mirror of
https://github.com/WenPai-org/wp-archiver.git
synced 2025-08-18 03:41:12 +08:00
680 lines
No EOL
27 KiB
PHP
680 lines
No EOL
27 KiB
PHP
<?php
|
|
if (!defined('ABSPATH')) {
|
|
exit;
|
|
}
|
|
|
|
class Archiver {
|
|
protected $slug = 'archiver';
|
|
protected $name;
|
|
protected $min_suffix = '';
|
|
protected $snapshot_max_count;
|
|
protected $wayback_machine_url_save = 'https://web.wenpai.net/save/';
|
|
protected $wayback_machine_url_fetch_archives = 'https://web.wenpai.net/cdx/';
|
|
protected $wayback_machine_url_view = 'https://web.archive.org/web/';
|
|
protected $current_permalink = '';
|
|
protected static $instance = null;
|
|
|
|
public static function get_instance($args = array()) {
|
|
if (null == self::$instance) {
|
|
self::$instance = new self($args);
|
|
}
|
|
return self::$instance;
|
|
}
|
|
|
|
public function __construct($args) {
|
|
$this->name = __('Archiver', 'archiver');
|
|
$this->snapshot_max_count = apply_filters('archiver_snapshot_max_count', 10);
|
|
$this->min_suffix = (defined('SCRIPT_DEBUG') && SCRIPT_DEBUG) ? '' : '.min';
|
|
add_action('init', array($this, 'setup_cron'));
|
|
add_action('rest_api_init', array($this, 'register_rest_routes'));
|
|
}
|
|
|
|
public function get_slug() {
|
|
return $this->slug;
|
|
}
|
|
|
|
public function run() {
|
|
add_action('wp_loaded', array($this, 'init'));
|
|
add_action('admin_init', array($this, 'admin_init'));
|
|
}
|
|
|
|
public function init() {
|
|
$this->set_locale();
|
|
add_action('wp_enqueue_scripts', array($this, 'register_scripts_and_styles'), 5);
|
|
add_action('admin_enqueue_scripts', array($this, 'register_scripts_and_styles'), 5);
|
|
add_action('wp_enqueue_scripts', array($this, 'enqueue_scripts'));
|
|
add_action('admin_enqueue_scripts', array($this, 'admin_enqueue_scripts'));
|
|
add_action('wp_ajax_archiver_immediate_snapshot', array($this, 'ajax_immediate_snapshot'));
|
|
if ($this->can_run()) {
|
|
add_action('save_post', array($this, 'trigger_post_snapshot'));
|
|
add_action('created_term', array($this, 'trigger_term_snapshot'), 10, 3);
|
|
add_action('edited_term', array($this, 'trigger_term_snapshot'), 10, 3);
|
|
add_action('profile_update', array($this, 'trigger_user_snapshot'), 10, 3);
|
|
add_action('admin_bar_menu', array($this, 'add_admin_bar_links'), 999);
|
|
} else {
|
|
add_action('admin_notices', array($this, 'do_admin_notice_disabled'));
|
|
}
|
|
}
|
|
|
|
public function admin_init() {
|
|
$this->add_post_meta_box();
|
|
$this->add_term_meta_box();
|
|
$this->add_user_meta_box();
|
|
}
|
|
|
|
public function setup_cron() {
|
|
$frequency = get_option('archiver_update_frequency', 'daily');
|
|
if (!wp_next_scheduled('archiver_process_urls')) {
|
|
wp_schedule_event(time(), $frequency, 'archiver_process_urls');
|
|
}
|
|
add_action('archiver_process_urls', array($this, 'process_urls_for_update'));
|
|
}
|
|
|
|
public function can_run() {
|
|
return apply_filters('archiver_can_run', __return_true());
|
|
}
|
|
|
|
protected function set_locale() {
