| 1 |
<?php |
| 2 |
/** |
| 3 |
* Content classification WordPress Ability implementation. |
| 4 |
* |
| 5 |
* @package WordPress\AI |
| 6 |
*/ |
| 7 |
|
| 8 |
declare( strict_types=1 ); |
| 9 |
|
| 10 |
namespace WordPress\AI\Abilities\Content_Classification; |
| 11 |
|
| 12 |
use WP_Error; |
| 13 |
use WP_Post; |
| 14 |
use WP_Post_Type; |
| 15 |
use WP_Taxonomy; |
| 16 |
use WordPress\AI\Abstracts\Abstract_Ability; |
| 17 |
use WordPress\AI\Experiments\Content_Classification\Content_Classification as Content_Classification_Experiment; |
| 18 |
|
| 19 |
use function WordPress\AI\get_post_context; |
| 20 |
use function WordPress\AI\normalize_content; |
| 21 |
|
| 22 |
/** |
| 23 |
* Content classification WordPress Ability. |
| 24 |
* |
| 25 |
* Generates taxonomy term suggestions based on post content analysis. |
| 26 |
* |
| 27 |
* @since 0.7.0 |
| 28 |
*/ |
| 29 |
class Content_Classification extends Abstract_Ability { |
| 30 |
|
| 31 |
/** |
| 32 |
* Default minimum confidence below which a suggestion is dropped. |
| 33 |
* |
| 34 |
* The system prompt instructs the model to use 0.5 = "somewhat |
| 35 |
* relevant" / 1.0 = "perfect match". In practice true positives |
| 36 |
* cluster at ≥0.85 and generic / popularity-driven false positives |
| 37 |
* sit in the 0.5–0.78 range, so filtering at 0.6 removes most |
| 38 |
* noise without meaningfully impacting recall. Filterable via |
| 39 |
* `wpai_content_classification_min_confidence`. |
| 40 |
* |
| 41 |
* @since 1.3.0 |
| 42 |
* |
| 43 |
* @var float |
| 44 |
*/ |
| 45 |
public const MIN_CONFIDENCE = 0.6; |
| 46 |
|
| 47 |
/** |
| 48 |
* {@inheritDoc} |
| 49 |
* |
| 50 |
* @since 0.8.0 |
| 51 |
*/ |
| 52 |
protected function guideline_categories(): array { |
| 53 |
return array( 'site', 'copy' ); |
| 54 |
} |
| 55 |
|
| 56 |
/** |
| 57 |
* Returns the input schema of the ability. |
| 58 |
* |
| 59 |
* @since 0.7.0 |
| 60 |
* |
| 61 |
* @return array<string, mixed> The input schema of the ability. |
| 62 |
*/ |
| 63 |
protected function input_schema(): array { |
| 64 |
return array( |
| 65 |
'type' => 'object', |
| 66 |
'properties' => array( |
| 67 |
'content' => array( |
| 68 |
'type' => 'string', |
| 69 |
'description' => esc_html__( 'Content to generate taxonomy suggestions for.', 'ai' ), |
| 70 |
), |
| 71 |
'post_id' => array( |
| 72 |
'type' => 'integer', |
| 73 |
'description' => esc_html__( 'Content from this post will be used to generate taxonomy suggestions. This overrides the content parameter if both are provided.', 'ai' ), |
| 74 |
), |
| 75 |
'taxonomy' => array( |
| 76 |
'type' => 'string', |
| 77 |
'default' => 'post_tag', |
| 78 |
'description' => esc_html__( 'The taxonomy to generate suggestions for (e.g., post_tag, category).', 'ai' ), |
| 79 |
), |
| 80 |
'strategy' => array( |
| 81 |
'type' => 'string', |
| 82 |
'default' => Content_Classification_Experiment::STRATEGY_EXISTING_ONLY, |
| 83 |
'description' => esc_html__( 'The suggestion strategy: existing_only or allow_new.', 'ai' ), |
| 84 |
), |
| 85 |
'max_suggestions' => array( |
| 86 |
'type' => 'integer', |
| 87 |
'minimum' => 1, |
| 88 |
'maximum' => 10, |
| 89 |
'default' => Content_Classification_Experiment::DEFAULT_MAX_SUGGESTIONS, |
| 90 |
'description' => esc_html__( 'Maximum number of suggestions to generate.', 'ai' ), |
| 91 |
), |
| 92 |
), |
| 93 |
); |
| 94 |
} |
| 95 |
|
| 96 |
/** |
| 97 |
* Returns the output schema of the ability. |
| 98 |
* |
| 99 |
* @since 0.7.0 |
| 100 |
* |
| 101 |
* @return array<string, mixed> The output schema of the ability. |
| 102 |
*/ |
| 103 |
protected function output_schema(): array { |
| 104 |
return array( |
| 105 |
'type' => 'object', |
| 106 |
'properties' => array( |
