← All changes
|
includes/admin/importers/class-snapshot-migrator.php
+786
-62
2.7.0
→
2.13.0
View file →
| @@ -54,8 +54,15 @@ | ||
| 54 | 54 | 'twitter_description' => '_thinkrank_twitter_description', |
| 55 | 55 | 'twitter_image' => '_thinkrank_twitter_image', |
| 56 | 56 | 'primary_category' => '_thinkrank_primary_category', |
| 57 | 57 | 'schema_type' => '_thinkrank_selected_schema_type', |
| 58 | + // WooCommerce product identifier (GTIN/MPN/ISBN). Free does not read | |
| 59 | + // it; ThinkRank Pro's Product_Fields does, under this exact key, so | |
| 60 | + // importing it here means the identifier is already in place when Pro | |
| 61 | + // is activated. Without it Google reports "missing identifier" on every | |
| 62 | + // product after a switch, which is a rich-result warning the user did | |
| 63 | + // not have before they migrated (#715). | |
| 64 | + 'product_identifier' => '_thinkrank_product_gtin', | |
| 58 | 65 | ]; |
| 59 | 66 | |
| 60 | 67 | /** |
| 61 | 68 | * Canonical robots meta fields. Composed into JSON-encoded |
| @@ -72,9 +79,17 @@ | ||
| 72 | 79 | |
| 73 | 80 | /** |
| 74 | 81 | * Data types that are migratable (have post/term/user meta mappings) |
| 75 | 82 | */ |
| 76 | - private const MIGRATABLE_TYPES = ['postmeta', 'termmeta', 'usermeta', 'redirections', '404_logs', 'settings']; | |
| 83 | + private const MIGRATABLE_TYPES = [ | |
| 84 | + 'postmeta', | |
| 85 | + 'termmeta', | |
| 86 | + 'usermeta', | |
| 87 | + 'redirections', | |
| 88 | + '404_logs', | |
| 89 | + 'settings', | |
| 90 | + Block_Converter::TYPE, | |
| 91 | + ]; | |
| 77 | 92 | |
| 78 | 93 | /** |
| 79 | 94 | * Settings-record `extended` keys that either migrate today or are safe to |
| 80 | 95 | * discard on cleanup (raw_options is pure capture-all insurance; a fresh |
| @@ -95,8 +110,16 @@ | ||
| 95 | 110 | 'role_capabilities', |
| 96 | 111 | 'image_seo', |
| 97 | 112 | 'sitemap_settings', |
| 98 | 113 | 'analytics_connected', |
| 114 | + 'focus_pages', | |
| 115 | + // Applied by migrate_site_identity() (#886). | |
| 116 | + 'robots_txt', | |
| 117 | + 'feed', | |
| 118 | + // Applied by migrate_markdown_for_ai() into ThinkRank Pro (#886). | |
| 119 | + // Without Pro it has no home, but neither is anything lost: cleanup | |
| 120 | + // removes a feature toggle, not content. | |
| 121 | + 'markdown_for_ai', | |
| 99 | 122 | // Capture-all raw buckets (whole source option sets stored verbatim). |
| 100 | 123 | // They live in the SNAPSHOT — cleanup never touches the snapshot — and |
| 101 | 124 | // a re-export recreates them, so they never block cleanup. |
| 102 | 125 | 'raw_options', |
| @@ -158,16 +181,42 @@ | ||
| 158 | 181 | if ($type === '404_logs') { |
| 159 | 182 | return $this->migrate_404_logs($plugin, $page); |
| 160 | 183 | } |
| 161 | 184 | |
| 185 | + if ($type === Block_Converter::TYPE) { | |
| 186 | + return $this->migrate_content_blocks($plugin, $page); | |
| 187 | + } | |
| 188 | + | |
| 162 | 189 | $chunk = Snapshot_Store::read_chunk($plugin, $type, $page); |
| 163 | 190 | if ($chunk === null || empty($chunk)) { |
| 191 | + // An empty chunk is not the end of the type. An exporter that pages | |
| 192 | + // one shared table and then splits the rows by kind writes nothing | |
| 193 | + // for a page whose rows all belonged to another kind — Squirrly | |
| 194 | + // reads the whole `qss` table that way, so a site whose terms and | |
| 195 | + // authors sit past the first page of posts has an empty chunk 1 and | |
| 196 | + // its real records in chunk 2. Ending the loop here dropped them | |
| 197 | + // silently, under a `complete` status, and cleanup then removed the | |
| 198 | + // source copy. Keep asking while the manifest says there are more | |
| 199 | + // chunks, exactly as the non-empty path below does. | |
| 200 | + $total_chunks = (int) ($manifest['types'][$type]['total_chunks'] ?? 0); | |
| 201 | + $has_more = $page < $total_chunks; | |
| 202 | + | |
| 203 | + // The last chunk of a sparse export can legitimately be the empty | |
| 204 | + // one — Squirrly's tail page holds only term rows — and it still | |
| 205 | + // ends the migration, so release the editors that mark_bulk() set | |
| 206 | + // polling. Without this they poll until the marker's own expiry. | |
| 207 | + if ($type === 'postmeta' && !$has_more) { | |
| 208 | + Metadata_Pending::clear_bulk(); | |
| 209 | + } | |
| 210 | + | |
| 164 | 211 | return [ |
| 165 | - 'status' => 'complete', | |
| 166 | - 'message' => 'No data in chunk', | |
| 167 | - 'has_more' => false, | |
| 168 | - 'processed' => 0, | |
| 169 | - 'skipped' => 0, | |
| 212 | + 'status' => $has_more ? 'processing' : 'complete', | |
| 213 | + 'message' => sprintf('No data in chunk %d', $page), | |
| 214 | + 'has_more' => $has_more, | |
| 215 | + 'page' => $page, | |
| 216 | + 'total_chunks' => $total_chunks, | |
| 217 | + 'processed' => 0, | |
| 218 | + 'skipped' => 0, | |
| 170 | 219 | ]; |
| 171 | 220 | } |
| 172 | 221 | |
| 173 | 222 | // Tell an open editor that SEO meta is being written right now, so its |
| @@ -197,8 +246,18 @@ | ||
| 197 | 246 | $skipped++; |
| 198 | 247 | continue; |
| 199 | 248 | } |
| 200 | 249 | |
| 250 | + // The object has to still exist. A source that keys its SEO by URL | |
| 251 | + // rather than by a foreign key keeps rows for content that was | |
| 252 | + // deleted years ago — Squirrly's `qss` table is keyed on a URL hash | |
| 253 | + // — and writing their meta creates orphan rows no screen can reach | |
| 254 | + // and no uninstall sweeps, while reporting them as migrated. | |
| 255 | + if (!$this->object_exists($object_type, $object_id)) { | |
| 256 | + $skipped++; | |
| 257 | + continue; | |
| 258 | + } | |
| 259 | + | |
| 201 | 260 | // Track migrated posts so their SEO score can be computed once the |
| 202 | 261 | // chunk's meta has landed (terms are not scored). |
| 203 | 262 | if ($object_type === 'post') { |
| 204 | 263 | $post_ids[$object_id] = true; |
| @@ -209,8 +268,25 @@ | ||
| 209 | 268 | // Collect focus keywords (primary + secondary) to seed the Pro |
| 210 | 269 | // Rank Tracker watch-list once the chunk is processed. |
| 211 | 270 | $this->collect_keywords($record, $data, $keywords); |
| 212 | 271 | |
| 272 | + // Term social fields travel in `extended`, not `data`: only the | |
| 273 | + // AIOSEO exporter puts them in the canonical bucket, the other four | |
| 274 | + // put the identical values one level out. The loop below walks | |
| 275 | + // `data`, so every term OG image and Twitter title/description was | |
| 276 | + // exported and then dropped. Fold them in for terms, without | |
| 277 | + // letting them win over a value the record already carries. | |
| 278 | + if ($object_type === 'term') { | |
| 279 | + $extended = is_array($record['extended'] ?? null) ? $record['extended'] : []; | |
| 280 | + foreach (['og_title', 'og_description', 'og_image', 'twitter_title', 'twitter_description', 'twitter_image'] as $social_key) { | |
| 281 | + if (!isset($data[$social_key]) || $data[$social_key] === '') { | |
| 282 | + if (isset($extended[$social_key]) && $extended[$social_key] !== '') { | |
| 283 | + $data[$social_key] = $extended[$social_key]; | |
| 284 | + } | |
| 285 | + } | |
| 286 | + } | |
| 287 | + } | |
| 288 | + | |
| 213 | 289 | foreach ($data as $canonical_key => $value) { |
