<?php
/**
 * LLMO診断エンジン: URLを取得・解析し、AI検索に引用されやすさを100点満点で採点する。
 */

if ( ! defined( 'ABSPATH' ) ) {
	exit;
}

class ACSAIV_Analyzer {

	const FETCH_TIMEOUT   = 10;
	const MAX_BODY_BYTES  = 2097152; // 2MB
	const AI_CRAWLERS     = array( 'GPTBot', 'ClaudeBot', 'PerplexityBot', 'Google-Extended', 'CCBot' );

	/**
	 * 診断を実行する。
	 *
	 * @param string $url 診断対象URL。
	 * @return array|WP_Error 診断結果。
	 */
	public function analyze( $url ) {
		$url = $this->validate_url( $url );
		if ( is_wp_error( $url ) ) {
			return $url;
		}

		$response = $this->safe_fetch( $url );
		if ( is_wp_error( $response ) ) {
			return $response;
		}

		$html      = $response['body'];
		$final_url = $response['final_url'];
		$parts     = wp_parse_url( $final_url );
		$origin    = $parts['scheme'] . '://' . $parts['host'] . ( isset( $parts['port'] ) ? ':' . $parts['port'] : '' );

		$dom = $this->load_dom( $html );

		$checks = array();

		// --- A. AIアクセス基盤 (30) ---
		$checks[] = $this->check_llms_txt( $origin );
		$checks[] = $this->check_robots( $origin );
		$checks[] = $this->check_sitemap( $origin );
		$checks[] = $this->check_https( $final_url );

		// --- B. 構造化・引用されやすさ (40) ---
		$jsonld   = $this->extract_jsonld_types( $dom );
		$checks[] = $this->check_jsonld( $jsonld );
		$checks[] = $this->check_faq_howto( $jsonld );
		$checks[] = $this->check_title_meta( $dom );
		$checks[] = $this->check_headings( $dom );
		$checks[] = $this->check_ogp( $dom );

		// --- C. 信頼シグナル (30) ---
		$checks[] = $this->check_author( $dom, $jsonld );
		$checks[] = $this->check_dates( $dom, $html, $jsonld );
		$checks[] = $this->check_outbound_links( $dom, $parts['host'] );
		$checks[] = $this->check_rss( $dom, $origin );

		$score = 0;
		$max   = 0;
		$cats  = array();
		foreach ( $checks as $c ) {
			$score += $c['passed'] ? $c['weight'] : 0;
			$max   += $c['weight'];
			if ( ! isset( $cats[ $c['category'] ] ) ) {
				$cats[ $c['category'] ] = array( 'score' => 0, 'max' => 0 );
			}
			$cats[ $c['category'] ]['score'] += $c['passed'] ? $c['weight'] : 0;
			$cats[ $c['category'] ]['max']   += $c['weight'];
		}

		return array(
			'url'        => $final_url,
			'score'      => $score,
			'max'        => $max,
			'rank'       => $this->rank( $score ),
			'categories' => $cats,
			'checks'     => $checks,
			'checked_at' => current_time( 'mysql' ),
		);
	}

	/** スコアから評価ランクを返す。 */
	private function rank( $score ) {
		if ( $score >= 85 ) {
			return 'A';
		}
		if ( $score >= 65 ) {
			return 'B';
		}
		if ( $score >= 40 ) {
			return 'C';
		}
		return 'D';
	}

	/* ---------------------------------------------------------------------
	 * URL検証・安全な取得（SSRF対策）
	 * ------------------------------------------------------------------- */