| 107 |
'suggestions' => array( |
| 108 |
'type' => 'array', |
| 109 |
'description' => esc_html__( 'Generated taxonomy term suggestions.', 'ai' ), |
| 110 |
'items' => array( |
| 111 |
'type' => 'object', |
| 112 |
'properties' => array( |
| 113 |
'term' => array( |
| 114 |
'type' => 'string', |
| 115 |
'description' => esc_html__( 'The suggested term name.', 'ai' ), |
| 116 |
), |
| 117 |
'confidence' => array( |
| 118 |
'type' => 'number', |
| 119 |
'description' => esc_html__( 'Confidence score between 0 and 1.', 'ai' ), |
| 120 |
), |
| 121 |
'is_new' => array( |
| 122 |
'type' => 'boolean', |
| 123 |
'description' => esc_html__( 'Whether this is a new term or an existing one.', 'ai' ), |
| 124 |
), |
| 125 |
'parent' => array( |
| 126 |
'type' => 'string', |
| 127 |
'description' => esc_html__( 'Parent term name for hierarchical taxonomies.', 'ai' ), |
| 128 |
), |
| 129 |
), |
| 130 |
), |
| 131 |
), |
| 132 |
), |
| 133 |
); |
| 134 |
} |
| 135 |
|
| 136 |
/** |
| 137 |
* Executes the ability with the given input arguments. |
| 138 |
* |
| 139 |
* @since 0.7.0 |
| 140 |
* |
| 141 |
* @param mixed $input The input arguments to the ability. |
| 142 |
* @return array{suggestions: array<array{term: string, confidence: float, is_new: bool, parent?: string}>}|\WP_Error The result of the ability execution, or a WP_Error on failure. |
| 143 |
*/ |
| 144 |
protected function execute_callback( $input ) { |
| 145 |
// Default arguments. |
| 146 |
$args = wp_parse_args( |
| 147 |
$input, |
| 148 |
array( |
| 149 |
'content' => null, |
| 150 |
'post_id' => null, |
| 151 |
'taxonomy' => 'post_tag', |
| 152 |
'strategy' => Content_Classification_Experiment::STRATEGY_EXISTING_ONLY, |
| 153 |
'max_suggestions' => (int) Content_Classification_Experiment::DEFAULT_MAX_SUGGESTIONS, |
| 154 |
), |
| 155 |
); |
| 156 |
|
| 157 |
// Validate taxonomy. |
| 158 |
if ( ! taxonomy_exists( $args['taxonomy'] ) ) { |
| 159 |
return new WP_Error( |
| 160 |
'invalid_taxonomy', |
| 161 |
/* translators: %s: Taxonomy name. */ |
| 162 |
sprintf( esc_html__( 'Taxonomy "%s" does not exist.', 'ai' ), sanitize_key( $args['taxonomy'] ) ) |
| 163 |
); |
| 164 |
} |
| 165 |
|
| 166 |
$assigned_terms = array(); |
| 167 |
|
| 168 |
// If a post ID is provided, ensure the post exists before using its content. |
| 169 |
if ( $args['post_id'] ) { |
| 170 |
$post = get_post( (int) $args['post_id'] ); |
| 171 |
|
| 172 |
if ( ! $post instanceof WP_Post ) { |
| 173 |
return new WP_Error( |
| 174 |
'post_not_found', |
| 175 |
/* translators: %d: Post ID. */ |
| 176 |
sprintf( esc_html__( 'Post with ID %d not found.', 'ai' ), absint( $args['post_id'] ) ) |
| 177 |
); |
| 178 |
} |
| 179 |
|
| 180 |
// Get the post context. |
| 181 |
$context = get_post_context( (int) $args['post_id'] ); |
| 182 |
|
| 183 |
// Default to the passed in content if it exists. |
| 184 |
if ( $args['content'] ) { |
| 185 |
$context['content'] = normalize_content( $args['content'] ); |
| 186 |
} |
| 187 |
|
| 188 |
// Get terms already assigned to this post for the taxonomy. |
| 189 |
$assigned = wp_get_object_terms( (int) $args['post_id'], $args['taxonomy'], array( 'fields' => 'names' ) ); |
| 190 |
if ( ! is_wp_error( $assigned ) ) { |
| 191 |
$assigned_terms = (array) $assigned; |
| 192 |
} |
| 193 |
} else { |
| 194 |
$context = array( |
| 195 |
'content' => normalize_content( $args['content'] ?? '' ), |
| 196 |
); |
| 197 |
} |
| 198 |
|
| 199 |
// If we have no content, return an error. |
| 200 |
if ( empty( $context['content'] ) ) { |
| 201 |
return new WP_Error( |
| 202 |
'content_not_provided', |
| 203 |
esc_html__( 'Content is required to generate taxonomy suggestions.', 'ai' ) |
| 204 |
); |
| 205 |
} |