| 214 | 290 | if (!isset(self::META_MAP[$canonical_key])) { |
| 215 | 291 | continue; |
| 216 | 292 | } |
| @@ -216,10 +292,13 @@ | ||
| 216 | 292 | } |
| 217 | 293 | |
| 218 | 294 | // Focus keywords are migrated as an array via the dedicated |
| 219 | 295 | // migrate_focus_keywords() below (which also keeps the legacy |
| 220 | - // single-value meta in sync), so skip the scalar write here. | |
| 221 | - if ($canonical_key === 'focus_keyword') { | |
| 296 | + // single-value meta in sync), so skip the scalar write here — | |
| 297 | + // but that writer only runs for posts, so skipping it for a | |
| 298 | + // term meant nobody wrote the term's focus keyword at all, | |
| 299 | + // even though get-term-seo reads `_thinkrank_focus_keyword`. | |
| 300 | + if ($canonical_key === 'focus_keyword' && $object_type === 'post') { | |
| 222 | 301 | continue; |
| 223 | 302 | } |
| 224 | 303 | |
| 225 | 304 | $thinkrank_key = self::META_MAP[$canonical_key]; |
| @@ -233,8 +312,12 @@ | ||
| 233 | 312 | if ($value === 0 && in_array($canonical_key, ['primary_category'], true)) { |
| 234 | 313 | continue; |
| 235 | 314 | } |
| 236 | 315 | |
| 316 | + // The meta writers unslash their value, so every write below | |
| 317 | + // slashes first. Unslashed, a title or description with a | |
| 318 | + // backslash in it lost it on the way in, and a JSON value lost | |
| 319 | + // the backslash of every `\"` and `\uXXXX` escape. | |
| 237 | 320 | if ($object_type === 'post') { |
| 238 | 321 | // Never overwrite existing ThinkRank data |
| 239 | 322 | $existing = get_post_meta($object_id, $thinkrank_key, true); |
| 240 | 323 | if ($existing !== '' && $existing !== false && $existing !== null) { |
| @@ -240,9 +323,9 @@ | ||
| 240 | 323 | if ($existing !== '' && $existing !== false && $existing !== null) { |
| 241 | 324 | continue; |
| 242 | 325 | } |
| 243 | 326 | |
| 244 | - update_post_meta($object_id, $thinkrank_key, $value); | |
| 327 | + update_post_meta($object_id, $thinkrank_key, wp_slash($value)); | |
| 245 | 328 | $record_had_writes = true; |
| 246 | 329 | } elseif ($object_type === 'term') { |
| 247 | 330 | $existing = get_term_meta($object_id, $thinkrank_key, true); |
| 248 | 331 | if ($existing !== '' && $existing !== false && $existing !== null) { |
| @@ -248,9 +331,9 @@ | ||
| 248 | 331 | if ($existing !== '' && $existing !== false && $existing !== null) { |
| 249 | 332 | continue; |
| 250 | 333 | } |
| 251 | 334 | |
| 252 | - update_term_meta($object_id, $thinkrank_key, $value); | |
| 335 | + update_term_meta($object_id, $thinkrank_key, wp_slash($value)); | |
| 253 | 336 | $record_had_writes = true; |
| 254 | 337 | } elseif ($object_type === 'user') { |
| 255 | 338 | $existing = get_user_meta($object_id, $thinkrank_key, true); |
| 256 | 339 | if ($existing !== '' && $existing !== false && $existing !== null) { |
| @@ -256,9 +339,9 @@ | ||
| 256 | 339 | if ($existing !== '' && $existing !== false && $existing !== null) { |
| 257 | 340 | continue; |
| 258 | 341 | } |
| 259 | 342 | |
| 260 | - update_user_meta($object_id, $thinkrank_key, $value); | |
| 343 | + update_user_meta($object_id, $thinkrank_key, wp_slash($value)); | |
| 261 | 344 | $record_had_writes = true; |
| 262 | 345 | } |
| 263 | 346 | } |
| 264 | 347 | |
| @@ -272,8 +355,17 @@ | ||
| 272 | 355 | if ($object_type === 'post' && $this->migrate_robots_payload($object_id, $data)) { |
| 273 | 356 | $record_had_writes = true; |
| 274 | 357 | } |
| 275 | 358 | |
| 359 | + // The same directives for a term. Every exporter emits term | |
| 360 | + // noindex/nofollow and ThinkRank stores them, but nothing wrote | |
| 361 | + // them — so a category the owner had deliberately kept out of the | |
| 362 | + // index came back indexable after the switch, which is the worst | |
| 363 | + // way for an import to be wrong. | |
| 364 | + if ($object_type === 'term' && $this->migrate_term_robots_payload($object_id, $data)) { | |
| 365 | + $record_had_writes = true; | |
| 366 | + } | |
| 367 | + | |
| 276 | 368 | // Pillar / cornerstone content flag (post meta only). |
| 277 | 369 | if ($object_type === 'post' && $this->migrate_pillar_content($object_id, $data)) { |
| 278 | 370 | $record_had_writes = true; |
| 279 | 371 | } |
| @@ -289,8 +381,15 @@ | ||
| 289 | 381 | if ($object_type === 'post' && $this->migrate_video_schema($object_id, $data, $record)) { |
| 290 | 382 | $record_had_writes = true; |
| 291 | 383 | } |
| 292 | 384 | |
| 385 | + // The post's own schemas as ThinkRank Pro Custom Schema entries | |
| 386 | + // scoped to it (extended.custom_schemas, #886). | |
| 387 | + if ($object_type === 'post' && !empty($record['extended']['custom_schemas']) && is_array($record['extended']['custom_schemas']) | |
| 388 | + && $this->migrate_custom_schemas($source_plugin, $record['extended']['custom_schemas']) > 0) { | |
| 389 | + $record_had_writes = true; | |
| 390 | + } | |
| 391 | + | |
| 293 | 392 | // Per-object redirect. SEOPress is the one source that stores a |
| 294 | 393 | // redirect as object meta rather than in a rules table, so its |
| 295 | 394 | // per-post redirects were exported into extended.redirect_* and |
| 296 | 395 | // then dropped for want of anywhere to put them. They have a home |
| @@ -305,9 +404,18 @@ | ||
| 305 | 404 | // Per-post "exclude from sitemap" flags. ThinkRank models sitemap |
| 306 | 405 | // exclusion as one comma-separated ID list on the sitemap settings |
| 307 | 406 | // rather than per-post meta, so collect the IDs and apply them once |
| 308 | 407 | // after the chunk (a settings write per post would be wasteful). |
| 309 | - if ($object_type === 'post' && !empty($record['extended']['exclude_sitemap'])) { | |
| 408 | + // Two spellings reach here: Rank Math's exporter emits | |
| 409 | + // `exclude_sitemap`, Squirrly's `exclude_from_sitemap`. Only the | |
| 410 | + // first was read, so every Squirrly `nositemap` flag was dropped | |
| 411 | + // and posts the owner had hidden reappeared in the sitemap. Accept | |
| 412 | + // both rather than renaming one, because snapshots already exported | |
| 413 | + // carry whichever spelling their exporter used at the time. | |
| 414 | + $excluded_from_sitemap = !empty($record['extended']['exclude_sitemap']) | |
| 415 | + || !empty($record['extended']['exclude_from_sitemap']); | |
| 416 | + | |
| 417 | + if ($object_type === 'post' && $excluded_from_sitemap) { | |
| 310 | 418 | $sitemap_excluded[] = $object_id; |
| 311 | 419 | } |
| 312 | 420 | |
| 313 | 421 | if ($record_had_writes) { |
| @@ -558,18 +666,22 @@ | ||
| 558 | 666 | // strings; but the restore's whole policy is to write the file's value |
| 559 | 667 | // verbatim, and a file holding one as an array would otherwise raise a |
| 560 | 668 | // TypeError that takes down the rest of the chunk with it. One bad key |
| 561 | 669 | // is worth skipping, not the records behind it. |
| 670 | + // | |
| 671 | + // wp_slash() because the meta writers unslash: a restored JSON value | |
| 672 | + // (schema form data, robots) would otherwise lose the backslash of | |
| 673 | + // every escaped quote and come back as invalid JSON. | |
| 562 | 674 | try { |
| 563 | 675 | switch ($object_type) { |
| 564 | 676 | case 'post': |
| 565 | - update_post_meta($object_id, $key, $value); | |
