UNIX for Dummies Questions & Answers

<?php
/** \action-csv.php
 * \copyright (c) 2012, Christopher Jay Cox, Licensed under GPLv2
 * \author Christopher Jay Cox
 * \version 20120318a
*/

function Html2CSV($page_html) {
        # This also preserves hyperlinks, we'll parse the targets and names later.
        $page_html = strip_tags($page_html, '<a><table><tr><th><td>');

        preg_match_all('/<tr[^>]*>(.*)<\/tr>/isU', $page_html, $trs);

        $ahrefexp = '/<a \s*[^>]*href=["\'](?P<href>[^"\']*)["\']\s*[^>]*>(?P<name>.*)<\s*\/a>/si';
        $csvout='';
        foreach ($trs[1] as $tr) {
                preg_match_all('/<t[hd][^>]*>(.*)<\/t[hd]>/isU', $tr, $tds);
                $first = true;
                foreach ($tds[1] as $td) {
                        # For CSV output, prefer blank to regular decode for nbsp
                        $td = preg_replace('/&nbsp;/', ' ', $td);
                        $td = preg_replace('/&quot;/', '"', $td);
                        # Double quotes must be escaped by another double quote
                        $td = preg_replace('/"/', '""', $td);
                        if (!$first) $csvout .= ',';
                                if (preg_match_all($ahrefexp, $td, $matches))
                                        $td = $matches['name'][0];
                                $csvout .= '"' . html_entity_decode($td, ENT_COMPAT, 'UTF-8') . '"';
                                $first = false;
                }
                $csvout .= "\n";
        }
        return $csvout;
}

$page_html = file_get_contents('/tmp/webpage.html');
$csvout=Html2CSV($page_html);
echo "$csvout"
?>