metasync / includes / class-metasync-seo-inventory-builder.php

class-metasync-seo-inventory-builder.php in Search Atlas SEO – OTTO AI SEO Automation for WordPress 2.6.22, at includes/class-metasync-seo-inventory-builder.php

279 lines 10.9 KB

Raw Download Zip

1	<?php
2	/**
3	* SEO Inventory Builder
4	*
5	* Shared builder that fetches all published posts/pages with their SEO metadata
6	* in bulk using cursor-based pagination. Used by both the REST endpoint and
7	* the MCP tool so the logic lives in one place.
8	*
9	* @package MetaSync
10	* @subpackage Includes
11	*/
12
13	if (!defined('ABSPATH')) {
14	exit;
15	}
16
17	class Metasync_SEO_Inventory_Builder {
18
19	/**
20	* Meta keys we read for every post (MCP field set — matches).
21	*/
22	const SEO_META_KEYS = [
23	'meta_title' => '_metasync_metatitle',
24	'meta_description' => '_metasync_metadesc',
25	'focus_keyword' => '_metasync_focus_keyword',
26	'robots' => '_metasync_robots_index',
27	'canonical_url' => '_metasync_canonical_url',
28	'og_title' => '_metasync_og_title',
29	'og_description' => '_metasync_og_description',
30	'og_image' => '_metasync_og_image',
31	'og_type' => '_metasync_og_type',
32	'twitter_title' => '_metasync_twitter_title',
33	'twitter_description' => '_metasync_twitter_description',
34	];
35
36	const META_TITLE_MAX_LENGTH = 60;
37	const META_DESCRIPTION_MAX_LENGTH = 160;
38	const DEFAULT_LIMIT = 200;
39	const MAX_LIMIT = 500;
40
41	/**
42	* Build the inventory response.
43	*
44	* @param array $args {
45	* @type string $post_type 'post', 'page', or 'any'. Default 'any'.
46	* @type string $post_status 'publish', 'draft', etc. Default 'publish'.
47	* @type int $limit Items per page (1–500). Default 200.
48	* @type int $cursor Post ID to start after (keyset pagination). Default 0.
49	* @type bool $include_issues Whether to add pre-computed issue flags. Default true.
50	* @type string $modified_after ISO 8601 date — only return posts modified after this.
51	* }
52	* @return array Inventory payload with items, total, next_cursor, has_more.
53	*/
54	public static function build(array $args = []): array {
55	$post_type = isset($args['post_type']) ? sanitize_text_field($args['post_type']) : 'any';
56	$post_status = isset($args['post_status']) ? sanitize_text_field($args['post_status']) : 'publish';
57	$limit = isset($args['limit']) ? absint($args['limit']) : self::DEFAULT_LIMIT;
58	$cursor = isset($args['cursor']) ? absint($args['cursor']) : 0;
59	$include_issues = isset($args['include_issues']) ? (bool) $args['include_issues'] : true;
60	$modified_after = isset($args['modified_after']) ? sanitize_text_field($args['modified_after']) : '';
61
62	$limit = max(1, min($limit, self::MAX_LIMIT));
63
64	$query_args = [
65	'post_type' => $post_type === 'any' ? ['post', 'page'] : $post_type,
66	'post_status' => $post_status,
67	'posts_per_page' => $limit + 1, // fetch one extra to detect has_more
68	'orderby' => 'ID',
69	'order' => 'ASC',
70	'no_found_rows' => true, // skip SQL_CALC_FOUND_ROWS for performance
71	'update_post_meta_cache' => true,
72	'update_post_term_cache' => false,
73	];
74
75	// Keyset pagination: only posts with ID > cursor
76	if ($cursor > 0) {
77	$query_args['where_id_gt'] = $cursor; // handled by filter below
78	add_filter('posts_where', [__CLASS__, 'filter_where_id_gt'], 10, 2);
79	}
80
81	// Modified-after filter
82	if (!empty($modified_after)) {
83	$query_args['date_query'] = [
84	[
85	'after' => $modified_after,
86	'column' => 'post_modified_gmt',
87	'inclusive' => false,
88	],
89	];
90	}
91
92	// Exclude custom/LPS pages — their SEO is self-managed, so the platform
93	// must not treat them as "missing SEO".
94	$query_args['meta_query'] = [metasync_get_custom_page_exclusion_meta_query()];
95
96	$query = new WP_Query($query_args);
97
98	// Remove the filter immediately so it doesn't affect other queries
99	remove_filter('posts_where', [__CLASS__, 'filter_where_id_gt'], 10);
100
101	$posts = $query->posts;
102	$has_more = count($posts) > $limit;
103
104	if ($has_more) {
105	array_pop($posts); // remove the extra sentinel post
106	}
107
108	// Meta cache is already primed by WP_Query (update_post_meta_cache => true)
109
110	$items = [];
111	$next_cursor = 0;
112
113	foreach ($posts as $post) {
114	$item = self::build_item($post, $include_issues);
115	$items[] = $item;
116	$next_cursor = $post->ID;
117	}
118
119	// Total count (separate lightweight query)
120	$total = self::get_total_count($post_type, $post_status, $modified_after);
121
122	return [
123	'items' => $items,
124	'total' => $total,
125	'next_cursor' => $has_more ? $next_cursor : null,
126	'has_more' => $has_more,
127	];
128	}
129
130	/**
131	* Build a single post's inventory item.
132	*
133	* @param WP_Post $post
134	* @param bool $include_issues
135	* @return array
136	*/
137	private static function build_item(WP_Post $post, bool $include_issues): array {
138	$post_id = $post->ID;
139
140	// Basic info
