searchanddisplace-ingest/app/Parser/DocxParser/Text.php

<?php

namespace App\Parser\DocxParser;

use App\Parser\DocxParser\Traits\Helper;

class Text
{

    use Helper;

    public function handle($textElement)
    {
        $data           = $this->getElementData($textElement);
        $data[ 'type' ] = 'text';

        return $data;
    }


    /**
     * @param $textElement
     *
     * @return array
     */
    private function getElementData($textElement)
    {
        $text = $textElement->getText();
        //if (strpos($text, 'PPOINTMENT AND GRANT OF LICENSE') !== false) {
        //    dd($textElement->getParent()->getDepth());
        //}
        $textData = $this->getNumberingFromText($text);

        if (strlen($textData[ 'content' ])) {
            $textData[ 'content' ] = $this->styleTheText($textData[ 'content' ], $textElement);
        }



        return $textData;

    }


    /**
     * @param $text
     *
     * @return array
     */
    private function getNumberingFromText($text)
    {

        $data = [];
        preg_match('/^([0-9.])([^(A-Z)(a-z) ]*)/', trim($text), $match);

        if ($match && isset($match[ 0 ]) && $match[ 0 ] !== '.') {
            $data[ 'content' ]   = trim(str_replace($match[ 0 ], '', $text));
            $data[ 'numbering' ] = $match[ 0 ];
        } else {
            $data[ 'content' ] = trim(preg_replace('/\t+/', '', $text));
        }

        return $data;
    }


    private function styleTheText($textString, $textObject)
    {

        $textStyle = [
            'font'      => $textObject->getFontStyle(),
            'paragraph' => $textObject->getParagraphStyle()
        ];

        $fontStyle   = $textStyle[ 'font' ]->getStyleValues();
        $inlineStyle = $this->getInlineStyles(array_merge($fontStyle[ 'style' ], $fontStyle[ 'basic' ]));

        return '<span'.(($inlineStyle) ? ' style="'.$inlineStyle.'"' : '').'>'.$this->getStyledText($textString,
                $fontStyle[ 'style' ]).'</span>';
    }


    /**
     * @param $styles
     *
     * @return string
     */
    private function getInlineStyles($styles)
    {
        $styleString    = '';
        $acceptedInline = [
            "dStrike"   => 'text-decoration: line-through;text-decoration-style: double;',
            "smallCaps" => 'text-transform: lowercase;',
            "allCaps"   => 'text-transform: capitalize;',
            "fgColor"   => 'background-color:'.$styles[ 'fgColor' ].';',
            "hidden"    => 'display:none;',
            "size"      => 'font-size:'.$styles[ 'size' ].'pt;',
            "color"     => 'color:#'.$styles[ 'color' ].';'
        ];

        foreach ($styles as $style => $value) {
            if (array_key_exists($style, $acceptedInline) && $value && ! in_array($value, ['none', 'auto'])) {
                $styleString .= $acceptedInline[ $style ];
            }
        }

        return $styleString;
    }


    /**
     * @param $text
     * @param $styles
     *
     * @return string
     */
    private function getStyledText($text, $styles)
    {
        $mappedStyle = [
            'bold'      => 'strong',
            'italic'    => 'i',
            'underline' => 'u',
            'strike'    => 'strike',
            "super"     => 'sup',
            "sub"       => 'sub',
        ];
        foreach ($styles as $style => $active) {
            if (array_key_exists($style, $mappedStyle) && $active && $active !== 'none') {
                $text = $this->appendHtmlStyle($text, $mappedStyle[ $style ]);
            }
        }

        return $text;
    }


    /**
     * @param $text
     * @param $styleType
     *
     * @return string
     */
    private function appendHtmlStyle($text, $styleType)
    {
        return "<$styleType>$text</$styleType>";
    }
}