(PHP 5, PHP 7, PHP 8)
DOMDocument::loadHTML — Load HTML from a string
The function parses the HTML contained in the string source.
Unlike loading XML, HTML does not have to be well-formed to load.
使用 Dom\HTMLDocument 来解析和处理现代 HTML 而不是 DOMDocument。
此函数使用 HTML 4 解析器解析输入。这跟现代 Web 浏览器使用的 HTML 5 的解析规则有所不同。根据输入，这可能会导致不同的 DOM 结构。因此，此函数不能安全地用于清理 HTML。
解析 HTML 时的行为取决于所使用的 libxml 版本，尤其是边缘条件和错误处理方面。对于符合
HTML5 规范的解析，请使用 PHP 8.4 中添加的 Dom\HTMLDocument::createFromString()
或 Dom\HTMLDocument::createFromFile()。
例如，某些 HTML 元素在遇到父元素时会默默关闭。HTML 4 和 HTML 5 中自动关闭父元素的规则不同，因此 DOMDocument 看到的最终 DOM 结构可能与 Web 浏览器看到的 DOM 结构不同，这可能使攻击者能够破坏最终的 HTML。
If an empty string is passed as the source,
a warning will be generated. This warning is not generated by libxml
and cannot be handled using libxml's error handling functions.
尽管非正确格式化的 HTML 仍应该被成功调入，但此函数会在遇到错误标记时产生 E_WARNING 错误。libxml 错误处理函数可以用来处理这类错误。
| 版本 | 说明 |
|---|---|
| 8.3.0 | This function now has a tentative bool return type. |
| 8.0.0 |
Calling this function statically will
now throw an Error.
Previously, an E_DEPRECATED was raised.
|
示例 #1 Creating a Document
<?php
$doc = new DOMDocument();
$doc->loadHTML("<html><body>Test<br></body></html>");
echo $doc->saveHTML();
?>
You can also load HTML as UTF-8 using this simple hack:<?php$doc = new DOMDocument();$doc->loadHTML('<?xml encoding="UTF-8">' . $html);// dirty fixforeach ($doc->childNodes as $item) if ($item->nodeType == XML_PI_NODE) $doc->removeChild($item); // remove hack$doc->encoding = 'UTF-8'; // insert proper?>If you are loading html content from any website, in "utf-8" encoding, when meta width content-type is not first child of HEAD, it would not be acknowledged by parser (encoding); So you can make this fix: function domLoadHTML($html) {$testDOM = new DOMDocument('1.0', 'UTF-8'); $testDOM->loadHTML($html); $charset = NULL; $searchInElemnt = function(&$item) use (&$searchInElemnt, &$charset) {if($item->childNodes) {foreach($item->childNodes as $childItem) {switch($childItem->nodeName) {case 'html': case 'head': $searchInElemnt($childItem); break; case 'meta': $attributes = array(); foreach ($childItem->attributes as $attr) {$attributes[mb_strtoupper($attr->localName)] = $attr->nodeValue; } if(array_key_exists('HTTP-EQUIV', $attributes) && (mb_strtoupper($attributes['HTTP-EQUIV']) == 'CONTENT-TYPE') && array_key_exists('CONTENT', $attributes) && preg_match('~[\s]*;[\s]*charset[\s]*=[\s]*([^\s]+)~', $attributes['CONTENT'], $matches)) {$charset = preg_replace('~[\s\']~', '', $matches[1]); } } } } }; $searchInElemnt($testDOM); if(isset($charset)) {$dom = new DOMDocument('1.0', $charset); $dom->loadHTML('<?xml encoding="'.$charset.'">'.$html); foreach ($dom->childNodes as $item) if($item->nodeType == XML_PI_NODE) {$dom->removeChild($item); } $dom->encoding = $charset; } else {$dom = $testDOM; } return $dom; };DOMDocument is very good at dealing with imperfect markup, but it throws warnings all over the place when it does. This isn't well documented here. The solution to this is to implement a separate aparatus for dealing with just these errors. Set libxml_use_internal_errors(true) before calling loadHTML. This will prevent errors from bubbling up to your default error handler. And you can then get at them (if you desire) using other libxml error functions. You can find more info here http://www.php.net/manual/en/ref.libxml.phpWhen using loadHTML() to process UTF-8 pages, you may meet the problem that the output of dom functions are not like the input. For example, if you want to get "Cạnh tranh", you will receive "Cạnh tranh". I suggest we use mb_convert_encoding before load UTF-8 page :
<?php
$pageDom = new DomDocument();
$searchPage = mb_convert_encoding($htmlUTF8Page, 'HTML-ENTITIES', "UTF-8");
@$pageDom->loadHTML($searchPage);
