This commit is contained in:
Javier Casares 2026-06-03 06:29:56 +00:00
commit 98b1cf0f3b
17 changed files with 2968 additions and 1313 deletions

View file

@ -8,6 +8,7 @@
namespace MediaRightsAudit\Admin;
use MediaRightsAudit\Admin\AttachmentDetailPage;
use MediaRightsAudit\External\HostnameFilter;
if ( ! class_exists( 'WP_List_Table' ) ) {
require_once ABSPATH . 'wp-admin/includes/class-wp-list-table.php';
@ -83,7 +84,7 @@ class MediaListTable extends \WP_List_Table {
'run_external_scan' => __( 'Run External Scan', 'robotstxt-mediaaudit' ),
'purge_external_data' => __( 'Purge External Data', 'robotstxt-mediaaudit' ),
'export_csv' => __( 'Export CSV', 'robotstxt-mediaaudit' ),
'export_external_csv' => __( 'Export External Results CSV', 'robotstxt-mediaaudit' ),
'export_csv_alerts' => __( 'Export CSV (Alerts only)', 'robotstxt-mediaaudit' ),
);
}
@ -279,6 +280,49 @@ class MediaListTable extends \WP_List_Table {
$es_val = $item['external_status'];
$status = is_string( $es_val ) ? $es_val : '';
// Check for alert domains first.
$top_domains_raw = isset( $item['top_domains_data'] ) && is_array( $item['top_domains_data'] )
? $item['top_domains_data']
: array();
$top_domains_data = array();
foreach ( $top_domains_raw as $td_key => $td_val ) {
if ( is_string( $td_key ) && is_int( $td_val ) ) {
$top_domains_data[ $td_key ] = $td_val;
}
}
$has_alert = ! empty( $top_domains_data ) && HostnameFilter::has_alert_domains( $top_domains_data );
if ( $has_alert ) {
$out = sprintf(
'<span class="mra-badge mra-badge-alert">%s</span>',
esc_html__( 'Alert', 'robotstxt-mediaaudit' )
);
// Sum counts for non-ignored domains.
$filtered_count = 0;
foreach ( $top_domains_data as $domain => $count ) {
if ( ! HostnameFilter::is_ignored( $domain ) ) {
$filtered_count += $count;
}
}
if ( $filtered_count > 0 ) {
$out .= sprintf(
' <span class="mra-badge mra-badge-match-count">%s</span>',
esc_html( number_format_i18n( $filtered_count ) )
);
}
$aid_val = $item['attachment_id'];
$aid = is_numeric( $aid_val ) ? (int) $aid_val : 0;
$out .= sprintf(
' <a href="#" class="mra-view-details" data-id="%d">%s</a>',
$aid,
esc_html__( 'View Results', 'robotstxt-mediaaudit' )
);
return $out;
}
$labels = array(
'pending' => __( 'Pending', 'robotstxt-mediaaudit' ),
'queued' => __( 'Queued', 'robotstxt-mediaaudit' ),
@ -385,6 +429,11 @@ class MediaListTable extends \WP_List_Table {
// External status filter.
echo '<select name="mra_external_status">';
echo '<option value="">' . esc_html__( 'All statuses', 'robotstxt-mediaaudit' ) . '</option>';
printf(
'<option value="alert"%s>%s</option>',
selected( $current_status, 'alert', false ),
esc_html__( 'Alert', 'robotstxt-mediaaudit' )
);
foreach ( self::EXTERNAL_STATUSES as $s ) {
printf(
'<option value="%s"%s>%s</option>',
@ -456,6 +505,13 @@ class MediaListTable extends \WP_List_Table {
$show_unused = ! empty( $_REQUEST['mra_unused'] );
// phpcs:enable WordPress.Security.NonceVerification.Recommended
// Detect alert filter before normalising external_status.
$filter_alert = ( 'alert' === $external_status );
if ( $filter_alert ) {
// Reuse existing matches WHERE logic to narrow the candidate set.