|
|
load_plugin_textdomain(
|
|
$this->slug,
|
|
false,
|
|
dirname(dirname(plugin_basename(__FILE__))) . '/languages/'
|
|
);
|
|
}
|
|
|
|
public function add_post_meta_box() {
|
|
$post_types = apply_filters('archiver_post_types', get_post_types());
|
|
add_meta_box(
|
|
'archiver_post',
|
|
__('Archives', 'archiver'),
|
|
array($this, 'output_archiver_metabox'),
|
|
$post_types,
|
|
'side',
|
|
'default'
|
|
);
|
|
}
|
|
|
|
public function add_term_meta_box() {
|
|
$taxonomies = apply_filters('archiver_taxonomies', get_taxonomies());
|
|
$archiver_taxonomy_slugs = array_map(
|
|
function($taxonomy) { return "archiver-" . $taxonomy; },
|
|
$taxonomies
|
|
);
|
|
add_meta_box(
|
|
'archiver_terms',
|
|
__('Archives', 'archiver'),
|
|
array($this, 'output_archiver_metabox'),
|
|
$archiver_taxonomy_slugs,
|
|
'side',
|
|
'default'
|
|
);
|
|
foreach ($taxonomies as $taxonomy) {
|
|
add_action("{$taxonomy}_edit_form", array($this, 'output_term_meta_box'));
|
|
}
|
|
}
|
|
|
|
public function output_term_meta_box() {
|
|
$object_type = get_current_screen()->taxonomy;
|
|
$this->output_manual_meta_box($object_type);
|
|
}
|
|
|
|
public function add_user_meta_box() {
|
|
add_meta_box(
|
|
'archiver_terms',
|
|
__('Archives', 'archiver'),
|
|
array($this, 'output_archiver_metabox'),
|
|
array('archiver-user'),
|
|
'side',
|
|
'default'
|
|
);
|
|
}
|
|
|
|
public function output_archiver_metabox() {
|
|
$url = $this->get_current_permalink();
|
|
$snapshots = $this->get_post_snapshots($url);
|
|
wp_nonce_field('archiver_immediate_snapshot', '_ajax_nonce');
|
|
echo '<input type="hidden" id="archiver-url" value="' . esc_attr($url) . '">';
|
|
echo '<input type="hidden" id="archiver_nonce" value="' . wp_create_nonce('archiver_immediate_snapshot') . '">';
|
|
echo '<div id="archiver-snapshots">';
|
|
if (empty($snapshots)) {
|
|
$urls_to_update = get_option('archiver_urls_to_update', array());
|
|
if (in_array($url, $urls_to_update)) {
|
|
esc_html_e('No archives yet. A snapshot request has been scheduled and will be processed soon.', 'archiver');
|
|
} else {
|
|
esc_html_e('There are no archives of this URL.', 'archiver');
|
|
}
|
|
} else {
|
|
$snapshots = array_slice($snapshots, 0, $this->snapshot_max_count);
|
|
$date_format = get_option('date_format');
|
|
$time_format = get_option('time_format');
|
|
echo '<ul>';
|
|
foreach ($snapshots as $snapshot) {
|
|
$date_time = date('Y-m-d H:i:s', strtotime($snapshot['timestamp']));
|
|
$adjusted_date = get_date_from_gmt($date_time);
|
|
$snapshot_url = $this->wayback_machine_url_view . $snapshot['timestamp'] . '/' . $snapshot['original'];
|
|
$date_time = date_i18n($date_format . ' @ ' . $time_format, strtotime($adjusted_date));
|
|
echo '<li><a href="' . esc_url($snapshot_url) . '" target="_blank">' . esc_html($date_time) . '</a></li>';
|
|
}
|
|
echo '</ul>';
|
|
}
|
|
echo '</div>';
|
|
echo '<hr />';
|
|
printf('<a href="%s" target="_external">%s</a>',
|
|
esc_url($this->wayback_machine_url_view . '*/' . $url),