?>To support HTML5 you have to disable xml error handling by add `LIBXML_NOERROR` as an option of loadHTML method.Example:<?php$doc = new DOMDocument();$doc->loadHTML("<html><body>Test<br><section>I'M UNSUPPORTED</section></body></html>", LIBXML_NOERROR);echo $doc->saveHTML();?>Pay attention when loading html that has a different charset than iso-8859-1. Since this method does not actively try to figure out what the html you are trying to load is encoded in (like most browsers do), you have to specify it in the html head. If, for instance, your html is in utf-8, make sure you have a meta tag in the html's head section:<head><meta http-equiv="Content-Type" content="text/html; charset=utf-8"/></head>If you do not specify the charset like this, all high-ascii bytes will be html-encoded. It is not enough to set the dom document you are loading the html in to UTF-8.If we are loading html5 tags such as <section>, <svg> there is following error:DOMDocument::loadHTML(): Tag section invalid in EntityWe can disable standard libxml errors (and enable user error handling) using libxml_use_internal_errors(true); before loadHTML();This is quite useful in phpunit custom assertions as given in following example (if using phpunit test cases):// Create a DOMDocument$dom = new DOMDocument();// fix html5/svg errorslibxml_use_internal_errors(true); // Load html $dom->loadHTML("<section></section>");$htmlNodes = $dom->getElementsByTagName('section');if ($htmlNodes->length == 0) { $this->assertFalse(TRUE);} else { $this->assertTrue(TRUE);}loadHTML() & loadHTMLFile() may always generate warnings if the html include some tags such as "nav, section, footer, etc" adopted as of HTML5 (in PHP 8.1.6).Try to run below.<?php$file_name = 'PHP Runtime Configuration - Manual.html'; // Download this file from "https://www.php.net/manual/en/session.configuration.php" in advance.$doc = new DOMDocument();$doc->loadHTMLFile($file_name); // if set "LIBXML_NOERROR" as 2nd arg, no errorecho $doc->saveHTML();// Warning: DOMDocument::loadHTMLFile(): Tag nav invalid in PHP Runtime Configuration - Manual.html, line: 63 in D:\xampp\htdocs\test\xml(dom)\loadHTML\index.php on line 6?>Warning: This does not function well with HTML5 elements such as SVG. Most of the advice on the Web is to turn off errors in order to have it work with HTML5.Remember: If you use an HTML5 doctype and a meta element like so<meta charset=utf-8">your HTML code will get interpreted as ISO-8859-something and non-ASCII chars will get converted into HTML entities. However the HTML4-like version will work (as has been pointed out 10 years ago by "bigtree at 29a"):<meta http-equiv="Content-Type" content="text/html; charset=utf-8">Be aware that this function doesn't actually understand HTML -- it fixes tag-soup input using the general rules of SGML, so it creates well-formed markup, but has no idea which element contexts are allowed.For example, with input like this where the first element isn't closed: <span>hello <div>world</div>loadHTML will change it to this, which is well-formed but invalid: <span>hello <div>world</div></span>Note that the elements of such document will have no namespace even with <html xmlns="http://www.w3.org/1999/xhtml">It should be noted that when any text is provided within the body tag
outside of a containing element, the DOMDocument will encapsulate that
text into a paragraph tag (<p>).
For example:
<?php
$doc = new DOMDocument();
$doc->loadHTML("<html><body>Test<br><div>Text</div></body></html>");
echo $doc->saveHTML();
?>
will yield:
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN"
"http://www.w3.org/TR/REC-html40/loose.dtd">
<html><body>
<p>Test<br></p>
<div>Text</div>
</body></html>
while:
<?php
$doc = new DOMDocument();
$doc->loadHTML(
"<html><body><i>Test</i><br><div>Text</div></body></html>");
echo $doc->saveHTML();
?>
will yield:
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN"
"http://www.w3.org/TR/REC-html40/loose.dtd">
<html><body>
<i>Test</i><br><div>Text</div>
</body></html>For those of you who want to get an external URL's class element, I have 2 usefull functions. In this example we get the '<h3 class="r">' elements back (search result headers) from google search:1. Check the URL (if it is reachable, existing)<?php# URL Checkfunction url_check($url) { $headers = @get_headers($url); return is_array($headers) ? preg_match('/^HTTP\\/\\d+\\.\\d+\\s+2\\d\\d\\s+.*$/',$headers[0]) : false; };?>2. Clean the element you want to get (remove all tags, tabs, new-lines etc.)<?php# Function to clean a stringfunction clean($text){ $clean = html_entity_decode(trim(str_replace(';','-',preg_replace('/\s+/S', " ", strip_tags($text)))));// remove everything return $clean; echo '\n';// throw a new line}?>After doing that, we can output the search result headers with following method:<?php$searchstring = 'djceejay';$url = 'http://www.google.de/webhp#q='.