<?php
/**
 * Bot user-agent patterns, matched case-insensitively as regex fragments,
 * mapped to WHY they crawl (Saddle_Analytics_Purpose — Phase 18's AI-crawler report).
 * Filter: saddle_analytics_bots. A filter may still append a plain numeric entry
 * ('mybot'); it drops as before and reports as purpose OTHER.
 * Cheapest, most specific first; the generic net last.
 */

defined( 'ABSPATH' ) || exit;
return array(
	// Training corpora.
	'GPTBot'              => Saddle_Analytics_Purpose::TRAINING,
	'ClaudeBot'           => Saddle_Analytics_Purpose::TRAINING,
	'CCBot'               => Saddle_Analytics_Purpose::TRAINING,
	'Google-Extended'     => Saddle_Analytics_Purpose::TRAINING,
	'Bytespider'          => Saddle_Analytics_Purpose::TRAINING,
	'Applebot-Extended'   => Saddle_Analytics_Purpose::TRAINING,
	'meta-externalagent'  => Saddle_Analytics_Purpose::TRAINING,
	// Live answer retrieval — a human asked a question about you just now.
	'ChatGPT-User'        => Saddle_Analytics_Purpose::AI_ANSWER,
	'Claude-User'         => Saddle_Analytics_Purpose::AI_ANSWER,
	'Perplexity-User'     => Saddle_Analytics_Purpose::AI_ANSWER,
	// Search/answer-engine indexing.
	'OAI-SearchBot'       => Saddle_Analytics_Purpose::INDEXING,
	'Claude-SearchBot'    => Saddle_Analytics_Purpose::INDEXING,
	'PerplexityBot'       => Saddle_Analytics_Purpose::INDEXING,
	'GoogleOther'         => Saddle_Analytics_Purpose::INDEXING,
	'Amazonbot'           => Saddle_Analytics_Purpose::INDEXING,
	// Previews, monitors, scripts — dropped, counted, not interesting.
	'facebookexternalhit' => Saddle_Analytics_Purpose::OTHER,
	// The generic net.
	'bot\b'               => Saddle_Analytics_Purpose::OTHER,
	'crawl'               => Saddle_Analytics_Purpose::OTHER,
	'spider'              => Saddle_Analytics_Purpose::OTHER,
	'slurp'               => Saddle_Analytics_Purpose::OTHER,
	'headless'            => Saddle_Analytics_Purpose::OTHER,
	'python-requests'     => Saddle_Analytics_Purpose::OTHER,
	'python-httpx'        => Saddle_Analytics_Purpose::OTHER,
	'wget/'               => Saddle_Analytics_Purpose::OTHER,
	'curl/'               => Saddle_Analytics_Purpose::OTHER,
	'go-http-client'      => Saddle_Analytics_Purpose::OTHER,
	'lighthouse'          => Saddle_Analytics_Purpose::OTHER,
	'pingdom'             => Saddle_Analytics_Purpose::OTHER,
	'uptimerobot'         => Saddle_Analytics_Purpose::OTHER,
);