XiaoMo/QueryList
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

fix: data callback

Browse Source
This commit is contained in:
Jaeger 2020-04-01 22:03:50 +08:00
parent c16826a573
commit faea883c6f
7 changed files with 58 additions and 45 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
composer.json
View File

@ -32,6 +32,9 @@
}, },
"require-dev": { "require-dev": {
"symfony/var-dumper": "^3.3", "symfony/var-dumper": "^3.3",
"phpunit/phpunit": "^7.5" "phpunit/phpunit": "^8.5"
},
"scripts": {
"test": "./vendor/bin/phpunit"
} }
} }

82
src/Dom/Query.php
View File

@ -52,7 +52,7 @@ class Query
{ {
$this->html = value($html); $this->html = value($html);
$this->destroyDocument(); $this->destroyDocument();
$this->document = phpQuery::newDocumentHTML($this->html,$charset); $this->document = phpQuery::newDocumentHTML($this->html, $charset);
return $this->ql; return $this->ql;
} }
@ -64,7 +64,7 @@ class Query
*/ */
public function getData(Closure $callback = null) public function getData(Closure $callback = null)
{ {
return is_null($callback) ? $this->data : $this->data->map($callback); return $this->handleData($this->data, $callback);
} }
/** /**
@ -125,7 +125,7 @@ class Query
*/ */
public function removeHead() public function removeHead()
{ {
$html = preg_replace('/<head.+?>.+<\/head>/is','<head></head>',$this->html); $html = preg_replace('/<head.+?>.+<\/head>/is', '<head></head>', $this->html);
$this->setHtml($html); $this->setHtml($html);
return $this->ql; return $this->ql;
} }
@ -139,24 +139,37 @@ class Query
public function query(Closure $callback = null) public function query(Closure $callback = null)
{ {
$this->data = $this->getList(); $this->data = $this->getList();
$callback && $this->data = $this->data->map($callback); $this->data = $this->handleData($this->data, $callback);
return $this->ql; return $this->ql;
} }
public function handleData(Collection $data, $callback)
{
if (is_callable($callback)) {
if (empty($this->range)) {
$data = collect($callback($data->all(), null));
} else {
$data = $data->map($callback);
}
}
return $data;
}
protected function getList() protected function getList()
{ {
$data = []; $data = [];
if (empty($this->range)) { if (empty($this->range)) {
foreach ($this->rules as $key => $reg_value){ foreach ($this->rules as $key => $reg_value) {
$rule = $this->parseRule($reg_value); $rule = $this->parseRule($reg_value);
$contentElements = $this->document->find($rule['selector']); $contentElements = $this->document->find($rule['selector']);
$data[$key] = $this->extractContent($contentElements, $key, $rule); $data[$key] = $this->extractContent($contentElements, $key, $rule);
} }
} else { } else {
$rangeElements = $this->document->find($this->range); $rangeElements = $this->document->find($this->range);
$i = 0; $i = 0;
foreach ($rangeElements as $element) { foreach ($rangeElements as $element) {
foreach ($this->rules as $key => $reg_value){ foreach ($this->rules as $key => $reg_value) {
$rule = $this->parseRule($reg_value); $rule = $this->parseRule($reg_value);
$contentElements = pq($element)->find($rule['selector']); $contentElements = pq($element)->find($rule['selector']);
$data[$i][$key] = $this->extractContent($contentElements, $key, $rule); $data[$i][$key] = $this->extractContent($contentElements, $key, $rule);
@ -175,7 +188,7 @@ class Query
$content = $this->allowTags($pqObj->html(), $rule['filter_tags']); $content = $this->allowTags($pqObj->html(), $rule['filter_tags']);
break; break;
case 'texts': case 'texts':
$content = (new Elements($pqObj))->map(function(Elements $element) use($rule){ $content = (new Elements($pqObj))->map(function (Elements $element) use ($rule) {
return $this->allowTags($element->html(), $rule['filter_tags']); return $this->allowTags($element->html(), $rule['filter_tags']);
})->all(); })->all();
break; break;
@ -183,7 +196,7 @@ class Query
$content = $this->stripTags($pqObj->html(), $rule['filter_tags']); $content = $this->stripTags($pqObj->html(), $rule['filter_tags']);
break; break;
case 'htmls': case 'htmls':
$content = (new Elements($pqObj))->map(function(Elements $element) use($rule){ $content = (new Elements($pqObj))->map(function (Elements $element) use ($rule) {
