<?php
/**
* Trait WordPress\Plugin_Check\Traits\AI_Analyzer
*
* @package plugin-check
*/
namespace WordPress\Plugin_Check\Traits;
use WordPress\Plugin_Check\Admin\Settings_Page;
use WordPress\Plugin_Check\Checker\Check_Context;
use WordPress\Plugin_Check\Checker\Check_Result;
use WP_Error;
/**
* Trait for analyzing check results for false positives using AI.
*
* Uses a batched approach inspired by the internal scanner: issues are grouped
* by check code prefix, each group gets a check-specific prompt loaded from
* the prompts/ directory, and all cases in a batch are sent in a single AI
* request for efficiency.
*
* @since 2.0.0
*/
trait AI_Analyzer {
/**
* Checks if AI analysis is available via WordPress core AI client.
*
* @since 2.0.0
*
* @return bool True if AI client is available, false otherwise.
*/
protected function is_ai_available() {
if ( ! function_exists( 'wp_ai_client_prompt' ) ) {
return false;
}
// Check WP 7.0+ AI support.
if ( function_exists( 'wp_supports_ai' ) && ! wp_supports_ai() ) {
return false;
}
return true;
}
/**
* Gets the maximum number of cases to send per AI batch request.
*
* @since 2.0.0
*
* @return int Batch size.
*/
protected function get_ai_batch_size() {
return 12;
}
/**
* Gets the maximum number of cases to analyze per check type.
*
* @since 2.0.0
*
* @return int Maximum case count.
*/
protected function get_ai_max_cases_per_check() {
return 24;
}
/**
* Gets the mapping of check code prefixes to their prompt template filenames.
*
* @since 2.0.0
*
* @return array Prompt map.
*/
protected function get_ai_prompt_map() {
return array(
'WordPress.Security.EscapeOutput' => 'ai-review-late-escaping.md',
'PluginCheck.CodeAnalysis.EscapeOutput' => 'ai-review-late-escaping.md',
'WordPress.Security.NonceVerification' => 'ai-review-nonce-verification.md',
'WordPress.Security.ValidatedSanitizedInput' => 'ai-review-sanitization.md',
'WordPress.DB.DirectDatabaseQuery' => 'ai-review-direct-db-queries.md',
'WordPress.DB.PreparedSQL' => 'ai-review-direct-db-queries.md',
'PluginCheck.CodeAnalysis.Obfuscation' => 'ai-review-code-obfuscation.md',
'PluginCheck.CodeAnalysis.SettingSanitization' => 'ai-review-setting-sanitization.md',
'PluginCheck.CodeAnalysis.PluginUpdater' => 'ai-review-plugin-updater.md',
);
}
/**
* Analyzes check results for false positives using batched AI requests.
*
* Issues are grouped by check code prefix, and each group is analyzed
* with a check-specific prompt. Only issues with severity below the
* configured threshold are analyzed.
*
* @since 2.0.0
*
* @param Check_Result $result Check result to analyze.
* @param Check_Context $check_context Check context instance.
* @param string $model_preference Optional model preference.
* @return array|WP_Error Array with 'analysis' and 'stats' keys, or WP_Error on failure.
*
* @SuppressWarnings(PHPMD.NPathComplexity)
*/
protected function analyze_results_with_ai( Check_Result $result, Check_Context $check_context, $model_preference = '' ) {
if ( ! $this->is_ai_available() ) {
return new WP_Error(
'ai_not_available',
__( 'AI analysis requires WordPress 7.0 or newer with AI support enabled.', 'plugin-check' )
);
}
$errors = $result->get_errors();
$warnings = $result->get_warnings();
if ( empty( $errors ) && empty( $warnings ) ) {
return $this->empty_ai_result();
}
// Collect all issues eligible for AI review, grouped by prompt type.
$grouped_issues = $this->collect_issues_for_ai( $errors, $warnings, $check_context );
if ( empty( $grouped_issues ) ) {
return $this->empty_ai_result();
}
// Process each group with its specific prompt.
