| 1 |
<?php |
| 2 |
/** |
| 3 |
* Mastodon importer file. |
| 4 |
* |
| 5 |
* @package Activitypub |
| 6 |
*/ |
| 7 |
|
| 8 |
namespace Activitypub\WP_Admin\Import; |
| 9 |
|
| 10 |
use Activitypub\Attachments; |
| 11 |
use Activitypub\Sanitize; |
| 12 |
|
| 13 |
use function Activitypub\is_activity_public; |
| 14 |
|
| 15 |
/** |
| 16 |
* Mastodon importer class. |
| 17 |
*/ |
| 18 |
class Mastodon { |
| 19 |
|
| 20 |
/** |
| 21 |
* Import file attachment ID. |
| 22 |
* |
| 23 |
* @var int |
| 24 |
*/ |
| 25 |
private static $import_id; |
| 26 |
|
| 27 |
/** |
| 28 |
* Archive folder. |
| 29 |
* |
| 30 |
* @var string |
| 31 |
*/ |
| 32 |
private static $archive; |
| 33 |
|
| 34 |
/** |
| 35 |
* Outbox file. |
| 36 |
* |
| 37 |
* @var array |
| 38 |
*/ |
| 39 |
private static $outbox; |
| 40 |
|
| 41 |
/** |
| 42 |
* Author ID. |
| 43 |
* |
| 44 |
* @var int |
| 45 |
*/ |
| 46 |
private static $author; |
| 47 |
|
| 48 |
/** |
| 49 |
* Whether to fetch attachments. |
| 50 |
* |
| 51 |
* @var bool |
| 52 |
*/ |
| 53 |
private static $fetch_attachments; |
| 54 |
|
| 55 |
/** |
| 56 |
* Dispatch |
| 57 |
*/ |
| 58 |
public static function dispatch() { |
| 59 |
// phpcs:ignore WordPress.Security.NonceVerification.Recommended |
| 60 |
$step = \absint( $_GET['step'] ?? 0 ); |
| 61 |
|
| 62 |
self::header(); |
| 63 |
|
| 64 |
switch ( $step ) { |
| 65 |
case 0: |
| 66 |
self::greet(); |
| 67 |
break; |
| 68 |
|
| 69 |
case 1: |
| 70 |
\check_admin_referer( 'import-upload' ); |
| 71 |
if ( self::handle_upload() ) { |
| 72 |
self::import_options(); |
| 73 |
} |
| 74 |
break; |
| 75 |
|
| 76 |
case 2: |
| 77 |
\check_admin_referer( 'import-mastodon' ); |
| 78 |
self::$import_id = \absint( $_POST['import_id'] ?? 0 ); |
| 79 |
self::$author = \absint( $_POST['author'] ?? \get_current_user_id() ); |
| 80 |
self::$fetch_attachments = ! empty( $_POST['fetch_attachments'] ); |
| 81 |
|
| 82 |
\set_time_limit( 0 ); |
| 83 |
self::import(); |
| 84 |
break; |
| 85 |
} |
| 86 |
|
| 87 |
self::footer(); |
| 88 |
} |
| 89 |
|
| 90 |
/** |
| 91 |
* Handle upload. |
| 92 |
* |
| 93 |
* @return bool |
| 94 |
*/ |
| 95 |
public static function handle_upload() { |
| 96 |
$error_message = \__( 'Sorry, there has been an error.', 'activitypub' ); |
| 97 |
|
| 98 |
\check_admin_referer( 'import-upload' ); |
| 99 |
|
| 100 |
if ( ! isset( $_FILES['import']['name'] ) ) { |
| 101 |
echo '<p><strong>' . \esc_html( $error_message ) . '</strong><br />'; |
| 102 |
\printf( |
| 103 |
/* translators: 1: php.ini, 2: post_max_size, 3: upload_max_filesize */ |
| 104 |
\esc_html__( 'File is empty. Please upload something more substantial. This error could also be caused by uploads being disabled in your %1$s file or by %2$s being defined as smaller than %3$s in %1$s.', 'activitypub' ), |
| 105 |
'php.ini', |
| 106 |
'post_max_size', |
| 107 |
'upload_max_filesize' |
| 108 |
); |
| 109 |
echo '</p>'; |
| 110 |
return false; |
| 111 |
} |
| 112 |
|
| 113 |
$file_info = \wp_check_filetype( \sanitize_file_name( $_FILES['import']['name'] ), array( 'zip' => 'application/zip' ) ); |
| 114 |
if ( 'application/zip' !== $file_info['type'] ) { |
| 115 |
echo '<p><strong>' . \esc_html( $error_message ) . '</strong><br />'; |
| 116 |
\esc_html_e( 'The uploaded file must be a ZIP archive. Please try again with the correct file format.', 'activitypub' ); |
| 117 |
echo '</p>'; |
| 118 |
return false; |
| 119 |
} |