|
|
esc_html__('See all snapshots ↗', 'archiver')
|
|
);
|
|
echo '<div style="margin-top: 10px;">';
|
|
echo '<button id="archiver-immediate-snapshot" class="button button-secondary">' . esc_html__('Archive Now', 'archiver') . '</button>';
|
|
echo '<span id="archiver-status" style="margin-left: 10px; display: none;"></span>';
|
|
echo '</div>';
|
|
}
|
|
|
|
public function ajax_immediate_snapshot() {
|
|
if (!check_ajax_referer('archiver_immediate_snapshot', '_ajax_nonce', false)) {
|
|
wp_send_json_error([
|
|
'message' => __('Security check failed. Please try again.', 'archiver')
|
|
], 403);
|
|
}
|
|
if (!current_user_can('edit_posts')) {
|
|
wp_send_json_error([
|
|
'message' => __('You do not have permission to perform this action.', 'archiver')
|
|
], 403);
|
|
}
|
|
$url = isset($_POST['url']) ? esc_url_raw($_POST['url']) : '';
|
|
if (empty($url)) {
|
|
wp_send_json_error([
|
|
'message' => __('Invalid URL provided.', 'archiver')
|
|
], 400);
|
|
}
|
|
$result = $this->trigger_wayback_machine_snapshot($url);
|
|
$snapshots = $this->fetch_snapshots_from_wayback($url);
|
|
if (!empty($snapshots)) {
|
|
$cache_key = 'archiver_snapshots_' . md5($url);
|
|
set_transient($cache_key, $snapshots, WEEK_IN_SECONDS);
|
|
$date_format = get_option('date_format');
|
|
$time_format = get_option('time_format');
|
|
$snapshot_list = [];
|
|
foreach ($snapshots as $snapshot) {
|
|
$date_time = date_i18n("$date_format @ $time_format", strtotime($snapshot['timestamp']));
|
|
$snapshot_url = $this->wayback_machine_url_view . $snapshot['timestamp'] . '/' . $snapshot['original'];
|
|
$snapshot_list[] = '<li><a href="'.esc_url($snapshot_url).'" target="_blank">'.esc_html($date_time).'</a></li>';
|
|
}
|
|
wp_send_json_success([
|
|
'message' => __('Snapshot created successfully!', 'archiver'),
|
|
'snapshots' => $snapshot_list
|
|
]);
|
|
} else {
|
|
wp_send_json_error([
|
|
'message' => __('Failed to retrieve snapshots.', 'archiver')
|
|
], 500);
|
|
}
|
|
}
|
|
|
|
public function add_admin_bar_links($wp_admin_bar) {
|
|
if (!current_user_can('edit_posts')) {
|
|
return;
|
|
}
|
|
$url = $this->get_current_permalink();
|
|
if (!$url) {
|
|
return;
|
|
}
|
|
$wp_admin_bar->add_node([
|
|
'id' => 'archiver',
|
|
'title' => '<span class="ab-icon dashicons dashicons-archive"></span><span class="ab-label">' . __('Archiver', 'archiver') . '</span>',
|
|
'href' => $this->wayback_machine_url_view . '*/' . $url,
|
|
'meta' => ['target' => '_blank']
|
|
]);
|
|
$snapshots = $this->get_post_snapshots();
|
|
$snapshot_count = is_wp_error($snapshots) ? 0 : count($snapshots);
|
|
if ($snapshot_count >= $this->snapshot_max_count) {
|
|
$snapshot_count = $this->snapshot_max_count . '+';
|
|
}
|
|
$wp_admin_bar->add_node([
|
|
'parent' => 'archiver',
|
|
'id' => 'archiver-snapshots',
|
|
'title' => __('Snapshots', 'archiver') . " ({$snapshot_count})",
|
|
'href' => $this->wayback_machine_url_view . '*/' . $url,
|
|
'meta' => ['target' => '_blank']
|
|
]);
|
|
