Code:

$code = new Decoda\Decoda('[quote=milesj]Hello, my name is [b]Miles Johnson[/b] :)[/quote] [b]Hello[/b] ;)');
$code->defaults();
$code->addHook(new \Decoda\Hook\EmoticonHook());
echo $code->parse();

Output:

<blockquote class="decoda-quote">
    <div class="decoda-quote-body">
        Hello, my name is <b>Miles Johnson</b> :)
    </div>
</blockquote>
<b>Hello</b> <img src="/images/wink.png" alt="">

As you can see the :) inside the quote is rendered as text instead of an image, which seems to have to do with persistContent being false in QuoteFilter.

| Q | A | | --- | --- | | Bug fix? | no | | New feature? | no | | BC breaks? | yes | | Deprecations? | no | | Tests pass? | yes | | Related tickets | | | License | MIT |

With PCRE assertions does not need a second pass anymore.

The callback can be also cleanup.

BC Breaks

• The first parameter of Decoda\Hook\EmoticonHook::_emoticonCallback() method has changed:

  Before:

  class MyEmoticonHook extends \Decoda\Hook\EmoticonHook
  {
      // ...
      protected function _emoticonCallback($matches) {
           $matches[0];       // contain the text that matches the full pattern
           $matches['left'];  // contain the text at the left of the smiley
           $matches[1];       // contain the text at the left of the smiley
           $matches['right']; // contain the text at the right of the smiley
           $matches[2];       // contain the text at the right of the smiley
      }
      // ...
  }
  

  Atfer:

  class MyEmoticonHook extends \Decoda\Hook\EmoticonHook
  {
      // ...
      protected function _emoticonCallback($matches) {
           $matches[0];        // contain the text that matches the full pattern only the smiley
           $matches[1];        // contain the text of the smiley
           $matches['smiley']; // contain the text of the smiley
      }
      // ...
  }
  

Can we get a way to strip all the tags so that:

test [b]some [i]text[/i] here[/b] with a [url]http://google.com[/url] tag

could become the following if we include content:

test some text here with a http://google.com tag

or if we exclude the content:

test tag

I am not sure how you call it whether you would do something like $code->parse(array('strip_tags'=>true, 'exclude_content'=>true)) or $code->stripTags(array('exclude_content'=>true))

I was also wondering instead of that last options to add another param in the FilterAbstract for 'stripContent' => false and then when running $code->stripTags() it would default to including the text within the tag but removing the tag itself. Then by setting 'stripContent' => true it would also strip out everything within the tag which would be useful for [code]blah blah[/code].

This functionality would allow us to get back to the original content for further parsing. For example, building search functionality against our content would not index the tags or the content within the tags that are not useful for our parsing.

Added the possibilty to switch to other templates instead of php. You could implement a template engine to use twig or smarty files for example. Removed the constant for the templates an changed it to a method of the template engine. This allows a change of the template path if necessary.

Hey,

For instance, if you center this text "Hello :)" in a wysiwyg editor, the text will eventually look like this when passed to Decoda: "[center]Hello :)[/center]"

The smiley will not get passed because it is right next to the /center tag..

I just updated code to be compatible with PHP 7.3.

Seems to be working correctly, however the PHP 7.3 test seems to hang on Travis while it's working without issue on our machines.

Please comment if you see changes to be made.

Hello,

I have problem with links and ClickableHook.

Prev: http://google.com/example/links => <a href="http://google.com/example/links">http://google.com/example/links</a>

Now: http://google.com/example/links => <a href="http://google.com">http://google.com</a>/example/links

We have a few threads with quite big posts, problem is that the parsing of these texts exceeds the maximum PHP runtime. I striped down Decoda to the bare minimum and then tested it on my local machine with one of the problematic Threads.

