<?php
/**
 * Microsoft AI visibility evidence from official Bing/Clarity exports.
 *
 * Admin-only. This deliberately does not scrape Microsoft dashboards and does
 * not mix citation evidence with crawler hits, referrals or search-demand rows.
 */

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

final class AI_O_Optimiser_Microsoft_AI_Visibility {
	const OPTION         = 'ai_o_optimiser_microsoft_ai_visibility_v1';
	const PAGE_SLUG      = 'ai-o-optimiser-microsoft-ai';
	const MAX_FILE_BYTES = 2097152;
	const MAX_ROWS       = 5000;
	const MAX_ITEMS      = 100;

	/** Register admin-only routes. */
	public static function boot() {
		add_action( 'admin_menu', array( __CLASS__, 'register_page' ) );
		add_action( 'admin_post_ai_o_optimiser_microsoft_ai_import', array( __CLASS__, 'handle_import' ) );
		add_action( 'admin_post_ai_o_optimiser_microsoft_ai_clear', array( __CLASS__, 'handle_clear' ) );
	}

	/** Dedicated administrator capability, with a safe fallback for tests/older installs. */
	private static function capability() {
		return defined( 'AI_O_Optimiser_SENSITIVE_CAPABILITY' ) ? AI_O_Optimiser_SENSITIVE_CAPABILITY : 'manage_options';
	}

	/** Register a visibility sub-page. */
	public static function register_page() {
		add_submenu_page(
			'ai-o-optimiser-dashboard',
			__( 'Microsoft AI visibility', 'ai-o-optimiser' ),
			__( 'Microsoft AI', 'ai-o-optimiser' ),
			self::capability(),
			self::PAGE_SLUG,
			array( __CLASS__, 'render_page' ),
			5
		);
	}

	/** Current imported evidence snapshot. */
	public static function snapshot() {
		$value = get_option( self::OPTION, array() );
		return is_array( $value ) ? $value : array();
	}

	/** Capability state for future Connector Hub/API adapters. */
	public static function capability_state() {
		$snapshot = self::snapshot();
		return array(
			'capability'      => 'platform_ai_visibility.microsoft',
			'schema_version'  => 'aio-evidence/1',
			'source'          => 'microsoft_ai_performance_export',
			'state'           => empty( $snapshot['imported_at'] ) ? 'not_imported' : 'available',
			'retrieved_at'    => isset( $snapshot['imported_at'] ) ? (string) $snapshot['imported_at'] : '',
			'evidence_family' => 'platform_ai_visibility',
		);
	}

	/** Normalise a CSV header into a comparison key. */
	private static function header_key( $value ) {
		$value = strtolower( trim( (string) $value ) );
		return preg_replace( '/[^a-z0-9]+/', '', $value );
	}

	/** First matching column index for any known alias. */
	private static function column_index( $headers, $aliases ) {
		$aliases = array_map( array( __CLASS__, 'header_key' ), $aliases );
		foreach ( $headers as $index => $header ) {
			if ( in_array( self::header_key( $header ), $aliases, true ) ) {
				return (int) $index;
			}
		}
		return -1;
	}

	/** Parse Microsoft percentage/number presentation into a float. */
	private static function numeric_value( $value ) {
		$value = str_replace( array( ',', '%', ' ' ), '', (string) $value );
		return is_numeric( $value ) ? (float) $value : 0.0;
	}

	/** Canonical host comparison for cited-page ownership. */
	private static function canonical_host( $host ) {
		$host = strtolower( trim( (string) $host ) );
		return 0 === strpos( $host, 'www.' ) ? substr( $host, 4 ) : $host;
	}

	/** Validate one cited-page URL against this WordPress site's host. */
	private static function cited_page_url( $value, $site_host ) {
		$url = esc_url_raw( trim( (string) $value ) );
		if ( '' === $url ) {
			return '';
		}
		$scheme = strtolower( (string) wp_parse_url( $url, PHP_URL_SCHEME ) );
		$host   = self::canonical_host( wp_parse_url( $url, PHP_URL_HOST ) );
		if ( ! in_array( $scheme, array( 'http', 'https' ), true ) || '' === $host ) {
			return '';
		}
		return self::canonical_host( $site_host ) === $host ? $url : '';
	}

