| @@ -5,8 +5,9 @@ | ||
| 5 | 5 | use WP_REST_Request; |
| 6 | 6 | use WPDeveloper\BetterDocs\Core\BaseAPI; |
| 7 | 7 | use WPDeveloper\BetterDocs\Core\SiteProfiler; |
| 8 | 8 | use WPDeveloper\BetterDocs\Core\Settings; |
| 9 | +use WPDeveloper\BetterDocs\Utils\AIUsage; | |
| 9 | 10 | use WPDeveloper\BetterDocs\Dependencies\DI\Container; |
| 10 | 11 | |
| 11 | 12 | /** |
| 12 | 13 | * REST surface for the AI "Generate Sample Docs" feature (Phase 2). |
| @@ -19,12 +20,52 @@ | ||
| 19 | 20 | * |
| 20 | 21 | * @since 4.5.3 |
| 21 | 22 | */ |
| 22 | 23 | class SampleDocs extends BaseAPI { |
| 23 | - /** Hard caps mirrored on the proxy (Phase 0 contract). */ | |
| 24 | + /** Hard caps mirrored on the proxy (Phase 0 contract) — legacy single-call docs. */ | |
| 24 | 25 | const MAX_CATEGORIES = 3; |
| 25 | 26 | const MAX_ARTICLES_PER_CATEGORY = 3; |
| 26 | 27 | |
| 28 | + /** | |
| 29 | + * FAQ caps — mirrored on the proxy. Upper bounds only: the AI designs the groups | |
| 30 | + * and the questions from the site's real content and right-sizes below these. The | |
| 31 | + * ceiling is generous (6 groups / 18 questions) so a content-rich site can get a | |
| 32 | + * real FAQ, but the AI is told to produce only as many as the content warrants and | |
| 33 | + * never to pad to the cap — a small site still gets a small FAQ. | |
| 34 | + */ | |
| 35 | + const MAX_FAQ_CATEGORIES = 6; | |
| 36 | + const MAX_FAQ_ARTICLES_PER_CATEGORY = 3; | |
| 37 | + const MAX_FAQ_ARTICLES_TOTAL = 18; | |
| 38 | + | |
| 39 | + /** | |
| 40 | + * Product FAQ caps. Higher than the general FAQ on purpose: this tab writes ONE group | |
| 41 | + * per real product category (and attaches it to that category), so the group cap has | |
| 42 | + * to cover every category the store actually has — with a smaller group cap, a | |
| 43 | + * store with many product categories simply lost half of them. | |
| 44 | + */ | |
| 45 | + const MAX_PRODUCT_FAQ_CATEGORIES = 12; | |
| 46 | + const MAX_PRODUCT_FAQ_ARTICLES_TOTAL = 36; | |
| 47 | + | |
| 48 | + /** | |
| 49 | + * How many detected site subjects (doc categories, key pages, product categories) | |
| 50 | + * to offer as FAQ scope chips. They are topics to cover, NOT groups — the AI decides | |
| 51 | + * how many groups a set of topics warrants (folding several into one, or dropping a | |
| 52 | + * thin one) and produces at most MAX_FAQ_CATEGORIES groups of its own design. | |
| 53 | + */ | |
| 54 | + const MAX_FAQ_TOPICS = 6; | |
| 55 | + | |
| 56 | + /** Product-category chips offered on the WooCommerce tab (one group is written per kept chip). */ | |
| 57 | + const MAX_PRODUCT_FAQ_TOPICS = 12; | |
| 58 | + | |
| 59 | + /** | |
| 60 | + * Deep "full knowledge base" caps (docs only) — mirrored on the proxy. These are | |
| 61 | + * upper bounds only: the AI right-sizes the KB to what the site actually needs and | |
| 62 | + * may return fewer. | |
| 63 | + */ | |
| 64 | + const MAX_KB_CATEGORIES = 6; | |
| 65 | + const MAX_KB_ARTICLES_PER_CATEGORY = 4; | |
| 66 | + const MAX_KB_ARTICLES_TOTAL = 18; | |
| 67 | + | |
| 27 | 68 | /** Default per-category palette/icons (matches the design mockup). */ |
| 28 | 69 | const PALETTE = [ '#00B884', '#3B82F6', '#8B5CF6', '#F59E0B', '#0EA5E9', '#EF4444' ]; |
| 29 | 70 | |
| 30 | 71 | /** |
| @@ -43,8 +84,11 @@ | ||
| 43 | 84 | |
| 44 | 85 | public function register() { |
| 45 | 86 | $this->post( 'sample-docs/detect', [ $this, 'detect' ] ); |
| 46 | 87 | $this->post( 'sample-docs/generate', [ $this, 'generate' ] ); |
| 88 | + // Deep KB (docs) — the multi-call outline→expand flow. | |
| 89 | + $this->post( 'sample-docs/outline', [ $this, 'outline' ] ); | |
| 90 | + $this->post( 'sample-docs/article', [ $this, 'article' ] ); | |
| 47 | 91 | $this->post( 'sample-docs/insert', [ $this, 'insert' ] ); |
| 48 | 92 | $this->post( 'sample-docs/undo', [ $this, 'undo' ] ); |
| 49 | 93 | } |
| 50 | 94 | |
| @@ -57,10 +101,16 @@ | ||
| 57 | 101 | return $this->error( 'feature_disabled', __( 'AI sample docs is disabled.', 'betterdocs' ), 403 ); |
| 58 | 102 | } |
| 59 | 103 | |
| 60 | 104 | $content_type = $this->content_type( $request ); |
| 61 | - $profile = $this->profiler->build( (bool) $request->get_param( 'fresh' ) ); | |
| 62 | 105 | |
| 106 | + // Every tab now reads its topic chips off the site's REAL content — the General FAQ | |
| 107 | + // and docs tabs from doc categories + key pages, the WooCommerce tab from product | |
| 108 | + // categories — so detect() needs the content digest for all of them. It's cached | |
| 109 | + // separately (and the docs outline call reuses the same cache), so this is one scan | |
| 110 | + // per site, not one per click. | |
| 111 | + $profile = $this->profiler->build( (bool) $request->get_param( 'fresh' ), true ); | |
| 112 | + | |
| 63 | 113 | /** |
| 64 | 114 | * Telemetry: site detection ran. |
| 65 | 115 | * |
| 66 | 116 | * @param string $type Detected site type. |
| @@ -86,36 +136,80 @@ | ||
| 86 | 136 | } |
| 87 | 137 | |
| 88 | 138 | $content_type = $this->content_type( $request ); |
| 89 | 139 | $profile = (array) $request->get_param( 'profile' ); |
| 90 | - $categories = $this->sanitize_categories( (array) $request->get_param( 'categories' ), false, $this->max_categories( $content_type ) ); | |
| 140 | + // For a general FAQ the incoming list is the owner's kept TOPICS (scope), not the | |
| 141 | + // groups to produce — so it is capped at MAX_FAQ_TOPICS, which is separate from the | |
| 142 | + // group cap. The AI decides how many of at most MAX_FAQ_CATEGORIES groups those | |
| 143 | + // topics warrant, folding several into one or dropping a thin one. | |
| 144 | + $categories = $this->sanitize_categories( | |
| 145 | + (array) $request->get_param( 'categories' ), | |
| 146 | + false, | |
| 147 | + 'docs' === $content_type | |
| 148 | + ? $this->max_categories( $content_type ) | |
| 149 | + : ( 'product_faq' === $content_type ? self::MAX_PRODUCT_FAQ_TOPICS : self::MAX_FAQ_TOPICS ), | |
| 150 | + $this->max_articles( $content_type ) | |
| 151 | + ); | |
| 91 | 152 | |
| 92 | 153 | if ( empty( $profile ) ) { |
| 93 | 154 | $profile = $this->profiler->build(); |
| 94 | 155 | } |
| 156 | + | |
| 157 | + // Product FAQs used to be generated deterministically from the store's WooCommerce | |
| 158 | + // SETTINGS (StoreFaqContent) — which produced generic store policy (Payments, | |
| 159 | + // Shipping, Returns) and said nothing about what the store actually sells. They | |
| 160 | + // now go through the AI like every other type, grounded in the real products and | |
| 161 | + // product categories from the content digest. | |
| 162 | + | |
| 163 | + // FAQ / product FAQ (and the legacy docs fallback) are grounded in the site's REAL | |
| 164 | + // content — | |
| 165 | + // the same homepage/About/page/post/product excerpts the deep docs flow uses — | |
| 166 | + // so questions and answers are specific to what the site offers, not generic. | |
| 167 | + // Enrich the profile when the client didn't send the content digest. | |
| 168 | + if ( empty( $profile['content'] ) ) { | |
| 169 | + $profile = $this->profiler->build( (bool) $request->get_param( 'fresh' ), true ); | |
| 170 | + } | |
| 95 | 171 | if ( empty( $categories ) ) { |
| 96 | 172 | $categories = $this->suggested_categories( $profile, $content_type ); |
| 97 | 173 | } |
| 98 | 174 | |
| 99 | - // Product FAQs are generated deterministically from the store's real | |
| 100 | - // WooCommerce settings — accurate, instant, never generic, and with no | |
| 101 | - // AI/proxy round-trip. | |
| 175 | + // The store-wide "Shipping, Returns & Payments" group is offered on the WooCommerce | |
| 176 | + // profile screen as a chip so the owner can see (and remove) it. It is NOT a product | |