$external_status = 'matches';
}
if ( ! in_array( $external_status, self::EXTERNAL_STATUSES, true ) ) {
$external_status = '';
}
@ -485,8 +541,6 @@ class MediaListTable extends \WP_List_Table {
$where = implode( ' AND ', $where_parts );
// phpcs:disable WordPress.DB.DirectDatabaseQuery,WordPress.DB.PreparedSQL.InterpolatedNotPrepared,WordPress.DB.PreparedSQL.NotPrepared
if ( 'usage_count' === $orderby ) {
$order_sql = "(SELECT COUNT(*) FROM {$wpdb->prefix}mra_media_usage mu WHERE mu.attachment_id = i.attachment_id) {$order}";
} elseif ( 'external_scanned_at' === $orderby ) {
@ -496,30 +550,110 @@ class MediaListTable extends \WP_List_Table {
$order_sql = "i.attachment_id {$order}";
}
$count_sql = "SELECT COUNT(*) FROM {$wpdb->prefix}mra_media_index i WHERE {$where}";
$items_sql = "SELECT i.*,
$select_sql = "SELECT i.*,
(SELECT COUNT(*) FROM {$wpdb->prefix}mra_media_usage mu WHERE mu.attachment_id = i.attachment_id) AS usage_count,
(SELECT COALESCE(SUM(er.match_count), 0) FROM {$wpdb->prefix}mra_external_results er WHERE er.attachment_id = i.attachment_id) AS total_match_count
FROM {$wpdb->prefix}mra_media_index i
WHERE {$where}
ORDER BY {$order_sql}
LIMIT %d OFFSET %d";
FROM {$wpdb->prefix}mra_media_index i";
$count_prepared = empty( $prepare_args )
? $count_sql
: $wpdb->prepare( $count_sql, ...$prepare_args );
$rows = array();
$total = 0;
$items_prepared = $wpdb->prepare(
$items_sql,
...array_merge( $prepare_args, array( $per_page, $offset ) )
);
if ( $filter_alert ) {
// Fetch ALL matching rows (no LIMIT), then PHP-filter by alert domains.
$all_sql = $select_sql . " WHERE {$where} ORDER BY {$order_sql}";
$total = (int) $wpdb->get_var( $count_prepared );
$rows = $wpdb->get_results( $items_prepared, ARRAY_A );
// phpcs:disable WordPress.DB.DirectDatabaseQuery,WordPress.DB.PreparedSQL.InterpolatedNotPrepared,WordPress.DB.PreparedSQL.NotPrepared
$all_prepared = empty( $prepare_args )
? $all_sql
: $wpdb->prepare( $all_sql, ...$prepare_args );
// phpcs:enable WordPress.DB.DirectDatabaseQuery,WordPress.DB.PreparedSQL.InterpolatedNotPrepared,WordPress.DB.PreparedSQL.NotPrepared
$all_rows = $wpdb->get_results( $all_prepared, ARRAY_A );
// phpcs:enable WordPress.DB.DirectDatabaseQuery,WordPress.DB.PreparedSQL.InterpolatedNotPrepared,WordPress.DB.PreparedSQL.NotPrepared
if ( ! $rows ) {
if ( ! is_array( $all_rows ) ) {
$all_rows = array();
}
// Bulk-fetch top_domains for all candidate rows.
$all_ids = array_map( 'intval', array_column( $all_rows, 'attachment_id' ) );
$top_domains_by_id = self::fetch_top_domains( $all_ids );
// PHP-filter: keep only rows that have at least one alert domain.
$filtered = array();
foreach ( $all_rows as $r ) {
if ( ! is_array( $r ) ) {
continue;
}
$aid_val = $r['attachment_id'] ?? null;
$aid = is_numeric( $aid_val ) ? (int) $aid_val : 0;
if ( HostnameFilter::has_alert_domains( $top_domains_by_id[ $aid ] ?? array() ) ) {
$filtered[] = $r;
}
}
$total = count( $filtered );
$rows = array_slice( $filtered, $offset, $per_page );
// Attach top_domains_data for the page slice.
foreach ( $rows as &$row ) {
$rid_val = $row['attachment_id'] ?? null;
$rid = is_numeric( $rid_val ) ? (int) $rid_val : 0;
$row['top_domains_data'] = $top_domains_by_id[ $rid ] ?? array();
}
unset( $row );
} else {
$count_sql = "SELECT COUNT(*) FROM {$wpdb->prefix}mra_media_index i WHERE {$where}";
$items_sql = $select_sql . " WHERE {$where} ORDER BY {$order_sql} LIMIT %d OFFSET %d";
// phpcs:disable WordPress.DB.DirectDatabaseQuery,WordPress.DB.PreparedSQL.InterpolatedNotPrepared,WordPress.DB.PreparedSQL.NotPrepared
$count_prepared = empty( $prepare_args )
? $count_sql
: $wpdb->prepare( $count_sql, ...$prepare_args );
$items_prepared = $wpdb->prepare(
$items_sql,
...array_merge( $prepare_args, array( $per_page, $offset ) )
);
$total = (int) $wpdb->get_var( $count_prepared );
$rows = $wpdb->get_results( $items_prepared, ARRAY_A );
// phpcs:enable WordPress.DB.DirectDatabaseQuery,WordPress.DB.PreparedSQL.InterpolatedNotPrepared,WordPress.DB.PreparedSQL.NotPrepared
if ( ! is_array( $rows ) ) {
$rows = array();
}
// Bulk-fetch top_domains only for rows with external_status = 'matches'.
