spark OpenHashMap 源码

spark OpenHashMap 代码

文件路径：/core/src/main/scala/org/apache/spark/util/collection/OpenHashMap.scala

/*
 * Licensed to the Apache Software Foundation (ASF) under one or more
 * contributor license agreements.  See the NOTICE file distributed with
 * this work for additional information regarding copyright ownership.
 * The ASF licenses this file to You under the Apache License, Version 2.0
 * (the "License"); you may not use this file except in compliance with
 * the License.  You may obtain a copy of the License at
 *
 *    http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

package org.apache.spark.util.collection

import scala.reflect.ClassTag

/**
 * A fast hash map implementation for nullable keys. This hash map supports insertions and updates,
 * but not deletions. This map is about 5X faster than java.util.HashMap, while using much less
 * space overhead.
 *
 * Under the hood, it uses our OpenHashSet implementation.
 *
 * NOTE: when using numeric type as the value type, the user of this class should be careful to
 * distinguish between the 0/0.0/0L and non-exist value
 */
private[spark]
class OpenHashMap[K : ClassTag, @specialized(Long, Int, Double) V: ClassTag](
    initialCapacity: Int)
  extends Iterable[(K, V)]
  with Serializable {

  def this() = this(64)

  protected var _keySet = new OpenHashSet[K](initialCapacity)

  // Init in constructor (instead of in declaration) to work around a Scala compiler specialization
  // bug that would generate two arrays (one for Object and one for specialized T).
  private var _values: Array[V] = _
  _values = new Array[V](_keySet.capacity)

  @transient private var _oldValues: Array[V] = null

  // Treat the null key differently so we can use nulls in "data" to represent empty items.
  private var haveNullValue = false
  private var nullValue: V = null.asInstanceOf[V]

  override def size: Int = if (haveNullValue) _keySet.size + 1 else _keySet.size

  /** Tests whether this map contains a binding for a key. */
  def contains(k: K): Boolean = {
    if (k == null) {
      haveNullValue
    } else {
      _keySet.getPos(k) != OpenHashSet.INVALID_POS
    }
  }

  /** Get the value for a given key */
  def apply(k: K): V = {
    if (k == null) {
      nullValue
    } else {
      val pos = _keySet.getPos(k)
      if (pos < 0) {
        null.asInstanceOf[V]
      } else {
        _values(pos)
      }
    }
  }

  /** Set the value for a key */
  def update(k: K, v: V): Unit = {
    if (k == null) {
      haveNullValue = true
      nullValue = v
    } else {
      val pos = _keySet.addWithoutResize(k) & OpenHashSet.POSITION_MASK
      _values(pos) = v
      _keySet.rehashIfNeeded(k, grow, move)
      _oldValues = null
    }
  }

  /**
   * If the key doesn't exist yet in the hash map, set its value to defaultValue; otherwise,
   * set its value to mergeValue(oldValue).
   *
   * @return the newly updated value.
   */
  def changeValue(k: K, defaultValue: => V, mergeValue: (V) => V): V = {
    if (k == null) {
      if (haveNullValue) {
        nullValue = mergeValue(nullValue)
      } else {
        haveNullValue = true
        nullValue = defaultValue
      }
      nullValue
    } else {
      val pos = _keySet.addWithoutResize(k)
      if ((pos & OpenHashSet.NONEXISTENCE_MASK) != 0) {
        val newValue = defaultValue
        _values(pos & OpenHashSet.POSITION_MASK) = newValue
        _keySet.rehashIfNeeded(k, grow, move)
        newValue
      } else {
        _values(pos) = mergeValue(_values(pos))
        _values(pos)
      }
    }
  }

  override def iterator: Iterator[(K, V)] = new Iterator[(K, V)] {
    var pos = -1
    var nextPair: (K, V) = computeNextPair()

    /** Get the next value we should return from next(), or null if we're finished iterating */
    def computeNextPair(): (K, V) = {
      if (pos == -1) {    // Treat position -1 as looking at the null value
        if (haveNullValue) {
          pos += 1
          return (null.asInstanceOf[K], nullValue)
        }
        pos += 1
      }
      pos = _keySet.nextPos(pos)
      if (pos >= 0) {
        val ret = (_keySet.getValue(pos), _values(pos))
        pos += 1
        ret
      } else {
        null
      }
    }

    def hasNext: Boolean = nextPair != null

    def next(): (K, V) = {
      val pair = nextPair
      nextPair = computeNextPair()
      pair
    }
  }

  private def grow(newCapacity: Int): Unit = {
    _oldValues = _values
    _values = new Array[V](newCapacity)
  }

  private def move(oldPos: Int, newPos: Int): Unit = {
    _values(newPos) = _oldValues(oldPos)
  }
}

相关信息

spark 源码目录

相关文章

spark AppendOnlyMap 源码

spark BitSet 源码

spark CompactBuffer 源码

spark ExternalAppendOnlyMap 源码

spark ExternalSorter 源码

spark ImmutableBitSet 源码

spark MedianHeap 源码

spark OpenHashSet 源码

spark PairsWriter 源码

spark PartitionedAppendOnlyMap 源码