| 177 | + // category, so pull it out of what goes to the AI, and remember whether the owner | |
| 178 | + // kept it — the deterministic group is only prepended below if they did. | |
| 179 | + $include_store_wide = false; | |
| 102 | 180 | if ( 'product_faq' === $content_type ) { |
| 103 | - return $this->success( $this->build_store_faqs( $profile, $content_type, $categories ) ); | |
| 181 | + $kept = []; | |
| 182 | + foreach ( $categories as $cat ) { | |
| 183 | + if ( ! empty( $cat['all_products'] ) ) { | |
| 184 | + $include_store_wide = true; | |
| 185 | + } else { | |
| 186 | + $kept[] = $cat; | |
| 187 | + } | |
| 188 | + } | |
| 189 | + $categories = $kept; | |
| 104 | 190 | } |
| 105 | 191 | |
| 192 | + // Subjects the owner removed on the profile screen must be skipped, not merely | |
| 193 | + // left out of the hint list — the AI sees the whole content digest and would | |
| 194 | + // otherwise write about them anyway. | |
| 195 | + $scope = $this->topic_scope( $profile, $content_type, $categories ); | |
| 196 | + | |
| 106 | 197 | $payload = [ |
| 107 | 198 | 'profile' => $profile, |
| 108 | 199 | 'categories' => $categories, |
| 109 | 200 | 'options' => [ |
| 110 | 201 | 'content_type' => $content_type, |
| 111 | - 'maxCategories' => self::MAX_CATEGORIES, | |
| 112 | - 'maxArticlesPerCategory' => self::MAX_ARTICLES_PER_CATEGORY, | |
| 202 | + 'maxCategories' => $this->max_categories( $content_type ), | |
| 203 | + 'maxArticlesPerCategory' => $this->max_articles( $content_type ), | |
| 113 | 204 | 'locale' => isset( $profile['site']['locale'] ) ? $profile['site']['locale'] : get_locale(), |
| 205 | + // Optional one-line steer from the owner ("what should these FAQs cover?"). | |
| 206 | + 'intent' => sanitize_text_field( (string) $request->get_param( 'intent' ) ), | |
| 207 | + 'exclude_topics' => $scope['exclude'], | |
| 114 | 208 | ], |
| 115 | 209 | ]; |
| 116 | 210 | |
| 117 | - $response = $this->call_proxy( $payload ); | |
| 211 | + $response = $this->call_proxy( $payload, $content_type ); | |
| 118 | 212 | |
| 119 | 213 | if ( is_wp_error( $response ) ) { |
| 120 | 214 | $code = $response->get_error_code(); |
| 121 | 215 | |
| @@ -126,8 +220,14 @@ | ||
| 126 | 220 | * @param string $code Error code. |
| 127 | 221 | */ |
| 128 | 222 | do_action( 'betterdocs_sample_docs_generation_failed', $content_type, $code ); |
| 129 | 223 | |
| 224 | + // NOTE: no store-policy fallback for the Product tab. StoreFaqContent generates | |
| 225 | + // generic store policy (Payments/Shipping/Returns), which is exactly what the | |
| 226 | + // product FAQ was fixed to stop producing — silently serving it on a proxy | |
| 227 | + // outage would just reintroduce the bug under a different trigger. Surface the | |
| 228 | + // error instead and let the user retry. | |
| 229 | + | |
| 130 | 230 | // Typed errors the UI maps to the quota / fallback screens. |
| 131 | 231 | $data = $response->get_error_data(); |
| 132 | 232 | $status = is_array( $data ) && isset( $data['status'] ) ? $data['status'] : 502; |
| 133 | 233 | |
| @@ -139,14 +239,32 @@ | ||
| 139 | 239 | * |
| 140 | 240 | * @param string $content_type docs|faq |
| 141 | 241 | * @param int $count Number of categories returned. |
| 142 | 242 | */ |
| 143 | - do_action( 'betterdocs_sample_docs_generated', $content_type, count( $response['categories'] ) ); | |
| 243 | + $categories = $response['categories']; | |
| 144 | 244 | |
| 245 | + // Layer 1 (WooCommerce only): prepend ONE deterministic, settings-grounded | |
| 246 | + // store-wide group (payments/shipping/returns/orders) flagged to show on every | |
| 247 | + // product page. The AI writes the per-category product groups (Layer 2); the | |
| 248 | + // store-wide policy answers come from the real Woo settings, not the AI, so they | |
| 249 | + // can never invent a return window or gateway the store doesn't have. | |
| 250 | + if ( 'product_faq' === $content_type && $include_store_wide ) { | |
| 251 | + $store_wide = $this->store_wide_group( $profile ); | |
| 252 | + if ( ! empty( $store_wide ) ) { | |
| 253 | + array_unshift( $categories, $store_wide ); | |
| 254 | + } | |
| 255 | + } | |
| 256 | + | |
| 257 | + do_action( 'betterdocs_sample_docs_generated', $content_type, count( $categories ) ); | |
| 258 | + | |
| 259 | + // Usage telemetry: one successful generation, bucketed by content type | |
| 260 | + // (docs/faq/product_faq). No single post here, so post_id = 0. | |
| 261 | + AIUsage::record( 'sample_docs', 0, $content_type ); | |
| 262 | + | |
| 145 | 263 | return $this->success( |
| 146 | 264 | [ |
| 147 | 265 | 'content_type' => $content_type, |
| 148 | - 'categories' => $response['categories'], | |
| 266 | + 'categories' => $categories, | |
| 149 | 267 | 'meta' => isset( $response['meta'] ) ? $response['meta'] : [], |
| 150 | 268 | ] |
| 151 | 269 | ); |
| 152 | 270 | } |
| @@ -151,34 +269,38 @@ | ||
| 151 | 269 | ); |
| 152 | 270 | } |
| 153 | 271 | |
| 154 | 272 | /** |
| 155 | - * Build the deterministic, store-grounded Product FAQ payload from the site's | |
| 156 | - * real WooCommerce settings (payments, shipping, returns/refunds, account). This | |
| 157 | - * is the primary generator for product_faq — accurate, instant, and AI-free. | |
| 273 | + * The consolidated store-wide product FAQ group (Layer 1), tagged so the builder | |
| 274 | + * flags it "show on all products". Built deterministically from the store's real | |
| 275 | + * WooCommerce settings via StoreFaqContent — no AI, no hallucinated policy. | |
| 158 | 276 | * |
| 159 | - * @return array { content_type, categories, meta } | |
| 277 | + * @return array|null A sanitized category array with `all_products => true`, or null. | |
| 160 | 278 | */ |
| 161 | - protected function build_store_faqs( $profile, $content_type, $selected = [] ) { | |
| 162 | - $store = $this->store_faq(); | |
| 163 | - $categories = $store ? $store->generate( (array) $profile ) : []; | |
| 164 | - // Honor the categories the user kept on the profile screen — the generator | |
| 165 | - // otherwise always emits the full curated set, so removing a chip had no | |
| 166 | - // effect on what got generated (fbs). | |
| 167 | - $categories = $this->filter_selected_categories( $categories, (array) $selected ); | |
| 168 | - $categories = $this->sanitize_categories( $categories, true, $this->max_categories( $content_type ) ); | |
| 279 | + protected function store_wide_group( array $profile ) { | |
| 280 | + $store = $this->store_faq(); | |
| 281 | + if ( null === $store ) { | |
| 282 | + return null; | |
| 283 | + } | |
| 169 | 284 | |
| 170 | - do_action( 'betterdocs_sample_docs_generated', $content_type, count( $categories ) ); | |
| 285 | + $group = $store->generate_consolidated( $profile ); | |
| 286 | + if ( empty( $group['articles'] ) ) { | |
| 287 | + return null; | |
| 288 | + } | |
| 171 | 289 | |
| 172 | - return [ | |
| 173 | - 'content_type' => $content_type, | |
| 174 | - 'categories' => $categories, | |
| 175 | - 'meta' => [ 'model' => 'betterdocs-local', 'source' => 'deterministic' ], | |
| 176 | - ]; | |
| 290 | + // Reuse the standard sanitizer (one group, its own question count — not the | |
| 291 | + // per-category caps), then tag it for the all-products routing in the builder. | |
| 292 | + $clean = $this->sanitize_categories( [ $group ], true, 1, count( $group['articles'] ) ); | |
| 293 | + if ( empty( $clean[0] ) ) { | |
| 294 | + return null; | |
| 295 | + } | |
| 296 | + | |
| 297 | + $clean[0]['all_products'] = true; | |
| 298 | + return $clean[0]; | |
| 177 | 299 | } |
| 178 | 300 | |
| 179 | 301 | /** |
| 180 | - * Lazily resolve the deterministic store-FAQ generator. | |