$searchstring;if(url_check($url)){ $doc = new DomDocument; $doc->validateOnParse = true; $doc->loadHtml(file_get_contents($url)); $output = clean($doc->getElementByClass('r')->textContent); echo $output . '<br>';}else{ echo 'URL not reachable!';// Throw message when URL not be called}?>Using loadHTML() automagically sets the doctype property of your DOMDocument instance(to the doctype in the html, or defaults to 4.0 Transitional). If you set the doctype with DOMImplementation it will be overridden.I assumed it was possible to set it and then load html with the doctype I defined(in order to decide the doctype at runtime), and ran into a huge headache trying to find out where my doctype was going. Hopefully this helps someone else.if you want to get rid of all the "DOMText elements containing ONLY whitespace", maybe try<?phpfunction loadHTML_noemptywhitespace(string $html, int $extra_flags = 0, int $exclude_flags = 0): DOMDocument{ $flags = LIBXML_HTML_NODEFDTD | LIBXML_NOBLANKS | LIBXML_NONET; $flags = ($flags | $extra_flags) & ~ $exclude_flags; $domd = new DOMDocument(); $domd->preserveWhiteSpace = false; @$domd->loadHTML('<?xml encoding="UTF-8">' . $html, $flags); $removeAnnoyingWhitespaceTextNodes = function (\DOMNode $node) use (&$removeAnnoyingWhitespaceTextNodes): void { if ($node->hasChildNodes()) { // Warning: it's important to do it backwards; if you do it forwards, the index for DOMNodeList might become invalidated; // that's why i don't use foreach() - don't change it (unless you know what you're doing, ofc) for ($i = $node->childNodes->length - 1; $i >= 0; --$i) { $removeAnnoyingWhitespaceTextNodes($node->childNodes->item($i)); } } if ($node->nodeType === XML_TEXT_NODE && !$node->hasChildNodes() && !$node->hasAttributes() && empty(trim($node->textContent))) { //echo "Removing annoying POS"; // var_dump($node); $node->parentNode->removeChild($node); } //elseif ($node instanceof DOMText) { echo "not removed"; var_dump($node, $node->hasChildNodes(), $node->hasAttributes(), trim($node->textContent)); } }; $removeAnnoyingWhitespaceTextNodes($domd); return $domd;}Beware of the "gotcha" (works as designed but not as expected): if you use loadHTML, you cannot validate the document. Validation is only for XML. Details here: http://bugs.php.net/bug.php?id=43771&edit=1If you use loadHTML() to process utf HTML string (eg in Vietnamese), you may experience result in garbage text, while some files were OK. Even your HTML already have meta charset like <meta http-equiv="content-type" content="text/html; charset=utf-8">I have discovered that, to help loadHTML() process utf file correctly, the meta tag should come first, before any utf string appear. For example, this HTML file<html> <head> <meta http-equiv="content-type" content="text/html; charset=utf-8"> <title> Vietnamese - Tiếng Việt</title> </head><body></body></html>will be OK with loadHTML() when <meta> tag appear <title> tag.But the file below will not regcornize by loadHTML() because <title> tag contains utf string appear before <meta> tag.<html> <head> <title> Vietnamese - Tiếng Việt</title> <meta http-equiv="content-type" content="text/html; charset=utf-8"> </head><body></body></html>Here is a function I wrote to capitalize the previous remarks about charset problems (UTF-8...) when using loadHTML and then DOM functions.It adds the charset meta tag just after <head> to improve automatic encoding detection, converts any specific character to an html entity, thus PHP DOM functions/attributes will return correct values.<?phpmb_detect_order("ASCII,UTF-8,ISO-8859-1,windows-1252,iso-8859-15");function loadNprepare($url,$encod='') { $content = file_get_contents($url); if (!empty($content)) { if (empty($encod)) $encod = mb_detect_encoding($content); $headpos = mb_strpos($content,'<head>'); if (FALSE=== $headpos) $headpos= mb_strpos($content,'<HEAD>'); if (FALSE!== $headpos) { $headpos+=6; $content = mb_substr($content,0,$headpos) . '<meta http-equiv="Content-Type" content="text/html; charset='.$encod.'">' .mb_substr($content,$headpos); } $content=mb_convert_encoding($content, 'HTML-ENTITIES', $encod); } $dom = new DomDocument; $res = $dom->loadHTML($content); if (!$res) return FALSE; return $dom;}?>NB: it uses mb_strpos/mb_substr instead of mb_ereg_replace because that seemed more efficient with huge html pages.