Report discarded log bytes

CHANGELOG.md

@@ -18,6 +18,7 @@
 - Fallback to distinct-id as user.id logging attribute when user is not set ([#4847](https://github.com/getsentry/sentry-java/pull/4847))
 - Report Timber.tag() as `timber.tag` log attribute ([#4845](https://github.com/getsentry/sentry-java/pull/4845))
 - Session Replay: Add screenshot strategy serialization to RRWeb events ([#4851](https://github.com/getsentry/sentry-java/pull/4851))
+- Report discarded log bytes ([#4871](https://github.com/getsentry/sentry-java/pull/4871))
 - Log why a properties file was not loaded ([#4879](https://github.com/getsentry/sentry-java/pull/4879))
 
 ### Dependencies

sentry/api/sentry.api

@@ -347,6 +347,7 @@ public final class io/sentry/DataCategory : java/lang/Enum {
 	public static final field Default Lio/sentry/DataCategory;
 	public static final field Error Lio/sentry/DataCategory;
 	public static final field Feedback Lio/sentry/DataCategory;
+	public static final field LogByte Lio/sentry/DataCategory;
 	public static final field LogItem Lio/sentry/DataCategory;
 	public static final field Monitor Lio/sentry/DataCategory;
 	public static final field Profile Lio/sentry/DataCategory;
@@ -7074,6 +7075,7 @@ public final class io/sentry/util/IntegrationUtils {
 public final class io/sentry/util/JsonSerializationUtils {
 	public fun <init> ()V
 	public static fun atomicIntegerArrayToList (Ljava/util/concurrent/atomic/AtomicIntegerArray;)Ljava/util/List;
+	public static fun byteSizeOf (Lio/sentry/ISerializer;Lio/sentry/ILogger;Lio/sentry/JsonSerializable;)J
 	public static fun bytesFrom (Lio/sentry/ISerializer;Lio/sentry/ILogger;Lio/sentry/JsonSerializable;)[B
 	public static fun calendarToMap (Ljava/util/Calendar;)Ljava/util/Map;
 }

sentry/src/main/java/io/sentry/DataCategory.java

@@ -10,6 +10,7 @@ public enum DataCategory {
   Session("session"),
   Attachment("attachment"),
   LogItem("log_item"),
+  LogByte("log_byte"),
   Monitor("monitor"),
   Profile("profile"),
   ProfileChunkUi("profile_chunk_ui"),

sentry/src/main/java/io/sentry/SentryClient.java

@@ -1184,13 +1184,21 @@ public void captureLog(@Nullable SentryLogEvent logEvent, @Nullable IScope scope
     }
 
     if (logEvent != null) {
+      final @NotNull SentryLogEvent tmpLogEvent = logEvent;
       logEvent = executeBeforeSendLog(logEvent);
 
       if (logEvent == null) {
         options.getLogger().log(SentryLevel.DEBUG, "Log Event was dropped by beforeSendLog");
         options
             .getClientReportRecorder()
             .recordLostEvent(DiscardReason.BEFORE_SEND, DataCategory.LogItem);
+        final @NotNull long logEventNumberOfBytes =
+            JsonSerializationUtils.byteSizeOf(
+                options.getSerializer(), options.getLogger(), tmpLogEvent);
+        options
+            .getClientReportRecorder()
+            .recordLostEvent(
+                DiscardReason.BEFORE_SEND, DataCategory.LogByte, logEventNumberOfBytes);
         return;
       }
 

sentry/src/main/java/io/sentry/clientreport/ClientReportRecorder.java

@@ -6,6 +6,7 @@
 import io.sentry.SentryEnvelopeItem;
 import io.sentry.SentryItemType;
 import io.sentry.SentryLevel;
+import io.sentry.SentryLogEvent;
 import io.sentry.SentryLogEvents;
 import io.sentry.SentryOptions;
 import io.sentry.protocol.SentrySpan;
@@ -104,9 +105,15 @@ public void recordLostEnvelopeItem(
         } else if (itemCategory.equals(DataCategory.LogItem)) {
           final @Nullable SentryLogEvents logs = envelopeItem.getLogs(options.getSerializer());
           if (logs != null) {
-            final long count = logs.getItems().size();
+            final @NotNull List<SentryLogEvent> items = logs.getItems();
+            final long count = items.size();
             recordLostEventInternal(reason.getReason(), itemCategory.getCategory(), count);
+            final long logBytes = envelopeItem.getData().length;
+            recordLostEventInternal(
+                reason.getReason(), DataCategory.LogByte.getCategory(), logBytes);
             executeOnDiscard(reason, itemCategory, count);
+          } else {
+            options.getLogger().log(SentryLevel.ERROR, "Unable to parse lost logs envelope item.");
           }
         } else {
           recordLostEventInternal(reason.getReason(), itemCategory.getCategory(), 1L);