| 302 | + * Lazily resolve the deterministic store-FAQ generator (Layer 1 answers). | |
| 181 | 303 | * |
| 182 | 304 | * @return \WPDeveloper\BetterDocs\Core\StoreFaqContent|null |
| 183 | 305 | */ |
| 184 | 306 | protected function store_faq() { |
| @@ -189,8 +311,118 @@ | ||
| 189 | 311 | return $this->container->get( $class ); |
| 190 | 312 | } |
| 191 | 313 | |
| 192 | 314 | /** |
| 315 | + * Deep KB step 1 — design the whole knowledge base. Sends the content-enriched | |
| 316 | + * site profile to the proxy's /outline endpoint and returns the information | |
| 317 | + * architecture + a job_token the article step reuses. | |
| 318 | + */ | |
| 319 | + public function outline( WP_REST_Request $request ) { | |
| 320 | + if ( ! $this->is_enabled() ) { | |
| 321 | + return $this->error( 'feature_disabled', __( 'AI sample docs is disabled.', 'betterdocs' ), 403 ); | |
| 322 | + } | |
| 323 | + | |
| 324 | + // The deep flow is documentation-only; FAQ/product_faq keep the single-call path. | |
| 325 | + $profile = (array) $request->get_param( 'profile' ); | |
| 326 | + if ( empty( $profile ) || empty( $profile['content'] ) ) { | |
| 327 | + // Enrich with the real content digest the outline call needs. | |
| 328 | + $profile = $this->profiler->build( (bool) $request->get_param( 'fresh' ), true ); | |
| 329 | + } | |
| 330 | + | |
| 331 | + $intent = sanitize_text_field( (string) $request->get_param( 'intent' ) ); | |
| 332 | + | |
| 333 | + // Honor the topics the owner kept on the profile screen. The outline call used to | |
| 334 | + // send NO categories at all, so removing a chip did nothing to the generated KB — | |
| 335 | + // the deep flow silently lost the "skip this category" behaviour the single-call | |
| 336 | + // flow had. Both the kept and the REMOVED subjects go to the proxy: naming what to | |
| 337 | + // skip is what actually keeps it out, since the AI still sees the whole content | |
| 338 | + // digest and would otherwise design that category right back in. | |
| 339 | + $scope = $this->topic_scope( $profile, 'docs', (array) $request->get_param( 'categories' ) ); | |
| 340 | + | |
| 341 | + $payload = [ | |
| 342 | + 'profile' => $profile, | |
| 343 | + 'options' => [ | |
| 344 | + 'content_type' => 'docs', | |
| 345 | + 'locale' => isset( $profile['site']['locale'] ) ? $profile['site']['locale'] : get_locale(), | |
| 346 | + // Upper bounds only — the proxy prompt tells the AI to right-size the KB | |
| 347 | + // to what the site genuinely needs and return fewer when appropriate. | |
| 348 | + 'max_categories' => self::MAX_KB_CATEGORIES, | |
| 349 | + 'max_articles' => self::MAX_KB_ARTICLES_TOTAL, | |
| 350 | + 'intent' => $intent, | |
| 351 | + 'topics' => $scope['include'], | |
| 352 | + 'exclude_topics' => $scope['exclude'], | |
| 353 | + ], | |
| 354 | + ]; | |
| 355 | + | |
| 356 | + $parsed = $this->proxy_request( 'outline', $payload, 30 ); | |
| 357 | + if ( is_wp_error( $parsed ) ) { | |
| 358 | + $data = $parsed->get_error_data(); | |
| 359 | + $http = is_array( $data ) && isset( $data['status'] ) ? (int) $data['status'] : 502; | |
| 360 | + $upstream = is_array( $data ) && isset( $data['upstream'] ) ? (int) $data['upstream'] : $http; | |
| 361 | + do_action( 'betterdocs_sample_docs_generation_failed', 'docs', $parsed->get_error_code() ); | |
| 362 | + return $this->error( $parsed->get_error_code(), $parsed->get_error_message(), $http, [ 'fallback' => 'static', 'proxy_status' => $upstream ] ); | |
| 363 | + } | |
| 364 | + | |
| 365 | + if ( empty( $parsed['outline']['categories'] ) || empty( $parsed['job_token'] ) ) { | |
| 366 | + return $this->error( 'proxy_error', __( 'The AI service returned an unexpected response.', 'betterdocs' ), 502 ); | |
| 367 | + } | |
| 368 | + | |
| 369 | + $outline = $this->sanitize_outline( $parsed['outline'] ); | |
| 370 | + | |
| 371 | + do_action( 'betterdocs_sample_docs_generated', 'docs', count( $outline['categories'] ) ); | |
| 372 | + | |
| 373 | + // Usage telemetry: the deep-KB outline is the once-per-generation success point | |
| 374 | + // for the docs flow (article() runs per-article and must NOT be counted). | |
| 375 | + AIUsage::record( 'sample_docs', 0, 'docs' ); | |
| 376 | + | |
| 377 | + return $this->success( | |
| 378 | + [ | |
| 379 | + 'content_type' => 'docs', | |
| 380 | + 'job_token' => sanitize_text_field( (string) $parsed['job_token'] ), | |
| 381 | + 'outline' => $outline, | |
| 382 | + 'meta' => isset( $parsed['meta'] ) ? $parsed['meta'] : [], | |
| 383 | + ] | |
| 384 | + ); | |
| 385 | + } | |
| 386 | + | |
| 387 | + /** | |
| 388 | + * Deep KB step 2 — expand one article of a previously issued outline. Thin pass- | |
| 389 | + * through to the proxy's /article endpoint; the React flow loops it per index. | |
| 390 | + */ | |
| 391 | + public function article( WP_REST_Request $request ) { | |
| 392 | + if ( ! $this->is_enabled() ) { | |
| 393 | + return $this->error( 'feature_disabled', __( 'AI sample docs is disabled.', 'betterdocs' ), 403 ); | |
| 394 | + } | |
| 395 | + | |
| 396 | + $job_token = sanitize_text_field( (string) $request->get_param( 'job_token' ) ); | |
| 397 | + $index = (int) $request->get_param( 'index' ); | |
| 398 | + | |
| 399 | + if ( '' === $job_token || $index < 0 ) { | |
| 400 | + return $this->error( 'bad_request', __( 'A job token and article index are required.', 'betterdocs' ), 400 ); | |
| 401 | + } | |
| 402 | + | |
| 403 | + $parsed = $this->proxy_request( 'article', [ 'job_token' => $job_token, 'index' => $index ], 30 ); | |
| 404 | + if ( is_wp_error( $parsed ) ) { | |
| 405 | + $data = $parsed->get_error_data(); | |
| 406 | + $http = is_array( $data ) && isset( $data['status'] ) ? (int) $data['status'] : 502; | |
| 407 | + $upstream = is_array( $data ) && isset( $data['upstream'] ) ? (int) $data['upstream'] : $http; | |
| 408 | + return $this->error( $parsed->get_error_code(), $parsed->get_error_message(), $http, [ 'proxy_status' => $upstream ] ); | |
| 409 | + } | |
| 410 | + | |
| 411 | + if ( empty( $parsed['article']['content_html'] ) ) { | |
| 412 | + return $this->error( 'proxy_error', __( 'The AI service returned an unexpected response.', 'betterdocs' ), 502 ); | |
| 413 | + } | |
| 414 | + | |
| 415 | + return $this->success( | |
| 416 | + [ | |
| 417 | + 'index' => isset( $parsed['index'] ) ? (int) $parsed['index'] : $index, | |
| 418 | + 'article' => $this->sanitize_kb_article( $parsed['article'] ), | |
| 419 | + 'meta' => isset( $parsed['meta'] ) ? $parsed['meta'] : [], | |
| 420 | + ] | |
| 421 | + ); | |
| 422 | + } | |
| 423 | + | |
| 424 | + /** | |
| 193 | 425 | * Insert step — DocBuilder creates the terms/posts. Wired in Phase 4. |
| 194 | 426 | */ |
| 195 | 427 | public function insert( WP_REST_Request $request ) { |
| 196 | 428 | if ( ! $this->is_enabled() ) { |
| @@ -202,10 +434,52 @@ | ||
| 202 | 434 | return $this->error( 'not_implemented', __( 'Inserting sample docs is not available yet.', 'betterdocs' ), 501 ); |
| 203 | 435 | } |
| 204 | 436 | |
| 205 | 437 | $content_type = $this->content_type( $request ); |
| 206 | - $categories = $this->sanitize_categories( (array) $request->get_param( 'categories' ), true, $this->max_categories( $content_type ) ); | |
| 207 | 438 | |
| 439 | + $raw = (array) $request->get_param( 'categories' ); | |
| 440 | + | |
| 441 | + // Docs come from the deep outline→expand flow (up to 8 categories × 6 articles); | |
| 442 | + // the 3×3 sanitizer would silently truncate them. FAQ/product_faq keep the caps. | |
| 443 | + if ( 'docs' === $content_type ) { | |
| 444 | + $categories = $this->sanitize_kb_categories( $raw ); | |
| 445 | + } elseif ( 'product_faq' === $content_type ) { | |
