| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
/** |
| 8 |
* Given a regex match against an old permalink structure, resolves and |
| 9 |
* validates a currently-published WordPress post: extracts named captures, |
| 10 |
* looks up the post by ID or by slug, and checks it against the structural |
| 11 |
* constraints implied by the captures (post type, date, author, category, |
| 12 |
* slug). |
| 13 |
* |
| 14 |
* Extracted from ABJ_404_Solution_OldPermalinkStructureResolver because |
| 15 |
* resolving a regex match to a validated post is a distinct concern from |
| 16 |
* compiling a structure into a regex or gathering candidate structures. |
| 17 |
*/ |
| 18 |
class ABJ_404_Solution_OldPermalinkPostResolver { |
| 19 |
|
| 20 |
/** @var ABJ_404_Solution_ContentRepositoryInterface */ |
| 21 |
private $contentRepository; |
| 22 |
|
| 23 |
/** @var ABJ_404_Solution_Logging */ |
| 24 |
private $logger; |
| 25 |
|
| 26 |
/** |
| 27 |
* @param ABJ_404_Solution_ContentRepositoryInterface $contentRepository |
| 28 |
* @param ABJ_404_Solution_Logging $logger |
| 29 |
*/ |
| 30 |
public function __construct( |
| 31 |
ABJ_404_Solution_ContentRepositoryInterface $contentRepository, |
| 32 |
ABJ_404_Solution_Logging $logger |
| 33 |
) { |
| 34 |
$this->contentRepository = $contentRepository; |
| 35 |
$this->logger = $logger; |
| 36 |
} |
| 37 |
|
| 38 |
/** |
| 39 |
* @param array<int|string, mixed> $matches Raw preg_match() result. |
| 40 |
* @param array{structure: string, post_types: array<int, string>} $candidate |
| 41 |
* @param array<string, mixed> $options |
| 42 |
* @return int|null |
| 43 |
*/ |
| 44 |
public function resolve(array $matches, array $candidate, array $options): ?int { |
| 45 |
$captures = $this->namedCaptures($matches); |
| 46 |
|
| 47 |
return isset($captures['post_id']) |
| 48 |
? $this->resolveByPostId($captures, $candidate, $options) |
| 49 |
: $this->resolveBySlug($captures, $candidate, $options); |
| 50 |
} |
| 51 |
|
| 52 |
/** |
| 53 |
* @param array<int|string, mixed> $matches |
| 54 |
* @return array<string, string> |
| 55 |
*/ |
| 56 |
private function namedCaptures(array $matches): array { |
| 57 |
$captures = array(); |
| 58 |
foreach ($matches as $key => $value) { |
| 59 |
if (is_string($key) && is_scalar($value)) { |
| 60 |
$captures[$key] = (string)$value; |
| 61 |
} |
| 62 |
} |
| 63 |
return $captures; |
| 64 |
} |
| 65 |
|
| 66 |
/** |
| 67 |
* @param array<string, string> $captures |
| 68 |
* @param array{structure: string, post_types: array<int, string>} $candidate |
| 69 |
* @param array<string, mixed> $options |
| 70 |
* @return int|null |
| 71 |
*/ |
| 72 |
private function resolveByPostId(array $captures, array $candidate, array $options): ?int { |
| 73 |
$postId = absint($captures['post_id'] ?? 0); |
| 74 |
if ($postId <= 0) { |
| 75 |
return null; |
| 76 |
} |
| 77 |
|
| 78 |
$post = $this->loadPost($postId); |
| 79 |
if ($post === null || !$this->postMatchesConstraints($post, $captures, $candidate, $options)) { |
| 80 |
return null; |
| 81 |
} |
| 82 |
|
| 83 |
$this->logger->debugMessage('Old permalink structure resolved by post ID: ' . $postId); |
| 84 |
return $postId; |
| 85 |
} |
| 86 |
|
| 87 |
/** |
| 88 |
* @param array<string, string> $captures |
| 89 |
* @param array{structure: string, post_types: array<int, string>} $candidate |
| 90 |
* @param array<string, mixed> $options |
| 91 |
* @return int|null |
| 92 |
*/ |
| 93 |
private function resolveBySlug(array $captures, array $candidate, array $options): ?int { |
| 94 |
$slug = $captures['postname'] ?? ($captures['pagename'] ?? ''); |
| 95 |
if ($slug === '') { |
| 96 |
return null; |
| 97 |
} |
| 98 |
|
| 99 |
$rows = $this->contentRepository->getPublishedPagesAndPostsIDs(array( |
| 100 |
'slug' => $slug, |
| 101 |
'limit_results' => '11', |
| 102 |
)); |
| 103 |
$matches = array(); |
| 104 |
foreach ($rows as $row) { |
| 105 |
$postId = $this->idFromRow($row); |
| 106 |
if ($postId <= 0) { |
| 107 |
continue; |
| 108 |
} |
| 109 |
$post = $this->loadPost($postId, $row); |
| 110 |
if ($post !== null && $this->postMatchesConstraints($post, $captures, $candidate, $options)) { |