| 120 |
|
| 121 |
$overrides = array( |
| 122 |
'test_form' => false, |
| 123 |
'test_type' => false, |
| 124 |
); |
| 125 |
|
| 126 |
$upload = \wp_handle_upload( $_FILES['import'], $overrides ); |
| 127 |
|
| 128 |
if ( isset( $upload['error'] ) ) { |
| 129 |
echo '<p><strong>' . \esc_html( $error_message ) . '</strong><br />'; |
| 130 |
echo \esc_html( $upload['error'] ) . '</p>'; |
| 131 |
return false; |
| 132 |
} |
| 133 |
|
| 134 |
// Construct the attachment array. |
| 135 |
$attachment = array( |
| 136 |
'post_title' => \wp_basename( $upload['file'] ), |
| 137 |
'post_content' => $upload['url'], |
| 138 |
'post_mime_type' => $upload['type'], |
| 139 |
'guid' => $upload['url'], |
| 140 |
'context' => 'import', |
| 141 |
'post_status' => 'private', |
| 142 |
); |
| 143 |
|
| 144 |
// Save the data. |
| 145 |
self::$import_id = \wp_insert_attachment( $attachment, $upload['file'] ); |
| 146 |
|
| 147 |
// Schedule a cleanup for one day from now in case of failed import or missing wp_import_cleanup() call. |
| 148 |
\wp_schedule_single_event( \time() + DAY_IN_SECONDS, 'importer_scheduled_cleanup', array( self::$import_id ) ); |
| 149 |
|
| 150 |
return true; |
| 151 |
} |
| 152 |
|
| 153 |
/** |
| 154 |
* Import options. |
| 155 |
*/ |
| 156 |
public static function import_options() { |
| 157 |
$author = 0; |
| 158 |
if ( isset( self::$outbox['orderedItems'][0] ) ) { |
| 159 |
$users = \get_users( |
| 160 |
array( |
| 161 |
'fields' => 'ID', |
| 162 |
// phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_meta_query |
| 163 |
'meta_query' => array( |
| 164 |
array( |
| 165 |
'key' => $GLOBALS['wpdb']->get_blog_prefix() . 'activitypub_also_known_as', |
| 166 |
'value' => self::$outbox['orderedItems'][0]['actor'], |
| 167 |
'compare' => 'LIKE', |
| 168 |
), |
| 169 |
), |
| 170 |
) |
| 171 |
); |
| 172 |
|
| 173 |
if ( ! empty( $users ) ) { |
| 174 |
$author = $users[0]; |
| 175 |
} |
| 176 |
} |
| 177 |
|
| 178 |
?> |
| 179 |
<form action="<?php echo \esc_url( \admin_url( 'admin.php?import=mastodon&step=2' ) ); ?>" method="post"> |
| 180 |
<?php \wp_nonce_field( 'import-mastodon' ); ?> |
| 181 |
<input type="hidden" name="import_id" value="<?php echo \esc_attr( self::$import_id ); ?>" /> |
| 182 |
<h3><?php \esc_html_e( 'Assign Author', 'activitypub' ); ?></h3> |
| 183 |
<p> |
| 184 |
<label for="author"><?php \esc_html_e( 'Author:', 'activitypub' ); ?></label> |
| 185 |
<?php |
| 186 |
\wp_dropdown_users( |
| 187 |
array( |
| 188 |
'name' => 'author', |
| 189 |
'id' => 'author', |
| 190 |
'show' => 'display_name_with_login', |
| 191 |
'selected' => $author, |
| 192 |
'capability' => 'activitypub', |
| 193 |
) |
| 194 |
); |
| 195 |
?> |
| 196 |
</p> |
| 197 |
<h3><?php \esc_html_e( 'Import Attachments', 'activitypub' ); ?></h3> |
| 198 |
<p> |
| 199 |
<input type="checkbox" value="1" name="fetch_attachments" id="import-attachments" checked /> |
| 200 |
<label for="import-attachments"><?php \esc_html_e( 'Download and import file attachments', 'activitypub' ); ?></label> |
| 201 |
</p> |
| 202 |
<p class="submit"> |
| 203 |
<input type="submit" class="button button-primary" value="<?php \esc_attr_e( 'Import', 'activitypub' ); ?>" /> |
| 204 |
</p> |
| 205 |
</form> |
| 206 |
<?php |
| 207 |
} |
| 208 |
|
| 209 |
/** |
| 210 |
* Import. |
| 211 |
*/ |
| 212 |
public static function import() { |
| 213 |
$error_message = \__( 'Sorry, there has been an error.', 'activitypub' ); |
