服务器之家:专注于服务器技术及软件下载分享
分类导航

PHP教程|ASP.NET教程|Java教程|ASP教程|编程技术|正则表达式|C/C++|IOS|C#|Swift|Android|JavaScript|易语言|

服务器之家 - 编程语言 - Java教程 - Java ArrayList.add 的实现方法

Java ArrayList.add 的实现方法

2021-06-10 14:22EthanSun Java教程

这篇文章主要介绍了Java ArrayList.add 的实现方法,小编觉得挺不错的,现在分享给大家,也给大家做个参考。一起跟随小编过来看看吧

arraylist是平时相当常用的list实现, 其中boolean add(e e) 的实现比较直接:

?
1
2
3
4
5
6
7
8
9
10
11
/**
 * appends the specified element to the end of this list.
 *
 * @param e element to be appended to this list
 * @return <tt>true</tt> (as specified by {@link collection#add})
 */
public boolean add(e e) {
  ensurecapacityinternal(size + 1); // increments modcount!!
  elementdata[size++] = e;
  return true;
}

有时候也使用 void add(int index, e element) 把元素插入到指定的index上. 在jdk中的实现是:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
/**
 * inserts the specified element at the specified position in this
 * list. shifts the element currently at that position (if any) and
 * any subsequent elements to the right (adds one to their indices).
 *
 * @param index index at which the specified element is to be inserted
 * @param element element to be inserted
 * @throws indexoutofboundsexception {@inheritdoc}
 */
public void add(int index, e element) {
  rangecheckforadd(index);
 
  ensurecapacityinternal(size + 1); // increments modcount!!
  system.arraycopy(elementdata, index, elementdata, index + 1,
           size - index);
  elementdata[index] = element;
  size++;
}

略有差别, 需要保证当前elementdata 数组容量够用, 然后把从index处一直到尾部的数组元素都向后挪一位. 最后把要插入的元素赋给数组的index处.

一直以来, 我都认为 system.arraycopy 这个native方法, 它的c++实现是调用底层的memcpy, 直接方便, 效率也没问题.

但今天看了openjdk的源码发现并非如此.

以openjdk8u60 为例, 在objarrayklass.cpp 中:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
void objarrayklass::copy_array(arrayoop s, int src_pos, arrayoop d,
                int dst_pos, int length, traps) {
 assert(s->is_objarray(), "must be obj array");
 
 if (!d->is_objarray()) {
  throw(vmsymbols::java_lang_arraystoreexception());
 }
 
 // check is all offsets and lengths are non negative
 if (src_pos < 0 || dst_pos < 0 || length < 0) {
  throw(vmsymbols::java_lang_arrayindexoutofboundsexception());
 }
 // check if the ranges are valid
 if ( (((unsigned int) length + (unsigned int) src_pos) > (unsigned int) s->length())
   || (((unsigned int) length + (unsigned int) dst_pos) > (unsigned int) d->length()) ) {
  throw(vmsymbols::java_lang_arrayindexoutofboundsexception());
 }
 
 // special case. boundary cases must be checked first
 // this allows the following call: copy_array(s, s.length(), d.length(), 0).
 // this is correct, since the position is supposed to be an 'in between point', i.e., s.length(),
 // points to the right of the last element.
 if (length==0) {
  return;
 }
 if (usecompressedoops) {
  narrowoop* const src = objarrayoop(s)->obj_at_addr<narrowoop>(src_pos);
  narrowoop* const dst = objarrayoop(d)->obj_at_addr<narrowoop>(dst_pos);
  do_copy<narrowoop>(s, src, d, dst, length, check);
 } else {
  oop* const src = objarrayoop(s)->obj_at_addr<oop>(src_pos);
  oop* const dst = objarrayoop(d)->obj_at_addr<oop>(dst_pos);
  do_copy<oop> (s, src, d, dst, length, check);
 }
}