$analysis_results = array();
$total_tokens = 0;
$input_tokens = 0;
$output_tokens = 0;
$false_positives = 0;
$issues_analyzed = 0;
$models_used = array();
$providers_used = array();
foreach ( $grouped_issues as $prompt_file => $cases ) {
$batch_result = $this->analyze_batch( $prompt_file, $cases, $check_context, $model_preference );
if ( is_wp_error( $batch_result ) ) {
continue;
}
foreach ( $batch_result['cases'] as $case_analysis ) {
$case_id = $case_analysis['case_id'];
if ( isset( $cases[ $case_id ] ) ) {
$original = $cases[ $case_id ];
$analysis_results[ $case_id ] = array(
'is_false_positive' => false === $case_analysis['issue'],
'reasoning' => sanitize_text_field( $case_analysis['short_explanation'] ),
'file' => $original['file'],
'line' => $original['line'],
'column' => $original['column'],
'code' => $original['code'],
'type' => $original['type'],
);
++$issues_analyzed;
if ( false === $case_analysis['issue'] ) {
++$false_positives;
}
}
}
if ( isset( $batch_result['token_usage']['total_tokens'] ) ) {
$total_tokens += (int) $batch_result['token_usage']['total_tokens'];
}
if ( isset( $batch_result['token_usage']['prompt_tokens'] ) ) {
$input_tokens += (int) $batch_result['token_usage']['prompt_tokens'];
}
if ( isset( $batch_result['token_usage']['completion_tokens'] ) ) {
$output_tokens += (int) $batch_result['token_usage']['completion_tokens'];
}
if ( ! empty( $batch_result['model_used'] ) ) {
$models_used[] = (string) $batch_result['model_used'];
}
if ( ! empty( $batch_result['provider_used'] ) ) {
$providers_used[] = (string) $batch_result['provider_used'];
}
}
return array(
'analysis' => $analysis_results,
'stats' => array(
'tokens_spent' => $total_tokens,
'input_tokens' => $input_tokens,
'output_tokens' => $output_tokens,
'false_positives' => $false_positives,
'issues_analyzed' => $issues_analyzed,
'model_used' => implode( ', ', array_unique( $models_used ) ),
'provider_used' => implode( ', ', array_unique( $providers_used ) ),
),
);
}
/**
* Collects issues eligible for AI review, grouped by prompt template.
*
* Only issues with severity below the configured threshold are included.
*
* @since 2.0.0
*
* @param array $errors Errors from Check_Result.
* @param array $warnings Warnings from Check_Result.
* @param Check_Context $check_context Check context instance.
* @return array Issues grouped by prompt filename. Each value is an associative
* array keyed by case_id with issue metadata.
*/
protected function collect_issues_for_ai( array $errors, array $warnings, Check_Context $check_context ) {
$error_threshold = $this->get_ai_severity_threshold( 'error' );
$warning_threshold = $this->get_ai_severity_threshold( 'warning' );
$grouped = array();
$counts = array(); // Track count per prompt to enforce limit.
// Process errors.
$this->collect_issues_from_collection( $errors, 'error', $error_threshold, $check_context, $grouped, $counts );
// Process warnings.
$this->collect_issues_from_collection( $warnings, 'warning', $warning_threshold, $check_context, $grouped, $counts );
return $grouped;
}
/**
* Collects issues from a single collection (errors or warnings).
*
* @since 2.0.0
*
* @param array $collection The errors or warnings collection.
* @param string $type 'error' or 'warning'.
* @param int $threshold Severity threshold.
* @param Check_Context $check_context Check context instance.
* @param array $grouped Reference to grouped issues array.
* @param array $counts Reference to counts per prompt.
*/
protected function collect_issues_from_collection( array $collection, $type, $threshold, Check_Context $check_context, array &$grouped, array &$counts ) {
foreach ( $collection as $file => $file_issues ) {
foreach ( $file_issues as $line => $line_issues ) {
foreach ( $line_issues as $column => $column_issues ) {
foreach ( $column_issues as $issue ) {
$severity = isset( $issue['severity'] ) ? (int) $issue['severity'] : 5;
if ( $severity >= $threshold ) {
continue;
}
$code = isset( $issue['code'] ) ? $issue['code'] : '';
$prompt_file = $this->get_prompt_for_code( $code );
if ( ! isset( $counts[ $prompt_file ] ) ) {
$counts[ $prompt_file ] = 0;
}
if ( $counts[ $prompt_file ] >= $this->get_ai_max_cases_per_check() ) {
continue;
}
$case_id = $this->get_issue_key( $file, $line, $column, $code );
if ( ! isset( $grouped[ $prompt_file ] ) ) {
$grouped[ $prompt_file ] = array();
}
$grouped[ $prompt_file ][ $case_id ] = array(
'file' => $file,
'line' => $line,
'column' => $column,
'code' => $code,
'message' => isset( $issue['message'] ) ? $issue['message'] : '',
'type' => $type,
);
++$counts[ $prompt_file ];
}
}
}
}
}
/**
* Analyzes a batch of issues with a specific prompt template.
