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

class-metasync-seo-inventory-builder.php in Search Atlas SEO – OTTO AI SEO Automation for WordPress 2.6.15, at includes/class-metasync-seo-inventory-builder.php

271 lines 10.4 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 WP-134).
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	$query = new WP_Query($query_args);
93
94	// Remove the filter immediately so it doesn't affect other queries
95	remove_filter('posts_where', [__CLASS__, 'filter_where_id_gt'], 10);
96
97	$posts = $query->posts;
98	$has_more = count($posts) > $limit;
99
100	if ($has_more) {
101	array_pop($posts); // remove the extra sentinel post
102	}
103
104	// Meta cache is already primed by WP_Query (update_post_meta_cache => true)
105
106	$items = [];
107	$next_cursor = 0;
108
109	foreach ($posts as $post) {
110	$item = self::build_item($post, $include_issues);
111	$items[] = $item;
112	$next_cursor = $post->ID;
113	}
114
115	// Total count (separate lightweight query)
116	$total = self::get_total_count($post_type, $post_status, $modified_after);
117
118	return [
119	'items' => $items,
120	'total' => $total,
121	'next_cursor' => $has_more ? $next_cursor : null,
122	'has_more' => $has_more,
123	];
124	}
125
126	/**
127	* Build a single post's inventory item.
128	*
129	* @param WP_Post $post
130	* @param bool $include_issues
131	* @return array
132	*/
133	private static function build_item(WP_Post $post, bool $include_issues): array {
134	$post_id = $post->ID;
135
136	// Basic info
137	// WP-499: Page-builder content (Divi [et_pb_*], Elementor, WPBakery) is stored
138	// as shortcodes whose handlers are not registered during a REST inventory request.
139	// Render what we can, then strip registered + leftover shortcode-style tags so raw
140	// builder markup does not leak into the content/word-count sent to SearchAtlas.
141	$content_text = do_shortcode($post->post_content);
142	$content_text = strip_shortcodes($content_text);
143	$content_text = preg_replace('/\[\/?[a-zA-Z][^\]]*\]/', '', $content_text);
144	$content_text = wp_strip_all_tags($content_text);
145	$word_count = str_word_count($content_text);
146
147	$item = [
148	'post_id' => $post_id,
149	'post_type' => $post->post_type,
150	'post_status' => $post->post_status,
151	'url' => get_permalink($post_id),
152	'slug' => $post->post_name,
153	'title' => $post->post_title,
154	'last_modified' => get_post_modified_time('c', true, $post_id),
155	'word_count' => $word_count,
156	];
157
158	// SEO sub-object — cast to string to guard against array meta
159	// (another plugin or manual DB edit could store non-string values,
160	// which would crash mb_strlen in compute_issues on PHP 8+)
161	$seo = [];
162	foreach (self::SEO_META_KEYS as $field => $meta_key) {
163	$value = get_post_meta($post_id, $meta_key, true);
164	$seo[$field] = is_string($value) ? $value : '';
165	}
166
167	// Schema types — handle both storage formats:
168	// 1. User/MCP format: ['types' => [['type' => 'article', ...], ...]]
169	// 2. OTTO format: ['otto_jsonld' => [...]]
170	$schema_data = get_post_meta($post_id, 'metasync_schema_markup', true);
171	$schema_types = [];
172	if (!empty($schema_data) && is_array($schema_data)) {
173	// Extract user-configured type names from 'types' array
174	if (isset($schema_data['types']) && is_array($schema_data['types'])) {
175	foreach ($schema_data['types'] as $type_entry) {
176	if (is_array($type_entry) && isset($type_entry['type'])) {
177	$schema_types[] = $type_entry['type'];
178	}
179	}
180	}
181	// Include OTTO/other top-level schema keys (e.g. 'otto_jsonld')
182	foreach (array_keys($schema_data) as $key) {
183	if (!in_array($key, ['enabled', 'types'], true)) {
184	$schema_types[] = $key;
185	}
186	}
187	}
188	$seo['schema_types'] = $schema_types;
189
190	$item['seo'] = $seo;
191
192	// Pre-computed issue flags
193	if ($include_issues) {
194	$item['issues'] = self::compute_issues($seo);
195	}
196
197	return $item;
198	}
199
200	/**
201	* Compute SEO issue flags for a post.
202	*
203	* @param array $seo The SEO sub-object.
204	* @return array Boolean issue flags.
205	*/
206	private static function compute_issues(array $seo): array {
207	$meta_title = $seo['meta_title'] ?? '';
208	$meta_desc = $seo['meta_description'] ?? '';
209
210	return [
211	'missing_meta_title' => empty($meta_title),
212	'missing_meta_description' => empty($meta_desc),
213	'meta_title_too_long' => mb_strlen($meta_title) > self::META_TITLE_MAX_LENGTH,
214	'meta_description_too_long' => mb_strlen($meta_desc) > self::META_DESCRIPTION_MAX_LENGTH,
215	'is_noindex' => ($seo['robots'] ?? '') === 'noindex',
216	'missing_og_image' => empty($seo['og_image'] ?? ''),
217	'missing_schema' => empty($seo['schema_types']),
218	'missing_focus_keyword' => empty($seo['focus_keyword'] ?? ''),
219	];
220	}
221
222	/**
223	* Get total count of matching posts (lightweight).
224	*
225	* @param string $post_type
226	* @param string $post_status
227	* @param string $modified_after
228	* @return int
229	*/
230	private static function get_total_count(string $post_type, string $post_status, string $modified_after): int {
231	$args = [
232	'post_type' => $post_type === 'any' ? ['post', 'page'] : $post_type,
233	'post_status' => $post_status,
234	'posts_per_page' => 1,
235	'fields' => 'ids',
236	'no_found_rows' => false, // we need found_rows for total
237	'update_post_meta_cache' => false,
238	'update_post_term_cache' => false,
239	];
240
241	if (!empty($modified_after)) {
242	$args['date_query'] = [
243	[
244	'after' => $modified_after,
245	'column' => 'post_modified_gmt',
246	'inclusive' => false,
247	],
248	];
249	}
250
251	$query = new WP_Query($args);
252	return (int) $query->found_posts;
253	}
254
255	/**
256	* WP_Query filter: WHERE ID > cursor for keyset pagination.
257	*
258	* @param string $where
259	* @param WP_Query $query
260	* @return string
261	*/
262	public static function filter_where_id_gt(string $where, WP_Query $query): string {
263	$cursor = isset($query->query_vars['where_id_gt']) ? absint($query->query_vars['where_id_gt']) : 0;
264	if ($cursor > 0) {
265	global $wpdb;
266	$where .= $wpdb->prepare(" AND {$wpdb->posts}.ID > %d", $cursor);
267	}
268	return $where;
269	}
270	}
271