| 446 | + // The store-wide group (Layer 1) is one extra group on top of the per-category | |
| 447 | + // cap, so it must not count against it — split it out, cap the per-category | |
| 448 | + // groups, then re-attach it. Otherwise the last product category is dropped. | |
| 449 | + $store_wide = []; | |
| 450 | + $per_cat = []; | |
| 451 | + foreach ( $raw as $cat ) { | |
| 452 | + if ( ! empty( $cat['all_products'] ) ) { | |
| 453 | + $store_wide[] = $cat; | |
| 454 | + } else { | |
| 455 | + $per_cat[] = $cat; | |
| 456 | + } | |
| 457 | + } | |
| 458 | + $categories = $this->sanitize_categories( | |
| 459 | + $per_cat, | |
| 460 | + true, | |
| 461 | + $this->max_categories( $content_type ), | |
| 462 | + $this->max_articles( $content_type ), | |
| 463 | + $this->max_articles_total( $content_type ) | |
| 464 | + ); | |
| 465 | + if ( ! empty( $store_wide[0] ) ) { | |
| 466 | + $clean = $this->sanitize_categories( [ $store_wide[0] ], true, 1, count( (array) ( $store_wide[0]['articles'] ?? [] ) ) ); | |
| 467 | + if ( ! empty( $clean[0] ) ) { | |
| 468 | + $clean[0]['all_products'] = true; | |
| 469 | + array_unshift( $categories, $clean[0] ); | |
| 470 | + } | |
| 471 | + } | |
| 472 | + } else { | |
| 473 | + $categories = $this->sanitize_categories( | |
| 474 | + $raw, | |
| 475 | + true, | |
| 476 | + $this->max_categories( $content_type ), | |
| 477 | + $this->max_articles( $content_type ), | |
| 478 | + $this->max_articles_total( $content_type ) | |
| 479 | + ); | |
| 480 | + } | |
| 481 | + | |
| 208 | 482 | $result = $builder->build( $categories, $content_type ); |
| 209 | 483 | |
| 210 | 484 | if ( is_wp_error( $result ) ) { |
| 211 | 485 | return $this->error( $result->get_error_code(), $result->get_error_message(), 400 ); |
| @@ -249,26 +523,61 @@ | ||
| 249 | 523 | /* Proxy plumbing */ |
| 250 | 524 | /* --------------------------------------------------------------------- */ |
| 251 | 525 | |
| 252 | 526 | /** |
| 253 | - * Sign + POST the payload to the hosted proxy, with one retry. | |
| 527 | + * Legacy single-call generation — sign + POST to the proxy and validate the | |
| 528 | + * { categories, meta } shape (used by the FAQ path). | |
| 254 | 529 | * |
| 255 | 530 | * @return array|\WP_Error Parsed { categories, meta } on success. |
| 256 | 531 | */ |
| 257 | - protected function call_proxy( array $payload ) { | |
| 258 | - $url = $this->proxy_url(); | |
| 532 | + protected function call_proxy( array $payload, $content_type = 'docs' ) { | |
| 533 | + $parsed = $this->proxy_request( '', $payload ); | |
| 534 | + if ( is_wp_error( $parsed ) ) { | |
| 535 | + return $parsed; | |
| 536 | + } | |
| 537 | + | |
| 538 | + if ( empty( $parsed['categories'] ) || ! is_array( $parsed['categories'] ) ) { | |
| 539 | + return $this->error( 'proxy_error', __( 'The AI service returned an unexpected response.', 'betterdocs' ), 502 ); | |
| 540 | + } | |
| 541 | + | |
| 542 | + // Enforce caps + sanitize defensively on our side too — with THIS content type's | |
| 543 | + // caps. Sanitizing an FAQ with the docs 3x3 defaults would silently throw away | |
| 544 | + // every group and question the proxy right-sized beyond the third. | |
| 545 | + $parsed['categories'] = $this->sanitize_categories( | |
| 546 | + $parsed['categories'], | |
| 547 | + true, | |
| 548 | + $this->max_categories( $content_type ), | |
| 549 | + $this->max_articles( $content_type ), | |
| 550 | + $this->max_articles_total( $content_type ) | |
| 551 | + ); | |
| 552 | + | |
| 553 | + return $parsed; | |
| 554 | + } | |
| 555 | + | |
| 556 | + /** | |
| 557 | + * Sign + POST a payload to a hosted-proxy action ('' = legacy generate, 'outline', | |
| 558 | + * 'article'), with one retry on transport failure. Returns the parsed JSON array on | |
| 559 | + * a 2xx response, or a typed WP_Error otherwise. Shape validation is the caller's job. | |
| 560 | + * | |
| 561 | + * @param string $action Sub-path under v1/sample-docs ('' | 'outline' | 'article'). | |
| 562 | + * @param array $payload Request body. | |
| 563 | + * @param int $timeout_base Base HTTP timeout in seconds (per single OpenAI call). | |
| 564 | + * @return array|\WP_Error | |
| 565 | + */ | |
| 566 | + protected function proxy_request( $action, array $payload, $timeout_base = 20 ) { | |
| 567 | + $url = $this->proxy_endpoint( $action ); | |
| 259 | 568 | $secret = $this->proxy_secret(); |
| 260 | 569 | $body = wp_json_encode( $payload ); |
| 261 | - $timeout = $this->request_timeout(); | |
| 570 | + $timeout = $this->request_timeout( $timeout_base ); | |
| 262 | 571 | |
| 263 | 572 | $args = [ |
| 264 | 573 | 'timeout' => $timeout, |
| 265 | 574 | 'headers' => [ |
| 266 | - 'Content-Type' => 'application/json', | |
| 267 | - 'Accept' => 'application/json', | |
| 268 | - 'X-BetterDocs-Site' => esc_url_raw( home_url() ), | |
| 269 | - 'X-BetterDocs-License' => $this->license_key(), | |
| 270 | - 'X-BetterDocs-Signature' => hash_hmac( 'sha256', $body, $secret ), | |
| 575 | + 'Content-Type' => 'application/json', | |
| 576 | + 'Accept' => 'application/json', | |
| 577 | + 'X-BetterDocs-Site' => esc_url_raw( home_url() ), | |
| 578 | + 'X-BetterDocs-License' => $this->license_key(), | |
| 579 | + 'X-BetterDocs-Signature' => hash_hmac( 'sha256', $body, $secret ), | |
| 271 | 580 | ], |
| 272 | 581 | 'body' => $body, |
| 273 | 582 | ]; |
| 274 | 583 | |
| @@ -299,9 +608,9 @@ | ||
| 299 | 608 | } |
| 300 | 609 | } |
| 301 | 610 | |
| 302 | 611 | if ( is_wp_error( $response ) ) { |
| 303 | - return $this->error( 'proxy_unreachable', __( 'Could not reach the BetterDocs AI service. Please try again.', 'betterdocs' ), 502 ); | |
| 612 | + return $this->error( 'proxy_unreachable', __( 'Could not reach the BetterDocs AI service. Please try again.', 'betterdocs' ), 502, [ 'upstream' => 0 ] ); | |
| 304 | 613 | } |
| 305 | 614 | |
| 306 | 615 | $status = wp_remote_retrieve_response_code( $response ); |
| 307 | 616 | $parsed = json_decode( wp_remote_retrieve_body( $response ), true ); |
| @@ -306,31 +615,32 @@ | ||
| 306 | 615 | $status = wp_remote_retrieve_response_code( $response ); |
| 307 | 616 | $parsed = json_decode( wp_remote_retrieve_body( $response ), true ); |
| 308 | 617 | |
| 309 | 618 | if ( 429 === $status || ( isset( $parsed['status'] ) && 'quota_exceeded' === $parsed['status'] ) ) { |
| 310 | - return $this->error( 'quota_exceeded', __( 'You have used your free AI generation for this site.', 'betterdocs' ), 429 ); | |
| 619 | + return $this->error( 'quota_exceeded', __( 'You have used your free AI generation for this site.', 'betterdocs' ), 429, [ 'upstream' => 429 ] ); | |
| 311 | 620 | } |
| 312 | 621 | |
| 313 | - if ( $status >= 400 || empty( $parsed['categories'] ) || ! is_array( $parsed['categories'] ) ) { | |
| 622 | + if ( $status >= 400 || ! is_array( $parsed ) ) { | |
| 314 | 623 | $message = isset( $parsed['message'] ) ? $parsed['message'] : __( 'The AI service returned an unexpected response.', 'betterdocs' ); |
| 315 | - return $this->error( 'proxy_error', $message, 502 ); | |
| 624 | + // Preserve the real upstream status under a distinct key so callers (and the | |
| 625 | + // UI) can special-case e.g. 404 (old proxy → classic fallback) or 410 (job | |
| 626 | + // expired → regenerate); error() itself overwrites data['status'] with $status. | |
| 627 | + return $this->error( 'proxy_error', $message, 502, [ 'upstream' => $status ? $status : 502 ] ); | |
| 316 | 628 | } |
| 317 | 629 | |
| 318 | - // Enforce caps + sanitize defensively on our side too. | |
| 319 | - $parsed['categories'] = $this->sanitize_categories( $parsed['categories'], true ); | |
| 320 | - | |
| 321 | 630 | return $parsed; |
| 322 | 631 | } |
| 323 | 632 | |
| 324 | 633 | /** |
