| 1 |
<?php |
| 2 |
/** |
| 3 |
* WPGraphQL delivery for the headless SEO payload. |
| 4 |
* |
| 5 |
* Registers one read-only field on every node type WPGraphQL exposes that this |
| 6 |
* plugin can describe, returning the object Metasync_Headless_Seo_Surface |
| 7 |
* resolved. No resolution logic lives here — this file is transport. |
| 8 |
* |
| 9 |
* **Where each entry point is attached, and why.** |
| 10 |
* |
| 11 |
* The surface has six entry points and WPGraphQL has a node for four of them. |
| 12 |
* Wherever a node exists the field goes on the node, because a node is an |
| 13 |
* address the frontend already has: it fetches a page with `nodeByUri(uri: …)` |
| 14 |
* and reads the SEO off whatever came back, in the same request, without |
| 15 |
* knowing which WordPress object type is behind the route. Attaching to the |
| 16 |
* node also means WPGraphQL has already decided the caller may see that object |
| 17 |
* before this field is ever reached. |
| 18 |
* |
| 19 |
* - a post type's `graphql_single_name` -> for_post() |
| 20 |
* - a taxonomy's `graphql_single_name` -> for_term() |
| 21 |
* - `User` -> for_author() |
| 22 |
* - `ContentType` -> for_post_type_archive() |
| 23 |
* |
| 24 |
* The front page and the blog index are the two that genuinely have no node of |
| 25 |
* their own — with `show_on_front = posts` there is no object at all — so they |
| 26 |
* get root fields on `RootQuery`: |
| 27 |
* |
| 28 |
* - `metasyncSeoForHomePage` -> for_home_page() |
| 29 |
* - `metasyncSeoForPostsPage` -> for_posts_page() |
| 30 |
* |
| 31 |
* Both take **no arguments**, which is the point rather than a convenience. |
| 32 |
* There is no id, name or slug a caller can hand a root field here, so no root |
| 33 |
* field can be pointed at an object the node path would have refused; every |
| 34 |
* caller-supplied identifier still arrives as a node WPGraphQL access-checked |
| 35 |
* first. See register_root_fields() for the alternatives that were rejected. |
| 36 |
* |
| 37 |
* **The field must never be able to break a page.** On a headless site the |
| 38 |
* frontend fetches content and SEO in a single GraphQL request, so a resolver |
| 39 |
* that throws does not degrade the SEO — it fails the whole content query and |
| 40 |
* the page renders as an error. Worse, a frontend using persisted (locked) query |
| 41 |
* documents cannot be quietly patched afterwards; every stored operation would |
| 42 |
* have to be rebuilt and redeployed. So: |
| 43 |
* |
| 44 |
* - the field and every child field are nullable; |
| 45 |
* - the resolver catches Throwable and returns null; |
| 46 |
* - registration itself is wrapped, so a malformed schema call cannot take the |
| 47 |
* whole schema down with it. |
| 48 |
* |
| 49 |
* Failing closed to null is the deliberate choice over an error a caller could |
| 50 |
* detect. A frontend that gets null renders the page with whatever SEO it can |
| 51 |
* derive itself; a frontend that gets an error renders nothing. |
| 52 |
* |
| 53 |
* The field is named `metasyncSeo` rather than `seo`. A headless install of the |
| 54 |
* kind this targets is very likely already serving a `seo` field from another |
| 55 |
* plugin, and colliding with it would break the site rather than extend it. The |
| 56 |
* root fields carry the same prefix for the same reason. |
| 57 |
* |
| 58 |
* Only render-facing values are exposed. No deployment revision, no internal |
| 59 |
* ids, no configuration — a headless GraphQL endpoint frequently has no auth |
| 60 |
* gate in front of it, so everything here should be safe to read anonymously. |
| 61 |
* Everything on the type is already public on the rendered page. |
| 62 |
* |
| 63 |
* @package Metasync |
| 64 |
* @subpackage Metasync/includes |
| 65 |
*/ |
| 66 |
|
| 67 |
if (!defined('ABSPATH')) { |
| 68 |
exit; # Exit if accessed directly |
| 69 |
} |
| 70 |
|
| 71 |
require_once dirname(__DIR__) . '/headless-refresh-history/class-metasync-headless-refresh-history-database.php'; |
| 72 |
|
| 73 |
class Metasync_Headless_Graphql |
| 74 |
{ |
| 75 |
/** |
| 76 |
* GraphQL object type name for the payload. |
| 77 |
*/ |
| 78 |
const TYPE_NAME = 'MetaSyncSeo'; |
| 79 |
|
| 80 |
/** |
| 81 |
* Field name added to each node type. |
| 82 |
*/ |
| 83 |
const FIELD_NAME = 'metasyncSeo'; |
| 84 |
|
| 85 |
/** |
| 86 |
* WPGraphQL's root query type, where the two archive-shaped entry points |
| 87 |
* that have no node of their own are attached. |
| 88 |
*/ |
| 89 |
const ROOT_TYPE_NAME = 'RootQuery'; |
| 90 |
|
| 91 |
/** |
| 92 |
* WPGraphQL's author node. Fixed rather than derived: unlike post types and |
| 93 |
* taxonomies, users are not a registry with a configurable GraphQL name. |
| 94 |
*/ |
| 95 |
const USER_TYPE_NAME = 'User'; |
| 96 |
|
| 97 |
/** |
| 98 |
* WPGraphQL's post-type node — the node `nodeByUri` returns for a post type |
| 99 |
* archive, and the one it returns for `/` on a site that shows posts there. |
| 100 |
*/ |
| 101 |
const CONTENT_TYPE_NAME = 'ContentType'; |
| 102 |
|
| 103 |
/** |
| 104 |
* Root field for the site front page. |
| 105 |
*/ |
| 106 |
const HOME_FIELD_NAME = 'metasyncSeoForHomePage'; |
| 107 |
|
| 108 |
/** |
| 109 |
* Root field for the blog index, when a static page holds it. |
| 110 |
*/ |
| 111 |
const POSTS_PAGE_FIELD_NAME = 'metasyncSeoForPostsPage'; |
| 112 |
|
| 113 |
/** |
| 114 |
* Most objects one GraphQL request may refresh from OTTO synchronously. |
| 115 |
* |
| 116 |
* A request asking for more than this many objects makes no live call at |
| 117 |
* all — not "the first five and then stop". Which is why the count cannot |
| 118 |
* be taken as resolvers fire: by the time the sixth object arrived, five |
| 119 |
* calls would already have gone out. See settle() for how the whole |
| 120 |
* request is counted before any of them runs. |
| 121 |
*/ |
| 122 |
const MAX_SYNC_REFRESH_OBJECTS = 5; |
| 123 |
|
| 124 |
/** |
| 125 |
* How many consecutive quiet turns prove a request has stopped growing. |
| 126 |
* |
| 127 |
* One is not enough. A lazy loader can be sitting in the deferred queue |
| 128 |
* behind us, so a single turn where nothing new arrived may simply be the |
| 129 |
* turn before that loader runs and introduces a whole new level of |
| 130 |
* objects. Two consecutive quiet turns mean we have been all the way round |
| 131 |
* the queue without the count moving. |
| 132 |
*/ |
| 133 |
const SETTLE_QUIET_TURNS = 2; |
| 134 |
|
| 135 |
/** |
| 136 |
* Every object this request asked to publish, keyed so that the same |
| 137 |
* object named twice — through a node and a connection, or through an |
| 138 |
* alias — counts once. |
| 139 |
* |
| 140 |
* @var array<string,bool> |
| 141 |
*/ |
| 142 |
private static $requested = array(); |
| 143 |
|
| 144 |
/** |
| 145 |
* The request's verdict: null until resolution settles, then whether the |
| 146 |
* request is small enough to refresh at all. |
| 147 |
* |
| 148 |
* Decided once, by whichever waiter first sees a settled request, and read |
| 149 |
* by every other waiter. Deciding per waiter would let the answer depend |
| 150 |
* on the order the client happened to write its selection set. |
| 151 |
* |
| 152 |
* @var bool|null |
| 153 |
*/ |
| 154 |
private static $verdict = null; |
| 155 |
|
| 156 |
/** |
| 157 |
* Live refreshes this request has started, across every pass. |
| 158 |
* |
| 159 |
* @var int |
| 160 |
*/ |
| 161 |
private static $refresh_calls = 0; |
| 162 |
|
| 163 |
/** |
| 164 |
* Objects already attempted this request, so a second selection of the |
| 165 |
* same object neither queues nor calls again. |
| 166 |
* |
| 167 |
* @var array<string,bool> |
| 168 |
*/ |
| 169 |
private static $attempted = array(); |
| 170 |
|
| 171 |
/** |
| 172 |
* What a refreshed object published, keyed the same way. |
| 173 |
* |
| 174 |
* Without this, a query naming one object twice would publish the |
| 175 |
* refreshed values for the first selection and the values captured before |
| 176 |
* the refresh for the second — two answers for one object in one response. |
| 177 |
