← All changes | includes/Core/SiteProfiler.php +254 -6 4.6.1 → 4.9.2 View file →

		@@ -27,23 +27,54 @@
27	27	*/
28	28	const CACHE_TTL = HOUR_IN_SECONDS;
29	29
30	30	/**
	31	+ * Transient key for the cached content digest (per-locale). Kept separate from the
	32	+ * base profile so the lightweight detect() path never carries the heavier excerpts.
	33	+ * @var string
	34	+ */
	35	+ const CONTENT_CACHE_KEY = 'betterdocs_site_content';
	36	+
	37	+ /**
31	38	* Build the full site profile.
32	39	*
33		- * @param bool $fresh Skip the cache and recompute.
	40	+ * @param bool $fresh Skip the cache and recompute.
	41	+ * @param bool $with_content Also attach the richer `content` digest (real page/post/
	42	+ * product excerpts). Only the AI KB "outline" call needs it;
	43	+ * detect() leaves it off to stay light.
34	44	* @return array
35	45	*/
36		- public function build( $fresh = false ) {
	46	+ public function build( $fresh = false, $with_content = false ) {
37	47	$cache_key = self::CACHE_KEY . '_' . get_locale();
38	48
	49	+ $profile = null;
39	50	if ( ! $fresh ) {
40	51	$cached = get_transient( $cache_key );
41	52	if ( is_array( $cached ) ) {
42		- return $cached;
	53	+ $profile = $cached;
43	54	}
44	55	}
45	56
	57	+ if ( null === $profile ) {
	58	+ $profile = $this->build_base();
	59	+ set_transient( $cache_key, $profile, self::CACHE_TTL );
	60	+ }
	61	+
	62	+ if ( $with_content ) {
	63	+ // Computed + cached separately so it never bloats the base profile transient.
	64	+ $profile['content'] = $this->content_digest( $fresh );
	65	+ }
	66	+
	67	+ return $profile;
	68	+ }
	69	+
	70	+ /**
	71	+ * Compute the base (content-free) profile. Split out of build() so the heavier
	72	+ * content digest can be attached on demand without polluting the base cache.
	73	+ *
	74	+ * @return array
	75	+ */
	76	+ private function build_base() {
46	77	$site = [
47	78	'title' => sanitize_text_field( get_bloginfo( 'name' ) ),
48	79	'tagline' => sanitize_text_field( get_bloginfo( 'description' ) ),
49	80	'url' => esc_url_raw( home_url() ),
		@@ -71,16 +102,232 @@
71	102	* Filter the computed site profile before it is cached/returned.
72	103	*
73	104	* @param array $profile
74	105	*/
75		- $profile = apply_filters( 'betterdocs_site_profile', $profile );
	106	+ return apply_filters( 'betterdocs_site_profile', $profile );
	107	+ }
76	108
77		- set_transient( $cache_key, $profile, self::CACHE_TTL );
	109	+ /**
	110	+ * Real, privacy-safe, size-bounded content excerpts that let the AI understand what
	111	+ * the site actually does (not just page/nav titles). Only published, public content
	112	+ * — already on the public web — is excerpted: no drafts, private, user or order data.
	113	+ *
	114	+ * Cached separately from the base profile and gated by filters so it can be trimmed
	115	+ * or disabled per site.
	116	+ *
	117	+ * @param bool $fresh Skip the content cache and recompute.
	118	+ * @return array
	119	+ */
	120	+ public function content_digest( $fresh = false ) {
	121	+ /**
	122	+ * Allow a site to opt out of sending real content excerpts to the AI proxy.