	/**
	 * Parse one official Microsoft CSV export.
	 *
	 * Supported shapes are Overview, Grounding queries and Cited pages. The
	 * parser is intentionally header-driven because Microsoft can reorder export
	 * columns without changing their meaning.
	 */
	public static function parse_csv_path( $path, $site_host = '' ) {
		$result = array(
			'type'          => '',
			'summary'       => array(),
			'queries'       => array(),
			'pages'         => array(),
			'rows_read'     => 0,
			'rows_rejected' => 0,
		);

		$handle = @fopen( $path, 'rb' ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_fopen -- bounded uploaded CSV, not WordPress filesystem content.
		if ( ! is_resource( $handle ) ) {
			return new WP_Error( 'microsoft_ai_open_failed', __( 'The Microsoft export could not be opened.', 'ai-o-optimiser' ) );
		}

		$headers = fgetcsv( $handle );
		if ( ! is_array( $headers ) || empty( $headers ) ) {
			fclose( $handle ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_fclose -- closes the bounded uploaded CSV handle opened directly above.
			return new WP_Error( 'microsoft_ai_empty_csv', __( 'The Microsoft export has no readable header row.', 'ai-o-optimiser' ) );
		}
		if ( isset( $headers[0] ) ) {
			$headers[0] = preg_replace( '/^\xEF\xBB\xBF/', '', (string) $headers[0] );
		}

		$query_col = self::column_index( $headers, array( 'Query', 'Grounding query', 'Grounding queries', 'Search query' ) );
		$page_col  = self::column_index( $headers, array( 'Page URL', 'Cited page', 'URL', 'Page' ) );
		$cite_col  = self::column_index( $headers, array( 'Citations', 'Citation count', 'Total citations' ) );
		$share_col = self::column_index( $headers, array( 'Share of authority', 'SoA', 'Citation share', 'Citation share %', 'Share' ) );
		$traffic_col = self::column_index( $headers, array( 'AI referral traffic', 'AI referrals', 'Referral traffic', 'AI-referred sessions' ) );
		$type_col  = self::column_index( $headers, array( 'Query type', 'Branded/non-branded', 'Brand type' ) );
		$intent_col = self::column_index( $headers, array( 'Intent' ) );
		$topic_col  = self::column_index( $headers, array( 'Topic', 'Topics' ) );

		if ( $query_col >= 0 ) {
			$result['type'] = 'queries';
		} elseif ( $page_col >= 0 ) {
			$result['type'] = 'pages';
		} elseif ( $cite_col >= 0 || $share_col >= 0 || $traffic_col >= 0 ) {
			$result['type'] = 'summary';
		} else {
			fclose( $handle ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_fclose -- closes the bounded uploaded CSV handle opened directly above.
			return new WP_Error( 'microsoft_ai_unknown_columns', __( 'This CSV does not look like a supported Microsoft AI Performance or Clarity citation export.', 'ai-o-optimiser' ) );
		}

		$site_host = $site_host ?: (string) wp_parse_url( home_url( '/' ), PHP_URL_HOST );
		$sum_citations = 0;
		while ( $result['rows_read'] < self::MAX_ROWS && false !== ( $row = fgetcsv( $handle ) ) ) {
			if ( ! is_array( $row ) ) {
				continue;
			}
			$result['rows_read']++;
			$citations = $cite_col >= 0 && isset( $row[ $cite_col ] ) ? (int) round( self::numeric_value( $row[ $cite_col ] ) ) : 0;
			$share     = $share_col >= 0 && isset( $row[ $share_col ] ) ? self::numeric_value( $row[ $share_col ] ) : 0.0;