*
* If the batch exceeds the configured batch size, it is split into sub-batches
* and each sub-batch is sent as a separate AI request.
*
* @since 2.0.0
*
* @param string $prompt_file Prompt template filename.
* @param array $cases Cases to analyze, keyed by case_id.
* @param Check_Context $check_context Check context instance.
* @param string $model_preference Optional model preference.
* @return array|WP_Error Array with 'cases' and 'token_usage' keys, or WP_Error.
*
* @SuppressWarnings(PHPMD.NPathComplexity)
*/
protected function analyze_batch( $prompt_file, array $cases, Check_Context $check_context, $model_preference = '' ) {
$issue_description = $this->load_prompt_template( $prompt_file );
if ( is_wp_error( $issue_description ) ) {
return $issue_description;
}
// Split into sub-batches if needed.
$batches = array_chunk( $cases, $this->get_ai_batch_size(), true );
$all_cases = array();
$total_tokens = 0;
$input_tokens = 0;
$output_tokens = 0;
$models_used = array();
$providers_used = array();
foreach ( $batches as $batch ) {
$result = $this->execute_batch_ai_request( $issue_description, $batch, $check_context, $model_preference );
if ( is_wp_error( $result ) ) {
continue;
}
if ( isset( $result['cases'] ) && is_array( $result['cases'] ) ) {
$all_cases = array_merge( $all_cases, $result['cases'] );
}
if ( isset( $result['token_usage']['total_tokens'] ) ) {
$total_tokens += (int) $result['token_usage']['total_tokens'];
}
if ( isset( $result['token_usage']['prompt_tokens'] ) ) {
$input_tokens += (int) $result['token_usage']['prompt_tokens'];
}
if ( isset( $result['token_usage']['completion_tokens'] ) ) {
$output_tokens += (int) $result['token_usage']['completion_tokens'];
}
if ( ! empty( $result['model_used'] ) ) {
$models_used[] = (string) $result['model_used'];
}
if ( ! empty( $result['provider_used'] ) ) {
$providers_used[] = (string) $result['provider_used'];
}
}
return array(
'cases' => $all_cases,
'token_usage' => array(
'prompt_tokens' => $input_tokens,
'completion_tokens' => $output_tokens,
'total_tokens' => $total_tokens,
),
'model_used' => implode( ', ', array_unique( $models_used ) ),
'provider_used' => implode( ', ', array_unique( $providers_used ) ),
);
}
/**
* Executes a single batched AI request for a group of cases.
*
* Builds a prompt following the internal scanner pattern:
* system instructions + issue description + cases list + output format.
*
* @since 2.0.0
*
* @param string $issue_description Issue description from prompt template.
* @param array $cases Cases to analyze, keyed by case_id.
* @param Check_Context $check_context Check context instance.
* @param string $model_preference Optional model preference.
* @return array|WP_Error Array with 'cases' and 'token_usage', or WP_Error.
*
* @SuppressWarnings(PHPMD.NPathComplexity)
*/
protected function execute_batch_ai_request( $issue_description, array $cases, Check_Context $check_context, $model_preference = '' ) {
$prompt = $this->build_batch_prompt( $issue_description, $cases, $check_context );
if ( ! function_exists( 'wp_ai_client_prompt' ) ) {
return new WP_Error(
'ai_client_not_available',
__( 'AI client is not available. This feature requires WordPress 7.0 or newer.', 'plugin-check' )
);
}
$builder = wp_ai_client_prompt( $prompt );
if ( is_wp_error( $builder ) ) {
return $builder;
}
// Apply model preference if provided.
if ( ! empty( $model_preference ) ) {
$builder = $this->apply_ai_model_preference( $builder, $model_preference );
if ( is_wp_error( $builder ) ) {
return $builder;
}
}
try {
// Try to generate a rich result first.