$wp_admin_bar->add_node([
|
|
'parent' => 'archiver',
|
|
'id' => 'archiver-trigger',
|
|
'title' => __('Trigger Snapshot', 'archiver') . ' <span class="ab-icon"></span>',
|
|
'href' => '#',
|
|
'meta' => [
|
|
'class' => 'archiver-trigger'
|
|
]
|
|
]);
|
|
}
|
|
|
|
public function get_post_snapshots($url = '') {
|
|
$url = $url ? $url : $this->get_current_permalink();
|
|
if (empty($url)) {
|
|
return array();
|
|
}
|
|
$cache_key = 'archiver_snapshots_' . md5($url);
|
|
$snapshots = wp_cache_get($cache_key);
|
|
if (false === $snapshots) {
|
|
$snapshots = get_transient('archiver_last_known_snapshots_' . md5($url));
|
|
if (false === $snapshots) {
|
|
$this->record_url_for_update($url);
|
|
$snapshots = array();
|
|
} else {
|
|
wp_cache_set($cache_key, $snapshots, '', HOUR_IN_SECONDS);
|
|
}
|
|
}
|
|
return $snapshots;
|
|
}
|
|
|
|
private function fetch_snapshots_from_wayback($url) {
|
|
$fetch_url = add_query_arg([
|
|
'url' => $url,
|
|
'output' => 'json',
|
|
], $this->wayback_machine_url_fetch_archives);
|
|
$response = wp_remote_get($fetch_url, array(
|
|
'timeout' => 30,
|
|
'sslverify' => false
|
|
));
|
|
if (is_wp_error($response)) {
|
|
error_log('Archiver: Failed to fetch snapshots for ' . $url . '. Error: ' . $response->get_error_message());
|
|
return array();
|
|
}
|
|
$response_code = wp_remote_retrieve_response_code($response);
|
|
if (200 != $response_code) {
|
|
error_log('Archiver: Failed to fetch snapshots for ' . $url . '. Status code: ' . $response_code);
|
|
return array();
|
|
}
|
|
$data = json_decode(wp_remote_retrieve_body($response), true);
|
|
if (empty($data)) {
|
|
error_log('Archiver: Empty response data for ' . $url);
|
|
return array();
|
|
}
|
|
return $this->process_snapshot_data($data);
|
|
}
|
|
|
|
private function record_url_for_update($url, $priority = false) {
|
|
$urls_to_update = get_option('archiver_urls_to_update', array());
|
|
if (!in_array($url, $urls_to_update)) {
|
|
if ($priority) {
|
|
array_unshift($urls_to_update, $url);
|
|
} else {
|
|
$urls_to_update[] = $url;
|
|
}
|
|
update_option('archiver_urls_to_update', $urls_to_update);
|
|
}
|
|
}
|
|
|
|
private function fetch_and_cache_snapshots($url) {
|
|
$fetch_url = add_query_arg([
|
|
'url' => $url,
|
|
'output' => 'json',
|
|
], $this->wayback_machine_url_fetch_archives);
|
|
$response = wp_remote_get($fetch_url, array(
|
|
'timeout' => 30,
|
|
'sslverify' => false
|
|
));
|
|
if (is_wp_error($response)) {
|
|
error_log('Archiver: Failed to fetch snapshots for ' . $url . '. Error: ' . $response->get_error_message());
|
|
return false;
|
|
}
|
|
$response_code = wp_remote_retrieve_response_code($response);
|
|
if (200 != $response_code) {
|
|
error_log('Archiver: Failed to fetch snapshots for ' . $url . '. Status code: ' . $response_code);
|
|
return false;
|
|
}
|
|
$data = json_decode(wp_remote_retrieve_body($response), true);
|
|
if (empty($data)) {
|
|
error_log('Archiver: Empty response data for ' . $url);
|
|
return false;
|
|
}
|
|