			if ( 'queries' === $result['type'] ) {
				$query = isset( $row[ $query_col ] ) ? sanitize_text_field( (string) $row[ $query_col ] ) : '';
				if ( '' === $query ) {
					$result['rows_rejected']++;
					continue;
				}
				$sum_citations += $citations;
				if ( count( $result['queries'] ) < self::MAX_ITEMS ) {
					$result['queries'][] = array(
						'query'      => substr( $query, 0, 300 ),
						'citations'  => $citations,
						'share'      => $share,
						'query_type' => $type_col >= 0 && isset( $row[ $type_col ] ) ? sanitize_text_field( (string) $row[ $type_col ] ) : '',
						'intent'     => $intent_col >= 0 && isset( $row[ $intent_col ] ) ? sanitize_text_field( (string) $row[ $intent_col ] ) : '',
						'topic'      => $topic_col >= 0 && isset( $row[ $topic_col ] ) ? sanitize_text_field( (string) $row[ $topic_col ] ) : '',
					);
				}
				continue;
			}

			if ( 'pages' === $result['type'] ) {
				$url = isset( $row[ $page_col ] ) ? self::cited_page_url( $row[ $page_col ], $site_host ) : '';
				if ( '' === $url ) {
					$result['rows_rejected']++;
					continue;
				}
				$sum_citations += $citations;
				if ( count( $result['pages'] ) < self::MAX_ITEMS ) {
					$result['pages'][] = array(
						'url'       => $url,
						'citations' => $citations,
						'share'     => $share,
					);
				}
				continue;
			}

			if ( 'summary' === $result['type'] ) {
				if ( $cite_col >= 0 && isset( $row[ $cite_col ] ) ) {
					$result['summary']['citations'] = $citations;
				}
				if ( $share_col >= 0 && isset( $row[ $share_col ] ) ) {
					$result['summary']['share_of_authority'] = $share;
				}
				if ( $traffic_col >= 0 && isset( $row[ $traffic_col ] ) ) {
					$result['summary']['ai_referral_traffic'] = (int) round( self::numeric_value( $row[ $traffic_col ] ) );
				}
				break;
			}
		}
		fclose( $handle ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_fclose -- closes the bounded uploaded CSV handle opened directly above.

		if ( 'queries' === $result['type'] ) {
			usort( $result['queries'], static function ( $a, $b ) { return (int) $b['citations'] <=> (int) $a['citations']; } );
			$result['summary']['citations_from_query_export'] = $sum_citations;
		}
		if ( 'pages' === $result['type'] ) {
			usort( $result['pages'], static function ( $a, $b ) { return (int) $b['citations'] <=> (int) $a['citations']; } );
			$result['summary']['citations_from_page_export'] = $sum_citations;
		}
		return $result;
	}

	/** Import handler for one official CSV export. */
	public static function handle_import() {
		if ( ! current_user_can( self::capability() ) ) {
			wp_die( esc_html__( 'You are not allowed to import Microsoft AI visibility evidence.', 'ai-o-optimiser' ) );
		}
		check_admin_referer( 'ai_o_optimiser_microsoft_ai_import' );

		$file = isset( $_FILES['microsoft_ai_csv'] ) && is_array( $_FILES['microsoft_ai_csv'] ) ? $_FILES['microsoft_ai_csv'] : array(); // phpcs:ignore WordPress.Security.ValidatedSanitizedInput.InputNotSanitized -- file metadata validated below; contents parsed as CSV.
		if ( empty( $file['tmp_name'] ) || UPLOAD_ERR_OK !== (int) ( $file['error'] ?? UPLOAD_ERR_NO_FILE ) ) {
			self::redirect_with_state( 'upload_error' );
		}
		$size = isset( $file['size'] ) ? (int) $file['size'] : 0;
		$name = isset( $file['name'] ) ? sanitize_file_name( (string) $file['name'] ) : '';
		if ( $size < 1 || $size > self::MAX_FILE_BYTES || 'csv' !== strtolower( pathinfo( $name, PATHINFO_EXTENSION ) ) ) {
			self::redirect_with_state( 'invalid_file' );
		}

		$parsed = self::parse_csv_path( (string) $file['tmp_name'] );
		if ( is_wp_error( $parsed ) ) {
			self::redirect_with_state( $parsed->get_error_code() );
		}

		$period_start = isset( $_POST['period_start'] ) ? sanitize_text_field( wp_unslash( $_POST['period_start'] ) ) : '';
		$period_end   = isset( $_POST['period_end'] ) ? sanitize_text_field( wp_unslash( $_POST['period_end'] ) ) : '';
		if ( ! self::valid_date( $period_start ) || ! self::valid_date( $period_end ) || $period_start > $period_end ) {
			self::redirect_with_state( 'invalid_period' );
		}