| 677 | + update_post_meta($object_id, $key, wp_slash($value)); | |
| 566 | 678 | return true; |
| 567 | 679 | case 'term': |
| 568 | - update_term_meta($object_id, $key, $value); | |
| 680 | + update_term_meta($object_id, $key, wp_slash($value)); | |
| 569 | 681 | return true; |
| 570 | 682 | case 'user': |
| 571 | - update_user_meta($object_id, $key, $value); | |
| 683 | + update_user_meta($object_id, $key, wp_slash($value)); | |
| 572 | 684 | return true; |
| 573 | 685 | } |
| 574 | 686 | } catch (\Throwable $e) { |
| 575 | 687 | if (defined('WP_DEBUG') && WP_DEBUG) { |
| @@ -1296,8 +1408,115 @@ | ||
| 1296 | 1408 | return !is_wp_error($result); |
| 1297 | 1409 | } |
| 1298 | 1410 | |
| 1299 | 1411 | /** |
| 1412 | + * The robots directives for a term. | |
| 1413 | + * | |
| 1414 | + * Deliberately narrower than migrate_robots_payload(): update-term-seo | |
| 1415 | + * writes `_thinkrank_robots_meta` and `_thinkrank_robots_meta_enabled` and | |
| 1416 | + * nothing else for a term, so the advanced directives a post supports have | |
| 1417 | + * nowhere to go here and are left in the snapshot rather than written to a | |
| 1418 | + * key no reader looks at. | |
| 1419 | + * | |
| 1420 | + * Same two rules as the post path — never overwrite an existing payload, | |
| 1421 | + * and never turn the override on for an all-false set, which is just | |
| 1422 | + * ThinkRank's default index/follow spelled out. | |
| 1423 | + * | |
| 1424 | + * @param int $term_id Target term ID. | |
| 1425 | + * @param array $data Canonical record data. | |
| 1426 | + * @return bool True when a payload was written. | |
| 1427 | + */ | |
| 1428 | + private function migrate_term_robots_payload(int $term_id, array $data): bool { | |
| 1429 | + $existing = get_term_meta($term_id, '_thinkrank_robots_meta', true); | |
| 1430 | + if (is_string($existing) && $existing !== '') { | |
| 1431 | + return false; | |
| 1432 | + } | |
| 1433 | + | |
| 1434 | + $robots = []; | |
| 1435 | + $has_active_directive = false; | |
| 1436 | + | |
| 1437 | + foreach (self::ROBOTS_FIELDS as $field) { | |
| 1438 | + if (!array_key_exists($field, $data)) { | |
| 1439 | + continue; | |
| 1440 | + } | |
| 1441 | + | |
| 1442 | + $value = $data[$field]; | |
| 1443 | + if ($value === '' || $value === null) { | |
| 1444 | + continue; | |
| 1445 | + } | |
| 1446 | + | |
| 1447 | + $robots[$field] = (bool) (int) $value; | |
| 1448 | + if ($robots[$field]) { | |
| 1449 | + $has_active_directive = true; | |
| 1450 | + } | |
| 1451 | + } | |
| 1452 | + | |
| 1453 | + if (!$has_active_directive) { | |
| 1454 | + return false; | |
| 1455 | + } | |
| 1456 | + | |
| 1457 | + $robots['index'] = empty($robots['noindex']); | |
| 1458 | + | |
| 1459 | + update_term_meta($term_id, '_thinkrank_robots_meta', wp_slash(wp_json_encode($robots))); | |
| 1460 | + update_term_meta($term_id, '_thinkrank_robots_meta_enabled', 1); | |
| 1461 | + | |
| 1462 | + return true; | |
| 1463 | + } | |
| 1464 | + | |
| 1465 | + /** | |
| 1466 | + * Carry the source's watched pages into ThinkRank Pro's Focus Pages. | |
| 1467 | + * | |
| 1468 | + * Squirrly keeps this list on its own servers, so the exporter reads it | |
| 1469 | + * live while the source plugin is still installed and connected — after | |
| 1470 | + * the switch there is nowhere left to read it from. See | |
| 1471 | + * Squirrly_Exporter::fetch_focus_pages(). | |
| 1472 | + * | |
| 1473 | + * Focus Pages is a Pro feature and a deliberately small, hand-picked list | |
| 1474 | + * (Settings::MAX_PAGES). Two rules follow from that: never touch a | |
| 1475 | + * selection the user has already made here, and never import more than | |
| 1476 | + * the cap. Without Pro the ids stay in the snapshot for a later run, the | |
| 1477 | + * same way per-object redirects wait for Pro's rules table. | |
| 1478 | + * | |
| 1479 | + * @param array $extended Extended settings payload. | |
| 1480 | + * @return bool True when at least one page was added. | |
| 1481 | + */ | |
| 1482 | + private function migrate_focus_pages(array $extended): bool { | |
| 1483 | + $ids = $extended['focus_pages'] ?? []; | |
| 1484 | + if (!is_array($ids) || $ids === []) { | |
| 1485 | + return false; | |
| 1486 | + } | |
| 1487 | + | |
| 1488 | + if (!class_exists('ThinkRank\\Pro\\Focus_Pages\\Settings')) { | |
| 1489 | + return false; | |
| 1490 | + } | |
| 1491 | + | |
| 1492 | + $settings = new \ThinkRank\Pro\Focus_Pages\Settings(); | |
| 1493 | + | |
| 1494 | + // A choice already made here outranks one carried over, exactly as | |
| 1495 | + // every other field in this class treats an existing value. | |
| 1496 | + if ($settings->get() !== []) { | |
| 1497 | + return false; | |
| 1498 | + } | |
| 1499 | + | |
| 1500 | + $added = false; | |
| 1501 | + foreach ($ids as $id) { | |
| 1502 | + $post_id = (int) $id; | |
| 1503 | + if ($post_id <= 0 || get_post($post_id) === null) { | |
| 1504 | + continue; | |
| 1505 | + } | |
| 1506 | + | |
| 1507 | + if (method_exists($settings, 'is_full') && $settings->is_full()) { | |
| 1508 | + break; | |
| 1509 | + } | |
| 1510 | + | |
| 1511 | + $settings->add($post_id); | |
| 1512 | + $added = true; | |
| 1513 | + } | |
| 1514 | + | |
| 1515 | + return $added; | |
| 1516 | + } | |
| 1517 | + | |
| 1518 | + /** | |
| 1300 | 1519 | * Migrate the pillar / cornerstone content flag to ThinkRank post meta. |
| 1301 | 1520 | * |
| 1302 | 1521 | * ThinkRank stores an enabled flag as the string '1'; the reader |
| 1303 | 1522 | * (Pillar_Content endpoint) matches meta_value = '1'. Never overwrites an |
| @@ -1397,9 +1616,9 @@ | ||
| 1397 | 1616 | if (is_string($existing) && $existing !== '') { |
| 1398 | 1617 | return false; |
| 1399 | 1618 | } |
| 1400 | 1619 | |
| 1401 | - update_post_meta($post_id, '_thinkrank_schema_form_data', wp_json_encode($review)); | |
| 1620 | + update_post_meta($post_id, '_thinkrank_schema_form_data', wp_slash(wp_json_encode($review))); | |
| 1402 | 1621 | |
| 1403 | 1622 | return true; |
| 1404 | 1623 | } |
| 1405 | 1624 | |
| @@ -1431,9 +1650,9 @@ | ||
| 1431 | 1650 | if (is_string($existing) && $existing !== '') { |
| 1432 | 1651 | return false; |
| 1433 | 1652 | } |
| 1434 | 1653 | |
| 1435 | - update_post_meta($post_id, '_thinkrank_schema_form_data', wp_json_encode($video)); | |
| 1654 | + update_post_meta($post_id, '_thinkrank_schema_form_data', wp_slash(wp_json_encode($video))); | |
| 1436 | 1655 | |
| 1437 | 1656 | return true; |
| 1438 | 1657 | } |
| 1439 | 1658 | |
| @@ -1513,13 +1732,13 @@ | ||
| 1513 | 1732 | } |
| 1514 | 1733 | |
| 1515 | 1734 | $wrote = false; |
| 1516 | 1735 | if (!empty($robots)) { |
| 1517 | - update_post_meta($post_id, '_thinkrank_robots_meta', wp_json_encode($robots)); | |
| 1736 | + update_post_meta($post_id, '_thinkrank_robots_meta', wp_slash(wp_json_encode($robots))); | |
| 1518 | 1737 | $wrote = true; |
| 1519 | 1738 | } |
| 1520 | 1739 | if (!empty($advanced)) { |
| 1521 | - update_post_meta($post_id, '_thinkrank_advanced_robots_meta', wp_json_encode($advanced)); | |
| 1740 | + update_post_meta($post_id, '_thinkrank_advanced_robots_meta', wp_slash(wp_json_encode($advanced))); | |
| 1522 | 1741 | $wrote = true; |
| 1523 | 1742 | } |
| 1524 | 1743 | if ($wrote) { |
| 1525 | 1744 | update_post_meta($post_id, '_thinkrank_robots_meta_enabled', 1); |