141	// Page-builder content (Divi [et_pb_*], Elementor, WPBakery) is stored
142	// as shortcodes whose handlers are not registered during a REST inventory request.
143	// Render what we can, then strip registered + leftover shortcode-style tags so raw
144	// builder markup does not leak into the content/word-count sent to SearchAtlas.
145	$content_text = do_shortcode($post->post_content);
146	$content_text = strip_shortcodes($content_text);
147	$content_text = preg_replace('/\[\/?[a-zA-Z][^\]]*\]/', '', $content_text);
148	$content_text = wp_strip_all_tags($content_text);
149	$word_count = str_word_count($content_text);
150
151	$item = [
152	'post_id' => $post_id,
153	'post_type' => $post->post_type,
154	'post_status' => $post->post_status,
155	'url' => get_permalink($post_id),
156	'slug' => $post->post_name,
157	'title' => $post->post_title,
158	'last_modified' => get_post_modified_time('c', true, $post_id),
159	'word_count' => $word_count,
160	];
161
162	// SEO sub-object — cast to string to guard against array meta
163	// (another plugin or manual DB edit could store non-string values,
164	// which would crash mb_strlen in compute_issues on PHP 8+)
165	$seo = [];
166	foreach (self::SEO_META_KEYS as $field => $meta_key) {
167	$value = get_post_meta($post_id, $meta_key, true);
168	$seo[$field] = is_string($value) ? $value : '';
169	}
170
171	// Schema types — handle both storage formats:
172	// 1. User/MCP format: ['types' => [['type' => 'article', ...], ...]]
173	// 2. OTTO format: ['otto_jsonld' => [...]]
174	$schema_data = get_post_meta($post_id, 'metasync_schema_markup', true);
175	$schema_types = [];
176	if (!empty($schema_data) && is_array($schema_data)) {
177	// Extract user-configured type names from 'types' array
178	if (isset($schema_data['types']) && is_array($schema_data['types'])) {
179	foreach ($schema_data['types'] as $type_entry) {
180	if (is_array($type_entry) && isset($type_entry['type'])) {
181	$schema_types[] = $type_entry['type'];
182	}
183	}
184	}
185	// Include OTTO/other top-level schema keys (e.g. 'otto_jsonld')
186	foreach (array_keys($schema_data) as $key) {
187	if (!in_array($key, ['enabled', 'types'], true)) {
188	$schema_types[] = $key;
189	}
190	}
191	}
192	$seo['schema_types'] = $schema_types;
193
194	$item['seo'] = $seo;
195
196	// Pre-computed issue flags
197	if ($include_issues) {
198	$item['issues'] = self::compute_issues($seo);
199	}
200
201	return $item;
202	}
203
204	/**
205	* Compute SEO issue flags for a post.
206	*
207	* @param array $seo The SEO sub-object.
208	* @return array Boolean issue flags.
209	*/
210	private static function compute_issues(array $seo): array {
211	$meta_title = $seo['meta_title'] ?? '';
212	$meta_desc = $seo['meta_description'] ?? '';
213
214	return [
215	'missing_meta_title' => empty($meta_title),
216	'missing_meta_description' => empty($meta_desc),
217	'meta_title_too_long' => mb_strlen($meta_title) > self::META_TITLE_MAX_LENGTH,
218	'meta_description_too_long' => mb_strlen($meta_desc) > self::META_DESCRIPTION_MAX_LENGTH,
219	'is_noindex' => ($seo['robots'] ?? '') === 'noindex',
220	'missing_og_image' => empty($seo['og_image'] ?? ''),
221	'missing_schema' => empty($seo['schema_types']),
222	'missing_focus_keyword' => empty($seo['focus_keyword'] ?? ''),
223	];
224	}
225
226	/**
227	* Get total count of matching posts (lightweight).
228	*
229	* @param string $post_type
230	* @param string $post_status
231	* @param string $modified_after
232	* @return int
233	*/
234	private static function get_total_count(string $post_type, string $post_status, string $modified_after): int {
235	$args = [
236	'post_type' => $post_type === 'any' ? ['post', 'page'] : $post_type,
237	'post_status' => $post_status,
238	'posts_per_page' => 1,
239	'fields' => 'ids',
240	'no_found_rows' => false, // we need found_rows for total
241	'update_post_meta_cache' => false,
242	'update_post_term_cache' => false,
243	// Exclude custom/LPS pages so the total matches the excluded item list
244	// in build() — otherwise the platform sees a count larger than the
245	// number of items it can page through.
246	'meta_query' => [metasync_get_custom_page_exclusion_meta_query()],
247	];
248
249	if (!empty($modified_after)) {
250	$args['date_query'] = [
251	[
252	'after' => $modified_after,
253	'column' => 'post_modified_gmt',
254	'inclusive' => false,
255	],
256	];
257	}
258
259	$query = new WP_Query($args);
260	return (int) $query->found_posts;
261	}
262
263	/**
264	* WP_Query filter: WHERE ID > cursor for keyset pagination.
265	*
266	* @param string $where
267	* @param WP_Query $query
268	* @return string
269	*/
270	public static function filter_where_id_gt(string $where, WP_Query $query): string {
271	$cursor = isset($query->query_vars['where_id_gt']) ? absint($query->query_vars['where_id_gt']) : 0;
272	if ($cursor > 0) {
273	global $wpdb;
274	$where .= $wpdb->prepare(" AND {$wpdb->posts}.ID > %d", $cursor);
275	}
276	return $where;
277	}
278	}
279