sentry/src/main/java/io/sentry/util/JsonSerializationUtils.java

@@ -65,4 +65,90 @@ public final class JsonSerializationUtils {
       return null;
     }
   }
+
+  /**
+   * Calculates the size in bytes of a serializable object when serialized to JSON without actually
+   * storing the serialized data. This is more memory efficient than {@link #bytesFrom(ISerializer,
+   * ILogger, JsonSerializable)} when you only need the size.
+   *
+   * @param serializer the serializer
+   * @param logger the logger
+   * @param serializable the serializable object
+   * @return the size in bytes, or 0 if serialization fails
+   */
+  public static long byteSizeOf(
+      final @NotNull ISerializer serializer,
+      final @NotNull ILogger logger,
+      final @Nullable JsonSerializable serializable) {
+    if (serializable == null) {
+      return 0;
+    }
+    try {
+      final ByteCountingWriter writer = new ByteCountingWriter();
+      serializer.serialize(serializable, writer);
+      return writer.getByteCount();
+    } catch (Throwable t) {
+      logger.log(SentryLevel.ERROR, "Could not calculate size of serializable", t);
+      return 0;
+    }
+  }
+
+  /**
+   * A Writer that counts the number of bytes that would be written in UTF-8 encoding without
+   * actually storing the data.
+   */
+  private static final class ByteCountingWriter extends Writer {
+    private long byteCount = 0L;
+
+    @Override
+    public void write(final char[] cbuf, final int off, final int len) {
+      for (int i = off; i < off + len; i++) {
+        byteCount += utf8ByteCount(cbuf[i]);
+      }
+    }
+
+    @Override
+    public void write(final int c) {
+      byteCount += utf8ByteCount((char) c);
+    }
+
+    @Override
+    public void write(final @NotNull String str, final int off, final int len) {
+      for (int i = off; i < off + len; i++) {
+        byteCount += utf8ByteCount(str.charAt(i));
+      }
+    }
+
+    @Override
+    public void flush() {
+      // Nothing to flush since we don't store data
+    }
+
+    @Override
+    public void close() {
+      // Nothing to close
+    }
+
+    public long getByteCount() {
+      return byteCount;
+    }
+
+    /**
+     * Calculates the number of bytes needed to encode a character in UTF-8.
+     *
+     * @param c the character
+     * @return the number of bytes (1-4)
+     */
+    private static int utf8ByteCount(final char c) {
+      if (c <= 0x7F) {
+        return 1; // ASCII
+      } else if (c <= 0x7FF) {
+        return 2; // 2-byte character
+      } else if (Character.isSurrogate(c)) {
+        return 2; // Surrogate pair, counted as 2 bytes each (total 4 for the pair)
+      } else {
+        return 3; // 3-byte character
+      }
+    }
+  }
 }

sentry/src/test/java/io/sentry/SentryClientTest.kt

@@ -293,7 +293,10 @@ class SentryClientTest {
 
     assertClientReport(
       fixture.sentryOptions.clientReportRecorder,
-      listOf(DiscardedEvent(DiscardReason.BEFORE_SEND.reason, DataCategory.LogItem.category, 1)),
+      listOf(
+        DiscardedEvent(DiscardReason.BEFORE_SEND.reason, DataCategory.LogItem.category, 1),
+        DiscardedEvent(DiscardReason.BEFORE_SEND.reason, DataCategory.LogByte.category, 109),
+      ),
     )
   }
 
