URLs
*
* @since 2.11
*
* @var null|array $image_ids_urls {
* @type string $base_url The URL for the original sized image.
* @type string ${$id} Contains the URLs associated to the IDs.
* }
*/
private $image_ids_urls = null;
/**
* All_in_One_SEO_Pack_Sitemap constructor.
*
* @since ?
*/
public function __construct() {
if ( get_class( $this ) === 'All_in_One_SEO_Pack_Sitemap' ) { // Set this up only when instantiated as this class.
$this->name = __( 'XML Sitemap', 'all-in-one-seo-pack' ); // Human-readable name of the plugin.
$this->prefix = 'aiosp_sitemap_'; // Option prefix.
$this->file = __FILE__; // The current file.
$this->extra_sitemaps = array();
$this->extra_sitemaps = apply_filters( $this->prefix . 'extra', $this->extra_sitemaps );
}
parent::__construct();
// TODO This could be move up to the class field default/initial values.
$this->comment_string = 'Sitemap %s generated by ' . AIOSEOP_PLUGIN_NAME . ' %s on %s';
$this->default_options = array(
'daily_cron' => array(
'name' => __( 'Schedule Updates', 'all-in-one-seo-pack' ),
'type' => 'select',
'initial_options' => array(
0 => __( 'No Schedule', 'all-in-one-seo-pack' ),
'daily' => __( 'Daily', 'all-in-one-seo-pack' ),
'weekly' => __( 'Weekly', 'all-in-one-seo-pack' ),
'monthly' => __( 'Monthly', 'all-in-one-seo-pack' ),
),
'default' => 0,
),
'indexes' => array(
'name' => __( 'Enable Sitemap Indexes', 'all-in-one-seo-pack' ),
'default' => 'on',
),
'max_posts' => array(
'name' => __( 'Maximum Posts Per Sitemap Page', 'all-in-one-seo-pack' ),
'type' => 'number',
'default' => 1000,
'condshow' => array(
"{$this->prefix}indexes" => 'on',
"{$this->prefix}indexes" => 'on',
),
),
'posttypes' => array(
'name' => __( 'Post Types', 'all-in-one-seo-pack' ),
'type' => 'multicheckbox',
'default' => 'all',
),
'taxonomies' => array(
'name' => __( 'Taxonomies', 'all-in-one-seo-pack' ),
'type' => 'multicheckbox',
'default' => 'all',
),
'archive' => array( 'name' => __( 'Include Date Archive Pages', 'all-in-one-seo-pack' ) ),
'author' => array( 'name' => __( 'Include Author Pages', 'all-in-one-seo-pack' ) ),
'images' => array( 'name' => __( 'Exclude Images', 'all-in-one-seo-pack' ) ),
'rewrite' => array(
'name' => __( 'Dynamically Generate Sitemap', 'all-in-one-seo-pack' ),
'default' => 'On',
),
);
$status_options = array(
'link' => array(
'default' => '',
'type' => 'html',
'label' => 'none',
'save' => false,
),
);
$this->layout = array(
'status' => array(
'name' => __( 'Sitemap Overview', 'all-in-one-seo-pack' ),
'help_link' => 'https://semperplugins.com/documentation/xml-sitemaps-module/#sitemap-overview',
'options' => array_keys( $status_options ),
),
'default' => array(
'name' => $this->name,
'help_link' => 'https://semperplugins.com/documentation/xml-sitemaps-module/#xml-sitemap',
'options' => array_keys( $this->default_options ),
),
);
$prio = array();
for ( $i = 0; $i <= 10; $i ++ ) {
$str = sprintf( '%0.1f', $i / 10.0 );
$prio[ $str ] = $str;
}
$arr_no = array( 'no' => __( 'Do Not Override', 'all-in-one-seo-pack' ) );
$arr_sel = array( 'sel' => __( 'Select Individual', 'all-in-one-seo-pack' ) );
$this->prio_sel = array_merge( $arr_no, $arr_sel, $prio );
$this->prio = array_merge( $arr_no, $prio );
$freq = array();
foreach ( array( 'always', 'hourly', 'daily', 'weekly', 'monthly', 'yearly', 'never' ) as $f ) {
$freq[ $f ] = $f;
}
$this->freq_sel = array_merge( $arr_no, $arr_sel, $freq );
$this->freq = array_merge( $arr_no, $freq );
foreach (
array(
'prio' => __( 'priority', 'all-in-one-seo-pack' ),
'freq' => __( 'frequency', 'all-in-one-seo-pack' ),
) as $k => $v
) {
$s = "{$k}_options";
$$s = array();
foreach (
array(
'homepage' => __( 'homepage', 'all-in-one-seo-pack' ),
'post' => __( 'posts', 'all-in-one-seo-pack' ),
'taxonomies' => __( 'taxonomies', 'all-in-one-seo-pack' ),
'archive' => __( 'archive pages', 'all-in-one-seo-pack' ),
'author' => __( 'author pages', 'all-in-one-seo-pack' ),
) as $opt => $val
) {
$arr = $$s;
if ( ( 'post' === $opt ) || ( 'taxonomies' === $opt ) ) {
$iopts = $this->{"{$k}_sel"};
} else {
$iopts = $this->$k;
}
$arr[ $k . '_' . $opt ] = array(
'name' => AIOSEOP_PHP_Functions::ucwords( $val ),
'type' => 'select',
'initial_options' => $iopts,
'default' => 'no',
);
if ( ( 'archive' === $opt ) || ( 'author' === $opt ) ) {
$arr[ $k . '_' . $opt ]['condshow'] = array( $this->prefix . $opt => 'on' );
}
$$s = $arr;
}
}
$addl_sitemap_options = array(
'rss_sitemap' => array( 'name' => __( 'Create RSS Sitemap', 'all-in-one-seo-pack' ) ),
);
/**
* Allows users to disable the Google News sitemap.
*
* @since 3.4.0
*
* @param bool Whether or not the Google News sitemap should be output. Defaults to true.
*/
if ( apply_filters( 'aioseo_news_sitemap_enabled', true ) ) {
$addl_sitemap_options['publication_name'] = array(
'name' => __( 'Google News Publication Name', 'all-in-one-seo-pack' ),
'type' => 'text',
'default' => get_bloginfo( 'name' ) ? get_bloginfo( 'name' ) : "",
);
$addl_sitemap_options['posttypes_news'] = array(
'name' => __( 'Google News Sitemap Post Types', 'all-in-one-seo-pack' ),
'type' => 'multicheckbox',
'default' => array( 'post' ),
);
}
$addl_pages_options = array(
'addl_instructions' => array(
'default' => '' . __( 'Enter information below for any additional links for your sitemap not already managed through WordPress.', 'all-in-one-seo-pack' ) . '
";
if ( ! empty( $args['value'] ) ) {
$buf .= "
\n";
foreach ( $args['value'] as $k => $v ) {
if ( is_object( $v ) ) {
$v = (array) $v;
}
$buf .= "\t{$k} {$v['prio']} {$v['freq']} {$v['mod']}
\n";
}
}
$args['options']['type'] = 'hidden';
if ( ! empty( $args['value'] ) ) {
$args['value'] = wp_json_encode( $args['value'] );
} else {
$args['options']['type'] = 'html';
}
if ( empty( $args['value'] ) ) {
$args['value'] = '';
}
$buf .= $this->get_option_html( $args );
$buf .= '
' . __( 'You can navigate to your sitemap(s) using the links below:', 'all-in-one-seo-pack' ) . '
';
$url = aioseop_home_url( '/' . 'sitemap.xml' );
$options[ $this->prefix . 'link' ] .= '' . __( 'XML Sitemap', 'all-in-one-seo-pack' ) . ' ' . __( 'RSS Sitemap', 'all-in-one-seo-pack' ) . ' ' . __( 'Google News Sitemap', 'all-in-one-seo-pack' ) . ' (' . __( 'This will lead to a 404 Not Found page if no posts have been published in the last 48 hours.', 'all-in-one-seo-pack' ) . ') ' . __( 'Video Sitemap', 'all-in-one-seo-pack' ) . ' ';
if ( '0' !== get_option( 'blog_public' ) ) {
$options[ $this->prefix . 'link' ] .= '' . __( 'Any changes are automatically submitted to search engines.', 'all-in-one-seo-pack' ) . '
';
}
if ( $this->option_isset( 'rewrite' ) ) {
switch( $this->prefix ) {
case 'aiosp_sitemap_': {
$url = aioseop_home_url( '/' . 'sitemap.xml' );
break;
}
case 'aiosp_video_sitemap_':{
$url = aioseop_home_url( '/' . 'video-sitemap.xml' );
break;
}
default:
break;
}
if ( $url ) {
$rule = $this->get_rewrite_url( $url );
$rules = $this->get_rewrite_rules();
if ( ! in_array( $rule, $rules, true ) ) {
$options[ $this->prefix . 'link' ] .= '' . __( 'Dynamic sitemap generation does not appear to be using the correct rewrite rules; please disable any other sitemap plugins or functionality on your site and reset your permalinks.', 'all-in-one-seo-pack' ) . ' ';
}
}
}
if ( ! get_option( 'blog_public' ) ) {
$privacy_link = '' . __( 'Reading Settings', 'all-in-one-seo-pack' ) . ' ';
/* translators: Link to settings to disable "Discourage search engines from indexing this site". */
$options[ $this->prefix . 'link' ] .= '' . sprintf( __( 'Warning: your privacy settings are configured to ask search engines to not index your site; you can change this under %s for your site.', 'all-in-one-seo-pack' ), $privacy_link );
}
$excl_terms = array();
if ( isset( $options[ $this->prefix . 'excl_terms' ] ) && is_array( $options[ $this->prefix . 'excl_terms' ] ) ) {
foreach ( $options[ $this->prefix . 'excl_terms' ] as $k1_taxonomy => $v1_tax_terms ) {
foreach ( $v1_tax_terms['terms'] as $v2_term ) {
$excl_terms[] = $k1_taxonomy . '-' . $v2_term;
}
}
}
$options[ $this->prefix . 'excl_terms' ] = $excl_terms;
return $options;
}
/**
* Filter Options
*
* Handle 'all' option for post types / taxonomies, further sanitization of filename, rewrites on for multisite, setting up addl pages option.
*
* @todo This needs nonce support.
*
* @since ?
* @since 3.0 Change saving 'excl_terms' to database with tax_query format for custom taxonomy support. (Pro #240)
*
* @param $options
* @return mixed
*/
public function filter_options( $options ) {
if ( ! isset( $this->default_options['posttypes']['initial_options'] ) ) {
$this->add_post_types();
}
// TODO Add `true` in 3rd argument with in_array(); which changes it to a strict comparison.
if ( is_array( $options[ "{$this->prefix}posttypes" ] ) && in_array( 'all', $options[ "{$this->prefix}posttypes" ] ) && is_array( $this->default_options['posttypes']['initial_options'] ) ) {
$options[ "{$this->prefix}posttypes" ] = array_keys( $this->default_options['posttypes']['initial_options'] );
}
// TODO Add `true` in 3rd argument with in_array(); which changes it to a strict comparison.
if ( is_array( $options[ "{$this->prefix}taxonomies" ] ) && in_array( 'all', $options[ "{$this->prefix}taxonomies" ] ) && is_array( $this->default_options['taxonomies']['initial_options'] ) ) {
$options[ "{$this->prefix}taxonomies" ] = array_keys( $this->default_options['taxonomies']['initial_options'] );
}
if ( is_multisite() ) {
$options[ $this->prefix . 'rewrite' ] = 'On';
}
if ( ! is_array( $options[ $this->prefix . 'addl_pages' ] ) ) {
$options[ $this->prefix . 'addl_pages' ] = wp_specialchars_decode( stripslashes_deep( $options[ $this->prefix . 'addl_pages' ] ), ENT_QUOTES );
$decoded = json_decode( $options[ $this->prefix . 'addl_pages' ] );
if ( null === $decoded ) {
$decoded = maybe_unserialize( $options[ $this->prefix . 'addl_pages' ] );
}
if ( ! is_array( $decoded ) ) {
$decoded = (array) $decoded;
}
if ( null === $decoded ) {
$decoded = $options[ $this->prefix . 'addl_pages' ];
}
$options[ $this->prefix . 'addl_pages' ] = $decoded;
}
if ( is_array( $options[ $this->prefix . 'addl_pages' ] ) ) {
foreach ( $options[ $this->prefix . 'addl_pages' ] as $k => $v ) {
if ( is_object( $v ) ) {
$options[ $this->prefix . 'addl_pages' ][ $k ] = (array) $v;
}
}
}
if ( isset( $options[ $this->prefix . 'addl_pages' ][0] ) ) {
unset( $options[ $this->prefix . 'addl_pages' ][0] );
}
// TODO Refactor all these... use a nonce, dump the incoming _Post into an array and use that.
if ( ! empty( $_POST[ $this->prefix . 'addl_url' ] ) ) {
foreach ( array( 'addl_url', 'addl_prio', 'addl_freq', 'addl_mod' ) as $field ) {
if ( ! empty( $_POST[ $this->prefix . $field ] ) ) {
// TODO Add/Change to filter_input().