		$current = self::snapshot();
		$same_period = isset( $current['period']['start'], $current['period']['end'] )
			&& $period_start === $current['period']['start']
			&& $period_end === $current['period']['end'];
		$snapshot = $same_period ? $current : array();
		$snapshot['schema_version']  = 'aio-evidence/1';
		$snapshot['source']          = 'microsoft_ai_performance_export';
		$snapshot['evidence_family'] = 'platform_ai_visibility';
		$snapshot['period']          = array( 'start' => $period_start, 'end' => $period_end );
		$snapshot['imported_at']     = gmdate( 'c' );
		$snapshot['summary']         = isset( $snapshot['summary'] ) && is_array( $snapshot['summary'] ) ? $snapshot['summary'] : array();
		$snapshot['datasets']        = isset( $snapshot['datasets'] ) && is_array( $snapshot['datasets'] ) ? $snapshot['datasets'] : array();

		if ( ! empty( $parsed['summary'] ) ) {
			$snapshot['summary'] = array_merge( $snapshot['summary'], $parsed['summary'] );
		}
		if ( 'queries' === $parsed['type'] ) {
			$snapshot['queries'] = $parsed['queries'];
		}
		if ( 'pages' === $parsed['type'] ) {
			$snapshot['pages'] = $parsed['pages'];
		}
		$snapshot['datasets'][ $parsed['type'] ] = array(
			'imported_at'    => $snapshot['imported_at'],
			'rows_read'      => (int) $parsed['rows_read'],
			'rows_rejected'  => (int) $parsed['rows_rejected'],
			'original_name'  => $name,
		);

		if ( false === get_option( self::OPTION, false ) ) {
			add_option( self::OPTION, $snapshot, '', 'no' );
		} else {
			update_option( self::OPTION, $snapshot, false );
		}
		self::redirect_with_state( 'imported' );
	}

	/** Clear imported evidence without touching any other visibility data. */
	public static function handle_clear() {
		if ( ! current_user_can( self::capability() ) ) {
			wp_die( esc_html__( 'You are not allowed to clear Microsoft AI visibility evidence.', 'ai-o-optimiser' ) );
		}
		check_admin_referer( 'ai_o_optimiser_microsoft_ai_clear' );
		delete_option( self::OPTION );
		self::redirect_with_state( 'cleared' );
	}

	private static function valid_date( $value ) {
		if ( ! preg_match( '/^\d{4}-\d{2}-\d{2}$/', (string) $value ) ) {
			return false;
		}
		$date = DateTime::createFromFormat( '!Y-m-d', (string) $value );
		return $date && $date->format( 'Y-m-d' ) === $value;
	}

	private static function redirect_with_state( $state ) {
		wp_safe_redirect( add_query_arg( array( 'page' => self::PAGE_SLUG, 'microsoft_ai' => sanitize_key( $state ) ), admin_url( 'admin.php' ) ) );
		exit;
	}

