Common: fix PiiPseudonymizerEnrichment for arrays and improve unit te…

…sts coverage (close #334)
snowplow · Sep 16, 2020 · b7e893f · b7e893f
1 parent 6d5e838
commit b7e893f
Showing 1 changed file with 128 additions and 0 deletions.
diff --git a/...cala/com.snowplowanalytics.snowplow.enrich.common/enrichments/EnrichmentManagerSpec.scala b/...cala/com.snowplowanalytics.snowplow.enrich.common/enrichments/EnrichmentManagerSpec.scala
@@ -449,6 +449,134 @@ class EnrichmentManagerSpec extends Specification with EitherMatchers {
       enriched.value must beLeft
     }
 
+    "fail to emit an EnrichedEvent if a context PII value that needs to be hashed is an empty object" >> {
+      val parameters = Map(
+        "e" -> "ue",
+        "tv" -> "js-0.13.1",
+        "p" -> "web",
+        "co" -> """
+          {
+            "schema": "iglu:com.snowplowanalytics.snowplow/contexts/jsonschema/1-0-0",
+            "data": [
+              {
+                "schema":"iglu:com.acme/email_sent/jsonschema/1-0-0",
+                "data": {
+                  "emailAddress": "hello@world.com",
+                  "emailAddress2": "foo@bar.org",
+                  "emailAddress3": {}
+                }
+              }
+            ]
+          }
+        """,
+        "ue_pr" -> """
+          {
+            "schema":"iglu:com.snowplowanalytics.snowplow/unstruct_event/jsonschema/1-0-0",
+            "data":{
+              "schema":"iglu:com.acme/email_sent/jsonschema/1-0-0",
+              "data": {
+                "emailAddress": "hello@world.com",
+                "emailAddress2": "foo@bar.org"
+              }
+            }
+          }"""
+      )
+      val rawEvent = RawEvent(api, parameters, None, source, context)
+      val enrichmentReg = EnrichmentRegistry[Id](
+        piiPseudonymizer = PiiPseudonymizerEnrichment(
+          List(
+            PiiJson(
+              fieldMutator = JsonMutators("contexts"),
+              schemaCriterion = SchemaCriterion("com.acme", "email_sent", "jsonschema", 1, 0, 0),
+              jsonPath = "$.emailAddress3"
+            )
+          ),
+          false,
+          PiiStrategyPseudonymize(
+            "MD5",
+            hashFunction = DigestUtils.sha256Hex(_: Array[Byte]),
+            "pepper123"
+          )
+        ).some
+      )
+      def enriched =
+        EnrichmentManager.enrichEvent(
+          enrichmentReg,
+          client,
+          processor,
+          timestamp,
+          rawEvent
+        )
+      enriched.value must beLeft
+    }
+
+    "fail to emit an EnrichedEvent if a PII value needs to be hashed in both co and ue and is invalid in one of them" >> {
+      val parameters = Map(
+        "e" -> "ue",
+        "tv" -> "js-0.13.1",
+        "p" -> "web",
+        "co" -> """
+          {
+            "schema": "iglu:com.snowplowanalytics.snowplow/contexts/jsonschema/1-0-0",
+            "data": [
+              {
+                "schema":"iglu:com.acme/email_sent/jsonschema/1-0-0",
+                "data": {
+                  "emailAddress": "hello@world.com",
+                  "emailAddress2": "foo@bar.org",
+                  "emailAddress3": {}
+                }
+              }
+            ]
+          }
+        """,
+        "ue_pr" -> """
+          {
+            "schema":"iglu:com.snowplowanalytics.snowplow/unstruct_event/jsonschema/1-0-0",
+            "data":{
+              "schema":"iglu:com.acme/email_sent/jsonschema/1-0-0",
+              "data": {
+                "emailAddress": "hello@world.com",
+                "emailAddress2": "foo@bar.org",
+                "emailAddress3": ""
+              }
+            }
+          }"""
+      )
+      val rawEvent = RawEvent(api, parameters, None, source, context)
+      val enrichmentReg = EnrichmentRegistry[Id](
+        piiPseudonymizer = PiiPseudonymizerEnrichment(
+          List(
+            PiiJson(
+              fieldMutator = JsonMutators("contexts"),
+              schemaCriterion = SchemaCriterion("com.acme", "email_sent", "jsonschema", 1, 0, 0),
+              jsonPath = "$.emailAddress3"
+            ),
+            PiiJson(
+              fieldMutator = JsonMutators("unstruct_event"),
+              schemaCriterion = SchemaCriterion("com.acme", "email_sent", "jsonschema", 1, 0, 0),
+              jsonPath = "$.emailAddress3"
+            )
+          ),
+          false,
+          PiiStrategyPseudonymize(
+            "MD5",
+            hashFunction = DigestUtils.sha256Hex(_: Array[Byte]),
+            "pepper123"
+          )
+        ).some
+      )
+      def enriched =
+        EnrichmentManager.enrichEvent(
+          enrichmentReg,
+          client,
+          processor,
+          timestamp,
+          rawEvent
+        )
+      enriched.value must beLeft
+    }
+
     "have a preference of 'ua' query string parameter over user agent of HTTP header" >> {
       val qs_ua = "Mozilla/5.0 (X11; Linux x86_64; rv:75.0) Gecko/20100101 Firefox/75.0"
       val parameters = Map(