| @@ -1560,9 +1779,9 @@ | ||
| 1560 | 1779 | $processed++; |
| 1561 | 1780 | } |
| 1562 | 1781 | } |
| 1563 | 1782 | |
| 1564 | - if (!empty($data['homepage_title']) || !empty($data['homepage_description']) || !empty($data['organization_name']) || !empty($data['organization_logo'])) { | |
| 1783 | + if (!empty($data['homepage_title']) || !empty($data['organization_name']) || !empty($data['organization_logo'])) { | |
| 1565 | 1784 | $site_identity = get_option('thinkrank_site_identity_settings', []); |
| 1566 | 1785 | $updated = false; |
| 1567 | 1786 | |
| 1568 | 1787 | if (!empty($data['homepage_title']) && empty($site_identity['homepage_title'])) { |
| @@ -1568,12 +1787,8 @@ | ||
| 1568 | 1787 | if (!empty($data['homepage_title']) && empty($site_identity['homepage_title'])) { |
| 1569 | 1788 | $site_identity['homepage_title'] = $data['homepage_title']; |
| 1570 | 1789 | $updated = true; |
| 1571 | 1790 | } |
| 1572 | - if (!empty($data['homepage_description']) && empty($site_identity['homepage_description'])) { | |
| 1573 | - $site_identity['homepage_description'] = $data['homepage_description']; | |
| 1574 | - $updated = true; | |
| 1575 | - } | |
| 1576 | 1791 | if (!empty($data['organization_name']) && empty($site_identity['organization_name'])) { |
| 1577 | 1792 | $site_identity['organization_name'] = $data['organization_name']; |
| 1578 | 1793 | $updated = true; |
| 1579 | 1794 | } |
| @@ -1587,21 +1802,10 @@ | ||
| 1587 | 1802 | $processed++; |
| 1588 | 1803 | } |
| 1589 | 1804 | } |
| 1590 | 1805 | |
| 1591 | - if (!empty($data['social_profiles'])) { | |
| 1592 | - $social = get_option('thinkrank_social_media_settings', []); | |
| 1593 | - $updated = false; | |
| 1594 | - | |
| 1595 | - foreach ($data['social_profiles'] as $platform => $url) { | |
| 1596 | - if (!empty($url) && empty($social[$platform])) { | |
| 1597 | - $social[$platform] = $url; | |
| 1598 | - $updated = true; | |
| 1599 | - } | |
| 1600 | - } | |
| 1601 | - | |
| 1602 | - if ($updated) { | |
| 1603 | - update_option('thinkrank_social_media_settings', $social); | |
| 1806 | + if (!empty($data['social_profiles']) && is_array($data['social_profiles'])) { | |
| 1807 | + if ($this->migrate_social_profiles($data['social_profiles'])) { | |
| 1604 | 1808 | $processed++; |
| 1605 | 1809 | } |
| 1606 | 1810 | } |
| 1607 | 1811 | |
| @@ -1658,8 +1862,23 @@ | ||
| 1658 | 1862 | $processed++; |
| 1659 | 1863 | } |
| 1660 | 1864 | } |
| 1661 | 1865 | |
| 1866 | + // Per-taxonomy and blog-index noindex, into the Content Type Matrix. | |
| 1867 | + // The templates beside them have no ThinkRank store and stay preserved. | |
| 1868 | + if ($this->migrate_archive_noindex($extended)) { | |
| 1869 | + $processed++; | |
| 1870 | + } | |
| 1871 | + | |
| 1872 | + // Site-wide custom schemas and Markdown for AI, both ThinkRank Pro. | |
| 1873 | + $global_schemas = $extended['custom_schemas']['entries'] ?? []; | |
| 1874 | + if (is_array($global_schemas) && $this->migrate_custom_schemas($plugin, $global_schemas) > 0) { | |
| 1875 | + $processed++; | |
| 1876 | + } | |
| 1877 | + if ($this->migrate_markdown_for_ai($extended)) { | |
| 1878 | + $processed++; | |
| 1879 | + } | |
| 1880 | + | |
| 1662 | 1881 | // Site-identity settings (homepage/org/breadcrumbs/local SEO) are served to |
| 1663 | 1882 | // the frontend from the wp_thinkrank_seo_settings table via the manager, not |
| 1664 | 1883 | // from the option written above — route them through the manager so they |
| 1665 | 1884 | // actually take effect. |
| @@ -1671,8 +1890,13 @@ | ||
| 1671 | 1890 | if ($this->migrate_image_seo($extended)) { |
| 1672 | 1891 | $processed++; |
| 1673 | 1892 | } |
| 1674 | 1893 | |
| 1894 | + // The pages the source had under active watch. | |
| 1895 | + if ($this->migrate_focus_pages($extended)) { | |
| 1896 | + $processed++; | |
| 1897 | + } | |
| 1898 | + | |
| 1675 | 1899 | // Sitemap inclusion settings. |
| 1676 | 1900 | if ($this->migrate_sitemap($extended)) { |
| 1677 | 1901 | $processed++; |
| 1678 | 1902 | } |
| @@ -1740,18 +1964,33 @@ | ||
| 1740 | 1964 | $manager = new \ThinkRank\SEO\Site_Identity_Manager(); |
| 1741 | 1965 | $current = $manager->get_settings('site'); |
| 1742 | 1966 | |
| 1743 | 1967 | // ThinkRank default seeds — only overwrite a value the user has not changed. |
| 1744 | - $seeds = [ | |
| 1745 | - 'homepage_title' => '%site_title% | %site_description%', | |
| 1746 | - 'site_name' => get_bloginfo('name'), | |
| 1747 | - 'logo_url' => '', | |
| 1748 | - 'breadcrumb_home_text' => 'Home', | |
| 1749 | - 'breadcrumb_separator' => '>', | |
| 1750 | - 'business_type' => '', | |
| 1751 | - 'business_name' => '', | |
| 1752 | - 'business_phone' => '', | |
| 1753 | - ]; | |
| 1968 | + // | |
| 1969 | + // The title formats come from Site_Identity_Manager rather than being | |
| 1970 | + // restated here. They were restated once, drifted (the homepage seed | |
| 1971 | + // still used a literal '|' after the shipped default moved to %sep%), | |
| 1972 | + // and the six per-context formats below were never listed at all — so | |
| 1973 | + // every shipped default read as "the user chose this" and no imported | |
| 1974 | + // title format was ever written. | |
| 1975 | + $seeds = array_merge( | |
| 1976 | + \ThinkRank\SEO\Site_Identity_Manager::TITLE_FORMAT_DEFAULTS, | |
| 1977 | + [ | |
| 1978 | + 'site_name' => get_bloginfo('name'), | |
| 1979 | + 'logo_url' => '', | |
| 1980 | + 'breadcrumb_home_text' => 'Home', | |
| 1981 | + // Two shipped values, both untouched. get_default_settings() | |
| 1982 | + // says '>' and the admin screen seeds '›' (as does the | |
| 1983 | + // breadcrumb renderer's own fallback), so which one a site | |
| 1984 | + // holds depends only on whether that screen has ever been | |
| 1985 | + // saved. Recognising one and not the other would skip the | |
| 1986 | + // imported separator on half of all installs. | |
| 1987 | + 'breadcrumb_separator' => ['>', '›'], | |
| 1988 | + 'business_type' => '', | |
| 1989 | + 'business_name' => '', | |
| 1990 | + 'business_phone' => '', | |
| 1991 | + ] | |
| 1992 | + ); | |
| 1754 | 1993 | |
| 1755 | 1994 | $updates = []; |
| 1756 | 1995 | $set = static function (string $key, $value) use (&$updates, $current, $seeds): void { |
| 1757 | 1996 | if ($value === '' || $value === null) { |
| @@ -1757,9 +1996,14 @@ | ||
| 1757 | 1996 | if ($value === '' || $value === null) { |
| 1758 | 1997 | return; |
| 1759 | 1998 | } |
| 1760 | 1999 | $cur = $current[$key] ?? null; |
| 1761 | - $is_default = !array_key_exists($key, $current) || $cur === '' || $cur === ($seeds[$key] ?? null); | |
| 2000 | + // A seed may list several values when more than one shipped default | |
| 2001 | + // is in circulation for the same field. | |
| 2002 | + $shipped = array_key_exists($key, $seeds) ? (array) $seeds[$key] : []; | |
| 2003 | + $is_default = !array_key_exists($key, $current) | |
| 2004 | + || $cur === '' | |
| 2005 | + || in_array($cur, $shipped, true); | |
| 1762 | 2006 | if ($is_default) { |
| 1763 | 2007 | $updates[$key] = $value; |
| 1764 | 2008 | } |
| 1765 | 2009 | }; |
| @@ -1770,8 +2014,12 @@ | ||
| 1770 | 2014 | // real template and beats the literal-resolved data.homepage_title, so it |
| 1771 | 2015 | // wins when the source provided one. |