$_POST[ $this->prefix . $field ] = esc_attr( wp_kses_post( $_POST[ $this->prefix . $field ] ) );
} else {
// TODO Add/Change to filter_input().
$_POST[ $this->prefix . $field ] = '';
}
}
if ( ! is_array( $options[ $this->prefix . 'addl_pages' ] ) ) {
$options[ $this->prefix . 'addl_pages' ] = array();
}
if ( aiosp_common::is_url_valid( $_POST[ $this->prefix . 'addl_url' ] ) ) {
$options[ $this->prefix . 'addl_pages' ][ $_POST[ $this->prefix . 'addl_url' ] ] = array(
// TODO Add/Change to filter_input().
'prio' => $_POST[ $this->prefix . 'addl_prio' ],
'freq' => $_POST[ $this->prefix . 'addl_freq' ],
'mod' => $_POST[ $this->prefix . 'addl_mod' ],
);
}
}
if ( ! empty( $_POST[ $this->prefix . 'excl_terms' ] ) ) {
$raw_excl_terms = filter_input( INPUT_POST, $this->prefix . 'excl_terms', FILTER_DEFAULT, FILTER_REQUIRE_ARRAY );
// Parse taxonomy terms {$taxonomy_slug}-{$term_id}.
$excl_terms = array();
foreach ( $raw_excl_terms as $v1_tax_term ) {
$term_id = explode( '-', $v1_tax_term );
$term_id = intval( end( $term_id ) );
$taxonomy_slug = sanitize_text_field( str_replace( '-' . $term_id, '', $v1_tax_term ) );
// Initialize taxonomy => terms array if not yet set.
if ( ! isset( $excl_terms[ $taxonomy_slug ] ) ) {
$excl_terms[ $taxonomy_slug ] = array(
'terms' => array(),
);
}
$excl_terms[ $taxonomy_slug ]['taxonomy'] = $taxonomy_slug;
$excl_terms[ $taxonomy_slug ]['terms'][] = $term_id;
}
$options[ $this->prefix . 'excl_terms' ] = $excl_terms;
}
return $options;
}
/**
* Gets Home Path
*
* If we're in wp-admin, use the WordPress function, otherwise we user our own version here.
* This only applies to static sitemaps.
*
* @since 2.3.6.1
*
* @return mixed|string
*/
public function get_home_path() {
if ( function_exists( 'get_home_path' ) ) {
return get_home_path();
}
$home = set_url_scheme( get_option( 'home' ), 'http' );
$siteurl = set_url_scheme( get_option( 'siteurl' ), 'http' );
if ( ! empty( $home ) && 0 !== strcasecmp( $home, $siteurl ) ) {
$wp_path_rel_to_home = str_ireplace( $home, '', $siteurl ); /* $siteurl - $home */
$pos = strripos( str_replace( '\\', '/', $_SERVER['SCRIPT_FILENAME'] ), trailingslashit( $wp_path_rel_to_home ) );
$home_path = substr( $_SERVER['SCRIPT_FILENAME'], 0, $pos );
$home_path = trailingslashit( $home_path );
} else {
$home_path = ABSPATH;
}
return str_replace( '\\', '/', $home_path );
}
/**
* Whitelist Static Sitemaps
*
* Whitelists files from static sitemap conflict warning.
* For right now, this is just externally produced news sitemaps until we figure out something better.
*
* @since 2.3.10.2
*
* @param $file
* @return string
*/
public function whitelist_static_sitemaps( $file ) {
$whitelist = array( 'sitemap_news.xml', 'sitemap-news.xml' );
if ( in_array( $file, $whitelist, true ) ) {
return '';
}
return $file;
}
/**
* Scan Match Files
*
* Scan for sitemaps on filesystem.
*
* @since ?
*
* @return array
*/
public function scan_match_files() {
$scan1 = '';
$files = array();
$filename = $this->get_filename();
if ( ! empty( $filename ) ) {
$scan1 = get_home_path() . $filename . '*.xml';
if ( empty( $scan1 ) ) {
return $files;
}
$home_path = get_home_path();
$filescan = $this->scandir( $home_path );
if ( ! empty( $filescan ) ) {
foreach ( $filescan as $f ) {
if ( ! empty( $scan1 ) && fnmatch( $scan1, $home_path . $f ) ) {
$f = $this->whitelist_static_sitemaps( $f );
$files[] = $home_path . $f;
continue;
}
}
}
return $files;
}
}
/**
* Do Sitemap Scan
*
* Handle deleting / renaming of conflicting sitemap files.
*
* @todo Add/Fix nonce.
*
* @since ?
*/
public function do_sitemap_scan() {
$msg = '';
if ( ! empty( $this->options[ "{$this->prefix}rewrite" ] ) && ( get_option( 'permalink_structure' ) === '' ) ) {
$msg = '
' . __( 'Warning: dynamic sitemap generation must have permalinks enabled.', 'all-in-one-seo-pack' ) . '
';
}
if ( ! empty( $_POST['aioseop_sitemap_rename_files'] ) || ! empty( $_POST['aioseop_sitemap_delete_files'] ) ) {
// TODO Add/Change to filter_input().
$nonce = $_POST['nonce-aioseop'];
if ( ! wp_verify_nonce( $nonce, 'aioseop-nonce' ) ) {
// TODO Change to wp_die().
die( __( 'Security Check - If you receive this in error, log out and back in to WordPress', 'all-in-one-seo-pack' ) );
}
if ( ! empty( $_POST['aioseop_sitemap_conflict'] ) ) {
$files = $this->scan_match_files();
foreach ( $files as $f => $file ) {
$files[ $f ] = realpath( $file );
}
foreach ( $_POST['aioseop_sitemap_conflict'] as $ren_file ) {
$ren_file = realpath( get_home_path() . $ren_file );
// TODO Add `true` in 3rd argument with in_array(); which changes it to a strict comparison.
if ( in_array( $ren_file, $files ) ) {
if ( ! empty( $_POST['aioseop_sitemap_delete_files'] ) ) {
if ( $this->delete_file( $ren_file ) ) {
/* translators: Shows which sitemap files have been deleted. */
$msg .= '' . sprintf( __( 'Deleted %s.', 'all-in-one-seo-pack' ), $ren_file ) . '
';
}
continue;
}
$count = 0;
do {
$ren_to = $ren_file . '._' . sprintf( '%03d', $count ) . '.old';
$count ++;
} while ( $this->file_exists( $ren_to ) && ( $count < 1000 ) );
if ( $count >= 1000 ) {
/* translators: Shows which sitemap files couldn't be renamed. */
$msg .= '' . sprintf( __( "Couldn't rename file %s!", 'all-in-one-seo-pack' ), $ren_file ) . '
';
} else {
$ren = $this->rename_file( $ren_file, $ren_to );
if ( $ren ) {
/* translators: Shows which sitemap files were renamed. */
$msg .= '' . sprintf( __( 'Renamed %1$s to %2$s.', 'all-in-one-seo-pack' ), $ren_file, $ren_to ) . '
';
}
}
} else {
/* translators: Shows which sitemap files couldn't be found. */
$msg .= '' . sprintf( __( "Couldn't find file %s!", 'all-in-one-seo-pack' ), $ren_file ) . '
';
}
}
}
} else {
$msg .= $this->scan_sitemaps();
}
if ( ! empty( $msg ) ) {
$this->output_error( $msg );
}
}
/**
* Scan Sitemaps
*
* Do the scan, return the results.
*
* @since ?
*
* @return string
*/
public function scan_sitemaps() {
$msg = '';
$files = $this->scan_match_files();
if ( ! empty( $files ) ) {
$msg = $this->sitemap_warning( $files );
}
return $msg;
}
/**
* Get Problem Files
*
* Get the list of potentially conflicting sitemap files, identify whether they came from us, are blank, or are of unknown origin.
*
* @since ?
* @since 2.3.10 Add the ability to see empty sitemap files as well.
*
* @param $files
* @param $msg
* @return array
*/
public function get_problem_files( $files, &$msg ) {
$problem_files = array();
$use_wpfs = true;
$wpfs = $this->get_filesystem_object();
if ( ! is_object( $wpfs ) ) {
$use_wpfs = false;
} else {
if ( 'direct' === $wpfs->method ) {
$use_wpfs = false;
}
}
foreach ( $files as $f ) {
if ( $this->is_file( $f ) ) {
$fn = $f;
$compressed = false;
if ( AIOSEOP_PHP_Functions::substr( $f, - 3 ) === '.gz' ) {
$compressed = true;
}
if ( $use_wpfs ) {
if ( $compressed ) { // Inefficient but necessary.
$file = $this->load_file( $fn );
if ( ! empty( $file ) ) {
$file = gzuncompress( $file, 4096 );
}
} else {
$file = $this->load_file( $fn, false, null, - 1, 4096 );
}
} else {
if ( $compressed ) {
$file_resource = gzopen( $fn, 'rb' );
$file = gzread( $file_resource, 4096 );
gzclose( $file_resource );
} else {
// TODO Change to `wp_remote_get()`.
$file = file_get_contents( $fn, false, null, 0, 4096 );
}
}
if ( ! empty( $file ) ) {
$matches = array();
if ( preg_match(
'//',
$file,
$matches
) ) {
if ( ! empty( $this->options[ "{$this->prefix}rewrite" ] ) ) {
/* translators: %1$s, %2$s, etc. are placeholders and should not be translated. %1$s expands to the name of a sitemap file, %2$s to the name of the plugin, All in One SEO Pack, %3$s is replaced with the plugin version number and %4$s with a date. */
$msg .= '' . sprintf(
__( "Warning: a static sitemap '%1\$s' generated by %2\$s %3\$s on %4\$s already exists that may conflict with dynamic sitemap generation.", 'all-in-one-seo-pack' ),
$f,
AIOSEOP_PLUGIN_NAME,
$matches[2],
$matches[3]
) . "
\n";
$problem_files[] = $f;
}
} else {
/* translators: Shows which 'unknown' file is conflicting with the current sitemap settings. */
$msg .= '' . sprintf( __( 'Potential conflict with unknown file %s.', 'all-in-one-seo-pack' ), $f ) . "
\n";
$problem_files[] = $f;
}
} else {
/* translators: Shows which files were removed. */
$msg .= '' . sprintf( __( 'Removed empty file %s.', 'all-in-one-seo-pack' ), $f ) . "
\n";
$problem_files[] = $f;
// This is causing all problem_files to be deleted automatically; which may be the intent.
// TODO Either create a separate variable for this set of problem_files, or a final loop to clean problem_files before returning.
foreach ( $problem_files as $f => $file ) {
$files[ $f ] = realpath( $file );
$this->delete_file( realpath( $file ) );
}
$problem_files = false; // Don't return anything. If it's blank, we'll take care of it here.
}
}
}
return $problem_files;
}
/**
* Sitemap Warning
*
* Display the warning and the form for conflicting sitemap files.
*
* @since ?
*
* @param $files
* @return string
*/
public function sitemap_warning( $files ) {
$msg = '';
$conflict = false;
$problem_files = $this->get_problem_files( $files, $msg );
if ( ! empty( $problem_files ) ) {
$conflict = true;
}
if ( $conflict ) {
foreach ( $problem_files as $p ) {
$msg .= "';
if ( is_multisite() ) {
// TODO Add esc_* function.
echo
'' . aiosp_common::esc_xml( 'title', get_blog_option( get_current_blog_id(), 'blogname' ) ) . ' ' .
'' . aiosp_common::esc_xml( 'description', get_blog_option( get_current_blog_id(), 'blogdescription' ) ) . ' ';
} else {
// TODO Add esc_* function.
echo
'' . aiosp_common::esc_xml( 'title', get_option( 'blogname' ) ) . ' ' .
