traffic_medium", "");
+        out2.assertFieldEquals("source_of_traffic_campaign", "direct");
+    }
+
     @Test
     public void testDoubleClick() throws InitializationException {
         // Test the DoubleClick case with presence of gclsrc parameter.