From 2013e4d2b03d7f5055a7260bbf43164ba9d0a562 Mon Sep 17 00:00:00 2001
From: Jaeger <JaegerCode@gmail.com>
Date: Fri, 22 Sep 2017 01:51:46 +0800
Subject: [PATCH] add Query
---
src/Dom/Elements.php | 132 ++++++++++++++++++++
src/Dom/Query.php | 206 +++++++++++++++++++++++++++++++
src/Kernel.php | 2 +-
src/QueryList.php | 217 ++++++---------------------------
src/Services/EncodeService.php | 23 +++-
5 files changed, 395 insertions(+), 185 deletions(-)
create mode 100644 src/Dom/Query.php
diff --git a/src/Dom/Elements.php b/src/Dom/Elements.php
index d093886..d778f96 100644
--- a/src/Dom/Elements.php
+++ b/src/Dom/Elements.php
@@ -9,8 +9,140 @@ namespace QL\Dom;
use phpQueryObject;
+/**
+ * Class Elements
+ * @package QL\Dom
+ *
+ * @method Elements toReference($var)
+ * @method documentFragment($state)
+ * @method Elements toRoot()
+ * @method Elements getDocumentIDRef($documentID)
+ * @method Elements getDocument()
+ * @method getDOMDocument()
+ * @method Elements getDocumentID()
+ * @method Elements unloadDocument()
+ * @method isHTML()
+ * @method isXHTML()
+ * @method isXML()
+ * @method serialize()
+ * @method serializeArray($submit)
+ * @method get($index,$callback1,$callback2,$callback3)
+ * @method getString($index,$callback1,$callback2,$callback3)
+ * @method getStrings($index,$callback1,$callback2,$callback3)
+ * @method newInstance($newStack)
+ * @method Elements find($selectors,$context,$noHistory)
+ * @method Elements is($selector,$nodes)
+ * @method Elements filterCallback($callback,$_skipHistory)
+ * @method Elements filter($selectors,$_skipHistory)
+ * @method load($url,$data,$callback)
+ * @method css()
+ * @method show()
+ * @method hide()
+ * @method Elements trigger($type,$data)
+ * @method Elements triggerHandler($type,$data)
+ * @method Elements bind($type,$data,$callback)
+ * @method unbind($type,$callback)
+ * @method Elements change($callback)
+ * @method Elements submit($callback)
+ * @method Elements click($callback)
+ * @method Elements wrapAllOld($wrapper)
+ * @method Elements wrapAll($wrapper)
+ * @method Elements wrapAllPHP($codeBefore,$codeAfter)
+ * @method Elements wrap($wrapper)
+ * @method Elements wrapPHP($codeBefore,$codeAfter)
+ * @method Elements wrapInner($wrapper)
+ * @method Elements wrapInnerPHP($codeBefore,$codeAfter)
+ * @method Elements contents()
+ * @method Elements contentsUnwrap()
+ * @method switchWith($markup)
+ * @method Elements eq($num)
+ * @method Elements size()
+ * @method Elements length()
+ * @method count()
+ * @method Elements end($level)
+ * @method Elements _clone()
+ * @method Elements replaceWithPHP($code)
+ * @method Elements replaceWith($content)
+ * @method Elements replaceAll($selector)
+ * @method Elements remove($selector)
+ * @method markup($markup,$callback1,$callback2,$callback3)
+ * @method markupOuter($callback1,$callback2,$callback3)
+ * @method html($html,$callback1,$callback2,$callback3)
+ * @method xml($xml,$callback1,$callback2,$callback3)
+ * @method htmlOuter($callback1,$callback2,$callback3)
+ * @method xmlOuter($callback1,$callback2,$callback3)
+ * @method Elements php($code)
+ * @method markupPHP($code)
+ * @method markupOuterPHP()
+ * @method Elements children($selector)
+ * @method Elements ancestors($selector)
+ * @method Elements append($content)
+ * @method Elements appendPHP($content)
+ * @method Elements appendTo($seletor)
+ * @method Elements prepend($content)
+ * @method Elements prependPHP($content)
+ * @method Elements prependTo($seletor)
+ * @method Elements before($content)
+ * @method Elements beforePHP($content)
+ * @method Elements insertBefore($seletor)
+ * @method Elements after($content)
+ * @method Elements afterPHP($content)
+ * @method Elements insertAfter($seletor)
+ * @method Elements insert($target,$type)
+ * @method index($subject)
+ * @method Elements slice($start,$end)
+ * @method Elements reverse()
+ * @method text($text,$callback1,$callback2,$callback3)