'' . aiosp_common::esc_xml( 'description', get_option( 'blogdescription' ) ) . ' ';
}
// remove urls that do not have the rss element.
$urls = array_filter( $urls, array( $this, 'include_in_rss' ) );
if ( false !== strpos( $sitemap_type, 'latest' ) ) {
// let's sort the array in descending order of date.
uasort( $urls, array( $this, 'sort_modifed_date_descending' ) );
$urls = array_slice( $urls, 0, apply_filters( $this->prefix . 'rss_latest_limit', 20 ) );
}
foreach ( $urls as $url ) {
// TODO Add esc_* function.
echo
'- ' .
'
' . aiosp_common::esc_xml( 'guid', $url['loc'] ) . ' ' .
'' . aiosp_common::esc_xml( 'title', $url['rss']['title'] ) . ' ' .
'' . aiosp_common::esc_xml( 'pubDate', $url['rss']['pubDate'] ) . ' ' .
' ';
}
echo ' prefix . 'xml_namespace',
array(
'xmlns' => 'http://www.sitemaps.org/schemas/sitemap/0.9',
'xmlns:image' => 'http://www.google.com/schemas/sitemap-image/1.1',
)
);
if ( ! empty( $namespaces ) ) {
$ns = array();
foreach ( $namespaces as $k => $v ) {
$ns[] = esc_attr( $k ) . '="' . esc_url( $v, array( 'http', 'https' ) ) . '"';
}
$xml_header .= join( "\r\n\t", $ns );
}
$xml_header .= '>' . "\r\n";
// TODO Add esc_* function.
echo $xml_header;
$count = 0;
foreach ( $urls as $url ) {
echo "\t\r\n";
if ( is_array( $url ) ) {
if ( isset( $url['rss'] ) ) {
unset( $url['rss'] );
}
foreach ( $url as $k => $v ) {
if ( ! empty( $v ) ) {
$v = aiosp_common::esc_xml( $k, $v );
if ( is_array( $v ) ) {
$buf = "\t\t\t<$k>\r\n";
foreach ( $v as $ext => $attr ) {
if ( is_array( $attr ) ) {
$buf = '';
// TODO Add esc_* function.
echo "\t\t<$k>\r\n";
foreach ( $attr as $a => $nested ) {
if ( is_array( $nested ) ) {
// TODO Add esc_* function.
echo "\t\t\t<$a>\r\n";
foreach ( $nested as $next => $nattr ) {
$value = aiosp_common::esc_xml( $next, $nattr );
// TODO Add esc_* function.
echo "\t\t\t\t<$next>$value\r\n";
}
// TODO Add esc_* function.
echo "\t\t\t\r\n";
} else {
$value = aiosp_common::esc_xml( $a, $nested );
// TODO Add esc_* function.
echo "\t\t\t<$a>$value\r\n";
}
}
// TODO Add esc_* function.
echo "\t\t\r\n";
} else {
$value = aiosp_common::esc_xml( $ext, $attr );
$buf .= "\t\t\t<$ext>$value\r\n";
}
}
if ( ! empty( $buf ) ) {
// TODO Add esc_* function.
echo $buf . "\t\t\r\n";
}
} else {
$value = aiosp_common::esc_xml( $k, $v );
// TODO Add esc_* function.
echo "\t\t<$k>$value\r\n";
}
}
}
} else {
$value = aiosp_common::esc_xml( 'loc', $url );
// TODO Add esc_* function.
echo "\t\t$value \r\n";
}
echo "\t \r\n";
if ( $count >= $max_items ) {
break;
}
}
echo ' ';
}
/**
* Output Sitemap Index
*
* Output the XML for a sitemap index.
*
* @since ?
*
* @param $urls
* @param string $comment
* @return null
*/
public function output_sitemap_index( $urls, $comment = '' ) {
$max_items = 50000;
if ( ! is_array( $urls ) ) {
return null;
}
echo '' . "\r\n\r\n";
// TODO Add esc_* function.
echo '\r\n";
$xsl_url = $this->get_sitemap_xsl();
// TODO Add esc_* function.
echo '' . "\r\n";
echo '' . "\r\n";
$count = 0;
foreach ( $urls as $url ) {
echo "\t\r\n";
if ( is_array( $url ) ) {
foreach ( $url as $k => $v ) {
// TODO Add `true` in 3rd argument with in_array(); which changes it to a strict comparison.
if ( ! in_array( $k, array( 'loc', 'lastmod' ) ) ) {
continue;
}
$v = aiosp_common::esc_xml( $k, $v );
// TODO Add esc_* function.
echo "\t\t<$k>$v\r\n";
}
} else {
$value = aiosp_common::esc_xml( 'loc', $url );
// TODO Add esc_* function.
echo "\t\t$value \r\n";
}
echo "\t \r\n";
$count ++;
if ( $count >= $max_items ) {
break;
}
}
echo ' ';
}
/**
* Build Sitemap Index
*
* Return an XML sitemap index as a string.
*
* @since ?
*
* @param $urls
* @param string $comment
* @return string
*/
public function build_sitemap_index( $urls, $comment = '' ) {
ob_start();
$this->output_sitemap_index( $urls, $comment );
return ob_get_clean();
}
/**
* Build Sitemap
*
* Return an XML sitemap as a string.
*
* @since ?
*
* @param $urls
* @param string $sitemap_type The type of sitemap viz. root, rss, rss_latest etc.. For static sitemaps, this would be empty.
* @param string $comment
* @return string
*/
public function build_sitemap( $urls, $sitemap_type, $comment = '' ) {
ob_start();
$this->output_sitemap( $urls, $sitemap_type, $comment );
return ob_get_clean();
}
/**
* Gets the last modified timestamp, priority and frequency for taxonomy terms.
*
* @since 3.4.0 Check if term has sitemap priority/frequency in termmeta. Renamed function to better reflect purpose.
*
* @param array $terms The taxonomy terms that need their sitemap meta info to be populated.
* @return array $populated_terms The taxonomy terms with their sitemap meta info.
*/
public function get_terms_data( $terms ) {
$populated_terms = array();
if ( is_array( $terms ) && ! empty( $terms ) ) {
foreach ( $terms as $term ) {
$pr_info = array();
$pr_info['loc'] = $this->get_term_link( $term, $term->taxonomy );
$pr_info['lastmod'] = $this->get_tax_term_timestamp( $term );
$pr_info['priority'] = $this->get_default_priority( 'taxonomies' );
$pr_info['changefreq'] = $this->get_default_frequency( 'taxonomies' );
if ( isset( $this->options[ $this->prefix . 'prio_taxonomies' ] ) && 'no' !== $this->options[ $this->prefix . 'prio_taxonomies' ] ) {
if ( 'sel' !== $this->options[ $this->prefix . 'prio_taxonomies' ] ) {
$pr_info['priority'] = $this->options[ $this->prefix . 'prio_taxonomies' ];
} elseif ( 'no' !== $this->options[ $this->prefix . 'prio_taxonomies_' . $term->taxonomy ] ) {
$pr_info['priority'] = $this->options[ $this->prefix . 'prio_taxonomies_' . $term->taxonomy ];
}
}
if ( isset( $this->options[ $this->prefix . 'freq_taxonomies' ] ) && 'no' !== $this->options[ $this->prefix . 'freq_taxonomies' ] ) {
if ( 'sel' !== $this->options[ $this->prefix . 'freq_taxonomies' ] ) {
$pr_info['changefreq'] = $this->options[ $this->prefix . 'freq_taxonomies' ];
} elseif ( 'no' !== $this->options[ $this->prefix . 'freq_taxonomies_' . $term->taxonomy ] ) {
$pr_info['changefreq'] = $this->options[ $this->prefix . 'freq_taxonomies_' . $term->taxonomy ];
}
}
$pr_info['image:image'] = $this->get_images_from_term( $term );
$pr_info['rss'] = array(
'title' => $term->name,
'description' => $term->description,
'pubDate' => $this->get_date_for_term( $term ),
);
$populated_terms[] = $pr_info;
}
}
return $populated_terms;
}
/**
* The get_tax_term_timestamp() function.
*
* Gets the Last Change timestamp for a taxonomy term.
*
* @since 3.2.0
* @since 3.4.0 Changed access modifier.
*
* @param object $term
* @return string $lastmod
*/
protected function get_tax_term_timestamp( $term ) {
$taxonomy_object = get_taxonomy( $term->taxonomy );
$lastmod = '';
// Loop through all attached post types and get timestamp of last modified assigned post.
foreach ( $taxonomy_object->object_type as $object_type ) {
$latest_modified_post = new WP_Query(
array(
'post_type' => $object_type,
'post_status' => 'publish',
'posts_per_page' => 1,
'orderby' => 'modified',
'order' => 'DESC',
'taxonomy' => $term->taxonomy,
'term' => $term->slug,
)
);
if ( $latest_modified_post->have_posts() ) {
$temp_lastmod = $latest_modified_post->posts[0]->post_modified_gmt;
if ( '' === $lastmod || ( $temp_lastmod > $lastmod ) ) {
$lastmod = $temp_lastmod;
}
}
}
$lastmod = date( 'Y-m-d\TH:i:s\Z', mysql2date( 'U', $lastmod ) );
return $lastmod;
}
/**
* Get Date for Term
*
* Return the date of the latest post in the given taxonomy term.
*
* @since 2.9.0
* @since 3.4.0 Changed access modifier.
*
* @param WP_Term $term The taxonomy term.
* @return string
*/
protected function get_date_for_term( $term ) {
$date = '';
$query = new WP_Query(
array(
'orderby' => 'post_date',
'order' => 'DESC',
'numberposts' => 1,
'post_type' => 'any',
'post_status' => 'publish',
'tax_query' => array(
array(
'taxonomy' => $term->taxonomy,
'terms' => $term->term_id,
),
),
)
);
if ( $query->have_posts() ) {
$timestamp = mysql2date( 'U', $query->post->post_modified_gmt );
$date = date( 'r', $timestamp );
}
return $date;
}
/**
* Get Term Permalinks
*
* Return a list of permalinks for an array of terms.
*
* @since ?
*
* @param $terms
* @return array
*/
public function get_term_permalinks( $terms ) {
$links = array();
if ( is_array( $terms ) ) {
foreach ( $terms as $term ) {
$url = $this->get_term_link( $term );
$links[] = $url;
}
}
return $links;
}
/**
* Get Archive Permalinks
*
* Return permalinks for archives.
*
* @since ?
*
* @param $posts
* @return array
*/
public function get_archive_permalinks( $posts ) {
$links = array();
$archives = array();
if ( is_array( $posts ) ) {
foreach ( $posts as $post ) {
$date = mysql2date( 'U', $post->post_date );
$year = date( 'Y', $date );
$month = date( 'm', $date );
$archives[ $year . '-' . $month ] = array( $year, $month );
}
}
$archives = array_keys( $archives );
foreach ( $archives as $d ) {
$links[] = get_month_link( $d[0], $d[1] );
}
return $links;
}
/**
* Get Author Permalink
*
* Return permalinks for authors.
*
* @since ?
*
* @param $posts
* @return array
*/
public function get_author_permalinks( $posts ) {
$links = array();
$authors = array();
if ( is_array( $posts ) ) {
foreach ( $posts as $post ) {
$authors[ $post->author_id ] = 1;
}
}
$authors = array_keys( $authors );
foreach ( $authors as $auth_id ) {
$links[] = get_author_posts_url( $auth_id );
}
return $links;
}
/**
* Get Post Permalink
*
* Return permalinks for posts.
*
* @since ?
*
* @param $posts
* @return array
*/
public function get_post_permalinks( $posts ) {
$links = array();
if ( is_array( $posts ) ) {
foreach ( $posts as $post ) {
$post->filter = 'sample';
$url = $this->get_permalink( $post );
$links[] = $url;
}
}
return $links;
}
/**
* Unparse URL
*
* Convert back from parse_url.
* Props to thomas at gielfeldt dot com.
*
* @since ?