| 206 |
|
| 207 |
// Generate the suggestions. |
| 208 |
$result = $this->generate_suggestions( |
| 209 |
$context, |
| 210 |
$args['taxonomy'], |
| 211 |
$args['strategy'], |
| 212 |
(int) $args['max_suggestions'], |
| 213 |
$assigned_terms |
| 214 |
); |
| 215 |
|
| 216 |
// If we have an error, return it. |
| 217 |
if ( is_wp_error( $result ) ) { |
| 218 |
return $result; |
| 219 |
} |
| 220 |
|
| 221 |
// If we have no results, return an error. |
| 222 |
if ( empty( $result ) ) { |
| 223 |
return new WP_Error( |
| 224 |
'no_results', |
| 225 |
esc_html__( 'No taxonomy suggestions were generated.', 'ai' ) |
| 226 |
); |
| 227 |
} |
| 228 |
|
| 229 |
return array( |
| 230 |
'suggestions' => $result, |
| 231 |
); |
| 232 |
} |
| 233 |
|
| 234 |
/** |
| 235 |
* Returns the permission callback of the ability. |
| 236 |
* |
| 237 |
* @since 0.7.0 |
| 238 |
* |
| 239 |
* @param mixed $args The input arguments to the ability. |
| 240 |
* @return bool|\WP_Error True if the user has permission, WP_Error otherwise. |
| 241 |
*/ |
| 242 |
protected function permission_callback( $args ) { |
| 243 |
$post_id = isset( $args['post_id'] ) ? absint( $args['post_id'] ) : null; |
| 244 |
|
| 245 |
if ( $post_id ) { |
| 246 |
$post = get_post( $post_id ); |
| 247 |
|
| 248 |
// Ensure the post exists. |
| 249 |
if ( ! $post instanceof WP_Post ) { |
| 250 |
return new WP_Error( |
| 251 |
'post_not_found', |
| 252 |
/* translators: %d: Post ID. */ |
| 253 |
sprintf( esc_html__( 'Post with ID %d not found.', 'ai' ), $post_id ) |
| 254 |
); |
| 255 |
} |
| 256 |
|
| 257 |
// Ensure the user has permission to edit this particular post. |
| 258 |
if ( ! current_user_can( 'edit_post', $post_id ) ) { |
| 259 |
return new WP_Error( |
| 260 |
'insufficient_capabilities', |
| 261 |
esc_html__( 'You do not have permission to generate taxonomy suggestions for this post.', 'ai' ) |
| 262 |
); |
| 263 |
} |
| 264 |
|
| 265 |
$post_type_obj = get_post_type_object( $post->post_type ); |
| 266 |
if ( ! $post_type_obj instanceof WP_Post_Type || empty( $post_type_obj->show_in_rest ) ) { |
| 267 |
return false; |
| 268 |
} |
| 269 |
} elseif ( ! current_user_can( 'edit_posts' ) ) { |
| 270 |
// Ensure the user has permission to edit posts in general. |
| 271 |
return new WP_Error( |
| 272 |
'insufficient_capabilities', |
| 273 |
esc_html__( 'You do not have permission to generate taxonomy suggestions.', 'ai' ) |
| 274 |
); |
| 275 |
} |
| 276 |
|
| 277 |
return true; |
| 278 |
} |
| 279 |
|
| 280 |
/** |
| 281 |
* Returns the meta of the ability. |
| 282 |
* |
| 283 |
* @since 0.7.0 |
| 284 |
* |
| 285 |
* @return array<string, mixed> The meta of the ability. |
| 286 |
*/ |
| 287 |
protected function meta(): array { |
| 288 |
return array( |
| 289 |
'show_in_rest' => true, |
| 290 |
); |
| 291 |
} |
| 292 |
|
| 293 |
/** |
| 294 |
* Generates taxonomy term suggestions from the given content. |
| 295 |
* |
| 296 |
* The LLM generates suggestions based purely on content analysis |
| 297 |
* and the currently assigned terms. Post-processing then matches |
| 298 |
* suggestions against existing terms and applies the strategy. |
| 299 |
* |
| 300 |
* @since 0.7.0 |
| 301 |
* |
| 302 |
* @param string|array<string, string> $context The context to generate suggestions from. |
| 303 |
* @param string $taxonomy The taxonomy to suggest terms for. |
| 304 |
* @param string $strategy The suggestion strategy. |
| 305 |
* @param int $max_suggestions The maximum number of suggestions. |
| 306 |
* @param array<string> $assigned_terms Terms already assigned to the post. |
| 307 |
* @return array<array{term: string, confidence: float, is_new: bool, parent?: string}>|\WP_Error The generated suggestions, or a WP_Error if there was an error. |
