Topic: Issue with calculating _wordmlunits: a non-numeric value encountered

Posted by j.ypma · 27-06-2017 - 12:21

Hi there,

I ran into an issue with phpdocx 7 when using: $docx->embedHTML($html);
This html is created by users using an editor and stored into the database.
When merging I sometimes get the error-message below:

inÃ‚ HTML2WordML.incÃ‚ (line 2563)

atÃ‚ HandleExceptions->handleError(2, 'A non-numeric value encountered', '/Users/jarichypma/projecten/inkoopwegwijzer5/phpdocx/Classes/Phpdocx/Transform/HTML2WordML.inc', 2563,Ã‚ array('sHtmlUnit' =>Ã‚ null, 'fontSize' =>Ã‚ false, 'match' =>Ã‚ array('', '', ''), 'widthType' => 'dxa'))inÃ‚ HTML2WordML.incÃ‚ (line 2563)

After some digging around I found out that the _wordMLUnits-method in HTML2WordML.inc raises this error. The problem occurs when the parameter $sHtmlUnit has a null-value.

I first tried to fix/sanitize my html, but that leads to nothing. So I wrote a small fix to check for a null-value. If so, return false right away.

    /**
     * Translates HTML units to Word ML units
     *
     * @access private
     * @param string $sHtmlUnit Units in HTML format
     * @param string $fontSize Font size, if applicable
     * @return array
     */
    private function _wordMLUnits($sHtmlUnit, $fontSize = false)
    {
        // Jarich.eu 2017-06-27
        // Check if $sHtmlUnit is not null, if it is null return right away
        if( is_null($sHtmlUnit) ){
            return false;
        }

        if (!preg_match('/^(-?\d*\.?\d*)(%|em|pt|px)?$/i', trim($sHtmlUnit), $match))
            return(array(0, 'dxa'));
        
        $match[1] = (strpos($match[1], '.') === 0 ? '0' : '') . $match[1];
        $match[2] = empty($match[2]) ? '' : $match[2];

        //if($match[2] != 'em' && $match[2] != 'px' && !empty($fontSize)) $match[2] = 'pt';

        switch ($match[2]) {
            case '%': //in WordML the precentage is given in fiftieths of a percent
                $widthType = 'pct';
                $width = 50 * $match[1];
                break;
            case 'em':
                $widthType = 'dxa';
                $width = 20 * $match[1] * $fontSize;
                break;
            case 'pt': //in WordML the width is given in twentieths of a point
                $widthType = 'dxa';
                $width = 20 * $match[1];
                break;
            case 'px': //a pixel is around 3/4 of a point
            default: //if no unit we asume is given in pixels
                $widthType = 'dxa';
                $width = 15 * $match[1];
        }

        return(array($width, $widthType));
    }

Is my approach the right one? And if so, can this fix be added to the codebase?

Regards,

Jarich

Posted by admin · 27-06-2017 - 13:34

Hello,

We move your patch to the dev team to be checked. We'll update this same topic.

Regards.

Posted by admin · 06-07-2017 - 11:39

Hello,

The dev team has checked your change. Although a null unit should happen, the best solution is returning a valid width and type if null:

// check if the unit is not set
if ($sHtmlUnit == null) {
    return(array(0, 'dxa'));
}

This is the whole method with the change applied:

    private function _wordMLUnits($sHtmlUnit, $fontSize = false)
    {
        // check if the unit is not set
        if ($sHtmlUnit == null) {
            return(array(0, 'dxa'));
        }

        if (!preg_match('/^(-?\d*\.?\d*)(%|em|pt|px)?$/i', trim($sHtmlUnit), $match)) {
            return(array(0, 'dxa'));
        }

        $match[1] = (strpos($match[1], '.') === 0 ? '0' : '') . $match[1];
        $match[2] = empty($match[2]) ? '' : $match[2];

        //if($match[2] != 'em' && $match[2] != 'px' && !empty($fontSize)) $match[2] = 'pt';

        switch ($match[2]) {
            case '%': //in WordML the precentage is given in fiftieths of a percent
                $widthType = 'pct';
                $width = 50 * $match[1];
                break;
            case 'em':
                $widthType = 'dxa';
                $width = 20 * $match[1] * $fontSize;
                break;
            case 'pt': //in WordML the width is given in twentieths of a point
                $widthType = 'dxa';
                $width = 20 * $match[1];
                break;
            case 'px': //a pixel is around 3/4 of a point
            default: //if no unit we asume is given in pixels
                $widthType = 'dxa';
                $width = 15 * $match[1];
        }

        return(array($width, $widthType));
    }

This change will be added in the next version of phpdocx.

Regards.

Forum