(don't merge) UID2-6805 email format analysis

pom.xml

@@ -6,7 +6,7 @@
 
     <groupId>com.uid2</groupId>
     <artifactId>uid2-operator</artifactId>
-    <version>5.70.6</version>
+    <version>5.70.7-alpha-323-SNAPSHOT</version>
 
     <properties>
         <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>

src/main/java/com/uid2/operator/vertx/UIDOperatorVerticle.java

@@ -910,6 +910,7 @@ private void handleTokenValidateV2(RoutingContext rc) {
                 recordTokenValidateStats(participantSiteId, "invalid_input");
                 return;
             }
+            recordRawEmailMetrics(input, rc.request().path());
 
             final Instant now = Instant.now();
             final String token = req.getString("token");
@@ -952,6 +953,7 @@ private void handleTokenGenerateV2(RoutingContext rc) {
 
             final InputUtil.InputVal input = this.getTokenInputV2(req);
             if (isTokenInputValid(input, rc)) {
+                recordRawEmailMetrics(input, rc.request().path());
                 final String apiContact = getApiContact(rc);
 
                 switch (validateUserConsent(req, apiContact)) {
@@ -1008,6 +1010,7 @@ private Future handleLogoutAsyncV2(RoutingContext rc) {
         final InputUtil.InputVal input = getTokenInputV2(req);
         final String uidTraceId = rc.request().getHeader(Audit.UID_TRACE_ID_HEADER);
         if (input != null && input.isValid()) {
+            recordRawEmailMetrics(input, rc.request().path());
             final Instant now = Instant.now();
 
             Promise promise = Promise.promise();
@@ -1232,6 +1235,12 @@ private void handleIdentityMapV2(RoutingContext rc) {
 
             if (!validateServiceLink(rc)) { return; }
 
+            if (v2Input.diiType().equals("email")) {
+                for (InputUtil.InputVal input : v2Input.inputList()) {
+                    recordRawEmailMetrics(input, rc.request().path());
+                }
+            }
+
             final JsonObject resp = processIdentityMapV2Response(rc, v2Input);
             ResponseUtil.SuccessV2(rc, resp);
         } catch (Exception e) {
@@ -1295,6 +1304,13 @@ private void handleIdentityMapV3(RoutingContext rc) {
 
             if (!validateServiceLink(rc)) { return; }
 
+            InputUtil.InputVal[] emailInputs = normalizedInput.get("email");
+            if (emailInputs != null) {
+                for (InputUtil.InputVal emailInput : emailInputs) {
+                    recordRawEmailMetrics(emailInput, rc.request().path());
+                }
+            }
+
             final JsonObject response = processIdentityMapV3Response(rc, normalizedInput);
             ResponseUtil.SuccessV2(rc, response);
         } catch (ClassCastException | JsonProcessingException processingException) {
@@ -1507,6 +1523,52 @@ public TokenVersion getRefreshTokenVersion(String s) {
         return null;
     }
 
+    private void recordRawEmailMetrics(InputUtil.InputVal input, String path) {
+        if (!input.isValid() || input.getInputType() != InputUtil.IdentityInputType.Raw
+                || input.getIdentityType() != IdentityType.Email) {
+            return;
+        }
+        String provided = input.getProvided();
+        int atIndex = provided.indexOf('@');
+        boolean isGmail = input.getNormalized().endsWith("@gmail.com");
+
+        boolean hasDot = atIndex > 0 && provided.lastIndexOf('.', atIndex) >= 0;
+        Counter.builder("uid2_operator_raw_email_dot_total")
+                .description("Count of valid raw emails processed, by presence of dot before @")
+                .tag("path", path)
+                .tag("has_dot", String.valueOf(hasDot))
+                .tag("is_gmail", String.valueOf(isGmail))
+                .register(Metrics.globalRegistry)
+                .increment();
+
+        int plusIndex = provided.indexOf('+');
+        boolean hasPlus = plusIndex >= 0 && plusIndex < atIndex;
+        Counter.builder("uid2_operator_raw_email_plus_total")
+                .description("Count of valid raw emails processed, by presence of + before @")
+                .tag("path", path)
+                .tag("has_plus", String.valueOf(hasPlus))
+                .tag("is_gmail", String.valueOf(isGmail))
+                .register(Metrics.globalRegistry)
+                .increment();
+
+        boolean hasTrailingDot = provided.charAt(provided.length() - 1) == '.';
+        Counter.builder("uid2_operator_raw_email_trailing_dot_total")
+                .description("Count of valid raw emails processed, by presence of trailing dot at end of address")
+                .tag("path", path)
+                .tag("has_trailing_dot", String.valueOf(hasTrailingDot))
+                .register(Metrics.globalRegistry)
+                .increment();
+
+        boolean hasWhitespace = provided.strip().chars().anyMatch(Character::isWhitespace);
+        Counter.builder("uid2_operator_raw_email_whitespace_total")
+                .description("Count of valid raw emails processed, by presence of internal whitespace and gmail domain")
+                .tag("path", path)
+                .tag("has_whitespace", String.valueOf(hasWhitespace))
+                .tag("is_gmail", String.valueOf(isGmail))
+                .register(Metrics.globalRegistry)
+                .increment();
+    }
+
     private void recordRefreshTokenVersionCount(String siteId, TokenVersion tokenVersion) {
         Counter.builder("uid2_refresh_token_received_count_total")
                 .description(String.format("Counter for the amount of refresh token %s received", tokenVersion.toString().toLowerCase()))

