Gigantic descriptors shouldn't overflow the Java string literal size limit.

3 files changed, 19 insertions, 2 deletions

diff --git a/java/pom.xml b/java/pom.xml
index 13208104..91874e79 100644
--- a/java/pom.xml
+++ b/java/pom.xml
@@ -112,6 +112,7 @@
                   <arg value="../src/google/protobuf/unittest_lite.proto" />
                   <arg value="../src/google/protobuf/unittest_import_lite.proto" />
                   <arg value="../src/google/protobuf/unittest_lite_imports_nonlite.proto" />
+                  <arg value="../src/google/protobuf/unittest_enormous_descriptor.proto" />
                 </exec>
               </tasks>
               <testSourceRoot>target/generated-test-sources</testSourceRoot>
diff --git a/java/src/main/java/com/google/protobuf/Descriptors.java b/java/src/main/java/com/google/protobuf/Descriptors.java
index df934bb4..0c162d5d 100644
--- a/java/src/main/java/com/google/protobuf/Descriptors.java
+++ b/java/src/main/java/com/google/protobuf/Descriptors.java
@@ -244,7 +244,8 @@ public final class Descriptors {
      * encoded in protocol buffer wire format.
      */
     public static void internalBuildGeneratedFileFrom(
-        final String descriptorData, final FileDescriptor[] dependencies,
+        final String[] descriptorDataParts,
+        final FileDescriptor[] dependencies,
         final InternalDescriptorAssigner descriptorAssigner) {
       // Hack:  We can't embed a raw byte array inside generated Java code
       //   (at least, not efficiently), but we can embed Strings.  So, the
@@ -255,9 +256,16 @@ public final class Descriptors {
       //   serialized form.  So, if we convert it to bytes in ISO-8859-1, we
       //   should get the original bytes that we want.
 
+      // descriptorData may contain multiple strings in order to get around the
+      // Java 64k string literal limit.
+      StringBuilder descriptorData = new StringBuilder();
+      for (String part : descriptorDataParts) {
+        descriptorData.append(part);
+      }
+
       final byte[] descriptorBytes;
       try {
-        descriptorBytes = descriptorData.getBytes("ISO-8859-1");
+        descriptorBytes = descriptorData.toString().getBytes("ISO-8859-1");
       } catch (UnsupportedEncodingException e) {
         throw new RuntimeException(
           "Standard encoding ISO-8859-1 not supported by JVM.", e);
diff --git a/java/src/test/java/com/google/protobuf/DescriptorsTest.java b/java/src/test/java/com/google/protobuf/DescriptorsTest.java
index 6fa56267..c5c38b27 100644
--- a/java/src/test/java/com/google/protobuf/DescriptorsTest.java
+++ b/java/src/test/java/com/google/protobuf/DescriptorsTest.java
@@ -397,4 +397,12 @@ public class DescriptorsTest extends TestCase {
       assertEquals(values1[i].toString(), values2[i].toString());
     }
   }
+
+  public void testEnormousDescriptor() throws Exception {
+    // The descriptor for this file is larger than 64k, yet it did not cause
+    // a compiler error due to an over-long string literal.
+    assertTrue(
+        UnittestEnormousDescriptor.getDescriptor()
+          .toProto().getSerializedSize() > 65536);
+  }
 }