$snapshots = $this->process_snapshot_data($data);
|
|
set_transient('archiver_last_known_snapshots_' . md5($url), $snapshots, WEEK_IN_SECONDS);
|
|
wp_cache_set('archiver_snapshots_' . md5($url), $snapshots, '', HOUR_IN_SECONDS);
|
|
return true;
|
|
}
|
|
|
|
private function trigger_wayback_machine_snapshot($url) {
|
|
$save_url = $this->wayback_machine_url_save . $url;
|
|
$response = wp_remote_get($save_url, array(
|
|
'timeout' => 10,
|
|
'sslverify' => false
|
|
));
|
|
if (is_wp_error($response)) {
|
|
error_log('Archiver: Failed to trigger Wayback Machine snapshot for ' . $url . '. Error: ' . $response->get_error_message());
|
|
return false;
|
|
}
|
|
$response_code = wp_remote_retrieve_response_code($response);
|
|
if (200 == $response_code) {
|
|
error_log('Archiver: Successfully triggered Wayback Machine snapshot for ' . $url);
|
|
return true;
|
|
}
|
|
error_log('Archiver: Failed to trigger Wayback Machine snapshot for ' . $url . '. Status: ' . $response_code);
|
|
return false;
|
|
}
|
|
|
|
private function process_snapshot_data($data) {
|
|
if (empty($data)) {
|
|
return array();
|
|
}
|
|
$field_columns = $data[0];
|
|
unset($data[0]);
|
|
$data = array_reverse($data);
|
|
$data = array_slice($data, 0, $this->snapshot_max_count);
|
|
$snapshots = array();
|
|
foreach ($data as $snapshot) {
|
|
$keyed_snapshot = array();
|
|
foreach ($snapshot as $i => $field) {
|
|
$keyed_snapshot[$field_columns[$i]] = $field;
|
|
}
|
|
$snapshots[] = $keyed_snapshot;
|
|
}
|
|
return $snapshots;
|
|
}
|
|
|
|
public function get_current_permalink() {
|
|
if (empty($this->current_permalink)) {
|
|
if (is_admin()) {
|
|
$this->current_permalink = $this->get_current_permalink_admin();
|
|
} else {
|
|
$this->current_permalink = $this->get_current_permalink_public();
|
|
}
|
|
}
|
|
return apply_filters('archiver_permalink', $this->current_permalink);
|
|
}
|
|
|
|
public function get_current_permalink_admin() {
|
|
$permalink = '';
|
|
$current_screen = get_current_screen();
|
|
$object_type = $current_screen->base;
|
|
switch ($object_type) {
|
|
case 'post':
|
|
global $post;
|
|
if ($post && $post->ID) {
|
|
$permalink = get_permalink($post->ID);
|
|
} else {
|
|
$post_id = isset($_GET['post']) ? intval($_GET['post']) : 0;
|
|
if ($post_id) {
|
|
$permalink = get_permalink($post_id);
|
|
}
|
|
}
|
|
break;
|
|
case 'term':
|
|
global $taxnow, $tag;
|
|
$taxonomy = $taxnow;
|
|
$term_id = intval($tag->term_id);
|
|
$permalink = get_term_link($term_id, $taxonomy);
|
|
break;
|
|
case 'profile':
|
|
case 'user-edit':
|
|
$user_id = !empty($_GET['user_id']) ? intval($_GET['user_id']) : get_current_user_id();
|
|
$permalink = get_author_posts_url($user_id);
|
|
break;
|
|
}
|
|
return apply_filters('archiver_permalink_admin', $permalink);
|
|
}
|
|
|
|
public function get_current_permalink_public() {
|
|
global $wp;
|
|
$permalink = home_url($wp->request);
|
|
if (!empty($_SERVER['QUERY_STRING'])) {
|
|
$permalink .= '?' . $_SERVER['QUERY_STRING'];
|
|
}
|
|
