DOM parser that allows HTML5-style </ in <script> tag

49
votes

Update: html5lib (bottom of question) seems to get close, I just need to improve my understanding of how it's used.

I am attempting to find an HTML5-compatible DOM parser for PHP 5.3. In particular, I need to access the following HTML-like CDATA within a script tag:

<script type="text/x-jquery-tmpl" id="foo">
    <table><tr><td>${name}</td></tr></table>
</script>

Most parsers will end parsing prematurely because HTML 4.01 ends script tag parsing when it finds ETAGO (</) inside a <script> tag. However, HTML5 allows for </ before </script>. All of the parsers I have tried so far have either failed, or they are so poorly documented that I haven't figured out if they work or not.

My requirements:

  1. Real parser, not regex hacks.
  2. Ability to load full pages or HTML fragments.
  3. Ability to pull script contents back out, selecting by the tag's id attribute.

Input:

<script id="foo"><td>bar</td></script>

Example of failing output (no closing </td>):

<script id="foo"><td>bar</script>

Some parsers and their results:


DOMDocument (fails)

Source:

<?php

header('Content-type: text/plain');
$d = new DOMDocument;
$d->loadHTML('<script id="foo"><td>bar</td></script>');
echo $d->saveHTML();

Output:

Warning: DOMDocument::loadHTML(): Unexpected end tag : td in Entity, line: 1 in /home/adam/public_html/2010/10/26/dom.php on line 5
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40/loose.dtd">
<html><head><script id="foo"><td>bar</script></head></html>


FluentDOM (fails)

Source:

<?php

header('Content-type: text/plain');
require_once 'FluentDOM/src/FluentDOM.php';
$html = "<html><head></head><body><script id='foo'><td></td></script></body></html>";
echo FluentDOM($html, 'text/html');

Output:

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40/loose.dtd">
<html><head></head><body><script id="foo"><td></script></body></html>


phpQuery (fails)

Source:

<?php

header('Content-type: text/plain');

require_once 'phpQuery.php';

phpQuery::newDocumentHTML(<<<EOF
<script type="text/x-jquery-tmpl" id="foo">
<td>test</td>
</script>
EOF
);

echo (string)pq('#foo');

Output:

<script type="text/x-jquery-tmpl" id="foo">
<td>test
</script>


html5lib (passes)

Possibly promising. Can I get at the contents of the script#foo tag?

Source:

<?php

header('Content-type: text/plain');

include 'HTML5/Parser.php';

$html = "<!DOCTYPE html><html><head></head><body><script id='foo'><td></td></script></body></html>";
$d = HTML5_Parser::parse($html);

echo $d->saveHTML();

Output:

<html><head></head><body><script id="foo"><td></td></script></body></html>
6
phpdomhtml
Note: when you try to parse HTML via loadHTML, DOM based libraries will use libxml's HTML parser module. If you load your snippet above with loadXML instead, there will be no errors, but of course, the page is expected to be valid XHTML then. Also see Best Methods to parse HTML but basically all DOM based parsers will likely produce the same results here. – Gordon
+1 for a good question. I wonder if it would be possible to use HTML comments or a CDATA block to delimit the code in the script tag, as one would do for Javascript? Or would that also get included in the output from the template? – Spudley
If html5lib passes isn't that the answer to this question? – Kwebble

6 Answers

11
votes

I had the same problem and apparently you can hack your way trough this by loading the document as XML, and save it as HTML :)

$d = new DOMDocument;
$d->loadXML('<script id="foo"><td>bar</td></script>');
echo $d->saveHTML();

But of course the markup must be error-free for loadXML to work.

5
votes

Re: html5lib

You click on the download tab and download the PHP version of the parser.

You untar the archive in a local folder

 tar -zxvf html5lib-php-0.1.tar.gz
 x html5lib-php-0.1/
 x html5lib-php-0.1/VERSION
 x html5lib-php-0.1/docs/
 ... etc

You change directories and create a file named hello.php

cd html5lib-php-0.1
touch hello.php

You place the following PHP code in hello.php

$html = '<html><head></head><body>
<script type="text/x-jquery-tmpl" id="foo">
<table><tr><td>${name}</td></tr></table>
</script> 
</body></html>';
$dom = HTML5_Parser::parse($html); 
var_dump($dom->saveXml()); 
echo "\nDone\n";

You run hello.php from the command line

php hello.php

The parser will parse the document tree, and return a DOMDocument object, which can be manipulated as any other DOMDocument object.

5
votes

FluentDOM uses the DOMDocument but blocks loading notices and warnings. It does not have an own parser. You can add your own loaders (For example one that uses the html5lib).

4
votes

I added comment tags (<!-- ... -->) in my jQuery template blocks (CDATA blocks also failed) and DOMDocument did not touch the internal HTML.

Then, before I used the jQuery templates, I wrote a script to remove the comments.

$(function() {
    $('script[type="text/x-jquery-tmpl"]').text(function() {
        // The comment node in this context is actually a text node.
        return $.trim($(this).text()).replace(/^<!--([\s\S]*)-->$/, '$1');
    });
});

Not ideal, but I wasn't sure of a better workaround.

4
votes

I just find out (in my case).

try to change parameters option of loadHTML using LIBXML_SCHEMA_CREATE in DOMDocument

$dom = new DOMDocument;

libxml_use_internal_errors(true);
//$dom->loadHTML($buffer, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
$dom->loadHTML($buffer, LIBXML_SCHEMA_CREATE);
3
votes

I ran into this exact problem.

PHP Dom Document parses the html inside a script tag and that can actually lead to a completely different dom.

Since I didn't want to use another library than DomDocument. I wrote a few lines that strips any script content, then you do what ever you need to do with dom document and afterwards you put that script content back.

Obviously the script content isn't available to your dom object because it's empty.

With the following lines of php code you can 'fix' this problem. Be warned that script tags in scripts tags will cause bug.

$scripts = array();
// this will select all script tags non-greedy. If you have a script tag in your script tag, it will cause problems.
preg_match_all("/((<script.*>)(.*))\/script>/sU", $html, $scripts);
// Make content of scripts empty
$html = str_replace($scripts[3], '', $html);

// Do DOM Document stuff here

// Put script contents back
$html = str_replace($scripts[2], $scripts[1], $html);

I hope this will help some people :-).