return $this->stripTags($element->html(), $rule['filter_tags']); return $this->stripTags($element->html(), $rule['filter_tags']);
})->all(); })->all();
break; break;
@ -191,7 +204,7 @@ class Query
$content = $this->stripTags($pqObj->htmlOuter(), $rule['filter_tags']); $content = $this->stripTags($pqObj->htmlOuter(), $rule['filter_tags']);
break; break;
case 'htmlOuters': case 'htmlOuters':
$content = (new Elements($pqObj))->map(function(Elements $element) use($rule){ $content = (new Elements($pqObj))->map(function (Elements $element) use ($rule) {
return $this->stripTags($element->htmlOuter(), $rule['filter_tags']); return $this->stripTags($element->htmlOuter(), $rule['filter_tags']);
})->all(); })->all();
break; break;
@ -200,7 +213,7 @@ class Query
break; break;
} }
if(is_callable($rule['handle_callback'])){ if (is_callable($rule['handle_callback'])) {
$content = call_user_func($rule['handle_callback'], $content, $ruleName); $content = call_user_func($rule['handle_callback'], $content, $ruleName);
} }
@ -220,49 +233,47 @@ class Query
/** /**
* 去除特定的html标签 * 去除特定的html标签
* @param string $html * @param string $html
* @param string $tags_str 多个标签名之间用空格隔开 * @param string $tags_str 多个标签名之间用空格隔开
* @return string * @return string
*/ */
protected function stripTags($html,$tags_str) protected function stripTags($html, $tags_str)
{ {
$tagsArr = $this->tag($tags_str); $tagsArr = $this->tag($tags_str);
$html = $this->removeTags($html,$tagsArr[1]); $html = $this->removeTags($html, $tagsArr[1]);
$p = array(); $p = array();
foreach ($tagsArr[0] as $tag) { foreach ($tagsArr[0] as $tag) {
$p[]="/(<(?:\/".$tag."|".$tag.")[^>]*>)/i"; $p[] = "/(<(?:\/" . $tag . "|" . $tag . ")[^>]*>)/i";
} }
$html = preg_replace($p,"",trim($html)); $html = preg_replace($p, "", trim($html));
return $html; return $html;
} }
/** /**
* 保留特定的html标签 * 保留特定的html标签
* @param string $html * @param string $html
* @param string $tags_str 多个标签名之间用空格隔开 * @param string $tags_str 多个标签名之间用空格隔开
* @return string * @return string
*/ */
protected function allowTags($html,$tags_str) protected function allowTags($html, $tags_str)
{ {
$tagsArr = $this->tag($tags_str); $tagsArr = $this->tag($tags_str);
$html = $this->removeTags($html,$tagsArr[1]); $html = $this->removeTags($html, $tagsArr[1]);
$allow = ''; $allow = '';
foreach ($tagsArr[0] as $tag) { foreach ($tagsArr[0] as $tag) {
$allow .= "<$tag> "; $allow .= "<$tag> ";
} }
return strip_tags(trim($html),$allow); return strip_tags(trim($html), $allow);
} }
protected function tag($tags_str) protected function tag($tags_str)
{ {
$tagArr = preg_split("/\s+/",$tags_str,-1,PREG_SPLIT_NO_EMPTY); $tagArr = preg_split("/\s+/", $tags_str, -1, PREG_SPLIT_NO_EMPTY);
$tags = array(array(),array()); $tags = array(array(), array());
foreach($tagArr as $tag) foreach ($tagArr as $tag) {
{ if (preg_match('/-(.+)/', $tag, $arr)) {
if(preg_match('/-(.+)/', $tag,$arr))
{
array_push($tags[1], $arr[1]); array_push($tags[1], $arr[1]);
}else{ } else {
array_push($tags[0], $tag); array_push($tags[0], $tag);
} }
} }
@ -271,17 +282,16 @@ class Query
/** /**
* 移除特定的html标签 * 移除特定的html标签
* @param string $html * @param string $html
* @param array $tags 标签数组 * @param array $tags 标签数组
* @return string * @return string
*/ */
protected function removeTags($html,$tags) protected function removeTags($html, $tags)
{ {
$tag_str = ''; $tag_str = '';
if(count($tags)) if (count($tags)) {
{
foreach ($tags as $tag) { foreach ($tags as $tag) {
$tag_str .= $tag_str?','.$tag:$tag; $tag_str .= $tag_str ? ',' . $tag : $tag;
} }
// phpQuery::$defaultCharset = $this->inputEncoding?$this->inputEncoding:$this->htmlEncoding; // phpQuery::$defaultCharset = $this->inputEncoding?$this->inputEncoding:$this->htmlEncoding;
$doc = phpQuery::newDocumentHTML($html); $doc = phpQuery::newDocumentHTML($html);
@ -294,7 +304,7 @@ class Query
protected function destroyDocument() protected function destroyDocument()
{ {
if($this->document instanceof phpQueryObject) { if ($this->document instanceof phpQueryObject) {
$this->document->unloadDocument(); $this->document->unloadDocument();
} }
} }