	/** Render imported Microsoft AI citation evidence and the import control. */
	public static function render_page() {
		if ( ! current_user_can( self::capability() ) ) {
			return;
		}
		$snapshot = self::snapshot();
		$summary  = isset( $snapshot['summary'] ) && is_array( $snapshot['summary'] ) ? $snapshot['summary'] : array();
		$queries  = isset( $snapshot['queries'] ) && is_array( $snapshot['queries'] ) ? $snapshot['queries'] : array();
		$pages    = isset( $snapshot['pages'] ) && is_array( $snapshot['pages'] ) ? $snapshot['pages'] : array();
		$status   = isset( $_GET['microsoft_ai'] ) ? sanitize_key( wp_unslash( $_GET['microsoft_ai'] ) ) : ''; // phpcs:ignore WordPress.Security.NonceVerification.Recommended -- read-only redirect result state.
		?>
		<div class="wrap ai-o-optimiser-wrap">
			<h1><?php esc_html_e( 'Microsoft AI visibility', 'ai-o-optimiser' ); ?></h1>
			<p class="description"><?php esc_html_e( 'Import official Microsoft AI Performance / Clarity citation exports. These are platform citation-visibility signals, not crawler hits, human visits, prompts, rankings or audience size.', 'ai-o-optimiser' ); ?></p>
			<?php if ( 'imported' === $status ) : ?><div class="notice notice-success inline"><p><?php esc_html_e( 'Microsoft AI visibility evidence imported.', 'ai-o-optimiser' ); ?></p></div><?php endif; ?>
			<?php if ( 'cleared' === $status ) : ?><div class="notice notice-success inline"><p><?php esc_html_e( 'Imported Microsoft AI visibility evidence cleared.', 'ai-o-optimiser' ); ?></p></div><?php endif; ?>
			<?php if ( $status && ! in_array( $status, array( 'imported', 'cleared' ), true ) ) : ?><div class="notice notice-error inline"><p><?php esc_html_e( 'The Microsoft export could not be imported. Check that it is a supported CSV export and that the selected period is valid.', 'ai-o-optimiser' ); ?></p></div><?php endif; ?>

			<?php if ( ! empty( $snapshot['imported_at'] ) ) : ?>
				<section class="ai-o-optimiser-panel">
					<h2><?php esc_html_e( 'Platform AI visibility', 'ai-o-optimiser' ); ?></h2>
					<p class="description"><?php echo esc_html( sprintf( /* translators: 1: period start, 2: period end, 3: import timestamp. */ __( 'Microsoft export period %1$s to %2$s · imported %3$s. Not observed means only that this import contains no matching evidence for the period.', 'ai-o-optimiser' ), $snapshot['period']['start'] ?? '', $snapshot['period']['end'] ?? '', $snapshot['imported_at'] ) ); ?></p>
					<div class="ai-o-optimiser-card-grid">
						<div class="ai-o-optimiser-card"><strong><?php esc_html_e( 'Citations', 'ai-o-optimiser' ); ?></strong><div class="ai-o-optimiser-card-value"><?php echo esc_html( isset( $summary['citations'] ) ? number_format_i18n( (int) $summary['citations'] ) : '—' ); ?></div><p><?php esc_html_e( 'Citation visibility reported by Microsoft.', 'ai-o-optimiser' ); ?></p></div>
						<div class="ai-o-optimiser-card"><strong><?php esc_html_e( 'Share of authority', 'ai-o-optimiser' ); ?></strong><div class="ai-o-optimiser-card-value"><?php echo esc_html( isset( $summary['share_of_authority'] ) ? number_format_i18n( (float) $summary['share_of_authority'], 2 ) . '%' : '—' ); ?></div><p><?php esc_html_e( 'Share reported for the imported Microsoft query set.', 'ai-o-optimiser' ); ?></p></div>
						<div class="ai-o-optimiser-card"><strong><?php esc_html_e( 'AI referral traffic', 'ai-o-optimiser' ); ?></strong><div class="ai-o-optimiser-card-value"><?php echo esc_html( isset( $summary['ai_referral_traffic'] ) ? number_format_i18n( (int) $summary['ai_referral_traffic'] ) : '—' ); ?></div><p><?php esc_html_e( 'Microsoft-reported referral sessions, kept separate from citations.', 'ai-o-optimiser' ); ?></p></div>
					</div>
				</section>

				<?php if ( $queries ) : ?>
				<section class="ai-o-optimiser-panel"><h2><?php esc_html_e( 'Grounding queries', 'ai-o-optimiser' ); ?></h2><p class="description"><?php esc_html_e( 'Grouped retrieval/query phrases from Microsoft. These are not exact user prompts.', 'ai-o-optimiser' ); ?></p><table class="widefat striped"><thead><tr><th><?php esc_html_e( 'Query', 'ai-o-optimiser' ); ?></th><th><?php esc_html_e( 'Citations', 'ai-o-optimiser' ); ?></th><th><?php esc_html_e( 'Share', 'ai-o-optimiser' ); ?></th><th><?php esc_html_e( 'Type', 'ai-o-optimiser' ); ?></th></tr></thead><tbody><?php foreach ( array_slice( $queries, 0, 25 ) as $row ) : ?><tr><td><?php echo esc_html( $row['query'] ); ?></td><td><?php echo esc_html( number_format_i18n( (int) $row['citations'] ) ); ?></td><td><?php echo esc_html( number_format_i18n( (float) $row['share'], 2 ) . '%' ); ?></td><td><?php echo esc_html( $row['query_type'] ?: '—' ); ?></td></tr><?php endforeach; ?></tbody></table></section>
				<?php endif; ?>