| 214 |
$file = \get_attached_file( self::$import_id ); |
| 215 |
|
| 216 |
\WP_Filesystem(); |
| 217 |
|
| 218 |
global $wp_filesystem; |
| 219 |
$import_folder = $wp_filesystem->wp_content_dir() . 'import/'; |
| 220 |
self::$archive = $import_folder . \basename( \basename( $file, '.txt' ), '.zip' ); |
| 221 |
|
| 222 |
// Clean up working directory. |
| 223 |
if ( $wp_filesystem->is_dir( self::$archive ) ) { |
| 224 |
$wp_filesystem->delete( self::$archive, true ); |
| 225 |
} |
| 226 |
|
| 227 |
// Unzip package to working directory. |
| 228 |
\unzip_file( $file, self::$archive ); |
| 229 |
self::maybe_unwrap_archive(); |
| 230 |
|
| 231 |
if ( ! $wp_filesystem->exists( self::$archive . '/outbox.json' ) ) { |
| 232 |
echo '<p><strong>' . \esc_html( $error_message ) . '</strong><br />'; |
| 233 |
echo \esc_html__( 'The archive does not contain an Outbox file, please try again.', 'activitypub' ) . '</p>'; |
| 234 |
return; |
| 235 |
} |
| 236 |
|
| 237 |
self::$outbox = \json_decode( $wp_filesystem->get_contents( self::$archive . '/outbox.json' ), true ); |
| 238 |
|
| 239 |
\wp_suspend_cache_invalidation(); |
| 240 |
\wp_defer_term_counting( true ); |
| 241 |
\wp_defer_comment_counting( true ); |
| 242 |
|
| 243 |
/** |
| 244 |
* Fires when the Mastodon import starts. |
| 245 |
*/ |
| 246 |
\do_action( 'import_start' ); |
| 247 |
|
| 248 |
$result = self::import_posts(); |
| 249 |
|
| 250 |
\wp_suspend_cache_invalidation( false ); |
| 251 |
\wp_defer_term_counting( false ); |
| 252 |
\wp_defer_comment_counting( false ); |
| 253 |
|
| 254 |
$wp_filesystem->delete( $import_folder, true ); |
| 255 |
\wp_import_cleanup( self::$import_id ); |
| 256 |
|
| 257 |
if ( \is_wp_error( $result ) ) { |
| 258 |
echo '<p><strong>' . \esc_html( $error_message ) . '</strong><br />'; |
| 259 |
echo \esc_html( $result->get_error_message() ) . '</p>'; |
| 260 |
} else { |
| 261 |
echo '<p>'; |
| 262 |
/* translators: Home URL */ |
| 263 |
\printf( \wp_kses_post( \__( 'All done. <a href="%s">Have fun!</a>', 'activitypub' ) ), \esc_url( \admin_url() ) ); |
| 264 |
echo '</p>'; |
| 265 |
} |
| 266 |
|
| 267 |
/** |
| 268 |
* Fires when the Mastodon import ends. |
| 269 |
*/ |
| 270 |
\do_action( 'import_end' ); |
| 271 |
} |
| 272 |
|
| 273 |
/** |
| 274 |
* Process posts. |
| 275 |
* |
| 276 |
* Uses a multi-pass approach: |
| 277 |
* 1. Categorize posts into regular posts and self-replies. |
| 278 |
* 2. Import regular posts (root posts and external replies) as WordPress posts. |
| 279 |
* 3. Import self-replies as comments on their parent posts. |
| 280 |
* |
| 281 |
* @return true|\WP_Error True on success, WP_Error on failure. |
| 282 |
*/ |
| 283 |
public static function import_posts() { |
| 284 |
$skipped = array(); |
| 285 |
$imported = 0; |
| 286 |
|
| 287 |
// Pass 1: Categorize posts. |
| 288 |
$posts_to_import = array(); |
| 289 |
$self_replies = array(); |
| 290 |
|
| 291 |
foreach ( self::$outbox['orderedItems'] as $post ) { |
| 292 |
// Skip boosts. |
| 293 |
if ( 'Announce' === $post['type'] ) { |
| 294 |
continue; |
| 295 |
} |
| 296 |
|
| 297 |
if ( ! is_activity_public( $post ) ) { |
| 298 |
continue; |
| 299 |
} |
| 300 |
|
| 301 |
if ( self::is_self_reply( $post ) ) { |
| 302 |
$self_replies[] = $post; |
| 303 |
} else { |
| 304 |
// Root posts and external replies are imported as WordPress posts. |
| 305 |
$posts_to_import[] = $post; |
| 306 |