| 1772 | 2016 | $title_formats = is_array($extended['title_formats'] ?? null) ? $extended['title_formats'] : []; |
| 1773 | 2017 | $set('homepage_title', $title_formats['homepage_title'] ?? ($data['homepage_title'] ?? '')); |
| 2018 | + // The homepage meta description. It used to go to the legacy | |
| 2019 | + // thinkrank_site_identity_settings option, which nothing reads, so the | |
| 2020 | + // homepage fell back to the tagline (#897). | |
| 2021 | + $set('homepage_description', trim((string) ($data['homepage_description'] ?? ''))); | |
| 1774 | 2022 | $set('site_name', $data['organization_name'] ?? ''); |
| 1775 | 2023 | $set('alternate_name', $data['alternate_name'] ?? ''); |
| 1776 | 2024 | $set('logo_url', $data['organization_logo'] ?? ''); |
| 1777 | 2025 | |
| @@ -1813,8 +2061,23 @@ | ||
| 1813 | 2061 | $set('breadcrumb_separator', $breadcrumbs['separator'] ?? ''); |
| 1814 | 2062 | $set('breadcrumb_prefix', $breadcrumbs['prefix'] ?? ''); |
| 1815 | 2063 | } |
| 1816 | 2064 | |
| 2065 | + // A hand-written robots.txt (extended.robots_txt.content) goes into | |
| 2066 | + // ThinkRank's Robots.txt textarea only while that is still empty: | |
| 2067 | + // a file the user already wrote here is never replaced (#886). | |
| 2068 | + $robots_txt = $extended['robots_txt'] ?? []; | |
| 2069 | + if (is_array($robots_txt)) { | |
| 2070 | + $set('robots_txt_content', trim((string) ($robots_txt['content'] ?? ''))); | |
| 2071 | + } | |
| 2072 | + | |
| 2073 | + // The "appeared first on" feed backlink (extended.feed.source_link). | |
| 2074 | + // Only switches it on: a source that had it off is no reason to turn | |
| 2075 | + // off a signature ThinkRank seeds for new installs. | |
| 2076 | + if (!empty($extended['feed']['source_link']) && empty($current['feed_source_link'])) { | |
| 2077 | + $updates['feed_source_link'] = true; | |
| 2078 | + } | |
| 2079 | + | |
| 1817 | 2080 | // Local SEO (extended.local_seo) — migrate the full NAP + geo when there |
| 1818 | 2081 | // is any meaningful business data (name, phone, address or coordinates), |
| 1819 | 2082 | // and enable the feature alongside it. Each field is written only while |
| 1820 | 2083 | // ThinkRank's Business Info still holds its default (non-destructive). |
| @@ -1961,9 +2224,10 @@ | ||
| 1961 | 2224 | } |
| 1962 | 2225 | |
| 1963 | 2226 | $app_id = trim((string) ($defaults['facebook_app_id'] ?? '')); |
| 1964 | 2227 | $og_image = trim((string) ($defaults['og_default_image'] ?? '')); |
| 1965 | - if ($app_id === '' && $og_image === '') { | |
| 2228 | + $twitter_image = trim((string) ($defaults['twitter_default_image'] ?? '')); | |
| 2229 | + if ($app_id === '' && $og_image === '' && $twitter_image === '') { | |
| 1966 | 2230 | return false; |
| 1967 | 2231 | } |
| 1968 | 2232 | |
| 1969 | 2233 | $manager = new \ThinkRank\SEO\Social_Meta_Manager(); |
| @@ -1972,11 +2236,19 @@ | ||
| 1972 | 2236 | $updates = []; |
| 1973 | 2237 | if ($app_id !== '' && empty($current['facebook_app_id'])) { |
| 1974 | 2238 | $updates['facebook_app_id'] = $app_id; |
| 1975 | 2239 | } |
| 1976 | - if ($og_image !== '' && empty($current['default_image'])) { | |
| 1977 | - $updates['default_image'] = $og_image; | |
| 2240 | + // `default_og_image` is the key Social_Meta_Manager declares for the | |
| 2241 | + // site context. `default_image` is only a legacy alias the front-end | |
| 2242 | + // readers still honour — saving under it is discarded, because a key | |
| 2243 | + // outside the allow-list never reaches the database. | |
| 2244 | + if ($og_image !== '' && empty($current['default_og_image'])) { | |
| 2245 | + $updates['default_og_image'] = $og_image; | |
| 1978 | 2246 | } |
| 2247 | + // The Twitter card fallback image (Slim SEO's default_twitter_image). | |
| 2248 | + if ($twitter_image !== '' && empty($current['default_twitter_image'])) { | |
| 2249 | + $updates['default_twitter_image'] = $twitter_image; | |
| 2250 | + } | |
| 1979 | 2251 | |
| 1980 | 2252 | if (empty($updates)) { |
| 1981 | 2253 | return false; |
| 1982 | 2254 | } |
| @@ -1984,8 +2256,97 @@ | ||
| 1984 | 2256 | return (bool) $manager->save_settings('site', null, $updates); |
| 1985 | 2257 | } |
| 1986 | 2258 | |
| 1987 | 2259 | /** |
| 2260 | + * Social profile URLs (data.social_profiles, platform => URL) into the | |
| 2261 | + * stores that are read (#898). | |
| 2262 | + * | |
| 2263 | + * They used to be written as `thinkrank_social_media_settings[<platform>]`, | |
| 2264 | + * a key nothing reads: the consumer looks for `<platform>_url`, the | |
| 2265 | + * Organization graph reads the Schema Manager's | |
| 2266 | + * `organization_social_<platform>`, and `twitter:site` comes from the | |
| 2267 | + * Social Meta `twitter_username`. So the profiles never reached `sameAs` | |
| 2268 | + * and the X handle stopped printing once the source plugin was off. Each | |
| 2269 | + * store is filled only where it is still empty. | |
| 2270 | + * | |
| 2271 | + * @param array $profiles platform => profile URL | |
| 2272 | + * @return bool True if anything was written | |
| 2273 | + */ | |
| 2274 | + private function migrate_social_profiles(array $profiles): bool { | |
| 2275 | + $profiles = array_filter(array_map( | |
| 2276 | + static fn($url): string => is_scalar($url) ? trim((string) $url) : '', | |
| 2277 | + $profiles | |
| 2278 | + )); | |
| 2279 | + if (empty($profiles)) { | |
| 2280 | + return false; | |
| 2281 | + } | |
| 2282 | + | |
| 2283 | + $written = false; | |
| 2284 | + | |
| 2285 | + $social = get_option('thinkrank_social_media_settings', []); | |
| 2286 | + $social = is_array($social) ? $social : []; | |
| 2287 | + $social_updated = false; | |
| 2288 | + foreach ($profiles as $platform => $url) { | |
| 2289 | + $key = sanitize_key((string) $platform) . '_url'; | |
| 2290 | + if (empty($social[$key])) { | |
| 2291 | + $social[$key] = esc_url_raw($url); | |
| 2292 | + $social_updated = true; | |
| 2293 | + } | |
| 2294 | + } | |
| 2295 | + if ($social_updated) { | |
| 2296 | + update_option('thinkrank_social_media_settings', $social); | |
| 2297 | + $written = true; | |
| 2298 | + } | |
| 2299 | + | |
| 2300 | + // The Organization node's sameAs. | |
| 2301 | + $schema_platforms = ['facebook', 'twitter', 'linkedin', 'instagram', 'youtube', 'pinterest', 'whatsapp', 'telegram']; | |
| 2302 | + $manager = $this->create_schema_manager(); | |
| 2303 | + if ($manager !== null) { | |
| 2304 | + $current = $manager->get_settings('site'); | |
| 2305 | + $updates = []; | |
| 2306 | + foreach ($profiles as $platform => $url) { | |
| 2307 | + $field = 'organization_social_' . $platform; | |
| 2308 | + if (in_array($platform, $schema_platforms, true) && filter_var($url, FILTER_VALIDATE_URL) && empty($current[$field])) { | |
| 2309 | + $updates[$field] = esc_url_raw($url); | |
| 2310 | + } | |
| 2311 | + } | |
| 2312 | + if (!empty($updates) && $manager->save_settings('site', null, $updates)) { | |
| 2313 | + $written = true; | |
| 2314 | + } | |
| 2315 | + } | |
| 2316 | + | |
| 2317 | + // twitter:site, from an X/Twitter profile URL. | |
| 2318 | + $handle = self::twitter_handle_from_url((string) ($profiles['twitter'] ?? '')); | |
| 2319 | + if ($handle !== '' && class_exists('ThinkRank\\SEO\\Social_Meta_Manager')) { | |
| 2320 | + $social_meta = new \ThinkRank\SEO\Social_Meta_Manager(); | |