* Only refreshed objects are recorded; everything else already publishes |
| 178 |
* what the resolver read. |
| 179 |
* |
| 180 |
* @var array<string,array> |
| 181 |
*/ |
| 182 |
private static $published = array(); |
| 183 |
|
| 184 |
/** |
| 185 |
* Monotonic start times for per-object history durations. |
| 186 |
* |
| 187 |
* @var array<string,float> |
| 188 |
*/ |
| 189 |
private static $history_started = array(); |
| 190 |
|
| 191 |
/** |
| 192 |
* Objects already written to refresh history in this GraphQL request. |
| 193 |
* |
| 194 |
* @var array<string,bool> |
| 195 |
*/ |
| 196 |
private static $history_recorded = array(); |
| 197 |
|
| 198 |
/** |
| 199 |
* Number of deferred settling callbacks waiting in the executor queue. |
| 200 |
* |
| 201 |
* @var int |
| 202 |
*/ |
| 203 |
private static $pending_rechecks = 0; |
| 204 |
|
| 205 |
/** @var int|null Current WPGraphQL request context identity. */ |
| 206 |
private static $request_context_id = null; |
| 207 |
|
| 208 |
/** @var array<int,array> Saved outer contexts for nested GraphQL calls. */ |
| 209 |
private static $request_context_stack = array(); |
| 210 |
|
| 211 |
/** |
| 212 |
* Hook registration into WPGraphQL's schema build. |
| 213 |
* |
| 214 |
* Safe to call unconditionally: `graphql_register_types` only ever fires |
| 215 |
* when WPGraphQL is active, so on a site without it this adds a callback |
| 216 |
* that is never invoked and changes nothing. |
| 217 |
* |
| 218 |
* @return void |
| 219 |
*/ |
| 220 |
public static function init() |
| 221 |
{ |
| 222 |
# Gated here as well as inside register(), so a non-headless site does not |
| 223 |
# even carry the hook. The ticket's rule is that with the mode off |
| 224 |
# execution follows exactly the path it does today, and an extra callback |
| 225 |
# on a WPGraphQL schema build is an observable difference even when it |
| 226 |
# returns immediately. The option read this costs is memoised and the |
| 227 |
# plugin already reads that option several times per request. |
| 228 |
# |
| 229 |
# The class_exists leg is the update-window case: this file and the |
| 230 |
# config class are separate files in the same auto-update, and WordPress |
| 231 |
# copies them one at a time. A request landing between the two copies |
| 232 |
# used to fatal here on every front-end load — the classmap cannot |
| 233 |
# autoload a file that is not on disk yet. Degrade to the no-headless |
| 234 |
# path instead; the site keeps serving until the update finishes. |
| 235 |
if (!class_exists('Metasync_Headless_Config') || !Metasync_Headless_Config::is_enabled()) { |
| 236 |
return; |
| 237 |
} |
| 238 |
|
| 239 |
add_action('graphql_register_types', array(__CLASS__, 'register')); |
| 240 |
add_action('graphql_before_execute', array(__CLASS__, 'begin_request_context'), 10, 1); |
| 241 |
# reset_request_state() takes no arguments; graphql_execute passes six, |
| 242 |
# so accept none and let WordPress hand the callback an empty arg list. |
| 243 |
add_action('graphql_execute', array(__CLASS__, 'reset_request_state'), 10, 0); |
| 244 |
add_action('graphql_after_execute', array(__CLASS__, 'end_request_context'), 10, 2); |
| 245 |
} |
| 246 |
|
| 247 |
/** |
| 248 |
* Forget everything the previous GraphQL request recorded about refreshes. |
| 249 |
* |
| 250 |
* @return void |
| 251 |
*/ |
| 252 |
public static function reset_request_state() |
| 253 |
{ |
| 254 |
self::$requested = array(); |
| 255 |
self::$verdict = null; |
| 256 |
self::$refresh_calls = 0; |
| 257 |
self::$attempted = array(); |
| 258 |
self::$published = array(); |
| 259 |
self::$history_started = array(); |
| 260 |
self::$history_recorded = array(); |
| 261 |
self::$pending_rechecks = 0; |
| 262 |
} |
| 263 |
|
| 264 |
public static function begin_request_context($request) |
| 265 |
{ |
| 266 |
self::$request_context_stack[] = array( |
| 267 |
'requested' => self::$requested, |
| 268 |
'verdict' => self::$verdict, |
| 269 |
'refresh_calls' => self::$refresh_calls, |
| 270 |
'attempted' => self::$attempted, |
| 271 |
'published' => self::$published, |
| 272 |
'history_started' => self::$history_started, |
| 273 |
'history_recorded' => self::$history_recorded, |
| 274 |
'pending_rechecks' => self::$pending_rechecks, |
| 275 |
); |
| 276 |
self::reset_request_state(); |
| 277 |
self::$request_context_id = is_object($request) ? spl_object_id($request) : null; |
| 278 |
} |
| 279 |
|
| 280 |
public static function end_request_context($response, $request) |
| 281 |
{ |
| 282 |
$context = array_pop(self::$request_context_stack); |
| 283 |
if ($context === null) { |
| 284 |
self::reset_request_state(); |
| 285 |
return; |
| 286 |
} |
| 287 |
|
| 288 |
self::$requested = $context['requested']; |
| 289 |
self::$verdict = $context['verdict']; |
| 290 |
self::$refresh_calls = $context['refresh_calls']; |
| 291 |
self::$attempted = $context['attempted']; |
| 292 |
self::$published = $context['published']; |
| 293 |
self::$history_started = $context['history_started']; |
| 294 |
self::$history_recorded = $context['history_recorded']; |
| 295 |
self::$pending_rechecks = $context['pending_rechecks']; |
| 296 |
self::$request_context_id = null; |
| 297 |
} |
| 298 |
|
| 299 |
/** |
| 300 |
* Register the type and its fields. |
| 301 |
* |
| 302 |
* The headless check happens here rather than at hook time so switching the |
| 303 |
* mode takes effect on the next request instead of needing a reload. With |
| 304 |
* the mode off nothing is registered and the schema is byte-identical to a |
| 305 |
* site without this plugin. The class_exists leg mirrors init(): during an |
| 306 |
* auto-update the file set on disk can be mixed, and a callback that only |
| 307 |
* exists to be inert must survive landing in one. |
| 308 |
* |
| 309 |
* @return void |
| 310 |
*/ |
| 311 |
public static function register() |
| 312 |
{ |
| 313 |
if (!class_exists('Metasync_Headless_Config') || !Metasync_Headless_Config::is_enabled()) { |
| 314 |
return; |
| 315 |
} |
| 316 |
|
| 317 |
if (!function_exists('register_graphql_object_type') || !function_exists('register_graphql_field')) { |
| 318 |
return; |
| 319 |
} |
| 320 |
|
| 321 |
try { |
| 322 |
self::register_type(); |
| 323 |
static::register_fields(); |
| 324 |
add_filter('graphql_connection_nodes', array(__CLASS__, 'prime_post_connection_nodes'), 10, 2); |
| 325 |
} catch (Throwable $e) { |
| 326 |
# A failed registration must not take down the whole schema — that |
| 327 |
# would break every query on the site, not just this field. |
| 328 |
# |
| 329 |
# Logged as an error rather than a warning because the consequence is |
| 330 |
# worse than a failed resolve: the field is absent from the schema, so |
| 331 |
# a persisted query asking for it fails validation with an unknown |
| 332 |
# field and takes the whole request with it. A resolve failure only |
| 333 |
# yields a null. This one needs to be noticed. |
| 334 |
self::log_failure('registration', $e, true); |
| 335 |
} |
| 336 |
} |
| 337 |
|
| 338 |
/** |
| 339 |
* The payload type. |
| 340 |
* |
| 341 |
* Every field is nullable, and the descriptions are written for the frontend |
| 342 |
* developer reading them in a schema explorer rather than for us. |
| 343 |
* |
| 344 |
* @return void |
| 345 |
*/ |
| 346 |
private static function register_type() |
| 347 |
{ |
| 348 |
register_graphql_object_type(self::TYPE_NAME, array( |
| 349 |
'description' => 'SEO values resolved by Search Atlas for this object, ready to render.', |
| 350 |
'fields' => array( |
| 351 |
'title' => array( |
| 352 |
'type' => 'String', |
| 353 |
'description' => 'Document title.', |
| 354 |
), |
| 355 |
'description' => array( |
| 356 |
'type' => 'String', |
| 357 |
'description' => 'Meta description.', |
| 358 |
), |
| 359 |
'keywords' => array( |
| 360 |
'type' => 'String', |
| 361 |
'description' => 'Focus keyword or keywords.', |
| 362 |
), |
| 363 |
'canonical' => array( |
| 364 |
'type' => 'String', |
| 365 |
'description' => 'Canonical URL, on the public frontend domain.', |
| 366 |
), |
| 367 |
'robots' => array( |
| 368 |
'type' => 'String', |
| 369 |
'description' => 'Robots directives, ready for a meta robots tag.', |