$result = null;
if ( is_callable( array( $builder, 'generate_text_result' ) ) ) {
$result = $builder->generate_text_result();
} elseif ( is_callable( array( $builder, 'generateTextResult' ) ) ) {
$result = $builder->generateTextResult();
}
if ( ! $result || is_wp_error( $result ) ) {
// Fallback to plain text generation.
$text = $builder->generate_text();
if ( is_wp_error( $text ) ) {
return $text;
}
return array(
'cases' => $this->parse_batch_response( (string) $text ),
'token_usage' => array(),
'model_used' => $this->normalize_ai_model_used( $model_preference ),
'provider_used' => $this->normalize_ai_provider_used( $model_preference ),
);
}
$text = method_exists( $result, 'to_text' ) ? $result->to_text() : ( method_exists( $result, 'toText' ) ? $result->toText() : '' );
$usage = $this->extract_ai_token_usage( $result );
$model = $this->extract_ai_model_used( $result );
$provider = $this->extract_ai_provider_used( $result );
return array(
'cases' => $this->parse_batch_response( $text ),
'token_usage' => $usage ? $usage : array(),
'model_used' => $model ? $model : $this->normalize_ai_model_used( $model_preference ),
'provider_used' => $provider ? $provider : $this->normalize_ai_provider_used( $model_preference ),
);
} catch ( \Throwable $e ) {
return new WP_Error(
'ai_request_failed',
sprintf(
/* translators: %s: Error message. */
__( 'AI analysis failed: %s', 'plugin-check' ),
$e->getMessage()
)
);
}
}
/**
* Builds the batched prompt following the internal scanner pattern.
*
* @since 2.0.0
*
* @param string $issue_description Issue description from prompt template.
* @param array $cases Cases to analyze, keyed by case_id.
* @param Check_Context $check_context Check context instance.
* @return string The complete prompt.
*/
protected function build_batch_prompt( $issue_description, array $cases, Check_Context $check_context ) {
$prompt = "You are an expert in WordPress security reviewing code for security, compatibility and performance.\n\n";
$prompt .= "You are given several cases to analyze. Each case references code in a WordPress plugin.\n";
$prompt .= "Do not trust on code comments to determine that something is not an issue.\n";
$prompt .= "Look up the code, understand the context and determine if there is specifically an issue with the following:\n\n";
$prompt .= $issue_description . "\n\n";
$prompt .= "## Cases\n\n";
foreach ( $cases as $case_id => $case ) {
$location = $case['file'] . ':' . $case['line'];
$code_context = $this->get_code_context_for_case( $case, $check_context );
$prompt .= '- Case ID ' . $case_id . ' : File and line "' . $location . '". ';
$prompt .= 'Issue message: "' . $case['message'] . '"';
if ( ! empty( $code_context ) ) {
$prompt .= "\n Code context:\n ```\n" . $code_context . "\n ```";
}
$prompt .= "\n\n";
}
$prompt .= "## Output\n\n";
$prompt .= "Respond ONLY with valid JSON matching this structure:\n";
$prompt .= "{\n";
$prompt .= ' "cases": [' . "\n";
$prompt .= " {\n";
$prompt .= ' "case_id": "the mentioned Case ID for each case",' . "\n";
$prompt .= ' "issue": true if there is a genuine issue (false if it is a false positive),' . "\n";
$prompt .= ' "short_explanation": "a very short explanation in one line"' . "\n";
$prompt .= " }\n";
$prompt .= " ]\n";
$prompt .= "}\n";
return $prompt;
}
/**
* Gets code context for a specific case.
*
* @since 2.0.0
*
* @param array $issue_case Case data with file, line, column.
* @param Check_Context $check_context Check context instance.
* @param int $context_lines Number of lines before and after.
* @return string Code context or empty string.