| 325 | - * HTTP timeout (seconds) for the proxy call, kept safely below PHP's | |
| 634 | + * HTTP timeout (seconds) for a proxy call, kept safely below PHP's | |
| 326 | 635 | * max_execution_time so a hung upstream returns a clean WP_Error (typed |
| 327 | 636 | * JSON + static fallback in the wizard) instead of fataling mid-cURL. |
| 328 | 637 | * |
| 638 | + * @param int $base Base timeout in seconds for a single OpenAI call. | |
| 329 | 639 | * @return int |
| 330 | 640 | */ |
| 331 | - protected function request_timeout() { | |
| 332 | - $timeout = 20; | |
| 641 | + protected function request_timeout( $base = 20 ) { | |
| 642 | + $timeout = max( 5, (int) $base ); | |
| 333 | 643 | $max_exec = (int) ini_get( 'max_execution_time' ); |
| 334 | 644 | |
| 335 | 645 | if ( $max_exec > 0 ) { |
| 336 | 646 | $timeout = min( $timeout, max( 5, $max_exec - 10 ) ); |
| @@ -339,13 +649,19 @@ | ||
| 339 | 649 | /** Filter the HTTP timeout (seconds) for hosted AI proxy requests. */ |
| 340 | 650 | return (int) apply_filters( 'betterdocs_ai_proxy_timeout', $timeout ); |
| 341 | 651 | } |
| 342 | 652 | |
| 343 | - protected function proxy_url() { | |
| 653 | + /** | |
| 654 | + * Full proxy endpoint URL for an action ('' | 'outline' | 'article'). | |
| 655 | + * | |
| 656 | + * @return string | |
| 657 | + */ | |
| 658 | + protected function proxy_endpoint( $action = '' ) { | |
| 344 | 659 | $base = get_option( 'betterdocs_ai_proxy_url', 'https://api.betterdocs.co/ai' ); |
| 345 | 660 | /** Filter the hosted proxy base URL. */ |
| 346 | 661 | $base = apply_filters( 'betterdocs_ai_proxy_url', $base ); |
| 347 | - return trailingslashit( $base ) . 'v1/sample-docs'; | |
| 662 | + $url = trailingslashit( $base ) . 'v1/sample-docs'; | |
| 663 | + return '' !== $action ? $url . '/' . ltrim( (string) $action, '/' ) : $url; | |
| 348 | 664 | } |
| 349 | 665 | |
| 350 | 666 | protected function proxy_secret() { |
| 351 | 667 | $secret = get_option( 'betterdocs_ai_proxy_secret', 'betterdocs-local-dev-secret' ); |
| @@ -376,83 +692,205 @@ | ||
| 376 | 692 | |
| 377 | 693 | /** |
| 378 | 694 | * Per-content-type category cap. Product FAQs cover the four store-ops groups |
| 379 | 695 | * (Payments & Billing, Shipping & Delivery, Returns & Refunds, Orders & Account); |
| 380 | - * docs and general FAQ keep the default cap. | |
| 696 | + * a general FAQ is right-sized by the AI up to the FAQ cap; docs keep the legacy | |
| 697 | + * single-call cap (the real KB goes through outline→expand). | |
| 381 | 698 | * |
| 382 | 699 | * @return int |
| 383 | 700 | */ |
| 384 | 701 | protected function max_categories( $content_type ) { |
| 385 | - return 'product_faq' === $content_type ? 4 : self::MAX_CATEGORIES; | |
| 702 | + if ( 'product_faq' === $content_type ) { | |
| 703 | + return self::MAX_PRODUCT_FAQ_CATEGORIES; | |
| 704 | + } | |
| 705 | + return 'docs' === $content_type ? self::MAX_CATEGORIES : self::MAX_FAQ_CATEGORIES; | |
| 386 | 706 | } |
| 387 | 707 | |
| 388 | 708 | /** |
| 709 | + * Per-content-type cap on entries within one category/group. | |
| 710 | + * | |
| 711 | + * @return int | |
| 712 | + */ | |
| 713 | + protected function max_articles( $content_type ) { | |
| 714 | + return 'docs' === $content_type ? self::MAX_ARTICLES_PER_CATEGORY : self::MAX_FAQ_ARTICLES_PER_CATEGORY; | |
| 715 | + } | |
| 716 | + | |
| 717 | + /** | |
| 718 | + * Per-content-type cap on total entries (0 = no separate total cap). | |
| 719 | + * | |
| 720 | + * @return int | |
| 721 | + */ | |
| 722 | + protected function max_articles_total( $content_type ) { | |
| 723 | + if ( 'product_faq' === $content_type ) { | |
| 724 | + return self::MAX_PRODUCT_FAQ_ARTICLES_TOTAL; | |
| 725 | + } | |
| 726 | + return 'docs' === $content_type ? 0 : self::MAX_FAQ_ARTICLES_TOTAL; | |
| 727 | + } | |
| 728 | + | |
| 729 | + /** | |
| 389 | 730 | * Deterministic, type-aware suggested categories (no AI) for the profile screen. |
| 390 | 731 | * |
| 391 | 732 | * @return array |
| 392 | 733 | */ |
| 393 | 734 | protected function suggested_categories( array $profile, $content_type ) { |
| 394 | - $type = isset( $profile['type'] ) ? $profile['type'] : 'general'; | |
| 735 | + // WooCommerce Product FAQ: the chips are the store's REAL product categories — | |
| 736 | + // this tab writes FAQs about what the store SELLS. (It used to show the canned | |
| 737 | + // store-policy groups from StoreFaqContent: Payments, Shipping, Returns…) | |
| 738 | + if ( 'product_faq' === $content_type ) { | |
| 739 | + return $this->product_topics( $profile ); | |
| 740 | + } | |
| 395 | 741 | |
| 396 | - // WooCommerce Product FAQ: the curated store starter set is the single | |
| 397 | - // source of truth (StoreFaqContent), so the chips/titles shown here match | |
| 398 | - // what the deterministic fallback generates. | |
| 399 | - if ( 'product_faq' === $content_type ) { | |
| 400 | - $store = $this->store_faq(); | |
| 401 | - $defs = $store ? $store->definition() : []; | |
| 402 | - $defs = array_slice( $defs, 0, $this->max_categories( $content_type ) ); | |
| 742 | + // General FAQ AND docs both show the SUBJECTS actually detected on the site (doc | |
| 743 | + // categories, key pages) as scope chips — no canned lists. Docs used to show a | |
| 744 | + // hardcoded preset structure ("Getting Started / Shipping & Delivery / Product | |
| 745 | + // Guides") with placeholder article titles, which the profile screen counted as a | |
| 746 | + // fixed "3 categories, 9 docs". That count was always misleading: the deep | |
| 747 | + // outline flow designs and right-sizes the REAL knowledge base from the site | |
| 748 | + // content, so the preset numbers never matched what got generated. The owner | |
| 749 | + // prunes the detected subjects to scope generation; the proxy designs the rest. | |
| 750 | + return $this->detected_topics( $profile ); | |
| 751 | + } | |
| 403 | 752 | |
| 404 | - $cats = []; | |
| 405 | - foreach ( $defs as $i => $group ) { | |
| 406 | - $cats[] = [ | |
| 407 | - 'id' => sanitize_key( 'sd_' . $i ), | |
| 408 | - 'name' => $group['name'], | |
| 409 | - 'icon' => isset( $group['icon'] ) ? sanitize_key( $group['icon'] ) : 'help', | |
| 410 | - 'color' => self::PALETTE[ $i % count( self::PALETTE ) ], | |
| 411 | - 'articles' => array_values( (array) $group['questions'] ), | |
| 412 | - ]; | |
| 753 | + /** | |
| 754 | + * Subjects actually detected ON the site, for the GENERAL FAQ profile screen — doc | |
| 755 | + * categories first (what the site already documents is what people ask about), then | |
| 756 | + * real key pages (Pricing, Security, Integrations…). These are SUBJECTS, not FAQ | |
| 757 | + * group names: the AI designs the groups, the questions and the answers itself. | |
| 758 | + * Pruning a chip scopes the FAQ. | |
| 759 | + * | |
| 760 | + * Product categories are NOT included: they are the WooCommerce tab's material | |
| 761 | + * (see product_topics()), and a general site FAQ has no business offering to write | |
| 762 | + * about "Hoodies". | |
| 763 | + * | |
| 764 | + * Returns [] on a site with no usable content — the AI then works from the site | |
| 765 | + * profile alone rather than from a canned list. | |
| 766 | + * | |
| 767 | + * @return array | |
| 768 | + */ | |
| 769 | + /** | |
| 770 | + * Subjects for the WooCommerce (Product FAQ) tab: what the store actually SELLS. | |
| 771 | + * | |
| 772 | + * Real product categories first (they map 1:1 onto the FAQ group the AI designs, so | |
| 773 | + * the group can then be assigned to that product category and show on those product | |
| 774 | + * pages), falling back to product names on a store with no categories. Never store | |
| 775 | + * policy — a shopper on a product page asks about the product. | |