| 370 |
), |
| 371 |
'ogTitle' => array( |
| 372 |
'type' => 'String', |
| 373 |
'description' => 'og:title.', |
| 374 |
), |
| 375 |
'ogDescription' => array( |
| 376 |
'type' => 'String', |
| 377 |
'description' => 'og:description.', |
| 378 |
), |
| 379 |
'ogImage' => array( |
| 380 |
'type' => 'String', |
| 381 |
'description' => 'og:image URL.', |
| 382 |
), |
| 383 |
'ogUrl' => array( |
| 384 |
'type' => 'String', |
| 385 |
'description' => 'og:url. Always the same value as canonical.', |
| 386 |
), |
| 387 |
'ogType' => array( |
| 388 |
'type' => 'String', |
| 389 |
'description' => 'og:type.', |
| 390 |
), |
| 391 |
'ogSiteName' => array( |
| 392 |
'type' => 'String', |
| 393 |
'description' => 'og:site_name.', |
| 394 |
), |
| 395 |
'ogLocale' => array( |
| 396 |
'type' => 'String', |
| 397 |
'description' => 'og:locale.', |
| 398 |
), |
| 399 |
'twitterCard' => array( |
| 400 |
'type' => 'String', |
| 401 |
'description' => 'twitter:card type.', |
| 402 |
), |
| 403 |
'twitterTitle' => array( |
| 404 |
'type' => 'String', |
| 405 |
'description' => 'twitter:title.', |
| 406 |
), |
| 407 |
'twitterDescription' => array( |
| 408 |
'type' => 'String', |
| 409 |
'description' => 'twitter:description.', |
| 410 |
), |
| 411 |
'twitterImage' => array( |
| 412 |
'type' => 'String', |
| 413 |
'description' => 'twitter:image URL.', |
| 414 |
), |
| 415 |
'schema' => array( |
| 416 |
'type' => 'String', |
| 417 |
'description' => 'JSON-LD for this object, serialised. Render verbatim inside a application/ld+json script tag; do not re-serialise it.', |
| 418 |
), |
| 419 |
'publicUrl' => array( |
| 420 |
'type' => 'String', |
| 421 |
'description' => 'URL this object is served from on the public frontend.', |
| 422 |
), |
| 423 |
), |
| 424 |
)); |
| 425 |
} |
| 426 |
|
| 427 |
/** |
| 428 |
* Add the field everywhere it belongs. |
| 429 |
* |
| 430 |
* The node types are collected into one map before anything is registered, |
| 431 |
* because a GraphQL type name may only be offered the field once: asking |
| 432 |
* WPGraphQL to add the same field to the same type twice raises. Two post |
| 433 |
* types, or two taxonomies, or a post type and a taxonomy can all share a |
| 434 |
* `graphql_single_name` — an install like that has other problems, but this |
| 435 |
* must not be one of them. |
| 436 |
* |
| 437 |
* The map is built with `+=` so the first claim on a name wins and the |
| 438 |
* registries are consulted in a fixed order. That makes this the single place |
| 439 |
* de-duplication happens; a second guard further down would look like |
| 440 |
* belt-and-braces but would mean neither could be tested, since either one |
| 441 |
* alone would satisfy the assertion. |
| 442 |
* |
| 443 |
* @return void |
| 444 |
*/ |
| 445 |
protected static function register_fields() |
| 446 |
{ |
| 447 |
$node_fields = array(); |
| 448 |
|
| 449 |
foreach (self::graphql_post_types() as $graphql_single_name) { |
| 450 |
$node_fields += array($graphql_single_name => array( |
| 451 |
'resolve' => array(__CLASS__, 'resolve'), |
| 452 |
'description' => 'SEO values resolved by Search Atlas for this object.', |
| 453 |
)); |
| 454 |
} |
| 455 |
|
| 456 |
# Terms are the type this feature exists for, and the registry is read |
| 457 |
# the same way the post types are — show_in_graphql and |
| 458 |
# graphql_single_name — so a custom taxonomy is covered without a code |
| 459 |
# change here. On a headless build the custom taxonomies are usually the |
| 460 |
# ones carrying the site's navigation. |
| 461 |
foreach (self::graphql_taxonomies() as $graphql_single_name) { |
| 462 |
$node_fields += array($graphql_single_name => array( |
| 463 |
'resolve' => array(__CLASS__, 'resolve_term'), |
| 464 |
'description' => 'SEO values resolved by Search Atlas for this term archive.', |
| 465 |
)); |
| 466 |
} |
| 467 |
|
| 468 |
# `User` is the node nodeByUri() returns for an author archive, so the |
| 469 |
# author payload goes there rather than on a root field taking an id. |
| 470 |
$node_fields += array(self::USER_TYPE_NAME => array( |
| 471 |
'resolve' => array(__CLASS__, 'resolve_author'), |
| 472 |
'description' => 'SEO values resolved by Search Atlas for this author archive.', |
| 473 |
)); |
| 474 |
|
| 475 |
# `ContentType` is the node nodeByUri() returns for a post type archive. |
| 476 |
$node_fields += array(self::CONTENT_TYPE_NAME => array( |
| 477 |
'resolve' => array(__CLASS__, 'resolve_content_type'), |
| 478 |
'description' => 'SEO values resolved by Search Atlas for the archive this post type is served at.', |
| 479 |
)); |
| 480 |
|
| 481 |
foreach ($node_fields as $type_name => $spec) { |
| 482 |
if ($type_name === '') { |
| 483 |
continue; |
| 484 |
} |
| 485 |
|
| 486 |
self::register_field($type_name, self::FIELD_NAME, $spec['resolve'], $spec['description']); |
| 487 |
} |
| 488 |
|
| 489 |
self::register_root_fields(); |
| 490 |
} |
| 491 |
|
| 492 |
/** |
| 493 |
* The two entry points with no node, as root fields. |
| 494 |
* |
| 495 |
* Neither takes an argument, and that is the whole design. The rejected |
| 496 |
* alternatives all did: |
| 497 |
* |
| 498 |
* - `metasyncSeoForAuthor(id: ID!)` and `metasyncSeoForArchive(postType: |
| 499 |
* String!)`. Both describe objects WPGraphQL already exposes as nodes it |
| 500 |
* access-checks, so a root field taking an identifier would add a second |
| 501 |
* way in that this plugin would have to keep as strict as WPGraphQL's own |
| 502 |
* — for ever, on both sides. The resolvers do refuse non-public objects, |
| 503 |
* so the two happen to agree today; agreeing by coincidence is not a |
| 504 |
* property worth depending on when the endpoint may have no auth gate. |
| 505 |
* - one root field with a discriminator, say `metasyncSeo(type: HOME | |
| 506 |
* ARCHIVE | AUTHOR, id: ID)`. GraphQL cannot express "this argument is |
| 507 |
* required only for that value of that other argument", so validation |
| 508 |
* would move into the resolver — where the only legal outcome is null, |
| 509 |
* because this field may never raise. A caller who passed the wrong |
| 510 |
* combination would get a silent null indistinguishable from "no data". |
| 511 |
* - a URI-resolving field of our own. The frontend does fetch by URI, but |
| 512 |
* `nodeByUri` already does that routing and it is a large, version- |
| 513 |
* sensitive re-implementation of WordPress's rewrite rules. Putting the |
| 514 |
* field on the nodes it returns reaches the same place in the same |
| 515 |
* request without owning any of it. |
| 516 |
* |
| 517 |
* @return void |
| 518 |
*/ |
| 519 |
private static function register_root_fields() |
| 520 |
{ |
| 521 |
self::register_field( |
| 522 |
self::ROOT_TYPE_NAME, |
| 523 |
self::HOME_FIELD_NAME, |
| 524 |
array(__CLASS__, 'resolve_home_page'), |
| 525 |
'SEO values resolved by Search Atlas for the site front page. Takes no arguments: there is exactly one front page, and on a site that shows posts there it is not backed by any node.' |
| 526 |
); |
| 527 |
|
| 528 |
self::register_field( |
| 529 |
self::ROOT_TYPE_NAME, |
| 530 |
self::POSTS_PAGE_FIELD_NAME, |
| 531 |
array(__CLASS__, 'resolve_posts_page'), |
| 532 |
'SEO values resolved by Search Atlas for the blog index when a static page holds it. Null when the blog index is the front page; metasyncSeoForHomePage describes it in that case.' |
| 533 |
); |
| 534 |
} |
| 535 |
|
| 536 |
/** |
| 537 |
* Register one field, containing any failure to that field alone. |
| 538 |
* |
| 539 |
* register_graphql_field() raises when the type does not exist or the field |
| 540 |
* name is already taken, and there are now several calls where there was |
| 541 |
* one. Left to register()'s single catch, the first bad name would abort the |
| 542 |
* loop and cost every field after it — and an *absent* field is the |
| 543 |
* expensive failure, not a failed resolve: a persisted query naming a field |
| 544 |
* the schema does not have fails validation and takes the whole request with |
| 545 |
* it, where a failed resolve only yields a null. So each registration is |