可以看到copy_array在做了各种检查之后, 最终copy的部分在do_copy方法中, 而这个方法实现如下:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
// either oop or narrowoop depending on usecompressedoops.
template <class t> void objarrayklass::do_copy(arrayoop s, t* src,
                arrayoop d, t* dst, int length, traps) {
 
 barrierset* bs = universe::heap()->barrier_set();
 // for performance reasons, we assume we are that the write barrier we
 // are using has optimized modes for arrays of references. at least one
 // of the asserts below will fail if this is not the case.
 assert(bs->has_write_ref_array_opt(), "barrier set must have ref array opt");
 assert(bs->has_write_ref_array_pre_opt(), "for pre-barrier as well.");
 
 if (s == d) {
  // since source and destination are equal we do not need conversion checks.
  assert(length > 0, "sanity check");
  bs->write_ref_array_pre(dst, length);
  copy::conjoint_oops_atomic(src, dst, length);
 } else {
  // we have to make sure all elements conform to the destination array
  klass* bound = objarrayklass::cast(d->klass())->element_klass();
  klass* stype = objarrayklass::cast(s->klass())->element_klass();
  if (stype == bound || stype->is_subtype_of(bound)) {
   // elements are guaranteed to be subtypes, so no check necessary
   bs->write_ref_array_pre(dst, length);
   copy::conjoint_oops_atomic(src, dst, length);
  } else {
   // slow case: need individual subtype checks
   // note: don't use obj_at_put below because it includes a redundant store check
   t* from = src;
   t* end = from + length;
   for (t* p = dst; from < end; from++, p++) {
    // xxx this is going to be slow.
    t element = *from;
    // even slower now
    bool element_is_null = oopdesc::is_null(element);
    oop new_val = element_is_null ? oop(null)
                   : oopdesc::decode_heap_oop_not_null(element);
    if (element_is_null ||
      (new_val->klass())->is_subtype_of(bound)) {
     bs->write_ref_field_pre(p, new_val);
     *p = element;
    } else {
     // we must do a barrier to cover the partial copy.
     const size_t pd = pointer_delta(p, dst, (size_t)heapoopsize);
     // pointer delta is scaled to number of elements (length field in
     // objarrayoop) which we assume is 32 bit.
     assert(pd == (size_t)(int)pd, "length field overflow");
     bs->write_ref_array((heapword*)dst, pd);
     throw(vmsymbols::java_lang_arraystoreexception());
     return;
    }
   }
  }
 }
 bs->write_ref_array((heapword*)dst, length);
}

可以看到, 在设定了heap barrier之后, 元素是在for循环中被一个个挪动的. 做的工作比我想象的要多.

如果有m个元素, 按照给定位置, 使用arraylist.add(int,e)逐个插入到一个长度为n的arraylist中, 复杂度应当是o(m*n), 或者o(m*(m+n)), 所以, 如果m和n都不小的话, 效率确实是不高的.

效率高一些的方法是, 建立m+n长度的数组或arraylist, 在给定位置赋值该m个要插入的元素, 其他位置依次赋值原n长度list的元素. 这样时间复杂度应当是o(m+n).

还有, 在前面的实现中, 我们可以看到有对ensurecapacityinternal(int) 的调用. 这个保证数组容量的实现主要在:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
/**
 * increases the capacity to ensure that it can hold at least the
 * number of elements specified by the minimum capacity argument.
 *
 * @param mincapacity the desired minimum capacity
 */
private void grow(int mincapacity) {
  // overflow-conscious code
  int oldcapacity = elementdata.length;
  int newcapacity = oldcapacity + (oldcapacity >> 1);
  if (newcapacity - mincapacity < 0)
    newcapacity = mincapacity;
  if (newcapacity - max_array_size > 0)
    newcapacity = hugecapacity(mincapacity);
  // mincapacity is usually close to size, so this is a win:
  elementdata = arrays.copyof(elementdata, newcapacity);
}

大家知道由于效率原因, arraylist容量增长不是正好按照要求的容量mincapacity来设计的, 新容量计算的主要逻辑是: 如果要求容量比当前容量的1.5倍大, 就按照要求容量重新分配空间; 否则按当前容量1.5倍增加. 当然不能超出integer.max_value了. oldcapacity + (oldcapacity >> 1) 实际就是当前容量1.5倍, 等同于(int) (oldcapacity * 1.5), 但因这段不涉及浮点运算只是移位, 显然效率高不少.

所以如果arraylist一个一个add元素的话, 容量是在不够的时候1.5倍增长的. 关于1.5这个数字, 或许是觉得2倍增长太快了吧. 也或许有实验数据的验证支撑.

关于这段代码中出现的arrays.copyof这个方法, 实现的是重新分配一段数组, 把elementdata赋值给新分配的空间, 如果新分配的空间大, 则后面赋值null, 如果分配空间比当前数组小则截断. 底层还是调用的system.arraycopy.

以上就是本文的全部内容,希望对大家的学习有所帮助,也希望大家多多支持服务器之家。

原文链接:https://segmentfault.com/a/1190000016910760

延伸 · 阅读

精彩推荐