	123	+ *
	124	+ * @param bool $enabled
	125	+ */
	126	+ if ( ! apply_filters( 'betterdocs_site_profile_include_content', true ) ) {
	127	+ return [];
	128	+ }
78	129
79		- return $profile;
	130	+ $cache_key = self::CONTENT_CACHE_KEY . '_' . get_locale();
	131	+ if ( ! $fresh ) {
	132	+ $cached = get_transient( $cache_key );
	133	+ if ( is_array( $cached ) ) {
	134	+ return $cached;
	135	+ }
	136	+ }
	137	+
	138	+ $content = [
	139	+ 'summary' => $this->home_about_summary(),
	140	+ 'pages' => $this->page_excerpts(),
	141	+ 'posts' => $this->post_excerpts(),
	142	+ ];
	143	+
	144	+ if ( class_exists( 'WooCommerce' ) ) {
	145	+ $content['products'] = $this->product_excerpts();
	146	+ }
	147	+
	148	+ $content = array_filter(
	149	+ $content,
	150	+ function ( $value ) {
	151	+ return ! ( is_array( $value ) && empty( $value ) ) && '' !== $value;
	152	+ }
	153	+ );
	154	+
	155	+ /**
	156	+ * Filter the computed content digest before it is cached/returned.
	157	+ *
	158	+ * @param array $content
	159	+ * @param SiteProfiler $profiler
	160	+ */
	161	+ $content = apply_filters( 'betterdocs_site_profile_content', $content, $this );
	162	+
	163	+ set_transient( $cache_key, $content, self::CACHE_TTL );
	164	+
	165	+ return $content;
80	166	}
81	167
82	168	/**
	169	+ * Short plain-text summary drawn from the front page and an About-style page — the
	170	+ * best single signal of "what is this site about".
	171	+ *
	172	+ * @return string
	173	+ */
	174	+ private function home_about_summary() {
	175	+ $parts = [];
	176	+
	177	+ $front_id = (int) get_option( 'page_on_front' );
	178	+ if ( 'page' === get_option( 'show_on_front' ) && $front_id > 0 ) {
	179	+ $ex = $this->excerpt_of( get_post_field( 'post_content', $front_id ), 300 );
	180	+ if ( '' !== $ex ) {
	181	+ $parts[] = $ex;
	182	+ }
	183	+ }
	184	+
	185	+ foreach ( [ 'about', 'about-us', 'company', 'who-we-are' ] as $slug ) {
	186	+ $page = get_page_by_path( $slug );
	187	+ if ( $page instanceof \WP_Post && 'publish' === $page->post_status ) {
	188	+ $ex = $this->excerpt_of( $page->post_content, 300 );
	189	+ if ( '' !== $ex ) {
	190	+ $parts[] = $ex;
	191	+ break;
	192	+ }
	193	+ }
	194	+ }
	195	+
	196	+ return $this->excerpt_of( implode( ' ', $parts ), 600 );
	197	+ }
	198	+
	199	+ /**
	200	+ * Title + short excerpt for the first few published pages (About, Pricing, Features…).
	201	+ *
	202	+ * @return array
	203	+ */
	204	+ private function page_excerpts() {
	205	+ // Exclude WooCommerce's utility pages (Shop/Cart/Checkout/My account) and the
	206	+ // privacy page: they carry no subject anyone documents or FAQs about, but on a
	207	+ // store they have the lowest menu_order and so used to consume the whole (small)
	208	+ // budget — starving the real content pages (About, Pricing, Financing, Delivery…).
	209	+ $exclude = array_filter(
	210	+ [
	211	+ (int) get_option( 'woocommerce_shop_page_id' ),
	212	+ (int) get_option( 'woocommerce_cart_page_id' ),
	213	+ (int) get_option( 'woocommerce_checkout_page_id' ),
	214	+ (int) get_option( 'woocommerce_myaccount_page_id' ),
	215	+ (int) get_option( 'wp_page_for_privacy_policy' ),
	216	+ ]
	217	+ );
	218	+
	219	+ $pages = get_posts(
	220	+ [
	221	+ 'post_type' => 'page',
	222	+ 'posts_per_page' => 15,
	223	+ 'orderby' => 'menu_order',
	224	+ 'order' => 'ASC',
	225	+ 'post_status' => 'publish',
	226	+ 'post__not_in' => $exclude,
	227	+ ]
	228	+ );
	229	+
	230	+ $out = [];
	231	+ foreach ( $pages as $page ) {
	232	+ $title = sanitize_text_field( get_the_title( $page ) );
	233	+ if ( '' === $title ) {
	234	+ continue;
	235	+ }
	236	+ $out[] = [
	237	+ 'title' => $title,
	238	+ 'excerpt' => $this->excerpt_of( $page->post_content, 150 ),
	239	+ ];
	240	+ }
	241	+
	242	+ return array_values( $out );
	243	+ }
	244	+
	245	+ /**
	246	+ * Title + short excerpt for recent published blog posts — signals real topics.