| 546 |
* isolated here, and register()'s catch stays as the backstop for everything |
| 547 |
* outside these calls. |
| 548 |
* |
| 549 |
* @param string $type_name |
| 550 |
* @param string $field_name |
| 551 |
* @param callable $resolver |
| 552 |
* @param string $description |
| 553 |
* @return void |
| 554 |
*/ |
| 555 |
private static function register_field($type_name, $field_name, $resolver, $description) |
| 556 |
{ |
| 557 |
try { |
| 558 |
register_graphql_field($type_name, $field_name, array( |
| 559 |
'type' => self::TYPE_NAME, |
| 560 |
'description' => $description, |
| 561 |
'resolve' => $resolver, |
| 562 |
)); |
| 563 |
} catch (Throwable $e) { |
| 564 |
self::log_failure('registration of ' . $field_name . ' on ' . $type_name, $e, true); |
| 565 |
} |
| 566 |
} |
| 567 |
|
| 568 |
/** |
| 569 |
* The post type objects WPGraphQL exposes. |
| 570 |
* |
| 571 |
* The seam is this raw read rather than the derived name list, so the |
| 572 |
* validation and de-duplication below stay real logic that tests actually |
| 573 |
* execute. Overriding the derived list instead would leave both untested. |
| 574 |
* |
| 575 |
* @return array |
| 576 |
*/ |
| 577 |
protected static function read_post_type_objects() |
| 578 |
{ |
| 579 |
return get_post_types(array('show_in_graphql' => true), 'objects'); |
| 580 |
} |
| 581 |
|
| 582 |
/** |
| 583 |
* The taxonomy objects WPGraphQL exposes. |
| 584 |
* |
| 585 |
* A seam for the same reason as the post types: the filtering below is real |
| 586 |
* logic, and a test that replaced the derived names would prove nothing |
| 587 |
* about it. |
| 588 |
* |
| 589 |
* @return array |
| 590 |
*/ |
| 591 |
protected static function read_taxonomy_objects() |
| 592 |
{ |
| 593 |
return get_taxonomies(array('show_in_graphql' => true), 'objects'); |
| 594 |
} |
| 595 |
|
| 596 |
/** |
| 597 |
* GraphQL type names of every post type exposed to GraphQL. |
| 598 |
* |
| 599 |
* @return string[] |
| 600 |
*/ |
| 601 |
private static function graphql_post_types() |
| 602 |
{ |
| 603 |
return self::graphql_type_names(self::read_registry('read_post_type_objects')); |
| 604 |
} |
| 605 |
|
| 606 |
/** |
| 607 |
* GraphQL type names of every taxonomy exposed to GraphQL. |
| 608 |
* |
| 609 |
* @return string[] |
| 610 |
*/ |
| 611 |
private static function graphql_taxonomies() |
| 612 |
{ |
| 613 |
return self::graphql_type_names(self::read_registry('read_taxonomy_objects')); |
| 614 |
} |
| 615 |
|
| 616 |
/** |
| 617 |
* Read one registry seam, treating a failure as an empty registry. |
| 618 |
* |
| 619 |
* Same reasoning as register_field(): a pass that cannot even read its |
| 620 |
* registry must cost only its own fields. Left to register()'s single catch, |
| 621 |
* a broken get_taxonomies() would abort before the root fields were reached |
| 622 |
* — and those depend on no registry at all, so there is no reason for them |
| 623 |
* to share its fate. |
| 624 |
* |
| 625 |
* @param string $seam Name of the protected read method to call. |
| 626 |
* @return mixed Whatever the seam returned, or an empty array on failure. |
| 627 |
*/ |
| 628 |
private static function read_registry($seam) |
| 629 |
{ |
| 630 |
try { |
| 631 |
# Called through static::class so a subclass substituting the seam is |
| 632 |
# honoured, exactly as a direct static:: call would be. |
| 633 |
return call_user_func(array(static::class, $seam)); |
| 634 |
} catch (Throwable $e) { |
| 635 |
self::log_failure('registry read via ' . $seam, $e, true); |
| 636 |
|
| 637 |
return array(); |
| 638 |
} |
| 639 |
} |
| 640 |
|
| 641 |
/** |
| 642 |
* Pull usable `graphql_single_name` values out of a registry read. |
| 643 |
* |
| 644 |
* Names are returned as they were found, duplicates and all. De-duplication |
| 645 |
* belongs to register_fields(), which is the only place that can see across |
| 646 |
* both registries and the fixed names; doing it here as well would mean |
| 647 |
* neither guard could be tested, because either alone would satisfy the |
| 648 |
* assertion. |
| 649 |
* |
| 650 |
* Shared by both registries because the property, the validation and the |
| 651 |
* failure mode are identical; the two registries differ in which function |
| 652 |
* reads them, which is what the seams above are for. |
| 653 |
* |
| 654 |
* @param mixed $objects Registry objects, as WordPress returns them. |
| 655 |
* @return string[] |
| 656 |
*/ |
| 657 |
private static function graphql_type_names($objects) |
| 658 |
{ |
| 659 |
if (!is_array($objects) && !($objects instanceof Traversable)) { |
| 660 |
return array(); |
| 661 |
} |
| 662 |
|
| 663 |
$names = array(); |
| 664 |
|
| 665 |
foreach ($objects as $object) { |
| 666 |
if (!is_object($object) || empty($object->graphql_single_name)) { |
| 667 |
continue; |
| 668 |
} |
| 669 |
|
| 670 |
if (!is_string($object->graphql_single_name)) { |
| 671 |
continue; |
| 672 |
} |
| 673 |
|
| 674 |
$names[] = $object->graphql_single_name; |
| 675 |
} |
| 676 |
|
| 677 |
return $names; |
| 678 |
} |
| 679 |
|
| 680 |
/** |
| 681 |
* Resolve the field on a post node. |
| 682 |
* |
| 683 |
* Returns null rather than raising for anything unexpected. A GraphQL error |
| 684 |
* here would fail the entire query the frontend used to fetch its content, |
| 685 |
* not just this field. |
| 686 |
* |
| 687 |
* @param mixed $source Node model WPGraphQL is resolving. |
| 688 |
* @return array|null |
| 689 |
*/ |
| 690 |
public static function resolve($source) |
| 691 |
{ |
| 692 |
try { |
| 693 |
$post_id = self::database_id_from_source($source); |
| 694 |
if ($post_id <= 0) { |
| 695 |
return null; |
| 696 |
} |
| 697 |
|
| 698 |
$payload = static::payload_for_post($post_id); |
| 699 |
|
| 700 |
return self::answer('post', $post_id, null, $payload, static function () use ($post_id) { |
| 701 |
return static::payload_for_post($post_id); |
| 702 |
}); |
| 703 |
} catch (Throwable $e) { |
| 704 |
self::log_failure('resolve', $e); |
| 705 |
|
| 706 |
return null; |
| 707 |
} |
| 708 |
} |
| 709 |
|
| 710 |
/** |
| 711 |
* Resolve the field on a term node. |
| 712 |
* |
| 713 |
* @param mixed $source Node model WPGraphQL is resolving. |
| 714 |
* @return array|null |
| 715 |
*/ |
| 716 |
public static function resolve_term($source) |
| 717 |
{ |
| 718 |
try { |
| 719 |
# `term_id` is accepted alongside the model's databaseId because a |
| 720 |
# filtered install can hand a resolver a plain WP_Term, and the |
| 721 |
# taxonomy does not have to be passed: for_term() looks the term up |
| 722 |
# by id and reads the taxonomy off what it finds. |
| 723 |
$term_id = self::database_id_from_source($source, array('databaseId', 'term_id', 'ID', 'id')); |
| 724 |
if ($term_id <= 0) { |
| 725 |
return null; |
| 726 |
} |
| 727 |
|
| 728 |
$payload = static::payload_for_term($term_id); |
| 729 |
|
| 730 |
# The taxonomy is read off the payload rather than off the node: the |
| 731 |
# surface already looked the term up to build this, and it is the |
| 732 |
# only place the taxonomy is known for certain. |
| 733 |
$taxonomy = $payload instanceof Metasync_Headless_Seo_Data |
| 734 |
? (string) $payload->get('object_sub_type') |
| 735 |
: ''; |
| 736 |
|
| 737 |
return self::answer('term', $term_id, $taxonomy, $payload, static function () use ($term_id) { |
| 738 |
return static::payload_for_term($term_id); |
| 739 |
}); |
| 740 |
} catch (Throwable $e) { |
| 741 |
self::log_failure('resolve_term', $e); |
| 742 |
|
| 743 |
return null; |
| 744 |
} |
| 745 |
} |
| 746 |
|
| 747 |
/** |
| 748 |
* Resolve the field on a user node. |
| 749 |
* |
| 750 |
* @param mixed $source Node model WPGraphQL is resolving. |
| 751 |
* @return array|null |
| 752 |
*/ |
| 753 |
public static function resolve_author($source) |
| 754 |
{ |
| 755 |
try { |
| 756 |
# `userId` is WPGraphQL's older name for the same value and is still |
| 757 |
# present on the model, deprecated; reading both means an install |
| 758 |
# pinned to an older release resolves rather than silently nulling. |
| 759 |
$user_id = self::database_id_from_source($source, array('databaseId', 'userId', 'ID', 'id')); |
| 760 |
