RocksDB#keyMayExist should not assume database values are unicode strings (#6186)

Summary:
Closes https://github.com/facebook/rocksdb/issues/6183
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6186

Differential Revision: D19201281

Pulled By: pdillinger

fbshipit-source-id: 1c96b4ea09e826f91e44b0009eba3de0991d9053

HISTORY.md

@@ -3,6 +3,7 @@
 ### Public API Change
 * Added a rocksdb::FileSystem class in include/rocksdb/file_system.h to encapsulate file creation/read/write operations, and an option DBOptions::file_system to allow a user to pass in an instance of rocksdb::FileSystem. If its a non-null value, this will take precendence over DBOptions::env for file operations. A new API rocksdb::FileSystem::Default() returns a platform default object. The DBOptions::env option and Env::Default() API will continue to be used for threading and other OS related functions, and where DBOptions::file_system is not specified, for file operations. For storage developers who are accustomed to rocksdb::Env, the interface in rocksdb::FileSystem is new and will probably undergo some changes as more storage systems are ported to it from rocksdb::Env. As of now, no env other than Posix has been ported to the new interface.
 * A new rocksdb::NewSstFileManager() API that allows the caller to pass in separate Env and FileSystem objects.
+* Changed Java API for RocksDB.keyMayExist functions to use Holder<byte[]> instead of StringBuilder, so that retrieved values need not decode to Strings.
 
 ### Bug Fixes
 * Fix a bug that can cause unnecessary bg thread to be scheduled(#6104).

java/CMakeLists.txt

@@ -144,6 +144,7 @@ set(JAVA_MAIN_CLASSES
   src/main/java/org/rocksdb/HdfsEnv.java
   src/main/java/org/rocksdb/HistogramData.java
   src/main/java/org/rocksdb/HistogramType.java
+  src/main/java/org/rocksdb/Holder.java
   src/main/java/org/rocksdb/IndexType.java
   src/main/java/org/rocksdb/InfoLogLevel.java
   src/main/java/org/rocksdb/IngestExternalFileOptions.java

java/rocksjni/rocksjni.cc

@@ -1588,127 +1588,156 @@ jobjectArray Java_org_rocksdb_RocksDB_multiGet__JJ_3_3B_3I_3I_3J(
 
 //////////////////////////////////////////////////////////////////////////////
 // rocksdb::DB::KeyMayExist
-jboolean key_may_exist_helper(JNIEnv* env, rocksdb::DB* db,
+bool key_may_exist_helper(JNIEnv* env, jlong jdb_handle, jlong jcf_handle,
-      const rocksdb::ReadOptions& read_opt,
+  jlong jread_opts_handle,
-      rocksdb::ColumnFamilyHandle* cf_handle,
+  jbyteArray jkey, jint jkey_offset, jint jkey_len,
-      jbyteArray jkey, jint jkey_off, jint jkey_len,
+  bool* has_exception, std::string* value, bool* value_found) {
-      jobject jstring_builder, bool* has_exception) {
+ auto* db = reinterpret_cast<rocksdb::DB*>(jdb_handle);
+ rocksdb::ColumnFamilyHandle* cf_handle;
+ if (jcf_handle == 0) {
+   cf_handle = db->DefaultColumnFamily();
+ } else {
+   cf_handle = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(jcf_handle);
+ }
+ rocksdb::ReadOptions read_opts =
+    jread_opts_handle == 0 ? rocksdb::ReadOptions()
+        : *(reinterpret_cast<rocksdb::ReadOptions*>(jread_opts_handle));
+
   jbyte* key = new jbyte[jkey_len];
-  env->GetByteArrayRegion(jkey, jkey_off, jkey_len, key);
+  env->GetByteArrayRegion(jkey, jkey_offset, jkey_len, key);
   if (env->ExceptionCheck()) {
     // exception thrown: ArrayIndexOutOfBoundsException
     delete[] key;
     *has_exception = true;
     return false;
   }
-
   rocksdb::Slice key_slice(reinterpret_cast<char*>(key), jkey_len);
 
-  std::string value;
+  const bool exists = db->KeyMayExist(
-  bool value_found = false;
+      read_opts, cf_handle, key_slice, value, value_found);
-  bool keyMayExist;
-  if (cf_handle != nullptr) {
-    keyMayExist =
-        db->KeyMayExist(read_opt, cf_handle, key_slice, &value, &value_found);
-  } else {
-    keyMayExist = db->KeyMayExist(read_opt, key_slice, &value, &value_found);
-  }
 
   // cleanup
   delete[] key;
 
