php 提取html中的文字或图片

//html富文本提取text普通文本内容
    private function getPlainTextFromHtml($html, $num = 'ALL')
    {
        // Remove the HTML tags
        $html = strip_tags($html);
        // Convert HTML entities to single characters
        $html     = html_entity_decode($html, ENT_QUOTES, 'UTF-8');
        $html_len = mb_strlen($html, 'UTF-8');
        // Make the string the desired number of characters
        // Note that substr is not good as it counts by bytes and not characters
        if ($num == 'ALL') {
            $n = strlen($html);
        } else {
            $n = $num;
        }
        $html = mb_substr($html, 0, $n, 'UTF-8');
        return $html;
    }

html提取图片

/*
    * $content 文章内容
    * $order 要获取哪张图片，ALL所有图片，0第一张图片
    */
    function getImgs($content, $order = 'ALL')
    {
        $pattern = "/<img .*?src=[\'|\"](.*?(?:[\.gif|\.jpg]))[\'|\"].*?[\/]?>/";
        preg_match_all($pattern, $content, $match);
        if (isset($match[1]) && !empty($match[1])) {
            if ($order === 'ALL') {
                return $match[1];
            }
            if (is_numeric($order) && isset($match[1][$order])) {
                return $match[1][$order];
            }
        }
        return '';
    }