/**
* Create a static version of the site
*
* @return void
*/
public function create_archive()
{
global $blog_id;
// TODO: Do ajax calls instead of just running forever and ever
set_time_limit(0);
// Create archive directory
$current_user = wp_get_current_user();
$archive_name = join('-', array($this->slug, $blog_id, time(), $current_user->user_login));
$this->archive_dir = trailingslashit($this->temp_files_dir . $archive_name);
if (!file_exists($this->archive_dir)) {
wp_mkdir_p($this->archive_dir);
}
// Add URLs to queue
$origin_url = sist_origin_url();
$destination_url = $this->destination_scheme . '://' . $this->destination_host;
$origin_path_length = strlen(parse_url($origin_url, PHP_URL_PATH));
$urls_queue = array_unique(array_merge(array(trailingslashit($origin_url)), preg_split("/\r\n|\n|\r/", $this->additional_urls)));
while (count($urls_queue)) {
$current_url = array_shift($urls_queue);
$response = Simply_Static_Url_Fetcher::fetch($current_url);
// If we get a WP_Error then somehow our request failed (e.g. space in URL)
// TODO: Keep a queue of failed urls too
if (is_wp_error($response)) {
continue;
}
$url_parts = parse_url($response->url);
// TODO: This could throw an `Undefined index` error on URLs without
// a path, e.g. http://www.example.com (no trailing slash)
$path = $url_parts['path'];
if ($origin_path_length > 1) {
// prevents removal of '/'
$path = substr($path, $origin_path_length);
}
$is_html = $response->is_html();
// If we get a 30x redirect...
if (in_array($response->code, array(301, 302, 303, 307))) {
$redirect_url = $response->get_redirect_url();
// WP likes to 301 redirect `/path` to `/path/` -- we want to
// check for this and just add the trailing slashed version
if ($redirect_url === trailingslashit($current_url)) {
$urls_queue = $this->add_url_to_queue($urls_queue, $redirect_url);
} else {
/// convert our potentially relative URL to an absolute URL
$redirect_url = sist_relative_to_absolute_url($redirect_url, $current_url);
if ($redirect_url) {
// check if this is a local URL
if (sist_is_local_url($redirect_url)) {
// add the redirected page to the queue
$urls_queue = $this->add_url_to_queue($urls_queue, $redirect_url);
// and update the URL
$redirect_url = str_replace($origin_url, $destination_url, $redirect_url);
}
$view = new Simply_Static_View();
$content = $view->set_template('redirect')->assign('redirect_url', $redirect_url)->render_to_string();
$this->save_url_to_file($path, $content, $is_html);
$this->export_log[] = $current_url;
}
}
continue;
}
// Not a 200 for the response code? Move on.
// TODO: Keep a queue of failed urls too
if ($response->code != 200) {
continue;
}
$this->export_log[] = $current_url;
// Fetch all URLs from the page and add them to the queue...
$urls = $response->extract_urls();
foreach ($urls as $url) {
$urls_queue = $this->add_url_to_queue($urls_queue, $url);
}
// Replace the origin URL with the destination URL
$response->replace_urls($destination_url);
// Save the page to our archive
$content = $response->body;
$this->save_url_to_file($path, $content, $is_html);
}
}
/**
* Check if URL starts with same URL as WordPress installation
*
* @param string $url URL to check
* @return boolean true if URL is local, false otherwise
*/
function sist_is_local_url($url)
{
return stripos($url, sist_origin_url()) === 0;
}
/**
* Replaces base URL
*
* @param string $origin_url
* @param string $destination_url
* @return void
*/
public function replace_urls($destination_url)
{
/* TODO: Might want to eventually rope this into extract_urls_from_html/
extract_urls_from_css so that we're only doing preg_replace/
str_replace once. Only reason I'm not doing that now is because of
the fix for wp_json_encode.
*/
if ($this->is_html() || $this->is_css()) {
// replace any instance of the origin url, whether it starts with https://, http://, or //
$response_body = preg_replace('/(https?:)?\\/\\/' . addcslashes(sist_origin_host(), '/') . '/i', $destination_url, $this->body);
// also replace wp_json_encode'd urls, as used by WP's `concatemoji`
$response_body = str_replace(addcslashes(sist_origin_url(), '/'), addcslashes($destination_url, '/'), $response_body);
$this->body = $response_body;
}
}