Home >Backend Development >PHP Tutorial >An example of PHP reconstruction and optimization - application of template method pattern_PHP tutorial
Recently optimized the php project, recorded the experience, and started working directly. . .
PHP is mainly used for page display in company projects. There is a view on the front end, and the view requests data from the back-end service. The data transmission format is json. Let’s look at the service code before optimization:
[php]
require_once('../../../global.php');
require_once(INCLUDE_PATH . '/discache/CacherManager.php');
require_once(INCLUDE_PATH.'/oracle_oci.php');
require_once(INCLUDE_PATH.'/caihui/cwsd.php');
header('Content-type: text/plain; charset=utf-8');
$max_age = isset($_GET['max-age']) ? $_GET['max-age']*1 : 15*60;
if($max_age < 30) {
$max_age = 30;
}
header('Cache-Control: max-age='.$max_age);
// 通过将url进行hash作为缓冲key
$url = $_SERVER['HTTP_HOST'].$_SERVER['REQUEST_URI'];
$url_hash = md5($url);
//echo "/finance/hs/marketdata/segment/${url_hash}.json";
if (!CacherManager::cachePageStart(CACHER_MONGO, "/finance/hs/marketdata/segment/${url_hash}.json", 60*60)) {
// 查询条件
$page = isset($_GET['page']) ? $_GET['page']*1 : 0;
$count = isset($_GET['count']) ? $_GET['count']*1 : 30;
$type = isset($_GET['type']) ? $_GET['type'] : 'query';
$sort = isset($_GET['sort']) ? $_GET['sort'] : 'symbol';
$order = isset($_GET['order']) ? $_GET['order'] : 'desc';
$callback = isset($_GET['callback']) ? $_GET['callback'] : null;
$fieldsstring = isset($_GET['fields']) ? $_GET['fields'] : null;
$querystring = isset($_GET['query']) ? $_GET['query'] : null;
$symbol=isset($_GET['symbol'])?$_GET['symbol']:'';
$date=isset($_GET['date'])?$_GET['date']:'';
if ($type == 'query') {
$queryObj = preg_split('/:|;/', $querystring, -1);
for($i=0; $i
if($queryObj[$i]=='symbol'){
$symbol = $queryObj[$i+1];
}
if($queryObj[$i]=='date'){
$date = $queryObj[$i+1];
}
}
}
// 查询列表
$oci = ntes_get_caihui_oci();
$stocklist = array();
$cwsd = new namespacedaocaihuiCwsd($oci);
$stockcurror = $cwsd->getCznlList($symbol,$date,$sort,$order,$count*($page),$count);
$sumrecords=$cwsd->getRecordCount($symbol,$date);
$i=0;
//var_dump($symbol,$date,$sort,$order,$count*($page),$count);
foreach($stockcurror as $item){
$item['RSMFRATIO1422']=isset($item['RSMFRATIO1422'])?number_format($item['RSMFRATIO1422'],2).'%':'--';
$item['RSMFRATIO1822']=isset($item['RSMFRATIO1822'])?number_format($item['RSMFRATIO1822'],2).'%':'--';
$item['RSMFRATIO22']=isset($item['RSMFRATIO22'])?number_format($item['RSMFRATIO22'],2).'%':'--';
$item['RSMFRATIO10']=isset($item['RSMFRATIO10'])?number_format($item['RSMFRATIO10'],2):'--';
$item['RSMFRATIO12']=isset($item['RSMFRATIO12'])?number_format($item['RSMFRATIO12'],2):'--';
$item['RSMFRATIO4']=isset($item['RSMFRATIO4'])?number_format($item['RSMFRATIO4'],2):'--';
$item['RSMFRATIO18']=isset($item['RSMFRATIO18'])?number_format($item['RSMFRATIO18'],2):'--';
$item['RSMFRATIO14']=isset($item['RSMFRATIO14'])?number_format($item['RSMFRATIO14'],2):'--';
$item['CODE']=$item['EXCHANGE'].$item['SYMBOL'];
//$item['REPORTDATE']=isset($item['REPORTDATE'])?$item['REPORTDATE']:'--';
$stocklist[$i] = $item;
$i=$i+1;
}
// Output results
$result = array();
//Page number, count per page, total number of results, pagecount, result list
$result['page'] = $page;
$result['count'] = $count;
$result['order'] = $order;
$result['total'] = $i;//$stockcurror->count();
$result['pagecount'] = ceil($sumrecords['SUMRECORD']/$count);
$result['time'] = date('Y-m-d H:i:s');
$result['list'] = $stocklist;
if(emptyempty($callback)){
echo json_encode($result);
}else{
echo $callback.'('.json_encode($result).');';
}
CacherManager::cachePageEnd();
}
?>
Let's take a look at the specific completion of this service:
1. Lines 6-16, prepare cache parameters and enable caching.
2. Lines 19-41, extract request parameters.
3. Lines 44-49, connect and query the database.
4. Lines 50-67, put the database query results into the array.
5. Lines 71-84, prepare json data.
6. Lines 86-87, turn off caching.
If you only look at this file, the problems are:
1. Lines 19-86, no indentation.
2. Line 44, the database will be reconnected with each request.
3. Lines 53-61, the repeated logic can be extracted as a function and then completed through iteration.
If most back-end services adopt this structure, then the problem is that all services need to go through a series of processes: opening cache, getting parameters, getting data, json conversion, and closing cache. In all processes, except for the logic of obtaining data, other processes are the same. There is a lot of repetitive logic in the code, which even gives people a "copy-paste" feeling, which seriously violates the DRY principle (Don't Repeat Yourself). Therefore, it needs to be reconstructed using object-oriented thinking. In the process of my reconstruction, I always kept one principle in mind - the principle of encapsulation change. The so-called encapsulation of changes is to distinguish between the constant and the variable in the system, and to encapsulate the variable, so that changes can be easily dealt with.
Through the above analysis, only the logic of obtaining data changes, and other logic remains unchanged. Therefore, the logic of obtaining data needs to be encapsulated. The specific encapsulation method can be inheritance or combination. I adopt the inheritance method. First, I abstract the service processing process as:
service(){
startCache();
getParam();
getData(); // Abstract method, implemented by subclasses
toJson();
Closecache ();
}
ServiceBase class is abstracted and inherited by subclasses to implement the corresponding logic of obtaining data. Subclasses do not need to deal with other logic such as parameter fetching and caching, as these are all handled by the ServiceBase class.
[php]
abstract class ServiceBase {
Public function __construct($cache_path, $cache_type, $max_age, $age_explore) {
// Get request parameters
$this->page = $this->getQueryParamDefault('page', 0, INT);
// Omit other logic for obtaining parameters
…
// Generate response
$this->response();
}
/**
* *
* Subclass implementation, returns data in array format
*/
abstract protected function data();
/**
* *
* Subclass implementation, returns the total number of all data
*/
abstract protected function total();
private function cache() {
$url = $_SERVER['HTTP_HOST'].$_SERVER['REQUEST_URI'];
$url_hash = md5($url);
$key = $this->cache_path.$url_hash.'.json';
If(!CacherManager::cachePageStart($this->cache_type, $key, $this->age_cache)){
$this->no_cache();
CacherManager::cachePageEnd();
}
}
private function no_cache(){
$data = $this->data();
$total = $this->total();
$this->send_data($data, $total);
}
Private function send_data($data, $total){
// Convert json, omit specific code
}
private function response() {
header('Content-type: text/plain; charset=utf-8');
header('Cache-Control: max-age='.$this->age_explore);
If($this->cache_type == NONE || self::$enable_cache == false){
$this->no_cache();
}else{
$this->cache();
}
}
}
This is the abstract parent class of each service. There are two abstract methods data and total. Data returns data in array format. Tatol is added due to paging. A specific service only needs to inherit ServiceBase and implement the data and total methods, and other logic is reused from the parent class. In fact, the optimized ServiceBase uses the template method pattern (Template Method). The parent class defines the algorithm processing process (the processing process of the service), and the subclass implements the steps of a specific change (the logic of obtaining data for the specific service). . By using the template method pattern, you can ensure that step changes are transparent to the client, and the logic in the parent class can be reused.
The following is the code of the above php using ServiceBase:
[php]
class CWSDService extends ServiceBase{
function __construct(){
parent::__construct();
$oci = ntes_get_caihui_oci();
$this->$cwsd = new namespacedaocaihuiCwsd($oci);
}
public function data(){
$stocklist = array();
$stockcurror = $this->cwsd->getCznlList($this->query_obj['symbol'],
$this->query_obj['symbol'], $sort, $order, $count*($page), $count);
$filter_list = array('RSMFRATIO1422', 'RSMFRATIO1822', 'RSMFRATIO22',
'RSMFRATIO10', 'RSMFRATIO12', 'RSMFRATIO4', 'RSMFRATIO18',
'RSMFRATIO14');
$i=0;
foreach($stockcurror as $item){
foreach($filter_list as $k)
$this->filter($item, $k);
$item['CODE']=$item['EXCHANGE'].$item['SYMBOL'];
$stocklist[$i] = $item;
$i=$i+1;
}
return $stocklist;
}
public function total(){
return $sumrecords=$this->cwsd->getRecordCount($this->query_obj['symbol'],
$this->query_obj['symbol']);
}
private function filter($item, $k){
isset($item[$k])?number_format($item[$k],2).'%':'--';
}
}
new CWSDService('/finance/hs/realtimedata/market/ab', MONGO, 30, 30);
代码量从87减少到32行,是因为大部分的逻辑都由父类完成,具体service只需要关注自己的业务逻辑就可以了。通过上面代码可以看出继承可以实现代码复用,多个子类中的相同的逻辑可以提取到父类中达到复用的目的;同时,继承也增加了父类和子类之间的耦合性,这也就是组合由于继承的方面,如果这个例子采用组合来封装变化,则具体的实现就是策略模式,将具体获取数据的逻辑看成是策略,不同的service就是不同的策略,由于时间原因,不再赘述。。。
摘自 chosen0ne的专栏