if ($user_id <= 0) { |
| 761 |
return null; |
| 762 |
} |
| 763 |
|
| 764 |
return self::shape_payload(static::payload_for_author($user_id)); |
| 765 |
} catch (Throwable $e) { |
| 766 |
self::log_failure('resolve_author', $e); |
| 767 |
|
| 768 |
return null; |
| 769 |
} |
| 770 |
} |
| 771 |
|
| 772 |
/** |
| 773 |
* Resolve the field on a post-type node. |
| 774 |
* |
| 775 |
* @param mixed $source Node model WPGraphQL is resolving. |
| 776 |
* @return array|null |
| 777 |
*/ |
| 778 |
public static function resolve_content_type($source) |
| 779 |
{ |
| 780 |
try { |
| 781 |
$post_type = self::post_type_from_source($source); |
| 782 |
if ($post_type === '') { |
| 783 |
return null; |
| 784 |
} |
| 785 |
|
| 786 |
# `post` has no archive route of its own. WordPress serves its |
| 787 |
# listing as the blog index, and get_post_type_archive_link('post') |
| 788 |
# returns that address rather than an archive of its own — which is |
| 789 |
# exactly why for_post_type_archive() refuses `post`. |
| 790 |
# |
| 791 |
# Routing it here is what stops the homepage being unreachable by |
| 792 |
# the path the frontend actually uses: on a site showing posts on |
| 793 |
# the front, `nodeByUri(uri: "/")` returns this very node, and |
| 794 |
# leaving it null would mean the customer's URI-driven query gets no |
| 795 |
# SEO for the site's most important page. |
| 796 |
# |
| 797 |
# Asked in this order, and neither call needs a config read here: |
| 798 |
# for_posts_page() answers only when a *distinct* static page holds |
| 799 |
# the index, and otherwise the index is the front page, which |
| 800 |
# for_home_page() owns. Each entry point already makes that decision |
| 801 |
# for itself, so this stays routing rather than a second copy of it. |
| 802 |
if ($post_type === 'post') { |
| 803 |
# One call, because the choice between the posts page and the |
| 804 |
# front page needs the reason behind a null and this layer |
| 805 |
# cannot see it. Asking for_posts_page() first and falling back |
| 806 |
# on null conflated "the front page is the blog index" with |
| 807 |
# "the assigned posts page is a draft", and answered the second |
| 808 |
# with the front page's own title and canonical. |
| 809 |
return self::shape_payload(static::payload_for_blog_index()); |
| 810 |
} |
| 811 |
|
| 812 |
return self::shape_payload(static::payload_for_post_type_archive($post_type)); |
| 813 |
} catch (Throwable $e) { |
| 814 |
self::log_failure('resolve_content_type', $e); |
| 815 |
|
| 816 |
return null; |
| 817 |
} |
| 818 |
} |
| 819 |
|
| 820 |
/** |
| 821 |
* Resolve the front page root field. |
| 822 |
* |
| 823 |
* @return array|null |
| 824 |
*/ |
| 825 |
public static function resolve_home_page() |
| 826 |
{ |
| 827 |
try { |
| 828 |
return self::shape_payload(static::payload_for_home_page()); |
| 829 |
} catch (Throwable $e) { |
| 830 |
self::log_failure('resolve_home_page', $e); |
| 831 |
|
| 832 |
return null; |
| 833 |
} |
| 834 |
} |
| 835 |
|
| 836 |
/** |
| 837 |
* Resolve the blog index root field. |
| 838 |
* |
| 839 |
* @return array|null |
| 840 |
*/ |
| 841 |
public static function resolve_posts_page() |
| 842 |
{ |
| 843 |
try { |
| 844 |
return self::shape_payload(static::payload_for_posts_page()); |
| 845 |
} catch (Throwable $e) { |
| 846 |
self::log_failure('resolve_posts_page', $e); |
| 847 |
|
| 848 |
return null; |
| 849 |
} |
| 850 |
} |
| 851 |
|
| 852 |
/** |
| 853 |
* Shape a resolved payload, or null when there is not one. |
| 854 |
* |
| 855 |
* Every resolver funnels through here so "the surface refused" and "the |
| 856 |
* surface returned something unexpected" become the same null in one place |
| 857 |
* rather than six. |
| 858 |
* |
| 859 |
* @param mixed $data |
| 860 |
* @return array|null |
| 861 |
*/ |
| 862 |
private static function shape_payload($data) |
| 863 |
{ |
| 864 |
if (!($data instanceof Metasync_Headless_Seo_Data)) { |
| 865 |
return null; |
| 866 |
} |
| 867 |
|
| 868 |
return self::shape($data); |
| 869 |
} |
| 870 |
|
| 871 |
/* ----------------------------------------------------------------- |
| 872 |
* Request-driven refresh |
| 873 |
* |
| 874 |
* Stored values are what GraphQL publishes. This block is only ever |
| 875 |
* allowed to make them *newer* — never slower to arrive than the |
| 876 |
* configured timeout, never absent because a refresh went wrong, and |
| 877 |
* never accompanied by a reason it went wrong. Anything that fails, |
| 878 |
* is blocked, or is simply not permitted yields the same answer the |
| 879 |
* resolver would have given without it. |
| 880 |
* ----------------------------------------------------------------- */ |
| 881 |
|
| 882 |
/** |
| 883 |
* Publish one object's SEO, refreshing it first when that is permitted. |
| 884 |
* |
| 885 |
* @param string $type 'post' or 'term'. |
| 886 |
* @param int $id WordPress object ID. |
| 887 |
* @param string|null $taxonomy Taxonomy name, terms only. |
| 888 |
* @param mixed $payload Stored payload as already resolved. |
| 889 |
* @param callable $reread Re-resolves the payload after a write. |
| 890 |
* @return array|\GraphQL\Deferred|null |
| 891 |
*/ |
| 892 |
private static function answer($type, $id, $taxonomy, $payload, callable $reread) |
| 893 |
{ |
| 894 |
if (!($payload instanceof Metasync_Headless_Seo_Data)) { |
| 895 |
# Nothing stored to publish, so nothing to refresh either — a |
| 896 |
# refresh writes over stored values, it does not invent them. |
| 897 |
return null; |
| 898 |
} |
| 899 |
|
| 900 |
$stored = self::shape($payload); |
| 901 |
|
| 902 |
# Asked before anything is queued, so a site with the live path off |
| 903 |
# carries none of the bookkeeping below. |
| 904 |
if (!static::live_refresh_available()) { |
| 905 |
return $stored; |
| 906 |
} |
| 907 |
|
| 908 |
$key = $type . ':' . (int) $id . (($taxonomy !== null && $taxonomy !== '') ? ':' . $taxonomy : ''); |
| 909 |
|
| 910 |
if (isset(self::$attempted[$key])) { |
| 911 |
# Already handled once in this request. A query naming the same |
| 912 |
# object twice — through a node and through a connection, say — |
| 913 |
# gets that answer again rather than a second call. Note it is the |
| 914 |
# *published* answer, not the one just re-read: one object must not |
| 915 |
# appear with two different titles in one response. |
| 916 |
return self::$published[$key] ?? $stored; |
| 917 |
} |
| 918 |
|
| 919 |
self::$requested[$key] = true; |
| 920 |
|
| 921 |
return static::defer(static function () use ($key, $type, $id, $taxonomy, $stored, $reread) { |
| 922 |
return self::settle($key, $type, $id, $taxonomy, $stored, $reread, -1, 0); |
| 923 |
}); |
| 924 |
} |
| 925 |
|
| 926 |
/** |
| 927 |
* Wait for the request to stop growing, then refresh if it is small enough. |
| 928 |
* |
| 929 |
* The rule is about the request, not about a pass: more than the ceiling |
| 930 |
* of objects and *none* of them is refreshed. That means the count has to |
| 931 |
* be final before the first call goes out, and under lazy resolution it is |
| 932 |
* not final when the first deferred callback runs. |
| 933 |
* |
| 934 |
* graphql-php keeps one FIFO queue of deferred callbacks, and completing |
| 935 |
* one can resolve fields that queue more. A connection is the ordinary |
| 936 |
* case: the parent's SEO field and the child loader are queued together, |
| 937 |
* so when the SEO callback runs, the children it will pull in have not |
| 938 |
* been counted yet. Answering there would refresh three objects out of a |
| 939 |
* query that turns out to name nine, and would do it or not depending on |
| 940 |
* whether the client wrote the connection before or after the SEO field. |
| 941 |
* |
| 942 |
* So a waiter that finds the count still moving hands its turn back to the |
| 943 |
* queue and looks again. When the count has held still for |
| 944 |
* SETTLE_QUIET_TURNS turns, resolution has produced everything it is going |
| 945 |
* to and the request can be judged. Costs about two extra turns per object |
| 946 |
* — no network, no database, just the queue that was going to drain anyway. |
| 947 |
* |
| 948 |