*
* @link http://www.php.net/manual/en/function.parse-url.php#106731
*
* @param $parsed_url
* @return string
*/
public function unparse_url( $parsed_url ) {
$scheme = isset( $parsed_url['scheme'] ) ? $parsed_url['scheme'] . '://' : '';
$host = isset( $parsed_url['host'] ) ? $parsed_url['host'] : '';
if ( ! empty( $host ) && empty( $scheme ) ) {
$scheme = '//';
}
$port = isset( $parsed_url['port'] ) ? ':' . $parsed_url['port'] : '';
$user = isset( $parsed_url['user'] ) ? $parsed_url['user'] : '';
$pass = isset( $parsed_url['pass'] ) ? ':' . $parsed_url['pass'] : '';
$pass = ( $user || $pass ) ? "$pass@" : '';
$path = isset( $parsed_url['path'] ) ? $parsed_url['path'] : '';
$query = isset( $parsed_url['query'] ) ? '?' . $parsed_url['query'] : '';
$fragment = isset( $parsed_url['fragment'] ) ? '#' . $parsed_url['fragment'] : '';
return "$scheme$user$pass$host$port$path$query$fragment";
}
/**
* Get Additional Page Only
*
* Return data for user entered additional pages.
*
* @since 2.3.6
* @since 2.3.12.3 Refactored to use aioseop_home_url() for compatibility purposes.
*
* @return array
*/
public function get_addl_pages_only() {
$pages = array();
if ( ! empty( $this->options[ $this->prefix . 'addl_pages' ] ) ) {
$siteurl = wp_parse_url( aioseop_home_url() );
foreach ( $this->options[ $this->prefix . 'addl_pages' ] as $k => $v ) {
$k = aiosp_common::make_url_valid_smartly( $k );
$url = wp_parse_url( $k );
if ( empty( $url['host'] ) ) {
$url['host'] = $siteurl['host'];
}
if ( ! empty( $url['path'] ) && substr( $url['path'], 0, 1 ) !== '/' ) {
$url['path'] = '/' . $url['path'];
}
$freq = '';
$prio = '';
$mod = '';
if ( ! empty( $v['mod'] ) ) {
$mod = $v['mod'];
}
if ( ! empty( $v['freq'] ) ) {
$freq = $v['freq'];
}
if ( ! empty( $v['prio'] ) ) {
$prio = $v['prio'];
}
if ( 'no' === $freq ) {
$freq = '';
}
if ( 'no' === $prio ) {
$prio = '';
}
$mod = date( 'Y-m-d\TH:i:s\Z', mysql2date( 'U', $mod ) );
$pages[] = array(
'loc' => $this->unparse_url( $url ),
'lastmod' => $mod,
'changefreq' => $freq,
'priority' => $prio,
);
}
}
$pages = apply_filters( $this->prefix . 'addl_pages_only', $pages );
return $pages;
}
/**
* Get Additional Pages
*
* Return data for user entered additional pages and extra pages.
*
* @since 2.3.6
* @since 2.3.12.3 Refactored to use aioseop_home_url() for compatibility purposes.
* @since 3.2.0 Do not include static homepage/posts page - #2126.
*
* @return array
*/
public function get_addl_pages() {
$home = array();
$home = array(
'loc' => aioseop_home_url(),
'changefreq' => $this->get_default_frequency( 'homepage' ),
'priority' => $this->get_default_priority( 'homepage' ),
'image:image' => $this->get_images_from_post( (int) get_option( 'page_on_front' ) ),
);
if ( isset( $this->options[ $this->prefix . 'prio_homepage' ] ) && 'no' !== $this->options[ $this->prefix . 'prio_homepage' ] ) {
if ( 'sel' !== $this->options[ $this->prefix . 'prio_homepage' ] ) {
$home['priority'] = $this->options[ $this->prefix . 'prio_homepage' ];
}
}
if ( isset( $this->options[ $this->prefix . 'freq_homepage' ] ) && 'no' !== $this->options[ $this->prefix . 'freq_homepage' ] ) {
if ( 'sel' !== $this->options[ $this->prefix . 'freq_homepage' ] ) {
$home['changefreq'] = $this->options[ $this->prefix . 'freq_homepage' ];
}
}
$posts = (int) get_option( 'page_for_posts' );
if ( $posts ) {
$posts = $this->get_permalink( $posts );
if ( $posts === $home['loc'] ) {
$posts = array();
} else {
$posts = array(
'loc' => $posts,
'changefreq' => $this->get_default_frequency( 'blog' ),
'priority' => $this->get_default_priority( 'blog' ),
);
}
} else {
$posts = array();
}
$pages = $this->get_addl_pages_only();
if ( ! empty( $home ) ) {
$pages[] = $home;
}
if ( ! empty( $posts ) ) {
$pages[] = $posts;
}
$pages = apply_filters( $this->prefix . 'addl_pages', $pages );
$pages = $this->get_homepage_timestamp( $pages );
$pages = $this->remove_addl_static_pages( $pages );
return $pages;
}
/**
* The remove_addl_static_pages() function.
*
* Removes the homepage/posts page from the Additional Pages index if it is static - #2126.
*
* @since 3.2.0
*
* @param array $pages
* @return array $pages
*/
private function remove_addl_static_pages( $pages ) {
$pages_to_remove = array();
if ( 0 !== (int) get_option( 'page_on_front' ) ) {
$homepage_url = get_site_url() . '/';
array_push( $pages_to_remove, $homepage_url );
}
$static_posts_page_id = (int) get_option( 'page_for_posts' );
if ( 0 !== $static_posts_page_id ) {
array_push( $pages_to_remove, get_permalink( $static_posts_page_id ) );
}
if ( count( $pages_to_remove ) > 0 ) {
return $this->remove_urls_from_sitemap_page( $pages, $pages_to_remove );
}
return $pages;
}
/**
* The remove_urls_from_sitemap_page() function.
*
* Removes URLs from a sitemap page. This is used both for indexes and pages within indexes.
*
* @since 3.2.0
*
* @param array $pages
* @param array $pages_to_remove
* @return array $pages
*/
private function remove_urls_from_sitemap_page( $pages, $pages_to_remove ) {
$count = count( $pages );
for ( $i = 0; $i < $count; $i++ ) {
if ( in_array( $pages[ $i ]['loc'], $pages_to_remove, true ) ) {
unset( $pages[ $i ] );
}
}
return $pages;
}
/**
* The get_homepage_timestamp() function.
*
* Gets the Last Change timestamp for the homepage if it isn't static.
*
* @since 3.2.0
*
* @param array $urls
* @return array $urls
*/
private function get_homepage_timestamp( $urls ) {
if ( 0 !== (int) get_option( 'page_on_front' ) ) {
return $urls;
}
$homepage_url = get_site_url() . '/';
$urls = $this->update_static_page_timestamp( $urls, $homepage_url );
return $urls;
}
/**
* The get_posts_page_timestamp() function.
*
* Gets the Last Change timestamp for the posts page.
*
* @since 3.2.0
*
* @param array $urls
* @return array $urls
*/
private function get_posts_page_timestamp( $urls ) {
$posts_page_id = (int) get_option( 'page_for_posts' );
if ( 0 === $posts_page_id ) {
return $urls;
}
$posts_page_url = get_permalink( $posts_page_id );
$urls = $this->update_static_page_timestamp( $urls, $posts_page_url );
return $urls;
}
/**
* The update_static_page_timestamp() function.
*
* Update the timestamp attribute for a static page.
*
* @since 3.2.0
*
* @param array $urls
* @param string $static_page_url
* @return array $urls
*/
private function update_static_page_timestamp( $urls, $static_page_url ) {
$lastmod = $this->get_last_modified_post_timestamp( 'post' );
if ( false === $lastmod ) {
return $urls;
}
$url_locs = array_combine( array_keys( $urls ), wp_list_pluck( $urls, 'loc' ) );
$index = array_search( $static_page_url, $url_locs );
if ( false === $index ) {
return $urls;
}
$urls[ $index ] = $this->insert_timestamp_as_second_attribute( $urls[ $index ], $lastmod );
return $urls;
}
/**
* The get_last_modified_post_timestamp() function.
*
* Gets the last modified post.
*
* @since 3.2.0
*
* @param string $post_type
* @return mixed Timestamp of the last modified post or false if there is none.
*/
private function get_last_modified_post_timestamp( $post_type ) {
$last_modified_post = new WP_Query(
array(
'post_type' => $post_type,
'post_status' => 'publish',
'posts_per_page' => 1,
'orderby' => 'modified',
'order' => 'DESC',
)
);
if ( $last_modified_post->have_posts() ) {
return $this->format_timestamp_as_lastmod_attribute( $last_modified_post );
}
return false;
}
/**
* The format_timestamp_as_lastmod_attribute() function.
*
* Formats the timestamp for a sitemap record in order to have valid sitemap schema.
*
* @since 3.2.0
*
* @param object $last_modified_post WP_Query for the last modified post.
* @return string $lastmod
*/
private function format_timestamp_as_lastmod_attribute( $last_modified_post ) {
$lastmod = $last_modified_post->posts[0]->post_modified_gmt;
return date( 'Y-m-d\TH:i:s\Z', mysql2date( 'U', $lastmod ) );
}
/**
* The insert_timestamp_as_second_attribute() function.
*
* Inserts the timestamp for a sitemap record as the second attribute.
* The lastmod subtag has to be inserted as second attribute in order to have valid schema.
*
* @since 3.2.0
*
* @param array $url
* @param string $lastmod
* @return array $url
*/
private function insert_timestamp_as_second_attribute( $url, $lastmod ) {
return array_slice( $url, 0, 1, true ) + array( 'lastmod' => $lastmod ) + array_slice( $url, 1, null, true );
}
/**
* Get Priority Calculation
*
* Scores posts based on date and relative comment count, if any.
*
* @since ?
*
* @param $date
* @param mixed $stats
* @return array
*/
public function get_prio_calc( $date, $stats ) {
static $cur_time = null;
if ( null === $cur_time ) {
$cur_time = time();
}
$time = $cur_time - mysql2date( 'U', $date );
if ( ! empty( $stats ) && isset( $stats['max'] ) && $stats['max'] ) {
$minadj = $time >> 3;
$maxadj = $time >> 1;
$avg = $stats['count'] / $stats['total'];
$calc = ( $stats['comment_count'] - $stats['min'] ) / $stats['max'];
$calc = $maxadj * $calc;
if ( $avg < $stats['comment_count'] ) {
$minadj = $time >> 2;
} else {
$maxadj = $time >> 2;
}
if ( $calc > $maxadj ) {
$calc = $maxadj;
}
if ( $calc < $minadj ) {
$calc = $minadj;
}
$time -= $calc;
}
$days = $time / ( 60 * 60 * 24 );
$prio_table = array(
'daily' => 7,
'weekly' => 30,
'monthly' => 210,
'yearly' => null,
);
$interval = 1.0;
$prev_days = 0;
foreach ( $prio_table as $change => $max_days ) {
$interval -= 0.3;
if ( null === $max_days ) {
$changefreq = $change;
$prio = 0.1;
break;
}
if ( $days < $max_days ) {
$int_days_max = $max_days - $prev_days;
$int_days = $days - $prev_days;
$prio = $interval + ( (int) ( 3 * ( ( $max_days - $int_days ) / $int_days_max ) ) / 10.0 );
$changefreq = $change;
break;
}
$prev_days = $max_days;
}
return array(
'lastmod' => $date,
'changefreq' => $changefreq,
'priority' => $prio,
);
}
/**
* Returns all date archive entries for the sitemap.
*
* @since 3.4.0 Fixed bug where archives couldn't fetch default priority/frequency + minor refactoring. Renamed function to better reflect purpose.
*
* @return array $date_archives All date archive sitemap entries.
*/
public function get_date_archive_data() {
global $wpdb;
$args = array(
'numberposts' => 50000,
'post_type' => 'post',
);
// Get monthly results.
