| 1 |
<?php |
| 2 |
/** |
| 3 |
* Statistics class file. |
| 4 |
* |
| 5 |
* Collects and stores ActivityPub statistics for monthly/annual reports. |
| 6 |
* |
| 7 |
* @package Activitypub |
| 8 |
*/ |
| 9 |
|
| 10 |
namespace Activitypub; |
| 11 |
|
| 12 |
use Activitypub\Collection\Actors; |
| 13 |
use Activitypub\Collection\Followers; |
| 14 |
use Activitypub\Collection\Outbox; |
| 15 |
use Activitypub\Comment; |
| 16 |
|
| 17 |
/** |
| 18 |
* Statistics class. |
| 19 |
* |
| 20 |
* Handles collection and storage of ActivityPub statistics. |
| 21 |
*/ |
| 22 |
class Statistics { |
| 23 |
|
| 24 |
/** |
| 25 |
* Option prefix for statistics storage. |
| 26 |
* |
| 27 |
* @var string |
| 28 |
*/ |
| 29 |
const OPTION_PREFIX = 'activitypub_stats_'; |
| 30 |
|
| 31 |
/** |
| 32 |
* Get the start and end date strings for a given month. |
| 33 |
* |
| 34 |
* @param int $year The year. |
| 35 |
* @param int $month The month (1-12). |
| 36 |
* |
| 37 |
* @return array { start: string, end: string } in 'Y-m-d H:i:s' format. |
| 38 |
*/ |
| 39 |
public static function get_month_date_range( $year, $month ) { |
| 40 |
$last_day = (int) \gmdate( 't', \gmmktime( 0, 0, 0, $month, 1, $year ) ); |
| 41 |
|
| 42 |
return array( |
| 43 |
'start' => \sprintf( '%d-%02d-01 00:00:00', $year, $month ), |
| 44 |
'end' => \sprintf( '%d-%02d-%02d 23:59:59', $year, $month, $last_day ), |
| 45 |
); |
| 46 |
} |
| 47 |
|
| 48 |
/** |
| 49 |
* Get the option name for monthly stats. |
| 50 |
* |
| 51 |
* @param int $user_id The user ID. |
| 52 |
* @param int $year The year. |
| 53 |
* @param int $month The month. |
| 54 |
* |
| 55 |
* @return string The option name. |
| 56 |
*/ |
| 57 |
public static function get_monthly_option_name( $user_id, $year, $month ) { |
| 58 |
return \sprintf( '%s%d_%d_%02d', self::OPTION_PREFIX, $user_id, $year, $month ); |
| 59 |
} |
| 60 |
|
| 61 |
/** |
| 62 |
* Get the option name for annual stats. |
| 63 |
* |
| 64 |
* @param int $user_id The user ID. |
| 65 |
* @param int $year The year. |
| 66 |
* |
| 67 |
* @return string The option name. |
| 68 |
*/ |
| 69 |
public static function get_annual_option_name( $user_id, $year ) { |
| 70 |
return \sprintf( '%s%d_%d_annual', self::OPTION_PREFIX, $user_id, $year ); |
| 71 |
} |
| 72 |
|
| 73 |
/** |
| 74 |
* Get monthly statistics. |
| 75 |
* |
| 76 |
* @param int $user_id The user ID. |
| 77 |
* @param int $year The year. |
| 78 |
* @param int $month The month. |
| 79 |
* |
| 80 |
* @return array|false The monthly stats array or false if not found. |
| 81 |
*/ |
| 82 |
public static function get_monthly_stats( $user_id, $year, $month ) { |
| 83 |
return \get_option( self::get_monthly_option_name( $user_id, $year, $month ), false ); |
| 84 |
} |
| 85 |
|
| 86 |
/** |
| 87 |
* Get annual summary statistics. |
| 88 |
* |
| 89 |
* @param int $user_id The user ID. |
| 90 |
* @param int $year The year. |
| 91 |
* |
| 92 |
* @return array|false The annual stats array or false if not found. |
| 93 |
*/ |
| 94 |
public static function get_annual_summary( $user_id, $year ) { |
| 95 |
return \get_option( self::get_annual_option_name( $user_id, $year ), false ); |
| 96 |
} |
| 97 |
|
| 98 |
/** |
| 99 |
* Save monthly statistics. |
| 100 |
* |
| 101 |
* @param int $user_id The user ID. |
| 102 |
* @param int $year The year. |
| 103 |
* @param int $month The month. |
| 104 |
* @param array $stats The stats array. |
| 105 |
* |
| 106 |
* @return bool True on success, false on failure. |
| 107 |
*/ |
| 108 |
public static function save_monthly_stats( $user_id, $year, $month, $stats ) { |
| 109 |
// Invalidate the REST API transient cache so fresh data is served. |
| 110 |
\delete_transient( 'activitypub_stats_' . $user_id ); |
| 111 |
|
| 112 |
return \update_option( self::get_monthly_option_name( $user_id, $year, $month ), $stats, false ); |
| 113 |
} |
| 114 |
|
| 115 |
/** |
| 116 |
* Save annual summary statistics. |
| 117 |
* |
| 118 |
* @param int $user_id The user ID. |
| 119 |
* @param int $year The year. |
| 120 |
* @param array $stats The stats array. |
| 121 |
* |
| 122 |
* @return bool True on success, false on failure. |
| 123 |
*/ |
| 124 |
public static function save_annual_summary( $user_id, $year, $stats ) { |
| 125 |
return \update_option( self::get_annual_option_name( $user_id, $year ), $stats, false ); |
| 126 |
} |
| 127 |
|
| 128 |
/** |
| 129 |
* Collect monthly statistics for a user. |
| 130 |
* |
| 131 |
* @param int $user_id The user ID. |
| 132 |
* @param int $year The year. |
| 133 |
* @param int $month The month. |
| 134 |
* |
| 135 |
* @return array The collected stats. |
| 136 |
*/ |
| 137 |
public static function collect_monthly_stats( $user_id, $year, $month ) { |
| 138 |
$month = (int) $month; |
| 139 |
$year = (int) $year; |
| 140 |
|
| 141 |
if ( $month < 1 || $month > 12 ) { |
| 142 |
return array(); |