-  // extract the value
+  return exists;
-  if (value_found && !value.empty()) {
+}
-    jobject jresult_string_builder =
+
-        rocksdb::StringBuilderJni::append(env, jstring_builder, value.c_str());
+
-    if (jresult_string_builder == nullptr) {
+/*
-      *has_exception = true;
+ * Class:     org_rocksdb_RocksDB
-      return false;
+ * Method:    keyMayExist
+ * Signature: (JJJ[BII)Z
+ */
+jboolean Java_org_rocksdb_RocksDB_keyMayExist(
+    JNIEnv* env, jobject, jlong jdb_handle, jlong jcf_handle,
+    jlong jread_opts_handle,
+    jbyteArray jkey, jint jkey_offset, jint jkey_len) {
+
+  bool has_exception = false;
+  std::string value;
+  bool value_found = false;
+
+  const bool exists = key_may_exist_helper(
+      env, jdb_handle, jcf_handle, jread_opts_handle,
+      jkey, jkey_offset, jkey_len,
+      &has_exception, &value, &value_found);
+
+  if (has_exception) {
+    // java exception already raised
+    return false;
+  }
+
+  return static_cast<jboolean>(exists);
+}
+
+/*
+ * Class:     org_rocksdb_RocksDB
+ * Method:    keyMayExistFoundValue
+ * Signature: (JJJ[BII)[[B
+ */
+jobjectArray Java_org_rocksdb_RocksDB_keyMayExistFoundValue(
+    JNIEnv* env, jobject, jlong jdb_handle, jlong jcf_handle,
+    jlong jread_opts_handle,
+    jbyteArray jkey, jint jkey_offset, jint jkey_len) {
+
+  bool has_exception = false;
+  std::string value;
+  bool value_found = false;
+
+  const bool exists = key_may_exist_helper(
+      env, jdb_handle, jcf_handle, jread_opts_handle,
+      jkey, jkey_offset, jkey_len,
+      &has_exception, &value, &value_found);
+
+  if (has_exception) {
+    // java exception already raised
+    return nullptr;
+  }
+
+  jbyte result_flags[1];
+  if (!exists) {
+    result_flags[0] = 0;
+  } else if (!value_found) {
+    result_flags[0] = 1;
+  } else {
+    // found
+    result_flags[0] = 2;
+  }
+
+  jobjectArray jresults = rocksdb::ByteJni::new2dByteArray(env, 2);
+  if (jresults == nullptr) {
+    // exception occurred
+    return nullptr;
+  }
+
+  // prepare the result flag
+  jbyteArray jresult_flags = env->NewByteArray(1);
+  if (jresult_flags == nullptr) {
+    // exception thrown: OutOfMemoryError
+    return nullptr;
+  }
+  env->SetByteArrayRegion(jresult_flags, 0, 1, result_flags);
+  if (env->ExceptionCheck()) {
+    // exception thrown: ArrayIndexOutOfBoundsException
+    env->DeleteLocalRef(jresult_flags);
+    return nullptr;
+  }
+  
+  env->SetObjectArrayElement(jresults, 0, jresult_flags);
+  if (env->ExceptionCheck()) {
+      // exception thrown: ArrayIndexOutOfBoundsException
+      env->DeleteLocalRef(jresult_flags);
+      return nullptr;
+  }
+
+  env->DeleteLocalRef(jresult_flags);
+
+  if (result_flags[0] == 2) {
+    // set the value
+    const jsize jvalue_len = static_cast<jsize>(value.size());
+    jbyteArray jresult_value = env->NewByteArray(jvalue_len);
+    if (jresult_value == nullptr) {
+      // exception thrown: OutOfMemoryError
+      return nullptr;
     }
+    env->SetByteArrayRegion(jresult_value, 0, jvalue_len, 
+        const_cast<jbyte*>(reinterpret_cast<const jbyte*>(value.data())));
+    if (env->ExceptionCheck()) {
+      // exception thrown: ArrayIndexOutOfBoundsException
+      env->DeleteLocalRef(jresult_value);
+      return nullptr;
+    }
+    env->SetObjectArrayElement(jresults, 1, jresult_value);
+    if (env->ExceptionCheck()) {
+      // exception thrown: ArrayIndexOutOfBoundsException
+      env->DeleteLocalRef(jresult_value);
+      return nullptr;
+    }
+
+    env->DeleteLocalRef(jresult_value);
   }
 