	/**
	 * URLを検証する。http/https限定・プライベートIP宛を拒否。
	 */
	public function validate_url( $url ) {
		$url = trim( (string) $url );
		if ( '' === $url ) {
			return new WP_Error( 'acsaiv_empty', __( 'Please enter a URL.', 'acs-ai-citation-readiness' ) );
		}
		if ( ! preg_match( '#^https?://#i', $url ) ) {
			$url = 'https://' . $url;
		}
		$url = esc_url_raw( $url, array( 'http', 'https' ) );
		if ( '' === $url ) {
			return new WP_Error( 'acsaiv_invalid', __( 'This is not a valid URL.', 'acs-ai-citation-readiness' ) );
		}
		$host = wp_parse_url( $url, PHP_URL_HOST );
		if ( ! $host || is_wp_error( $this->assert_public_host( $host ) ) ) {
			return new WP_Error( 'acsaiv_blocked', __( 'This URL cannot be checked. Please enter the URL of a publicly accessible website.', 'acs-ai-citation-readiness' ) );
		}
		return $url;
	}

	/**
	 * ホストがグローバルIPに解決されることを確認する。
	 */
	private function assert_public_host( $host ) {
		// IPリテラルの場合はそのまま判定。
		$ip = $host;
		if ( ! filter_var( $host, FILTER_VALIDATE_IP ) ) {
			$resolved = gethostbyname( $host );
			if ( $resolved === $host ) {
				return new WP_Error( 'acsaiv_dns', 'DNS解決に失敗' );
			}
			$ip = $resolved;
		}
		if ( ! filter_var( $ip, FILTER_VALIDATE_IP, FILTER_FLAG_NO_PRIV_RANGE | FILTER_FLAG_NO_RES_RANGE ) ) {
			return new WP_Error( 'acsaiv_private', 'プライベートIP宛は拒否' );
		}
		return true;
	}

	/**
	 * サイズ・タイムアウト・リダイレクトを制限してURLを取得する。
	 */
	private function safe_fetch( $url, $head_only = false ) {
		$args = array(
			'timeout'             => self::FETCH_TIMEOUT,
			'redirection'         => 3,
			'limit_response_size' => self::MAX_BODY_BYTES,
			'user-agent'          => 'Mozilla/5.0 (compatible; ACS-AI-Visibility/1.0; +https://acs-developer.com/)',
		);

		$response = $head_only ? wp_remote_head( $url, $args ) : wp_remote_get( $url, $args );
		if ( is_wp_error( $response ) ) {
			return new WP_Error( 'acsaiv_fetch', __( 'Could not fetch the site. Please check the URL.', 'acs-ai-citation-readiness' ) );
		}
		$code = wp_remote_retrieve_response_code( $response );
		if ( $code < 200 || $code >= 400 ) {
			/* translators: %d: HTTP status code. */
			return new WP_Error( 'acsaiv_http', sprintf( __( 'The site returned an error (HTTP %d).', 'acs-ai-citation-readiness' ), $code ) );
		}
		return array(
			'body'      => wp_remote_retrieve_body( $response ),
			'code'      => $code,
			'final_url' => $url,
		);
	}

	/** サブリソース（robots.txt等）の存在確認。200かつ本文ありでtrue。 */
	private function resource_exists( $url, &$body = null ) {
		$res = $this->safe_fetch( $url );
		if ( is_wp_error( $res ) ) {
			return false;
		}
		$body = $res['body'];
		return '' !== trim( (string) $body );
	}

	private function load_dom( $html ) {
		$dom = new DOMDocument();
		libxml_use_internal_errors( true );
		$dom->loadHTML( '<?xml encoding="UTF-8">' . $html );
		libxml_clear_errors();
		return $dom;
	}

	/* ---------------------------------------------------------------------
	 * チェック項目
	 * ------------------------------------------------------------------- */

	private function result( $category, $label, $weight, $passed, $advice_ok, $advice_ng ) {
		return array(
			'category' => $category,
			'label'    => $label,
			'weight'   => $weight,
			'passed'   => (bool) $passed,
			'advice'   => $passed ? $advice_ok : $advice_ng,
		);
	}

