Three ways of infinite classification (iteration, recursion, reference)

There are two general classification tree structures:

• One is the adjacency list, which is the form of id and parent id.

• The other is nested set, which is the form of left and right values.

The left and right value formquery is more efficient and does not require recursion, etc. It is recommended to use, but it is not as simple and intuitive as the pid form, and some are old The structural design of databases similar to regions has always been in the form of pid (it seems that there are algorithms that can convert the two, but I won’t go into details), so. . .

The following are all in the form of adjacency list. The data table is similar to the format of id, pid, name.

Usually, this is achieved by reading all the data from the database and then assembling the array. Of course, you can also query the database every time you recurse, but it will cause database pressure, and It is not easy to encapsulate into a method and is not recommended.

Currently, there are three commonly used methods. Let’s implement the select drop-down menu display style:

1. The first is the most commonly used and common, and also the least efficient. The recursive method: just keep foreachlooprecursion.

function getTreeOptions3($list, $pid = 0)
{    $options = [];    foreach ($list as $key => $value) {        if ($value['id'] == $pid) {//查看是否为子元素，如果是则递归继续查询
            $options[$value['id']] = $value['name'];            unset($list[$key]);//销毁已查询的，减轻下次递归时查询数量
            $optionsTmp = $this->getTreeOptions3($list, $value['id']);//递归
            if (!empty($optionsTmp)) {                $options = array_merge($options, $optionsTmp);
            }
        }
    }    return $options;
}

2. The second method is to use the recursion of pushing and popping into the stack to calculate. The efficiency is better than the previous one, but It's also quite slow. The process is to first reverse the array, then take out the top-level array and push it onto the stack, start the while loop, first pop one out of the stack to find if there is a child node under it, and if there is a child node, push this child node into the stack as well. The next while loop will check the child nodes, and so on:

function getTreeOptions2($list, $pid = 0)
{    $tree = [];    if (!empty($list)) {        //先将数组反转，因为后期出栈时会优先出最上面的
        $list = array_reverse($list);        //先取出顶级的来压入数组$stack中，并将在$list中的删除掉
        $stack = [];        foreach ($list as $key => $value) {            if ($value['pid'] == $pid) {                array_push($stack,$value);                unset($list[$key]);
            }
        }        while (count($stack)) {            //先从栈中取出第一项
            $info = array_pop($stack);            //查询剩余的$list中pid与其id相等的，也就是查找其子节点
            foreach ($list as $key => $child) {                if ($child[pid] == $info['id']) {                    //如果有子节点则入栈，while循环中会继续查找子节点的下级
                    array_push($stack,  $child);                    unset($list[$key]);
                }
            }            //组装成下拉菜单格式
            $tree[$info['id']] = $info['name'];
        }
    }    return $tree;
}

3. Use to quote Processing, this is really clever and the most efficient, you can refer to here:

/**
* 先生成类似下面的形式的数据
* [
*  'id'=>1,
*  'pid'=>0,
*  'items'=>[
*      'id'=>2,
*      'pid'=>'1'
*       。。。
*  ]
* ]*/function getTree($list, $pid = 0)
{    $tree = [];    if (!empty($list)) {        //先修改为以id为下标的列表
        $newList = [];        foreach ($list as $k => $v) {            $newList[$v['id']] = $v;
        }        //然后开始组装成特殊格式
        foreach ($newList as $value) {            if ($pid == $value['pid']) {//先取出顶级
                $tree[] = &$newList[$value['id']];
            } elseif (isset($newList[$value['pid']])) {//再判定非顶级的pid是否存在，如果存在，则再pid所在的数组下面加入一个字段items，来将本身存进去
                $newList[$value['pid']]['items'][] = &$newList[$value['id']];
            }
        }
    }    return $tree;
}

Then recursively generate the select drop-down menu required , due to the special format above, the recursion is very fast:

function formatTree($tree)
{    $options = [];    if (!empty($tree)) {        foreach ($tree as $key => $value) {            $options[$value['id']] = $value['name'];            if (isset($value['items'])) {//查询是否有子节点
                $optionsTmp = $this->formatTree($value['items']);                if (!empty($optionsTmp)) {                    $options = array_merge($options, $optionsTmp);
                }
            }
        }
    }    return $options;
}

The above three are suitable for small amounts of data. , it doesn’t matter which one is used, but it is very obvious for large amounts of data. Efficiency comparison of test results using 4000 regional data:

• The first method (recursive) is time-consuming : About 8.9441471099854

• The second method (iteration) takes about 6.7250330448151

• The third method (reference) takes time: 0.028863906860352 Left and right

Let me go, this gap, the third method is really outrageous. But remind me again, this is only when reading a lot of data at one time. When the amount of data is very small, the difference is almost the same. You don’t have to use the most efficient one. It can also be achieved through other methods such as lazy loading.

Encapsulate a class by the way, you can add some padding or something. More details can be found in the following classes:

  1   8  */  9  10 class Tree 11 { 12     /** 13      * 图标 14      */ 15     public $icon = '└'; 16     /** 17      * 填充 18      */ 19     public $blank = '   '; 20     /** 21      * 默认ID字段名称 22      */ 23     public $idName = 'id'; 24     /** 25      * 默认PID字段名称 26      */ 27     public $pidName = 'pid'; 28     /** 29      * 默认名称字段名称 30      */ 31     public $titleName = 'name'; 32     /** 33      * 默认子元素字段名称 34      */ 35     public $childrenName = 'items'; 36  37     /** 38      * 构造函数，可覆盖默认字段值 39      * @param array $config 40      */ 41     function construct($config = []) 42     { 43         if (!empty($config)) { 44             foreach ($config as $name => $value) { 45                 $this->$name = $value; 46             } 47         } 48     } 49  50     /** 51      * 生成下拉菜单可用树列表的方法 52      * 经测试4000条地区数据耗时0.02左右，比另外两种方法快超级多 53      * 流程是先通过引用方法来生成一种特殊树结构，再通过递归来解析这种特殊的结构 54      * @param array $list 55      * @param int $pid 56      * @param int $level 57      * @return array 58      */ 59     public function getTreeOptions($list, $pid = 0, $level = 0) 60     { 61         //先生成特殊规格的树 62         $tree = $this->getTree($list, $pid); 63         //再组装成select需要的形式 64         return $this->formatTree($tree, $level); 65     } 66  67     /** 68      * 通过递归来解析特殊的树结构来组装成下拉菜单所需要的样式 69      * @param array $tree 特殊规格的数组 70      * @param int $level 71      * @return array 72      */ 73     protected function formatTree($tree, $level = 0) 74     { 75         $options = []; 76         if (!empty($tree)) { 77             $blankStr = str_repeat($this->blank, $level) . $this->icon; 78             if ($level == 0) {//第一次无需有图标及空格 79                 $blankStr = ''; 80             } 81             foreach ($tree as $key => $value) { 82                 $options[$value[$this->idName]] = $blankStr . $value[$this->titleName]; 83                 if (isset($value[$this->childrenName])) {//查询是否有子节点 84                     $optionsTmp = $this->formatTree($value[$this->childrenName], $level + 1); 85                     if (!empty($optionsTmp)) { 86                         $options = array_merge($options, $optionsTmp); 87                     } 88                 } 89             } 90         } 91         return $options; 92     } 93  94     /** 95      * 生成类似下种格式的树结构 96      * 利用了引用&来实现，参照：http://blog.csdn.net/gxdvip/article/details/24434801 97      * [ 98      *  'id'=>1, 99      *  'pid'=>0,100      *  'items'=>[101      *      'id'=>2,102      *      'pid'=>'1'103      *       。。。104      *  ]105      * ]106      * @param array $list107      * @param int $pid108      * @return array109      */110     protected function getTree($list, $pid = 0)111     {112         $tree = [];113         if (!empty($list)) {114             //先修改为以id为下标的列表115             $newList = [];116             foreach ($list as $k => $v) {117                 $newList[$v[$this->idName]] = $v;118             }119             //然后开始组装成特殊格式120             foreach ($newList as $value) {121                 if ($pid == $value[$this->pidName]) {122                     $tree[] = &$newList[$value[$this->idName]];123                 } elseif (isset($newList[$value[$this->pidName]])) {124                     $newList[$value[$this->pidName]][$this->childrenName][] = &$newList[$value[$this->idName]];125                 }126             }127         }128         return $tree;129     }130 131     /**132      * 第二种方法，利用出入栈迭代来实现133      * 经测试4000条地区数据耗时6.5s左右，比较慢134      * @param $list135      * @param int $pid136      * @param int $level137      * @return array138      */139     public function getTreeOptions2($list, $pid = 0, $level = 0)140     {141         $tree = [];142         if (!empty($list)) {143 144             //先将数组反转，因为后期出栈时会有限出最上面的145             $list = array_reverse($list);146             //先取出顶级的来压入数组$stack中，并将在$list中的删除掉147             $stack = [];148             foreach ($list as $key => $value) {149                 if ($value[$this->pidName] == $pid) {150                     array_push($stack, ['data' => $value, 'level' => $level]);//将层级记录下来，方便填充空格151                     unset($list[$key]);152                 }153             }154             while (count($stack)) {155                 //先从栈中取出第一项156                 $info = array_pop($stack);157                 //查询剩余的$list中pid与其id相等的，也就是查找其子节点158                 foreach ($list as $key => $child) {159                     if ($child[$this->pidName] == $info['data'][$this->idName]) {160                         //如果有子节点则入栈，while循环中会继续查找子节点的下级161                         array_push($stack, ['data' => $child, 'level' => $info['level'] + 1]);162                         unset($list[$key]);163                     }164                 }165                 //组装成下拉菜单格式166                 $blankStr = str_repeat($this->blank, $info['level']) . $this->icon;167                 if ($info['level'] == 0) {//第一次无需有图标及空格168                     $blankStr = '';169                 }170                 $tree[$info['data'][$this->idName]] = $blankStr . $info['data'][$this->titleName];171             }172         }173         return $tree;174     }175 176     /**177      * 第三种普通列表转为下拉菜单可用的树列表178      * 经测试4000条地区数据耗时8.7s左右，最慢179      * @param array $list 原数组180      * @param int $pid 起始pid181      * @param int $level 起始层级182      * @return array183      */184     public function getTreeOptions3($list, $pid = 0, $level = 0)185     {186         $options = [];187         if (!empty($list)) {188             $blankStr = str_repeat($this->blank, $level) . $this->icon;189             if ($level == 0) {//第一次无需有图标及空格190                 $blankStr = '';191             }192             foreach ($list as $key => $value) {193                 if ($value[$this->pidName] == $pid) {194                     $options[$value[$this->idName]] = $blankStr . $value[$this->titleName];195                     unset($list[$key]);//销毁已查询的，减轻下次递归时查询数量196                     $optionsTmp = $this->getTreeOptions3($list, $value[$this->idName], $level + 1);//递归197                     if (!empty($optionsTmp)) {198                         $options = array_merge($options, $optionsTmp);199                     }200                 }201             }202         }203         return $options;204     }205 }

View Code

Record the above, if you reprint, please indicate the source address

The above is the detailed content of Three ways of infinite classification (iteration, recursion, reference). For more information, please follow other related articles on the PHP Chinese website!