} |
| 307 |
} |
| 308 |
|
| 309 |
// Pass 2: Import regular posts as WordPress posts. |
| 310 |
$source_to_post_id = array(); |
| 311 |
foreach ( $posts_to_import as $post ) { |
| 312 |
$created = false; |
| 313 |
$result = self::import_as_post( $post, $created ); |
| 314 |
|
| 315 |
if ( \is_wp_error( $result ) ) { |
| 316 |
return $result; |
| 317 |
} |
| 318 |
|
| 319 |
// Map it either way: an already-imported post is still the parent for its replies. |
| 320 |
$source_to_post_id[ $post['object']['id'] ] = $result; |
| 321 |
|
| 322 |
if ( $created ) { |
| 323 |
++$imported; |
| 324 |
} else { |
| 325 |
$skipped[] = $post['object']['id']; |
| 326 |
} |
| 327 |
} |
| 328 |
|
| 329 |
// Pass 3: Import self-replies as comments (sorted by date for correct threading). |
| 330 |
\usort( |
| 331 |
$self_replies, |
| 332 |
static function ( $a, $b ) { |
| 333 |
return \strtotime( $a['published'] ) <=> \strtotime( $b['published'] ); |
| 334 |
} |
| 335 |
); |
| 336 |
|
| 337 |
$source_to_comment_id = array(); |
| 338 |
$comments_skipped = array(); |
| 339 |
$comments_imported = 0; |
| 340 |
|
| 341 |
foreach ( $self_replies as $post ) { |
| 342 |
$result = self::import_as_comment( $post, $source_to_post_id, $source_to_comment_id ); |
| 343 |
|
| 344 |
if ( $result ) { |
| 345 |
++$comments_imported; |
| 346 |
} else { |
| 347 |
$comments_skipped[] = $post['object']['id']; |
| 348 |
} |
| 349 |
} |
| 350 |
|
| 351 |
// Output results. |
| 352 |
if ( ! empty( $skipped ) ) { |
| 353 |
echo '<p>' . \esc_html__( 'Skipped posts:', 'activitypub' ) . '<br>'; |
| 354 |
echo \wp_kses( \implode( '<br>', $skipped ), array( 'br' => array() ) ); |
| 355 |
echo '</p>'; |
| 356 |
} |
| 357 |
|
| 358 |
if ( ! empty( $comments_skipped ) ) { |
| 359 |
echo '<p>' . \esc_html__( 'Skipped comments:', 'activitypub' ) . '<br>'; |
| 360 |
echo \wp_kses( \implode( '<br>', $comments_skipped ), array( 'br' => array() ) ); |
| 361 |
echo '</p>'; |
| 362 |
} |
| 363 |
|
| 364 |
/* translators: %s: Number of posts */ |
| 365 |
echo '<p>' . \esc_html( \sprintf( \_n( 'Imported %s post.', 'Imported %s posts.', $imported, 'activitypub' ), \number_format_i18n( $imported ) ) ) . '</p>'; |
| 366 |
|
| 367 |
if ( $comments_imported > 0 ) { |
| 368 |
/* translators: %s: Number of comments */ |
| 369 |
echo '<p>' . \esc_html( \sprintf( \_n( 'Imported %s comment from self-reply threads.', 'Imported %s comments from self-reply threads.', $comments_imported, 'activitypub' ), \number_format_i18n( $comments_imported ) ) ) . '</p>'; |
| 370 |
} |
| 371 |
|
| 372 |
return true; |
| 373 |
} |
| 374 |
|
| 375 |
/** |
| 376 |
* Check if a post is a self-reply (thread continuation). |
| 377 |
* |
| 378 |
* A self-reply is when a user replies to their own post, creating a thread. |
| 379 |
* |
| 380 |
* @param array $post The Mastodon activity. |
| 381 |
* |
| 382 |
* @return bool True if replying to own post. |
| 383 |
*/ |
| 384 |
private static function is_self_reply( $post ) { |
| 385 |
if ( empty( $post['object']['inReplyTo'] ) ) { |
| 386 |
return false; |
| 387 |
} |
| 388 |
|
| 389 |
/* |
| 390 |
* Compare base URLs (actor URL should be a prefix of inReplyTo for self-replies). |
| 391 |
* |
| 392 |
* Example: |
| 393 |
* - actor: https://mastodon.social/users/example |
| 394 |
* - inReplyTo: https://mastodon.social/users/example/statuses/123 |
| 395 |
* |
| 396 |
* Adding a trailing slash ensures we don't match partial usernames |