| 111 |
$matches[] = $postId; |
| 112 |
} |
| 113 |
} |
| 114 |
|
| 115 |
$matches = array_values(array_unique($matches)); |
| 116 |
if (count($matches) !== 1) { |
| 117 |
if (count($matches) > 1) { |
| 118 |
$this->logger->debugMessage('Old permalink structure ambiguous for slug: ' . $slug); |
| 119 |
} |
| 120 |
return null; |
| 121 |
} |
| 122 |
|
| 123 |
$this->logger->debugMessage('Old permalink structure resolved by slug: ' . $slug); |
| 124 |
return (int)$matches[0]; |
| 125 |
} |
| 126 |
|
| 127 |
/** |
| 128 |
* @param mixed $row |
| 129 |
* @return int |
| 130 |
*/ |
| 131 |
private function idFromRow($row): int { |
| 132 |
if (is_object($row) && isset($row->id) && is_scalar($row->id)) { |
| 133 |
return absint($row->id); |
| 134 |
} |
| 135 |
if (is_object($row) && isset($row->ID) && is_scalar($row->ID)) { |
| 136 |
return absint($row->ID); |
| 137 |
} |
| 138 |
if (is_array($row) && isset($row['id']) && is_scalar($row['id'])) { |
| 139 |
return absint($row['id']); |
| 140 |
} |
| 141 |
if (is_array($row) && isset($row['ID']) && is_scalar($row['ID'])) { |
| 142 |
return absint($row['ID']); |
| 143 |
} |
| 144 |
return 0; |
| 145 |
} |
| 146 |
|
| 147 |
/** |
| 148 |
* @param int $postId |
| 149 |
* @param mixed|null $fallback |
| 150 |
* @return object|null |
| 151 |
*/ |
| 152 |
private function loadPost(int $postId, $fallback = null): ?object { |
| 153 |
if (function_exists('get_post')) { |
| 154 |
$post = get_post($postId); |
| 155 |
if (is_object($post)) { |
| 156 |
return $post; |
| 157 |
} |
| 158 |
} |
| 159 |
return is_object($fallback) ? $fallback : null; |
| 160 |
} |
| 161 |
|
| 162 |
/** |
| 163 |
* @param object $post |
| 164 |
* @param array<string, string> $captures |
| 165 |
* @param array{structure: string, post_types: array<int, string>} $candidate |
| 166 |
* @param array<string, mixed> $options |
| 167 |
* @return bool |
| 168 |
*/ |
| 169 |
private function postMatchesConstraints(object $post, array $captures, array $candidate, array $options): bool { |
| 170 |
$postId = isset($post->ID) ? absint($post->ID) : (isset($post->id) ? absint($post->id) : 0); |
| 171 |
$status = isset($post->post_status) ? (string)$post->post_status |
| 172 |
: (function_exists('get_post_status') ? (string)get_post_status($postId) : ''); |
| 173 |
if (!in_array($status, array('publish', 'published'), true)) { |
| 174 |
return false; |
| 175 |
} |
| 176 |
|
| 177 |
$postType = isset($post->post_type) ? sanitize_key((string)$post->post_type) : ''; |
| 178 |
$allowedTypes = $this->recognizedPostTypes($options, $candidate['post_types']); |
| 179 |
if ($postType === '' || !in_array($postType, $allowedTypes, true)) { |
| 180 |
return false; |
| 181 |
} |
| 182 |
|
| 183 |
// Positive-evidence check: when the captured old permalink specifies |
| 184 |
// a slug segment, the candidate post must actually carry a matching |
| 185 |
// post_name, not merely "no post_name to contradict it". The DB |
| 186 |
// query in getPublishedPagesAndPostsIDs() usually pre-filters by |
| 187 |
// slug, but for a UTF8MB4 slug it drops the SQL-level slug clause |
| 188 |
// entirely (PublishedContentRepository::buildPostSlugClause()) and |
| 189 |
// returns every published post/page of the recognized types -- this |
| 190 |
// check is the only remaining filter in that path. An empty |
| 191 |
// post_name must not be treated as "no evidence against a match"; |
| 192 |
// it must be treated as "no evidence for one". |
| 193 |
$postName = isset($post->post_name) ? (string)$post->post_name : ''; |
| 194 |
$slugCapture = $captures['postname'] ?? ($captures['pagename'] ?? null); |
| 195 |
if ($slugCapture !== null && $postName !== $slugCapture) { |
| 196 |
return false; |
| 197 |
} |
| 198 |
|
| 199 |
if (!$this->dateMatches($post, $captures)) { |
| 200 |
return false; |
| 201 |
} |
| 202 |
|
| 203 |
if (isset($captures['author']) && !$this->authorMatches($post, $captures['author'])) { |
| 204 |
return false; |
| 205 |
} |
| 206 |
|
| 207 |
if (isset($captures['category']) && !$this->categoryMatches($postId, $captures['category'])) { |
| 208 |
return false; |
