Re: [PHP] importNode issue

Michael A. Peters wrote:
> I'm experiencing a slight problem with importNODE putting unwanted 
> carriage returns in the the output.
> 
> Here's my function:
> 
> // syntax highlighting
> include_once('Text/Highlighter.php');
> 
> function syntaxHighlight($dom,$lang,$code) {
>    $hl =& Text_Highlighter::factory($lang);
>    $out = $hl->highlight($code);
>    //die($out);
>    $tmpDOM = new DOMDocument('1.0','UTF-8');
>    $tmpDOM->loadXML($out);
>    $foo = $tmpDOM->saveXML();
>    //die($foo);
> 
>    $nodeList = $tmpDOM->getElementsByTagName('div');
>    $impDIV = $nodeList->item(0);
>    $returnDIV = $dom->importNode($impDIV,true);
> 
>    return $returnDIV;
>    }
> 
> -=-
> 
> Here's my test:
> 
> $code  ="<?php" . "\n\n";
> $code .="require_once('/path/to/something');" . "\n";
> $code .="function somefunc(\$myfoo,\$mybar) {" . "\n";
> $code .="   \$myfoobar = \$myfoo . \$mybar;" . "\n";
> $code .="   return \$myfoobar;" . "\n";
> $code .="   }" . "\n";
> $code .="?>" . "\n";
> 
> $fooTest = syntaxHighlight($dom,'PHP',$code);
> 
> -=-
> 
> If I uncomment the die($out) - I get what I expect spit to the screen, 
> view source shows code that will do what I want.
> 
> If instead I uncomment die($foo) - I also get what I expect spit to 
> screen. view source shows code that will do what I want.
> 
> However, if the function is allowed to continue, the imported div has 
> carriage returns between each and every </span><span> which of course 
> completely breaks the browser display because they are inside a 
> <pre></pre> node.
> 
> Anyone know why importNode does this and how to fix it?
> 
> The only (untried) solution I can think of is to replace each carriage 
> return with a <br /> and every space with &#160; and then replace the 
> <pre> with a <div class='monospace'> or some such hackery before running 
> loadXML() on it. But I would rather not do that.
> 
> php 5.2.12 built against libxml 2.6.26
> 

Found the solution - the problem was where Text/Highlighter.php was 
putting the newline in the code it generates.

$out = preg_replace("/<span class=\"hl-code\">\n/","\n<span 
class=\"hl-code\">",$out);

fixes the issue, I don't have to set

$dom->formatOutput = false;

to avoid broken display now.
I think it's a DOMDocument bug, well, maybe, it shouldn't do any 
modifications with newlines inside a pre node with formatOutput - but I 
suppose it has no way of knowing what the pre node is use for since html 
5 doctype doesn't identify itself as html.

But anyway, that preg_replace fixes it.
I may send a demo of problem and patch to the pear Text/Highlighter.php 
maintainer so that the preg_replace isn't needed.