| 776 | + * | |
| 777 | + * @return array | |
| 778 | + */ | |
| 779 | + protected function product_topics( array $profile ) { | |
| 780 | + $woo = isset( $profile['woocommerce'] ) && is_array( $profile['woocommerce'] ) ? $profile['woocommerce'] : []; | |
| 781 | + $content = isset( $profile['content'] ) && is_array( $profile['content'] ) ? $profile['content'] : []; | |
| 782 | + | |
| 783 | + $names = []; | |
| 784 | + | |
| 785 | + if ( ! empty( $woo['product_categories'] ) ) { | |
| 786 | + $names = array_map( 'strval', (array) $woo['product_categories'] ); | |
| 787 | + } | |
| 788 | + | |
| 789 | + // No product categories (a small store selling a handful of items): fall back to | |
| 790 | + // the products themselves. | |
| 791 | + if ( empty( $names ) ) { | |
| 792 | + if ( ! empty( $content['products'] ) && is_array( $content['products'] ) ) { | |
| 793 | + foreach ( $content['products'] as $product ) { | |
| 794 | + if ( ! empty( $product['title'] ) ) { | |
| 795 | + $names[] = (string) $product['title']; | |
| 796 | + } | |
| 797 | + } | |
| 798 | + } elseif ( ! empty( $woo['sample_products'] ) ) { | |
| 799 | + $names = array_map( 'strval', (array) $woo['sample_products'] ); | |
| 413 | 800 | } |
| 801 | + } | |
| 414 | 802 | |
| 415 | - return $cats; | |
| 803 | + $chips = $this->topic_chips( $names, self::MAX_PRODUCT_FAQ_TOPICS ); | |
| 804 | + | |
| 805 | + // Offer the always-available store-wide group as the FIRST chip, so the owner can | |
| 806 | + // see — and, by removing it, opt out of — the deterministic "Shipping, Returns & | |
| 807 | + // Payments" group that shows on every product page. It is NOT a product category; | |
| 808 | + // the `all_products` flag tells generate() to route it to the store-wide layer | |
| 809 | + // rather than the AI. | |
| 810 | + $store = $this->store_faq(); | |
| 811 | + if ( $store ) { | |
| 812 | + $def = $store->consolidated_definition(); | |
| 813 | + array_unshift( | |
| 814 | + $chips, | |
| 815 | + [ | |
| 816 | + 'id' => 'sd_storewide', | |
| 817 | + 'name' => $def['name'], | |
| 818 | + 'icon' => isset( $def['icon'] ) ? sanitize_key( $def['icon'] ) : 'truck', | |
| 819 | + 'color' => self::PALETTE[0], | |
| 820 | + 'all_products' => true, | |
| 821 | + // Real question titles so the profile screen shows a count. | |
| 822 | + 'articles' => array_values( (array) $def['questions'] ), | |
| 823 | + ] | |
| 824 | + ); | |
| 416 | 825 | } |
| 417 | 826 | |
| 418 | - if ( 'faq' === $content_type ) { | |
| 419 | - // FAQ groups: product-store questions on a WooCommerce/ecommerce | |
| 420 | - // site, general site questions otherwise. | |
| 421 | - $names = 'ecommerce' === $type | |
| 422 | - ? [ 'Shipping & Delivery', 'Returns & Refunds', 'Payments & Orders' ] | |
| 423 | - : [ 'Getting Started', 'Account & Billing', 'Troubleshooting' ]; | |
| 424 | - } else { | |
| 425 | - // Documentation categories — article-style guides, NOT FAQ groups (no | |
| 426 | - // "FAQs" category here; FAQs are a separate content type). The hosted | |
| 427 | - // proxy refines these names to fit the specific site and writes real | |
| 428 | - // article titles + bodies for each. | |
| 429 | - $presets = [ | |
| 430 | - 'ecommerce' => [ 'Getting Started', 'Shipping & Delivery', 'Product Guides' ], | |
| 431 | - 'course_lms' => [ 'Getting Started', 'Course Access', 'Lessons & Content' ], | |
| 432 | - 'digital_downloads' => [ 'Getting Started', 'Downloads & Licenses', 'Installation' ], | |
| 433 | - 'membership' => [ 'Getting Started', 'Membership & Plans', 'Member Benefits' ], | |
| 434 | - 'community' => [ 'Getting Started', 'Using the Community', 'Your Profile' ], | |
| 435 | - 'business_site' => [ 'Getting Started', 'Our Services', 'How-to Guides' ], | |
| 436 | - 'general' => [ 'Getting Started', 'Guides & How-tos', 'Features & Settings' ], | |
| 437 | - ]; | |
| 438 | - $names = isset( $presets[ $type ] ) ? $presets[ $type ] : $presets['general']; | |
| 827 | + return $chips; | |
| 828 | + } | |
| 829 | + | |
| 830 | + protected function detected_topics( array $profile ) { | |
| 831 | + $topics = isset( $profile['topics'] ) && is_array( $profile['topics'] ) ? $profile['topics'] : []; | |
| 832 | + $content = isset( $profile['content'] ) && is_array( $profile['content'] ) ? $profile['content'] : []; | |
| 833 | + | |
| 834 | + $names = []; | |
| 835 | + | |
| 836 | + // 1. What the site already documents. | |
| 837 | + if ( ! empty( $topics['doc_categories'] ) ) { | |
| 838 | + $names = array_merge( $names, array_map( 'strval', (array) $topics['doc_categories'] ) ); | |
| 439 | 839 | } |
| 440 | 840 | |
| 441 | - // Never suggest more than the cap, so the preview matches what actually | |
| 442 | - // gets generated and inserted. | |
| 443 | - $names = array_slice( $names, 0, $this->max_categories( $content_type ) ); | |
| 841 | + // 2. Real key pages — minus the boilerplate every WP site has, which nobody | |
| 842 | + // writes an FAQ group about. | |
| 843 | + if ( ! empty( $content['pages'] ) && is_array( $content['pages'] ) ) { | |
| 844 | + foreach ( $content['pages'] as $page ) { | |
| 845 | + $title = isset( $page['title'] ) ? (string) $page['title'] : ''; | |
| 846 | + if ( '' !== $title && ! $this->is_boilerplate_page( $title ) ) { | |
| 847 | + $names[] = $title; | |
| 848 | + } | |
| 849 | + } | |
| 850 | + } | |
| 444 | 851 | |
| 445 | - $icons = [ 'book', 'truck', 'help', 'refund' ]; | |
| 852 | + // NOTE: product categories are deliberately NOT topics here. They belong to the | |
| 853 | + // WooCommerce (Product FAQ) tab, which is generated per product category — pulling | |
| 854 | + // them into the General FAQ made a general site FAQ offer to write about | |
| 855 | + // "Clothing", "Hoodies", "Music". | |
| 446 | 856 | |
| 857 | + return $this->topic_chips( $names ); | |
| 858 | + } | |
| 859 | + | |
| 860 | + /** | |
| 861 | + * Turn a raw list of subject names into profile-screen chips: de-duped | |
| 862 | + * case-insensitively (first spelling wins) and bounded to MAX_FAQ_TOPICS. | |
| 863 | + * | |
| 864 | + * @return array | |
| 865 | + */ | |
| 866 | + protected function topic_chips( array $names, $max = self::MAX_FAQ_TOPICS ) { | |
| 867 | + $max = max( 1, (int) $max ); | |
| 868 | + $seen = []; | |
| 869 | + $clean = []; | |
| 870 | + foreach ( $names as $name ) { | |
| 871 | + $name = trim( wp_strip_all_tags( (string) $name ) ); | |
| 872 | + $key = strtolower( $name ); | |
| 873 | + if ( '' === $name || isset( $seen[ $key ] ) ) { | |
| 874 | + continue; | |
| 875 | + } | |
| 876 | + $seen[ $key ] = true; | |
| 877 | + $clean[] = $name; | |
| 878 | + if ( count( $clean ) >= $max ) { | |
| 879 | + break; | |
| 880 | + } | |
| 881 | + } | |
| 882 | + | |
| 447 | 883 | $cats = []; |
| 448 | - foreach ( $names as $i => $name ) { | |
| 884 | + foreach ( $clean as $i => $name ) { | |
| 449 | 885 | $cats[] = [ |
| 450 | 886 | 'id' => sanitize_key( 'sd_' . $i ), |
| 451 | - 'name' => $name, | |
| 452 | - 'icon' => $icons[ $i % count( $icons ) ], | |
| 887 | + 'name' => sanitize_text_field( $name ), | |
| 888 | + 'icon' => 'help', | |
| 453 | 889 | 'color' => self::PALETTE[ $i % count( self::PALETTE ) ], |
| 454 | - 'articles' => $this->seed_articles( $name, $content_type ), | |
| 890 | + // No seeded questions: the AI writes them. The FAQ profile screen counts | |
| 891 | + // topics, not questions, so nothing needs a placeholder here. | |
| 892 | + 'articles' => [], | |
| 455 | 893 | ]; |
| 456 | 894 | } |
| 457 | 895 | |
| 458 | 896 | return $cats; |
| @@ -458,28 +896,67 @@ | ||
| 458 | 896 | return $cats; |
| 459 | 897 | } |
| 460 | 898 | |
| 461 | 899 | /** |
| 462 | - * Placeholder article/question titles so the profile-step counters render. | |