				<?php if ( $pages ) : ?>
				<section class="ai-o-optimiser-panel"><h2><?php esc_html_e( 'Cited pages', 'ai-o-optimiser' ); ?></h2><table class="widefat striped"><thead><tr><th><?php esc_html_e( 'Page', 'ai-o-optimiser' ); ?></th><th><?php esc_html_e( 'Citations', 'ai-o-optimiser' ); ?></th><th><?php esc_html_e( 'Share', 'ai-o-optimiser' ); ?></th></tr></thead><tbody><?php foreach ( array_slice( $pages, 0, 25 ) as $row ) : ?><tr><td><a href="<?php echo esc_url( $row['url'] ); ?>" target="_blank" rel="noopener noreferrer"><?php echo esc_html( $row['url'] ); ?></a></td><td><?php echo esc_html( number_format_i18n( (int) $row['citations'] ) ); ?></td><td><?php echo esc_html( number_format_i18n( (float) $row['share'], 2 ) . '%' ); ?></td></tr><?php endforeach; ?></tbody></table></section>
				<?php endif; ?>
			<?php endif; ?>

			<section class="ai-o-optimiser-panel">
				<h2><?php esc_html_e( 'Import Microsoft export', 'ai-o-optimiser' ); ?></h2>
				<p><?php esc_html_e( 'Export the Overview, Grounding queries or Cited pages view from Microsoft AI Performance / Clarity as CSV, then import it here. Import multiple CSVs for the same period to combine the views.', 'ai-o-optimiser' ); ?></p>
				<form method="post" enctype="multipart/form-data" action="<?php echo esc_url( admin_url( 'admin-post.php' ) ); ?>">
					<?php wp_nonce_field( 'ai_o_optimiser_microsoft_ai_import' ); ?>
					<input type="hidden" name="action" value="ai_o_optimiser_microsoft_ai_import" />
					<table class="form-table" role="presentation"><tbody>
					<tr><th><label for="ai-o-microsoft-ai-file"><?php esc_html_e( 'CSV export', 'ai-o-optimiser' ); ?></label></th><td><input type="file" id="ai-o-microsoft-ai-file" name="microsoft_ai_csv" accept=".csv,text/csv" required /><p class="description"><?php echo esc_html( sprintf( /* translators: %s file size. */ __( 'CSV only, maximum %s.', 'ai-o-optimiser' ), size_format( self::MAX_FILE_BYTES ) ) ); ?></p></td></tr>
					<tr><th><label for="ai-o-microsoft-ai-start"><?php esc_html_e( 'Period start', 'ai-o-optimiser' ); ?></label></th><td><input type="date" id="ai-o-microsoft-ai-start" name="period_start" required /></td></tr>
					<tr><th><label for="ai-o-microsoft-ai-end"><?php esc_html_e( 'Period end', 'ai-o-optimiser' ); ?></label></th><td><input type="date" id="ai-o-microsoft-ai-end" name="period_end" required /></td></tr>
					</tbody></table>
					<?php submit_button( __( 'Import Microsoft AI evidence', 'ai-o-optimiser' ) ); ?>
				</form>
				<?php if ( ! empty( $snapshot ) ) : ?><form method="post" action="<?php echo esc_url( admin_url( 'admin-post.php' ) ); ?>" style="margin-top:12px"><?php wp_nonce_field( 'ai_o_optimiser_microsoft_ai_clear' ); ?><input type="hidden" name="action" value="ai_o_optimiser_microsoft_ai_clear" /><?php submit_button( __( 'Clear imported Microsoft evidence', 'ai-o-optimiser' ), 'secondary', 'submit', false ); ?></form><?php endif; ?>
			</section>
		</div>
		<?php
	}
}