$sql_query = $wpdb->prepare( "
SELECT
YEAR(post_date) AS `year`,
MONTH(post_date) AS `month`
FROM {$wpdb->posts}
WHERE post_type = %s AND post_status = 'publish'
GROUP BY
YEAR(post_date),
MONTH(post_date)
ORDER BY post_date ASC LIMIT %d",
$args['post_type'],
$args['numberposts']
);
$key = hash( 'sha256', $sql_query );
$last_changed = wp_cache_get_last_changed( 'posts' );
$key = "aioseop_get_date_archive_data:$key:$last_changed";
$date_results = wp_cache_get( $key, 'posts' );
if ( ! $date_results ) {
$date_results = $wpdb->get_results( $sql_query );
wp_cache_set( $key, $date_results, 'posts' );
}
$priority = $this->get_default_priority( 'archive' );
$frequency = $this->get_default_frequency( 'archive' );
if ( isset( $this->options[ $this->prefix . 'prio_archive' ] ) && 'no' !== $this->options[ $this->prefix . 'prio_archive' ] ) {
if ( 'sel' !== $this->options[ $this->prefix . 'prio_archive' ] ) {
$priority = $this->options[ $this->prefix . 'prio_archive' ];
}
}
if ( isset( $this->options[ $this->prefix . 'freq_archive' ] ) && 'no' !== $this->options[ $this->prefix . 'freq_archive' ] ) {
if ( 'sel' !== $this->options[ $this->prefix . 'freq_archive' ] ) {
$frequency = $this->options[ $this->prefix . 'freq_archive' ];
}
}
if ( $date_results ) {
$year = null;
foreach ( $date_results as $date_result ) {
if ( $year !== $date_result->year ) {
$year = $date_result->year;
$date_archives[] = array(
'loc' => get_year_link( $date_result->year ),
'changefreq' => $frequency,
'priority' => $priority,
);
}
$date_archives[] = array(
'loc' => get_month_link( $date_result->year, $date_result->month ),
'changefreq' => $frequency,
'priority' => $priority,
);
}
}
return $date_archives;
}
/**
* Returns a post archive sitemap entry if the post type supports one.
*
* @since 3.2.0 Don't fetch WooCommerce shop page twice - #2126
* @since 3.4.0 Full refactoring. Renamed function to better reflect purpose.
*
* @param string $post_types The post types that have to be checked for archive pages.
*
* @return array $post_archive Post archive sitemap entry with its meta info.
*/
private function get_post_archive_data( $post_types ) {
$post_archives = array();
if ( 'aiosp_video_sitemap_' === $this->prefix ) {
return $post_archives;
}
$post_types_with_archives = get_post_types(
array(
'has_archive' => true,
'_builtin' => false,
),
'names'
);
$default_priority = $this->get_default_priority( 'archive' );
$default_frequency = $this->get_default_frequency( 'archive' );
if ( ! is_array( $post_types ) ) {
$post_types = array( $post_types );
}
foreach ( $post_types as $post_type ) {
// WooCommerce product archive page is the shop page (which we include in the pages-sitemap index).
if ( aioseop_is_woocommerce_active() && 'product' === $post_type ) {
continue;
}
$args = array(
'post_status' => 'publish',
'post_type' => $post_type,
);
// Adding this here in the case we decide to include post archives in the video sitemap.
if ( 'aiosp_video_sitemap_' === $this->prefix ) {
$args['meta_query'] =
array(
array(
'key' => '_aioseop_oembed_info',
'compare' => 'EXISTS',
),
);
}
$query = new WP_Query( $args );
if ( 0 === $query->found_posts || ! in_array( $post_type, $post_types_with_archives, true ) ) {
continue;
}
if ( ! in_array( $post_type, apply_filters( "{$this->prefix}include_post_types_archives", array( $post_type ) ) ) ) {
continue;
}
$post_archive = array(
'loc' => get_post_type_archive_link( $post_type ),
'changefreq' => $default_frequency,
'priority' => $default_priority,
);
$post_archive = $this->get_post_archive_timestamp( $post_archive, $post_type );
array_push( $post_archives, $post_archive );
}
return $post_archives;
}
/**
* Populates a post archive sitemap entry with its last modified timestamp.
*
* Last modified timestamp is based on the last modified timestamp of the first assigned post.
*
* @since 3.2.0
* @since 3.4.0 Expect single post archive instead of array. Renamed function for better function cohesion.
*
* @param array $post_archive The post archive.
* @param string $post_type The post type to which the archive belongs.
* @return array $post_archive The post archive with its last modified timestamp.
*/
private function get_post_archive_timestamp( $post_archive, $post_type ) {
$lastmod = $this->get_last_modified_post_timestamp( $post_type );
if ( false === $lastmod ) {
return $post_archive;
}
$post_archive = $this->insert_timestamp_as_second_attribute( $post_archive, $lastmod );
return $post_archive;
}
/**
* Returns all author archive entries for the sitemap.
*
* @since
* @since 3.4.0 Fixed bug where archives couldn't fetch default priority/frequency + minor refactoring. Renamed function to better reflect purpose.
*
* @return array $author_archives All author archive sitemap entries.
*/
public function get_author_archive_data() {
$entries = array();
if (
! aioseop_last_modified_post() ||
! $this->options[ $this->prefix . 'author' ]
) {
return $entries;
}
$args = array(
'has_published_posts' => array( 'post' ),
);
$authors = get_users( $args );
if ( ! $authors ) {
return $entries;
}
foreach ( $authors as $author ) {
$args = array(
'author' => $author->ID,
);
$entries[] = array(
'loc' => get_author_posts_url( $author->ID ),
'lastmod' => aioseop_last_modified_post( $args ) ? aioseop_last_modified_post( $args )->post_modified_gmt : '',
'changefreq' => $this->get_default_frequency( 'author' ),
'priority' => $this->get_default_priority( 'author' ),
);
}
return apply_filters( 'aioseo_sitemap_author', $entries );
}
/**
* Get Comment Count Stats
*
* Return comment statistics on an array of posts.
*
* @since ?
*
* @param $posts
* @return array|int
*/
public function get_comment_count_stats( $posts ) {
$count = 0;
$total = 0.0;
$min = null;
$max = 0;
if ( is_array( $posts ) ) {
foreach ( $posts as $post ) {
if ( ! empty( $post->comment_count ) ) {
$cnt = $post->comment_count;
$count ++;
$total += $cnt;
if ( null === $min ) {
$min = $cnt;
}
if ( $max < $cnt ) {
$max = $cnt;
}
if ( $min > $cnt ) {
$min = $cnt;
}
}
}
}
if ( $count ) {
return array(
'max' => $max,
'min' => $min,
'total' => $total,
'count' => $cnt,
);
} else {
return 0;
}
}
/**
* Gets the last modified timestamp, priority and frequency for posts.
*
* @since 3.4.0 Check if post has sitemap priority/frequency in postmeta. Renamed function to better reflect purpose.
*
* @param $posts The posts that need their sitemap meta info to be populated.
* @param bool $prio_override
* @param bool $freq_override
* @param string $linkfunc
* @param string $type Type of entity being fetched viz. author, post etc.
* @return array $populated_posts The posts with their sitemap meta info.
*/
public function get_posts_data( $posts, $prio_override = false, $freq_override = false, $linkfunc = 'get_permalink', $type = 'post' ) {
$populated_posts = array();
$args = array(
'prio_override' => $prio_override,
'freq_override' => $freq_override,
'linkfunc' => $linkfunc,
);
if ( $prio_override && $freq_override ) {
$stats = 0;
} else {
$stats = $this->get_comment_count_stats( $posts );
}
if ( is_array( $posts ) ) {
foreach ( $posts as $key => $post ) {
// Determine if we check the post for images.
$is_single = true;
$post->filter = 'sample';
$timestamp = null;
if ( 'get_permalink' === $linkfunc ) {
$url = $this->get_permalink( $post );
} else {
$url = call_user_func( $linkfunc, $post );
$is_single = false;
}
if ( strpos( $url, '__trashed' ) !== false ) {
// excluded trashed urls.
continue;
}
$date = $post->post_modified_gmt;
if ( '0000-00-00 00:00:00' === $date ) {
$date = $post->post_date_gmt;
}
if ( '0000-00-00 00:00:00' !== $date ) {
$timestamp = $date;
$date = date( 'Y-m-d\TH:i:s\Z', mysql2date( 'U', $date ) );
} else {
$date = 0;
}
if ( $prio_override && $freq_override ) {
$pr_info = array(
'lastmod' => $date,
'changefreq' => null,
'priority' => null,
);
} else {
if ( empty( $post->comment_count ) ) {
$stat = 0;
} else {
$stat = $stats;
}
if ( ! empty( $stat ) ) {
$stat['comment_count'] = $post->comment_count;
}
$pr_info = $this->get_prio_calc( $date, $stat );
}
if ( $freq_override ) {
$pr_info['changefreq'] = $freq_override;
}
if ( $prio_override ) {
$pr_info['priority'] = $prio_override;
}
if ( isset( $this->options[ $this->prefix . 'prio_post' ] ) && 'no' !== $this->options[ $this->prefix . 'prio_post' ] ) {
if ( 'sel' !== $this->options[ $this->prefix . 'prio_post' ] ) {
$pr_info['priority'] = $this->options[ $this->prefix . 'prio_post' ];
} elseif ( 'no' !== $this->options[ $this->prefix . 'prio_post_' . $post->post_type ] ) {
$pr_info['priority'] = $this->options[ $this->prefix . 'prio_post_' . $post->post_type ];
}
}
if ( isset( $this->options[ $this->prefix . 'freq_post' ] ) && 'no' !== $this->options[ $this->prefix . 'freq_post' ] ) {
if ( 'sel' !== $this->options[ $this->prefix . 'freq_post' ] ) {
$pr_info['changefreq'] = $this->options[ $this->prefix . 'freq_post' ];
} elseif ( 'no' !== $this->options[ $this->prefix . 'freq_post_' . $post->post_type ] ) {
$pr_info['changefreq'] = $this->options[ $this->prefix . 'freq_post_' . $post->post_type ];
}
}
$pr_info = array(
'loc' => $url,
) + $pr_info; // Prepend loc to the array.
if ( is_float( $pr_info['priority'] ) ) {
$pr_info['priority'] = sprintf( '%0.1F', $pr_info['priority'] );
}
// add the rss specific data.
if ( $timestamp ) {
$title = null;
switch ( $type ) {
case 'author':
$title = get_the_author_meta( 'display_name', $key );
break;
default:
$title = get_the_title( $post );
break;
}
// RSS expects the GMT date.
$timestamp = mysql2date( 'U', $post->post_modified_gmt );
$pr_info['rss'] = array(
'title' => $title,
'description' => get_post_field( 'post_excerpt', $post->ID ),
'pubDate' => date( 'r', $timestamp ),
'timestamp ' => $timestamp,
'post_type' => $post->post_type,
);
}
$pr_info['image:image'] = $is_single ? $this->get_images_from_post( $post ) : null;
$pr_info = apply_filters( $this->prefix . 'prio_item_filter', $pr_info, $post, $args );
if ( ! empty( $pr_info ) ) {
$populated_posts[] = $pr_info;
}
}
}
return $populated_posts;
}
/**
* Get Images from Term
*
* Return the images attached to the term.
*
* @since 2.4.0
* @since 3.0.0 Remove check for WP 4.4
* @since 3.4.0 Changed access modifier.
*
* @param WP_Term $term the term object.
* @return array
*/
protected function get_images_from_term( $term ) {
if ( ! aiosp_include_images() ) {
return array();
}
$images = array();
$thumbnail_id = get_term_meta( $term->term_id, 'thumbnail_id', true );
if ( $thumbnail_id ) {
$image = wp_get_attachment_url( $thumbnail_id );
if ( $image ) {
$images['image:image'] = array(
'image:loc' => $image,
'image:caption' => wp_get_attachment_caption( $thumbnail_id ),
'image:title' => get_the_title( $thumbnail_id ),
);
}
}
return $images;
}
/**
* Get Images from Post
*
* Return the images from the post.
*
* @todo Add ~`get_attachment_postid_to_url()` function.
* @todo Benchmark `wp_get_attachment_image_src()` & `wp_get_attachment_url()`.
* @todo Look into using 'wp_get_attachment_image_url()'.
*
* @since 2.4.0
* @since 2.11.0 Optimization #2008 - Reduce the need to convert url to id.
* @since 3.4.0 Changed access modifier.
*
* @param WP_Post|int $post the post object.
* @return array
*/
protected function get_images_from_post( $post ) {
if ( ! aiosp_include_images() ) {
return array();
}
$rtn_image_attributes = array();
$post_image_ids = array();
$post_image_urls = array();
$transient_update = false;
if ( is_numeric( $post ) ) {
if ( 0 === $post ) {
return null;
}
$post = get_post( $post );
}
if ( 'attachment' === $post->post_type ) {
if ( false === strpos( $post->post_mime_type, 'image/' ) ) {
// Ignore all attachments except images.
return null;
}
$attributes = wp_get_attachment_image_src( $post->ID );
if ( $attributes ) {
$rtn_image_attributes[] = array(
'image:loc' => $this->aioseop_clean_url( $attributes[0] ),
'image:caption' => wp_get_attachment_caption( $post->ID ),
'image:title' => get_the_title( $post->ID ),
);
}
return $rtn_image_attributes;
}
// Set Image IDs w/ URLs.
if ( is_null( $this->image_ids_urls ) ) {
// Get Transient/Cache data.
if ( is_multisite() ) {
$this->image_ids_urls = get_site_transient( 'aioseop_multisite_attachment_ids_urls' );
} else {
$this->image_ids_urls = get_transient( 'aioseop_attachment_ids_urls' );
}
// Set default if no data exists.
if ( false === $this->image_ids_urls ) {
$this->image_ids_urls = array();
}
}
/**
* Static attachment cache, 1 query vs. n posts.