| 463 | - * The proxy replaces these with real generated content. | |
| 900 | + * Pages that are site *plumbing* — a cart, a login form, an index — and so carry no | |
| 901 | + * subject anyone asks a question about. | |
| 464 | 902 | * |
| 465 | - * @return array | |
| 903 | + * Policy pages (privacy, terms, refunds) are deliberately NOT skipped: they are | |
| 904 | + * plumbing for documentation but they are prime FAQ material ("Do you sell my data?", | |
| 905 | + * "Can I get a refund?"). Skipping them meant a site with no docs yet — the normal | |
| 906 | + * case when generating samples — detected zero topics, while the AI, which reads the | |
| 907 | + * whole content digest rather than these chips, went on to build "Privacy & Data" and | |
| 908 | + * "Terms & Licensing" groups from them anyway. | |
| 909 | + * | |
| 910 | + * @return bool | |
| 466 | 911 | */ |
| 467 | - protected function seed_articles( $name, $content_type ) { | |
| 468 | - if ( 'faq' === $content_type ) { | |
| 469 | - return [ | |
| 470 | - /* translators: %s: suggested FAQ group / category name. */ | |
| 471 | - sprintf( __( 'What is %s?', 'betterdocs' ), $name ), | |
| 472 | - __( 'Common questions', 'betterdocs' ), | |
| 473 | - __( 'Tips & best practices', 'betterdocs' ), | |
| 474 | - ]; | |
| 912 | + protected function is_boilerplate_page( $title ) { | |
| 913 | + $skip = [ 'home', 'homepage', 'front page', 'sample page', 'blog', 'news', 'shop', 'store', 'cart', 'checkout', 'my account', 'account', 'login', 'log in', 'register', 'sign up', 'search results', '404', 'page not found' ]; | |
| 914 | + return in_array( strtolower( trim( wp_strip_all_tags( (string) $title ) ) ), $skip, true ); | |
| 915 | + } | |
| 916 | + | |
| 917 | + /** | |
| 918 | + * What the owner kept, and what they REMOVED, on the profile screen. | |
| 919 | + * | |
| 920 | + * Removing a chip must actually skip that subject — for docs, FAQ and product FAQ | |
| 921 | + * alike. Sending only the kept list isn't enough: the AI also sees the site's full | |
| 922 | + * content digest, so a removed subject happily reappears unless it is named as | |
| 923 | + * off-limits. The removed set is derived server-side by diffing the deterministic | |
| 924 | + * suggestion list against what the client sent back, so no client change (and no | |
| 925 | + * trust in the client) is needed. | |
| 926 | + * | |
| 927 | + * @param array $kept Categories the client sent back (each with a 'name'). | |
| 928 | + * @param string $content_type docs|faq|product_faq | |
| 929 | + * @return array { include: string[], exclude: string[] } | |
| 930 | + */ | |
| 931 | + protected function topic_scope( array $profile, $content_type, array $kept ) { | |
| 932 | + $suggested = $this->suggested_categories( $profile, $content_type ); | |
| 933 | + | |
| 934 | + $kept_names = []; | |
| 935 | + foreach ( $kept as $cat ) { | |
| 936 | + $name = is_array( $cat ) ? ( isset( $cat['name'] ) ? $cat['name'] : '' ) : $cat; | |
| 937 | + $name = trim( wp_strip_all_tags( (string) $name ) ); | |
| 938 | + if ( '' !== $name ) { | |
| 939 | + $kept_names[ $this->normalize_category_name( $name ) ] = $name; | |
| 940 | + } | |
| 475 | 941 | } |
| 476 | 942 | |
| 943 | + $exclude = []; | |
| 944 | + foreach ( $suggested as $cat ) { | |
| 945 | + // The store-wide chip is not a real topic the AI writes about (it's the | |
| 946 | + // deterministic Layer-1 group), so never push it into the AI's exclude list. | |
| 947 | + if ( is_array( $cat ) && ! empty( $cat['all_products'] ) ) { | |
| 948 | + continue; | |
| 949 | + } | |
| 950 | + $name = isset( $cat['name'] ) ? (string) $cat['name'] : ''; | |
| 951 | + if ( '' !== $name && ! isset( $kept_names[ $this->normalize_category_name( $name ) ] ) ) { | |
| 952 | + $exclude[] = $name; | |
| 953 | + } | |
| 954 | + } | |
| 955 | + | |
| 477 | 956 | return [ |
| 478 | - /* translators: %s: suggested category name. */ | |
| 479 | - sprintf( __( '%s — overview', 'betterdocs' ), $name ), | |
| 480 | - __( 'Key concepts', 'betterdocs' ), | |
| 481 | - __( 'Step-by-step guide', 'betterdocs' ), | |
| 957 | + 'include' => array_values( $kept_names ), | |
| 958 | + 'exclude' => $exclude, | |
| 482 | 959 | ]; |
| 483 | 960 | } |
| 484 | 961 | |
| 485 | 962 | /** |
| @@ -532,22 +1009,30 @@ | ||
| 532 | 1009 | * Sanitize + cap an incoming categories array (used both for the user's edited |
| 533 | 1010 | * list and the proxy response). |
| 534 | 1011 | * |
| 535 | 1012 | * @param bool $with_content Keep article body HTML (proxy response) vs titles only. |
| 536 | - * @param int $max Category cap (defaults to MAX_CATEGORIES; product_faq uses more). | |
| 1013 | + * @param int $max Category cap (defaults to MAX_CATEGORIES; FAQ uses more). | |
| 1014 | + * @param int $max_articles Per-category entry cap. | |
| 1015 | + * @param int $max_total Total entry cap across all categories (0 = none). | |
| 537 | 1016 | * @return array |
| 538 | 1017 | */ |
| 539 | - protected function sanitize_categories( array $categories, $with_content = false, $max = self::MAX_CATEGORIES ) { | |
| 540 | - $max = max( 1, (int) $max ); | |
| 541 | - $clean = []; | |
| 1018 | + protected function sanitize_categories( array $categories, $with_content = false, $max = self::MAX_CATEGORIES, $max_articles = self::MAX_ARTICLES_PER_CATEGORY, $max_total = 0 ) { | |
| 1019 | + $max = max( 1, (int) $max ); | |
| 1020 | + $max_articles = max( 1, (int) $max_articles ); | |
| 1021 | + $max_total = max( 0, (int) $max_total ); | |
| 1022 | + $total = 0; | |
| 1023 | + $clean = []; | |
| 542 | 1024 | foreach ( array_slice( $categories, 0, $max ) as $i => $cat ) { |
| 543 | - if ( empty( $cat['name'] ) ) { | |
| 1025 | + if ( empty( $cat['name'] ) || ( $max_total > 0 && $total >= $max_total ) ) { | |
| 544 | 1026 | continue; |
| 545 | 1027 | } |
| 546 | 1028 | |
| 547 | 1029 | $articles = []; |
| 548 | 1030 | $raw = isset( $cat['articles'] ) && is_array( $cat['articles'] ) ? $cat['articles'] : []; |
| 549 | - foreach ( array_slice( $raw, 0, self::MAX_ARTICLES_PER_CATEGORY ) as $article ) { | |
| 1031 | + foreach ( array_slice( $raw, 0, $max_articles ) as $article ) { | |
| 1032 | + if ( $max_total > 0 && $total >= $max_total ) { | |
| 1033 | + break; | |
| 1034 | + } | |
| 550 | 1035 | if ( is_array( $article ) ) { |
| 551 | 1036 | $entry = [ |
| 552 | 1037 | 'title' => sanitize_text_field( isset( $article['title'] ) ? $article['title'] : '' ), |
| 553 | 1038 | 'excerpt' => sanitize_text_field( isset( $article['excerpt'] ) ? $article['excerpt'] : '' ), |
| @@ -556,13 +1041,15 @@ | ||
| 556 | 1041 | $entry['content_html'] = wp_kses_post( isset( $article['content_html'] ) ? $article['content_html'] : '' ); |
| 557 | 1042 | } |
| 558 | 1043 | if ( '' !== $entry['title'] ) { |
| 559 | 1044 | $articles[] = $entry; |
| 1045 | + $total++; | |
| 560 | 1046 | } |
| 561 | 1047 | } else { |
| 562 | 1048 | $title = sanitize_text_field( $article ); |
| 563 | 1049 | if ( '' !== $title ) { |
| 564 | 1050 | $articles[] = $with_content ? [ 'title' => $title, 'content_html' => '', 'excerpt' => '' ] : $title; |
| 1051 | + $total++; | |
| 565 | 1052 | } |
| 566 | 1053 | } |
| 567 | 1054 | } |
| 568 | 1055 | |
| @@ -568,8 +1055,147 @@ | ||
| 568 | 1055 | |
| 569 | 1056 | $clean[] = [ |
| 570 | 1057 | 'id' => sanitize_key( isset( $cat['id'] ) ? $cat['id'] : 'sd_' . $i ), |
| 571 | 1058 | 'name' => sanitize_text_field( $cat['name'] ), |
| 1059 | + 'description' => isset( $cat['description'] ) ? sanitize_text_field( $cat['description'] ) : '', | |
| 1060 | + 'icon' => isset( $cat['icon'] ) ? sanitize_key( $cat['icon'] ) : 'book', | |