* @param string $key |
| 949 |
* @param string $type |
| 950 |
* @param int $id |
| 951 |
* @param string|null $taxonomy |
| 952 |
* @param array $stored |
| 953 |
* @param callable $reread |
| 954 |
* @param int $seen Object count at this waiter's previous turn. |
| 955 |
* @param int $quiet Consecutive turns the count has not moved. |
| 956 |
* @return array|\GraphQL\Deferred |
| 957 |
*/ |
| 958 |
private static function settle($key, $type, $id, $taxonomy, $stored, callable $reread, $seen, $quiet) |
| 959 |
{ |
| 960 |
if (self::$verdict === null) { |
| 961 |
$total = count(self::$requested); |
| 962 |
|
| 963 |
if ($total !== $seen) { |
| 964 |
# Still arriving. Nothing can be decided yet, by us or anyone. |
| 965 |
return static::defer(static function () use ($key, $type, $id, $taxonomy, $stored, $reread, $total) { |
| 966 |
return self::settle($key, $type, $id, $taxonomy, $stored, $reread, $total, 0); |
| 967 |
}); |
| 968 |
} |
| 969 |
|
| 970 |
if ($quiet < self::SETTLE_QUIET_TURNS - 1 |
| 971 |
&& (!static::queue_is_empty() || self::$pending_rechecks > 0)) { |
| 972 |
# Held still, but there is still work queued that could bring |
| 973 |
# more objects. Take another turn before trusting the count. |
| 974 |
$next = $quiet + 1; |
| 975 |
|
| 976 |
return static::defer(static function () use ($key, $type, $id, $taxonomy, $stored, $reread, $total, $next) { |
| 977 |
return self::settle($key, $type, $id, $taxonomy, $stored, $reread, $total, $next); |
| 978 |
}); |
| 979 |
} |
| 980 |
|
| 981 |
# Settled. Judge the request once, for every object in it, so the |
| 982 |
# answer cannot depend on selection order. |
| 983 |
self::$verdict = ($total <= self::MAX_SYNC_REFRESH_OBJECTS); |
| 984 |
} |
| 985 |
|
| 986 |
return self::refresh_then_publish($key, $type, $id, $taxonomy, $stored, $reread); |
| 987 |
} |
| 988 |
|
| 989 |
/** |
| 990 |
* Run the refresh this object queued, then publish whatever is stored. |
| 991 |
* |
| 992 |
* @param string $key |
| 993 |
* @param string $type |
| 994 |
* @param int $id |
| 995 |
* @param string|null $taxonomy |
| 996 |
* @param array $stored |
| 997 |
* @param callable $reread |
| 998 |
* @return array |
| 999 |
*/ |
| 1000 |
private static function refresh_then_publish($key, $type, $id, $taxonomy, $stored, callable $reread) |
| 1001 |
{ |
| 1002 |
try { |
| 1003 |
if (isset(self::$attempted[$key])) { |
| 1004 |
# Claimed by an earlier callback in this same request. answer() |
| 1005 |
# cannot catch this case: at resolve time no callback has run |
| 1006 |
# yet, so both selections of the object look unclaimed. |
| 1007 |
return self::$published[$key] ?? $stored; |
| 1008 |
} |
| 1009 |
|
| 1010 |
if (!self::$verdict || self::$refresh_calls >= self::MAX_SYNC_REFRESH_OBJECTS) { |
| 1011 |
# Either the request named more objects than may be refreshed |
| 1012 |
# synchronously, or the budget is already spent. Both publish |
| 1013 |
# what is stored. |
| 1014 |
self::$attempted[$key] = true; |
| 1015 |
self::record_history( |
| 1016 |
'__request__', |
| 1017 |
'request', |
| 1018 |
0, |
| 1019 |
!self::$verdict ? 'request_too_large' : 'budget_exhausted', |
| 1020 |
count(self::$requested) |
| 1021 |
); |
| 1022 |
|
| 1023 |
return $stored; |
| 1024 |
} |
| 1025 |
|
| 1026 |
# Marked before the call, not after: if the call throws, this object |
| 1027 |
# must still not be retried later in the same request. |
| 1028 |
self::$attempted[$key] = true; |
| 1029 |
self::$history_started[$key] = microtime(true); |
| 1030 |
|
| 1031 |
# Charged before the call, not after. A throw must still cost the |
| 1032 |
# budget, or a request whose refreshes all fail could attempt an |
| 1033 |
# unbounded number of them. |
| 1034 |
self::$refresh_calls++; |
| 1035 |
|
| 1036 |
$outcome = static::refresh_one($type, (int) $id, $taxonomy); |
| 1037 |
self::record_history($key, $type, $id, $outcome, count(self::$requested)); |
| 1038 |
|
| 1039 |
if ($outcome === Metasync_Headless_Refresh_Job::OUTCOME_FRESH |
| 1040 |
|| $outcome === Metasync_Headless_Refresh_Job::OUTCOME_DISABLED |
| 1041 |
|| $outcome === Metasync_Headless_Refresh_Job::OUTCOME_NO_URL |
| 1042 |
|| $outcome === Metasync_Headless_Refresh_Job::OUTCOME_BREAKER_OPEN |
| 1043 |
|| $outcome === Metasync_Headless_Refresh_Job::OUTCOME_LOCKED |
| 1044 |
|| $outcome === Metasync_Headless_Refresh_Job::OUTCOME_RATE_LIMITED) { |
| 1045 |
# None of these reached the network: each returns before the |
| 1046 |
# fetch. Refund the charge, or a run of cooldown hits would |
| 1047 |
# crowd out the objects that would actually refresh. |
| 1048 |
self::$refresh_calls--; |
| 1049 |
} |
| 1050 |
|
| 1051 |
if ($outcome !== Metasync_Headless_Refresh_Job::OUTCOME_REFRESHED) { |
| 1052 |
# Fresh, empty, unchanged, locked, rate-limited, breaker-open, |
| 1053 |
# failed: in every one of those the stored values are still the |
| 1054 |
# right answer, and which one it was is ours to know, not the |
| 1055 |
# client's. They are deliberately indistinguishable from here. |
| 1056 |
return $stored; |
| 1057 |
} |
| 1058 |
|
| 1059 |
$refreshed = $reread(); |
| 1060 |
$answer = $refreshed instanceof Metasync_Headless_Seo_Data ? self::shape($refreshed) : $stored; |
| 1061 |
|
| 1062 |
self::$published[$key] = $answer; |
| 1063 |
|
| 1064 |
return $answer; |
| 1065 |
} catch (Throwable $e) { |
| 1066 |
# This runs from graphql-php's deferred queue, outside the |
| 1067 |
# resolver's own try. Without this catch a refresh failure would |
| 1068 |
# reach the executor and fail the whole content query — the one |
| 1069 |
# thing this class must never do. |
| 1070 |
# |
| 1071 |
# static:: rather than self:: because this is the one failure path |
| 1072 |
# a subclass has to be able to observe: it is the only one that |
| 1073 |
# does not surface as a null the caller can see. |
| 1074 |
static::log_failure('refresh of ' . $type . ' ' . (int) $id, $e); |
| 1075 |
|
| 1076 |
return $stored; |
| 1077 |
} |
| 1078 |
} |
| 1079 |
|
| 1080 |
/** |
| 1081 |
* Persist one bounded, privacy-safe refresh history row. |
| 1082 |
* |
| 1083 |
* @param string $key |
| 1084 |
* @param string $type |
| 1085 |
* @param int $id |
| 1086 |
* @param string $outcome |
| 1087 |
* @param int $object_count |
| 1088 |
* @return void |
| 1089 |
*/ |
| 1090 |
private static function record_history($key, $type, $id, $outcome, $object_count) |
| 1091 |
{ |
| 1092 |
if ($outcome === Metasync_Headless_Refresh_Job::OUTCOME_FRESH |
| 1093 |
|| isset(self::$history_recorded[$key])) { |
| 1094 |
return; |
| 1095 |
} |
| 1096 |
|
| 1097 |
self::$history_recorded[$key] = true; |
| 1098 |
$started = self::$history_started[$key] ?? microtime(true); |
| 1099 |
$duration = (int) round((microtime(true) - $started) * 1000); |
| 1100 |
$reason_map = array( |
| 1101 |
Metasync_Headless_Refresh_Job::OUTCOME_FRESH => 'timestamp_fresh', |
| 1102 |
Metasync_Headless_Refresh_Job::OUTCOME_REFRESHED => 'live_success', |
| 1103 |
Metasync_Headless_Refresh_Job::OUTCOME_UNCHANGED => 'live_unchanged', |
| 1104 |
Metasync_Headless_Refresh_Job::OUTCOME_EMPTY => 'empty_response', |
| 1105 |
Metasync_Headless_Refresh_Job::OUTCOME_NO_URL => 'no_frontend_url', |
| 1106 |
Metasync_Headless_Refresh_Job::OUTCOME_LOCKED => 'refresh_locked', |
| 1107 |
Metasync_Headless_Refresh_Job::OUTCOME_RATE_LIMITED => 'rate_limited', |
| 1108 |
Metasync_Headless_Refresh_Job::OUTCOME_BREAKER_OPEN => 'breaker_open', |
| 1109 |
Metasync_Headless_Refresh_Job::OUTCOME_FAILED => 'transport_failed', |
| 1110 |
Metasync_Headless_Refresh_Job::OUTCOME_DISABLED => 'refresh_disabled', |
| 1111 |
'request_too_large' => 'request_too_large', |
| 1112 |
'budget_exhausted' => 'budget_exhausted', |
| 1113 |
); |
| 1114 |
$reason = $reason_map[$outcome] ?? 'refresh_outcome'; |
| 1115 |
|
| 1116 |
Metasync_Headless_Refresh_History::record( |
| 1117 |
$type, |
| 1118 |
(int) $id, |
| 1119 |
$outcome, |
| 1120 |
$reason, |
| 1121 |
$duration, |
| 1122 |
(int) $object_count |
| 1123 |
); |
| 1124 |
} |
| 1125 |
|
| 1126 |
/** |
| 1127 |
* Has graphql-php run out of deferred work? |
| 1128 |
* |
| 1129 |
* A seam because the transport tests drive settling without the real |