	private function check_llms_txt( $origin ) {
		$body   = null;
		$exists = $this->resource_exists( $origin . '/llms.txt', $body );
		// HTMLが返ってくる404ページ対策: llms.txtはMarkdown/プレーンテキストのはず。
		if ( $exists && false !== stripos( substr( (string) $body, 0, 500 ), '<html' ) ) {
			$exists = false;
		}
		return $this->result(
			'A',
			'llms.txt',
			12,
			$exists,
			__( 'llms.txt is in place. AI can easily understand your site structure.', 'acs-ai-citation-readiness' ),
			__( 'llms.txt is missing. Add an entry-point file that tells AI what your site is about (this plugin can generate it automatically).', 'acs-ai-citation-readiness' )
		);
	}

	private function check_robots( $origin ) {
		$body    = null;
		$blocked = array();
		if ( $this->resource_exists( $origin . '/robots.txt', $body ) ) {
			$blocked = $this->blocked_ai_crawlers( (string) $body );
		}
		$passed = empty( $blocked );
		return $this->result(
			'A',
			__( 'AI crawler access (robots.txt)', 'acs-ai-citation-readiness' ),
			10,
			$passed,
			__( 'Major AI crawlers are not blocked.', 'acs-ai-citation-readiness' ),
			/* translators: %s: comma-separated list of blocked AI crawler names. */
			sprintf( __( 'robots.txt blocks %s. Consider allowing them if you want to be cited.', 'acs-ai-citation-readiness' ), implode( ', ', $blocked ) )
		);
	}

	/**
	 * robots.txtを解析し、サイト全体をDisallowされているAIクローラー名の配列を返す。
	 */
	public function blocked_ai_crawlers( $robots_txt ) {
		$blocked        = array();
		$current_agents = array();
		$rules_started  = false;

		foreach ( preg_split( '/\r\n|\r|\n/', $robots_txt ) as $line ) {
			$line = trim( preg_replace( '/#.*$/', '', $line ) );
			if ( '' === $line || false === strpos( $line, ':' ) ) {
				continue;
			}
			list( $field, $value ) = array_map( 'trim', explode( ':', $line, 2 ) );
			$field = strtolower( $field );

			if ( 'user-agent' === $field ) {
				if ( $rules_started ) {
					$current_agents = array();
					$rules_started  = false;
				}
				$current_agents[] = $value;
			} elseif ( 'disallow' === $field ) {
				$rules_started = true;
				if ( '/' === $value ) {
					foreach ( self::AI_CRAWLERS as $bot ) {
						foreach ( $current_agents as $agent ) {
							if ( 0 === strcasecmp( $agent, $bot ) ) {
								$blocked[] = $bot;
							}
						}
					}
				}
			} else {
				$rules_started = true;
			}
		}
		return array_values( array_unique( $blocked ) );
	}

	private function check_sitemap( $origin ) {
		$exists = $this->resource_exists( $origin . '/sitemap.xml' )
			|| $this->resource_exists( $origin . '/sitemap_index.xml' )
			|| $this->resource_exists( $origin . '/wp-sitemap.xml' );
		return $this->result(
			'A',
			'sitemap.xml',
			4,
			$exists,
			__( 'Sitemap found.', 'acs-ai-citation-readiness' ),
			__( 'No sitemap found. Crawlers may index your site less efficiently.', 'acs-ai-citation-readiness' )
		);
	}

	private function check_https( $url ) {
		$passed = 0 === stripos( $url, 'https://' );
		return $this->result(
			'A',
			'HTTPS',
			4,
			$passed,
			__( 'Served over HTTPS.', 'acs-ai-citation-readiness' ),
			__( 'Not served over HTTPS. Always-on SSL is strongly recommended because it directly affects trust evaluation.', 'acs-ai-citation-readiness' )
		);
	}