| 308 |
*/ |
| 309 |
protected function generate_suggestions( $context, string $taxonomy, string $strategy, int $max_suggestions, array $assigned_terms = array() ) { |
| 310 |
// Convert the context to a string if it's an array. |
| 311 |
if ( is_array( $context ) ) { |
| 312 |
$context = implode( |
| 313 |
"\n", |
| 314 |
array_map( |
| 315 |
static function ( $key, $value ) { |
| 316 |
return sprintf( |
| 317 |
'%s: %s', |
| 318 |
ucwords( str_replace( '_', ' ', $key ) ), |
| 319 |
$value |
| 320 |
); |
| 321 |
}, |
| 322 |
array_keys( $context ), |
| 323 |
$context |
| 324 |
) |
| 325 |
); |
| 326 |
} |
| 327 |
|
| 328 |
// When using existing_only strategy, send the top terms to the LLM |
| 329 |
// so it can select from actual terms rather than guessing. |
| 330 |
$available_terms = array(); |
| 331 |
if ( Content_Classification_Experiment::STRATEGY_EXISTING_ONLY === $strategy ) { |
| 332 |
$available_terms = $this->get_top_terms( $taxonomy ); |
| 333 |
} |
| 334 |
|
| 335 |
/** |
| 336 |
* Filters the candidate pool of existing terms surfaced to the model as |
| 337 |
* `<available-terms>`. |
| 338 |
* |
| 339 |
* Default behaviour: the top 100 terms ordered by usage count when the |
| 340 |
* `existing_only` strategy is in use; an empty array otherwise. The |
| 341 |
* system instruction treats this pool as candidates ("use only when |
| 342 |
* they genuinely fit; relevance outweighs popularity") rather than as |
| 343 |
* required choices, so the popularity ordering is not load-bearing for |
| 344 |
* relevance — it's just the default. Sites can return a relevance- |
| 345 |
* ranked pool (e.g. via embeddings) here without touching prompt code. |
| 346 |
* |
| 347 |
* Returning an empty array suppresses the `<available-terms>` block |
| 348 |
* entirely, which lets the model rely purely on content + assigned |
| 349 |
* terms. |
| 350 |
* |
| 351 |
* @since 1.3.0 |
| 352 |
* |
| 353 |
* @param array<string> $available_terms The default candidate pool (term names). |
| 354 |
* @param string $taxonomy The taxonomy slug being suggested for. |
| 355 |
* @param string $strategy The suggestion strategy |
| 356 |
* (`existing_only` or `allow_new`). |
| 357 |
* @return array<string> The filtered candidate pool. |
| 358 |
*/ |
| 359 |
$available_terms = (array) apply_filters( |
| 360 |
'wpai_content_classification_available_terms', |
| 361 |
$available_terms, |
| 362 |
$taxonomy, |
| 363 |
$strategy |
| 364 |
); |
| 365 |
|
| 366 |
// Piece together the various prompt parts. |
| 367 |
$prompt_parts = array(); |
| 368 |
|
| 369 |
$prompt_parts[] = $this->build_taxonomy_descriptor( $taxonomy ); |
| 370 |
$prompt_parts[] = '<content>' . $context . '</content>'; |
| 371 |
|
| 372 |
// If we have currently assigned terms, add them to the prompt to avoid redundant suggestions. |
| 373 |
if ( ! empty( $assigned_terms ) ) { |
| 374 |
$prompt_parts[] = '<assigned-terms>' . implode( ', ', $assigned_terms ) . '</assigned-terms>'; |
| 375 |
} |
| 376 |
|
| 377 |
// Surface the candidate pool: the existing terms fetched for the |
| 378 |
// existing_only strategy by default, or whatever the |
| 379 |
// `wpai_content_classification_available_terms` filter injected. |
| 380 |
if ( ! empty( $available_terms ) ) { |
| 381 |
$prompt_parts[] = '<available-terms>' . implode( ', ', $available_terms ) . '</available-terms>'; |
| 382 |
} |
| 383 |
|
| 384 |
$prompt = implode( "\n", $prompt_parts ); |
| 385 |
|
| 386 |
/** |
| 387 |
* Filters the prompt string before it is sent to the AI model for taxonomy suggestion generation. |
| 388 |
* |
| 389 |
* Allows developers to modify, augment, or replace the prompt that the AI analyzes |
| 390 |