return apply_filters('archiver_permalink_public', $permalink);
|
|
}
|
|
|
|
public function register_scripts_and_styles() {
|
|
wp_register_script(
|
|
'archiver',
|
|
ARCHIVER_PLUGIN_DIR_URL . 'js/archiver' . $this->min_suffix . '.js',
|
|
array('jquery', 'wp-api-request', 'wp-i18n'),
|
|
filemtime(ARCHIVER_PLUGIN_DIR_PATH . 'js/archiver' . $this->min_suffix . '.js'),
|
|
true
|
|
);
|
|
wp_register_style(
|
|
'archiver',
|
|
ARCHIVER_PLUGIN_DIR_URL . 'css/archiver' . $this->min_suffix . '.css',
|
|
array('dashicons'),
|
|
filemtime(ARCHIVER_PLUGIN_DIR_PATH . 'css/archiver' . $this->min_suffix . '.css')
|
|
);
|
|
wp_localize_script('archiver', 'archiver', array(
|
|
'ajax_url' => admin_url('admin-ajax.php'),
|
|
'rest_url' => rest_url('archiver/v1/trigger-snapshot'),
|
|
'nonce' => wp_create_nonce('wp_rest'),
|
|
'url' => $this->get_current_permalink(),
|
|
'i18n' => array(
|
|
'triggering' => __('Triggering snapshot...', 'archiver'),
|
|
'success' => __('Snapshot triggered successfully!', 'archiver'),
|
|
'error' => __('Failed to trigger snapshot.', 'archiver')
|
|
)
|
|
));
|
|
}
|
|
|
|
public function enqueue_scripts() {
|
|
$url = $this->get_current_permalink();
|
|
if (!$url) {
|
|
return;
|
|
}
|
|
wp_enqueue_script('archiver');
|
|
wp_enqueue_style('archiver');
|
|
}
|
|
|
|
public function admin_enqueue_scripts($hook) {
|
|
if ('tools_page_archiver-settings' === $hook) {
|
|
wp_enqueue_style('archiver');
|
|
}
|
|
wp_enqueue_script('archiver');
|
|
}
|
|
|
|
public function do_admin_notice_disabled() {
|
|
$id = 'archiver-notice-disabled';
|
|
$dismiss_notice_key = 'archiver_dismiss_notice_' . $id;
|
|
if (get_user_meta(get_current_user_id(), $dismiss_notice_key)) {
|
|
return;
|
|
}
|
|
$class = 'archiver-notice notice notice-error is-dismissible';
|
|
$message = __("Archiver is currently disabled via the <code>archiver_can_run</code> filter.", 'archiver');
|
|
printf('<div id="%s" class="%s"><p>%s</p></div>', $id, $class, $message);
|
|
}
|
|
|
|
public function register_rest_routes() {
|
|
register_rest_route('archiver/v1', '/trigger-snapshot', [
|
|
'methods' => 'POST',
|
|
'callback' => array($this, 'rest_trigger_snapshot'),
|
|
'permission_callback' => function() {
|
|
return current_user_can('edit_posts');
|
|
}
|
|
]);
|
|
}
|
|
|
|
public function rest_trigger_snapshot($request) {
|
|
if (!current_user_can('edit_posts')) {
|
|
return new WP_Error('rest_forbidden', __('You do not have permissions to trigger snapshots.', 'archiver'), array('status' => 401));
|
|
}
|
|
$url = $request->get_param('url');
|
|
if (empty($url)) {
|
|
return new WP_Error('rest_invalid_param', __('Invalid URL parameter.', 'archiver'), array('status' => 400));
|
|
}
|
|
$this->record_url_for_update($url, true);
|
|
return new WP_REST_Response(array(
|
|
'success' => true,
|
|
'message' => __('Snapshot request recorded and will be processed soon.', 'archiver')
|
|
), 200);
|
|
}
|
|
|
|
public function process_urls_for_update() {
|
|
$urls_to_update = get_option('archiver_urls_to_update', array());