| 209 |
} |
| 210 |
|
| 211 |
return true; |
| 212 |
} |
| 213 |
|
| 214 |
/** |
| 215 |
* @param array<string, mixed> $options |
| 216 |
* @param array<int, string> $candidatePostTypes |
| 217 |
* @return array<int, string> |
| 218 |
*/ |
| 219 |
private function recognizedPostTypes(array $options, array $candidatePostTypes): array { |
| 220 |
$raw = isset($options['recognized_post_types']) && is_scalar($options['recognized_post_types']) |
| 221 |
? (string)$options['recognized_post_types'] |
| 222 |
: "page\npost\nproduct"; |
| 223 |
$types = preg_split('/[\s,]+/', $raw) ?: array(); |
| 224 |
$types = array_values(array_filter(array_map(static function($type): string { |
| 225 |
return sanitize_key((string)$type); |
| 226 |
}, $types))); |
| 227 |
|
| 228 |
if (empty($types)) { |
| 229 |
$types = array('page', 'post', 'product'); |
| 230 |
} |
| 231 |
|
| 232 |
if (!empty($candidatePostTypes)) { |
| 233 |
$types = array_values(array_intersect($types, $candidatePostTypes)); |
| 234 |
} |
| 235 |
|
| 236 |
return $types; |
| 237 |
} |
| 238 |
|
| 239 |
/** |
| 240 |
* $post->post_date is a WP-stored value expressed in the site's |
| 241 |
* configured timezone (Settings > General), the same convention WP |
| 242 |
* core uses to generate the %year%/%monthnum%/%day% permalink tags |
| 243 |
* these captures came from. Parsing and formatting it both anchor to |
| 244 |
* the WP site timezone (SiteTimezone) rather than PHP's implicit |
| 245 |
* default timezone, matching the convention established by |
| 246 |
* RedirectScheduleTimezone -- raw strtotime()/date() would parse and |
| 247 |
* re-format using PHP's default timezone instead, which is fragile: |
| 248 |
* it only stays a no-op because parsing and formatting happen to use |
| 249 |
* the same implicit zone today, a coupling a future refactor could |
| 250 |
* easily break. |
| 251 |
* |
| 252 |
* @param object $post |
| 253 |
* @param array<string, string> $captures |
| 254 |
* @return bool |
| 255 |
*/ |
| 256 |
private function dateMatches(object $post, array $captures): bool { |
| 257 |
if (!isset($captures['year']) && !isset($captures['monthnum']) && !isset($captures['day'])) { |
| 258 |
return true; |
| 259 |
} |
| 260 |
$date = isset($post->post_date) ? (string)$post->post_date : ''; |
| 261 |
if ($date === '') { |
| 262 |
return false; |
| 263 |
} |
| 264 |
try { |
| 265 |
$postDateTime = new DateTimeImmutable($date, ABJ_404_Solution_SiteTimezone::resolve()); |
| 266 |
} catch (Exception $e) { |
| 267 |
$this->logger->warn('OldPermalinkPostResolver: unparseable post_date "' . $date . |
| 268 |
'" on post ID ' . (isset($post->ID) ? (string)$post->ID : 'unknown') . ': ' . $e->getMessage()); |
| 269 |
return false; |
| 270 |
} |
| 271 |
if (isset($captures['year']) && $postDateTime->format('Y') !== $captures['year']) { |
| 272 |
return false; |
| 273 |
} |
| 274 |
if (isset($captures['monthnum']) && $postDateTime->format('m') !== $captures['monthnum']) { |
| 275 |
return false; |
| 276 |
} |
| 277 |
if (isset($captures['day']) && $postDateTime->format('d') !== $captures['day']) { |
| 278 |
return false; |
| 279 |
} |
| 280 |
return true; |
| 281 |
} |
| 282 |
|
| 283 |
/** @param object $post @param string $authorSlug @return bool */ |
| 284 |
private function authorMatches(object $post, string $authorSlug): bool { |
| 285 |
if (!function_exists('get_userdata') || !isset($post->post_author)) { |
| 286 |
return false; |
| 287 |
} |
| 288 |
$user = get_userdata(absint($post->post_author)); |
| 289 |
return is_object($user) && isset($user->user_nicename) && (string)$user->user_nicename === $authorSlug; |
| 290 |
} |
| 291 |
|
| 292 |
/** @param int $postId @param string $categoryPath @return bool */ |
| 293 |
private function categoryMatches(int $postId, string $categoryPath): bool { |
| 294 |
if (!function_exists('has_category')) { |
| 295 |
return false; |
| 296 |
} |
| 297 |
$segments = array_values(array_filter(explode('/', $categoryPath))); |
| 298 |
if (empty($segments)) { |
| 299 |
return false; |
| 300 |
} |
| 301 |
return (bool)has_category(end($segments), $postId); |
| 302 |
} |
| 303 |
} |
| 304 |
|
