HTML5 Tools + Drupal 7 + HTML5 PHP breaks </address> tag #109

sylus · 2016-05-02T14:34:22Z

It seems when using HTML5 Tools + HTML5 PHP for Drupal 7 that address tags are broken.

We originally added the HTML5 PHP library to fix issues with video tags which worked. Except now it has been proven when saving a piece of content with the opening + closing address tags that the closing tags get removed on node render. CKEditor source view shows the close address tag as still present. On node view and inspect page source the closing address tag is removed.

Was wondering if had any ideas about this? ^_^

mattfarina · 2016-05-02T21:57:11Z

@sylus Can you point me to the code using Masterminds\HTML5 that's causing this problem? Is it in a module?

If you look in Drupal\Component\Utility\Html you'll see it uses \DOMDocument instead of the html5 parser here. For example, see the load function.

I'm curious where this is happening.

sylus · 2016-05-05T18:45:37Z

Hey thanks for the response! :)

We are using a patch to get html5_tools to work with html5-php from this issue:

http://drupal.org/node/2415305
https://www.drupal.org/files/issues/replace_core_text-2415305-7.patch

When looking at the markup for the body field in devel I see the following:

value (String, 40 characters ) <address>Testing address tag</address>
...
safe_value (String, 39 characters ) <address>Testing address tag

So it is only the safe_value which has the issue and gets called by the following:

function html5_tools_filter_info() {
...
  $filters['filter_html5corrector'] = array(
    'title' =>  t('Correct faulty and chopped off HTML5'),
    'process callback' => '_html5_tools_filter_htmlcorrector',
    'weight' => 10,
  );
...
}

/**
 * Implements callback_filter_process().
 *
 * Scans the input and makes sure that HTML tags are properly closed.
 */
function _html5_tools_filter_htmlcorrector($text) {
  return html5_tools_filter_dom_serialize(html5_tools_filter_dom_load($text));
}

function html5_tools_filter_dom_serialize($dom_document) {
  $body_node = $dom_document->getElementsByTagName('body')->item(0);
  $body_content = '';
  foreach ($body_node->getElementsByTagName('script') as $node) {
    filter_dom_serialize_escape_cdata_element($dom_document, $node);
  }
  foreach ($body_node->getElementsByTagName('style') as $node) {
    filter_dom_serialize_escape_cdata_element($dom_document, $node, '/*', '*/');
  }
  $html5 = html5_tools_get_html5_instance();
  foreach ($body_node->childNodes as $child_node) {
    $body_content .= $html5->saveHTML($child_node);
  }
  return preg_replace('|<([^> ]*)/>|i', '<$1 />', $body_content);
}

function html5_tools_filter_dom_load($text) {
  $html5 = html5_tools_get_html5_instance();

  $text = <<<EOL
<!DOCTYPE html>
<html>
<head><meta http-equiv="Content-Type" content="text/html; charset=utf-8"></head>
<body>$text</body>
</html>
EOL;

  $dom_document = $html5->loadHTML($text);
  return $dom_document;
}

If i don't use that filter then the issues goes away. Hope this helps!

sylus · 2016-05-09T17:29:15Z

I believe I have reduced it down to this:

  $html5 = html5_tools_get_html5_instance();
  $text = <<<EOL
<!DOCTYPE html>
<html>
<head><meta http-equiv="Content-Type" content="text/html; charset=utf-8"></head>
<body>
<address>Example</address>
<video src="http://www.google.com"/>
</body>
</html>
EOL;

  $dom_document = $html5->loadHTML($text);

  $body_node = $dom_document->getElementsByTagName('body')->item(0);
  $html5 = html5_tools_get_html5_instance();
  foreach ($body_node->childNodes as $child_node) {
    $body_content .= $html5->saveHTML($child_node);
  }

Where $body_content is returned as:

<address>Example
<video src="http://www.google.com">&nbsp;</video>

Noticably the video gets fixed but address isn't.

sylus · 2016-05-09T17:44:55Z

When I changed the following:

        "address" => 89, // NORMAL | VOID_TAG | AUTOCLOSE_P | BLOCK_TAG

to

        "address" => 65, // NORMAL | BLOCK_TAG

I then got correct output.

<address>Example</address>
<video src="http://www.google.com">&nbsp;</video>

mattfarina · 2016-05-09T17:57:11Z

Looking at the spec for address I see that neither the start or end tag can be omitted. I'm not sure why AUTOCLOSE_P is on this. I think that's what's causing this issue.

@sylus Do you want to update this on master and the 2.x branch or shall I? I'm always open to PRs.

    "address" => 65, // NORMAL | BLOCK_TAG

sylus · 2016-05-09T18:07:30Z

Thanks a bunch submitted a P.R. ^_^

#109

mattfarina · 2016-05-10T14:14:42Z

Version 2.2.1 has been released with this fix included. Thanks for the contribution.

sylus · 2016-05-10T14:36:28Z

Thank you for the awesomeness that is html5-php :)

mattfarina added the bug label May 9, 2016

sylus mentioned this issue May 9, 2016

Correct address tag to reflect proper specification. #110

Merged

mattfarina pushed a commit that referenced this issue May 10, 2016

Correct address to reflect proper specfiication.

ec39dc6

#109

mattfarina closed this as completed May 10, 2016

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

HTML5 Tools + Drupal 7 + HTML5 PHP breaks </address> tag #109

HTML5 Tools + Drupal 7 + HTML5 PHP breaks </address> tag #109

sylus commented May 2, 2016

mattfarina commented May 2, 2016

sylus commented May 5, 2016 •

edited

Loading

sylus commented May 9, 2016 •

edited

Loading

sylus commented May 9, 2016 •

edited

Loading

mattfarina commented May 9, 2016

sylus commented May 9, 2016

mattfarina commented May 10, 2016

sylus commented May 10, 2016 •

edited

Loading

HTML5 Tools + Drupal 7 + HTML5 PHP breaks </address> tag #109

HTML5 Tools + Drupal 7 + HTML5 PHP breaks </address> tag #109

Comments

sylus commented May 2, 2016

mattfarina commented May 2, 2016

sylus commented May 5, 2016 • edited Loading

sylus commented May 9, 2016 • edited Loading

sylus commented May 9, 2016 • edited Loading

mattfarina commented May 9, 2016

sylus commented May 9, 2016

mattfarina commented May 10, 2016

sylus commented May 10, 2016 • edited Loading

sylus commented May 5, 2016 •

edited

Loading

sylus commented May 9, 2016 •

edited

Loading

sylus commented May 9, 2016 •

edited

Loading

sylus commented May 10, 2016 •

edited

Loading