| 143 |
} |
| 144 |
|
| 145 |
$range = self::get_month_date_range( $year, $month ); |
| 146 |
$start = $range['start']; |
| 147 |
$end = $range['end']; |
| 148 |
|
| 149 |
// Count new followers gained this month (by post_date in followers table). |
| 150 |
$followers_count = Followers::count_in_range( $user_id, $start, $end ); |
| 151 |
|
| 152 |
$stats = array( |
| 153 |
'posts_count' => self::count_federated_posts_in_range( $user_id, $start, $end ), |
| 154 |
'followers_count' => $followers_count, |
| 155 |
'followers_total' => self::get_follower_count( $user_id ), |
| 156 |
'top_posts' => self::get_top_posts( $user_id, $start, $end, 5 ), |
| 157 |
'top_multiplicator' => self::get_top_multiplicator( $user_id, $start, $end ), |
| 158 |
'collected_at' => \gmdate( 'Y-m-d H:i:s' ), |
| 159 |
); |
| 160 |
|
| 161 |
// Add counts for each comment type dynamically. |
| 162 |
foreach ( \array_keys( self::get_comment_types_for_stats() ) as $type ) { |
| 163 |
$stats[ $type . '_count' ] = self::count_engagement_in_range( $user_id, $start, $end, $type ); |
| 164 |
} |
| 165 |
|
| 166 |
self::save_monthly_stats( $user_id, $year, $month, $stats ); |
| 167 |
|
| 168 |
return $stats; |
| 169 |
} |
| 170 |
|
| 171 |
/** |
| 172 |
* Compile annual summary from monthly stats. |
| 173 |
* |
| 174 |
* @param int $user_id The user ID. |
| 175 |
* @param int $year The year. |
| 176 |
* |
| 177 |
* @return array The annual summary. |
| 178 |
*/ |
| 179 |
public static function compile_annual_summary( $user_id, $year ) { |
| 180 |
// Initialize totals dynamically based on registered comment types. |
| 181 |
$comment_types = \array_keys( self::get_comment_types_for_stats() ); |
| 182 |
$totals = array( 'posts_count' => 0 ); |
| 183 |
foreach ( $comment_types as $type ) { |
| 184 |
$totals[ $type . '_count' ] = 0; |
| 185 |
} |
| 186 |
|
| 187 |
$most_active_month = null; |
| 188 |
$most_active_engagement = 0; |
| 189 |
$first_month_stats = null; |
| 190 |
$last_month_stats = null; |
| 191 |
$all_multiplicators = array(); |
| 192 |
|
| 193 |
for ( $month = 1; $month <= 12; $month++ ) { |
| 194 |
$stats = self::get_monthly_stats( $user_id, $year, $month ); |
| 195 |
|
| 196 |
if ( ! $stats ) { |
| 197 |
continue; |
| 198 |
} |
| 199 |
|
| 200 |
// Track first and last months with data. |
| 201 |
if ( ! $first_month_stats ) { |
| 202 |
$first_month_stats = $stats; |
| 203 |
} |
| 204 |
$last_month_stats = $stats; |
| 205 |
|
| 206 |
// Sum totals dynamically. |
| 207 |
$totals['posts_count'] += $stats['posts_count'] ?? 0; |
| 208 |
foreach ( $comment_types as $type ) { |
| 209 |
$key = $type . '_count'; |
| 210 |
$totals[ $key ] += $stats[ $key ] ?? 0; |
| 211 |
} |
| 212 |
|
| 213 |
// Calculate engagement for this month (sum of all comment type counts). |
| 214 |
$engagement = 0; |
| 215 |
foreach ( $comment_types as $type ) { |
| 216 |
$engagement += $stats[ $type . '_count' ] ?? 0; |
| 217 |
} |
| 218 |
|
| 219 |
if ( $engagement > $most_active_engagement ) { |
| 220 |
$most_active_engagement = $engagement; |
| 221 |
$most_active_month = $month; |
| 222 |
} |
| 223 |
|
| 224 |
// Aggregate multiplicators. |
| 225 |
if ( ! empty( $stats['top_multiplicator'] ) && ! empty( $stats['top_multiplicator']['url'] ) ) { |
| 226 |
$url = $stats['top_multiplicator']['url']; |
| 227 |
if ( ! isset( $all_multiplicators[ $url ] ) ) { |
| 228 |
$all_multiplicators[ $url ] = array( |
| 229 |
'name' => $stats['top_multiplicator']['name'], |
| 230 |
'url' => $url, |
| 231 |
'count' => 0, |
| 232 |
); |
| 233 |
} |
| 234 |
$all_multiplicators[ $url ]['count'] += $stats['top_multiplicator']['count'] ?? 0; |
| 235 |
} |
| 236 |
} |
| 237 |
|
| 238 |
// Find top multiplicator for the year. |
| 239 |
$top_multiplicator = null; |
| 240 |
if ( ! empty( $all_multiplicators ) ) { |
| 241 |
\usort( |
| 242 |
$all_multiplicators, |
| 243 |
function ( $a, $b ) { |
| 244 |
return $b['count'] - $a['count']; |
| 245 |
} |
| 246 |
); |
| 247 |
$top_multiplicator = \reset( $all_multiplicators ); |
| 248 |
} |
| 249 |
|
| 250 |
// Build summary with dynamic comment type counts. |
| 251 |
// Calculate followers_start: total at start of first month (total minus gained that month). |
| 252 |
// Monthly stats store: followers_count (gained this month), followers_total (total at end of month). |
| 253 |
$followers_start = 0; |
| 254 |
if ( $first_month_stats ) { |
| 255 |
$followers_start = ( $first_month_stats['followers_total'] ?? 0 ) - ( $first_month_stats['followers_count'] ?? 0 ); |
| 256 |
} |
| 257 |
|
| 258 |
// Get top posts for the full year. |
| 259 |
$year_start = \sprintf( '%d-01-01 00:00:00', $year ); |
| 260 |
$year_end = \sprintf( '%d-12-31 23:59:59', $year ); |
| 261 |
|
| 262 |
$summary = array( |
| 263 |
'posts_count' => $totals['posts_count'], |
| 264 |
'most_active_month' => $most_active_month, |
| 265 |
