/**
* Print an XML sitemap conforming to the Sitemaps.org protocol.
* Outputs an XML list of up to the latest 1000 posts.
*
* @module sitemaps
*
* @link http://sitemaps.org/protocol.php Sitemaps.org protocol.
*/
function jetpack_print_sitemap()
{
global $wpdb, $post;
$xml = get_transient('jetpack_sitemap');
if ($xml) {
header('Content-Type: ' . jetpack_sitemap_content_type(), true);
echo $xml;
die;
}
// Compatibility with PHP 5.3 and older
if (!defined('ENT_XML1')) {
define('ENT_XML1', 16);
}
/**
* Filter the post types that will be included in sitemap.
*
* @module sitemaps
*
* @since 3.9.0
*
* @param array $post_types Array of post types.
*/
$post_types = apply_filters('jetpack_sitemap_post_types', array('post', 'page'));
$post_types_in = array();
foreach ((array) $post_types as $post_type) {
$post_types_in[] = $wpdb->prepare('%s', $post_type);
}
$post_types_in = join(",", $post_types_in);
// use direct query instead because get_posts was acting too heavy for our needs
//$posts = get_posts( array( 'numberposts'=>1000, 'post_type'=>$post_types, 'post_status'=>'published' ) );
$posts = $wpdb->get_results("SELECT ID, post_type, post_modified_gmt, comment_count FROM {$wpdb->posts} WHERE post_status='publish' AND post_type IN ({$post_types_in}) ORDER BY post_modified_gmt DESC LIMIT 1000");
if (empty($posts)) {
status_header(404);
}
header('Content-Type: ' . jetpack_sitemap_content_type());
$initstr = jetpack_sitemap_initstr(get_bloginfo('charset'));
$tree = simplexml_load_string($initstr);
// If we did not get a valid string, force UTF-8 and try again.
if (false === $tree) {
$initstr = jetpack_sitemap_initstr('UTF-8');
$tree = simplexml_load_string($initstr);
}
unset($initstr);
$latest_mod = '';
foreach ($posts as $post) {
setup_postdata($post);
/**
* Filter condition to allow skipping specific posts in sitemap.
*
* @module sitemaps
*
* @since 3.9.0
*
* @param bool $skip Current boolean. False by default, so no post is skipped.
* @param WP_POST $post Current post object.
*/
if (apply_filters('jetpack_sitemap_skip_post', false, $post)) {
continue;
}
$post_latest_mod = null;
$url = array('loc' => esc_url(get_permalink($post->ID)));
// If this post is configured to be the site home, skip since it's added separately later
if (untrailingslashit(get_permalink($post->ID)) == untrailingslashit(get_option('home'))) {
continue;
}
// Mobile node specified in http://support.google.com/webmasters/bin/answer.py?hl=en&answer=34648
$url['mobile:mobile'] = '';
// Image node specified in http://support.google.com/webmasters/bin/answer.py?hl=en&answer=178636
// These attachments were produced with batch SQL earlier in the script
if (!post_password_required($post->ID)) {
$media = array();
$methods = array('from_thumbnail' => false, 'from_slideshow' => false, 'from_gallery' => false, 'from_attachment' => false, 'from_html' => false);
foreach ($methods as $method => $value) {
$methods[$method] = true;
$images_collected = Jetpack_PostImages::get_images($post->ID, $methods);
if (is_array($images_collected)) {
$media = array_merge($media, $images_collected);
}
$methods[$method] = false;
}
$images = array();
foreach ($media as $item) {
if (!isset($item['type']) || 'image' != $item['type']) {
continue;
}
$one_image = array();
if (isset($item['src'])) {
$one_image['image:loc'] = esc_url($item['src']);
$one_image['image:title'] = sanitize_title_with_dashes($name = pathinfo($item['src'], PATHINFO_FILENAME));
}
$images[] = $one_image;
}
if (!empty($images)) {
$url['image:image'] = $images;
}
}
if ($post->post_modified_gmt && $post->post_modified_gmt != '0000-00-00 00:00:00') {
$post_latest_mod = $post->post_modified_gmt;
}
if ($post->comment_count > 0) {
// last modified based on last comment
$latest_comment_datetime = jetpack_get_approved_comments_max_datetime($post->ID);
if (!empty($latest_comment_datetime)) {
if (is_null($post_latest_mod) || $latest_comment_datetime > $post_latest_mod) {
$post_latest_mod = $latest_comment_datetime;
}
}
unset($latest_comment_datetime);
}
if (!empty($post_latest_mod)) {
$latest_mod = max($latest_mod, $post_latest_mod);
$url['lastmod'] = jetpack_w3cdate_from_mysql($post_latest_mod);
}
unset($post_latest_mod);
if ($post->post_type == 'page') {
$url['changefreq'] = 'weekly';
$url['priority'] = '0.6';
// set page priority above default priority of 0.5
} else {
$url['changefreq'] = 'monthly';
}
/**
* Filter associative array with data to build <url> node and its descendants for current post.