*
* Concepts like this should be followed; although this could possibly be improved (maybe as a wrapped function) but is still good code.
*/
static $post_thumbnails;
if ( is_null( $post_thumbnails ) || defined( 'AIOSEOP_UNIT_TESTING' ) ) {
global $wpdb;
$post_thumbnails = $wpdb->get_results(
"SELECT post_ID, meta_value FROM {$wpdb->postmeta} WHERE meta_key = '_thumbnail_id'",
ARRAY_A
);
if ( $post_thumbnails ) {
$post_thumbnails = array_combine(
wp_list_pluck( $post_thumbnails, 'post_ID' ),
wp_list_pluck( $post_thumbnails, 'meta_value' )
);
}
}
if ( isset( $post_thumbnails[ $post->ID ] ) ) {
$post_image_ids[] = intval( $post_thumbnails[ $post->ID ] );
}
$post_image_ids = array_merge( $post_image_ids, $this->get_gallery_image_ids( $post ) );
$this->get_gallery_images( $post, $post_image_urls );
// Get image URLs from content.
$content = $post->post_content;
$content .= $this->get_content_from_galleries( $content );
$this->parse_content_for_images( $content, $post_image_urls );
if ( ! empty( $post_image_urls ) ) {
// Remove any invalid/empty images.
$post_image_urls = array_filter( $post_image_urls, array( $this, 'is_image_url_valid' ) );
// If possible, get ID from URL, and store the post's attachment ID => URL value.
// This is to base the attachment query on the ID instead of the URL; which is less SQL intense.
foreach ( $post_image_urls as $k1_index => $v1_image_url ) {
$v1_image_url = aiosp_common::absolutize_url( $v1_image_url );
$post_image_urls[ $k1_index ] = $v1_image_url;
$v1_image_url = $this->prepare_url( $v1_image_url );
$attachment_id = aiosp_common::attachment_url_to_postid( $v1_image_url );
if ( $attachment_id ) {
if ( ! isset( $this->image_ids_urls[ $attachment_id ] ) ) {
// Use transient/cache data.
$this->image_ids_urls[ $attachment_id ] = array( $v1_image_url );
$transient_update = true;
} else {
// If transient/cache data is already set, and URL is not already stored.
if ( ! in_array( $v1_image_url, $this->image_ids_urls[ $attachment_id ], true ) ) {
$this->image_ids_urls[ $attachment_id ][] = $v1_image_url;
$transient_update = true;
}
}
// Store and use ID instead.
array_push( $post_image_ids, $attachment_id );
unset( $post_image_urls[ $k1_index ] );
}
}
}
// Site's Images.
if ( $post_image_ids ) {
// Filter out duplicates.
$post_image_ids = array_unique( $post_image_ids );
foreach ( $post_image_ids as $v1_image_id ) {
// Set base URL to display later in this instance, or later (transient/cache) instances.
// Converting ID from URL can also be heavy on memory & time.
if ( ! isset( $this->image_ids_urls[ $v1_image_id ] ) ) {
// Sets any remaining post image IDs that weren't converted from URL.
$this->image_ids_urls[ $v1_image_id ] = array(
'base_url' => $this->aioseop_clean_url( wp_get_attachment_url( $v1_image_id ) ),
);
$transient_update = true;
} else {
if ( empty( $this->image_ids_urls[ $v1_image_id ]['base_url'] ) ) {
$this->image_ids_urls[ $v1_image_id ]['base_url'] = $this->aioseop_clean_url( wp_get_attachment_url( $v1_image_id ) );
$transient_update = true;
}
}
// Set return variable for image data/attributes.
$rtn_image_attributes[] = array(
'image:loc' => $this->image_ids_urls[ $v1_image_id ]['base_url'],
'image:caption' => wp_get_attachment_caption( $v1_image_id ),
'image:title' => get_the_title( $v1_image_id ),
);
}
}
// External/Custom images remaining.
if ( ! empty( $post_image_urls ) ) {
foreach ( $post_image_urls as $v1_image_url ) {
$rtn_image_attributes[] = array(
'image:loc' => $v1_image_url,
);
}
}
if ( $transient_update ) {
add_action( 'shutdown', array( $this, 'set_transient_attachment_ids_urls' ) );
}
return $rtn_image_attributes;
}
/**
* Prepares a given image URL for further processing by removing the image dimensions from the slug.
*
* @since 3.6.0
*
* @param string $url The image URL.
* @return string The prepared image URL.
*/
private function prepare_url( $url ) {
$upload_dir = wp_get_upload_dir();
$upload_dir_url = $upload_dir['baseurl'];
if ( ! preg_match( "#$upload_dir_url.*#", $url ) ) {
return $url;
}
return preg_replace( '#(-[0-9]*x[0-9]*)#', '', $url );
}
/**
* Set Transient Attachment IDs => URLS
*
* Set Transient for Image IDs => URLs
*
* @since 2.11
*/
public function set_transient_attachment_ids_urls() {
if ( is_multisite() ) {
set_site_transient( 'aioseop_multisite_attachment_ids_urls', $this->image_ids_urls, DAY_IN_SECONDS );
} else {
set_transient( 'aioseop_attachment_ids_urls', $this->image_ids_urls, DAY_IN_SECONDS );
}
}
/**
* Get Gallery Images
*
* Fetch images from WP, Jetpack and WooCommerce galleries.
*
* @since 2.4.2
* @since 2.11 Optimization #2008 - Reduce the need to convert url to id.
*
* @param WP_Post $post The post.
* @param array $images the array of images.
*/
private function get_gallery_images( $post, &$images ) {
if ( false === apply_filters( 'aioseo_include_images_in_wp_gallery', true ) ) {
return;
}
// Check images galleries in the content. DO NOT run the_content filter here as it might cause issues with other shortcodes.
if ( has_shortcode( $post->post_content, 'gallery' ) ) {
/*
* TODO Investigate other alternatives to retrieve ID instead. Specifically Jetpack data.
*
* Is this even necessary? Jetpack uses many of the WP functions, some of which may already be in use.
* This is also limited to 1 source, and doesn't check other sources once a value is obtained.
*
* @link https://hayashikejinan.com/wp-content/uploads/jetpack_api/classes/Jetpack_PostImages.html
*/
if ( class_exists( 'Jetpack_PostImages' ) ) {
// Get the jetpack gallery images.
$jetpack = Jetpack_PostImages::get_images( $post->ID );
if ( $jetpack ) {
foreach ( $jetpack as $jetpack_image ) {
$images[] = $jetpack_image['src'];
}
}
}
}
$images = array_unique( $images );
}
/**
* Get Gallery Image IDs
*
* @uses get_post_galleries()
* @link https://developer.wordpress.org/reference/functions/get_post_galleries/
*
* @since 2.11
*
* @param WP_Post $post
* @return array
*/
private function get_gallery_image_ids( $post ) {
$rtn_image_ids = array();
if ( false === apply_filters( 'aioseo_include_images_in_wp_gallery', true ) ) {
return $rtn_image_ids;
}
// Check images galleries in the content. DO NOT run the_content filter here as it might cause issues with other shortcodes.
if ( has_shortcode( $post->post_content, 'gallery' ) ) {
// Get the default WP gallery images.
$galleries = get_post_galleries( $post, false );
if ( ! empty( $galleries ) ) {
foreach ( $galleries as $gallery ) {
$gallery_ids = explode( ',', $gallery['ids'] );
if ( ! empty( $gallery_ids ) ) {
foreach ( $gallery_ids as $image_id ) {
// Skip if invalid id.
if ( ! is_numeric( $image_id ) ) {
continue;
}
$image_id = intval( $image_id );
array_push( $rtn_image_ids, $image_id );
}
}
}
}
}
// Check WooCommerce product gallery.
if ( class_exists( 'WooCommerce' ) ) {
$wc_image_ids = get_post_meta( $post->ID, '_product_image_gallery', true );
if ( ! empty( $woo_images ) ) {
$wc_image_ids = array_filter( explode( ',', $wc_image_ids ) );
foreach ( $wc_image_ids as $image_id ) {
if ( is_numeric( $image_id ) ) {
$image_id = intval( $image_id );
array_push( $rtn_image_ids, $image_id );
}
}
}
}
return array_unique( $rtn_image_ids );
}
/**
* Get Content from Galleries
*
* Parses the content to find out if specified images galleries exist and if they do, parse them for images.
* Supports NextGen.
*
* @since 2.4.2
*
* @param string $content The post content.
* @return string
*/
private function get_content_from_galleries( $content ) {
// Support for NextGen Gallery.
static $gallery_types;
$gallery_types = array( 'ngg', 'ngg_images' );
$types = apply_filters( 'aioseop_gallery_shortcodes', $gallery_types );
$gallery_content = '';
if ( ! $types ) {
return $gallery_content;
}
$found = array();
if ( $types ) {
foreach ( $types as $type ) {
if ( has_shortcode( $content, $type ) ) {
$found[] = $type;
}
}
}
// If none of the shortcodes-of-interest are found, bail.
if ( empty( $found ) ) {
return $gallery_content;
}
$galleries = array();
if ( ! preg_match_all( '/' . get_shortcode_regex() . '/s', $content, $matches, PREG_SET_ORDER ) ) {
return $gallery_content;
}
// Collect the shortcodes and their attributes.
foreach ( $found as $type ) {
foreach ( $matches as $shortcode ) {
if ( $type === $shortcode[2] ) {
$attributes = shortcode_parse_atts( $shortcode[3] );
if ( '' === $attributes ) { // Valid shortcode without any attributes.
$attributes = array();
}
$galleries[ $shortcode[2] ] = $attributes;
}
}
}
// Recreate the shortcodes and then render them to get the HTML content.
if ( $galleries ) {
foreach ( $galleries as $shortcode => $attributes ) {
$code = '[' . $shortcode;
foreach ( $attributes as $key => $value ) {
$code .= " $key=$value";
}
$code .= ']';
$gallery_content .= aioseop_do_shortcodes( $code );
}
}
return $gallery_content;
}
/**
* AIOSEOP Clean URL
*
* Cleans the URL so that its acceptable in the sitemap.
*
* @since 2.4.1
*
* @param string $url The image url.
* @return string
*/
public function aioseop_clean_url( $url ) {
// remove the query string.
$url = strtok( $url, '?' );
// make the url XML-safe.
$url = htmlspecialchars( $url, ENT_COMPAT, 'UTF-8' );
// Make the url absolute, if its relative.
$url = aiosp_common::absolutize_url( $url );
return apply_filters( 'aioseop_clean_url', $url );
}
/**
* The is_image_url_valid() function.
*
* Checks whether the image URL is valid.
*
* @since 2.4.1
* @since 2.4.3 Compatibility with Pre v4.7 wp_parse_url().
* @since 2.11.0 Sitemap Optimization #2008 - Changed to a more appropriate name.
* @since 3.0.0 Remove checks for old WP versions.
* @since 3.2.0 Remove redundant code.
*
* @param string $image The image src.
* @return bool
*/
public function is_image_url_valid( $image ) {
// Bail if empty image.
if ( empty( $image ) ) {
return false;
}
$image = aiosp_common::absolutize_url( $image );
$extn = pathinfo( $image, PATHINFO_EXTENSION );
$allowed = apply_filters( 'aioseop_allowed_image_extensions', self::$image_extensions );
$match = false;
foreach ( $allowed as $extension ) {
$pattern = "#${extension}.*#";
if ( preg_match( $pattern, $extn ) ) {
$match = true;
}
}
// Bail if image does not refer to an image file otherwise Google Search Console might reject the sitemap.
if ( ! $match ) {
return false;
}
$image_host = wp_parse_url( $image, PHP_URL_HOST );
$host = wp_parse_url( home_url(), PHP_URL_HOST );
if ( $image_host !== $host ) {
// Allowed hosts will be provided in a wildcard format i.e. img.yahoo.* or *.akamai.*.