|
|
$processed_urls = array();
|
|
$selected_post_types = get_option('archiver_post_types', array('post', 'page'));
|
|
$normalized_urls = array();
|
|
$valid_urls = array();
|
|
$total_archived = (int) get_option('archiver_total_archived', 0);
|
|
$failed_snapshots = (int) get_option('archiver_failed_snapshots', 0);
|
|
|
|
foreach ($urls_to_update as $url) {
|
|
$normalized_url = $this->normalize_url($url);
|
|
if (empty($normalized_url)) {
|
|
$processed_urls[] = $url;
|
|
continue;
|
|
}
|
|
if (!in_array($normalized_url, $normalized_urls)) {
|
|
$normalized_urls[] = $normalized_url;
|
|
$valid_urls[$normalized_url] = $url;
|
|
} else {
|
|
$processed_urls[] = $url;
|
|
error_log("Archiver: Duplicate URL removed: $url");
|
|
}
|
|
}
|
|
|
|
foreach ($normalized_urls as $normalized_url) {
|
|
$url = $valid_urls[$normalized_url];
|
|
$post_id = $this->get_post_id_from_url($url);
|
|
if ($post_id) {
|
|
$post = get_post($post_id);
|
|
if (!$post || $post->post_status === 'draft' || $post->post_status === 'private' || !empty($post->post_password) || $this->is_preview_url($url)) {
|
|
$processed_urls[] = $url;
|
|
continue;
|
|
}
|
|
}
|
|
if (!$post_id || ($post_id && in_array(get_post_type($post_id), $selected_post_types))) {
|
|
$fetch_result = $this->fetch_and_cache_snapshots($normalized_url);
|
|
$processed_urls[] = $url;
|
|
if ($fetch_result) {
|
|
$snapshot_result = $this->trigger_wayback_machine_snapshot($normalized_url);
|
|
if ($snapshot_result) {
|
|
$total_archived++;
|
|
} else {
|
|
$failed_snapshots++;
|
|
}
|
|
} else {
|
|
$failed_snapshots++;
|
|
}
|
|
if (count($processed_urls) >= 5) {
|
|
break;
|
|
}
|
|
} else {
|
|
$processed_urls[] = $url;
|
|
}
|
|
}
|
|
|
|
$remaining_urls = array_diff($urls_to_update, $processed_urls);
|
|
update_option('archiver_urls_to_update', array_values($remaining_urls));
|
|
update_option('archiver_last_run', current_time('mysql'));
|
|
update_option('archiver_total_archived', $total_archived);
|
|
update_option('archiver_failed_snapshots', $failed_snapshots);
|
|
error_log("Archiver: Updated pending URLs: " . print_r($remaining_urls, true));
|
|
}
|
|
|
|
private function normalize_url($url) {
|
|
$parsed_url = parse_url($url);
|
|
$normalized = isset($parsed_url['scheme']) ? $parsed_url['scheme'] . '://' : 'https://';
|
|
$normalized .= isset($parsed_url['host']) ? $parsed_url['host'] : '';
|
|
$normalized .= isset($parsed_url['path']) ? $parsed_url['path'] : '/';
|
|
$normalized = preg_replace('/^www\./i', '', $normalized);
|
|
$normalized = rtrim($normalized, '/');
|
|
if (isset($parsed_url['query'])) {
|
|
parse_str($parsed_url['query'], $query_params);
|
|
if (isset($query_params['preview'])) {
|
|
error_log("Archiver: Excluding preview URL: $url");
|
|
return '';
|
|
}
|
|
if (isset($query_params['p'])) {
|
|
$post_id = intval($query_params['p']);
|
|
$post = get_post($post_id);
|
|
if ($post && $post->post_status === 'publish' && empty($post->post_password)) {
|
|