-  *has_exception = false;
+  return jresults;
-  return static_cast<jboolean>(keyMayExist);
-}
-
-/*
- * Class:     org_rocksdb_RocksDB
- * Method:    keyMayExist
- * Signature: (J[BIILjava/lang/StringBuilder;)Z
- */
-jboolean Java_org_rocksdb_RocksDB_keyMayExist__J_3BIILjava_lang_StringBuilder_2(
-    JNIEnv* env, jobject, jlong jdb_handle,
-    jbyteArray jkey, jint jkey_off, jint jkey_len, jobject jstring_builder) {
-  auto* db = reinterpret_cast<rocksdb::DB*>(jdb_handle);
-  bool has_exception = false;
-  return key_may_exist_helper(env, db, rocksdb::ReadOptions(), nullptr, jkey,
-      jkey_off, jkey_len, jstring_builder, &has_exception);
-}
-
-/*
- * Class:     org_rocksdb_RocksDB
- * Method:    keyMayExist
- * Signature: (J[BIIJLjava/lang/StringBuilder;)Z
- */
-jboolean
-Java_org_rocksdb_RocksDB_keyMayExist__J_3BIIJLjava_lang_StringBuilder_2(
-    JNIEnv* env, jobject, jlong jdb_handle,
-    jbyteArray jkey, jint jkey_off, jint jkey_len,
-    jlong jcf_handle, jobject jstring_builder) {
-  auto* db = reinterpret_cast<rocksdb::DB*>(jdb_handle);
-  auto* cf_handle = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(jcf_handle);
-  if (cf_handle != nullptr) {
-    bool has_exception = false;
-    return key_may_exist_helper(env, db, rocksdb::ReadOptions(), cf_handle,
-        jkey, jkey_off, jkey_len, jstring_builder,
-        &has_exception);
-  } else {
-    rocksdb::RocksDBExceptionJni::ThrowNew(
-        env, rocksdb::Status::InvalidArgument("Invalid ColumnFamilyHandle."));
-    return true;
-  }
-}
-
-/*
- * Class:     org_rocksdb_RocksDB
- * Method:    keyMayExist
- * Signature: (JJ[BIILjava/lang/StringBuilder;)Z
- */
-jboolean
-Java_org_rocksdb_RocksDB_keyMayExist__JJ_3BIILjava_lang_StringBuilder_2(
-    JNIEnv* env, jobject, jlong jdb_handle, jlong jread_options_handle,
-    jbyteArray jkey, jint jkey_off, jint jkey_len, jobject jstring_builder) {
-  auto* db = reinterpret_cast<rocksdb::DB*>(jdb_handle);
-  auto& read_options =
-      *reinterpret_cast<rocksdb::ReadOptions*>(jread_options_handle);
-  bool has_exception = false;
-  return key_may_exist_helper(env, db, read_options, nullptr, jkey, jkey_off,
-      jkey_len, jstring_builder, &has_exception);
-}
-
-/*
- * Class:     org_rocksdb_RocksDB
- * Method:    keyMayExist
- * Signature: (JJ[BIIJLjava/lang/StringBuilder;)Z
- */
-jboolean
-Java_org_rocksdb_RocksDB_keyMayExist__JJ_3BIIJLjava_lang_StringBuilder_2(
-    JNIEnv* env, jobject, jlong jdb_handle, jlong jread_options_handle,
-    jbyteArray jkey, jint jkey_off, jint jkey_len, jlong jcf_handle,
-    jobject jstring_builder) {
-  auto* db = reinterpret_cast<rocksdb::DB*>(jdb_handle);
-  auto& read_options =
-      *reinterpret_cast<rocksdb::ReadOptions*>(jread_options_handle);
-  auto* cf_handle = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(jcf_handle);
-  if (cf_handle != nullptr) {
-    bool has_exception = false;
-    return key_may_exist_helper(env, db, read_options, cf_handle, jkey,
-        jkey_off, jkey_len, jstring_builder, &has_exception);
-  } else {
-    rocksdb::RocksDBExceptionJni::ThrowNew(
-        env, rocksdb::Status::InvalidArgument("Invalid ColumnFamilyHandle."));
-    return true;
-  }
 }
 
 /*

java/src/main/java/org/rocksdb/Holder.java

@@ -0,0 +1,46 @@
+// Copyright (c) 2016, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+package org.rocksdb;
+
+/**
+ * Simple instance reference wrapper.
+ */
+public class Holder<T> {
+  private /* @Nullable */ T value;
+
+  /**
+   * Constructs a new Holder with null instance.
+   */
+  public Holder() {
+  }
+
+  /**
+   * Constructs a new Holder.
+   *
+   * @param value the instance or null
+   */
+  public Holder(/* @Nullable */ final T value) {
+    this.value = value;
+  }
+
+  /**
+   * Get the instance reference.
+   *
+   * @return value the instance reference or null
+   */
+  public /* @Nullable */ T getValue() {
+    return value;
+  }
+
+  /**
+   * Set the instance reference.
+   *
+   * @param value the instance reference or null
+   */
+  public void setValue(/* @Nullable */ final T value) {
+    this.value = value;
+  }
+}

java/src/main/java/org/rocksdb/RocksDB.java

@@ -2189,68 +2189,94 @@ public class RocksDB extends RocksObject {
 
   /**
    * If the key definitely does not exist in the database, then this method
-   * returns false, else true.
+   * returns null, else it returns an instance of KeyMayExistResult
    *
-   * This check is potentially lighter-weight than invoking DB::Get(). One way
+   * If the caller wants to obtain value when the key
-   * to make this lighter weight is to avoid doing any IOs.
+   * is found in memory, then {@code valueHolder} must be set.
+   *
+   * This check is potentially lighter-weight than invoking
+   * {@link #get(byte[])}. One way to make this lighter weight is to avoid
+   * doing any IOs.
    *
    * @param key byte array of a key to search for
-   * @param value StringBuilder instance which is a out parameter if a value is
+   * @param valueHolder non-null to retrieve the value if it is found, or null
-   *    found in block-cache.
+   *     if the value is not needed. If non-null, upon return of the function,
-   * @return boolean value indicating if key does not exist or might exist.
+   *     the {@code value} will be set if it could be retrieved.
+   *
+   * @return false if the key definitely does not exist in the database,
+   *     otherwise true.
    */
-  public boolean keyMayExist(final byte[] key, final StringBuilder value) {
+  public boolean keyMayExist(final byte[] key,
-    return keyMayExist(nativeHandle_, key, 0, key.length, value);
+      /* @Nullable */ final Holder<byte[]> valueHolder) {
+    return keyMayExist(key, 0, key.length, valueHolder);
   }
 