'followers_start' => $followers_start, |
| 266 |
'followers_end' => $last_month_stats ? ( $last_month_stats['followers_total'] ?? 0 ) : self::get_follower_count( $user_id ), |
| 267 |
'followers_net_change' => 0, |
| 268 |
'top_multiplicator' => $top_multiplicator, |
| 269 |
'top_posts' => self::get_top_posts( $user_id, $year_start, $year_end, 5 ), |
| 270 |
'compiled_at' => \gmdate( 'Y-m-d H:i:s' ), |
| 271 |
); |
| 272 |
|
| 273 |
// Add comment type totals dynamically. |
| 274 |
foreach ( $comment_types as $type ) { |
| 275 |
$summary[ $type . '_count' ] = $totals[ $type . '_count' ]; |
| 276 |
} |
| 277 |
|
| 278 |
$summary['followers_net_change'] = $summary['followers_end'] - $summary['followers_start']; |
| 279 |
|
| 280 |
self::save_annual_summary( $user_id, $year, $summary ); |
| 281 |
|
| 282 |
return $summary; |
| 283 |
} |
| 284 |
|
| 285 |
/** |
| 286 |
* Count published posts in a date range. |
| 287 |
* |
| 288 |
* Counts published posts in ActivityPub-enabled post types. |
| 289 |
* |
| 290 |
* @param int $user_id The user ID. |
| 291 |
* @param string $start Start date (Y-m-d H:i:s). |
| 292 |
* @param string $end End date (Y-m-d H:i:s). |
| 293 |
* |
| 294 |
* @return int The post count. |
| 295 |
*/ |
| 296 |
public static function count_federated_posts_in_range( $user_id, $start, $end ) { |
| 297 |
global $wpdb; |
| 298 |
|
| 299 |
$post_subquery = self::get_post_ids_subquery( $user_id, $start, $end ); |
| 300 |
|
| 301 |
// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery |
| 302 |
// phpcs:disable WordPress.DB.DirectDatabaseQuery.NoCaching |
| 303 |
// phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared |
| 304 |
// phpcs:disable WordPress.DB.PreparedSQL.NotPrepared |
| 305 |
$count = $wpdb->get_var( |
| 306 |
"SELECT COUNT(*) FROM ({$post_subquery}) AS posts" |
| 307 |
); |
| 308 |
// phpcs:enable |
| 309 |
|
| 310 |
return (int) $count; |
| 311 |
} |
| 312 |
|
| 313 |
/** |
| 314 |
* Count engagement (likes, reposts, comments, quotes) in a date range. |
| 315 |
* |
| 316 |
* @param int $user_id The user ID. |
| 317 |
* @param string $start Start date (Y-m-d H:i:s). |
| 318 |
* @param string $end End date (Y-m-d H:i:s). |
| 319 |
* @param string|null $type Optional. The engagement type ('like', 'repost', 'comment', 'quote'). |
| 320 |
* |
| 321 |
* @return int The engagement count. |
| 322 |
*/ |
| 323 |
public static function count_engagement_in_range( $user_id, $start, $end, $type = null ) { |
| 324 |
global $wpdb; |
| 325 |
|
| 326 |
// Use a subquery to avoid loading all post IDs into memory. |
| 327 |
$post_subquery = self::get_post_ids_subquery( $user_id ); |
| 328 |
|
| 329 |
$type_clause = ''; |
| 330 |
if ( $type ) { |
| 331 |
$type_clause = $wpdb->prepare( ' AND c.comment_type = %s', $type ); |
| 332 |
} else { |
| 333 |
// Get all comment types tracked in statistics (includes federated comments via filter). |
| 334 |
$comment_types = \array_keys( self::get_comment_types_for_stats() ); |
| 335 |
if ( ! empty( $comment_types ) ) { |
| 336 |
$placeholders_types = \implode( ', ', \array_fill( 0, \count( $comment_types ), '%s' ) ); |
| 337 |
// phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared, WordPress.DB.PreparedSQLPlaceholders.UnfinishedPrepare |
| 338 |
$type_clause = $wpdb->prepare( " AND c.comment_type IN ($placeholders_types)", $comment_types ); |
| 339 |
} |
| 340 |
} |
| 341 |
|
| 342 |
// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery |
| 343 |
// phpcs:disable WordPress.DB.DirectDatabaseQuery.NoCaching |
| 344 |
// phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared |
| 345 |
$count = $wpdb->get_var( |
| 346 |
$wpdb->prepare( |
| 347 |
"SELECT COUNT(DISTINCT c.comment_ID) FROM {$wpdb->comments} c |
| 348 |
INNER JOIN {$wpdb->commentmeta} cm ON c.comment_ID = cm.comment_id |
| 349 |
WHERE c.comment_post_ID IN ({$post_subquery}) |
| 350 |
AND cm.meta_key = 'protocol' |
| 351 |
AND cm.meta_value = 'activitypub' |
| 352 |
AND c.comment_date_gmt >= %s |
| 353 |
AND c.comment_date_gmt <= %s |
| 354 |
{$type_clause}", |
| 355 |
$start, |
| 356 |
$end |
| 357 |
) |
| 358 |
); |
| 359 |
// phpcs:enable |
| 360 |
|
| 361 |
return (int) $count; |
| 362 |
} |
| 363 |
|
| 364 |
/** |
| 365 |
* Get top performing posts in a date range. |
| 366 |
* |
| 367 |
* @param int $user_id The user ID. |
| 368 |
* @param string $start Start date (Y-m-d H:i:s). |
| 369 |
* @param string $end End date (Y-m-d H:i:s). |
| 370 |
* @param int $limit Maximum number of posts to return. |
| 371 |
* |
| 372 |
* @return array Array of top posts with engagement data. |
| 373 |
*/ |
| 374 |
public static function get_top_posts( $user_id, $start, $end, $limit = 5 ) { |
| 375 |
global $wpdb; |
| 376 |
|
| 377 |
// Use a subquery with date range to only consider posts published in the period. |
| 378 |
$post_subquery = self::get_post_ids_subquery( $user_id, $start, $end ); |