@@ -312,7 +315,10 @@ class SentryClientTest {
 
     assertClientReport(
       fixture.sentryOptions.clientReportRecorder,
-      listOf(DiscardedEvent(DiscardReason.BEFORE_SEND.reason, DataCategory.LogItem.category, 1)),
+      listOf(
+        DiscardedEvent(DiscardReason.BEFORE_SEND.reason, DataCategory.LogItem.category, 1),
+        DiscardedEvent(DiscardReason.BEFORE_SEND.reason, DataCategory.LogByte.category, 109),
+      ),
     )
   }
 

sentry/src/test/java/io/sentry/clientreport/ClientReportTest.kt

@@ -377,6 +377,9 @@ class ClientReportTest {
     val logItem =
       clientReport!!.discardedEvents!!.first { it.category == DataCategory.LogItem.category }
     assertEquals(2, logItem.quantity)
+    val logByte =
+      clientReport!!.discardedEvents!!.first { it.category == DataCategory.LogByte.category }
+    assertEquals(226, logByte.quantity)
   }
 
   private fun givenClientReportRecorder(

sentry/src/test/java/io/sentry/util/JsonSerializationUtilsTest.kt

@@ -3,18 +3,28 @@ package io.sentry.util
 import io.sentry.ILogger
 import io.sentry.JsonSerializable
 import io.sentry.JsonSerializer
+import io.sentry.ObjectWriter
+import io.sentry.SentryLogEvent
+import io.sentry.SentryLogLevel
+import io.sentry.SentryOptions
+import io.sentry.protocol.SentryId
 import java.io.Writer
 import java.util.Calendar
 import java.util.concurrent.atomic.AtomicIntegerArray
 import kotlin.test.Test
 import kotlin.test.assertContentEquals
 import kotlin.test.assertEquals
 import kotlin.test.assertNull
+import kotlin.test.assertTrue
 import org.mockito.invocation.InvocationOnMock
 import org.mockito.kotlin.any
 import org.mockito.kotlin.mock
 
 class JsonSerializationUtilsTest {
+
+  private val serializer = JsonSerializer(SentryOptions())
+  private val logger: ILogger = mock()
+
   @Test
   fun `serializes calendar to map`() {
     val calendar = Calendar.getInstance()
@@ -74,4 +84,57 @@ class JsonSerializationUtilsTest {
 
     assertNull(actualBytes, "Mocker error should be captured and null returned.")
   }
+
+  @Test
+  fun `byteSizeOf returns same size as bytesFrom for ASCII`() {
+    val logEvent = SentryLogEvent(SentryId(), 1234567890.0, "Hello ASCII", SentryLogLevel.INFO)
+
+    val actualBytes = JsonSerializationUtils.bytesFrom(serializer, logger, logEvent)
+    val byteSize = JsonSerializationUtils.byteSizeOf(serializer, logger, logEvent)
+
+    assertEquals(
+      (actualBytes?.size ?: -1).toLong(),
+      byteSize,
+      "byteSizeOf should match actual byte array length",
+    )
+    assertTrue(byteSize > 0, "byteSize should be positive")
+  }
+
+  @Test
+  fun `byteSizeOf returns same size as bytesFrom for UTF-8 characters`() {
+    // Mix of 1-byte, 2-byte, 3-byte and 4-byte UTF-8 characters
+    val logEvent =
+      SentryLogEvent(SentryId(), 1234567890.0, "Hello 世界 café 🎉 🚀", SentryLogLevel.WARN)
+
+    val actualBytes = JsonSerializationUtils.bytesFrom(serializer, logger, logEvent)
+    val byteSize = JsonSerializationUtils.byteSizeOf(serializer, logger, logEvent)
+
+    assertEquals(
+      (actualBytes?.size ?: -1).toLong(),
+      byteSize,
+      "byteSizeOf should match actual byte array length for UTF-8",
+    )
+    assertTrue(byteSize > 0, "byteSize should be positive")
+  }
+
+  @Test
+  fun `byteSizeOf returns 0 on serialization error`() {
+    val serializable =
+      object : JsonSerializable {
+        override fun serialize(writer: ObjectWriter, logger: ILogger) {
+          throw RuntimeException("Serialization error")
+        }
+      }
+
+    val byteSize = JsonSerializationUtils.byteSizeOf(serializer, logger, serializable)
+
+    assertEquals(0, byteSize, "byteSizeOf should return 0 on error")
+  }
+
+  @Test
+  fun `byteSizeOf returns 0 on null serializable`() {
+    val byteSize = JsonSerializationUtils.byteSizeOf(serializer, logger, null)
+
+    assertEquals(0, byteSize, "byteSizeOf should return 0 on null serializable")
+  }
 }