* when generating taxonomy term suggestions. |
| 391 |
* |
| 392 |
* @since 0.7.0 |
| 393 |
* |
| 394 |
* @param string $prompt The prompt string to be sent to the AI model. |
| 395 |
* @param string|array<string, string> $context The context to generate suggestions from. |
| 396 |
* @param string $taxonomy The taxonomy slug being suggested for (e.g., 'post_tag', 'category'). |
| 397 |
* @param array<string> $assigned_terms Terms already assigned to the post. |
| 398 |
* @param array<string> $available_terms Available terms to suggest from. |
| 399 |
*/ |
| 400 |
$prompt = (string) apply_filters( 'wpai_content_classification_prompt', $prompt, $context, $taxonomy, $assigned_terms, $available_terms ); |
| 401 |
|
| 402 |
$prompt_builder = $this->get_prompt_builder( $prompt ); |
| 403 |
|
| 404 |
if ( is_wp_error( $prompt_builder ) ) { |
| 405 |
return $prompt_builder; |
| 406 |
} |
| 407 |
|
| 408 |
// Generate the suggestions using the AI client with structured output. |
| 409 |
$result = $prompt_builder->generate_text(); |
| 410 |
|
| 411 |
if ( is_wp_error( $result ) ) { |
| 412 |
return $result; |
| 413 |
} |
| 414 |
|
| 415 |
// Parse, match against existing terms, filter, and limit. |
| 416 |
$suggestions = $this->parse_suggestions( $result, $strategy, $assigned_terms, $taxonomy, $max_suggestions ); |
| 417 |
|
| 418 |
if ( is_wp_error( $suggestions ) ) { |
| 419 |
return $suggestions; |
| 420 |
} |
| 421 |
|
| 422 |
/** |
| 423 |
* Filters the parsed taxonomy suggestions before they are returned to the client. |
| 424 |
* |
| 425 |
* Allows developers to modify, reorder, add, or remove suggestions after the AI |
| 426 |
* has generated them and they have been parsed into structured data. |
| 427 |
* |
| 428 |
* Each suggestion is an associative array with the keys: |
| 429 |
* - 'term' (string) The suggested term name. |
| 430 |
* - 'confidence' (float) Confidence score between 0 and 1. |
| 431 |
* - 'is_new' (bool) Whether the term is new or already exists on the site. |
| 432 |
* - 'parent' (string) Optional. Parent term name for hierarchical taxonomies. |
| 433 |
* |
| 434 |
* @since 0.7.0 |
| 435 |
* |
| 436 |
* @param array<array{term: string, confidence: float, is_new: bool, parent?: string}> $suggestions The parsed suggestions. |
| 437 |
* @param string $taxonomy The taxonomy slug (e.g., 'post_tag', 'category'). |
| 438 |
* @param string $strategy The suggestion strategy ('existing_only' or 'allow_new'). |
| 439 |
*/ |
| 440 |
return (array) apply_filters( 'wpai_content_classification_suggestions', $suggestions, $taxonomy, $strategy ); |
| 441 |
} |
| 442 |
|
| 443 |
/** |
| 444 |
* Builds the `<taxonomy …/>` descriptor block sent to the model. |
| 445 |
* |
| 446 |
* Surfaces the human label, the description, whether the taxonomy is |
| 447 |
* hierarchical, and a coarse `kind` (`category` vs `tag`) so the |
| 448 |
* model can reason about intent (broad/thematic vs specific) rather |
| 449 |
* than guessing from the raw slug. The system instruction branches |
| 450 |
* on `kind`. |
| 451 |
* |
| 452 |
* @since 1.3.0 |
| 453 |
* |
| 454 |
* @param string $taxonomy The taxonomy slug. |
| 455 |
* @return string The descriptor block (e.g. |
| 456 |
* `<taxonomy name="category" label="Categories" kind="category" hierarchical="true">…description…</taxonomy>`), |
| 457 |
* or an empty string when the taxonomy does not exist. |
| 458 |
*/ |
| 459 |
private function build_taxonomy_descriptor( string $taxonomy ): string { |
| 460 |
$taxonomy = sanitize_key( $taxonomy ); |
| 461 |
|
| 462 |
if ( '' === $taxonomy || ! taxonomy_exists( $taxonomy ) ) { |
| 463 |
return ''; |
| 464 |
} |
| 465 |
|
| 466 |
$tax_object = get_taxonomy( $taxonomy ); |