| 379 |
|
| 380 |
// Use the same comment type source as all other statistics methods. |
| 381 |
$comment_types = \array_keys( self::get_comment_types_for_stats() ); |
| 382 |
if ( empty( $comment_types ) ) { |
| 383 |
return array(); |
| 384 |
} |
| 385 |
|
| 386 |
$placeholders_types = \implode( ', ', \array_fill( 0, \count( $comment_types ), '%s' ) ); |
| 387 |
|
| 388 |
// phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared, WordPress.DB.PreparedSQLPlaceholders.UnfinishedPrepare |
| 389 |
$type_clause = $wpdb->prepare( "AND c.comment_type IN ({$placeholders_types})", $comment_types ); |
| 390 |
|
| 391 |
// Get engagement counts per post (only engagement within the date range). |
| 392 |
// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery |
| 393 |
// phpcs:disable WordPress.DB.DirectDatabaseQuery.NoCaching |
| 394 |
// phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared |
| 395 |
$results = $wpdb->get_results( |
| 396 |
$wpdb->prepare( |
| 397 |
"SELECT c.comment_post_ID as post_id, COUNT(c.comment_ID) as engagement_count |
| 398 |
FROM {$wpdb->comments} c |
| 399 |
INNER JOIN {$wpdb->commentmeta} cm ON c.comment_ID = cm.comment_id |
| 400 |
WHERE c.comment_post_ID IN ({$post_subquery}) |
| 401 |
AND cm.meta_key = 'protocol' |
| 402 |
AND cm.meta_value = 'activitypub' |
| 403 |
{$type_clause} |
| 404 |
AND c.comment_date_gmt >= %s |
| 405 |
AND c.comment_date_gmt <= %s |
| 406 |
GROUP BY c.comment_post_ID |
| 407 |
ORDER BY engagement_count DESC |
| 408 |
LIMIT %d", |
| 409 |
$start, |
| 410 |
$end, |
| 411 |
$limit |
| 412 |
), |
| 413 |
ARRAY_A |
| 414 |
); |
| 415 |
// phpcs:enable |
| 416 |
|
| 417 |
$top_posts = array(); |
| 418 |
foreach ( $results as $result ) { |
| 419 |
$post = \get_post( $result['post_id'] ); |
| 420 |
if ( $post ) { |
| 421 |
$top_posts[] = array( |
| 422 |
'post_id' => $result['post_id'], |
| 423 |
'title' => \html_entity_decode( \get_the_title( $post ), ENT_QUOTES, 'UTF-8' ), |
| 424 |
'url' => \get_permalink( $post ), |
| 425 |
'edit_url' => \get_edit_post_link( $post, 'raw' ), |
| 426 |
'engagement_count' => (int) $result['engagement_count'], |
| 427 |
); |
| 428 |
} |
| 429 |
} |
| 430 |
|
| 431 |
return $top_posts; |
| 432 |
} |
| 433 |
|
| 434 |
/** |
| 435 |
* Get the top multiplicator (actor who boosted content the most) in a date range. |
| 436 |
* |
| 437 |
* @param int $user_id The user ID. |
| 438 |
* @param string $start Start date (Y-m-d H:i:s). |
| 439 |
* @param string $end End date (Y-m-d H:i:s). |
| 440 |
* |
| 441 |
* @return array|null Actor data or null if none found. |
| 442 |
*/ |
| 443 |
public static function get_top_multiplicator( $user_id, $start, $end ) { |
| 444 |
global $wpdb; |
| 445 |
|
| 446 |
// Use a subquery to avoid loading all post IDs into memory. |
| 447 |
$post_subquery = self::get_post_ids_subquery( $user_id ); |
| 448 |
|
| 449 |
// Get actor who boosted the most. |
| 450 |
// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery |
| 451 |
// phpcs:disable WordPress.DB.DirectDatabaseQuery.NoCaching |
| 452 |
// phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared |
| 453 |
$result = $wpdb->get_row( |
| 454 |
$wpdb->prepare( |
| 455 |
"SELECT c.comment_author as name, c.comment_author_url as url, COUNT(c.comment_ID) as boost_count |
| 456 |
FROM {$wpdb->comments} c |
| 457 |
INNER JOIN {$wpdb->commentmeta} cm ON c.comment_ID = cm.comment_id |
| 458 |
WHERE c.comment_post_ID IN ({$post_subquery}) |
| 459 |
AND cm.meta_key = 'protocol' |
| 460 |
AND cm.meta_value = 'activitypub' |
| 461 |
AND c.comment_type = 'repost' |
| 462 |
AND c.comment_date_gmt >= %s |
| 463 |
AND c.comment_date_gmt <= %s |
| 464 |
GROUP BY c.comment_author_url |
| 465 |
ORDER BY boost_count DESC |
| 466 |
LIMIT 1", |
| 467 |
$start, |
| 468 |
$end |
| 469 |
), |
| 470 |
ARRAY_A |
| 471 |
); |
| 472 |
// phpcs:enable |
| 473 |
|
| 474 |
if ( ! $result || empty( $result['url'] ) ) { |
| 475 |
return null; |
| 476 |
} |
| 477 |
|
| 478 |
return array( |
| 479 |
'name' => $result['name'], |
| 480 |
'url' => $result['url'], |
| 481 |
'count' => (int) $result['boost_count'], |
| 482 |
); |
| 483 |
} |
| 484 |
|
| 485 |
/** |
| 486 |
* Get current follower count for a user. |
| 487 |
* |
| 488 |
* @param int $user_id The user ID. |
| 489 |
* |
| 490 |
* @return int The follower count. |
| 491 |
*/ |
| 492 |
public static function get_follower_count( $user_id ) { |
| 493 |
return Followers::count( $user_id ); |
| 494 |
} |
| 495 |
|
| 496 |
/** |
| 497 |
* Get all active user IDs that have ActivityPub enabled. |
| 498 |
* |
| 499 |
* @return int[] Array of user IDs including BLOG_USER_ID if enabled. |
| 500 |
*/ |
| 501 |
public static function get_active_user_ids() { |
| 502 |
return Actors::get_all_ids(); |
| 503 |
} |
| 504 |
|
| 505 |
/** |
| 506 |