// And we will convert that into a regular expression for matching.
$whitelist = apply_filters( 'aioseop_images_allowed_from_hosts', array() );
$allowed = false;
if ( $whitelist ) {
foreach ( $whitelist as $pattern ) {
if ( preg_match( '/' . str_replace( '*', '.*', $pattern ) . '/', $image_host ) === 1 ) {
$allowed = true;
break;
}
}
}
return $allowed;
}
return true;
}
/**
* Parse Content for Images
*
* Parse the post for images.
*
* @since 2.9.1
*
* @param string $content the post content.
* @param array $images the array of images.
*/
public function parse_content_for_images( $content, &$images ) {
// These tags should be WITHOUT trailing space because some plugins such as the nextgen gallery put newlines immediately after loadHTML( $content );
libxml_clear_errors();
// phpcs:ignore WordPress.NamingConventions.ValidVariableName.UsedPropertyNotSnakeCase
$dom->preserveWhiteSpace = false;
$matches = $dom->getElementsByTagName( 'img' );
foreach ( $matches as $match ) {
$images[] = $match->getAttribute( 'src' );
}
} else {
// Fall back to regex, but also report an error.
global $img_err_msg;
if ( ! isset( $img_err_msg ) ) {
// we will log this error message only once, not per post.
$img_err_msg = true;
$this->debug_message( 'DOMDocument not found; using REGEX' );
}
preg_match_all( '/options[ "{$this->prefix}indexes" ] ) ) {
$args['number'] = $this->max_posts;
$args['offset'] = $page * $this->max_posts;
}
$args['taxonomy'] = $this->show_or_hide_taxonomy( $taxonomies );
$args['exclude'] = array();
if ( $this->option_isset( 'excl_terms' ) ) {
foreach ( $taxonomies as $v1_taxonomy ) {
if ( isset( $this->options[ $this->prefix . 'excl_terms' ][ $v1_taxonomy ] ) ) {
$args['exclude'] = array_merge( $args['exclude'], $this->options[ $this->prefix . 'excl_terms' ][ $v1_taxonomy ]['terms'] );
}
}
}
if ( empty( $taxonomies ) ) {
return apply_filters( 'aioseop_sitemap_exclude_tax_terms', $args );
}
$noindexed_tax_terms = array();
// Get globally noindexed taxonomies.
foreach ( $taxonomies as $taxonomy ) {
$tax_name = $taxonomy;
if ( 'post_tag' === $tax_name ) {
$tax_name = 'tags';
}
if ( isset( $aioseop_options[ "aiosp_${tax_name}_noindex" ] ) &&
$aioseop_options[ "aiosp_${tax_name}_noindex" ]
) {
$noindexed_tax_terms = array_merge( $noindexed_tax_terms, get_terms( $taxonomy ) );
}
if ( isset( $aioseop_options['aiosp_tax_noindex'] ) &&
is_array( $aioseop_options['aiosp_tax_noindex'] ) &&
in_array( $tax_name, $aioseop_options['aiosp_tax_noindex'] )
) {
$noindexed_tax_terms = array_merge( $noindexed_tax_terms, get_terms( $taxonomy ) );
}
}
// Get individually noindexed taxonomy terms.
$noindexed_tax_terms = array_merge(
$noindexed_tax_terms,
get_terms(
array(
'meta_key' => '_aioseop_noindex',
'meta_value' => 'on',
)
)
);
foreach ( $noindexed_tax_terms as $tax_term ) {
array_push( $args['exclude'], $tax_term->term_id );
}
/**
* The aioseop_sitemap_exclude_tax_terms filter hook.
*
* Allows users to exclude (or include) taxonomy terms from the sitemap.
*
* @since 2.9
* @since 3.2.0 Rename filter hook & remove redundant params.
*
* @param array $args {
* @type array $taxonomy Name of the taxonomy that is being included in the sitemap.
* @type array $exclude IDs of taxonomy terms of the relevant taxonomy that need to be excluded.
* }
*/
$args = apply_filters( 'aioseop_sitemap_exclude_tax_terms', $args );
return $args;
}
/**
* Set Post Args
*
* Return excluded categories and pages for post queries.
*
* @since ?
* @since 3.0 Change 'excl_terms' to tax_query format. (Pro #240)
*
* @param $args
* @return mixed
*/
public function set_post_args( $args ) {
if ( $this->option_isset( 'excl_terms' ) ) {
foreach ( $this->options[ $this->prefix . 'excl_terms' ] as $k1_taxonomy => $v1_tax_terms ) {
if ( ! isset( $args['tax_query'] ) ) {
$args['tax_query'] = array(
'relation' => 'AND',
);
}
$args['tax_query'][] = array(
'taxonomy' => $k1_taxonomy,
'terms' => $v1_tax_terms['terms'],
'operator' => 'NOT IN',
);
}
}
if ( $this->option_isset( 'excl_pages' ) ) {
$args['exclude'] = $this->options[ $this->prefix . 'excl_pages' ];
}
return $args;
}
/**
* Returns all sitemap entries for a custom post type.
*
* @since 3.2.0 Update Last Change timestamp for WooCommerce shop page.
* @since 3.4.0 Renamed function to better reflect purpose.
*
* @param string $include
* @param string $status The editorial status of the post (e.g. "publish", "draft", "private").
* @param int $page_number The number of the page.
* @return array The custom post entries with their metadata (last modified timestamp, priority, frequency).
*/
public function get_custom_posts_data( $include = 'any', $status = 'publish', $page_number = 0 ) {
$posts = array();
$page_query = array();
if ( ! empty( $this->options[ "{$this->prefix}indexes" ] ) ) {
$page_query = array( 'offset' => $page_number * $this->max_posts );
}
if ( ( 'publish' === $status ) && ( 'attachment' === $include ) ) {
$status = 'inherit';
}
if ( is_array( $include ) ) {
$pos = array_search( 'attachment', $include );
if ( false !== $pos ) {
unset( $include[ $pos ] );
$att_args = array(
'post_type' => 'attachment',
'post_status' => 'inherit',
);
$att_args = array_merge( $att_args, $page_query );
$posts = $this->get_all_post_type_data( $att_args );
}
}
$args = array(
'post_type' => $include,
'post_status' => $status,
);
$args = array_merge( $args, $page_query );
$args = $this->set_post_args( $args );
$posts = array_merge( $this->get_all_post_type_data( $args ), $posts );
$links = $this->get_posts_data( $posts, $this->get_default_priority( 'post' ), $this->get_default_frequency( 'post' ) );
$links = array_merge( $this->get_post_archive_data( $include ), $links );
$is_sitemap_indexes_disabled = empty( $this->options['aiosp_sitemap_indexes'] );
if ( $is_sitemap_indexes_disabled || ( ! $is_sitemap_indexes_disabled && 'page' === $include ) ) {
$links = $this->get_posts_page_timestamp( $links );
$links = $this->get_prio_freq_static_homepage( $links );
$links = $this->get_prio_freq_static_blogpage( $links );
$links = $this->update_woocommerce_shop_timestamp( $links );
}
if ( 'page' === $include ) {
$links = $this->removeWooCommercePages( $links );
}
return $links;
}
/**
* Excludes the Cart, Checkout and My Account pages from the sitemap if WooCommerce noindex them.
*
* @since 3.6.0
*
* @param array $urls The URLs.
* @return array $remainingUrls The filtered URLs.
*/
private function removeWooCommercePages( $urls ) {
// Check if WooCommerce is noindexing its own pages.
if ( ! aioseop_is_woocommerce_active() || ! has_action( 'wp_head', 'wc_page_noindex' ) ) {
return $urls;
}
$pages = array(
wc_get_cart_url(),
wc_get_checkout_url(),
wc_get_page_permalink( 'myaccount' )
);
$remainingUrls = array();
foreach ( $urls as $url ) {
$isNoindexed = false;
foreach( $pages as $page ) {
if ( $page === $url['loc'] ) {
$isNoindexed = true;
break;
}
}
if ( ! $isNoindexed ) {
$remainingUrls[] = $url;
}
}
return $remainingUrls;
}
/**
* The get_prio_freq_static_homepage() function.
*
* Sets the priority and frequency for the homepage if it is static.
*
* @since 3.2.0
*
* @param array $links
* @return array $links
*/
private function get_prio_freq_static_homepage( $links ) {
if ( 0 === (int) get_option( 'page_on_front' ) || 'page' !== get_option( 'show_on_front' ) ) {
return $links;
}
$prio = 'no';
$freq = 'no';
if ( isset( $this->options['aiosp_sitemap_prio_homepage'] ) ) {
$prio = $this->options['aiosp_sitemap_prio_homepage'];
}
if ( isset( $this->options['aiosp_sitemap_freq_homepage'] ) ) {
$freq = $this->options['aiosp_sitemap_freq_homepage'];
}
$homepage_url = get_site_url() . '/';
$homepage_index = array_search( $homepage_url, array_column( $links, 'loc' ) ); // phpcs:ignore PHPCompatibility.FunctionUse.NewFunctions.array_columnFound
if ( ! $homepage_url ) {
return $links;
}
if ( 'no' !== $prio ) {
$links[ $homepage_index ]['priority'] = $prio;
}
if ( 'no' !== $freq ) {
$links[ $homepage_index ]['changefreq'] = $freq;
}
return $links;
}
/**
* Sets the priority and frequency for the static blogpage.
*
* @since 3.4.0
*
* @param array $links
* @return array $links
*/
protected function get_prio_freq_static_blogpage( $links ) {
$blogpage_id = (int) get_option( 'page_for_posts' );
$permalink = get_permalink( $blogpage_id );
if ( 0 === $blogpage_id || 'page' !== get_option( 'show_on_front' ) ) {
return $links;
}
$blogpage_index = array_search( $permalink, array_column( $links, 'loc' ) ); // phpcs:ignore PHPCompatibility.FunctionUse.NewFunctions.array_columnFound
if ( isset( $this->options[ $this->prefix . 'prio_post' ] ) && 'no' !== $this->options[ $this->prefix . 'prio_post' ] ) {
if ( 'sel' !== $this->options[ $this->prefix . 'prio_post' ] ) {
$links[ $blogpage_index ]['priority'] = $this->options[ $this->prefix . 'prio_post' ];
} elseif ( 'no' !== $this->options[ $this->prefix . 'prio_post_' . 'page' ] ) {
$links[ $blogpage_index ]['priority'] = $this->options[ $this->prefix . 'prio_post_' . 'page' ];
}
}
if ( isset( $this->options[ $this->prefix . 'freq_post' ] ) && 'no' !== $this->options[ $this->prefix . 'freq_post' ] ) {
if ( 'sel' !== $this->options[ $this->prefix . 'freq_post' ] ) {
$links[ $blogpage_index ]['changefreq'] = $this->options[ $this->prefix . 'freq_post' ];
} elseif ( 'no' !== $this->options[ $this->prefix . 'freq_post_' . 'page' ] ) {
$links[ $blogpage_index ]['changefreq'] = $this->options[ $this->prefix . 'freq_post_' . 'page' ];
}
}
return $links;
}
/**
* The update_woocommerce_shop_timestamp() function.
*
* Updates the Last Change timestamp for the WooCommerce shop page based on the last modified product - #2126.
*
* @since 3.2.0
*
* @param array $links
* @return array $links
*/
private function update_woocommerce_shop_timestamp( $links ) {
if ( ! aioseop_is_woocommerce_active() ) {
return $links;
}
$shop_page_url = get_permalink( wc_get_page_id( 'shop' ) );
$shop_page_index = array_search( $shop_page_url, array_column( $links, 'loc' ) ); // phpcs:ignore PHPCompatibility.FunctionUse.NewFunctions.array_columnFound
if ( ! $shop_page_index ) {
return $links;
}
// TODO Use get_last_modified_post_timestamp() instead when #2721 is merged.
$latest_modified_product = new WP_Query(
array(
'post_type' => 'product',
'post_status' => 'publish',
'posts_per_page' => 1,
'orderby' => 'modified',
'order' => 'DESC',
)
);
if ( $latest_modified_product->have_posts() ) {
$timestamp = $latest_modified_product->posts[0]->post_modified_gmt;
$lastmod = date( 'Y-m-d\TH:i:s\Z', mysql2date( 'U', $timestamp ) );
// Last Change timestamp needs to be inserted as second attribute in order to have valid sitemap schema.