*
* @module sitemaps
*
* @since 3.9.0
*
* @param array $url Data to build parent and children nodes for current post.
* @param int $post_id Current post ID.
*/
$url_node = apply_filters('jetpack_sitemap_url', $url, $post->ID);
jetpack_sitemap_array_to_simplexml(array('url' => $url_node), $tree);
unset($url);
}
wp_reset_postdata();
$blog_home = array('loc' => esc_url(get_option('home')), 'changefreq' => 'daily', 'priority' => '1.0');
if (!empty($latest_mod)) {
$blog_home['lastmod'] = jetpack_w3cdate_from_mysql($latest_mod);
header('Last-Modified:' . mysql2date('D, d M Y H:i:s', $latest_mod, 0) . ' GMT');
}
/**
* Filter associative array with data to build <url> node and its descendants for site home.
*
* @module sitemaps
*
* @since 3.9.0
*
* @param array $blog_home Data to build parent and children nodes for site home.
*/
$url_node = apply_filters('jetpack_sitemap_url_home', $blog_home);
jetpack_sitemap_array_to_simplexml(array('url' => $url_node), $tree);
unset($blog_home);
/**
* Filter data before rendering it as XML.
*
* @module sitemaps
*
* @since 3.9.0
*
* @param SimpleXMLElement $tree Data tree for sitemap.
* @param string $latest_mod Date of last modification.
*/
$tree = apply_filters('jetpack_print_sitemap', $tree, $latest_mod);
$xml = $tree->asXML();
unset($tree);
if (!empty($xml)) {
set_transient('jetpack_sitemap', $xml, DAY_IN_SECONDS);
echo $xml;
}
die;
}
/**
* Print an XML sitemap conforming to the Sitemaps.org protocol.
* Outputs an XML list of up to the latest 1000 posts.
*
* @module sitemaps
*
* @link http://sitemaps.org/protocol.php Sitemaps.org protocol.
*/
function jetpack_print_sitemap()
{
global $wpdb;
$xml = get_transient('jetpack_sitemap');
if ($xml) {
header('Content-Type: ' . jetpack_sitemap_content_type(), true);
echo $xml;
die;
}
// Compatibility with PHP 5.3 and older
if (!defined('ENT_XML1')) {
define('ENT_XML1', 16);
}
/**
* Filter the post types that will be included in sitemap.
*
* @module sitemaps
*
* @since 3.9.0
*
* @param array $post_types Array of post types.
*/
$post_types = apply_filters('jetpack_sitemap_post_types', array('post', 'page'));
$post_types_in = array();
foreach ((array) $post_types as $post_type) {
$post_types_in[] = $wpdb->prepare('%s', $post_type);
}
$post_types_in = join(",", $post_types_in);
// use direct query instead because get_posts was acting too heavy for our needs
//$posts = get_posts( array( 'numberposts'=>1000, 'post_type'=>$post_types, 'post_status'=>'published' ) );
$posts = $wpdb->get_results("SELECT ID, post_type, post_modified_gmt, comment_count FROM {$wpdb->posts} WHERE post_status='publish' AND post_type IN ({$post_types_in}) ORDER BY post_modified_gmt DESC LIMIT 1000");
if (empty($posts)) {
status_header(404);
}
header('Content-Type: ' . jetpack_sitemap_content_type());
$initstr = jetpack_sitemap_initstr(get_bloginfo('charset'));
$tree = simplexml_load_string($initstr);
// If we did not get a valid string, force UTF-8 and try again.
if (false === $tree) {
$initstr = jetpack_sitemap_initstr('UTF-8');
$tree = simplexml_load_string($initstr);
}
// Acquire necessary attachment data for all of the posts in a performant manner
$attachment_parents = wp_list_pluck($posts, 'ID');
$post_attachments = array();
while ($sub_posts = array_splice($attachment_parents, 0, 100)) {
$post_parents = implode(',', array_map('intval', $sub_posts));
// Get the attachment IDs for all posts. We need to see how many
// attachments each post parent has and limit it to 5.
$query = "SELECT ID, post_parent FROM {$wpdb->posts} WHERE post_parent IN ({$post_parents}) AND post_type='attachment' AND ( post_mime_type='image/jpeg' OR post_mime_type='image/png' ) LIMIT 0,1000;";
$all_attachments = $wpdb->get_results($query);
$selected_attachments = array();
$attachment_count = array();
foreach ($all_attachments as $attachment) {
if (!isset($attachment_count[$attachment->post_parent])) {
$attachment_count[$attachment->post_parent] = 0;
}
// Skip this particular attachment if we already have 5 for the post
if ($attachment_count[$attachment->post_parent] >= 5) {
continue;
}
$selected_attachments[] = $attachment->ID;
$attachment_count[$attachment->post_parent]++;
}
// bail if there weren't any attachments to avoid an extra query
if (empty($selected_attachments)) {
continue;
}
// Get more of the attachment object for the attachments we actually care about
$attachment_ids = implode(',', array_map('intval', $selected_attachments));
$query = "SELECT p.ID, p.post_parent, p.post_title, p.post_excerpt, p.guid FROM {$wpdb->posts} as p WHERE p.ID IN ({$attachment_ids}) AND p.post_type='attachment' AND ( p.post_mime_type='image/jpeg' OR p.post_mime_type='image/png' ) LIMIT 500;";
$attachments = $wpdb->get_results($query);
$post_attachments = array_merge($post_attachments, $attachments);
}
unset($initstr);
$latest_mod = '';
foreach ($posts as $post) {
/**
* Filter condition to allow skipping specific posts in sitemap.