* Get statistics for the current period. |
| 507 |
* |
| 508 |
* Always queries live data for the current period to include recent engagement. |
| 509 |
* |
| 510 |
* @param int $user_id The user ID. |
| 511 |
* @param string $period The period ('month', 'year', 'all'). |
| 512 |
* |
| 513 |
* @return array The statistics. |
| 514 |
*/ |
| 515 |
public static function get_current_stats( $user_id, $period = 'month' ) { |
| 516 |
$now = \current_time( 'timestamp' ); // phpcs:ignore WordPress.DateTime.CurrentTimeTimestamp.Requested |
| 517 |
|
| 518 |
switch ( $period ) { |
| 519 |
case 'year': |
| 520 |
$start = \gmdate( 'Y-01-01 00:00:00', $now ); |
| 521 |
$end = \gmdate( 'Y-12-31 23:59:59', $now ); |
| 522 |
break; |
| 523 |
|
| 524 |
case 'all': |
| 525 |
$start = '1970-01-01 00:00:00'; |
| 526 |
$end = \gmdate( 'Y-m-d 23:59:59', $now ); |
| 527 |
break; |
| 528 |
|
| 529 |
case 'month': |
| 530 |
default: |
| 531 |
$start = \gmdate( 'Y-m-01 00:00:00', $now ); |
| 532 |
$end = \gmdate( 'Y-m-t 23:59:59', $now ); |
| 533 |
break; |
| 534 |
} |
| 535 |
|
| 536 |
$stats = array( |
| 537 |
'posts_count' => self::count_federated_posts_in_range( $user_id, $start, $end ), |
| 538 |
'followers_total' => self::get_follower_count( $user_id ), |
| 539 |
'top_posts' => self::get_top_posts( $user_id, $start, $end, 3 ), |
| 540 |
'top_multiplicator' => self::get_top_multiplicator( $user_id, $start, $end ), |
| 541 |
'period' => $period, |
| 542 |
'start' => $start, |
| 543 |
'end' => $end, |
| 544 |
); |
| 545 |
|
| 546 |
// Add counts for each comment type dynamically. |
| 547 |
foreach ( \array_keys( self::get_comment_types_for_stats() ) as $type ) { |
| 548 |
$stats[ $type . '_count' ] = self::count_engagement_in_range( $user_id, $start, $end, $type ); |
| 549 |
} |
| 550 |
|
| 551 |
return $stats; |
| 552 |
} |
| 553 |
|
| 554 |
/** |
| 555 |
* Get rolling monthly breakdown (last X months). |
| 556 |
* |
| 557 |
* Returns stats for the last X months, crossing year boundaries as needed. |
| 558 |
* |
| 559 |
* @param int $user_id The user ID. |
| 560 |
* @param int $num_months Optional. Number of months to return. Defaults to 12. |
| 561 |
* |
| 562 |
* @return array Array of monthly stats ordered chronologically. |
| 563 |
*/ |
| 564 |
public static function get_rolling_monthly_breakdown( $user_id, $num_months = 12 ) { |
| 565 |
$now = \current_time( 'timestamp' ); // phpcs:ignore WordPress.DateTime.CurrentTimeTimestamp.Requested |
| 566 |
$months = array(); |
| 567 |
$comment_types = \array_keys( self::get_comment_types_for_stats() ); |
| 568 |
|
| 569 |
// Start from (num_months - 1) months ago and go to current month. |
| 570 |
for ( $i = $num_months - 1; $i >= 0; $i-- ) { |
| 571 |
$timestamp = \strtotime( "-{$i} months", $now ); |
| 572 |
$year = (int) \gmdate( 'Y', $timestamp ); |
| 573 |
$month = (int) \gmdate( 'n', $timestamp ); |
| 574 |
|
| 575 |
$month_data = self::get_month_data( $user_id, $year, $month, $comment_types ); |
| 576 |
$month_data['year'] = $year; |
| 577 |
$month_data['month'] = $month; |
| 578 |
|
| 579 |
$months[] = $month_data; |
| 580 |
} |
| 581 |
|
| 582 |
return $months; |
| 583 |
} |
| 584 |
|
| 585 |
/** |
| 586 |
* Get data for a single month. |
| 587 |
* |
| 588 |
* @param int $user_id The user ID. |
| 589 |
* @param int $year The year. |
| 590 |
* @param int $month The month. |
| 591 |
* @param array $comment_types Array of comment type slugs. |
| 592 |
* |
| 593 |
* @return array Month data with posts_count, engagement, and type counts. |
| 594 |
*/ |
| 595 |
private static function get_month_data( $user_id, $year, $month, $comment_types ) { |
| 596 |
$now = \current_time( 'timestamp' ); // phpcs:ignore WordPress.DateTime.CurrentTimeTimestamp.Requested |
| 597 |
$current_year = (int) \gmdate( 'Y', $now ); |
| 598 |
$current_month = (int) \gmdate( 'n', $now ); |
| 599 |
|
| 600 |
// Always query live for the current month to include recent engagement. |
| 601 |
$is_current_month = ( $year === $current_year && $month === $current_month ); |
| 602 |
|
| 603 |
// Check for stored monthly stats first (but not for current month). |
| 604 |
$stored_stats = $is_current_month ? false : self::get_monthly_stats( $user_id, $year, $month ); |
| 605 |
|
| 606 |
if ( $stored_stats ) { |
| 607 |
// Use stored data. |
| 608 |
$engagement = 0; |
| 609 |
foreach ( $comment_types as $type ) { |
| 610 |
$engagement += $stored_stats[ $type . '_count' ] ?? 0; |
| 611 |
} |
| 612 |
|
| 613 |
$month_data = array( |
| 614 |
'month' => $month, |
| 615 |
'posts_count' => $stored_stats['posts_count'] ?? 0, |
| 616 |
'engagement' => $engagement, |
| 617 |
); |
| 618 |
|
| 619 |
// Add counts for each comment type from stored stats. |
| 620 |
foreach ( $comment_types as $type ) { |
| 621 |
$month_data[ $type . '_count' ] = $stored_stats[ $type . '_count' ] ?? 0; |
| 622 |
} |
| 623 |
} else { |
