/**
* Converts a given string to our xml friendly text.
* This step involves purifying the string
*
* @param String $string Input string to reformat
* @return String Reformatted string (optional HTML -> Markdown, UTF-8)
*/
public function xml_ready($string, $convert_to_markdown = true)
{
static $purifier = null;
static $fixer = null;
static $markdown = null;
if ($purifier === null) {
$purifier_config = HTMLPurifier_Config::createDefault();
$purifier_config->set('Cache.SerializerPath', realpath($GLOBALS['TMP_PATH']));
$purifier = new HTMLPurifier($purifier_config);
$markdown = new HTML_To_Markdown();
$markdown->set_option('strip_tags', true);
}
$string = studip_utf8encode($string);
$string = $purifier->purify($string);
if ($convert_to_markdown) {
$string = $markdown->convert($string);
$string = preg_replace('/\\[\\]\\((\\w+:\\/\\/.*?)\\)/', '', $string);
$string = preg_replace('/\\[(\\w+:\\/\\/.*?)\\/?\\]\\(\\1\\/?\\s+"(.*?)"\\)/isxm', '$2: $1', $string);
$string = preg_replace('/\\[(\\w+:\\/\\/.*?)\\/?\\]\\(\\1\\/?\\)/isxm', '$1', $string);
$string = preg_replace('/\\[(.*?)\\]\\((\\w+:\\/\\/.*?)\\)/', '$1: $2', $string);
}
$string = preg_replace('/[\\x00-\\x08\\x0b\\x0c\\x0e-\\x1f]/', '', $string);
$string = trim($string);
$string = htmlspecialchars($string, ENT_QUOTES, 'UTF-8');
return $string;
}
/**
* In addition to markdown conversion, strips images and treats h3-6 as paragraphs.
* @param string $html
* @return string empty if conversion fails
*/
public static function convert($html)
{
$converter = new HTML_To_Markdown();
$html = preg_replace('/<img[^>]*>/', '', $html);
$html = preg_replace('/<(\\/)?h[3-6]>/', '<$1p>', $html);
$converter->set_option('header_style', 'postmatic');
$converter->set_option('strip_tags', true);
$markdown = $converter->convert($html);
return $markdown ? $markdown : '';
}
public static function fromHtml($html, $detail = false)
{
# Undo markdown
require_once dirname(__FILE__) . '/html-to-markdown/HTML_To_Markdown.php';
$markdown = new HTML_To_Markdown($html);
if (!$detail) {
return self::deparseBlock($markdown->output());
}
$detail = array('markdown' => $markdown->output(), 'human' => self::deparseBlock($markdown->output()));
return $detail;
}
/**
* Create the Calendar Event
*
* @param $title
* @param $startDate
* @param null $description
* @param int $duration
* @param null $location
*/
public function __construct($title, $startDate, $description = null, $duration = 60, $location = null)
{
if ($description) {
$markdown = new HTML_To_Markdown();
$markdown->set_option('strip_tags', true);
$noImageText = preg_replace(array("/<img[^>]+\\>/i", "/<hr[^>]+\\>/i"), "", $description);
$markdown->convert($noImageText);
$this->description = $markdown;
}
if (is_numeric($startDate)) {
$startTimestamp = new DateTime();
$startTimestamp->setTimestamp($startDate);
} else {
$startTimestamp = new DateTime($startDate);
}
$startTimestamp->setTimezone(new DateTimeZone('UTC'));
$this->startDate = $startTimestamp->format('Ymd\\THis\\Z');
$this->duration = $duration;
$this->endDate = $startTimestamp->modify('+' . $duration . ' minutes')->format('Ymd\\THis\\Z');
$this->location = $location;
$this->title = strip_tags($title);
}
public function test_set_option()
{
$markdown = new HTML_To_Markdown();
$markdown->set_option('strip_tags', true);
$markdown->convert('<span>Strip</span>');
$this->assertEquals('Strip', $markdown->__toString());
}
$sql = 'SELECT post_title, post_name, post_date_gmt, post_status, post_content FROM `' . $config['prefix'] . '_posts` WHERE post_status = "publish" AND post_content != ""';
$rs = $conn->query($sql);
if ($rs === false) {
trigger_error('Wrong SQL: ' . $sql . ' Error: ' . $conn->error, E_USER_ERROR);
} else {
$rows_returned = $rs->num_rows;
// dumper($rs->fetch_assoc());
while ($row = $rs->fetch_assoc()) {
// process the posts
$date = date('Y-m-d', strtotime($row['post_date_gmt']));
$filename = $row['post_name'] . '.md';
if ($row['post_status'] == 'publish') {
$filename = $date . '-' . $filename;
}
// convert content to md
$temp = new HTML_To_Markdown($row['post_content']);
$md = $temp->output();
// change url and assets directory
foreach ($config['old_wp_url'] as $arr) {
// change upload
$md = str_replace($arr . '/wp-content/uploads', 'images', $md);
$md = str_replace($arr, '', $md);
}
// insert title in content;
$md = "#" . $row['post_title'] . "\n\n" . $md;
// create post in posts directory
file_put_contents('posts/' . $filename, $md);
}
}
echo 'Done! Go <a href="gen.php">generate the index now »</a>';
//// =========================
protected function htmlToMarkdown($html)
{
$markdown = new \HTML_To_Markdown($html, $this->options);
return $markdown->output();
}
function sanitize($v)
{
if (is_object($v)) {
return $v;
}
$v = trim($v);
if (!$v) {
return '';
}
if (preg_match($this->tag('p'), $v)) {
$markdown = new HTML_To_Markdown($v, array('strip_tags' => true, 'remove_nodes' => 'img'));
$v = $markdown->output();
$v = preg_replace('/\\[\\s*\\]/', '[link]', $v);
} else {
$v = strip_tags($v);
$v = preg_replace("/\n/", " \n", $v);
// Convert line breaks to markdown style
}
$v = trim($v, " \t\n\r\v Â");
return $v;
}
public static function text($html = null, $overrides = null)
{
$md = new \HTML_To_Markdown($overrides);
return $md->convert($html);
}
/**
* @param string $html
* @return string
*/
protected function htmlToMarkdown($html)
{
$markdown = new \HTML_To_Markdown($html, array('header_style' => 'atx', 'bold_style' => '__', 'italic_style' => '_'));
return $markdown->output();
}
/**
* Convert HTML to markdown
*
* @since 1.0.0
*
* @param string $html
* @return string
*/
protected function convertToMarkdown($html)
{
$converter = new HTML_To_Markdown();
$converter->set_option('strip_tags', false);
$converter->set_option('header_style', $this->headingStyle);
return $converter->convert($html);
}