| 1130 |
* must still be able to answer the question. |
| 1131 |
* |
| 1132 |
* @return bool |
| 1133 |
*/ |
| 1134 |
protected static function queue_is_empty() |
| 1135 |
{ |
| 1136 |
if (!class_exists('GraphQL\Executor\Promise\Adapter\SyncPromiseQueue')) { |
| 1137 |
# Without the queue we cannot tell whether more work is pending, so |
| 1138 |
# we assume it is: the extra turn is cheap and the alternative is |
| 1139 |
# judging a request that has not finished arriving. |
| 1140 |
return false; |
| 1141 |
} |
| 1142 |
|
| 1143 |
return \GraphQL\Executor\Promise\Adapter\SyncPromiseQueue::isEmpty(); |
| 1144 |
} |
| 1145 |
|
| 1146 |
/** |
| 1147 |
* Is the request-driven refresh path usable at all? |
| 1148 |
* |
| 1149 |
* Deferral is part of the answer, not an optimisation: without it the |
| 1150 |
* object count cannot be known before the first call, and a path that |
| 1151 |
* cannot enforce the ceiling must not run. A site whose GraphQL cannot |
| 1152 |
* defer serves stored values, which is what it does today. |
| 1153 |
* |
| 1154 |
* @return bool |
| 1155 |
*/ |
| 1156 |
protected static function live_refresh_available() |
| 1157 |
{ |
| 1158 |
if (!class_exists('Metasync_Headless_Refresh_Job') || !class_exists('GraphQL\Deferred')) { |
| 1159 |
return false; |
| 1160 |
} |
| 1161 |
|
| 1162 |
return (bool) apply_filters('metasync_headless_graphql_live_refresh', true); |
| 1163 |
} |
| 1164 |
|
| 1165 |
/** |
| 1166 |
* Wrap a callback so graphql-php runs it after the current pass is counted. |
| 1167 |
* |
| 1168 |
* @param callable $callback |
| 1169 |
* @return \GraphQL\Deferred |
| 1170 |
*/ |
| 1171 |
protected static function defer(callable $callback) |
| 1172 |
{ |
| 1173 |
self::$pending_rechecks++; |
| 1174 |
|
| 1175 |
return new \GraphQL\Deferred(function () use ($callback) { |
| 1176 |
self::$pending_rechecks = max(0, self::$pending_rechecks - 1); |
| 1177 |
return $callback(); |
| 1178 |
}); |
| 1179 |
} |
| 1180 |
|
| 1181 |
/** |
| 1182 |
* Refresh one object, the single point where transport calls the job. |
| 1183 |
* |
| 1184 |
* A seam for the same reason as the resolution seams above: the transport |
| 1185 |
* tests drive counting, deferral, deduplication and fallback, and none of |
| 1186 |
* those need a live OTTO round trip to be real. |
| 1187 |
* |
| 1188 |
* @param string $type |
| 1189 |
* @param int $id |
| 1190 |
* @param string|null $taxonomy |
| 1191 |
* @return string One of the job's OUTCOME_* constants. |
| 1192 |
*/ |
| 1193 |
protected static function refresh_one($type, $id, $taxonomy) |
| 1194 |
{ |
| 1195 |
return Metasync_Headless_Refresh_Job::refresh_single_object($type, (int) $id, $taxonomy, 2); |
| 1196 |
} |
| 1197 |
|
| 1198 |
/* ----------------------------------------------------------------- |
| 1199 |
* Resolution seams |
| 1200 |
* |
| 1201 |
* The single points where transport meets resolution, one per entry |
| 1202 |
* point. Kept separate so a REST route can share them, so a bulk |
| 1203 |
* resolver can be substituted for a listing query without touching |
| 1204 |
* the shaping below, and so the transport tests can drive registration, |
| 1205 |
* gating, routing and fail-closed behaviour without dragging the whole |
| 1206 |
* resolver — and the ambient WordPress stubs — in behind them. |
| 1207 |
* ----------------------------------------------------------------- */ |
| 1208 |
|
| 1209 |
/** |
| 1210 |
* The resolved payload for one post. |
| 1211 |
* |
| 1212 |
* @param int $post_id |
| 1213 |
* @return Metasync_Headless_Seo_Data|null |
| 1214 |
*/ |
| 1215 |
protected static function payload_for_post($post_id) |
| 1216 |
{ |
| 1217 |
return Metasync_Headless_Seo_Surface::for_post($post_id); |
| 1218 |
} |
| 1219 |
|
| 1220 |
/** |
| 1221 |
* Resolve a page of post payloads in one pass. |
| 1222 |
* |
| 1223 |
* @param int[] $post_ids |
| 1224 |
* @return array |
| 1225 |
*/ |
| 1226 |
protected static function payload_for_posts(array $post_ids) |
| 1227 |
{ |
| 1228 |
return Metasync_Headless_Seo_Surface::for_posts($post_ids); |
| 1229 |
} |
| 1230 |
|
| 1231 |
/** |
| 1232 |
* Resolve a post connection's SEO payloads after WPGraphQL has loaded its |
| 1233 |
* nodes, so the per-node field resolver can reuse the request memo. |
| 1234 |
* |
| 1235 |
* WPGraphQL exposes connection IDs/nodes filters, but no field-specific |
| 1236 |
* batch resolver. This nodes filter is the practical boundary: it sees the |
| 1237 |
* complete connection page and runs only when metasyncSeo was requested. |
| 1238 |
* |
| 1239 |
* @param mixed $nodes Whatever the filter chain passes; guarded at runtime |
| 1240 |
* because a third-party filter is free to hand back |
| 1241 |
* something other than the node array. |
| 1242 |
* @param mixed $resolver |
| 1243 |
* @return array |
| 1244 |
*/ |
| 1245 |
public static function prime_post_connection_nodes($nodes, $resolver) |
| 1246 |
{ |
| 1247 |
if (!is_array($nodes) || !is_object($resolver)) { |
| 1248 |
return $nodes; |
| 1249 |
} |
| 1250 |
|
| 1251 |
try { |
| 1252 |
if (!method_exists($resolver, 'get_loader_name') || $resolver->get_loader_name() !== 'post') { |
| 1253 |
return $nodes; |
| 1254 |
} |
| 1255 |
|
| 1256 |
if (!self::connection_requests_seo($resolver)) { |
| 1257 |
return $nodes; |
| 1258 |
} |
| 1259 |
|
| 1260 |
$ids = array(); |
| 1261 |
foreach ($nodes as $node) { |
| 1262 |
$id = self::database_id_from_source($node); |
| 1263 |
if ($id > 0) { |
| 1264 |
$ids[] = $id; |
| 1265 |
} |
| 1266 |
} |
| 1267 |
|
| 1268 |
if (!empty($ids)) { |
| 1269 |
static::payload_for_posts($ids); |
| 1270 |
} |
| 1271 |
} catch (Throwable $e) { |
| 1272 |
self::log_failure('connection priming', $e); |
| 1273 |
} |
| 1274 |
|
| 1275 |
return $nodes; |
| 1276 |
} |
| 1277 |
|
| 1278 |
/** |
| 1279 |
* @param mixed $resolver |
| 1280 |
* @return bool |
| 1281 |
*/ |
| 1282 |
private static function connection_requests_seo($resolver) |
| 1283 |
{ |
| 1284 |
if (!method_exists($resolver, 'get_info')) { |
| 1285 |
return false; |
| 1286 |
} |
| 1287 |
|
| 1288 |
$info = $resolver->get_info(); |
| 1289 |
if (!is_object($info) || !method_exists($info, 'getFieldSelection')) { |
| 1290 |
return false; |
| 1291 |
} |
| 1292 |
|
| 1293 |
return self::selection_contains_field($info->getFieldSelection(5), self::FIELD_NAME); |
| 1294 |
} |
| 1295 |
|
| 1296 |
/** |
| 1297 |
* @param mixed $selection |
| 1298 |
* @param string $field |
| 1299 |
* @return bool |
| 1300 |
*/ |
| 1301 |
private static function selection_contains_field($selection, $field) |
| 1302 |
{ |
| 1303 |
if (!is_array($selection)) { |
| 1304 |
return false; |
| 1305 |
} |
| 1306 |
|
| 1307 |
foreach ($selection as $name => $children) { |
| 1308 |
if ((string) $name === $field || self::selection_contains_field($children, $field)) { |
| 1309 |
return true; |
| 1310 |
} |
| 1311 |
} |
| 1312 |
|
| 1313 |
return false; |
| 1314 |
} |
| 1315 |
|
| 1316 |
/** |
| 1317 |
* The resolved payload for one term. |
| 1318 |
* |
| 1319 |
* @param int $term_id |
| 1320 |
* @return Metasync_Headless_Seo_Data|null |
| 1321 |
*/ |
| 1322 |
protected static function payload_for_term($term_id) |
| 1323 |
{ |
| 1324 |
return Metasync_Headless_Seo_Surface::for_term($term_id); |
| 1325 |
} |
| 1326 |
|
| 1327 |
/** |
| 1328 |
* The resolved payload for one author archive. |
| 1329 |
* |
| 1330 |
* @param int $user_id |
| 1331 |
* @return Metasync_Headless_Seo_Data|null |
| 1332 |
*/ |
| 1333 |
protected static function payload_for_author($user_id) |
| 1334 |
{ |
| 1335 |
return Metasync_Headless_Seo_Surface::for_author($user_id); |
| 1336 |
} |
| 1337 |
|
| 1338 |
/** |
| 1339 |
* The resolved payload for one post type archive. |
| 1340 |
* |
| 1341 |
* @param string $post_type |
| 1342 |
* @return Metasync_Headless_Seo_Data|null |
| 1343 |
*/ |
| 1344 |
protected static function payload_for_post_type_archive($post_type) |
| 1345 |
{ |
| 1346 |
return Metasync_Headless_Seo_Surface::for_post_type_archive($post_type); |
| 1347 |
} |
| 1348 |
|
| 1349 |
/** |
| 1350 |
* The resolved payload for the front page. |
| 1351 |
* |
| 1352 |