| 397 |
* (e.g., "example" shouldn't match "example2"). |
| 398 |
*/ |
| 399 |
return \str_starts_with( $post['object']['inReplyTo'], \rtrim( $post['actor'], '/' ) . '/' ); |
| 400 |
} |
| 401 |
|
| 402 |
/** |
| 403 |
* Import a single activity as a WordPress post. |
| 404 |
* |
| 405 |
* @param array $post The Mastodon activity. |
| 406 |
* @param bool $created Set to false when the post was already imported and the |
| 407 |
* return value is the existing id rather than a new one. |
| 408 |
* |
| 409 |
* @return int|\WP_Error Post ID, either newly created or the one already imported, |
| 410 |
* or WP_Error on failure. |
| 411 |
*/ |
| 412 |
private static function import_as_post( $post, &$created ) { |
| 413 |
/* |
| 414 |
* An imported post is handled the same as a federated one: `Sanitize::content()` |
| 415 |
* normalizes and sanitizes the content. The |
| 416 |
* import runs as a user with `unfiltered_html`, so kses filters are not installed |
| 417 |
* for this request and `wp_insert_post()` would otherwise store whatever the archive |
| 418 |
* contained; these posts are published publicly, so they are the wider exposure. |
| 419 |
*/ |
| 420 |
$post_data = array( |
| 421 |
'post_author' => self::$author, |
| 422 |
'post_date' => $post['published'], |
| 423 |
// Slashed like the federated path: wp_insert_post() unslashes what it is given. |
| 424 |
'post_excerpt' => \wp_slash( \is_string( $post['object']['summary'] ?? null ) ? \wp_strip_all_tags( $post['object']['summary'] ) : '' ), |
| 425 |
'post_content' => \wp_slash( Sanitize::content( $post['object']['content'] ?? '' ) ), |
| 426 |
'post_status' => 'publish', |
| 427 |
'post_type' => 'post', |
| 428 |
'meta_input' => array( '_source_id' => $post['object']['id'] ), |
| 429 |
'tags_input' => \array_map( |
| 430 |
static function ( $tag ) { |
| 431 |
if ( 'Hashtag' === $tag['type'] ) { |
| 432 |
return \ltrim( $tag['name'], '#' ); |
| 433 |
} |
| 434 |
|
| 435 |
return ''; |
| 436 |
}, |
| 437 |
$post['object']['tag'] ?? array() |
| 438 |
), |
| 439 |
); |
| 440 |
|
| 441 |
/** |
| 442 |
* Filter the post data before inserting it into the database. |
| 443 |
* |
| 444 |
* @param array $post_data The post data to be inserted. |
| 445 |
* @param array $post The Mastodon Create activity. |
| 446 |
*/ |
| 447 |
$post_data = \apply_filters( 'activitypub_import_mastodon_post_data', $post_data, $post ); |
| 448 |
|
| 449 |
/* |
| 450 |
* Match on the archive's own id first. Falling back to `post_exists()` alone keys |
| 451 |
* de-duplication on an exact `post_content` match, so any change to what we store |
| 452 |
* (a new sanitizer, for one) makes everything imported by an older version stop |
| 453 |
* matching and come back as a duplicate. |
| 454 |
*/ |
| 455 |
$post_exists = self::get_post_by_source_id( $post['object']['id'] ?? '', $post_data['post_type'] ); |
| 456 |
|
| 457 |
if ( ! $post_exists ) { |
| 458 |
$post_exists = \post_exists( '', $post_data['post_content'], $post_data['post_date'], $post_data['post_type'] ); |
| 459 |
} |
| 460 |
|
| 461 |
/** |
| 462 |
* Filter ID of the existing post corresponding to post currently importing. |
| 463 |
* |
| 464 |
* Return 0 to force the post to be imported. Filter the ID to be something else |
| 465 |
* to override which existing post is mapped to the imported post. |
| 466 |
* |
| 467 |
* @see post_exists() |
| 468 |
* |
