前言
前幾天研究Spark分區器的時候,Range分區器的原理中邊界劃定時就用到了分塊查找算法,當時不知道這個名詞,今天學習的時候,發現原理就是分塊查找啊。多學習肯定沒錯的,一下子加速我的理解。
代碼
package day15
import day14.Utils
/**
* Created by doctorq on 2017/6/30.
* 分塊查找:適用於表結構動態變化的場景,比順序查找快,但不如折半查找
* 表索引有序,值無需有序
*/
object BlockSearch extends App with Utils {
/**
* 獲取索引表
*
* @param unSort
* @param sizePerBlock 每個塊的數據最大個數
* @return
*/
def getIndexArray(search: Array[Int], sizePerBlock: Int): Array[Int] = {
val blockSize = search.length / sizePerBlock + 1
// println(s"分成${blockSize}塊")
//索引表,順序的
val indexArray = new Array[Int](blockSize)
var i = 0
var j = 0
while (i < array.length) {
val maxIndex = i + sizePerBlock - 1
if (maxIndex > array.length - 1) {
indexArray(j) = array(array.length - 1)
} else {
indexArray(j) = array(i + sizePerBlock - 1)
}
j += 1
i += sizePerBlock
}
indexArray
}
/**
*
* @param search 待查找表
* @param sizePerBlock 每個塊的最大數據個數
* @param find
* @return
*/
def blockSearch(search: Array[Int], sizePerBlock: Int, find: Int): Int = {
//獲得索引表
val indexArray = getIndexArray(array, sizePerBlock)
val blockIndex: Int = findBlockIndex(indexArray, find)
if (blockIndex != -1) {
println(s"處於第${blockIndex}塊中")
var start = blockIndex * sizePerBlock - 1 + 1
if (start < 0) start = 0
val end = (blockIndex + 1) * sizePerBlock - 1
println(s"$start ~ $end")
for (i <- start to end) {
if (search(i) == find) return i
}
}
return -1
}
def findBlockIndex(indexArray: Array[Int], find: Int): Int = {
for (i <- 0 until indexArray.length) {
if (indexArray(i) >= find) {
return i
}
}
return -1
}
val array = Array[Int](2, 4, 5, 63, 5, 6, 2, 3, 89, 34, 34, 7, 8).sorted
printlnArray(array)
// println(s"元素個數: ${array.size}")
var length = 3
val result = blockSearch(array, length, 4)
result match {
case -1 => println("no found")
case _ => println(s"location:${result}")
}
}