| 467 |
|
| 468 |
if ( ! $tax_object instanceof WP_Taxonomy ) { |
| 469 |
return ''; |
| 470 |
} |
| 471 |
|
| 472 |
$is_hierarchical = is_taxonomy_hierarchical( $taxonomy ); |
| 473 |
$kind = $is_hierarchical ? 'category' : 'tag'; |
| 474 |
|
| 475 |
$label = ! empty( $tax_object->labels->name ) |
| 476 |
? wp_strip_all_tags( (string) $tax_object->labels->name ) |
| 477 |
: $taxonomy; |
| 478 |
|
| 479 |
$description = trim( wp_strip_all_tags( (string) $tax_object->description ) ); |
| 480 |
|
| 481 |
// esc_attr() guards the attribute values so a label containing quotes or |
| 482 |
// angle brackets can't break the pseudo-XML the model reads. |
| 483 |
return sprintf( |
| 484 |
'<taxonomy name="%1$s" label="%2$s" kind="%3$s" hierarchical="%4$s">%5$s</taxonomy>', |
| 485 |
esc_attr( $taxonomy ), |
| 486 |
esc_attr( $label ), |
| 487 |
esc_attr( $kind ), |
| 488 |
$is_hierarchical ? 'true' : 'false', |
| 489 |
$description |
| 490 |
); |
| 491 |
} |
| 492 |
|
| 493 |
/** |
| 494 |
* Get the prompt builder for generating taxonomy term suggestions. |
| 495 |
* |
| 496 |
* @since 0.7.0 |
| 497 |
* |
| 498 |
* @param string $prompt The prompt to use for generating taxonomy term suggestions. |
| 499 |
* @return \WP_AI_Client_Prompt_Builder|\WP_Error The prompt builder, or a WP_Error on failure. |
| 500 |
*/ |
| 501 |
private function get_prompt_builder( string $prompt ) { |
| 502 |
$prompt_builder = wp_ai_client_prompt( $prompt ) |
| 503 |
->using_system_instruction( $this->get_system_instruction() ) |
| 504 |
->as_json_response( $this->suggestions_schema() ); |
| 505 |
|
| 506 |
$prompt_builder = $this->filter_prompt_builder( $prompt_builder, Content_Classification_Experiment::class, array(), $prompt ); |
| 507 |
|
| 508 |
return $this->ensure_text_generation_supported( |
| 509 |
$prompt_builder, |
| 510 |
esc_html__( 'Term generation failed. Please ensure you have a connected provider that supports text generation.', 'ai' ) |
| 511 |
); |
| 512 |
} |
| 513 |
|
| 514 |
/** |
| 515 |
* Returns the JSON schema for structured output from the AI model. |
| 516 |
* |
| 517 |
* @since 0.7.0 |
| 518 |
* |
| 519 |
* @return array<string, mixed> The JSON schema for structured output. |
| 520 |
*/ |
| 521 |
protected function suggestions_schema(): array { |
| 522 |
return array( |
| 523 |
'type' => 'object', |
| 524 |
'properties' => array( |
| 525 |
'suggestions' => array( |
| 526 |
'type' => 'array', |
| 527 |
'items' => array( |
| 528 |
'type' => 'object', |
| 529 |
'properties' => array( |
| 530 |
'term' => array( 'type' => 'string' ), |
| 531 |
'confidence' => array( 'type' => 'number' ), |
| 532 |
), |
| 533 |
'required' => array( 'term', 'confidence' ), |
| 534 |
'additionalProperties' => false, |
| 535 |
), |
| 536 |
), |
| 537 |
), |
| 538 |
'required' => array( 'suggestions' ), |
| 539 |
'additionalProperties' => false, |
| 540 |
); |
| 541 |
} |
| 542 |
|
| 543 |
/** |
| 544 |
* Parses the AI response into structured suggestions. |
| 545 |
* |
| 546 |
* Matches LLM suggestions against existing terms (case-insensitive), |
| 547 |
* filters out assigned terms, applies the strategy, sorts by confidence, |
| 548 |
* and limits to the requested number of suggestions. |
| 549 |
* |
| 550 |
* @since 0.7.0 |
| 551 |
* |
| 552 |
* @param string $response The raw AI response. |
| 553 |
* @param string $strategy The suggestion strategy ('existing_only' or 'allow_new'). |
| 554 |
* @param array<string> $assigned_terms Terms already assigned to the post. |
| 555 |
* @param string $taxonomy The taxonomy to suggest terms for. |
| 556 |
* @param int $max_suggestions The maximum number of suggestions to return. |
| 557 |