| 469 |
* @param int $post_exists Post ID, or 0 if post did not exist. |
| 470 |
* @param array $post_data The post array to be inserted. |
| 471 |
*/ |
| 472 |
$post_exists = \apply_filters( 'wp_import_existing_post', $post_exists, $post_data ); |
| 473 |
|
| 474 |
if ( $post_exists ) { |
| 475 |
/* |
| 476 |
* Report it as already imported, but hand the id back: pass 3 maps self-replies |
| 477 |
* onto their parent through this return value, and a reply whose parent we |
| 478 |
* skipped would otherwise find no parent and be dropped. |
| 479 |
*/ |
| 480 |
$created = false; |
| 481 |
|
| 482 |
return $post_exists; |
| 483 |
} |
| 484 |
|
| 485 |
$created = true; |
| 486 |
|
| 487 |
$post_id = \wp_insert_post( $post_data, true ); |
| 488 |
|
| 489 |
if ( \is_wp_error( $post_id ) ) { |
| 490 |
return $post_id; |
| 491 |
} |
| 492 |
|
| 493 |
\set_post_format( $post_id, 'status' ); |
| 494 |
|
| 495 |
// Process attachments if enabled. |
| 496 |
if ( self::$fetch_attachments && ! empty( $post['object']['attachment'] ) ) { |
| 497 |
// Prepend archive path to attachment URLs for local files. |
| 498 |
$attachments = \array_map( array( self::class, 'prepend_archive_path' ), $post['object']['attachment'] ); |
| 499 |
|
| 500 |
Attachments::import( $attachments, $post_id, self::$author ); |
| 501 |
} |
| 502 |
|
| 503 |
return $post_id; |
| 504 |
} |
| 505 |
|
| 506 |
/** |
| 507 |
* Find a post already imported under an archive object id. |
| 508 |
* |
| 509 |
* @since 9.3.0 |
| 510 |
* |
| 511 |
* @param string $source_id The archive object id. |
| 512 |
* @param string $post_type The post type to look in. |
| 513 |
* |
| 514 |
* @return int The post ID, or 0 when the object has not been imported yet. |
| 515 |
*/ |
| 516 |
private static function get_post_by_source_id( $source_id, $post_type ) { |
| 517 |
if ( ! \is_string( $source_id ) || '' === $source_id ) { |
| 518 |
return 0; |
| 519 |
} |
| 520 |
|
| 521 |
/* |
| 522 |
* `suppress_filters` is left at the get_posts() default of true on purpose: a |
| 523 |
* de-duplication probe is the last query a third-party `posts_where` should be |
| 524 |
* able to rewrite, since hiding the existing post brings the duplicates back. |
| 525 |
*/ |
| 526 |
|
| 527 |
/* |
| 528 |
* Every status, trash included. `post_exists()` below takes no status argument, so |
| 529 |
* it always matched a trashed import on content; `'any'` would be narrower, because |
| 530 |
* it drops the statuses flagged `exclude_from_search`. |
| 531 |
*/ |
| 532 |
$posts = \get_posts( |
| 533 |
array( |
| 534 |
'post_type' => $post_type, |
| 535 |
'post_status' => \get_post_stati(), |
| 536 |
'numberposts' => 1, |
| 537 |
'fields' => 'ids', |
| 538 |
'meta_key' => '_source_id', // phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_meta_key |
| 539 |
'meta_value' => $source_id, // phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_meta_value |
| 540 |
) |
| 541 |
); |
| 542 |
|
| 543 |
return $posts ? (int) $posts[0] : 0; |
| 544 |
} |
| 545 |
|
| 546 |
/** |
| 547 |
* Import a self-reply as a comment on its parent post. |
| 548 |
* |
| 549 |
* @param array $post The Mastodon activity. |
| 550 |
* @param array $source_to_post_id Mapping of source IDs to WordPress post IDs. |
| 551 |
* @param array $source_to_comment_id Mapping of source IDs to WordPress comment IDs (passed by reference). |
| 552 |
* |
| 553 |