*/
protected function get_code_context_for_case( array $issue_case, Check_Context $check_context, $context_lines = 10 ) {
$file_path = $check_context->path( '/' ) . $issue_case['file'];
if ( ! file_exists( $file_path ) || ! is_readable( $file_path ) ) {
return '';
}
$file_content = file_get_contents( $file_path ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_get_contents_file_get_contents
if ( empty( $file_content ) ) {
return '';
}
return $this->get_code_context( $file_content, $issue_case['line'], $context_lines );
}
/**
* Gets code context around a specific line.
*
* @since 2.0.0
*
* @param string $file_content Full file content.
* @param int $line Line number (1-based).
* @param int $context Number of lines before and after.
* @return string Code context with line numbers.
*/
protected function get_code_context( $file_content, $line, $context = 10 ) {
if ( empty( $file_content ) ) {
return '';
}
$lines = explode( "\n", $file_content );
$start = max( 0, $line - $context - 1 );
$end = min( count( $lines ), $line + $context );
$context_lines = array();
for ( $i = $start; $i < $end; $i++ ) {
$line_num = $i + 1;
$marker = ( $line_num === (int) $line ) ? ' >>>' : ' ';
$context_lines[] = sprintf( '%s %4d | %s', $marker, $line_num, $lines[ $i ] );
}
return implode( "\n", $context_lines );
}
/**
* Parses the batched AI response into individual case results.
*
* @since 2.0.0
*
* @param string $response_text AI response text.
* @return array Array of case results.
*
* @SuppressWarnings(PHPMD.NPathComplexity)
*/
protected function parse_batch_response( $response_text ) {
if ( empty( $response_text ) ) {
return array();
}
// Remove markdown code fences if present.
$text = preg_replace( '/^```(?:json)?\s*\n?/m', '', $response_text );
$text = preg_replace( '/\n?```\s*$/m', '', $text );
$text = trim( $text );
// Try to find JSON object in the response.
$json_start = strpos( $text, '{' );
$json_end = strrpos( $text, '}' );
if ( false === $json_start || false === $json_end || $json_end <= $json_start ) {
return array();
}
$json_text = substr( $text, $json_start, $json_end - $json_start + 1 );
$decoded = json_decode( $json_text, true );
if ( JSON_ERROR_NONE !== json_last_error() || ! is_array( $decoded ) ) {
return array();
}
if ( ! isset( $decoded['cases'] ) || ! is_array( $decoded['cases'] ) ) {
return array();
}
$results = array();
foreach ( $decoded['cases'] as $case ) {
if ( ! isset( $case['case_id'] ) ) {
continue;
}
$results[] = array(
'case_id' => (string) $case['case_id'],
'issue' => isset( $case['issue'] ) ? (bool) $case['issue'] : true,
'short_explanation' => isset( $case['short_explanation'] ) ? (string) $case['short_explanation'] : '',
);
}
return $results;
}
/**
* Determines the prompt template filename for a given check code.
*
* @since 2.0.0
*
* @param string $code The check code (e.g., 'WordPress.Security.EscapeOutput.OutputNotEscaped').
* @return string Prompt template filename.
*/
protected function get_prompt_for_code( $code ) {
foreach ( $this->get_ai_prompt_map() as $prefix => $prompt_file ) {
if ( 0 === strpos( $code, $prefix ) ) {
return $prompt_file;
}
}
return 'ai-review-generic.md';
}
/**
* Loads a prompt template from the prompts/ directory.
*
* @since 2.0.0
*
* @param string $filename Prompt template filename.
* @return string|WP_Error Prompt content or WP_Error.
*/
protected function load_prompt_template( $filename ) {
if ( ! defined( 'WP_PLUGIN_CHECK_PLUGIN_DIR_PATH' ) ) {
return new WP_Error( 'plugin_constant_not_defined', __( 'Plugin constant not defined.', 'plugin-check' ) );
}
$path = WP_PLUGIN_CHECK_PLUGIN_DIR_PATH . 'prompts/' . $filename;
if ( ! file_exists( $path ) ) {
return new WP_Error(
'prompt_not_found',
sprintf(
/* translators: %s: Prompt filename. */
__( 'AI prompt template not found: %s', 'plugin-check' ),
$filename
)
);
}
$contents = (string) file_get_contents( $path ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_get_contents_file_get_contents
$contents = trim( $contents );
if ( empty( $contents ) ) {
return new WP_Error( 'prompt_empty', __( 'AI prompt template is empty.', 'plugin-check' ) );
}
return $contents;
}
/**
* Gets the AI severity threshold for a given type.
