Fix repeated string serialization for JSON. (#6888)

Co-authored-by: Jack Berg <[email protected]>

Author: jsuereth

exporters/common/src/main/java/io/opentelemetry/exporter/internal/marshal/JsonSerializer.java

@@ -122,6 +122,22 @@ public void writeString(
     generator.writeString(string);
   }
 
+  @Override
+  public void writeRepeatedString(ProtoFieldInfo field, byte[][] utf8Bytes) throws IOException {
+    generator.writeArrayFieldStart(field.getJsonName());
+    for (byte[] value : utf8Bytes) {
+      // Marshalers encoded String into UTF-8 bytes to optimize for binary serialization where
+      // we are able to avoid the encoding process happening twice, one for size computation and one
+      // for actual writing. JsonGenerator actually has a writeUTF8String that would be able to
+      // accept
+      // this, but it only works when writing to an OutputStream, but not to a String like we do for
+      // writing to logs. It's wasteful to take a String, convert it to bytes, and convert back to
+      // the same String but we can see if this can be improved in the future.
+      generator.writeString(new String(value, StandardCharsets.UTF_8));
+    }
+    generator.writeEndArray();
+  }
+
   @Override
   public void writeBytes(ProtoFieldInfo field, byte[] value) throws IOException {
     generator.writeBinaryField(field.getJsonName(), value);

exporters/common/src/main/java/io/opentelemetry/exporter/internal/marshal/MarshalerUtil.java

@@ -110,6 +110,16 @@ private static int sizeRepeatedFixed64(ProtoFieldInfo field, int numValues) {
     return size;
   }
 
+  /** Returns the size of a repeated string field. */
+  @SuppressWarnings("AvoidObjectArrays")
+  public static int sizeRepeatedString(ProtoFieldInfo field, byte[][] utf8Bytes) {
+    int size = 0;
+    for (byte[] i : utf8Bytes) {
+      size += MarshalerUtil.sizeBytes(field, i);
+    }
+    return size;
+  }
+
   /**
    * Returns the size of a repeated uint64 field.
    *

exporters/common/src/main/java/io/opentelemetry/exporter/internal/marshal/ProtoSerializer.java

@@ -163,6 +163,13 @@ public void writeString(
     StatelessMarshalerUtil.writeUtf8(output, string, utf8Length, context);
   }
 
+  @Override
+  public void writeRepeatedString(ProtoFieldInfo field, byte[][] utf8Bytes) throws IOException {
+    for (byte[] value : utf8Bytes) {
+      writeString(field, value);
+    }
+  }
+
   @Override
   public void writeBytes(ProtoFieldInfo field, byte[] value) throws IOException {
     output.writeUInt32NoTag(field.getTag());

exporters/common/src/main/java/io/opentelemetry/exporter/internal/marshal/Serializer.java

@@ -220,6 +220,18 @@ public void serializeString(ProtoFieldInfo field, byte[] utf8Bytes) throws IOExc
     writeString(field, utf8Bytes);
   }
 
+  /**
+   * Serializes a protobuf {@code repeated string} field. {@code utf8Bytes} is the UTF8 encoded
+   * bytes of the strings to serialize.
+   */
+  @SuppressWarnings("AvoidObjectArrays")
+  public void serializeRepeatedString(ProtoFieldInfo field, byte[][] utf8Bytes) throws IOException {
+    if (utf8Bytes.length == 0) {
+      return;
+    }
+    writeRepeatedString(field, utf8Bytes);
+  }
+
   /**
    * Serializes a protobuf {@code string} field. {@code string} is the value to be serialized and
    * {@code utf8Length} is the length of the string after it is encoded in UTF8. This method reads
@@ -246,6 +258,11 @@ public abstract void writeString(
       ProtoFieldInfo field, String string, int utf8Length, MarshalerContext context)
       throws IOException;
 
+  /** Writes a protobuf {@code repeated string} field, even if it matches the default value. */
+  @SuppressWarnings("AvoidObjectArrays")
+  public abstract void writeRepeatedString(ProtoFieldInfo field, byte[][] utf8Bytes)
+      throws IOException;
+
   /** Serializes a protobuf {@code bytes} field. */
   public void serializeBytes(ProtoFieldInfo field, byte[] value) throws IOException {
     if (value.length == 0) {

exporters/otlp/profiles/src/main/java/io/opentelemetry/exporter/otlp/profiles/ProfileMarshaler.java

@@ -149,9 +149,7 @@ protected void writeTo(Serializer output) throws IOException {
     output.serializeRepeatedMessage(Profile.ATTRIBUTE_TABLE, attributeMarshalers);
     output.serializeRepeatedMessage(Profile.ATTRIBUTE_UNITS, attributeUnitMarshalers);
     output.serializeRepeatedMessage(Profile.LINK_TABLE, linkMarshalers);
-    for (byte[] i : stringTable) {
-      output.serializeString(Profile.STRING_TABLE, i);
-    }
+    output.serializeRepeatedString(Profile.STRING_TABLE, stringTable);
     output.serializeInt64(Profile.DROP_FRAMES, dropFrames);
     output.serializeInt64(Profile.KEEP_FRAMES, keepFrames);
     output.serializeInt64(Profile.TIME_NANOS, timeNanos);
@@ -192,9 +190,7 @@ private static int calculateSize(
     size += MarshalerUtil.sizeRepeatedMessage(Profile.ATTRIBUTE_TABLE, attributeMarshalers);
     size += MarshalerUtil.sizeRepeatedMessage(Profile.ATTRIBUTE_UNITS, attributeUnitMarshalers);
     size += MarshalerUtil.sizeRepeatedMessage(Profile.LINK_TABLE, linkMarshalers);
-    for (byte[] i : stringTable) {
-      size += MarshalerUtil.sizeBytes(Profile.STRING_TABLE, i);
-    }
+    size += MarshalerUtil.sizeRepeatedString(Profile.STRING_TABLE, stringTable);
     size += MarshalerUtil.sizeInt64(Profile.DROP_FRAMES, dropFrames);
     size += MarshalerUtil.sizeInt64(Profile.KEEP_FRAMES, keepFrames);
     size += MarshalerUtil.sizeInt64(Profile.TIME_NANOS, timeNanos);

exporters/otlp/profiles/src/test/java/io/opentelemetry/exporter/otlp/profiles/ProfilesRequestMarshalerTest.java

@@ -275,7 +275,7 @@ private static ProfileData sampleProfileData() {
         Attributes.empty(),
         Collections.emptyList(),
         Collections.emptyList(),
-        Collections.emptyList(),
+        listOf("foo", "bar"),
         3,
         4,
         5,
@@ -304,6 +304,8 @@ private static Profile.Builder sampleProfileBuilder() {
                         .AGGREGATION_TEMPORALITY_CUMULATIVE)
                 .build())
         .addAllComment(listOf(8L, 9L))
+        .addStringTable("foo")
+        .addStringTable("bar")
         .setDefaultSampleType(10);
   }