* @return int|false Comment ID on success, false if parent not found or skipped. |
| 554 |
*/ |
| 555 |
private static function import_as_comment( $post, $source_to_post_id, &$source_to_comment_id ) { |
| 556 |
$in_reply_to = $post['object']['inReplyTo']; |
| 557 |
|
| 558 |
// Find parent - could be a post or another comment. |
| 559 |
$parent_post_id = null; |
| 560 |
$parent_comment_id = 0; |
| 561 |
|
| 562 |
if ( isset( $source_to_post_id[ $in_reply_to ] ) ) { |
| 563 |
// Replying to a root post or external reply. |
| 564 |
$parent_post_id = $source_to_post_id[ $in_reply_to ]; |
| 565 |
} elseif ( isset( $source_to_comment_id[ $in_reply_to ] ) ) { |
| 566 |
// Replying to another comment (nested thread). |
| 567 |
$parent_comment_id = $source_to_comment_id[ $in_reply_to ]; |
| 568 |
$parent_comment = \get_comment( $parent_comment_id ); |
| 569 |
|
| 570 |
if ( $parent_comment ) { |
| 571 |
$parent_post_id = $parent_comment->comment_post_ID; |
| 572 |
} |
| 573 |
} |
| 574 |
|
| 575 |
// If we couldn't find the parent, skip this comment. |
| 576 |
if ( ! $parent_post_id ) { |
| 577 |
return false; |
| 578 |
} |
| 579 |
|
| 580 |
// Check for duplicate. |
| 581 |
$existing_comments = \get_comments( |
| 582 |
array( |
| 583 |
'post_id' => $parent_post_id, |
| 584 |
'meta_key' => 'source_id', // phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_meta_key |
| 585 |
'meta_value' => $post['object']['id'], // phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_meta_value |
| 586 |
'number' => 1, |
| 587 |
) |
| 588 |
); |
| 589 |
|
| 590 |
if ( ! empty( $existing_comments ) ) { |
| 591 |
// Already imported, add to mapping and skip. |
| 592 |
$source_to_comment_id[ $post['object']['id'] ] = $existing_comments[0]->comment_ID; |
| 593 |
|
| 594 |
return false; |
| 595 |
} |
| 596 |
|
| 597 |
$comment_data = array( |
| 598 |
'comment_post_ID' => $parent_post_id, |
| 599 |
'comment_parent' => $parent_comment_id, |
| 600 |
'comment_author' => \wp_slash( \get_the_author_meta( 'display_name', self::$author ) ), |
| 601 |
|
| 602 |
/* |
| 603 |
* Sanitize the archive's content explicitly. The import runs as a user with |
| 604 |
* `unfiltered_html`, so `kses_init()` installs none of the kses filters for |
| 605 |
* this request: neither `wp_insert_comment()` nor the `pre_comment_*` chain |
| 606 |
* would touch this value. Core then prints `comment_content` unescaped, on |
| 607 |
* the front end and in the Dashboard "Activity" widget. |
| 608 |
* |
| 609 |
* The comment allowlist, not the post one, so this field is cleaned the same |
| 610 |
* way Collection\Interactions cleans a live-federated reply. |
| 611 |
*/ |
| 612 |
// Slashed like the federated path: wp_insert_comment() unslashes what it is given. |
| 613 |
'comment_content' => \wp_slash( Sanitize::comment_content( $post['object']['content'] ?? '' ) ), |
| 614 |
'comment_date' => $post['published'], |
| 615 |
'user_id' => self::$author, |
| 616 |
'comment_approved' => 1, |
| 617 |
); |
| 618 |
|
| 619 |
$comment_id = \wp_insert_comment( $comment_data ); |
| 620 |
|
| 621 |
if ( $comment_id ) { |
| 622 |
\update_comment_meta( $comment_id, 'source_id', $post['object']['id'] ); |
| 623 |
|
| 624 |
$source_to_comment_id[ $post['object']['id'] ] = $comment_id; |
| 625 |
} |
| 626 |
|
| 627 |
return $comment_id; |
| 628 |
} |
| 629 |
|
| 630 |
/** |
| 631 |
* Header. |
| 632 |
*/ |
| 633 |
public static function header() { |
| 634 |