| 2321 | + $current = $social_meta->get_settings('site'); | |
| 2322 | + if (empty($current['twitter_username']) && $social_meta->save_settings('site', null, ['twitter_username' => $handle])) { | |
| 2323 | + $written = true; | |
| 2324 | + } | |
| 2325 | + } | |
| 2326 | + | |
| 2327 | + return $written; | |
| 2328 | + } | |
| 2329 | + | |
| 2330 | + /** | |
| 2331 | + * The handle in an x.com / twitter.com profile URL, or '' when the URL is | |
| 2332 | + * not one (`https://x.com/arsenal` → `arsenal`). Handles are 1-15 letters, | |
| 2333 | + * digits or underscores, which is also what Social Meta accepts. | |
| 2334 | + * | |
| 2335 | + * @param string $url Profile URL | |
| 2336 | + */ | |
| 2337 | + public static function twitter_handle_from_url(string $url): string { | |
| 2338 | + $host = strtolower((string) wp_parse_url($url, PHP_URL_HOST)); | |
| 2339 | + if (!in_array(preg_replace('/^(www\.|mobile\.)/', '', $host), ['x.com', 'twitter.com'], true)) { | |
| 2340 | + return ''; | |
| 2341 | + } | |
| 2342 | + $segment = explode('/', trim((string) wp_parse_url($url, PHP_URL_PATH), '/'))[0] ?? ''; | |
| 2343 | + $segment = ltrim(rawurldecode($segment), '@'); | |
| 2344 | + | |
| 2345 | + return preg_match('/^[A-Za-z0-9_]{1,15}$/', $segment) ? $segment : ''; | |
| 2346 | + } | |
| 2347 | + | |
| 2348 | + /** | |
| 1988 | 2349 | * Migrate the source plugin's Pinterest site-verification code into |
| 1989 | 2350 | * ThinkRank's core `pinterest_site_verification` setting. Never overwrites |
| 1990 | 2351 | * a configured code. |
| 1991 | 2352 | * |
| @@ -2226,18 +2587,40 @@ | ||
| 2226 | 2587 | $skipped++; |
| 2227 | 2588 | } |
| 2228 | 2589 | } |
| 2229 | 2590 | |
| 2591 | + // Report the same has_more every other type does. Hardcoding false | |
| 2592 | + // here was invisible in the admin, which iterates 1..total_chunks, and | |
| 2593 | + // silently truncated the MCP/ability import, which loops on has_more | |
| 2594 | + // alone: a site with more than one chunk of rules got its first | |
| 2595 | + // hundred and a clean `complete`. | |
| 2596 | + $has_more = $page < (int) ($this->chunk_total($plugin, 'redirections')); | |
| 2597 | + | |
| 2230 | 2598 | return [ |
| 2231 | - 'status' => 'complete', | |
| 2232 | - 'message' => sprintf('Migrated %d redirections, skipped %d (page %d)', $processed, $skipped, $page), | |
| 2233 | - 'has_more' => false, | |
| 2234 | - 'processed' => $processed, | |
| 2235 | - 'skipped' => $skipped, | |
| 2599 | + 'status' => $has_more ? 'processing' : 'complete', | |
| 2600 | + 'message' => sprintf('Migrated %d redirections, skipped %d (page %d)', $processed, $skipped, $page), | |
| 2601 | + 'has_more' => $has_more, | |
| 2602 | + 'page' => $page, | |
| 2603 | + 'total_chunks' => $this->chunk_total($plugin, 'redirections'), | |
| 2604 | + 'processed' => $processed, | |
| 2605 | + 'skipped' => $skipped, | |
| 2236 | 2606 | ]; |
| 2237 | 2607 | } |
| 2238 | 2608 | |
| 2239 | 2609 | /** |
| 2610 | + * How many chunks the manifest declares for a type, or 0 when unknown. | |
| 2611 | + * | |
| 2612 | + * @param string $plugin Source slug. | |
| 2613 | + * @param string $type Snapshot type. | |
| 2614 | + * @return int | |
| 2615 | + */ | |
| 2616 | + private function chunk_total(string $plugin, string $type): int { | |
| 2617 | + $manifest = Snapshot_Store::get_manifest($plugin); | |
| 2618 | + | |
| 2619 | + return (int) ($manifest['types'][$type]['total_chunks'] ?? 0); | |
| 2620 | + } | |
| 2621 | + | |
| 2622 | + /** | |
| 2240 | 2623 | * Migrate a chunk of logged 404 hits into ThinkRank Pro's 404 Monitor. |
| 2241 | 2624 | * Pro-gated exactly like migrate_redirections(). |
| 2242 | 2625 | * |
| 2243 | 2626 | * @param string $plugin Plugin slug |
| @@ -2287,18 +2670,130 @@ | ||
| 2287 | 2670 | $skipped++; |
| 2288 | 2671 | } |
| 2289 | 2672 | } |
| 2290 | 2673 | |
| 2674 | + $has_more = $page < $this->chunk_total($plugin, '404_logs'); | |
| 2675 | + | |
| 2291 | 2676 | return [ |
| 2292 | - 'status' => 'complete', | |
| 2293 | - 'message' => sprintf('Migrated %d 404 logs, skipped %d (page %d)', $processed, $skipped, $page), | |
| 2294 | - 'has_more' => false, | |
| 2295 | - 'processed' => $processed, | |
| 2296 | - 'skipped' => $skipped, | |
| 2677 | + 'status' => $has_more ? 'processing' : 'complete', | |
| 2678 | + 'message' => sprintf('Migrated %d 404 logs, skipped %d (page %d)', $processed, $skipped, $page), | |
| 2679 | + 'has_more' => $has_more, | |
| 2680 | + 'page' => $page, | |
| 2681 | + 'total_chunks' => $this->chunk_total($plugin, '404_logs'), | |
| 2682 | + 'processed' => $processed, | |
| 2683 | + 'skipped' => $skipped, | |
| 2297 | 2684 | ]; |
| 2298 | 2685 | } |
| 2299 | 2686 | |
| 2300 | 2687 | /** |
| 2688 | + * Rewrite a chunk of posts' Rank Math FAQ / HowTo blocks into ThinkRank's | |
| 2689 | + * own blocks. | |
| 2690 | + * | |
| 2691 | + * Unlike every other type here this does not write meta — it edits | |
| 2692 | + * `post_content` in place, because that is where the blocks live. The | |
| 2693 | + * snapshot chunk carries post ids only, so the conversion always runs | |
| 2694 | + * against the post as it stands now rather than a stale copy. | |
| 2695 | + * | |
| 2696 | + * The conflict strategy is deliberately ignored. A Rank Math block and a | |
| 2697 | + * ThinkRank block are not two values competing for one field: the Rank Math | |
| 2698 | + * one is broken markup that needs replacing, and any ThinkRank block | |
| 2699 | + * already in the post is simply left alone by the converter. | |
| 2700 | + * | |
| 2701 | + * @param string $plugin Plugin slug | |
| 2702 | + * @param int $page Chunk number | |
| 2703 | + * @return array Migration result | |
| 2704 | + */ | |
| 2705 | + private function migrate_content_blocks(string $plugin, int $page): array { | |
| 2706 | + $chunk = Snapshot_Store::read_chunk($plugin, Block_Converter::TYPE, $page); | |
| 2707 | + $total_chunks = $this->chunk_total($plugin, Block_Converter::TYPE); | |
| 2708 | + | |
| 2709 | + if ($chunk === null || empty($chunk)) { | |
| 2710 | + $has_more = $page < $total_chunks; | |
| 2711 | + | |
| 2712 | + return [ | |
| 2713 | + 'status' => $has_more ? 'processing' : 'complete', | |
| 2714 | + 'message' => 'No content blocks in chunk', | |
| 2715 | + 'has_more' => $has_more, | |
| 2716 | + 'page' => $page, | |
| 2717 | + 'processed' => 0, | |
| 2718 | + 'skipped' => 0, | |
| 2719 | + 'failed' => 0, | |
| 2720 | + 'failures' => [], | |
| 2721 | + ]; | |
| 2722 | + } | |
| 2723 | + | |
| 2724 | + // Rewriting a few hundred posts is well past the default execution | |
| 2725 | + // window on shared hosting, and a timeout mid-chunk would leave the | |
| 2726 | + // migration looking stalled. | |
| 2727 | + if (function_exists('set_time_limit')) { | |
| 2728 | + @set_time_limit(300); // phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged | |
| 2729 | + } | |
| 2730 | + | |
| 2731 | + $processed = 0; | |
| 2732 | + $skipped = 0; | |
| 2733 | + $failed = 0; | |
| 2734 | + $failures = []; | |
| 2735 | + $blocks = 0; | |
| 2736 | + | |
| 2737 | + foreach ($chunk as $record) { | |
| 2738 | + $post_id = (int) ($record['object_id'] ?? ($record['data']['post_id'] ?? 0)); | |
| 2739 | + if ($post_id < 1) { | |
| 2740 | + $skipped++; | |
| 2741 | + continue; | |
