php queryList函数,QueryList/QueryList.php at master · baijunyao/QueryList · GitHub
namespace QL;
use phpQuery,Exception,ReflectionClass;
use Monolog\Logger;
use Monolog\Handler\StreamHandler;
/**
* QueryList
*
* 一個基于phpQuery的通用列表采集類
*
* @author Jaeger
* @email 734708094@qq.com
* @link http://git.oschina.net/jae/QueryList
* @version 3.1.2
*
* @example
*
//獲取CSDN移動開發欄目下的文章列表標題
$hj = QueryList::Query('http://mobile.csdn.net/',array("title"=>array('.unit h1','text')));
print_r($hj->data);
//回調函數1
function callfun1($content,$key)
{
return '回調函數1：'.$key.'-'.$content;
}
class HJ{
//回調函數2
static public function callfun2($content,$key)
{
return '回調函數2：'.$key.'-'.$content;
}
}
//獲取CSDN文章頁下面的文章標題和內容
$url = 'http://www.csdn.net/article/2014-06-05/2820091-build-or-buy-a-mobile-game-backend';
$rules = array(
'title'=>array('h1','text','','callfun1'), //獲取純文本格式的標題,并調用回調函數1
'summary'=>array('.summary','text','-input strong'), //獲取純文本的文章摘要，但保strong標簽并去除input標簽
'content'=>array('.news_content','html','div a -.copyright'), //獲取html格式的文章內容，但過濾掉div和a標簽,去除類名為copyright的元素
'callback'=>array('HJ','callfun2') //調用回調函數2作為全局回調函數
);
$rang = '.left';
$hj = QueryList::Query($url,$rules,$rang);
print_r($hj->data);
//繼續獲取右邊相關熱門文章列表的標題以及鏈接地址
$hj->setQuery(array('title'=>array('','text'),'url'=>array('a','href')),'#con_two_2 li');
//輸出數據
echo $hj->getData();
*/
class QueryList
{
public $data;
public $html;
private $page;
private $pqHtml;
private $outputEncoding = false;
private $inputEncoding = false;
private $htmlEncoding;
public static $logger = null;
public static $instances;
public function __construct() {
}
/**
* 靜態方法，訪問入口
* @param string $page 要抓取的網頁URL地址(支持https);或者是html源代碼
* @param array $rules 【選擇器數組】說明：格式array("名稱"=>array("選擇器","類型"[,"標簽過濾列表"][,"回調函數"]),.......[,"callback"=>"全局回調函數"]);
* 【選擇器】說明:可以為任意的jQuery選擇器語法
* 【類型】說明：值 "text" ,"html" ,"HTML標簽屬性" ,
* 【標簽過濾列表】:可選，要過濾的選擇器名，多個用空格隔開,當標簽名前面添加減號(-)時(此時標簽可以為任意的元素選擇器)，表示移除該標簽以及標簽內容，否則當【類型】值為text時表示需要保留的HTML標簽，為html時表示要過濾掉的HTML標簽
* 【回調函數】/【全局回調函數】：可選，字符串(函數名) 或 數組(array("類名","類的靜態方法"))，回調函數應有倆個參數，第一個參數是選擇到的內容，第二個參數是選擇器數組下標，回調函數會覆蓋全局回調函數
*
* @param string $range 【塊選擇器】：指 先按照規則 選出 幾個大塊 ，然后再分別再在塊里面 進行相關的選擇
* @param string $outputEncoding【輸出編碼格式】指要以什么編碼輸出(UTF-8,GB2312,.....)，防止出現亂碼,如果設置為 假值 則不改變原字符串編碼
* @param string $inputEncoding 【輸入編碼格式】明確指定輸入的頁面編碼格式(UTF-8,GB2312,.....)，防止出現亂碼,如果設置為 假值 則自動識別
* @param bool|false $removeHead 【是否移除頁面頭部區域】 亂碼終極解決方案
* @return mixed
*/
public static function Query($page,array $rules, $range = '', $outputEncoding = null, $inputEncoding = null,$removeHead = false)
{
return self::getInstance()->_query($page, $rules, $range, $outputEncoding, $inputEncoding,$removeHead);
}
/**
* 運行QueryList擴展
* @param $class
* @param array $args
* @return mixed
* @throws Exception
*/
public static function run($class,$args = array())
{
$extension = self::getInstance("QL\\Ext\\{$class}");
return $extension->run($args);
}
/**
* 日志設置
* @param $handler
*/
public static function setLog($handler)
{
if(class_exists(Logger::class))
{
if(is_string($handler))
{
$handler = new StreamHandler($handler,Logger::INFO);
}
self::$logger = new Logger('QueryList');
self::$logger->pushHandler($handler);
}else{
throw new Exception("You need to install the package [monolog/monolog]");
}
}
/**
* 獲取任意實例
* @return mixed
* @throws Exception
*/
public static function getInstance()
{
$args = func_get_args();
count($args) || $args = array('QL\QueryList');
$key = md5(serialize($args));
$className = array_shift($args);
if(!class_exists($className)) {
throw new Exception("no class {$className}");
}
if(!isset(self::$instances[$key])) {
$rc = new ReflectionClass($className);
self::$instances[$key] = $rc->newInstanceArgs($args);
}
return self::$instances[$key];
}
/**
* 獲取目標頁面源碼(主要用于調試)
* @param bool|true $rel
* @return string