echo '<div class="wrap">'; |
| 635 |
echo '<h2>' . \esc_html__( 'Import from Mastodon (Beta)', 'activitypub' ) . '</h2>'; |
| 636 |
} |
| 637 |
|
| 638 |
/** |
| 639 |
* Footer. |
| 640 |
*/ |
| 641 |
public static function footer() { |
| 642 |
echo '</div>'; |
| 643 |
} |
| 644 |
|
| 645 |
/** |
| 646 |
* Intro. |
| 647 |
*/ |
| 648 |
public static function greet() { |
| 649 |
echo '<div class="narrow">'; |
| 650 |
echo '<p>' . \wp_kses( |
| 651 |
\sprintf( |
| 652 |
/* translators: %s: URL to Mastodon export documentation */ |
| 653 |
\__( 'This importer allows you to bring your Mastodon posts into your WordPress site. For a smooth import experience, check out the <a href="%s" target="_blank">Mastodon documentation</a>.', 'activitypub' ), |
| 654 |
'https://docs.joinmastodon.org/user/moving/#export' |
| 655 |
), |
| 656 |
array( |
| 657 |
'a' => array( |
| 658 |
'href' => array(), |
| 659 |
'target' => array(), |
| 660 |
), |
| 661 |
) |
| 662 |
) . '</p>'; |
| 663 |
echo '<p>' . \esc_html__( 'Here’s how to get started:', 'activitypub' ) . '</p>'; |
| 664 |
|
| 665 |
echo '<ol>'; |
| 666 |
echo '<li>' . \wp_kses( \__( 'Log in to your Mastodon account and go to <strong>Preferences > Import and Export</strong>.', 'activitypub' ), array( 'strong' => array() ) ) . '</li>'; |
| 667 |
echo '<li>' . \esc_html__( 'Request a new archive of your data and wait for the email notification.', 'activitypub' ) . '</li>'; |
| 668 |
echo '<li>' . \wp_kses( \__( 'Download the archive file (it will be a <code>.zip</code> file).', 'activitypub' ), array( 'code' => array() ) ) . '</li>'; |
| 669 |
echo '<li>' . \esc_html__( 'Upload that file below to begin the import process.', 'activitypub' ) . '</li>'; |
| 670 |
echo '</ol>'; |
| 671 |
|
| 672 |
\wp_import_upload_form( 'admin.php?import=mastodon&step=1' ); |
| 673 |
echo '</div>'; |
| 674 |
} |
| 675 |
|
| 676 |
/** |
| 677 |
* Prepend archive path to local attachment URLs. |
| 678 |
* |
| 679 |
* @param array $attachment The attachment array. |
| 680 |
* |
| 681 |
* @return array The attachment array with updated URL. |
| 682 |
*/ |
| 683 |
private static function prepend_archive_path( $attachment ) { |
| 684 |
if ( ! empty( $attachment['url'] ) && ! \preg_match( '#^https?://#i', $attachment['url'] ) ) { |
| 685 |
$attachment['url'] = self::$archive . $attachment['url']; |
| 686 |
} |
| 687 |
|
| 688 |
return $attachment; |
| 689 |
} |
| 690 |
|
| 691 |
/** |
| 692 |
* Detect and unwrap single nested directory in archive. |
| 693 |
* |
| 694 |
* Some Mastodon exports wrap all files in a root folder. This method |
| 695 |
* detects this pattern and updates the archive path to point inside it. |
| 696 |
*/ |
| 697 |
private static function maybe_unwrap_archive() { |
| 698 |
global $wp_filesystem; |
| 699 |
|
| 700 |
$files = $wp_filesystem->dirlist( self::$archive ); |
| 701 |
|
| 702 |
// Check if there's exactly one directory at root level. |
| 703 |
if ( \count( $files ) !== 1 ) { |
| 704 |
return; |
| 705 |
} |
| 706 |
|
| 707 |
$first = \reset( $files ); |
| 708 |
if ( 'd' !== $first['type'] ) { |
| 709 |
return; |
| 710 |
} |
| 711 |
|
| 712 |
// Check if outbox.json exists inside the nested directory. |
| 713 |
$nested_path = self::$archive . '/' . $first['name']; |
| 714 |
if ( $wp_filesystem->exists( $nested_path . '/outbox.json' ) ) { |
| 715 |
self::$archive = $nested_path; |
| 716 |
} |
| 717 |
} |
| 718 |
} |
| 719 |
|