*
* @since 2.0.0
*
* @param string $type 'error' or 'warning'.
* @return int Severity threshold.
*/
protected function get_ai_severity_threshold( $type ) {
if ( class_exists( Settings_Page::class ) ) {
$default = 'error' === $type ? Settings_Page::get_severity_errors() : Settings_Page::get_severity_warnings();
} else {
$default = 'error' === $type ? 7 : 6;
}
/**
* Filters the AI severity threshold.
*
* @since 2.0.0
*
* @param int $threshold Threshold from settings (7 for errors, 6 for warnings).
* @param string $type 'error' or 'warning'.
*/
return (int) apply_filters( 'wp_plugin_check_ai_severity_threshold', $default, $type );
}
/**
* Applies a model preference to the prompt builder.
*
* @since 2.0.0
*
* @param object $builder Prompt builder instance.
* @param string $model_preference Model preference string.
* @return object|WP_Error Updated builder or WP_Error.
*/
protected function apply_ai_model_preference( $builder, $model_preference ) {
if ( empty( $model_preference ) ) {
return $builder;
}
$preference = trim( (string) $model_preference );
// Parse provider::model format.
foreach ( array( '::', '|', ':' ) as $separator ) {
if ( false !== strpos( $preference, $separator ) ) {
list( $provider, $model ) = array_map( 'trim', explode( $separator, $preference, 2 ) );
if ( '' !== $provider && '' !== $model ) {
$preference = array( $provider, $model );
break;
}
}
}
try {
$result = $builder->using_model_preference( $preference );
return $result ? $result : $builder;
} catch ( \Exception $e ) {
return new WP_Error(
'model_preference_error',
sprintf(
/* translators: %s: Exception message. */
__( 'Failed to apply model preference: %s', 'plugin-check' ),
$e->getMessage()
)
);
}
}
/**
* Extracts token usage from a result object.
*
* @since 2.0.0
*
* @param object $result Result object.
* @return array|null Token usage array or null.
*
* @SuppressWarnings(PHPMD.CyclomaticComplexity)
* @SuppressWarnings(PHPMD.NPathComplexity)
*/
protected function extract_ai_token_usage( $result ) {
$usage = null;
if ( method_exists( $result, 'get_token_usage' ) ) {
$usage = $result->get_token_usage();
} elseif ( method_exists( $result, 'getTokenUsage' ) ) {
$usage = $result->getTokenUsage();
}
if ( ! $usage || ! is_object( $usage ) ) {
return null;
}
$prompt_tokens = method_exists( $usage, 'get_prompt_tokens' ) ? $usage->get_prompt_tokens() : ( method_exists( $usage, 'getPromptTokens' ) ? $usage->getPromptTokens() : null );
$prompt_tokens = null === $prompt_tokens && method_exists( $usage, 'get_input_tokens' ) ? $usage->get_input_tokens() : $prompt_tokens;
$prompt_tokens = null === $prompt_tokens && method_exists( $usage, 'getInputTokens' ) ? $usage->getInputTokens() : $prompt_tokens;
$completion_tokens = method_exists( $usage, 'get_completion_tokens' ) ? $usage->get_completion_tokens() : ( method_exists( $usage, 'getCompletionTokens' ) ? $usage->getCompletionTokens() : null );
$completion_tokens = null === $completion_tokens && method_exists( $usage, 'get_output_tokens' ) ? $usage->get_output_tokens() : $completion_tokens;
$completion_tokens = null === $completion_tokens && method_exists( $usage, 'getOutputTokens' ) ? $usage->getOutputTokens() : $completion_tokens;
$total_tokens = method_exists( $usage, 'get_total_tokens' ) ? $usage->get_total_tokens() : ( method_exists( $usage, 'getTotalTokens' ) ? $usage->getTotalTokens() : null );
if ( null === $total_tokens && null !== $prompt_tokens && null !== $completion_tokens ) {
$total_tokens = $prompt_tokens + $completion_tokens;
}
if ( null === $prompt_tokens && null === $completion_tokens && null === $total_tokens ) {
return null;
}
return array_filter(
array(
'prompt_tokens' => $prompt_tokens,
'completion_tokens' => $completion_tokens,
'total_tokens' => $total_tokens,
),
static function ( $value ) {
return null !== $value;
}
);
}
/**
* Extracts the model used from an AI result object.