* @return array<array{term: string, confidence: float, is_new: bool, parent?: string}>|\WP_Error Parsed suggestions or error. |
| 558 |
*/ |
| 559 |
private function parse_suggestions( string $response, string $strategy, array $assigned_terms, string $taxonomy, int $max_suggestions ) { |
| 560 |
$decoded = json_decode( $response, true ); |
| 561 |
|
| 562 |
if ( ! is_array( $decoded ) || ! isset( $decoded['suggestions'] ) || ! is_array( $decoded['suggestions'] ) ) { |
| 563 |
return new WP_Error( |
| 564 |
'invalid_response', |
| 565 |
esc_html__( 'Could not parse AI response as valid suggestions.', 'ai' ) |
| 566 |
); |
| 567 |
} |
| 568 |
|
| 569 |
// Only fetch existing terms when we need them for post-processing (existing_only strategy). |
| 570 |
$existing_terms = Content_Classification_Experiment::STRATEGY_EXISTING_ONLY === $strategy |
| 571 |
? $this->get_existing_terms( $taxonomy ) |
| 572 |
: array(); |
| 573 |
|
| 574 |
// Build a lowercase → original name lookup for existing terms. |
| 575 |
// We don't use slugs here because the LLM may generate terms that don't match the taxonomy slug. |
| 576 |
if ( ! empty( $existing_terms ) ) { |
| 577 |
$existing_terms = array_combine( array_map( 'strtolower', $existing_terms ), $existing_terms ); |
| 578 |
} |
| 579 |
|
| 580 |
/** |
| 581 |
* Filters the minimum confidence threshold for a suggestion to be |
| 582 |
* returned. Suggestions with confidence below this value are dropped |
| 583 |
* before sorting and limiting to `max_suggestions`. |
| 584 |
* |
| 585 |
* @since 1.3.0 |
| 586 |
* |
| 587 |
* @param float $min_confidence The minimum confidence (0.0–1.0). |
| 588 |
* @param string $taxonomy The taxonomy slug being suggested for. |
| 589 |
* @param string $strategy The suggestion strategy. |
| 590 |
* @return float The filtered minimum confidence. |
| 591 |
*/ |
| 592 |
$min_confidence = (float) apply_filters( |
| 593 |
'wpai_content_classification_min_confidence', |
| 594 |
self::MIN_CONFIDENCE, |
| 595 |
$taxonomy, |
| 596 |
$strategy |
| 597 |
); |
| 598 |
// Clamp to the valid range so a stray filter return can't disable the floor entirely. |
| 599 |
$min_confidence = max( 0.0, min( 1.0, $min_confidence ) ); |
| 600 |
|
| 601 |
// Build a lowercase set of assigned terms for filtering. |
| 602 |
$assigned_terms = array_map( 'strtolower', $assigned_terms ); |
| 603 |
$suggestions = array(); |
| 604 |
foreach ( $decoded['suggestions'] as $item ) { |
| 605 |
if ( ! is_array( $item ) || empty( $item['term'] ) ) { |
| 606 |
continue; |
| 607 |
} |
| 608 |
|
| 609 |
$term = sanitize_text_field( trim( $item['term'] ) ); |
| 610 |
$term_lower = strtolower( $term ); |
| 611 |
$is_new = ! isset( $existing_terms[ $term_lower ] ); |
| 612 |
$confidence = isset( $item['confidence'] ) ? (float) $item['confidence'] : 0.5; |
| 613 |
// Clamp first so the floor compares against the effective confidence |
| 614 |
// rather than malformed model output. |
| 615 |
$confidence = max( 0.0, min( 1.0, $confidence ) ); |
| 616 |
|
| 617 |
// Drop suggestions below the relevance floor. |
| 618 |
if ( $confidence < $min_confidence ) { |
| 619 |
continue; |
| 620 |
} |
| 621 |
|
| 622 |
// Skip terms already assigned to the post. |
| 623 |
// The agent should avoid suggesting these, but just in case we'll check here as well. |
| 624 |
if ( in_array( $term_lower, $assigned_terms, true ) ) { |
| 625 |
continue; |
| 626 |
} |
| 627 |
|
| 628 |
// For existing_only strategy, skip terms that don't exist. |
| 629 |
if ( Content_Classification_Experiment::STRATEGY_EXISTING_ONLY === $strategy && $is_new ) { |
| 630 |
continue; |
| 631 |
} |
| 632 |
|
| 633 |
// Use the original capitalized name for existing terms. |
| 634 |
if ( ! $is_new ) { |
| 635 |
$term = $existing_terms[ $term_lower ]; |