*
* @module sitemaps
*
* @since 3.9.0
*
* @param bool $skip Current boolean. False by default, so no post is skipped.
* @param WP_POST $post Current post object.
*/
if (apply_filters('jetpack_sitemap_skip_post', false, $post)) {
continue;
}
$post_latest_mod = null;
$url = array('loc' => esc_url(get_permalink($post->ID)));
// If this post is configured to be the site home, skip since it's added separately later
if (untrailingslashit(get_permalink($post->ID)) == untrailingslashit(get_option('home'))) {
continue;
}
// Mobile node specified in http://support.google.com/webmasters/bin/answer.py?hl=en&answer=34648
$url['mobile:mobile'] = '';
// Image node specified in http://support.google.com/webmasters/bin/answer.py?hl=en&answer=178636
// These attachments were produced with batch SQL earlier in the script
if (!post_password_required($post->ID) && ($attachments = wp_filter_object_list($post_attachments, array('post_parent' => $post->ID)))) {
$url['image:image'] = array();
foreach ($attachments as $attachment) {
$attachment_url = wp_get_attachment_url($attachment->ID);
if ($attachment_url) {
$url['image:image']['loc'] = esc_url($attachment_url);
}
// Only include title if not empty.
/** This filter is documented in wp-includes/feed.php */
if ($attachment_title = apply_filters('the_title_rss', $attachment->post_title)) {
$url['image:image']['title'] = html_entity_decode(esc_html($attachment_title), ENT_XML1);
}
// Only include caption if not empty.
/** This filter is documented in wp-includes/feed.php */
if ($attachment_caption = apply_filters('the_excerpt_rss', $attachment->post_excerpt)) {
$url['image:image']['caption'] = html_entity_decode(esc_html($attachment_caption), ENT_XML1);
}
}
}
if ($post->post_modified_gmt && $post->post_modified_gmt != '0000-00-00 00:00:00') {
$post_latest_mod = $post->post_modified_gmt;
}
if ($post->comment_count > 0) {
// last modified based on last comment
$latest_comment_datetime = jetpack_get_approved_comments_max_datetime($post->ID);
if (!empty($latest_comment_datetime)) {
if (is_null($post_latest_mod) || $latest_comment_datetime > $post_latest_mod) {
$post_latest_mod = $latest_comment_datetime;
}
}
unset($latest_comment_datetime);
}
if (!empty($post_latest_mod)) {
$latest_mod = max($latest_mod, $post_latest_mod);
$url['lastmod'] = jetpack_w3cdate_from_mysql($post_latest_mod);
}
unset($post_latest_mod);
if ($post->post_type == 'page') {
$url['changefreq'] = 'weekly';
$url['priority'] = '0.6';
// set page priority above default priority of 0.5
} else {
$url['changefreq'] = 'monthly';
}
/**
* Filter associative array with data to build <url> node and its descendants for current post.
*
* @module sitemaps
*
* @since 3.9.0
*
* @param array $url Data to build parent and children nodes for current post.
* @param int $post_id Current post ID.
*/
$url_node = apply_filters('jetpack_sitemap_url', $url, $post->ID);
jetpack_sitemap_array_to_simplexml(array('url' => $url_node), $tree);
unset($url);
}
$blog_home = array('loc' => esc_url(get_option('home')), 'changefreq' => 'daily', 'priority' => '1.0');
if (!empty($latest_mod)) {
$blog_home['lastmod'] = jetpack_w3cdate_from_mysql($latest_mod);
header('Last-Modified:' . mysql2date('D, d M Y H:i:s', $latest_mod, 0) . ' GMT');
}
/**
* Filter associative array with data to build <url> node and its descendants for site home.
*
* @module sitemaps
*
* @since 3.9.0
*
* @param array $blog_home Data to build parent and children nodes for site home.
*/
$url_node = apply_filters('jetpack_sitemap_url_home', $blog_home);
jetpack_sitemap_array_to_simplexml(array('url' => $url_node), $tree);
unset($blog_home);
/**
* Filter data before rendering it as XML.
*
* @module sitemaps
*
* @since 3.9.0
*
* @param SimpleXMLElement $tree Data tree for sitemap.
* @param string $latest_mod Date of last modification.
*/
$tree = apply_filters('jetpack_print_sitemap', $tree, $latest_mod);
$xml = $tree->asXML();
unset($tree);
if (!empty($xml)) {
set_transient('jetpack_sitemap', $xml, DAY_IN_SECONDS);
echo $xml;
}
die;
}