*
* @since 2.0.0
*
* @param object $result Result object.
* @return string Model identifier or empty string.
*/
protected function extract_ai_model_used( $result ) {
foreach ( array( 'get_model_metadata', 'getModelMetadata', 'get_model', 'getModel', 'get_model_id', 'getModelId', 'get_model_name', 'getModelName' ) as $method ) {
if ( ! method_exists( $result, $method ) ) {
continue;
}
$model = $result->$method();
if ( is_string( $model ) && '' !== trim( $model ) ) {
return trim( $model );
}
if ( is_object( $model ) ) {
foreach ( array( 'get_id', 'getId', 'get_name', 'getName' ) as $model_method ) {
if ( method_exists( $model, $model_method ) ) {
$value = $model->$model_method();
if ( is_string( $value ) && '' !== trim( $value ) ) {
return trim( $value );
}
}
}
}
}
return '';
}
/**
* Extracts the provider used from an AI result object.
*
* @since 2.0.0
*
* @param object $result Result object.
* @return string Provider identifier or empty string.
*/
protected function extract_ai_provider_used( $result ) {
foreach ( array( 'get_provider_metadata', 'getProviderMetadata', 'get_provider', 'getProvider', 'get_provider_id', 'getProviderId', 'get_provider_name', 'getProviderName' ) as $method ) {
if ( ! method_exists( $result, $method ) ) {
continue;
}
$provider = $result->$method();
if ( is_string( $provider ) && '' !== trim( $provider ) ) {
return trim( $provider );
}
if ( is_object( $provider ) ) {
foreach ( array( 'get_id', 'getId', 'get_name', 'getName' ) as $provider_method ) {
if ( method_exists( $provider, $provider_method ) ) {
$value = $provider->$provider_method();
if ( is_string( $value ) && '' !== trim( $value ) ) {
return trim( $value );
}
}
}
}
}
return '';
}
/**
* Normalizes a configured model preference for display.
*
* @since 2.0.0
*
* @param string $model_preference Model preference.
* @return string Model identifier or empty string.
*/
protected function normalize_ai_model_used( $model_preference ) {
$model_preference = trim( (string) $model_preference );
if ( '' === $model_preference ) {
return '';
}
foreach ( array( '::', '|', ':' ) as $separator ) {
if ( false !== strpos( $model_preference, $separator ) ) {
$parts = array_map( 'trim', explode( $separator, $model_preference, 2 ) );
return isset( $parts[1] ) && '' !== $parts[1] ? $parts[1] : $model_preference;
}
}
return $model_preference;
}
/**
* Normalizes a configured model preference provider for display.
*
* @since 2.0.0
*
* @param string $model_preference Model preference.
* @return string Provider identifier or empty string.
*/
protected function normalize_ai_provider_used( $model_preference ) {
$model_preference = trim( (string) $model_preference );
if ( '' === $model_preference ) {
return '';
}
foreach ( array( '::', '|', ':' ) as $separator ) {
if ( false !== strpos( $model_preference, $separator ) ) {
$parts = array_map( 'trim', explode( $separator, $model_preference, 2 ) );
return isset( $parts[0] ) && '' !== $parts[0] ? $parts[0] : '';
}
}
return '';
}
/**
* Returns an empty AI result structure.
*
* @since 2.0.0
*
* @return array Empty result with zeroed stats.
*/
protected function empty_ai_result() {
return array(
'analysis' => array(),
'stats' => array(
'tokens_spent' => 0,
'input_tokens' => 0,
'output_tokens' => 0,
'false_positives' => 0,
'issues_analyzed' => 0,
'model_used' => '',
'provider_used' => '',
),
);
}
/**
* Generates a unique key for an issue.
*
* @since 2.0.0
*
* @param string $file File path.
* @param int $line Line number.
* @param int $column Column number.
* @param string $code Issue code.
* @return string Unique key.
*/
protected function get_issue_key( $file, $line, $column, $code ) {
return md5( $file . ':' . $line . ':' . $column . ':' . $code );
}
}