   /**
    * If the key definitely does not exist in the database, then this method
-   * returns false, else true.
+   * returns null, else it returns an instance of KeyMayExistResult
    *
-   * This check is potentially lighter-weight than invoking DB::Get(). One way
+   * If the caller wants to obtain value when the key
-   * to make this lighter weight is to avoid doing any IOs.
+   * is found in memory, then {@code valueHolder} must be set.
+   *
+   * This check is potentially lighter-weight than invoking
+   * {@link #get(byte[], int, int)}. One way to make this lighter weight is to
+   * avoid doing any IOs.
    *
    * @param key byte array of a key to search for
    * @param offset the offset of the "key" array to be used, must be
    *     non-negative and no larger than "key".length
    * @param len the length of the "key" array to be used, must be non-negative
    *     and no larger than "key".length
-   * @param value StringBuilder instance which is a out parameter if a value is
+   * @param valueHolder non-null to retrieve the value if it is found, or null
-   *    found in block-cache.
+   *     if the value is not needed. If non-null, upon return of the function,
+   *     the {@code value} will be set if it could be retrieved.
    *
-   * @return boolean value indicating if key does not exist or might exist.
+   * @return false if the key definitely does not exist in the database,
+   *     otherwise true.
    */
-  public boolean keyMayExist(final byte[] key, final int offset, final int len,
+  public boolean keyMayExist(final byte[] key,
-      final StringBuilder value) {
+      final int offset, final int len,
-    checkBounds(offset, len, key.length);
+      /* @Nullable */ final Holder<byte[]> valueHolder) {
-    return keyMayExist(nativeHandle_, key, offset, len, value);
+    return keyMayExist((ColumnFamilyHandle)null, key, offset, len, valueHolder);
   }
 
   /**
    * If the key definitely does not exist in the database, then this method
-   * returns false, else true.
+   * returns null, else it returns an instance of KeyMayExistResult
    *
-   * This check is potentially lighter-weight than invoking DB::Get(). One way
+   * If the caller wants to obtain value when the key
-   * to make this lighter weight is to avoid doing any IOs.
+   * is found in memory, then {@code valueHolder} must be set.
+   *
+   * This check is potentially lighter-weight than invoking
+   * {@link #get(ColumnFamilyHandle,byte[])}. One way to make this lighter
+   * weight is to avoid doing any IOs.
    *
    * @param columnFamilyHandle {@link ColumnFamilyHandle} instance
    * @param key byte array of a key to search for
-   * @param value StringBuilder instance which is a out parameter if a value is
+   * @param valueHolder non-null to retrieve the value if it is found, or null
-   *    found in block-cache.
+   *     if the value is not needed. If non-null, upon return of the function,
-   * @return boolean value indicating if key does not exist or might exist.
+   *     the {@code value} will be set if it could be retrieved.
+   *
+   * @return false if the key definitely does not exist in the database,
+   *     otherwise true.
    */
-  public boolean keyMayExist(final ColumnFamilyHandle columnFamilyHandle,
+  public boolean keyMayExist(
-      final byte[] key, final StringBuilder value) {
+      final ColumnFamilyHandle columnFamilyHandle, final byte[] key,
-    return keyMayExist(nativeHandle_, key, 0, key.length,
+      /* @Nullable */ final Holder<byte[]> valueHolder) {
-        columnFamilyHandle.nativeHandle_, value);
+    return keyMayExist(columnFamilyHandle, key, 0, key.length,
+        valueHolder);
   }
 
   /**
    * If the key definitely does not exist in the database, then this method
-   * returns false, else true.
+   * returns null, else it returns an instance of KeyMayExistResult
    *
-   * This check is potentially lighter-weight than invoking DB::Get(). One way
+   * If the caller wants to obtain value when the key
-   * to make this lighter weight is to avoid doing any IOs.
+   * is found in memory, then {@code valueHolder} must be set.
+   *
+   * This check is potentially lighter-weight than invoking
+   * {@link #get(ColumnFamilyHandle, byte[], int, int)}. One way to make this
+   * lighter weight is to avoid doing any IOs.
    *
    * @param columnFamilyHandle {@link ColumnFamilyHandle} instance
    * @param key byte array of a key to search for
@@ -2258,42 +2284,58 @@ public class RocksDB extends RocksObject {
    *    non-negative and no larger than "key".length
    * @param len the length of the "key" array to be used, must be non-negative
    *    and no larger than "key".length
-   * @param value StringBuilder instance which is a out parameter if a value is
+   * @param valueHolder non-null to retrieve the value if it is found, or null
-   *    found in block-cache.
+   *     if the value is not needed. If non-null, upon return of the function,
-   * @return boolean value indicating if key does not exist or might exist.
+   *     the {@code value} will be set if it could be retrieved.
+   *
+   * @return false if the key definitely does not exist in the database,
+   *     otherwise true.
    */
-  public boolean keyMayExist(final ColumnFamilyHandle columnFamilyHandle,
+  public boolean keyMayExist(
-      final byte[] key, int offset, int len, final StringBuilder value) {
+      final ColumnFamilyHandle columnFamilyHandle,
-    checkBounds(offset, len, key.length);
+      final byte[] key, int offset, int len,
-    return keyMayExist(nativeHandle_, key, offset, len,
+      /* @Nullable */ final Holder<byte[]> valueHolder) {
-        columnFamilyHandle.nativeHandle_, value);
+    return keyMayExist(columnFamilyHandle, null, key, offset, len,
+        valueHolder);
   }
 