$matches_ids = array();
foreach ( $rows as $r ) {
if ( ! is_array( $r ) ) {
continue;
}
$es_val = $r['external_status'] ?? null;
if ( 'matches' === $es_val ) {
$aid_val = $r['attachment_id'] ?? null;
if ( is_numeric( $aid_val ) ) {
$matches_ids[] = (int) $aid_val;
}
}
}
$top_domains_by_id = empty( $matches_ids )
? array()
: self::fetch_top_domains( $matches_ids );
foreach ( $rows as &$row ) {
if ( ! is_array( $row ) ) {
continue;
}
$rid_val = $row['attachment_id'] ?? null;
$rid = is_numeric( $rid_val ) ? (int) $rid_val : 0;
$row['top_domains_data'] = $top_domains_by_id[ $rid ] ?? array();
}
unset( $row );
}
if ( empty( $rows ) ) {
$this->items = array();
} else {
// Bulk-load usages for this page (avoid N+1).
@ -533,6 +667,9 @@ class MediaListTable extends \WP_List_Table {
}
foreach ( $rows as &$row ) {
if ( ! is_array( $row ) ) {
continue;
}
$rid_val = $row['attachment_id'] ?? null;
$rid = is_numeric( $rid_val ) ? (int) $rid_val : 0;
$row['usages_data'] = $usage_by[ $rid ] ?? array();
@ -562,6 +699,74 @@ class MediaListTable extends \WP_List_Table {
// Public query helpers
// -------------------------------------------------------------------------
/**
* Fetches and merges top_domains for a set of attachment IDs.
* Multiple providers for the same attachment have their domain counts summed.
*
* @param array<int, int> $attachment_ids Attachment IDs to look up.
*
* @return array<int, array<string, int>> attachment_id domain count
*/
public static function fetch_top_domains( array $attachment_ids ): array {
global $wpdb;
if ( empty( $attachment_ids ) ) {
return array();
}
$placeholders = implode( ',', array_fill( 0, count( $attachment_ids ), '%d' ) );
// phpcs:disable WordPress.DB.DirectDatabaseQuery,WordPress.DB.PreparedSQL.InterpolatedNotPrepared,WordPress.DB.PreparedSQLPlaceholders.UnfinishedPrepare
$rows = $wpdb->get_results(
$wpdb->prepare(
"SELECT attachment_id, top_domains FROM {$wpdb->prefix}mra_external_results WHERE attachment_id IN ({$placeholders})",
...$attachment_ids
),
ARRAY_A
);
// phpcs:enable WordPress.DB.DirectDatabaseQuery,WordPress.DB.PreparedSQL.InterpolatedNotPrepared,WordPress.DB.PreparedSQLPlaceholders.UnfinishedPrepare
if ( ! is_array( $rows ) ) {
return array();
}
$result = array();
foreach ( $rows as $row ) {
if ( ! is_array( $row ) ) {
continue;
}
$aid_val = $row['attachment_id'] ?? null;
$aid = is_numeric( $aid_val ) ? (int) $aid_val : 0;
if ( $aid <= 0 ) {
continue;
}
$td_val = $row['top_domains'] ?? null;
$td_raw = is_string( $td_val ) ? json_decode( $td_val, true ) : null;
$domains = is_array( $td_raw ) ? $td_raw : array();
if ( ! isset( $result[ $aid ] ) ) {
$result[ $aid ] = array();
}
foreach ( $domains as $domain => $count ) {
if ( ! is_string( $domain ) ) {
continue;
}
$c = is_numeric( $count ) ? (int) $count : 0;
if ( isset( $result[ $aid ][ $domain ] ) ) {
$result[ $aid ][ $domain ] += $c;
} else {
$result[ $aid ][ $domain ] = $c;
}
}
}
return $result;
}
/**
* Fetches usage rows for a set of attachment IDs in one query.
*