*/
public function getHtml($rel = true)
{
return $rel?$this->qpHtml:$this->html;
}
/**
* 獲取采集結果數據
* @param callback $callback
* @return array
*/
public function getData($callback = null)
{
if(is_callable($callback)){
return array_map($callback,$this->data);
}
return $this->data;
}
/**
* 重新設置選擇器
* @param $rules
* @param string $range
* @param string $outputEncoding
* @param string $inputEncoding
* @param bool|false $removeHead
* @return QueryList
*/
public function setQuery(array $rules, $range = '',$outputEncoding = null, $inputEncoding = null,$removeHead = false)
{
return $this->_query($this->html,$rules, $range, $outputEncoding, $inputEncoding,$removeHead);
}
private function _query($page,array $rules, $range, $outputEncoding, $inputEncoding,$removeHead)
{
$this->data = array();
$this->page = $page;
$this->html = $this->_isURL($this->page)?$this->_request($this->page):$this->page;
$outputEncoding && $this->outputEncoding = $outputEncoding;
$inputEncoding && $this->inputEncoding = $inputEncoding;
$removeHead && $this->html = $this->_removeHead($this->html);
$this->pqHtml = '';
if(empty($this->html)){
$this->_log('The received content is empty!','error');
trigger_error('The received content is empty!',E_USER_NOTICE);
}
//獲取編碼格式
$this->htmlEncoding = $this->inputEncoding?$this->inputEncoding:$this->_getEncode($this->html);
// $this->html = $this->_removeTags($this->html,array('script','style'));
$this->regArr = $rules;
$this->regRange = $range;
$this->_getList();
$this->_log();
return $this;
}
private function _getList()
{
$this->inputEncoding && phpQuery::$defaultCharset = $this->inputEncoding;
$document = phpQuery::newDocumentHTML($this->html);
$this->qpHtml = $document->htmlOuter();
if (!empty($this->regRange)) {
$robj = pq($document)->find($this->regRange);
$i = 0;
foreach ($robj as $item) {
while (list($key, $reg_value) = each($this->regArr)) {
if($key=='callback')continue;
$tags = isset($reg_value[2])?$reg_value[2]:'';
$iobj = pq($item)->find($reg_value[0]);
switch ($reg_value[1]) {
case 'text':
$this->data[$i][$key] = $this->_allowTags(pq($iobj)->html(),$tags);
break;
case 'html':
$this->data[$i][$key] = $this->_stripTags(pq($iobj)->html(),$tags);
break;
default:
$this->data[$i][$key] = pq($iobj)->attr($reg_value[1]);
break;
}
if(isset($reg_value[3])){
$this->data[$i][$key] = call_user_func($reg_value[3],$this->data[$i][$key],$key);
}else if(isset($this->regArr['callback'])){
$this->data[$i][$key] = call_user_func($this->regArr['callback'],$this->data[$i][$key],$key);
}
}
//重置數組指針
reset($this->regArr);
$i++;
}
} else {
while (list($key, $reg_value) = each($this->regArr)) {
if($key=='callback')continue;
$document = phpQuery::newDocumentHTML($this->html);
$tags = isset($reg_value[2])?$reg_value[2]:'';
$lobj = pq($document)->find($reg_value[0]);
$i = 0;
foreach ($lobj as $item) {
switch ($reg_value[1]) {
case 'text':
$this->data[$i][$key] = $this->_allowTags(pq($item)->html(),$tags);
break;
case 'html':
$this->data[$i][$key] = $this->_stripTags(pq($item)->html(),$tags);
break;
default:
$this->data[$i][$key] = pq($item)->attr($reg_value[1]);
break;
}
if(isset($reg_value[3])){
$this->data[$i][$key] = call_user_func($reg_value[3],$this->data[$i][$key],$key);
}else if(isset($this->regArr['callback'])){
$this->data[$i][$key] = call_user_func($this->regArr['callback'],$this->data[$i][$key],$key);
}
$i++;
}
}
}
if ($this->outputEncoding) {
//編碼轉換
$this->data = $this->_arrayConvertEncoding($this->data, $this->outputEncoding, $this->htmlEncoding);
}
phpQuery::$documents = array();
}
/**
* URL請求
* @param $url
* @return string
*/
private function _request($url)
{
if(function_exists('curl_init')){
$ch = curl_init();
curl_setopt($ch, CURLOPT_URL, $url);
curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, false);
curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
curl_setopt($ch, CURLOPT_AUTOREFERER, true);
curl_setopt($ch, CURLOPT_REFERER, $url);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
curl_setopt($ch, CURLOPT_USERAGENT, 'Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/46.0.2490.71 Safari/537.36');
$result = curl_exec($ch);
curl_close($ch);
}elseif(version_compare(PHP_VERSION, '5.0.0')>=0){
$opts = array(
'http' => array(
'header' => "Referer:{$url}"
)
);
$result = file_get_contents($url,false,stream_context_create($opts));
}else{
$result = file_get_contents($url);
}
return $result;
}
/**
* 移除頁面head區域代碼
* @param $html
* @return mixed
*/
private function _removeHead($html)
{
return preg_replace('/
.+/is','',$html);}
/**
* 獲取文件編碼
* @param $string
* @return string
*/
private function _getEncode($string)
{
return mb_detect_encoding($string, array('ASCII', 'GB2312', 'GBK', 'UTF-8'));
}
/**
* 轉換數組值的編碼格式
* @param array $arr
* @param string $toEncoding
* @param string $fromEncoding
* @return array
*/
private function _arrayConvertEncoding($arr, $toEncoding, $fromEncoding)
{
eval('$arr = '.iconv($fromEncoding, $toEncoding.'//IGNORE', var_export($arr,TRUE)).';');
return $arr;
}
/**
* 簡單的判斷一下參數是否為一個URL鏈接
* @param string $str
* @return boolean
*/
private function _isURL($str)
{
if (preg_match('/^http(s)?:\\/\\/.+/', $str)) {
return true;
}
return false;
}
/**
* 去除特定的html標簽
* @param string $html
* @param string $tags_str 多個標簽名之間用空格隔開
* @return string
*/
private function _stripTags($html,$tags_str)
{
$tagsArr = $this->_tag($tags_str);
$html = $this->_removeTags($html,$tagsArr[1]);
$p = array();
foreach ($tagsArr[0] as $tag) {
$p[]="/(]*>)/i";
}
$html = preg_replace($p,"",trim($html));
return $html;
}
/**
* 保留特定的html標簽
* @param string $html
* @param string $tags_str 多個標簽名之間用空格隔開
* @return string
*/
private function _allowTags($html,$tags_str)
{
$tagsArr = $this->_tag($tags_str);
$html = $this->_removeTags($html,$tagsArr[1]);
$allow = '';
foreach ($tagsArr[0] as $tag) {
$allow .= " ";
}
return strip_tags(trim($html),$allow);
}
private function _tag($tags_str)
{
$tagArr = preg_split("/\s+/",$tags_str,-1,PREG_SPLIT_NO_EMPTY);
$tags = array(array(),array());
foreach($tagArr as $tag)
{
if(preg_match('/-(.+)/', $tag,$arr))
{
array_push($tags[1], $arr[1]);
}else{
array_push($tags[0], $tag);
}
}
return $tags;
}
/**
* 移除特定的html標簽
* @param string $html
* @param array $tags 標簽數組
* @return string
*/
private function _removeTags($html,$tags)
{
$tag_str = '';
if(count($tags))
{
foreach ($tags as $tag) {
$tag_str .= $tag_str?','.$tag:$tag;
}
phpQuery::$defaultCharset = $this->inputEncoding?$this->inputEncoding:$this->htmlEncoding;
$doc = phpQuery::newDocumentHTML($html);
pq($doc)->find($tag_str)->remove();
$html = pq($doc)->htmlOuter();
$doc->unloadDocument();
}
return $html;
}
/**
* 打印日志
* @param string $message
* @param string $level
*/
private function _log($message = '',$level = 'info')
{
if(!is_null(self::$logger))
{
$url = $this->_isURL($this->page)?$this->page:'[html]';
$count = count($this->data);
$level = empty($level)?($count?'info':'warning'):$level;
$message = empty($message)?($count?'Get data successfully':'Get data failed'):$message;
self::$logger->$level($message,array(
'page' => $url,
'count' => $count
));
}
}
}
/*
class Autoload
{
public static function load($className)
{
$files = array(
sprintf('%s/extensions/%s.php',__DIR__,$className),
sprintf('%s/extensions/vendors/%s.php',__DIR__,$className)
);
foreach ($files as $file) {
if(is_file($file)){
require $file;
return true;
}
}
return false;
}
}
spl_autoload_register(array('Autoload','load'));
*/
總結
以上是生活随笔為你收集整理的php queryList函数,QueryList/QueryList.php at master · baijunyao/QueryList · GitHub的全部內容，希望文章能夠幫你解決所遇到的問題。
- 上一篇： oracle ora 00279,ORA
- 下一篇： 动态规划算法php,php算法学习之动态