   /**
    * If the key definitely does not exist in the database, then this method
-   * returns false, else true.
+   * returns null, else it returns an instance of KeyMayExistResult
    *
-   * This check is potentially lighter-weight than invoking DB::Get(). One way
+   * If the caller wants to obtain value when the key
-   * to make this lighter weight is to avoid doing any IOs.
+   * is found in memory, then {@code valueHolder} must be set.
+   *
+   * This check is potentially lighter-weight than invoking
+   * {@link #get(ReadOptions, byte[])}. One way to make this
+   * lighter weight is to avoid doing any IOs.
    *
    * @param readOptions {@link ReadOptions} instance
    * @param key byte array of a key to search for
-   * @param value StringBuilder instance which is a out parameter if a value is
+   * @param valueHolder non-null to retrieve the value if it is found, or null
-   *    found in block-cache.
+   *     if the value is not needed. If non-null, upon return of the function,
-   * @return boolean value indicating if key does not exist or might exist.
+   *     the {@code value} will be set if it could be retrieved.
+   *
+   * @return false if the key definitely does not exist in the database,
+   *     otherwise true.
    */
-  public boolean keyMayExist(final ReadOptions readOptions,
+  public boolean keyMayExist(
-      final byte[] key, final StringBuilder value) {
+      final ReadOptions readOptions, final byte[] key,
-    return keyMayExist(nativeHandle_, readOptions.nativeHandle_,
+      /* @Nullable */ final Holder<byte[]> valueHolder) {
-        key, 0, key.length, value);
+    return keyMayExist(readOptions, key, 0, key.length,
+        valueHolder);
   }
 
   /**
    * If the key definitely does not exist in the database, then this method
-   * returns false, else true.
+   * returns null, else it returns an instance of KeyMayExistResult
    *
-   * This check is potentially lighter-weight than invoking DB::Get(). One way
+   * If the caller wants to obtain value when the key
-   * to make this lighter weight is to avoid doing any IOs.
+   * is found in memory, then {@code valueHolder} must be set.
+   *
+   * This check is potentially lighter-weight than invoking
+   * {@link #get(ReadOptions, byte[], int, int)}. One way to make this
+   * lighter weight is to avoid doing any IOs.
    *
    * @param readOptions {@link ReadOptions} instance
    * @param key byte array of a key to search for
@@ -2301,65 +2343,103 @@ public class RocksDB extends RocksObject {
    *     non-negative and no larger than "key".length
    * @param len the length of the "key" array to be used, must be non-negative
    *     and no larger than "key".length
-   * @param value StringBuilder instance which is a out parameter if a value is
+   * @param valueHolder non-null to retrieve the value if it is found, or null
-   *    found in block-cache.
+   *     if the value is not needed. If non-null, upon return of the function,
-   * @return boolean value indicating if key does not exist or might exist.
+   *     the {@code value} will be set if it could be retrieved.
+   *
+   * @return false if the key definitely does not exist in the database,
+   *     otherwise true.
    */
-  public boolean keyMayExist(final ReadOptions readOptions,
+  public boolean keyMayExist(
+      final ReadOptions readOptions,
       final byte[] key, final int offset, final int len,
-      final StringBuilder value) {
+      /* @Nullable */ final Holder<byte[]> valueHolder) {
-    checkBounds(offset, len, key.length);
+    return keyMayExist(null, readOptions,
-    return keyMayExist(nativeHandle_, readOptions.nativeHandle_,
+        key, offset, len, valueHolder);
-        key, offset, len, value);
   }
 
   /**
    * If the key definitely does not exist in the database, then this method
-   * returns false, else true.
+   * returns null, else it returns an instance of KeyMayExistResult
    *
-   * This check is potentially lighter-weight than invoking DB::Get(). One way
+   * If the caller wants to obtain value when the key
-   * to make this lighter weight is to avoid doing any IOs.
+   * is found in memory, then {@code valueHolder} must be set.
+   *
+   * This check is potentially lighter-weight than invoking
+   * {@link #get(ColumnFamilyHandle, ReadOptions, byte[])}. One way to make this
+   * lighter weight is to avoid doing any IOs.
    *
-   * @param readOptions {@link ReadOptions} instance
    * @param columnFamilyHandle {@link ColumnFamilyHandle} instance
+   * @param readOptions {@link ReadOptions} instance
    * @param key byte array of a key to search for
-   * @param value StringBuilder instance which is a out parameter if a value is
+   * @param valueHolder non-null to retrieve the value if it is found, or null
-   *    found in block-cache.
+   *     if the value is not needed. If non-null, upon return of the function,
-   * @return boolean value indicating if key does not exist or might exist.
+   *     the {@code value} will be set if it could be retrieved.
+   *
+   * @return false if the key definitely does not exist in the database,
+   *     otherwise true.
    */
-  public boolean keyMayExist(final ReadOptions readOptions,
+  public boolean keyMayExist(
-      final ColumnFamilyHandle columnFamilyHandle, final byte[] key,
+      final ColumnFamilyHandle columnFamilyHandle,
-      final StringBuilder value) {
+      final ReadOptions readOptions, final byte[] key,
-    return keyMayExist(nativeHandle_, readOptions.nativeHandle_,
+      /* @Nullable */ final Holder<byte[]> valueHolder) {
-        key, 0, key.length, columnFamilyHandle.nativeHandle_,
+    return keyMayExist(columnFamilyHandle, readOptions,
-        value);
+        key, 0, key.length, valueHolder);
   }
 