public function bbDecodeForum($text)
{
  $textlen = strlen($text);
  if ($this->stopwatch) {
    $this->stopwatch
      ->start('bbDecodeForum: decode Post ' . $esPostData['id'] . ' with ' . $textlen . 'chars.');
  }
 $config = array(
   'xhtmlOutput' => false,
   'strictMode' => false,
   'escapeHtml' => true
 );
 $codedText = new Decoda($text, $config);
 $engine = new \Decoda\Engine\PhpEngine();
 $codedText
   ->setEngine($engine)
   ->setLocale('de-de')
   ->addFilter(new BBCode\BasicFilter())
   ->setMaxNewlines(3)
   ->setLineBreaks(true);
 $return = $codedText->parse();
 if ($this->stopwatch) {
   $this->stopwatch
     ->stop('bbDecodeForum: decode Post ' . $esPostData['id'] . ' with ' . $textlen . 'chars.');
 }
 return $return;
}

But still it need a big chunk of CPU time:

Is there anything I can do to increase performance? Thanks in advance :)

-Florian

Following BBCode

[image]/some/url.png[/image]

should be parsed as

<img src="/some/url.png alt="">

but it parsed to instead

<img src="/some/url.png" alt="">/some/url.png</img>

At the same time [img] tag works as supposed.

TODO:

• [x] Add some tests
• [x] Fix bugs founds

This PR try to improved emoticons detection.

When I use emoticon hook and I want to convert this string ':) :) :)' Parser convert only the first ":)"

But if I add spaces between smileys, it works

[url=https://example.net/]Dies soll ein Link sein[/url]

[url=https://example.net/]Dies soll ein Link[/url]

[url=https://example.net/]Dies soll ein[/url]

[url=https://example.net/]Dies soll[/url]

[url=https://example.net/]Dies[/url]

https://example.net/ 

becomes

<div class="body bbcode">
Dies soll ein Link sein<br><br>
Dies soll ein Link<br><br>
Dies soll ein<br><br>
Dies soll<br><br>
<a href="http://Dies">Dies</a><br><br>
<a href="https://example.net">https://example.net</a>/
</div>

So only the last two (with no space in the title) works.

The trailing slash at the end getting removed from the last is only a minor issue. The main one for sure is the linking not working for the first 4

Not sure if thats a problem for normal use cases, but using HTML escaping or h() in CakePHP you get

Some "<b>demo</b>" string

transformed into

Some "<b>demo</b>" string

with >/</" escaped

But with this BBCode parser the " remain it seems:

Some "<b>demo</b>" string

Depending on where those chars appear it could break some layouts maybe. But not sure.

I tried to first escape h($text) before adding it into the converter, but then the tags needing " characters to work wont parse anymore.

The Video standalone filter is nice for starters. But the issue is that you need to also maintain those video links/URLs/code all the time. It could be beneficial to let a dedicated library take care of that E.g. https://github.com/dereuromark/media-embed

To have BC and the choice we could add an optional MediaEmbedVideoFilter which only adds the dependency as require-dev for testing If you want to use you add the real dependency, otherwise no harm.

The working code can be seen here: https://github.com/dereuromark/cakephp-markup/blob/master/src/Bbcode/Decoda/VideoFilter.php

What do you think? Worth porting into, or should this be kept outside as own package shipping?

Is there a reason the string has to be put into the constructor, and the actual config happens at runtime? It makes it difficult for DI and libraries/config to adjust the object ones in constructing and then using it.

If you look into e.g. https://commonmark.thephpleague.com/1.5/extensions/github-flavored-markdown/ or https://github.com/kzykhys/Ciconia/blob/master/src/Ciconia/Ciconia.php#L42

Then you usually make the builder itself stateless and instead wrap statefulness internally.

$env = .. // Can contain all the filters and stuff
$options =  [
    'xhtmlOutput' => true,
    'strictMode' => false,
    'escapeHtml' => true
];

// We can pass defaults here if we want
$instance = new Decoda($env, $options);

// We can also set options here per convertion
$html = $instance->convert($bbcodeText, $options);
$html2 = $instance->convert($bbcodeText2, $otherOptions);

What do you think?

There will also be no need for reset() then as it keeps the wrapper stateless and reuses the objects here.

Is there anyway to force the img tag to use https when accessing a site over https? A few of my users are using non-secure links and I'd rather the images just not work on https then transfer insecure images.