<?php
/**
 * Builds ai.txt.
 *
 * @package AcidStudios\AEO
 */

declare(strict_types=1);

namespace AcidStudios\AEO\Generators;

defined( 'ABSPATH' ) || exit;

/**
 * Mirrors the AI crawler policy into `/ai.txt`.
 *
 * `ai.txt` has no standard behind it, so this file is deliberately secondary:
 * the authoritative policy stays in robots.txt and in the `Content-Signal`
 * header, and the header comment says so. The per-bot groups are rendered by
 * `RobotsGenerator::generate_groups()` rather than repeated here, so the two
 * files cannot drift apart. Sitemaps are left out: this file is about
 * permission, not about discovery.
 *
 * Pure PHP, like every other generator.
 */
final class AiTxtGenerator {

	/**
	 * The line that points readers at the authoritative policy.
	 *
	 * @var string
	 */
	private const CANONICAL_NOTE = '# Canonical policy: robots.txt + Content-Signal. This file mirrors it for AI-specific tooling.';

	/**
	 * Comment that marks the file as this plugin's output.
	 *
	 * @var string
	 */
	private const MARKER = '# generated by Acid AEO';

	/**
	 * Renders the file.
	 *
	 * @param array<string,mixed> $data The policy array `Generators\RobotsGenerator` takes,
	 *                                  plus `owner`, `contact`, `license_label`, `license_url`
	 *                                  and `links` (`context`, `summary`, `humans`, `security`).
	 * @return string File body, `\n` separated, with a trailing newline.
	 */
	public function generate( array $data ): string {
		$blocks = [
			$this->header( $data ),
			( new RobotsGenerator() )->generate_groups( $data ),
			$this->content_signal_line( $data ),
			$this->footer( $data ),
		];

		return implode( "\n\n", array_filter( $blocks, static fn( string $block ): bool => '' !== $block ) ) . "\n";
	}

	/**
	 * The comment block that opens the file.
	 *
	 * @param array<string,mixed> $data Input array.
	 * @return string Header comment lines.
	 */
	private function header( array $data ): string {
		$site_name = $this->line( $data['site_name'] ?? '' );
		$title     = '# AI crawler policy';

		if ( '' !== $site_name ) {
			$title .= ' for ' . $site_name;
		}

		return implode( "\n", [ $title, self::CANONICAL_NOTE, self::MARKER ] );
	}

	/**
	 * The `Content-Signal` directive, serialized exactly as robots.txt serializes it.
	 *
	 * @param array<string,mixed> $data Input array.
	 * @return string Line, or an empty string when there is no signal.
	 */
	private function content_signal_line( array $data ): string {
		$signal = RobotsGenerator::content_signal_line( is_array( $data['content_signal'] ?? null ) ? $data['content_signal'] : [] );

		return '' === $signal ? '' : 'Content-Signal: ' . $signal;
	}

	/**
	 * The comment block that closes the file: who owns the content and where to read more.
	 *
	 * @param array<string,mixed> $data Input array.
	 * @return string Comment lines, or an empty string when nothing is known.
	 */
	private function footer( array $data ): string {
		$links = is_array( $data['links'] ?? null ) ? $data['links'] : [];
		$lines = [];

		foreach ( [
			'Owner'    => $data['owner'] ?? '',
			'Contact'  => $data['contact'] ?? '',
			'License'  => $this->license( $data ),
			'Context'  => $links['context'] ?? '',
			'Summary'  => $links['summary'] ?? '',
			'Humans'   => $links['humans'] ?? '',
			'Security' => $links['security'] ?? '',
		] as $name => $value ) {
			$value = $this->line( $value );

			if ( '' !== $value ) {
				$lines[] = '# ' . $name . ': ' . $value;
			}
		}

		return implode( "\n", $lines );
	}

	/**
	 * The content licence, as a label followed by its URL.
	 *
	 * A custom licence is stored as a URL and its label is that same URL, so the
	 * two are only joined when they differ.
	 *
	 * @param array<string,mixed> $data Input array.
	 * @return string Licence, empty when none is set.
	 */
	private function license( array $data ): string {
		$label = $this->line( $data['license_label'] ?? '' );
		$url   = $this->line( $data['license_url'] ?? '' );

		if ( '' === $url || $url === $label ) {
			return $label;
		}

		return '' !== $label ? $label . ' — ' . $url : $url;
	}

	/**
	 * A value as a single trimmed line.
	 *
	 * @param mixed $value Raw value.
	 * @return string Line, empty when the value is not text.
	 */
	private function line( $value ): string {
		if ( ! is_scalar( $value ) ) {
			return '';
		}

		return trim( (string) preg_replace( '/\s+/', ' ', (string) $value ) );
	}
}