| 2742 | + } | |
| 2743 | + | |
| 2744 | + $result = Block_Converter::convert_post($post_id); | |
| 2745 | + | |
| 2746 | + if ('converted' === $result['status']) { | |
| 2747 | + $processed++; | |
| 2748 | + $blocks += $result['converted']; | |
| 2749 | + continue; | |
| 2750 | + } | |
| 2751 | + | |
| 2752 | + // A post the converter refused (broken block markup, a PCRE | |
| 2753 | + // failure) was left untouched and still holds its Rank Math | |
| 2754 | + // blocks. Folding it into `skipped` made it indistinguishable from | |
| 2755 | + // a post that was simply already converted, so it is counted and | |
| 2756 | + // named on its own. | |
| 2757 | + if ('error' === $result['status']) { | |
| 2758 | + $failed++; | |
| 2759 | + $failures[] = ['post_id' => $post_id, 'message' => $result['message']]; | |
| 2760 | + continue; | |
| 2761 | + } | |
| 2762 | + | |
| 2763 | + // `unchanged` is the normal outcome of a re-run, not a failure. | |
| 2764 | + $skipped++; | |
| 2765 | + } | |
| 2766 | + | |
| 2767 | + $has_more = $page < $total_chunks; | |
| 2768 | + | |
| 2769 | + if (!$has_more) { | |
| 2770 | + // The detector caches its scan for an hour; without clearing it the | |
| 2771 | + // Migration screen keeps offering blocks that are no longer there. | |
| 2772 | + (new Import_Detector())->clear_cache(); | |
| 2773 | + } | |
| 2774 | + | |
| 2775 | + return [ | |
| 2776 | + 'status' => $has_more ? 'processing' : 'complete', | |
| 2777 | + 'message' => sprintf( | |
| 2778 | + 'Converted %d FAQ/HowTo blocks in %d posts, skipped %d, failed %d (page %d)', | |
| 2779 | + $blocks, | |
| 2780 | + $processed, | |
| 2781 | + $skipped, | |
| 2782 | + $failed, | |
| 2783 | + $page | |
| 2784 | + ), | |
| 2785 | + 'has_more' => $has_more, | |
| 2786 | + 'page' => $page, | |
| 2787 | + 'total_chunks' => $total_chunks, | |
| 2788 | + 'processed' => $processed, | |
| 2789 | + 'skipped' => $skipped, | |
| 2790 | + 'failed' => $failed, | |
| 2791 | + 'failures' => $failures, | |
| 2792 | + ]; | |
| 2793 | + } | |
| 2794 | + | |
| 2795 | + /** | |
| 2301 | 2796 | * Build ThinkRank Pro's Redirections store, when Pro is active. |
| 2302 | 2797 | * |
| 2303 | 2798 | * Split out (and protected) so tests can substitute a fake — the real store |
| 2304 | 2799 | * writes to Pro's tables. Pro lazily creates them via Schema::ensure(). |
| @@ -2698,11 +3193,39 @@ | ||
| 2698 | 3193 | foreach ($extended as $key => $value) { |
| 2699 | 3194 | if (in_array($key, self::HANDLED_EXTENDED_SETTINGS, true) || empty($value)) { |
| 2700 | 3195 | continue; |
| 2701 | 3196 | } |
| 3197 | + // A bucket the migrator applies part of blocks cleanup only for | |
| 3198 | + // the part it does not. | |
| 3199 | + if (is_array($value) && empty($this->unapplied_remainder((string) $key, $value))) { | |
| 3200 | + continue; | |
| 3201 | + } | |
| 3202 | + // Custom schemas have a home only in ThinkRank Pro. Without it | |
| 3203 | + // they are preserved, and cleanup would delete the source's | |
| 3204 | + // copy, including every post's own schemas. | |
| 3205 | + if ($key === 'custom_schemas') { | |
| 3206 | + if ($this->create_custom_schema_repository() !== null) { | |
| 3207 | + continue; | |
| 3208 | + } | |
| 3209 | + $count = count(is_array($value['entries'] ?? null) ? $value['entries'] : []) + (int) ($value['post_count'] ?? 0); | |
| 3210 | + $buckets[] = [ | |
| 3211 | + 'key' => 'settings.custom_schemas', | |
| 3212 | + 'label' => (string) self::extended_bucket_label('custom_schemas'), | |
| 3213 | + 'count' => max(1, $count), | |
| 3214 | + ]; | |
| 3215 | + continue; | |
| 3216 | + } | |
| 3217 | + if ($key === 'schema_templates' && is_array($value)) { | |
| 3218 | + $buckets[] = [ | |
| 3219 | + 'key' => 'settings.schema_templates', | |
| 3220 | + 'label' => (string) self::extended_bucket_label('schema_templates'), | |
| 3221 | + 'count' => count($value), | |
| 3222 | + ]; | |
| 3223 | + continue; | |
| 3224 | + } | |
| 2702 | 3225 | $buckets[] = [ |
| 2703 | 3226 | 'key' => 'settings.' . $key, |
| 2704 | - 'label' => (string) $key, | |
| 3227 | + 'label' => self::extended_bucket_label((string) $key) ?? ucfirst(str_replace('_', ' ', (string) $key)), | |
| 2705 | 3228 | 'count' => 1, |
| 2706 | 3229 | ]; |
| 2707 | 3230 | } |
| 2708 | 3231 | } |
| @@ -2707,8 +3230,209 @@ | ||
| 2707 | 3230 | } |
| 2708 | 3231 | } |
| 2709 | 3232 | |
| 2710 | 3233 | return $buckets; |
| 3234 | + } | |
| 3235 | + | |
| 3236 | + /** | |
| 3237 | + * Save canonical custom schema entries into ThinkRank Pro's Custom Schema. | |
| 3238 | + * | |
| 3239 | + * Each entry's id is derived from the source and its `key`, so running | |
| 3240 | + * the migration again finds the entry it wrote and leaves it — including | |
| 3241 | + * any edit made since. An entry the Repository refuses (an invalid | |
| 3242 | + * condition, an oversized document) is skipped; the source still holds | |
| 3243 | + * it, and the snapshot keeps the rendered JSON. | |
| 3244 | + * | |
| 3245 | + * @param string $plugin Source plugin slug | |
| 3246 | + * @param array $entries Canonical entries {key, title, enabled, json, conditions} | |
| 3247 | + * @return int Entries written | |
| 3248 | + */ | |
| 3249 | + private function migrate_custom_schemas(string $plugin, array $entries): int { | |
| 3250 | + $repository = $this->create_custom_schema_repository(); | |
| 3251 | + if ($repository === null) { | |
| 3252 | + return 0; | |
| 3253 | + } | |
| 3254 | + | |
| 3255 | + $written = 0; | |
| 3256 | + foreach ($entries as $entry) { | |
| 3257 | + if (!is_array($entry) || trim((string) ($entry['json'] ?? '')) === '') { | |
| 3258 | + continue; | |
| 3259 | + } | |
| 3260 | + $id = 'cs_' . substr(md5($plugin . '|' . (string) ($entry['key'] ?? wp_json_encode($entry))), 0, 12); | |
| 3261 | + if ($repository->get($id) !== null) { | |
| 3262 | + continue; | |
| 3263 | + } | |
| 3264 | + | |
| 3265 | + try { | |
| 3266 | + $repository->save([ | |
| 3267 | + 'id' => $id, | |
| 3268 | + 'title' => (string) ($entry['title'] ?? ''), | |
| 3269 | + 'enabled' => !empty($entry['enabled']), | |
| 3270 | + 'json' => (string) $entry['json'], | |
| 3271 | + 'conditions' => is_array($entry['conditions'] ?? null) ? $entry['conditions'] : ['include' => [], 'exclude' => []], | |
| 3272 | + ]); | |
| 3273 | + $written++; | |
| 3274 | + } catch (\Throwable $e) { | |
| 3275 | + continue; | |
| 3276 | + } | |
| 3277 | + } | |
| 3278 | + | |
| 3279 | + return $written; | |
| 3280 | + } | |
| 3281 | + | |
| 3282 | + /** | |
| 3283 | + * ThinkRank Pro's Custom Schema repository, when Pro is active. | |
| 3284 | + * | |
| 3285 | + * @return object|null | |
| 3286 | + */ | |
| 3287 | + protected function create_custom_schema_repository(): ?object { | |
| 3288 | + return class_exists('ThinkRank\\Pro\\Schema\\Repository') ? new \ThinkRank\Pro\Schema\Repository() : null; | |
| 3289 | + } | |
| 3290 | + | |
| 3291 | + /** | |
| 3292 | + * Turn on ThinkRank Pro's Markdown for AI with the source's post types | |
| 3293 | + * (extended.markdown_for_ai), while its settings were never saved. | |
| 3294 | + * | |
| 3295 | + * @param array $extended Canonical settings `extended` payload | |
| 3296 | + * @return bool True if written | |
| 3297 | + */ | |
| 3298 | + private function migrate_markdown_for_ai(array $extended): bool { | |