	/** ページ内のJSON-LD @type を全て集める。 */
	private function extract_jsonld_types( DOMDocument $dom ) {
		$types = array();
		foreach ( $dom->getElementsByTagName( 'script' ) as $script ) {
			if ( 'application/ld+json' !== strtolower( (string) $script->getAttribute( 'type' ) ) ) {
				continue;
			}
			$data = json_decode( $script->textContent, true );
			if ( null === $data ) {
				continue;
			}
			$stack = array( $data );
			while ( $stack ) {
				$node = array_pop( $stack );
				if ( ! is_array( $node ) ) {
					continue;
				}
				if ( isset( $node['@type'] ) ) {
					foreach ( (array) $node['@type'] as $t ) {
						if ( is_string( $t ) ) {
							$types[] = $t;
						}
					}
				}
				foreach ( $node as $v ) {
					if ( is_array( $v ) ) {
						$stack[] = $v;
					}
				}
			}
		}
		return array_values( array_unique( $types ) );
	}

	private function check_jsonld( array $types ) {
		return $this->result(
			'B',
			__( 'Structured data (JSON-LD)', 'acs-ai-citation-readiness' ),
			10,
			! empty( $types ),
			/* translators: %s: comma-separated list of schema.org types found. */
			sprintf( __( 'Structured data found (%s).', 'acs-ai-citation-readiness' ), implode( ', ', array_slice( $types, 0, 5 ) ) ),
			__( 'No structured data found. Provide machine-readable markup so AI can understand what your pages mean.', 'acs-ai-citation-readiness' )
		);
	}

	private function check_faq_howto( array $types ) {
		$passed = (bool) array_intersect( array( 'FAQPage', 'HowTo', 'Question' ), $types );
		return $this->result(
			'B',
			__( 'FAQ / HowTo schema', 'acs-ai-citation-readiness' ),
			8,
			$passed,
			__( 'FAQ/HowTo schema found. Question-and-answer formats are easy for AI to cite in answers.', 'acs-ai-citation-readiness' ),
			__( 'No FAQ/HowTo schema found. Structuring frequent questions as Q&A makes them easier to cite (insert them with the FAQ shortcode in this plugin).', 'acs-ai-citation-readiness' )
		);
	}

	private function check_title_meta( DOMDocument $dom ) {
		$title = '';
		$nodes = $dom->getElementsByTagName( 'title' );
		if ( $nodes->length > 0 ) {
			$title = trim( $nodes->item( 0 )->textContent );
		}
		$desc = $this->get_meta( $dom, 'description' );

		$title_len = mb_strlen( $title );
		$desc_len  = mb_strlen( $desc );
		$passed    = $title_len >= 10 && $title_len <= 70 && $desc_len >= 50 && $desc_len <= 200;
		return $this->result(
			'B',
			__( 'title / meta description', 'acs-ai-citation-readiness' ),
			8,
			$passed,
			__( 'The title and meta description have appropriate lengths.', 'acs-ai-citation-readiness' ),
			__( 'The title or meta description is missing or unsuitable. AI uses summaries as answer material, so write a description that conveys the conclusion of the page.', 'acs-ai-citation-readiness' )
		);
	}

	private function check_headings( DOMDocument $dom ) {
		$h1 = $dom->getElementsByTagName( 'h1' )->length;
		$h2 = $dom->getElementsByTagName( 'h2' )->length;
		$passed = ( 1 === $h1 && $h2 >= 2 );
		return $this->result(
			'B',
			__( 'Heading structure', 'acs-ai-citation-readiness' ),
			8,
			$passed,
			__( 'One h1 and multiple h2 headings: AI can follow the structure easily.', 'acs-ai-citation-readiness' ),
			/* translators: 1: number of h1 headings, 2: number of h2 headings. */
			sprintf( __( 'Heading structure can be improved (h1: %1$d / h2: %2$d). Use exactly one h1 and divide the body with h2 headings.', 'acs-ai-citation-readiness' ), $h1, $h2 )
		);
	}

	private function check_ogp( DOMDocument $dom ) {
		$og_title = $this->get_meta( $dom, 'og:title', 'property' );
		$og_desc  = $this->get_meta( $dom, 'og:description', 'property' );
		$passed   = '' !== $og_title && '' !== $og_desc;
		return $this->result(
			'B',
			'OGP',
			6,
			$passed,
			__( 'OGP tags are set.', 'acs-ai-citation-readiness' ),
			__( 'OGP tags (og:title / og:description) are missing. This hurts previews on social media and in AI interfaces.', 'acs-ai-citation-readiness' )
		);
	}