   /**
    * If the key definitely does not exist in the database, then this method
-   * returns false, else true.
+   * returns null, else it returns an instance of KeyMayExistResult
    *
-   * This check is potentially lighter-weight than invoking DB::Get(). One way
+   * If the caller wants to obtain value when the key
-   * to make this lighter weight is to avoid doing any IOs.
+   * is found in memory, then {@code valueHolder} must be set.
+   *
+   * This check is potentially lighter-weight than invoking
+   * {@link #get(ColumnFamilyHandle, ReadOptions, byte[], int, int)}.
+   * One way to make this lighter weight is to avoid doing any IOs.
    *
-   * @param readOptions {@link ReadOptions} instance
    * @param columnFamilyHandle {@link ColumnFamilyHandle} instance
+   * @param readOptions {@link ReadOptions} instance
    * @param key byte array of a key to search for
    * @param offset the offset of the "key" array to be used, must be
    *     non-negative and no larger than "key".length
    * @param len the length of the "key" array to be used, must be non-negative
    *     and no larger than "key".length
-   * @param value StringBuilder instance which is a out parameter if a value is
+   * @param valueHolder non-null to retrieve the value if it is found, or null
-   *    found in block-cache.
+   *     if the value is not needed. If non-null, upon return of the function,
-   * @return boolean value indicating if key does not exist or might exist.
+   *     the {@code value} will be set if it could be retrieved.
+   *
+   * @return false if the key definitely does not exist in the database,
+   *     otherwise true.
    */
-  public boolean keyMayExist(final ReadOptions readOptions,
+  public boolean keyMayExist(
-      final ColumnFamilyHandle columnFamilyHandle, final byte[] key,
+      final ColumnFamilyHandle columnFamilyHandle,
-      final int offset, final int len, final StringBuilder value) {
+      final ReadOptions readOptions,
+      final byte[] key, final int offset, final int len,
+      /* @Nullable */ final Holder<byte[]> valueHolder) {
     checkBounds(offset, len, key.length);
-    return keyMayExist(nativeHandle_, readOptions.nativeHandle_,
+    if (valueHolder == null) {
-        key, offset, len, columnFamilyHandle.nativeHandle_,
+      return keyMayExist(nativeHandle_,
-        value);
+          columnFamilyHandle == null ? 0 : columnFamilyHandle.nativeHandle_,
+          readOptions == null ? 0 : readOptions.nativeHandle_,
+          key, offset, len);
+    } else {
+      final byte[][] result = keyMayExistFoundValue(
+          nativeHandle_,
+          columnFamilyHandle == null ? 0 : columnFamilyHandle.nativeHandle_,
+          readOptions == null ? 0 : readOptions.nativeHandle_,
+          key, offset, len);
+      if (result[0][0] == 0x0) {
+        valueHolder.setValue(null);
+        return false;
+      } else if (result[0][0] == 0x1) {
+        valueHolder.setValue(null);
+        return true;
+      } else {
+        valueHolder.setValue(result[1]);
+        return true;
+      }
+    }
   }
 
   /**
@@ -4158,19 +4238,12 @@ public class RocksDB extends RocksObject {
   private native byte[][] multiGet(final long dbHandle, final long rOptHandle,
       final byte[][] keys, final int[] keyOffsets, final int[] keyLengths,
       final long[] columnFamilyHandles);
-  private native boolean keyMayExist(final long handle, final byte[] key,
+  private native boolean keyMayExist(
-      final int keyOffset, final int keyLength,
+      final long handle, final long cfHandle, final long readOptHandle,
-      final StringBuilder stringBuilder);
+      final byte[] key, final int keyOffset, final int keyLength);
-  private native boolean keyMayExist(final long handle, final byte[] key,
+  private native byte[][] keyMayExistFoundValue(
-      final int keyOffset, final int keyLength, final long cfHandle,
+      final long handle, final long cfHandle, final long readOptHandle,
-      final StringBuilder stringBuilder);
+      final byte[] key, final int keyOffset, final int keyLength);
-  private native boolean keyMayExist(final long handle,
-      final long optionsHandle, final byte[] key, final int keyOffset,
-      final int keyLength, final StringBuilder stringBuilder);
-  private native boolean keyMayExist(final long handle,
-      final long optionsHandle, final byte[] key, final int keyOffset,
-      final int keyLength, final long cfHandle,
-      final StringBuilder stringBuilder);
   private native long iterator(final long handle);
   private native long iterator(final long handle, final long readOptHandle);
   private native long iteratorCF(final long handle, final long cfHandle);