| 624 |
// Query live data. |
| 625 |
$range = self::get_month_date_range( $year, $month ); |
| 626 |
$start = $range['start']; |
| 627 |
$end = $range['end']; |
| 628 |
|
| 629 |
$month_data = array( |
| 630 |
'month' => $month, |
| 631 |
'posts_count' => self::count_federated_posts_in_range( $user_id, $start, $end ), |
| 632 |
'engagement' => 0, |
| 633 |
); |
| 634 |
|
| 635 |
// Query each type and sum for total engagement (avoids extra N+1 total query). |
| 636 |
foreach ( $comment_types as $type ) { |
| 637 |
$type_count = self::count_engagement_in_range( $user_id, $start, $end, $type ); |
| 638 |
$month_data[ $type . '_count' ] = $type_count; |
| 639 |
$month_data['engagement'] += $type_count; |
| 640 |
} |
| 641 |
} |
| 642 |
|
| 643 |
return $month_data; |
| 644 |
} |
| 645 |
|
| 646 |
/** |
| 647 |
* Get period-over-period comparison (current month vs previous month). |
| 648 |
* |
| 649 |
* Reuses pre-computed current stats when available to avoid duplicate queries. |
| 650 |
* Falls back to live queries for current month if no stats are provided. |
| 651 |
* |
| 652 |
* @param int $user_id The user ID. |
| 653 |
* @param array|null $current_stats Optional. Pre-computed current month stats from get_current_stats(). |
| 654 |
* |
| 655 |
* @return array Comparison data with current values and changes from previous month. |
| 656 |
*/ |
| 657 |
public static function get_period_comparison( $user_id, $current_stats = null ) { |
| 658 |
$now = \current_time( 'timestamp' ); // phpcs:ignore WordPress.DateTime.CurrentTimeTimestamp.Requested |
| 659 |
|
| 660 |
// Previous month (handles year boundary). |
| 661 |
$prev_timestamp = \strtotime( '-1 month', $now ); |
| 662 |
$prev_year = (int) \gmdate( 'Y', $prev_timestamp ); |
| 663 |
$prev_month = (int) \gmdate( 'n', $prev_timestamp ); |
| 664 |
|
| 665 |
// Check for stored stats (only for previous month - current month is always live). |
| 666 |
$prev_stats = self::get_monthly_stats( $user_id, $prev_year, $prev_month ); |
| 667 |
|
| 668 |
// Reuse pre-computed current stats or query live. |
| 669 |
if ( $current_stats ) { |
| 670 |
$current_posts = $current_stats['posts_count'] ?? 0; |
| 671 |
} else { |
| 672 |
$current_start = \gmdate( 'Y-m-01 00:00:00', $now ); |
| 673 |
$current_end = \gmdate( 'Y-m-t 23:59:59', $now ); |
| 674 |
$current_posts = self::count_federated_posts_in_range( $user_id, $current_start, $current_end ); |
| 675 |
} |
| 676 |
|
| 677 |
$current_start = $current_start ?? \gmdate( 'Y-m-01 00:00:00', $now ); |
| 678 |
$current_end = $current_end ?? \gmdate( 'Y-m-t 23:59:59', $now ); |
| 679 |
|
| 680 |
$current_followers = Followers::count_in_range( $user_id, $current_start, $current_end ); |
| 681 |
|
| 682 |
// Get previous month data (from stored stats or live query). |
| 683 |
if ( $prev_stats ) { |
| 684 |
$prev_posts = $prev_stats['posts_count'] ?? 0; |
| 685 |
$prev_followers = $prev_stats['followers_count'] ?? 0; |
| 686 |
} else { |
| 687 |
$prev_start = \gmdate( 'Y-m-01 00:00:00', $prev_timestamp ); |
| 688 |
$prev_end = \gmdate( 'Y-m-t 23:59:59', $prev_timestamp ); |
| 689 |
$prev_posts = self::count_federated_posts_in_range( $user_id, $prev_start, $prev_end ); |
| 690 |
$prev_followers = Followers::count_in_range( $user_id, $prev_start, $prev_end ); |
| 691 |
} |
| 692 |
|
| 693 |
$comparison = array( |
| 694 |
'posts' => array( |
| 695 |
'current' => $current_posts, |
| 696 |
'change' => $current_posts - $prev_posts, |
| 697 |
), |
| 698 |
'followers' => array( |
| 699 |
'current' => $current_followers, |
| 700 |
'change' => $current_followers - $prev_followers, |
| 701 |
), |
| 702 |
); |
| 703 |
|
| 704 |
// Add comparison for each comment type tracked in statistics (includes federated comments). |
| 705 |
$comment_types = \array_keys( self::get_comment_types_for_stats() ); |
| 706 |
foreach ( $comment_types as $type ) { |
| 707 |
// Reuse pre-computed current stats or query live. |
| 708 |
if ( $current_stats && isset( $current_stats[ $type . '_count' ] ) ) { |
| 709 |
$current_count = $current_stats[ $type . '_count' ]; |
| 710 |
} else { |
| 711 |
$current_count = self::count_engagement_in_range( $user_id, $current_start, $current_end, $type ); |
| 712 |
} |
| 713 |
|
| 714 |
// Use stored stats for previous month if available. |
| 715 |
if ( $prev_stats ) { |
| 716 |
$prev_count = $prev_stats[ $type . '_count' ] ?? 0; |
| 717 |
} else { |
| 718 |
$prev_count = self::count_engagement_in_range( $user_id, $prev_start, $prev_end, $type ); |
| 719 |
} |
| 720 |
|
| 721 |
$comparison[ $type ] = array( |
| 722 |
'current' => $current_count, |
| 723 |
'change' => $current_count - $prev_count, |
| 724 |
); |
| 725 |
} |
| 726 |
|
| 727 |
return $comparison; |
| 728 |
} |
| 729 |
|
| 730 |
/** |
| 731 |
* Get comment types to track in statistics. |
| 732 |
* |
| 733 |