	private function check_author( DOMDocument $dom, array $types ) {
		$passed = (bool) array_intersect( array( 'Person', 'Organization' ), $types );
		if ( ! $passed ) {
			// 運営者情報系ページへのリンクで代替判定。
			foreach ( $dom->getElementsByTagName( 'a' ) as $a ) {
				$href = strtolower( (string) $a->getAttribute( 'href' ) );
				$text = (string) $a->textContent;
				if ( preg_match( '#(about|company|profile|author)#', $href )
					|| false !== mb_strpos( $text, '運営者' )
					|| false !== mb_strpos( $text, '会社概要' ) ) {
					$passed = true;
					break;
				}
			}
		}
		return $this->result(
			'C',
			__( 'Author / publisher info', 'acs-ai-citation-readiness' ),
			10,
			$passed,
			__( 'Author or publisher information is present.', 'acs-ai-citation-readiness' ),
			__( 'No author or publisher information found. Knowing who wrote the content is the core of trust evaluation (E-E-A-T) in the AI era. Add an about page and Person/Organization schema.', 'acs-ai-citation-readiness' )
		);
	}

	private function check_dates( DOMDocument $dom, $html, array $types ) {
		$passed = $dom->getElementsByTagName( 'time' )->length > 0
			|| false !== strpos( $html, 'datePublished' )
			|| false !== strpos( $html, 'dateModified' );
		return $this->result(
			'C',
			__( 'Date information', 'acs-ai-citation-readiness' ),
			10,
			$passed,
			__( 'Published and modified dates are present.', 'acs-ai-citation-readiness' ),
			__( 'No date information found. AI values freshness, so expose published and modified dates in machine-readable form.', 'acs-ai-citation-readiness' )
		);
	}

	private function check_outbound_links( DOMDocument $dom, $host ) {
		$count = 0;
		foreach ( $dom->getElementsByTagName( 'a' ) as $a ) {
			$href = (string) $a->getAttribute( 'href' );
			$h    = wp_parse_url( $href, PHP_URL_HOST );
			if ( $h && 0 !== strcasecmp( $h, $host ) ) {
				$count++;
			}
		}
		$passed = $count >= 1;
		return $this->result(
			'C',
			__( 'Citations / outbound links', 'acs-ai-citation-readiness' ),
			6,
			$passed,
			__( 'Outbound reference links are present. Citing sources is a trust signal.', 'acs-ai-citation-readiness' ),
			__( 'No outbound source links found. Linking to primary sources that back your claims raises trust with AI.', 'acs-ai-citation-readiness' )
		);
	}

	private function check_rss( DOMDocument $dom, $origin ) {
		$passed = false;
		foreach ( $dom->getElementsByTagName( 'link' ) as $link ) {
			if ( false !== stripos( (string) $link->getAttribute( 'type' ), 'rss' ) ) {
				$passed = true;
				break;
			}
		}
		if ( ! $passed ) {
			$passed = $this->resource_exists( $origin . '/feed/' );
		}
		return $this->result(
			'C',
			__( 'RSS feed', 'acs-ai-citation-readiness' ),
			4,
			$passed,
			__( 'RSS feed found. Updates propagate easily to AI and crawlers.', 'acs-ai-citation-readiness' ),
			__( 'No RSS feed found. A machine-readable update feed encourages more frequent crawling.', 'acs-ai-citation-readiness' )
		);
	}

	/** meta要素の値を取得する。 */
	private function get_meta( DOMDocument $dom, $name, $attr = 'name' ) {
		foreach ( $dom->getElementsByTagName( 'meta' ) as $meta ) {
			if ( 0 === strcasecmp( (string) $meta->getAttribute( $attr ), $name ) ) {
				return trim( (string) $meta->getAttribute( 'content' ) );
			}
		}
		return '';
	}
}