java/src/test/java/org/rocksdb/CompactionFilterFactoryTest.java

@@ -55,7 +55,8 @@ public class CompactionFilterFactoryTest {
           rocksDb.compactRange(cfHandles.get(1));
 
           assertThat(rocksDb.get(cfHandles.get(1), key1)).isEqualTo(value1);
-          assertThat(rocksDb.keyMayExist(cfHandles.get(1), key2, new StringBuilder())).isFalse();
+          final boolean exists = rocksDb.keyMayExist(cfHandles.get(1), key2, null);
+          assertThat(exists).isFalse();
         } finally {
           for (final ColumnFamilyHandle cfHandle : cfHandles) {
             cfHandle.close();

java/src/test/java/org/rocksdb/KeyMayExistTest.java

@@ -13,6 +13,7 @@ import java.util.ArrayList;
 import java.util.Arrays;
 import java.util.List;
 
+import static java.nio.charset.StandardCharsets.UTF_8;
 import static org.assertj.core.api.Assertions.assertThat;
 
 public class KeyMayExistTest {
@@ -41,81 +42,118 @@ public class KeyMayExistTest {
       try {
         assertThat(columnFamilyHandleList.size()).
             isEqualTo(2);
-        db.put("key".getBytes(), "value".getBytes());
+        db.put("key".getBytes(UTF_8), "value".getBytes(UTF_8));
         // Test without column family
-        StringBuilder retValue = new StringBuilder();
+        final Holder<byte[]> holder = new Holder<>();
-        boolean exists = db.keyMayExist("key".getBytes(), retValue);
+        boolean exists = db.keyMayExist("key".getBytes(UTF_8), holder);
+        assertThat(exists).isTrue();
+        assertThat(holder.getValue()).isNotNull();
+        assertThat(new String(holder.getValue(), UTF_8)).isEqualTo("value");
+
+        exists = db.keyMayExist("key".getBytes(UTF_8), null);
         assertThat(exists).isTrue();
-        assertThat(retValue.toString()).isEqualTo("value");
 
         // Slice key
-        StringBuilder builder = new StringBuilder("prefix");
+        final StringBuilder builder = new StringBuilder("prefix");
-        int offset = builder.toString().length();
+        final int offset = builder.toString().length();
         builder.append("slice key 0");
-        int len = builder.toString().length() - offset;
+        final int len = builder.toString().length() - offset;
         builder.append("suffix");
 
-        byte[] sliceKey = builder.toString().getBytes();
+        final byte[] sliceKey = builder.toString().getBytes(UTF_8);
-        byte[] sliceValue = "slice value 0".getBytes();
+        final byte[] sliceValue = "slice value 0".getBytes(UTF_8);
         db.put(sliceKey, offset, len, sliceValue, 0, sliceValue.length);
 
-        retValue = new StringBuilder();
+        exists = db.keyMayExist(sliceKey, offset, len, holder);
-        exists = db.keyMayExist(sliceKey, offset, len, retValue);
+        assertThat(exists).isTrue();
+        assertThat(holder.getValue()).isNotNull();
+        assertThat(holder.getValue()).isEqualTo(sliceValue);
+
+        exists = db.keyMayExist(sliceKey, offset, len, null);
         assertThat(exists).isTrue();
-        assertThat(retValue.toString().getBytes()).isEqualTo(sliceValue);
 
         // Test without column family but with readOptions
         try (final ReadOptions readOptions = new ReadOptions()) {
-          retValue = new StringBuilder();
+          exists = db.keyMayExist(readOptions, "key".getBytes(UTF_8), holder);
-          exists = db.keyMayExist(readOptions, "key".getBytes(), retValue);
           assertThat(exists).isTrue();
-          assertThat(retValue.toString()).isEqualTo("value");
+          assertThat(holder.getValue()).isNotNull();
+          assertThat(new String(holder.getValue(), UTF_8)).isEqualTo("value");
 
-          retValue = new StringBuilder();
+          exists = db.keyMayExist(readOptions, "key".getBytes(UTF_8), null);
-          exists = db.keyMayExist(readOptions, sliceKey, offset, len, retValue);
+          assertThat(exists).isTrue();
+
+          exists = db.keyMayExist(readOptions, sliceKey, offset, len, holder);
+          assertThat(exists).isTrue();
+          assertThat(holder.getValue()).isNotNull();
+          assertThat(holder.getValue()).isEqualTo(sliceValue);
+
+          exists = db.keyMayExist(readOptions, sliceKey, offset, len, null);
           assertThat(exists).isTrue();
-          assertThat(retValue.toString().getBytes()).isEqualTo(sliceValue);
         }
 