| 1061 | + 'color' => sanitize_hex_color( isset( $cat['color'] ) ? $cat['color'] : '' ) ?: self::PALETTE[ $i % count( self::PALETTE ) ], | |
| 1062 | + // Product FAQ: the WooCommerce product category this group is about, so the | |
| 1063 | + // builder can attach the group to it (the FAQ then shows on those product | |
| 1064 | + // pages instead of nowhere). | |
| 1065 | + 'product_category' => isset( $cat['product_category'] ) ? sanitize_text_field( $cat['product_category'] ) : '', | |
| 1066 | + // Product FAQ Layer 1: the consolidated store-wide group, flagged so the | |
| 1067 | + // builder assigns it to "all products" rather than one category. | |
| 1068 | + 'all_products' => ! empty( $cat['all_products'] ), | |
| 1069 | + 'articles' => $articles, | |
| 1070 | + ]; | |
| 1071 | + } | |
| 1072 | + | |
| 1073 | + return $clean; | |
| 1074 | + } | |
| 1075 | + | |
| 1076 | + /* --------------------------------------------------------------------- */ | |
| 1077 | + /* Deep KB sanitizers */ | |
| 1078 | + /* --------------------------------------------------------------------- */ | |
| 1079 | + | |
| 1080 | + /** | |
| 1081 | + * Sanitize the proxy's designed outline for the client: clamp counts, keep the | |
| 1082 | + * per-article specs (index/type/slug/summary), and assign a display color per | |
| 1083 | + * category. Does NOT run the 3×3 sanitizer. | |
| 1084 | + * | |
| 1085 | + * @return array { total_articles, categories:[ { id, name, slug, icon, color, description, articles:[…] } ] } | |
| 1086 | + */ | |
| 1087 | + protected function sanitize_outline( array $outline ) { | |
| 1088 | + $cats_in = isset( $outline['categories'] ) && is_array( $outline['categories'] ) ? $outline['categories'] : []; | |
| 1089 | + | |
| 1090 | + $categories = []; | |
| 1091 | + $total = 0; | |
| 1092 | + foreach ( array_slice( $cats_in, 0, self::MAX_KB_CATEGORIES ) as $i => $cat ) { | |
| 1093 | + if ( empty( $cat['name'] ) || $total >= self::MAX_KB_ARTICLES_TOTAL ) { | |
| 1094 | + continue; | |
| 1095 | + } | |
| 1096 | + | |
| 1097 | + $articles = []; | |
| 1098 | + $arts_in = isset( $cat['articles'] ) && is_array( $cat['articles'] ) ? $cat['articles'] : []; | |
| 1099 | + foreach ( array_slice( $arts_in, 0, self::MAX_KB_ARTICLES_PER_CATEGORY ) as $art ) { | |
| 1100 | + if ( $total >= self::MAX_KB_ARTICLES_TOTAL || empty( $art['title'] ) ) { | |
| 1101 | + continue; | |
| 1102 | + } | |
| 1103 | + $links = []; | |
| 1104 | + if ( isset( $art['links'] ) && is_array( $art['links'] ) ) { | |
| 1105 | + foreach ( $art['links'] as $l ) { | |
| 1106 | + $links[] = sanitize_title( (string) $l ); | |
| 1107 | + } | |
| 1108 | + } | |
| 1109 | + $articles[] = [ | |
| 1110 | + 'index' => isset( $art['index'] ) ? (int) $art['index'] : $total, | |
| 1111 | + 'type' => sanitize_key( isset( $art['type'] ) ? $art['type'] : 'guide' ), | |
| 1112 | + 'title' => sanitize_text_field( $art['title'] ), | |
| 1113 | + 'slug' => sanitize_title( isset( $art['slug'] ) ? $art['slug'] : $art['title'] ), | |
| 1114 | + 'summary' => sanitize_text_field( isset( $art['summary'] ) ? $art['summary'] : '' ), | |
| 1115 | + 'links' => array_values( array_filter( $links ) ), | |
| 1116 | + ]; | |
| 1117 | + $total++; | |
| 1118 | + } | |
| 1119 | + | |
| 1120 | + if ( empty( $articles ) ) { | |
| 1121 | + continue; | |
| 1122 | + } | |
| 1123 | + | |
| 1124 | + $categories[] = [ | |
| 1125 | + 'id' => sanitize_key( isset( $cat['id'] ) ? 'sd_' . $cat['id'] : 'sd_' . $i ), | |
| 1126 | + 'name' => sanitize_text_field( $cat['name'] ), | |
| 1127 | + 'slug' => sanitize_title( isset( $cat['slug'] ) ? $cat['slug'] : $cat['name'] ), | |
| 1128 | + 'icon' => sanitize_key( isset( $cat['icon'] ) ? $cat['icon'] : 'book' ), | |
| 1129 | + 'color' => self::PALETTE[ count( $categories ) % count( self::PALETTE ) ], | |
| 1130 | + 'description' => sanitize_text_field( isset( $cat['description'] ) ? $cat['description'] : '' ), | |
| 1131 | + 'articles' => $articles, | |
| 1132 | + ]; | |
| 1133 | + } | |
| 1134 | + | |
| 1135 | + return [ | |
| 1136 | + 'total_articles' => $total, | |
| 1137 | + 'categories' => $categories, | |
| 1138 | + ]; | |
| 1139 | + } | |
| 1140 | + | |
| 1141 | + /** | |
| 1142 | + * Sanitize a single expanded article. Keeps the `#bd-link--slug` cross-link | |
| 1143 | + * sentinels (fragment hrefs survive wp_kses_post) for the builder to resolve. | |
| 1144 | + * | |
| 1145 | + * @return array { title, slug, type, category_slug, content_html, excerpt } | |
| 1146 | + */ | |
| 1147 | + protected function sanitize_kb_article( array $article ) { | |
| 1148 | + return [ | |
| 1149 | + 'title' => sanitize_text_field( isset( $article['title'] ) ? $article['title'] : '' ), | |
| 1150 | + 'slug' => sanitize_title( isset( $article['slug'] ) ? $article['slug'] : '' ), | |
| 1151 | + 'type' => sanitize_key( isset( $article['type'] ) ? $article['type'] : 'guide' ), | |
| 1152 | + 'category_slug' => sanitize_title( isset( $article['category_slug'] ) ? $article['category_slug'] : '' ), | |
| 1153 | + 'content_html' => wp_kses_post( isset( $article['content_html'] ) ? $article['content_html'] : '' ), | |
| 1154 | + 'excerpt' => sanitize_text_field( isset( $article['excerpt'] ) ? $article['excerpt'] : '' ), | |
| 1155 | + ]; | |
| 1156 | + } | |
| 1157 | + | |
| 1158 | + /** | |
| 1159 | + * Sanitize + cap an assembled deep KB (categories with expanded articles) for | |
| 1160 | + * insertion. Preserves article slug/type (needed for cross-link resolution and | |
| 1161 | + * intro/quickstart ordering) and keeps content HTML. | |
| 1162 | + * | |
| 1163 | + * @return array | |
| 1164 | + */ | |
| 1165 | + protected function sanitize_kb_categories( array $categories ) { | |
| 1166 | + $clean = []; | |
| 1167 | + $total = 0; | |
| 1168 | + | |
| 1169 | + foreach ( array_slice( $categories, 0, self::MAX_KB_CATEGORIES ) as $i => $cat ) { | |
| 1170 | + if ( empty( $cat['name'] ) || $total >= self::MAX_KB_ARTICLES_TOTAL ) { | |
| 1171 | + continue; | |
| 1172 | + } | |
| 1173 | + | |
| 1174 | + $articles = []; | |
| 1175 | + $raw = isset( $cat['articles'] ) && is_array( $cat['articles'] ) ? $cat['articles'] : []; | |
| 1176 | + foreach ( array_slice( $raw, 0, self::MAX_KB_ARTICLES_PER_CATEGORY ) as $article ) { | |
| 1177 | + if ( $total >= self::MAX_KB_ARTICLES_TOTAL || ! is_array( $article ) || empty( $article['title'] ) ) { | |
| 1178 | + continue; | |
| 1179 | + } | |
| 1180 | + $articles[] = [ | |
| 1181 | + 'title' => sanitize_text_field( $article['title'] ), | |
| 1182 | + 'slug' => sanitize_title( isset( $article['slug'] ) ? $article['slug'] : $article['title'] ), | |
| 1183 | + 'type' => sanitize_key( isset( $article['type'] ) ? $article['type'] : 'guide' ), | |
| 1184 | + 'content_html' => wp_kses_post( isset( $article['content_html'] ) ? $article['content_html'] : '' ), | |
| 1185 | + 'excerpt' => sanitize_text_field( isset( $article['excerpt'] ) ? $article['excerpt'] : '' ), | |
| 1186 | + ]; | |
| 1187 | + $total++; | |
| 1188 | + } | |
| 1189 | + | |
| 1190 | + if ( empty( $articles ) ) { | |
| 1191 | + continue; | |
| 1192 | + } | |
| 1193 | + | |
| 1194 | + $clean[] = [ | |
| 1195 | + 'id' => sanitize_key( isset( $cat['id'] ) ? $cat['id'] : 'sd_' . $i ), | |
| 1196 | + 'name' => sanitize_text_field( $cat['name'] ), | |
| 1197 | + 'slug' => sanitize_title( isset( $cat['slug'] ) ? $cat['slug'] : $cat['name'] ), | |
| 572 | 1198 | 'description' => isset( $cat['description'] ) ? sanitize_text_field( $cat['description'] ) : '', |
| 573 | 1199 | 'icon' => isset( $cat['icon'] ) ? sanitize_key( $cat['icon'] ) : 'book', |
| 574 | 1200 | 'color' => sanitize_hex_color( isset( $cat['color'] ) ? $cat['color'] : '' ) ?: self::PALETTE[ $i % count( self::PALETTE ) ], |
| 575 | 1201 | 'articles' => $articles, |