| 3299 | + $markdown = $extended['markdown_for_ai'] ?? []; | |
| 3300 | + if (!is_array($markdown) || empty($markdown['enabled']) || !class_exists('ThinkRank\\Pro\\Markdown_For_AI\\Settings')) { | |
| 3301 | + return false; | |
| 3302 | + } | |
| 3303 | + if (get_option('thinkrank_pro_markdown_for_ai', null) !== null) { | |
| 3304 | + return false; | |
| 3305 | + } | |
| 3306 | + | |
| 3307 | + $types = array_values(array_filter( | |
| 3308 | + array_map('strval', is_array($markdown['post_types'] ?? null) ? $markdown['post_types'] : ['post']), | |
| 3309 | + 'post_type_exists' | |
| 3310 | + )); | |
| 3311 | + | |
| 3312 | + return (bool) (new \ThinkRank\Pro\Markdown_For_AI\Settings())->save([ | |
| 3313 | + 'enabled' => true, | |
| 3314 | + 'post_types' => !empty($types) ? $types : ['post'], | |
| 3315 | + ]); | |
| 3316 | + } | |
| 3317 | + | |
| 3318 | + /** | |
| 3319 | + * What a preserved settings bucket holds, as the cleanup warning names it. | |
| 3320 | + * | |
| 3321 | + * The warning exists so someone can decide whether to delete their old | |
| 3322 | + * plugin's data; printing `code_injection` does not help them decide | |
| 3323 | + * (#900). Every bucket an exporter can leave unapplied has an entry here, | |
| 3324 | + * and a test fails when one is added without. | |
| 3325 | + * | |
| 3326 | + * @param string $key Extended bucket key | |
| 3327 | + * @return string|null Label, or null when the bucket has none | |
| 3328 | + */ | |
| 3329 | + public static function extended_bucket_label(string $key): ?string { | |
| 3330 | + $labels = [ | |
| 3331 | + 'code_injection' => __('Header, body and footer code', 'thinkrank'), | |
| 3332 | + 'redirect_settings' => __('Redirect behaviour settings', 'thinkrank'), | |
| 3333 | + 'taxonomy_settings' => __('Taxonomy templates and images', 'thinkrank'), | |
| 3334 | + 'post_type_archive_settings' => __('Post type archive templates and images', 'thinkrank'), | |
| 3335 | + 'no_category_base' => __('Removing /category/ from category URLs', 'thinkrank'), | |
| 3336 | + 'webmaster_tools' => __('Webmaster tools verification codes', 'thinkrank'), | |
| 3337 | + 'access_control' => __('Role access settings', 'thinkrank'), | |
| 3338 | + 'og_frontpage_title' => __('Homepage social title', 'thinkrank'), | |
| 3339 | + 'og_frontpage_desc' => __('Homepage social description', 'thinkrank'), | |
| 3340 | + 'og_frontpage_image' => __('Homepage social image', 'thinkrank'), | |
| 3341 | + 'custom_schemas' => __('Custom schemas (need ThinkRank Pro)', 'thinkrank'), | |
| 3342 | + 'schema_templates' => __('Schema templates with no ThinkRank equivalent', 'thinkrank'), | |
| 3343 | + ]; | |
| 3344 | + | |
| 3345 | + return $labels[$key] ?? null; | |
| 3346 | + } | |
| 3347 | + | |
| 3348 | + /** | |
| 3349 | + * What is left of a partially-applied extended bucket once the keys | |
| 3350 | + * migrate_archive_noindex() consumes are removed. | |
| 3351 | + * | |
| 3352 | + * `taxonomy_settings` and `post_type_archive_settings` carry a `noindex` | |
| 3353 | + * per context that is applied, beside templates and images that are not. | |
| 3354 | + * Any other bucket is returned whole. | |
| 3355 | + * | |
| 3356 | + * @param string $key Extended bucket key | |
| 3357 | + * @param array $value Bucket contents | |
| 3358 | + * @return array The unapplied part | |
| 3359 | + */ | |
| 3360 | + private function unapplied_remainder(string $key, array $value): array { | |
| 3361 | + if ($key !== 'taxonomy_settings' && $key !== 'post_type_archive_settings') { | |
| 3362 | + return $value; | |
| 3363 | + } | |
| 3364 | + | |
| 3365 | + $remainder = []; | |
| 3366 | + foreach ($value as $context => $settings) { | |
| 3367 | + if (!is_array($settings)) { | |
| 3368 | + continue; | |
| 3369 | + } | |
| 3370 | + // Only the blog index has an archive entity to receive a noindex. | |
| 3371 | + $applied = $key === 'taxonomy_settings' || (string) $context === 'post'; | |
| 3372 | + $rest = $applied ? array_diff_key($settings, ['noindex' => true]) : $settings; | |
| 3373 | + if (!empty($rest)) { | |
| 3374 | + $remainder[$context] = $rest; | |
| 3375 | + } | |
| 3376 | + } | |
| 3377 | + | |
| 3378 | + return $remainder; | |
| 3379 | + } | |
| 3380 | + | |
| 3381 | + /** | |
| 3382 | + * Apply per-taxonomy and blog-index noindex through the Content Type | |
| 3383 | + * Matrix (`taxonomy:<slug>` / `archive:blog`), the entities the front end | |
| 3384 | + * reads robots directives for on those archives. | |
| 3385 | + * | |
| 3386 | + * Reads `extended.taxonomy_settings[<taxonomy>].noindex` (Yoast's and | |
| 3387 | + * Slim SEO's shape) and `extended.post_type_archive_settings.post.noindex`. | |
| 3388 | + * An entity whose robots were already customised in ThinkRank is left | |
| 3389 | + * alone, and a source that did NOT noindex an archive writes nothing — | |
| 3390 | + * "indexed" is the default and restating it would pin the entity. | |
| 3391 | + * | |
| 3392 | + * @param array $extended Canonical settings `extended` payload | |
| 3393 | + * @return bool True if any entity was written | |
| 3394 | + */ | |
| 3395 | + private function migrate_archive_noindex(array $extended): bool { | |
| 3396 | + if (!class_exists('ThinkRank\\SEO\\Content_Type_Settings')) { | |
| 3397 | + return false; | |
| 3398 | + } | |
| 3399 | + | |
| 3400 | + // Only public taxonomies are matrix entities (Content_Type_Settings::get_entities()). | |
| 3401 | + $entities = []; | |
| 3402 | + $taxonomies = is_array($extended['taxonomy_settings'] ?? null) ? $extended['taxonomy_settings'] : []; | |
| 3403 | + foreach ($taxonomies as $taxonomy => $settings) { | |
| 3404 | + if (!is_array($settings) || empty($settings['noindex']) || !taxonomy_exists((string) $taxonomy)) { | |
| 3405 | + continue; | |
| 3406 | + } | |
| 3407 | + $object = get_taxonomy((string) $taxonomy); | |
| 3408 | + if (is_object($object) && !empty($object->public)) { | |
| 3409 | + $entities[] = \ThinkRank\SEO\Content_Type_Settings::PREFIX_TAXONOMY . $taxonomy; | |
| 3410 | + } | |
| 3411 | + } | |
| 3412 | + $archives = is_array($extended['post_type_archive_settings'] ?? null) ? $extended['post_type_archive_settings'] : []; | |
| 3413 | + if (!empty($archives['post']['noindex'])) { | |
| 3414 | + $entities[] = \ThinkRank\SEO\Content_Type_Settings::ENTITY_BLOG_INDEX; | |
| 3415 | + } | |
| 3416 | + | |
| 3417 | + $written = false; | |
| 3418 | + foreach ($entities as $entity_key) { | |
| 3419 | + $existing = \ThinkRank\SEO\Content_Type_Settings::get_entity_settings($entity_key); | |
| 3420 | + if (!empty($existing['robots_meta_enabled'])) { | |
| 3421 | + continue; | |
| 3422 | + } | |
| 3423 | + $robots = \ThinkRank\SEO\Content_Type_Settings::default_robots_meta($entity_key); | |
| 3424 | + $robots['index'] = false; | |
| 3425 | + $robots['noindex'] = true; | |
| 3426 | + if (\ThinkRank\SEO\Content_Type_Settings::update_entity_settings($entity_key, [ | |
| 3427 | + 'robots_meta_enabled' => true, | |
| 3428 | + 'robots_meta' => $robots, | |
| 3429 | + ])) { | |
| 3430 | + $written = true; | |
| 3431 | + } | |
| 3432 | + } | |
| 3433 | + | |
| 3434 | + return $written; | |
| 2711 | 3435 | } |
| 2712 | 3436 | |
| 2713 | 3437 | /** |
| 2714 | 3438 | * Fold post IDs the source excluded from its sitemap into ThinkRank's |