+ * @method Elements plugin($class,$file)
+ * @method extend($class,$file)
+ * @method Elements _next($selector)
+ * @method Elements _prev($selector)
+ * @method Elements prev($selector)
+ * @method Elements prevAll($selector)
+ * @method Elements nextAll($selector)
+ * @method Elements siblings($selector)
+ * @method Elements not($selector)
+ * @method Elements add($selector)
+ * @method Elements parent($selector)
+ * @method Elements parents($selector)
+ * @method stack($nodeTypes)
+ * @method attr($attr,$value)
+ * @method Elements attrPHP($attr,$code)
+ * @method Elements removeAttr($attr)
+ * @method val($val)
+ * @method Elements andSelf()
+ * @method Elements addClass($className)
+ * @method Elements addClassPHP($className)
+ * @method hasClass($className)
+ * @method Elements removeClass($className)
+ * @method Elements toggleClass($className)
+ * @method Elements _empty()
+ * @method Elements each($callback,$param1,$param2,$param3)
+ * @method Elements callback($callback,$param1,$param2,$param3)
+ * @method data($key,$value)
+ * @method removeData($key)
+ * @method rewind()
+ * @method current()
+ * @method key()
+ * @method Elements next($cssSelector)
+ * @method valid()
+ * @method offsetExists($offset)
+ * @method offsetGet($offset)
+ * @method offsetSet($offset,$value)
+ * @method offsetUnset($offset)
+ * @method whois($oneNode)
+ * @method Elements dump()
+ * @method dumpWhois()
+ * @method dumpLength()
+ * @method dumpTree($html,$title)
+ * @method dumpDie()
+ */
+
class Elements
{
+ /**
+ * @var phpQueryObject
+ */
protected $elements;
/**
diff --git a/src/Dom/Query.php b/src/Dom/Query.php
new file mode 100644
index 0000000..ad32c46
--- /dev/null
+++ b/src/Dom/Query.php
@@ -0,0 +1,206 @@
+<?php
+/**
+ * Created by PhpStorm.
+ * User: Jaeger <JaegerCode@gmail.com>
+ * Date: 2017/9/21
+ */
+
+namespace QL\Dom;
+
+use phpQuery;
+use QL\QueryList;
+
+class Query
+{
+ protected $html;
+ protected $document;
+ protected $rules;
+ protected $range = null;
+ protected $ql;
+
+
+ public function __construct(QueryList $ql)
+ {
+ $this->ql = $ql;
+ }
+
+
+ /**
+ * @return mixed
+ */
+ public function getHtml()
+ {
+ return $this->html;
+ }
+
+ public function setHtml($html)
+ {
+ $this->html = $html;
+ $this->document = phpQuery::newDocumentHTML($this->html);
+ return $this->ql;
+ }
+
+ public function find($selector)
+ {
+ return (new Dom($this->document))->find($selector);
+ }
+
+ public function rules(array $rules)
+ {
+ $this->rules = $rules;
+ return $this->ql;
+ }
+
+
+ public function range($range)
+ {
+ $this->range = $range;
+ return $this->ql;
+ }
+
+ public function removeHead()
+ {
+ $html = preg_replace('/<head.+?>.+<\/head>/is','<head></head>',$this->html);
+ $this->setHtml($html);
+ return $this->ql;
+ }
+
+ public function query($callback = null)
+ {
+ $data = $this->getList();
+ return is_null($callback)?$data:$data->map($callback);
+ }
+
+ protected function getList()
+ {
+ $data = [];
+ $document = $this->document;
+ if (!empty($this->range)) {
+ $robj = pq($document)->find($this->range);
+ $i = 0;
+ foreach ($robj as $item) {
+ foreach ($this->rules as $key => $reg_value){
+ $tags = $reg_value[2] ?? '';
+ $iobj = pq($item)->find($reg_value[0]);
+ switch ($reg_value[1]) {
+ case 'text':
+ $data[$i][$key] = $this->allowTags(pq($iobj)->html(),$tags);
+ break;
+ case 'html':
+ $data[$i][$key] = $this->stripTags(pq($iobj)->html(),$tags);
+ break;
+ default:
+ $data[$i][$key] = pq($iobj)->attr($reg_value[1]);
+ break;
+ }
+
+ if(isset($reg_value[3])){
+ $data[$i][$key] = call_user_func($reg_value[3],$data[$i][$key],$key);
+ }
+ }
+ $i++;
+ }
+ } else {
+ foreach ($this->rules as $key => $reg_value){
+ $tags = $reg_value[2] ?? '';
+ $lobj = pq($document)->find($reg_value[0]);
+ $i = 0;
+ foreach ($lobj as $item) {
+ switch ($reg_value[1]) {
+ case 'text':
+ $data[$i][$key] = $this->allowTags(pq($item)->html(),$tags);
+ break;
+ case 'html':
+ $data[$i][$key] = $this->stripTags(pq($item)->html(),$tags);
