<?php
/**
 * Extracts text, headings, and links from post HTML
 *
 * @package Ntrnllnk
 */

namespace Ntrnllnk;

defined( 'ABSPATH' ) || exit;

/**
 * Extracts text, headings, and links from post HTML
 *
 * Like the other core classes, it does not call WordPress, so that it can be tested without it.
 */
final class Extract {

	/**
	 * Elements whose boundaries separate words
	 */
	private const ELEMENTS_BLOCK = 'address|article|aside|blockquote|br|dd|div|dl|dt|figcaption|figure|footer|h[1-6]|header|hr|li|nav|ol|p|pre|section|table|td|th|tr|ul';

	/**
	 * Returns the plain text of HTML, with whitespace collapsed
	 *
	 * @param string $html HTML.
	 */
	public static function text( string $html ): string {
		$html = preg_replace( '#<!--.*?-->#s', '', $html ) ?? '';
		$html = preg_replace( '#<(script|style)\b[^>]*>.*?</\1>#is', '', $html ) ?? '';
		$html = preg_replace( '#<(/?)(' . self::ELEMENTS_BLOCK . ')\b#i', ' <$1$2', $html ) ?? '';
		$text = html_entity_decode( strip_tags( $html ), ENT_QUOTES | ENT_HTML5, 'UTF-8' ); // phpcs:ignore WordPress.WP.AlternativeFunctions.strip_tags_strip_tags -- Kept WordPress-independent

		return trim( preg_replace( '/[\s\x{00A0}]+/u', ' ', $text ) ?? '' );
	}

	/**
	 * Returns the plain text of each heading
	 *
	 * @param string $html HTML.
	 * @return string[]
	 */
	public static function headings( string $html ): array {
		preg_match_all( '#<h([1-6])\b[^>]*>(.*?)</h\1>#is', $html, $matches );

		return array_map( [ self::class, 'text' ], $matches[2] );
	}

	/**
	 * Returns the highest heading level (the lowest number), or null if there are no headings
	 *
	 * @param string $html HTML.
	 */
	public static function heading_level_top( string $html ): ?int {
		preg_match_all( '#<h([1-6])\b#i', $html, $matches );

		return $matches[1] ? (int) min( $matches[1] ) : null;
	}

	/**
	 * Returns the unique, normalized targets of all links
	 *
	 * Root-relative targets are resolved against `$url_base`, if given; other relative targets are skipped.
	 *
	 * @param string $html HTML.
	 * @param string $url_base Site URL, e.g., `https://example.com`.
	 * @return string[]
	 */
	public static function links( string $html, string $url_base = '' ): array {
		return array_values( array_unique( array_filter( array_map( [ self::class, 'normalize_url' ], self::hrefs( $html, $url_base ) ) ) ) );
	}

	/**
	 * Returns the unique, absolute HTTP(S) targets of all links, as they are
	 *
	 * Root-relative targets are resolved against `$url_base`, if given; other relative targets are skipped.
	 *
	 * @param string $html HTML.
	 * @param string $url_base Site URL, e.g., `https://example.com`.
	 * @return string[]
	 */
	public static function hrefs( string $html, string $url_base = '' ): array {
		preg_match_all( '#<a\b[^>]*?\shref\s*=\s*(?:"([^"]*)"|\'([^\']*)\'|([^\s>]+))#i', $html, $matches, PREG_SET_ORDER | PREG_UNMATCHED_AS_NULL );

		$origin_base = self::origin( $url_base );
		$hrefs       = [];
		foreach ( $matches as $match ) {
			$href = trim( html_entity_decode( $match[1] ?? $match[2] ?? $match[3] ?? '', ENT_QUOTES | ENT_HTML5, 'UTF-8' ) );
			if ( str_starts_with( $href, '//' ) ) {
				$href = 'https:' . $href;
			} elseif ( str_starts_with( $href, '/' ) ) {
				if ( null === $origin_base ) {
					continue;
				}
				$href = $origin_base . $href;
			}
			if ( null !== self::normalize_url( $href ) ) {
				$hrefs[ $href ] = true;
			}
		}

		return array_keys( $hrefs );
	}

	/**
	 * Returns an absolute HTTP(S) URL as host and path, without “www.,” query, fragment, and trailing slash, and with the path decoded (but for slashes, which would change it), so that it is the same however encoded
	 *
	 * @param string $url URL.
	 */
	public static function normalize_url( string $url ): ?string {
		// Non-ASCII characters encoded first, as `parse_url()` replaces some of their bytes
		$url   = (string) preg_replace_callback( '/[^\x00-\x7F]+/', fn( array $matches ): string => rawurlencode( $matches[0] ), $url );
		$parts = parse_url( $url ); // phpcs:ignore WordPress.WP.AlternativeFunctions.parse_url_parse_url -- Kept WordPress-independent
		if ( ! is_array( $parts ) || empty( $parts['host'] ) || ! in_array( strtolower( $parts['scheme'] ?? '' ), [ 'http', 'https' ], true ) ) {
			return null;
		}
		$host = preg_replace( '/^www\./', '', strtolower( $parts['host'] ) );

		return $host . rtrim( rawurldecode( (string) preg_replace( '/%2F/i', '%252F', $parts['path'] ?? '' ) ), '/' );
	}

	/**
	 * Returns the ID of the post a URL names in its query (`?p=123`, `?page_id=123`), or null if it names none
	 *
	 * @param string $url URL.
	 */
	public static function id( string $url ): ?int {
		parse_str( (string) parse_url( $url, PHP_URL_QUERY ), $query ); // phpcs:ignore WordPress.WP.AlternativeFunctions.parse_url_parse_url -- Kept WordPress-independent
		$id = $query['p'] ?? $query['page_id'] ?? null;

		return is_string( $id ) && ctype_digit( $id ) ? (int) $id : null;
	}

	/**
	 * Returns scheme and host of a URL, or null if there are none
	 *
	 * @param string $url URL.
	 */
	private static function origin( string $url ): ?string {
		$parts = parse_url( $url ); // phpcs:ignore WordPress.WP.AlternativeFunctions.parse_url_parse_url -- Kept WordPress-independent
		if ( ! is_array( $parts ) || empty( $parts['scheme'] ) || empty( $parts['host'] ) ) {
			return null;
		}

		return $parts['scheme'] . '://' . $parts['host'] . ( isset( $parts['port'] ) ? ':' . $parts['port'] : '' );
	}
}