* By default includes all registered ActivityPub comment types. |
| 734 |
* Use the 'activitypub_stats_comment_types' filter to add additional types. |
| 735 |
* |
| 736 |
* @return array Array of comment type data with slug, label, and singular. |
| 737 |
*/ |
| 738 |
public static function get_comment_types_for_stats() { |
| 739 |
$comment_types = Comment::get_comment_types(); |
| 740 |
$result = array(); |
| 741 |
|
| 742 |
foreach ( $comment_types as $slug => $type ) { |
| 743 |
$result[ $slug ] = array( |
| 744 |
'slug' => $slug, |
| 745 |
'label' => $type['label'] ?? \ucfirst( $slug ), |
| 746 |
'singular' => $type['singular'] ?? \ucfirst( $slug ), |
| 747 |
); |
| 748 |
} |
| 749 |
|
| 750 |
// Add federated comments (replies) which use the standard 'comment' type. |
| 751 |
if ( ! isset( $result['comment'] ) ) { |
| 752 |
$result['comment'] = array( |
| 753 |
'slug' => 'comment', |
| 754 |
'label' => \__( 'Comments', 'activitypub' ), |
| 755 |
'singular' => \__( 'Comment', 'activitypub' ), |
| 756 |
); |
| 757 |
} |
| 758 |
|
| 759 |
/** |
| 760 |
* Filter the comment types tracked in statistics. |
| 761 |
* |
| 762 |
* Allows adding additional comment types to be tracked |
| 763 |
* in the statistics dashboard. |
| 764 |
* |
| 765 |
* @param array $result Array of comment type data with slug, label, and singular. |
| 766 |
*/ |
| 767 |
return \apply_filters( 'activitypub_stats_comment_types', $result ); |
| 768 |
} |
| 769 |
|
| 770 |
/** |
| 771 |
* Backfill historical statistics for all active users. |
| 772 |
* |
| 773 |
* This method processes statistics in batches to avoid timeouts. |
| 774 |
* It only collects stats for completed months (not the current month). |
| 775 |
* |
| 776 |
* @param int $batch_size Optional. Number of months to process per batch. Default 12. |
| 777 |
* @param int $user_index Optional. The current user index being processed. Default 0. |
| 778 |
* @param int $year Optional. The year being processed. Default 0 (will determine earliest year). |
| 779 |
* @param int $month Optional. The month being processed. Default 1. |
| 780 |
* |
| 781 |
* @return array|null Array with batch info if more processing needed, null if complete. |
| 782 |
*/ |
| 783 |
public static function backfill_historical_stats( $batch_size = 12, $user_index = 0, $year = 0, $month = 1 ) { |
| 784 |
$user_ids = self::get_active_user_ids(); |
| 785 |
|
| 786 |
if ( empty( $user_ids ) || $user_index >= \count( $user_ids ) ) { |
| 787 |
return null; // All done. |
| 788 |
} |
| 789 |
|
| 790 |
$user_id = $user_ids[ $user_index ]; |
| 791 |
$now = \current_time( 'timestamp' ); // phpcs:ignore WordPress.DateTime.CurrentTimeTimestamp.Requested |
| 792 |
$current_year = (int) \gmdate( 'Y', $now ); |
| 793 |
$current_month = (int) \gmdate( 'n', $now ); |
| 794 |
|
| 795 |
// Determine the earliest year with data if not set. |
| 796 |
if ( 0 === $year ) { |
| 797 |
$year = self::get_earliest_data_year( $user_id ); |
| 798 |
if ( ! $year ) { |
| 799 |
// No data for this user, move to next user. |
| 800 |
return array( |
| 801 |
'batch_size' => $batch_size, |
| 802 |
'user_index' => $user_index + 1, |
| 803 |
'year' => 0, |
| 804 |
'month' => 1, |
| 805 |
); |
| 806 |
} |
| 807 |
} |
| 808 |
|
| 809 |
$months_processed = 0; |
| 810 |
|
| 811 |
// Process months for this user. |
| 812 |
while ( $months_processed < $batch_size ) { |
| 813 |
// Skip the current month - it's still in progress and should always be queried live. |
| 814 |
// Only process completed months (before the current month). |
| 815 |
if ( $year > $current_year || ( $year === $current_year && $month >= $current_month ) ) { |
| 816 |
// Move to next user. |
| 817 |
return array( |
| 818 |
'batch_size' => $batch_size, |
| 819 |
'user_index' => $user_index + 1, |
| 820 |
'year' => 0, |
| 821 |
'month' => 1, |
| 822 |
); |
| 823 |
} |
| 824 |
|
| 825 |
// Check if stats already exist for this month. |
| 826 |
$existing = self::get_monthly_stats( $user_id, $year, $month ); |
| 827 |
if ( ! $existing ) { |
| 828 |
// Collect stats for this month. |
| 829 |
self::collect_monthly_stats( $user_id, $year, $month ); |
| 830 |
} |
| 831 |
|
| 832 |
++$months_processed; |
| 833 |
++$month; |
| 834 |
|
| 835 |
// Move to next year if needed. |
| 836 |
if ( $month > 12 ) { |
| 837 |
$month = 1; |
| 838 |
++$year; |
| 839 |
} |
| 840 |
} |
| 841 |
|
| 842 |
// More months to process for this user. |
| 843 |
return array( |
| 844 |
'batch_size' => $batch_size, |
| 845 |
'user_index' => $user_index, |
| 846 |
'year' => $year, |
| 847 |
'month' => $month, |
| 848 |
); |
| 849 |
} |
| 850 |
|
| 851 |
/** |
| 852 |
* Get a prepared SQL subquery that returns post IDs for a user. |
| 853 |
* |
| 854 |
* This avoids loading all post IDs into PHP memory by using a SQL subquery |
| 855 |
* that can be embedded in other queries via IN (...). |
| 856 |
* |
