从XML元素中删除开始和结束空格

Question

How can I remove all spacing characters before and after a XML field?

<data version="2.0">

  <field> 

     1 

  </field>        

  <field something=" some attribute here... "> 

     2  

  </field>

</data>

Notice that spacing before 1 and 2 and 'some attribute here...', I want to remove that with PHP.

if(($xml = simplexml_load_file($file)) === false) die();

print_r($xml);

Also the data doesn't appear to be string, I need to append (string) before each variable. Why?

Answer 1

You may want to use something like this:

$str = file_get_contents($file);
$str = preg_replace('~\s*(<([^>]*)>[^<]*</\2>|<[^>]*>)\s*~','$1',$str);
$xml = simplexml_load_string($xml,'SimpleXMLElement', LIBXML_NOCDATA);

I haven't tried this, but you can find more on this at http://www.lonhosford.com/lonblog/2011/01/07/php-simplexml-load-xml-file-preserve-cdata-remove-whitespace-between-nodes-and-return-json/ .

Note that the spaces between the opening and closing brackets ( <x> _space_ </x> ) and the attributes ( <x attr=" _space_ "> ) are actually part of the XML document's data (in contrast with the spaces between <x> _space_ <y> ), so I would suggest that the source you use should be a bit less messy with spaces.

Answer 2

To do that in PHP you first have to convert the document into a DOMDocument so that you can address the nodes you want to normalize the whitespace within properly via DOMXPath . The (xpath in) SimpleXMLElement is too limited to access text-nodes precisely enough as it would be needed for this operation.

An Xpath-query to access all text-nodes that are within leaf-elements and all attributes is:

//*[not(*)]/text() | //@*

Given that $xml is a SimpleXMLElement you could do white-space normalization like in the following example:

$doc   = dom_import_simplexml($xml)->ownerDocument;
$xpath = new DOMXPath($doc);
foreach ($xpath->query('//*[not(*)]/text()|//@*') as $node) {
    /** @var $node DOMText|DOMAttr */
    $node->nodeValue = trim(preg_replace('~\s+~u', ' ', $node->nodeValue), ' ');
}

You could perhaps stretch this to all text-nodes ( as suggested in related Q&A ), but this might require document normalization under circumstance. As text() in Xpath does not differ between text-nodes and Cdata-sections, you might want to skip on these type of nodes ( DOMCdataSection ) or expand them into text-nodes when loading the document (use the LIBXML_NOCDATA option for that) to achieve more useful results.

---

Also the data doesn't appear to be string, I need to append (string) before each variable. Why?

Because it's an object of type SimpleXMLElement , if you want the string value of such an object (element), you need to cast it to string. See as well the following reference question:

• Forcing a SimpleXML Object to a string, regardless of context

---

And last but not least: don't trust print_r or var_dump when you use it on a SimpleXMLElement : it's not showing the truth. Eg you could override __toString() which could also solve your issue:

class TrimXMLElement extends SimpleXMLElement
{
    public function __toString()
    {
        return trim(preg_replace('~\s+~u', ' ', parent::__toString()), ' ');
    }
}

$xml = simplexml_load_string($buffer, 'TrimXMLElement');

print_r($xml);

Even though casting to string would normally apply (eg with echo ), the output of print_r still would not reflect these changes. So better not rely on it, it can never show the whole picture.

---

Full example code to this answer ( Online Demo ):

<?php
/**
 * Remove starting and ending spaces from XML elements
 *
 * @link https://stackoverflow.com/a/31793566/367456
 */

$buffer = <<<XML
<data version="2.0">

  <field>

     1

  </field>

  <field something=" some attribute here... ">

     2 <![CDATA[ 34 ]]>

  </field>

</data>
XML;

class TrimXMLElement extends SimpleXMLElement implements JsonSerializable
{
    public function __toString()
    {
        return trim(preg_replace('~\s+~u', ' ', parent::__toString()), ' ');
    }

    function jsonSerialize()
    {
        $array = (array) $this;

        array_walk_recursive($array, function(&$value) {
            if (is_string($value)) {
                $value  = trim(preg_replace('~\s+~u', ' ', $value), ' ');
            }
        });

        return $array;
    }
}

$xml = simplexml_load_string($buffer, 'TrimXMLElement', LIBXML_NOCDATA);

print_r($xml);
echo json_encode($xml);

$xml = simplexml_load_string($buffer, null, LIBXML_NOCDATA);

$doc = dom_import_simplexml($xml)->ownerDocument;
$doc->normalizeDocument();
$doc->normalize();

$xpath = new DOMXPath($doc);
foreach ($xpath->query('//*[not(*)]/text()|//@*') as $node) {
    /** @var $node DOMText|DOMAttr|DOMCdataSection */
    if ($node instanceof DOMCdataSection) {
        continue;
    }
    $node->nodeValue = trim(preg_replace('~\s+~u', ' ', $node->nodeValue), ' ');
}

echo $xml->asXML();

Answer 3

Since simplexml_load_file() reads data into an array, you could do something like this:

function TrimArray($input){

    if (!is_array($input))
        return trim($input);

    return array_map('TrimArray', $input);
}