+ break;
+ default:
+ $data[$i][$key] = pq($item)->attr($reg_value[1]);
+ break;
+ }
+
+ if(isset($reg_value[3])){
+ $data[$i][$key] = call_user_func($reg_value[3],$data[$i][$key],$key);
+ }
+
+ $i++;
+ }
+ }
+ }
+// phpQuery::$documents = array();
+ return collect($data);
+ }
+
+ /**
+ * 去除特定的html标签
+ * @param string $html
+ * @param string $tags_str 多个标签名之间用空格隔开
+ * @return string
+ */
+ protected function stripTags($html,$tags_str)
+ {
+ $tagsArr = $this->tag($tags_str);
+ $html = $this->removeTags($html,$tagsArr[1]);
+ $p = array();
+ foreach ($tagsArr[0] as $tag) {
+ $p[]="/(<(?:\/".$tag."|".$tag.")[^>]*>)/i";
+ }
+ $html = preg_replace($p,"",trim($html));
+ return $html;
+ }
+
+ /**
+ * 保留特定的html标签
+ * @param string $html
+ * @param string $tags_str 多个标签名之间用空格隔开
+ * @return string
+ */
+ protected function allowTags($html,$tags_str)
+ {
+ $tagsArr = $this->tag($tags_str);
+ $html = $this->removeTags($html,$tagsArr[1]);
+ $allow = '';
+ foreach ($tagsArr[0] as $tag) {
+ $allow .= "<$tag> ";
+ }
+ return strip_tags(trim($html),$allow);
+ }
+
+ protected 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标签
+ * @param string $html
+ * @param array $tags 标签数组
+ * @return string
+ */
+ protected function removeTags($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;
+ }
+}
\ No newline at end of file
diff --git a/src/Kernel.php b/src/Kernel.php
index 616686e..c2c004b 100644
--- a/src/Kernel.php
+++ b/src/Kernel.php
@@ -49,7 +49,7 @@ class Kernel
$this->binds[$name] = $provider;
}
- public function getBind(string $name)
+ public function getService(string $name)
{
if(!$this->binds->offsetExists($name)){
throw new ServiceNotFoundException("Service: {$name} not found!");
diff --git a/src/QueryList.php b/src/QueryList.php
index a0dac3b..b81b042 100644
--- a/src/QueryList.php
+++ b/src/QueryList.php
@@ -1,8 +1,4 @@
<?php
-namespace QL;
-use phpQuery;
-use QL\Dom\Dom;
-
/**
* QueryList
*
@@ -15,12 +11,27 @@ use QL\Dom\Dom;
*
*/
+namespace QL;
+use phpQuery;
+use QL\Dom\Query;
+
+
+/**
+ * Class QueryList
+ * @package QL
+ *
+ * @method QueryList getHtml()
+ * @method QueryList setHtml($html)
+ * @method Dom\Elements find($selector)
+ * @method QueryList rules(array $rules)
+ * @method QueryList range($range)
+ * @method QueryList removeHead()
+ * @method \Illuminate\Support\Collection query($callback = null)
+ * @method QueryList encoding(string $outputEncoding,string $inputEncoding = null)
+ */
class QueryList
{
- protected $html;
- protected $document;
- protected $rules;
- protected $range = null;
+ protected $query;
protected $kernel;
/**
@@ -28,195 +39,41 @@ class QueryList
*/
public function __construct()
{
+ $this->query = new Query($this);
$this->kernel = (new Kernel($this))->bootstrap();
}
public function __call($name, $arguments)
{
- return $this->kernel->getBind($name)->call($this,...$arguments);
- }
-
-
- /**
- * @return mixed
- */
- public function getHtml()
- {
- return $this->html;
- }
-
- /**
- * @param $html
- * @return $this
- */
- public function setHtml($html)
- {
- $this->html = $html;
- $this->document = phpQuery::newDocumentHTML($this->html);
- return $this;
- }
-
- public function find($selector)
- {
- return (new Dom($this->document))->find($selector);
- }
-
- public function rules(array $rules)
- {
- $this->rules = $rules;
- return $this;
- }
-
-
- public function range($range)
- {
- $this->range = $range;
- return $this;
- }
-
- public function removeHead()
- {
- $html = preg_replace('/<head.+?>.+<\/head>/is','<head></head>',$this->html);
- $this->setHtml($html);
- return $this;
- }
-
- public function query($callback = null)
- {
- $data = $this->getList();
- return is_null($callback)?$data:$data->map($callback);
- }
-
- protected function getList()
- {
- $data = [];
- $document = $this->document;
- if (!empty($this->range)) {
- $robj = pq($document)->find($this->range);
- $i = 0;
- foreach ($robj as $item) {
- foreach ($this->rules as $key => $reg_value){
