sirius-press/plugins/sirius-press-sia-export/includes/class-spe-mapper.php

<?php
/**
 * Where each URL lands in the name's bucket.
 *
 * WordPress URLs are directories: `/hello-world/`. Object storage has no
 * directories and no index resolution of its own, so every such URL becomes
 * `hello-world/index.html` and the resolver's directory-index rescue does the
 * rest. URLs that already name a file — `/robots.txt`, `/wp-content/…/app.css`
 * — keep their own name.
 *
 * The one rule this file exists to enforce is that the mapping is total and
 * reversible enough to be safe: anything that cannot be represented as a
 * gateway-legal path returns '' rather than being coerced into some other
 * file's name. A path collision in a static export is silent data loss — one
 * page overwrites another and nobody notices until a reader hits the wrong
 * article.
 *
 * @package SiriusPress
 */

defined( 'ABSPATH' ) || exit;

final class SPE_Mapper {

	/** Extensions treated as assets rather than pages. */
	const ASSET_EXTENSIONS = array(
		'css', 'js', 'mjs', 'json', 'map',
		'png', 'jpg', 'jpeg', 'gif', 'webp', 'avif', 'svg', 'ico',
		'woff', 'woff2', 'ttf', 'eot', 'otf',
		'mp3', 'mp4', 'webm', 'ogg', 'wav', 'pdf', 'zip', 'txt', 'xml',
	);

	/**
	 * Bucket path for a site-relative URL path.
	 *
	 * @param string $url_path e.g. '/blog/hello/' or '/wp-content/x.css'.
	 * @return string '' when the path cannot be exported.
	 */
	public static function path_for_url_path( $url_path ) {
		$home_path = (string) wp_parse_url( home_url( '/' ), PHP_URL_PATH );
		$url_path  = '/' . ltrim( (string) $url_path, '/' );

		// A site installed in a subdirectory serves its own root at that
		// prefix; the export has the name's root, so the prefix comes off.
		if ( '' !== $home_path && '/' !== $home_path && 0 === strpos( $url_path, $home_path ) ) {
			$url_path = '/' . ltrim( substr( $url_path, strlen( $home_path ) ), '/' );
		}

		$url_path = rawurldecode( $url_path );
		$trimmed  = trim( $url_path, '/' );

		if ( '' === $trimmed ) {
			return 'index.html';
		}

		// A URL naming a file keeps its name; a URL naming a directory gets an
		// index.html inside it. "Names a file" means the last segment has an
		// extension AND the URL did not end in a slash — `/about.us/` is a
		// page called "about.us", not a file.
		$slash = strrpos( $trimmed, '/' );
		$last  = false === $slash ? $trimmed : substr( $trimmed, $slash + 1 );
		$path  = ( false !== strpos( $last, '.' ) && '/' !== substr( $url_path, -1 ) )
			? $trimmed
			: $trimmed . '/index.html';

		// Everything the gateway would reject is better refused here, where
		// there is still a chance to leave the original URL alone.
		return SP_Gateway::clean_path( $path );
	}

	/** Bucket path for a post, page or any public post type. */
	public static function path_for_post( $post ) {
		$post = get_post( $post );
		if ( ! $post ) {
			return '';
		}
		$permalink = get_permalink( $post );
		if ( ! $permalink ) {
			return '';
		}
		return self::path_for_url_path( (string) wp_parse_url( $permalink, PHP_URL_PATH ) );
	}

	public static function is_asset( $path ) {
		$ext = strtolower( (string) pathinfo( $path, PATHINFO_EXTENSION ) );
		return in_array( $ext, self::ASSET_EXTENSIONS, true ) && 'html' !== $ext;
	}

	/**
	 * Everything a full export should cover.
	 *
	 * @return array<string,array{url:string,kind:string}> path => descriptor.
	 */
	public static function full_site() {
		$targets = array();

		$add = function ( $url, $kind = 'page' ) use ( &$targets ) {
			if ( ! $url ) {
				return;
			}
			$path = self::path_for_url_path( (string) wp_parse_url( $url, PHP_URL_PATH ) );
			if ( '' === $path ) {
				return;
			}
			$targets[ $path ] = array(
				'url'  => $url,
				'kind' => $kind,
			);
		};

		$add( home_url( '/' ) );

		// Every published thing with a public permalink.
		$types = get_post_types( array( 'public' => true ), 'names' );
		unset( $types['attachment'] );
		$posts = get_posts(
			array(
				'post_type'        => array_values( $types ),
				'post_status'      => 'publish',
				'numberposts'      => -1,
				'fields'           => 'ids',
				'suppress_filters' => false,
			)
		);
		foreach ( $posts as $post_id ) {
			$add( get_permalink( $post_id ) );
		}

		// Archives people actually navigate through. Date archives are left
		// out by default: on a long-running blog they multiply into hundreds
		// of near-identical pages that nothing links to.
		foreach ( get_taxonomies( array( 'public' => true ), 'names' ) as $taxonomy ) {
			$terms = get_terms(
				array(
					'taxonomy'   => $taxonomy,
					'hide_empty' => true,
				)
			);
			if ( is_wp_error( $terms ) ) {
				continue;
			}
			foreach ( $terms as $term ) {
				$link = get_term_link( $term );
				if ( ! is_wp_error( $link ) ) {
					$add( $link, 'archive' );
				}
			}
		}

		if ( 'page' === get_option( 'show_on_front' ) && get_option( 'page_for_posts' ) ) {
			$add( get_permalink( (int) get_option( 'page_for_posts' ) ), 'archive' );
		}

		/**
		 * Filters the full list of export targets.
		 *
		 * @param array $targets path => {url, kind}.
		 */
		return apply_filters( 'sirius_press_export_targets', $targets );
	}

	/**
	 * The paths that go stale when one post changes.
	 *
	 * A post edit does not only change the post: it changes the home page, the
	 * archives it belongs to, and the posts either side of it if the theme
	 * shows previous/next links. Being generous here costs a few uploads;
	 * being stingy leaves a reader looking at a stale index.
	 *
	 * @return array<string,array{url:string,kind:string}>
	 */
	public static function affected_by_post( $post ) {
		$post    = get_post( $post );
		$targets = array();
		if ( ! $post ) {
			return $targets;
		}

		$add = function ( $url, $kind = 'page' ) use ( &$targets ) {
			if ( ! $url || is_wp_error( $url ) ) {
				return;
			}
			$path = SPE_Mapper::path_for_url_path( (string) wp_parse_url( $url, PHP_URL_PATH ) );
			if ( '' !== $path ) {
				$targets[ $path ] = array(
					'url'  => $url,
					'kind' => $kind,
				);
			}
		};

		if ( 'publish' === $post->post_status ) {
			$add( get_permalink( $post ) );
		}
		$add( home_url( '/' ) );

		if ( 'page' === get_option( 'show_on_front' ) && get_option( 'page_for_posts' ) ) {
			$add( get_permalink( (int) get_option( 'page_for_posts' ) ), 'archive' );
		}

		foreach ( get_object_taxonomies( $post->post_type, 'names' ) as $taxonomy ) {
			$tax = get_taxonomy( $taxonomy );
			if ( ! $tax || ! $tax->public ) {
				continue;
			}
			foreach ( (array) get_the_terms( $post, $taxonomy ) as $term ) {
				if ( $term instanceof WP_Term ) {
					$add( get_term_link( $term ), 'archive' );
				}
			}
		}

		/**
		 * Filters which paths a post change invalidates.
		 *
		 * @param array   $targets
		 * @param WP_Post $post
		 */
		return apply_filters( 'sirius_press_affected_paths', $targets, $post );
	}
}