| 636 |
} |
| 637 |
|
| 638 |
$suggestion = array( |
| 639 |
'term' => $term, |
| 640 |
'confidence' => $confidence, |
| 641 |
'is_new' => $is_new, |
| 642 |
); |
| 643 |
|
| 644 |
// Only preserve parent for hierarchical taxonomies, and strip it |
| 645 |
// when the AI returns the taxonomy slug itself as the parent. |
| 646 |
if ( |
| 647 |
! empty( $item['parent'] ) |
| 648 |
&& is_taxonomy_hierarchical( $taxonomy ) |
| 649 |
&& strtolower( trim( $item['parent'] ) ) !== strtolower( $taxonomy ) |
| 650 |
) { |
| 651 |
$suggestion['parent'] = sanitize_text_field( trim( $item['parent'] ) ); |
| 652 |
} |
| 653 |
|
| 654 |
$suggestions[] = $suggestion; |
| 655 |
} |
| 656 |
|
| 657 |
// Sort by confidence descending. |
| 658 |
usort( |
| 659 |
$suggestions, |
| 660 |
static function ( $a, $b ) { |
| 661 |
return $b['confidence'] <=> $a['confidence']; |
| 662 |
} |
| 663 |
); |
| 664 |
|
| 665 |
// Limit to max suggestions. |
| 666 |
return array_slice( $suggestions, 0, $max_suggestions ); |
| 667 |
} |
| 668 |
|
| 669 |
/** |
| 670 |
* Gets existing terms for a taxonomy. |
| 671 |
* |
| 672 |
* @since 0.7.0 |
| 673 |
* |
| 674 |
* @param string $taxonomy The taxonomy to get terms for. |
| 675 |
* @return array<string> List of existing term names. |
| 676 |
*/ |
| 677 |
private function get_existing_terms( string $taxonomy ): array { |
| 678 |
$terms = get_terms( |
| 679 |
array( |
| 680 |
'taxonomy' => $taxonomy, |
| 681 |
'hide_empty' => false, |
| 682 |
'fields' => 'names', |
| 683 |
) |
| 684 |
); |
| 685 |
|
| 686 |
if ( is_wp_error( $terms ) ) { |
| 687 |
return array(); |
| 688 |
} |
| 689 |
|
| 690 |
return (array) $terms; |
| 691 |
} |
| 692 |
|
| 693 |
/** |
| 694 |
* Gets the top terms for a taxonomy, ordered by usage count. |
| 695 |
* |
| 696 |
* Used to provide the LLM with a set of existing terms to select from |
| 697 |
* when using the existing_only strategy, improving match quality. |
| 698 |
* |
| 699 |
* @since 0.7.0 |
| 700 |
* |
| 701 |
* @param string $taxonomy The taxonomy to get terms for. |
| 702 |
* @param int $limit Maximum number of terms to return. |
| 703 |
* @return array<string> List of term names ordered by count descending. |
| 704 |
*/ |
| 705 |
private function get_top_terms( string $taxonomy, int $limit = 100 ): array { |
| 706 |
/** |
| 707 |
* Filters the maximum number of existing terms fetched for the |
| 708 |
* candidate pool surfaced to the model under the `existing_only` |
| 709 |
* strategy. |
| 710 |
* |
| 711 |
* The default of 100 suits most sites; large taxonomies may benefit |
| 712 |
* from a higher cap, while smaller sites can lower it to reduce the |
| 713 |
* prompt token count. |
| 714 |
* |
| 715 |
* @since 1.3.0 |
| 716 |
* |
| 717 |
* @param int $limit The maximum number of terms to fetch. |
| 718 |
* @param string $taxonomy The taxonomy slug being suggested for. |
| 719 |
* @return int The filtered limit. |
| 720 |
*/ |
| 721 |
$limit = (int) apply_filters( 'wpai_content_classification_candidate_pool_size', $limit, $taxonomy ); |
| 722 |
|
| 723 |
// A non-positive limit would make get_terms() return everything; |
| 724 |
// fall back to the default so the pool stays bounded. |
| 725 |
if ( $limit < 1 ) { |
| 726 |
$limit = 100; |
| 727 |
} |
| 728 |
|
| 729 |
$terms = get_terms( |
| 730 |
array( |
| 731 |
'taxonomy' => $taxonomy, |
| 732 |
'hide_empty' => false, |
| 733 |
'fields' => 'names', |
| 734 |
'orderby' => 'count', |
| 735 |
'order' => 'DESC', |
| 736 |
'number' => $limit, |
| 737 |
) |
| 738 |
); |
| 739 |
|
| 740 |
if ( is_wp_error( $terms ) ) { |
| 741 |
return array(); |
| 742 |
} |
| 743 |
|
| 744 |
return (array) $terms; |
| 745 |
} |
| 746 |
} |
| 747 |
|