$normalized = get_permalink($post_id);
|
|
error_log("Archiver: Normalized $url to $normalized");
|
|
} else {
|
|
error_log("Archiver: Excluding draft/protected URL: $url");
|
|
return '';
|
|
}
|
|
}
|
|
}
|
|
$normalized = rtrim($normalized, '/');
|
|
error_log("Archiver: Normalized URL: $url -> $normalized");
|
|
return $normalized;
|
|
}
|
|
|
|
private function get_post_id_from_url($url) {
|
|
$post_id = url_to_postid($url);
|
|
if (!$post_id) {
|
|
$parsed_url = parse_url($url);
|
|
if (isset($parsed_url['query'])) {
|
|
parse_str($parsed_url['query'], $query_params);
|
|
if (isset($query_params['p'])) {
|
|
$post_id = intval($query_params['p']);
|
|
}
|
|
}
|
|
}
|
|
return $post_id;
|
|
}
|
|
|
|
private function is_preview_url($url) {
|
|
$parsed_url = parse_url($url);
|
|
if (isset($parsed_url['query'])) {
|
|
parse_str($parsed_url['query'], $query_params);
|
|
return isset($query_params['preview']) && $query_params['preview'] === 'true';
|
|
}
|
|
return false;
|
|
}
|
|
|
|
public function trigger_post_snapshot($post_id) {
|
|
if ('publish' != get_post_status($post_id) || wp_is_post_revision($post_id)) {
|
|
return;
|
|
}
|
|
$url = get_permalink($post_id);
|
|
$this->trigger_url_snapshot($url);
|
|
}
|
|
|
|
public function trigger_term_snapshot($term_id, $taxonomy_id, $taxonomy) {
|
|
$url = get_term_link($term_id, $taxonomy);
|
|
if (is_wp_error($url)) {
|
|
return;
|
|
}
|
|
$this->trigger_url_snapshot($url);
|
|
}
|
|
|
|
public function trigger_user_snapshot($user_id) {
|
|
$url = get_author_posts_url($user_id);
|
|
$this->trigger_url_snapshot($url);
|
|
}
|
|
|
|
public function trigger_url_snapshot($url) {
|
|
if (empty($url)) {
|
|
return new WP_Error('empty_url', __('URL cannot be empty.', 'archiver'));
|
|
}
|
|
$last_snapshot = get_transient('archiver_last_snapshot_' . md5($url));
|
|
if ($last_snapshot) {
|
|
return new WP_Error('snapshot_throttled', __('Snapshot already taken recently.', 'archiver'));
|
|
}
|
|
$response = wp_safe_remote_get($this->wayback_machine_url_save . $url, array(
|
|
'timeout' => 30,
|
|
'sslverify' => false
|
|
));
|
|
if (is_wp_error($response)) {
|
|
return $response;
|
|
}
|
|
$response_code = wp_remote_retrieve_response_code($response);
|
|
if (200 === $response_code) {
|
|
set_transient('archiver_last_snapshot_' . md5($url), time(), HOUR_IN_SECONDS);
|
|
return true;
|
|
} else {
|
|
return new WP_Error('snapshot_failed', __('Failed to trigger snapshot.', 'archiver'));
|
|
}
|
|
}
|
|
|
|
public function reschedule_cron_task($frequency) {
|
|
$timestamp = wp_next_scheduled('archiver_process_urls');
|
|
if ($timestamp) {
|
|
wp_unschedule_event($timestamp, 'archiver_process_urls');
|
|
}
|
|
wp_schedule_event(time(), $frequency, 'archiver_process_urls');
|
|
}
|
|
|
|
public static function deactivate() {
|
|
$timestamp = wp_next_scheduled('archiver_process_urls');
|
|
if ($timestamp) {
|
|
wp_unschedule_event($timestamp, 'archiver_process_urls');
|
|
}
|
|
delete_option('archiver_urls_to_update');
|
|
}
|
|
} |