src/test/java/com/uid2/operator/UIDOperatorVerticleTest.java

@@ -5253,6 +5253,67 @@ void asyncBatchRequestEnabledLogsCorrectMessage(Vertx vertx, VertxTestContext te
         testContext.completeNow();
     }
 
+    @Test
+    void rawEmailMetricsRecorded(Vertx vertx, VertxTestContext testContext) {
+        fakeAuth(201, Role.GENERATOR, Role.MAPPER);
+        setupSalts();
+        setupKeys();
+
+        // /v2/token/generate — one email per characteristic being tested
+        String emailHash = TokenUtils.getIdentityHashString("hash@example.com");
+        sendTokenGenerate(vertx, new JsonObject().put("email", "john.doe@example.com"), 200, json -> {});  // dot, no plus, non-gmail
+        sendTokenGenerate(vertx, new JsonObject().put("email", "johndoe@example.com"), 200, json -> {});   // no dot, non-gmail
+        sendTokenGenerate(vertx, new JsonObject().put("email", "john.doe@gmail.com"), 200, json -> {});    // dot, gmail
+        sendTokenGenerate(vertx, new JsonObject().put("email", "john+tag@example.com"), 200, json -> {});  // plus, non-gmail
+        sendTokenGenerate(vertx, new JsonObject().put("email", "john+tag@gmail.com"), 200, json -> {});    // plus, gmail
+        sendTokenGenerate(vertx, new JsonObject().put("email", "john@example.com."), 200, json -> {});     // trailing dot
+        sendTokenGenerate(vertx, new JsonObject().put("email_hash", emailHash), 200, json -> {             // hash - should not record
+
+            // dot metric
+            assertEquals(1, Metrics.globalRegistry
+                    .get("uid2_operator_raw_email_dot_total")
+                    .tag("path", "/v2/token/generate").tag("has_dot", "true").tag("is_gmail", "false")
+                    .counter().count());
+            assertEquals(1, Metrics.globalRegistry
+                    .get("uid2_operator_raw_email_dot_total")
+                    .tag("path", "/v2/token/generate").tag("has_dot", "true").tag("is_gmail", "true")
+                    .counter().count());
+            // has_dot=false, is_gmail=false count is 3 (johndoe, john+tag@example.com, john@example.com.) — hash did not increment
+            assertEquals(3, Metrics.globalRegistry
+                    .get("uid2_operator_raw_email_dot_total")
+                    .tag("path", "/v2/token/generate").tag("has_dot", "false").tag("is_gmail", "false")
+                    .counter().count());
+
+            // plus metric
+            assertEquals(1, Metrics.globalRegistry
+                    .get("uid2_operator_raw_email_plus_total")
+                    .tag("path", "/v2/token/generate").tag("has_plus", "true").tag("is_gmail", "false")
+                    .counter().count());
+            assertEquals(1, Metrics.globalRegistry
+                    .get("uid2_operator_raw_email_plus_total")
+                    .tag("path", "/v2/token/generate").tag("has_plus", "true").tag("is_gmail", "true")
+                    .counter().count());
+
+            // trailing dot metric
+            assertEquals(1, Metrics.globalRegistry
+                    .get("uid2_operator_raw_email_trailing_dot_total")
+                    .tag("path", "/v2/token/generate").tag("has_trailing_dot", "true")
+                    .counter().count());
+
+            // /v2/identity/map — batch with two emails to verify counter reaches 2
+            JsonObject mapReq = new JsonObject().put("email", new JsonArray()
+                    .add("a.b@example.com")
+                    .add("c.d@example.com"));
+            send(vertx, "v2/identity/map", mapReq, 200, mapJson -> {
+                assertEquals(2, Metrics.globalRegistry
+                        .get("uid2_operator_raw_email_dot_total")
+                        .tag("path", "/v2/identity/map").tag("has_dot", "true").tag("is_gmail", "false")
+                        .counter().count());
+                testContext.completeNow();
+            });
+        });
+    }
+
     @Test
     void asyncBatchRequestDisabledLogsCorrectMessage(Vertx vertx, VertxTestContext testContext) {
         // Verify that when enable_async_batch_request is false, the correct log message is emitted