| 857 |
* @param int $user_id The user ID. |
| 858 |
* @param string|null $start Optional start date (Y-m-d H:i:s). |
| 859 |
* @param string|null $end Optional end date (Y-m-d H:i:s). |
| 860 |
* |
| 861 |
* @return string Prepared SQL subquery string. |
| 862 |
*/ |
| 863 |
private static function get_post_ids_subquery( $user_id, $start = null, $end = null ) { |
| 864 |
global $wpdb; |
| 865 |
|
| 866 |
$post_types = (array) \get_option( 'activitypub_support_post_types', array( 'post' ) ); |
| 867 |
$type_placeholders = \implode( ', ', \array_fill( 0, \count( $post_types ), '%s' ) ); |
| 868 |
$params = $post_types; |
| 869 |
|
| 870 |
$author_clause = ''; |
| 871 |
if ( Actors::BLOG_USER_ID !== $user_id ) { |
| 872 |
$author_clause = ' AND post_author = %d'; |
| 873 |
$params[] = $user_id; |
| 874 |
} |
| 875 |
|
| 876 |
$date_clause = ''; |
| 877 |
if ( $start && $end ) { |
| 878 |
// Use COALESCE to fall back to post_date when post_date_gmt is empty. |
| 879 |
$date_clause = " AND COALESCE(NULLIF(post_date_gmt, '0000-00-00 00:00:00'), post_date) >= %s AND COALESCE(NULLIF(post_date_gmt, '0000-00-00 00:00:00'), post_date) <= %s"; |
| 880 |
$params[] = $start; |
| 881 |
$params[] = $end; |
| 882 |
} |
| 883 |
|
| 884 |
// phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared |
| 885 |
// phpcs:disable WordPress.DB.PreparedSQLPlaceholders.ReplacementsWrongNumber |
| 886 |
// phpcs:disable WordPress.DB.PreparedSQLPlaceholders.UnfinishedPrepare |
| 887 |
return $wpdb->prepare( |
| 888 |
"SELECT ID FROM {$wpdb->posts} WHERE post_status = 'publish' AND post_type IN ({$type_placeholders}){$author_clause}{$date_clause}", |
| 889 |
$params |
| 890 |
); |
| 891 |
// phpcs:enable |
| 892 |
} |
| 893 |
|
| 894 |
/** |
| 895 |
* Get the earliest year that has ActivityPub data for a user. |
| 896 |
* |
| 897 |
* @param int $user_id The user ID. |
| 898 |
* |
| 899 |
* @return int|null The earliest year with data, or null if no data. |
| 900 |
*/ |
| 901 |
private static function get_earliest_data_year( $user_id ) { |
| 902 |
global $wpdb; |
| 903 |
|
| 904 |
// Use a subquery to avoid loading all post IDs into memory. |
| 905 |
$post_subquery = self::get_post_ids_subquery( $user_id ); |
| 906 |
|
| 907 |
// Find earliest comment with ActivityPub protocol. |
| 908 |
// The $post_subquery is already prepared via $wpdb->prepare() in get_post_ids_subquery(), |
| 909 |
// so the outer query is safe despite not using prepare() itself. |
| 910 |
// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery |
| 911 |
// phpcs:disable WordPress.DB.DirectDatabaseQuery.NoCaching |
| 912 |
// phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared |
| 913 |
// phpcs:disable WordPress.DB.PreparedSQL.NotPrepared |
| 914 |
$earliest_date = $wpdb->get_var( |
| 915 |
"SELECT MIN(c.comment_date_gmt) FROM {$wpdb->comments} c |
| 916 |
INNER JOIN {$wpdb->commentmeta} cm ON c.comment_ID = cm.comment_id |
| 917 |
WHERE c.comment_post_ID IN ({$post_subquery}) |
| 918 |
AND cm.meta_key = 'protocol' |
| 919 |
AND cm.meta_value = 'activitypub'" |
| 920 |
); |
| 921 |
// phpcs:enable |
| 922 |
|
| 923 |
if ( ! $earliest_date ) { |
| 924 |
// No ActivityPub data, check outbox instead. |
| 925 |
$outbox_args = array( |
| 926 |
'post_type' => Outbox::POST_TYPE, |
| 927 |
'posts_per_page' => 1, |
| 928 |
'orderby' => 'date', |
| 929 |
'order' => 'ASC', |
| 930 |
// phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_meta_query |
| 931 |
'meta_query' => array( |
| 932 |
array( |
| 933 |
'key' => '_activitypub_activity_type', |
| 934 |
'value' => 'Create', |
| 935 |
), |
| 936 |
), |
| 937 |
); |
| 938 |
|
| 939 |
if ( Actors::BLOG_USER_ID !== $user_id ) { |
| 940 |
$outbox_args['author'] = $user_id; |
| 941 |
} |
| 942 |
|
| 943 |
$earliest_outbox = \get_posts( $outbox_args ); |
| 944 |
|
| 945 |
/* |
| 946 |
* Prefer `post_date_gmt`; fall back to local `post_date` when the GMT field |
| 947 |
* is empty or the MySQL zero-date (rare corruption seen on some production |
| 948 |
* sites). The zero-date is a truthy string, so it has to be detected |
| 949 |
* explicitly rather than via `?:`. |
| 950 |
* |
| 951 |
* Convert the local `post_date` fallback to GMT first so the later |
| 952 |
* `strtotime()` + `gmdate()` flow consistently derives the year in UTC. |
| 953 |
*/ |
| 954 |
$earliest_date = ''; |
| 955 |
if ( $earliest_outbox ) { |
| 956 |
$gmt = $earliest_outbox[0]->post_date_gmt; |
| 957 |
$earliest_date = ( empty( $gmt ) || '0000-00-00 00:00:00' === $gmt ) |
| 958 |
? \get_gmt_from_date( $earliest_outbox[0]->post_date ) |
| 959 |
: $gmt; |
| 960 |
} |
| 961 |
} |
| 962 |
|
| 963 |
$timestamp = \strtotime( (string) $earliest_date ); |
| 964 |
if ( false === $timestamp ) { |
| 965 |
return null; |
| 966 |
} |
| 967 |
|
| 968 |
return (int) \gmdate( 'Y', $timestamp ); |
| 969 |
} |
| 970 |
} |
| 971 |
|