         // Test with column family
-        retValue = new StringBuilder();
+        exists = db.keyMayExist(columnFamilyHandleList.get(0), "key".getBytes(UTF_8),
-        exists = db.keyMayExist(columnFamilyHandleList.get(0), "key".getBytes(),
+            holder);
-            retValue);
+        assertThat(exists).isTrue();
+        assertThat(holder.getValue()).isNotNull();
+        assertThat(new String(holder.getValue(), UTF_8)).isEqualTo("value");
+
+        exists = db.keyMayExist(columnFamilyHandleList.get(0), "key".getBytes(UTF_8),
+            null);
         assertThat(exists).isTrue();
-        assertThat(retValue.toString()).isEqualTo("value");
 
         // Test slice sky with column family
-        retValue = new StringBuilder();
         exists = db.keyMayExist(columnFamilyHandleList.get(0), sliceKey, offset, len,
-            retValue);
+            holder);
+        assertThat(exists).isTrue();
+        assertThat(holder.getValue()).isNotNull();
+        assertThat(holder.getValue()).isEqualTo(sliceValue);
+
+        exists = db.keyMayExist(columnFamilyHandleList.get(0), sliceKey, offset, len,
+            null);
         assertThat(exists).isTrue();
-        assertThat(retValue.toString().getBytes()).isEqualTo(sliceValue);
 
         // Test with column family and readOptions
         try (final ReadOptions readOptions = new ReadOptions()) {
-          retValue = new StringBuilder();
+          exists = db.keyMayExist(columnFamilyHandleList.get(0), readOptions,
-          exists = db.keyMayExist(readOptions,
+              "key".getBytes(UTF_8), holder);
-              columnFamilyHandleList.get(0), "key".getBytes(),
+          assertThat(exists).isTrue();
-              retValue);
+          assertThat(holder.getValue()).isNotNull();
+          assertThat(new String(holder.getValue(), UTF_8)).isEqualTo("value");
+
+          exists = db.keyMayExist(columnFamilyHandleList.get(0), readOptions,
+              "key".getBytes(UTF_8), null);
           assertThat(exists).isTrue();
-          assertThat(retValue.toString()).isEqualTo("value");
 
           // Test slice key with column family and read options
-          retValue = new StringBuilder();
+          exists = db.keyMayExist(columnFamilyHandleList.get(0), readOptions,
-          exists = db.keyMayExist(readOptions,
+              sliceKey, offset, len, holder);
-              columnFamilyHandleList.get(0), sliceKey, offset, len,
+          assertThat(exists).isTrue();
-              retValue);
+          assertThat(holder.getValue()).isNotNull();
+          assertThat(holder.getValue()).isEqualTo(sliceValue);
+
+          exists = db.keyMayExist(columnFamilyHandleList.get(0), readOptions,
+              sliceKey, offset, len, null);
           assertThat(exists).isTrue();
-          assertThat(retValue.toString().getBytes()).isEqualTo(sliceValue);
         }
 
-        // KeyMayExist in CF1 must return false
+        // KeyMayExist in CF1 must return null value
-        assertThat(db.keyMayExist(columnFamilyHandleList.get(1),
+        exists = db.keyMayExist(columnFamilyHandleList.get(1),
-            "key".getBytes(), retValue)).isFalse();
+            "key".getBytes(UTF_8), holder);
+        assertThat(exists).isFalse();
+        assertThat(holder.getValue()).isNull();
+        exists = db.keyMayExist(columnFamilyHandleList.get(1),
+            "key".getBytes(UTF_8), null);
+        assertThat(exists).isFalse();
 
         // slice key
-        assertThat(db.keyMayExist(columnFamilyHandleList.get(1),
+        exists = db.keyMayExist(columnFamilyHandleList.get(1),
-           sliceKey, 1, 3, retValue)).isFalse();
+            sliceKey, 1, 3, holder);
+        assertThat(exists).isFalse();
+        assertThat(holder.getValue()).isNull();
+        exists = db.keyMayExist(columnFamilyHandleList.get(1),
+            sliceKey, 1, 3, null);
+        assertThat(exists).isFalse();
       } finally {
         for (final ColumnFamilyHandle columnFamilyHandle :
             columnFamilyHandleList) {
@@ -124,4 +162,31 @@ public class KeyMayExistTest {
       }
     }
   }
+
+  @Test
+  public void keyMayExistNonUnicodeString() throws RocksDBException {
+    try (final Options options = new Options()
+        .setCreateIfMissing(true)
+        .setCreateMissingColumnFamilies(true);
+         final RocksDB db = RocksDB.open(options,
+             dbFolder.getRoot().getAbsolutePath())) {
+      final byte key[] = "key".getBytes(UTF_8);
+      final byte value[] = { (byte)0x80 };  // invalid unicode code-point
+      db.put(key, value);
+
+      final byte buf[] = new byte[10];
+      final int read = db.get(key, buf);
+      assertThat(read).isEqualTo(1);
+      assertThat(buf).startsWith(value);
+
+      final Holder<byte[]> holder = new Holder<>();
+      boolean exists = db.keyMayExist("key".getBytes(UTF_8), holder);
+      assertThat(exists).isTrue();
+      assertThat(holder.getValue()).isNotNull();
+      assertThat(holder.getValue()).isEqualTo(value);
+
+      exists = db.keyMayExist("key".getBytes(UTF_8), null);
+      assertThat(exists).isTrue();
+    }
+  }
 }