	247	+ *
	248	+ * @return array
	249	+ */
	250	+ private function post_excerpts() {
	251	+ $posts = get_posts(
	252	+ [
	253	+ 'post_type' => 'post',
	254	+ 'posts_per_page' => 8,
	255	+ 'orderby' => 'date',
	256	+ 'order' => 'DESC',
	257	+ 'post_status' => 'publish',
	258	+ ]
	259	+ );
	260	+
	261	+ $out = [];
	262	+ foreach ( $posts as $post ) {
	263	+ $title = sanitize_text_field( get_the_title( $post ) );
	264	+ if ( '' === $title ) {
	265	+ continue;
	266	+ }
	267	+ $raw = '' !== trim( (string) $post->post_excerpt ) ? $post->post_excerpt : $post->post_content;
	268	+ $out[] = [
	269	+ 'title' => $title,
	270	+ 'excerpt' => $this->excerpt_of( $raw, 120 ),
	271	+ ];
	272	+ }
	273	+
	274	+ return array_values( $out );
	275	+ }
	276	+
	277	+ /**
	278	+ * Title + short description for a few recent WooCommerce products.
	279	+ *
	280	+ * @return array
	281	+ */
	282	+ private function product_excerpts() {
	283	+ $products = get_posts(
	284	+ [
	285	+ 'post_type' => 'product',
	286	+ 'posts_per_page' => 8,
	287	+ 'orderby' => 'date',
	288	+ 'order' => 'DESC',
	289	+ 'post_status' => 'publish',
	290	+ ]
	291	+ );
	292	+
	293	+ $out = [];
	294	+ foreach ( $products as $product ) {
	295	+ $title = sanitize_text_field( get_the_title( $product ) );
	296	+ if ( '' === $title ) {
	297	+ continue;
	298	+ }
	299	+ $raw = '' !== trim( (string) $product->post_excerpt ) ? $product->post_excerpt : $product->post_content;
	300	+ $out[] = [
	301	+ 'title' => $title,
	302	+ 'excerpt' => $this->excerpt_of( $raw, 120 ),
	303	+ ];
	304	+ }
	305	+
	306	+ return array_values( $out );
	307	+ }
	308	+
	309	+ /**
	310	+ * Turn raw post content into a clean, bounded, single-line plain-text excerpt.
	311	+ * Strips shortcodes + tags (handles page-builder markup) and collapses whitespace.
	312	+ *
	313	+ * @param string $raw
	314	+ * @param int $chars
	315	+ * @return string
	316	+ */
	317	+ private function excerpt_of( $raw, $chars ) {
	318	+ $text = wp_strip_all_tags( strip_shortcodes( (string) $raw ) );
	319	+ $text = trim( preg_replace( '/\s+/', ' ', $text ) );
	320	+ if ( '' === $text ) {
	321	+ return '';
	322	+ }
	323	+ if ( mb_strlen( $text ) > $chars ) {
	324	+ $text = rtrim( mb_substr( $text, 0, $chars ) ) . '…';
	325	+ }
	326	+ return sanitize_text_field( $text );
	327	+ }
	328	+
	329	+ /**
83	330	* Primary site type, keyed off active plugins (highest signal).
84	331	* First match wins.
85	332	*
86	333	* @return string
		@@ -613,8 +860,9 @@
613	860	* @return void
614	861	*/
615	862	public function flush() {
616	863	delete_transient( self::CACHE_KEY . '_' . get_locale() );
	864	+ delete_transient( self::CONTENT_CACHE_KEY . '_' . get_locale() );
617	865	}
618	866
619	867	/**
620	868	* Active plugins on this site (network-active included).