// TODO Use insert_timestamp_as_second_attribute() instead when #2721 is merged.
$links[ $shop_page_index ] = array_slice( $links[ $shop_page_index ], 0, 1, true ) + array( 'lastmod' => $lastmod ) + array_slice( $links[ $shop_page_index ], 1, null, true );
}
return $links;
}
/**
* Get All Permalinks
*
* Return a list of all permalinks.
*
* @since ?
*
* @param string $include
* @param string $status
* @return array
*/
public function get_all_permalinks( $include = 'any', $status = 'publish' ) {
$args = array(
'post_type' => $include,
'post_status' => $status,
);
$args = $this->set_post_args( $args );
$posts = $this->get_all_post_type_data( $args );
$links = $this->get_post_permalinks( $posts );
if ( $this->option_isset( 'archive' ) ) {
$links = array_merge( $links, $this->get_archive_permalinks( $posts ) );
}
if ( $this->option_isset( 'author' ) ) {
$links = array_merge( $links, $this->get_author_permalinks( $posts ) );
}
return $links;
}
/**
* Cache Structure
*
* Static memory cache for permalink_structure option.
*
* @since ?
*
* @param $pre
* @return null
*/
public function cache_structure( $pre ) {
return $this->cache_struct;
}
/**
* Cache Home
*
* Static memory cache for home option.
*
* @since ?
*
* @param $pre
* @return null
*/
public function cache_home( $pre ) {
return $this->cache_home;
}
/**
* Cache Options
*
* Cache permalink_structure and home for repeated sitemap queries.
*
* @since ?
*/
public function cache_options() {
static $start = true;
if ( $start ) {
$this->cache_struct = get_option( 'permalink_structure' );
if ( ! empty( $this->cache_struct ) ) {
add_filter( 'pre_option_permalink_structure', array( $this, 'cache_structure' ) );
}
$this->cache_home = get_option( 'home' );
if ( ! empty( $this->cache_home ) ) {
add_filter( 'pre_option_home', array( $this, 'cache_home' ) );
}
$start = false;
}
}
/**
* Get Term Link
*
* Call get_term_link with caching in place.
*
* @since ?
*
* @param $term
* @param string $taxonomy
* @return string|WP_Error
*/
public function get_term_link( $term, $taxonomy = '' ) {
static $start = true;
if ( $start ) {
$this->cache_options();
$start = false;
}
return get_term_link( $term, $taxonomy );
}
/**
* Get Permalink
*
* Call get_permalink with caching in place.
*
* @since ?
*
* @param $post
* @return false|string
*/
public function get_permalink( $post ) {
static $start = true;
if ( $start ) {
$this->cache_options();
$start = false;
}
return aioseop_get_permalink( $post );
}
/**
* Show or hide the taxonomy/taxonomies.
*
* @since 3.0.0
*
* @param array $taxonomy The array of taxonomy slugs.
*
* @return array The array of taxonomy slugs that need to be shown.
*/
private function show_or_hide_taxonomy( $taxonomy ) {
/**
* Determines whether to show or hide the taxonomy/taxonomies.
*
* @since 3.0.0
*
* @param array $taxonomy The array of taxonomy slugs.
*/
return apply_filters( "{$this->prefix}show_taxonomy", $taxonomy );
}
/**
* Get All Terms Counts
*
* Return term counts using wp_count_terms().
*
* @since ?
*
* @param $args
* @return array|int|mixed|null|WP_Error
*/
public function get_all_term_counts( $args ) {
$term_counts = null;
if ( ! empty( $args ) && ! empty( $args['taxonomy'] ) ) {
// TODO Add `true` in 3rd argument with in_array(); which changes it to a strict comparison.
if ( ! is_array( $args['taxonomy'] ) || ( count( $args['taxonomy'] ) == 1 ) ) {
if ( is_array( $args['taxonomy'] ) ) {
$args['taxonomy'] = array_shift( $args['taxonomy'] );
}
$term_counts = wp_count_terms( $this->show_or_hide_taxonomy( $args['taxonomy'] ), array( 'hide_empty' => true ) );
} else {
foreach ( $args['taxonomy'] as $taxonomy ) {
if ( 'all' === $taxonomy ) {
continue;
}
$term_counts[ $taxonomy ] = wp_count_terms( $this->show_or_hide_taxonomy( $taxonomy ), array( 'hide_empty' => true ) );
}
}
}
$term_counts = apply_filters( $this->prefix . 'term_counts', $term_counts, $args );
return $term_counts;
}
/**
* Get All Post Counts
*
* Return post counts.
*
* @since ?
* @since 2.4.3 Refactored to use get_post_count() instead of wp_count_posts().
*
* @param $args
* @return array
*/
public function get_all_post_counts( $args ) {
$post_counts = array();
$status = 'inherit';
if ( ! empty( $args['post_status'] ) ) {
$status = $args['post_status'];
}
if ( ! empty( $args ) && ! empty( $args['post_type'] ) ) {
// #884: removed hard-to-understand code here which suspected $args['post_type'] to NOT be an array. Do not see any case in which this is likely to happen.
foreach ( $args['post_type'] as $post_type ) {
if ( 'all' === $post_type ) {
continue;
}
$post_type_count = (array) wp_count_posts( $post_type );
$post_counts[ $post_type ] = 0;
if ( 'attachment' === $post_type ) {
$post_counts[ $post_type ] = $post_type_count['inherit'];
} elseif ( ! empty( $post_type_count[ $status ] ) ) {
$post_counts[ $post_type ] = $post_type_count[ $status ];
}
}
}
$post_counts = apply_filters( $this->prefix . 'post_counts', $post_counts, $args );
return $post_counts;
}
/**
* Modify Post Params for External Plugins
*
* Modify the post arguments in case third-party plugins are being used e.g. WPML.
*
* @since 2.4.5
*
* @param $args
*/
public function modify_post_params_for_external_plugins( &$args ) {
// if WPML is being used, do not suppress filters.
if ( defined( 'ICL_SITEPRESS_VERSION' ) ) {
$args['suppress_filters'] = false;
}
$args = apply_filters( $this->prefix . 'modify_post_params', $args );
}
/**
* Get total post count.
*
* @param $args
*
* @return int
*/
public function get_total_post_count( $args ) {
$total = 0;
$counts = $this->get_all_post_counts( $args );
if ( ! empty( $counts ) ) {
foreach ( $counts as $count ) {
$total += $count;
}
}
return $total;
}
/**
* Get All Post Type Data
*
* Return post data using get_posts().
*
* @since ?
* @since 3.0 Changed to exclude noindex post types & posts. #1382
*
* @global array $aioseop_options
*
* @param array $args Query Arguments.
* @return array|mixed
*/
public function get_all_post_type_data( $args ) {
global $aioseop_options;
$defaults = array(
'numberposts' => $this->max_posts,
'offset' => 0,
'category' => 0,
'orderby' => 'post_date',
'order' => 'ASC',
'include' => array(),
'exclude' => array(),
'post_type' => 'any',
'meta_query' => array(),
'cache_results' => false,
'no_found_rows' => true,
);
$this->modify_post_params_for_external_plugins( $defaults );
/*
* Filter to exclude password protected posts.
* TODO: move to its own function and call it from here, returning whatever is appropriate.
* @since 2.3.12
*/
if ( apply_filters( 'aioseop_sitemap_include_password_posts', true ) === false ) {
$defaults['has_password'] = false;
}
$args = wp_parse_args( $args, $defaults );
if ( empty( $args['post_type'] ) ) {
return apply_filters( $this->prefix . 'post_filter', array(), $args );
}
$exclude_slugs = array();
if ( ! empty( $args['exclude'] ) ) {
$exclude = preg_split( '/[\s,]+/', trim( $args['exclude'] ) );
if ( ! empty( $exclude ) ) {
foreach ( $exclude as $k => $v ) {
if ( ! is_numeric( $v ) ) {
$exclude_slugs[] = $v;
unset( $exclude[ $k ] );
}
}
if ( ! empty( $exclude_slugs ) ) {
$args['exclude'] = implode( ',', $exclude );
}
}
}
if ( ! is_array( $args['exclude'] ) ) {
$args['exclude'] = explode( ',', $args['exclude'] );
}
// Exclude (method) query args.
$ex_args = $args;
$ex_args['meta_query'] = array(
'relation' => 'OR',
array(
'key' => '_aioseop_sitemap_exclude',
'value' => 'on',
'compare' => '=',
),
array(
'key' => '_aioseop_noindex',
'value' => 'on',
'compare' => '=',
),
);
// This needs to be -1 so that excluding posts isn't restricted to affect posts to not be excluded properly.
$ex_args['posts_per_page'] = -1;
$ex_args['fields'] = 'ids';
// Exclude (method) query.
$q_exclude = new WP_Query( $ex_args );
if ( ! empty( $q_exclude->posts ) ) {
$args['exclude'] = array_merge( $args['exclude'], $q_exclude->posts );
}
$this->excludes = array_merge( $args['exclude'], $exclude_slugs ); // Add excluded slugs and IDs to class var.
// Avoid if possible.
// Include (method) query args for including posts that may have been excluded;
// for example, exclude post type, but include certain posts.
// NOTE: Do NOT use this for basic including. It's best to avoid an additional query.
$args_include = array(
'post_type' => array(),
'meta_query' => array(
'relation' => 'OR',
array(
'key' => '_aioseop_noindex',
'value' => 'off',
'compare' => '=',
),
),
'posts_per_page' => $this->max_posts,
);
// Exclude from main query, and check if a Query Include is needed.
// Check for NoIndex Post Types, BUT also check for Index on NoIdex Post Type.
if ( is_array( $aioseop_options['aiosp_cpostnoindex'] ) ) {
// Check if wp_query_args post_type is an array or string.
if ( is_array( $args['post_type'] ) ) {
foreach ( $args['post_type'] as $index => $post_type ) {
if ( in_array( $post_type, $aioseop_options['aiosp_cpostnoindex'], true ) ) {
$args_include['post_type'][] = $post_type;
unset( $args['post_type'][ $index ] );
}
}
} else {
if ( in_array( $args['post_type'], $aioseop_options['aiosp_cpostnoindex'], true ) ) {
$args_include['post_type'][] = $args['post_type'];
$q_include = new WP_Query( $args_include );
// Return posts on single post type query, since no additional query is needed.
return $q_include->posts;
}
}
}
// Avoid if possible.
// Include (method) query.
// NOTE: Do NOT use this for basic including. It's best to avoid an additional query.
$posts_include = array();
if ( ! empty( $args_include['post_type'] ) ) {
$q_include = new WP_Query( $args_include );
// When posts exists from the include method, add to $posts_include to add to final query.
if ( ! empty( $q_include->posts ) ) {
$posts_include = $q_include->posts;
}
}
// TODO: consider using WP_Query instead of get_posts to improve consistency (does not improve performance).
/**
* {$module_prefix}post_query
*
* Arguments to use on get_posts().
*
* @since ?
*
* @param array $args {
* Arguments/params for get_posts.
* @see get_posts()
* @link https://developer.wordpress.org/reference/functions/get_posts/
* }
*/
$posts = get_posts( apply_filters( $this->prefix . 'post_query', $args ) );
// TODO Possibly change to exclude with post__not_in.
// Hardcoded exclude concept.
if ( ! empty( $exclude_slugs ) ) {
foreach ( $posts as $k => $v ) {
// TODO Add `true` in 3rd argument with in_array(); which changes it to a strict comparison.
if ( in_array( $v->post_name, $exclude_slugs ) ) {
unset( $posts[ $k ] );
}
}
}
// Hardcoded include concept.
foreach ( $posts_include as $k1_post_id => $v1_post ) {
$posts[ $k1_post_id ] = $v1_post;
}
/**
* {$module_prefix}post_filter
*
* Posts from finalized query for sitemap data.
*
* @since ?
*
* @param array $posts {
* @type WP_Post ${$post_id} {
* @see WP_Post object for more information.
* @link https://codex.wordpress.org/Class_Reference/WP_Post
* }
* }
* @param array $args {
* Arguments/params for get_posts.
* @see get_posts()
* @link https://developer.wordpress.org/reference/functions/get_posts/
* }
*/
$posts = apply_filters( $this->prefix . 'post_filter', $posts, $args );
return $posts;
}
}
}