2
tests/Dom/FindTest.php
View File

@ -17,7 +17,7 @@ class FindTest extends TestCaseBase
protected $html; protected $html;
protected $ql; protected $ql;
public function setUp() protected function setUp(): void
{ {
$this->html = $this->getSnippet('snippet-1'); $this->html = $this->getSnippet('snippet-1');
$this->ql = QueryList::html($this->html); $this->ql = QueryList::html($this->html);

2
tests/Dom/RulesTest.php
View File

@ -18,7 +18,7 @@ class RulesTest extends TestCaseBase
protected $html; protected $html;
protected $ql; protected $ql;
public function setUp() protected function setUp(): void
{ {
$this->html = $this->getSnippet('snippet-2'); $this->html = $this->getSnippet('snippet-2');
$this->ql = QueryList::html($this->html); $this->ql = QueryList::html($this->html);

2
tests/Feature/HttpTest.php
View File

@ -18,7 +18,7 @@ class HttpTest extends TestCaseBase
{ {
protected $urls; protected $urls;
public function setUp() protected function setUp(): void
{ {
$this->urls = [ $this->urls = [
'http://httpbin.org/get?name=php', 'http://httpbin.org/get?name=php',

6
tests/Feature/InstanceTest.php
View File

@ -16,7 +16,7 @@ class InstanceTest extends TestCaseBase
{ {
protected $html; protected $html;
public function setUp() protected function setUp(): void
{ {
$this->html = $this->getSnippet('snippet-1'); $this->html = $this->getSnippet('snippet-1');
} }
@ -38,11 +38,11 @@ class InstanceTest extends TestCaseBase
public function get_new_object() public function get_new_object()
{ {
$ql = (new QueryList())->html($this->html); $ql = (new QueryList())->html($this->html);
$ql2 = new QueryList(); $ql2 = (new QueryList())->html('');
$this->assertNotEquals($ql->getHtml(),$ql2->getHtml()); $this->assertNotEquals($ql->getHtml(),$ql2->getHtml());
$ql = QueryList::range('')->html($this->html); $ql = QueryList::range('')->html($this->html);
$ql2 = QueryList::range(''); $ql2 = QueryList::range('')->html('');
$this->assertNotEquals($ql->getHtml(),$ql2->getHtml()); $this->assertNotEquals($ql->getHtml(),$ql2->getHtml());
} }
} }

4
tests/Feature/MethodTest.php
View File

@ -16,7 +16,7 @@ class MethodTest extends TestCaseBase
{ {
protected $html; protected $html;
public function setUp() protected function setUp(): void
{ {
$this->html = $this->getSnippet('snippet-1'); $this->html = $this->getSnippet('snippet-1');
} }
@ -30,7 +30,7 @@ class MethodTest extends TestCaseBase
$qlHtml = QueryList::pipe(function(QueryList $ql) use($html){ $qlHtml = QueryList::pipe(function(QueryList $ql) use($html){
$ql->setHtml($html); $ql->setHtml($html);
return $ql; return $ql;
})->getHtml(); })->getHtml(false);
$this->assertEquals($html,$qlHtml); $this->assertEquals($html,$qlHtml);
} }
} }