* @return Metasync_Headless_Seo_Data|null |
| 1353 |
*/ |
| 1354 |
protected static function payload_for_home_page() |
| 1355 |
{ |
| 1356 |
return Metasync_Headless_Seo_Surface::for_home_page(); |
| 1357 |
} |
| 1358 |
|
| 1359 |
/** |
| 1360 |
* The resolved payload for the blog index. |
| 1361 |
* |
| 1362 |
* @return Metasync_Headless_Seo_Data|null |
| 1363 |
*/ |
| 1364 |
/** |
| 1365 |
* The resolved payload for whatever serves as the blog index. |
| 1366 |
* |
| 1367 |
* @return Metasync_Headless_Seo_Data|null |
| 1368 |
*/ |
| 1369 |
protected static function payload_for_blog_index() |
| 1370 |
{ |
| 1371 |
return Metasync_Headless_Seo_Surface::for_blog_index(); |
| 1372 |
} |
| 1373 |
|
| 1374 |
protected static function payload_for_posts_page() |
| 1375 |
{ |
| 1376 |
return Metasync_Headless_Seo_Surface::for_posts_page(); |
| 1377 |
} |
| 1378 |
|
| 1379 |
/** |
| 1380 |
* Map the payload onto the GraphQL field names. |
| 1381 |
* |
| 1382 |
* Written out rather than derived from the payload keys so adding a field to |
| 1383 |
* the resolver cannot silently publish it: anything new has to be added here |
| 1384 |
* and to the type deliberately. On an endpoint that may have no auth gate, |
| 1385 |
* that is the property worth having. |
| 1386 |
* |
| 1387 |
* @param Metasync_Headless_Seo_Data $data |
| 1388 |
* @return array |
| 1389 |
*/ |
| 1390 |
private static function shape(Metasync_Headless_Seo_Data $data) |
| 1391 |
{ |
| 1392 |
$schema = $data->get_schema(); |
| 1393 |
|
| 1394 |
return array( |
| 1395 |
'title' => $data->get_title(), |
| 1396 |
'description' => $data->get_description(), |
| 1397 |
'keywords' => $data->get_keywords(), |
| 1398 |
'canonical' => $data->get_canonical(), |
| 1399 |
'robots' => $data->get_robots(), |
| 1400 |
|
| 1401 |
'ogTitle' => $data->get_og_title(), |
| 1402 |
'ogDescription' => $data->get_og_description(), |
| 1403 |
'ogImage' => $data->get_og_image(), |
| 1404 |
'ogUrl' => $data->get_og_url(), |
| 1405 |
'ogType' => (string) $data->get('og_type'), |
| 1406 |
'ogSiteName' => (string) $data->get('og_site_name'), |
| 1407 |
'ogLocale' => (string) $data->get('og_locale'), |
| 1408 |
|
| 1409 |
'twitterCard' => (string) $data->get('twitter_card'), |
| 1410 |
'twitterTitle' => $data->get_twitter_title(), |
| 1411 |
'twitterDescription' => $data->get_twitter_description(), |
| 1412 |
'twitterImage' => $data->get_twitter_image(), |
| 1413 |
|
| 1414 |
# Null rather than "[]" when there is no graph: a frontend testing |
| 1415 |
# truthiness before writing a script tag should not have to know that |
| 1416 |
# an empty JSON array means "nothing to render". |
| 1417 |
'schema' => $schema === array() ? null : wp_json_encode($schema), |
| 1418 |
|
| 1419 |
'publicUrl' => $data->get_public_url(), |
| 1420 |
); |
| 1421 |
} |
| 1422 |
|
| 1423 |
/** |
| 1424 |
* Pull a database ID out of whatever WPGraphQL hands the resolver. |
| 1425 |
* |
| 1426 |
* The model exposes `databaseId`; older shapes and plain WordPress objects |
| 1427 |
* expose `ID`. Both are accepted rather than assuming one, since guessing |
| 1428 |
* wrong means the field silently resolves to null on every request — the |
| 1429 |
* failure mode that looks exactly like "there is no data". |
| 1430 |
* |
| 1431 |
* The property list is a parameter because each node type spells it |
| 1432 |
* differently, and because a name that means one thing on one model can mean |
| 1433 |
* something else on another: `id` on a WPGraphQL model is the relay global |
| 1434 |
* id, a base64 string, so it is read last and only survives the is_numeric() |
| 1435 |
* test on a plain object that happens to use it. |
| 1436 |
* |
| 1437 |
* @param mixed $source |
| 1438 |
* @param string[] $properties Property names to try, in order. |
| 1439 |
* @return int |
| 1440 |
*/ |
| 1441 |
private static function database_id_from_source($source, $properties = array('databaseId', 'ID', 'id')) |
| 1442 |
{ |
| 1443 |
if (is_numeric($source)) { |
| 1444 |
return (int) $source; |
| 1445 |
} |
| 1446 |
|
| 1447 |
if (!is_object($source)) { |
| 1448 |
return 0; |
| 1449 |
} |
| 1450 |
|
| 1451 |
foreach ($properties as $property) { |
| 1452 |
if (!isset($source->$property)) { |
| 1453 |
continue; |
| 1454 |
} |
| 1455 |
|
| 1456 |
$value = $source->$property; |
| 1457 |
if (is_numeric($value)) { |
| 1458 |
return (int) $value; |
| 1459 |
} |
| 1460 |
} |
| 1461 |
|
| 1462 |
return 0; |
| 1463 |
} |
| 1464 |
|
| 1465 |
/** |
| 1466 |
* Pull a post type name out of a ContentType node. |
| 1467 |
* |
| 1468 |
* `name` is the WordPress post type key on WPGraphQL's model — not the |
| 1469 |
* GraphQL type name, which is `graphqlSingleName`. `post_type` is accepted |
| 1470 |
* as well so a plain registered post type object also works. |
| 1471 |
* |
| 1472 |
* A plain string is accepted too, so this reads the same way as the id |
| 1473 |
* extraction above: whatever came in, either it names a post type or it does |
| 1474 |
* not. |
| 1475 |
* |
| 1476 |
* @param mixed $source |
| 1477 |
* @return string |
| 1478 |
*/ |
| 1479 |
private static function post_type_from_source($source) |
| 1480 |
{ |
| 1481 |
if (is_string($source)) { |
| 1482 |
return trim($source); |
| 1483 |
} |
| 1484 |
|
| 1485 |
if (!is_object($source)) { |
| 1486 |
return ''; |
| 1487 |
} |
| 1488 |
|
| 1489 |
foreach (array('name', 'post_type') as $property) { |
| 1490 |
if (!isset($source->$property)) { |
| 1491 |
continue; |
| 1492 |
} |
| 1493 |
|
| 1494 |
$value = $source->$property; |
| 1495 |
if (is_string($value) && trim($value) !== '') { |
| 1496 |
return trim($value); |
| 1497 |
} |
| 1498 |
} |
| 1499 |
|
| 1500 |
return ''; |
| 1501 |
} |
| 1502 |
|
| 1503 |
/** |
| 1504 |
* Record a failure without surfacing it to the caller. |
| 1505 |
* |
| 1506 |
* The whole point of this class is that the frontend never sees an error, so |
| 1507 |
* a failure that is invisible to us as well would be undiagnosable. |
| 1508 |
* |
| 1509 |
* Logging is itself wrapped, and that is not belt-and-braces. This runs |
| 1510 |
* inside a catch block, and a throw from inside a catch is not caught by the |
| 1511 |
* try it belongs to — so a logger that fails for any reason (an unwritable |
| 1512 |
* log path, a missing constant, a full disk) would send the original |
| 1513 |
* exception straight on to WPGraphQL and fail the entire content query. The |
| 1514 |
* one thing this class must never do. |
| 1515 |
* |
| 1516 |
* The severity arrives as a boolean rather than as a logger constant for the |
| 1517 |
* same reason. A caller writing `Metasync_Error_Logger::SEVERITY_ERROR` in |
| 1518 |
* its argument list resolves that class *before* entering this method, so on |
| 1519 |
* an install where the class is somehow absent the constant lookup throws |
| 1520 |
* from inside the caller's catch block — the exact shape this method exists |
| 1521 |
* to prevent, reintroduced by the call to it. Passing a bool means every |
| 1522 |
* mention of the class stays inside the try below. |
| 1523 |
* |
| 1524 |
* @param string $stage Where it happened. |
| 1525 |
* @param Throwable $e |
| 1526 |
* @param bool $is_error Log as an error rather than a warning. |
| 1527 |
* @return void |
| 1528 |
*/ |
| 1529 |
protected static function log_failure($stage, $e, $is_error = false) |
| 1530 |
{ |
| 1531 |
try { |
| 1532 |
if (!class_exists('Metasync_Error_Logger')) { |
| 1533 |
return; |
| 1534 |
} |
| 1535 |
|
| 1536 |
Metasync_Error_Logger::log( |
| 1537 |
Metasync_Error_Logger::CATEGORY_HEADLESS_DELIVERY, |
| 1538 |
$is_error ? Metasync_Error_Logger::SEVERITY_ERROR : Metasync_Error_Logger::SEVERITY_WARNING, |
| 1539 |
'Headless GraphQL SEO field failed during ' . $stage, |
| 1540 |
array( |
| 1541 |
'stage' => $stage, |
| 1542 |
'exception' => $e->getMessage(), |
| 1543 |
) |
| 1544 |
); |
| 1545 |
} catch (Throwable $ignored) { |
| 1546 |
# Nothing left to do: reporting the reporting failure would have the |
| 1547 |
# same problem. Losing a log line is survivable; breaking the page |
| 1548 |
# the visitor asked for is not. |
| 1549 |
return; |
| 1550 |
} |
| 1551 |
} |
| 1552 |
} |
| 1553 |
|