- $tags = $reg_value[2] ?? '';
- $iobj = pq($item)->find($reg_value[0]);
- switch ($reg_value[1]) {
- case 'text':
- $data[$i][$key] = $this->allowTags(pq($iobj)->html(),$tags);
- break;
- case 'html':
- $data[$i][$key] = $this->stripTags(pq($iobj)->html(),$tags);
- break;
- default:
- $data[$i][$key] = pq($iobj)->attr($reg_value[1]);
- break;
- }
-
- if(isset($reg_value[3])){
- $data[$i][$key] = call_user_func($reg_value[3],$data[$i][$key],$key);
- }
- }
- $i++;
- }
- } else {
- foreach ($this->rules as $key => $reg_value){
- $tags = $reg_value[2] ?? '';
- $lobj = pq($document)->find($reg_value[0]);
- $i = 0;
- foreach ($lobj as $item) {
- switch ($reg_value[1]) {
- case 'text':
- $data[$i][$key] = $this->allowTags(pq($item)->html(),$tags);
- break;
- case 'html':
- $data[$i][$key] = $this->stripTags(pq($item)->html(),$tags);
- break;
- default:
- $data[$i][$key] = pq($item)->attr($reg_value[1]);
- break;
- }
-
- if(isset($reg_value[3])){
- $data[$i][$key] = call_user_func($reg_value[3],$data[$i][$key],$key);
- }
-
- $i++;
- }
- }
+ if(method_exists($this->query,$name)){
+ $result = $this->query->$name(...$arguments);
+ }else{
+ $result = $this->kernel->getService($name)->call($this,...$arguments);
}
- phpQuery::$documents = array();
- return collect($data);
+ return $result;
}
- /**
- * 去除特定的html标签
- * @param string $html
- * @param string $tags_str 多个标签名之间用空格隔开
- * @return string
- */
- protected function stripTags($html,$tags_str)
+ public static function __callStatic($name, $arguments)
{
- $tagsArr = $this->tag($tags_str);
- $html = $this->removeTags($html,$tagsArr[1]);
- $p = array();
- foreach ($tagsArr[0] as $tag) {
- $p[]="/(<(?:\/".$tag."|".$tag.")[^>]*>)/i";
- }
- $html = preg_replace($p,"",trim($html));
- return $html;
+ $instance = self::getInstance();
+ return $instance->$name(...$arguments);
}
- /**
- * 保留特定的html标签
- * @param string $html
- * @param string $tags_str 多个标签名之间用空格隔开
- * @return string
- */
- protected function allowTags($html,$tags_str)
+ public function __destruct()
{
- $tagsArr = $this->tag($tags_str);
- $html = $this->removeTags($html,$tagsArr[1]);
- $allow = '';
- foreach ($tagsArr[0] as $tag) {
- $allow .= "<$tag> ";
- }
- return strip_tags(trim($html),$allow);
+ $this->destruct();
}
- protected function tag($tags_str)
+ public static function getInstance()
{
- $tagArr = preg_split("/\s+/",$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;
+ $instance = new self();
+ return $instance;
}
- /**
- * 移除特定的html标签
- * @param string $html
- * @param array $tags 标签数组
- * @return string
- */
- protected function removeTags($html,$tags)
+ public function destruct()
{
- $tag_str = '';
- if(count($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 $html;
+ phpQuery::$documents = [];
}
+
+
}
\ No newline at end of file
diff --git a/src/Services/EncodeService.php b/src/Services/EncodeService.php
index 2c5a612..6f03255 100644
--- a/src/Services/EncodeService.php
+++ b/src/Services/EncodeService.php
@@ -3,20 +3,35 @@
* Created by PhpStorm.
* User: Jaeger <JaegerCode@gmail.com>
* Date: 2017/9/20
+ * 编码转换服务
*/
namespace QL\Services;
+use QL\QueryList;
+
class EncodeService
{
- public static function convert($ql,string $outputEncoding,string $inputEncoding = null)
+ public static function convert(QueryList $ql,string $outputEncoding,string $inputEncoding = null)
{
- dump($outputEncoding,$inputEncoding);
+ $html = $ql->getHtml();
+ $inputEncoding || $inputEncoding = self::detect($html);
+ $html = iconv($inputEncoding,$outputEncoding.'//IGNORE',$html);
+ $ql->setHtml($html);
return $ql;
}
- public static function detect()
+ /**
+ * Attempts to detect the encoding
+ * @param $string
+ * @return bool|false|mixed|string
+ */
+ public static function detect($string)
{
-
+ $charset=mb_detect_encoding($string, array('ASCII', 'GB2312', 'GBK', 'UTF-8'),true);
+ if(strtolower($charset)=='cp936')
+ $charset='GBK';
+ return $charset;
}
+
}
\ No newline at end of file