OPML Import: Support OPML without nesting (#403)

akirk · Nov 29, 2024 · 5cd4454 · 5cd4454
1 parent 975a86c
commit 5cd4454
Show file tree

Hide file tree

Showing 5 changed files with 192 additions and 43 deletions.
diff --git a/includes/class-import.php b/includes/class-import.php
@@ -18,57 +18,95 @@
  * @author Alex Kirk
  */
 class Import {
-	public static function opml( $opml ) {
-		$opml = simplexml_load_string( $opml );
-		if ( ! $opml ) {
-			return new \WP_Error( 'friends_import_opml_error', __( 'Failed to parse OPML.', 'friends' ) );
+
+	private static function get_feed_from_opml_node( $friend ) {
+		$xml_url = (string) $friend['xmlUrl'];
+		if ( ! $xml_url ) {
+			return null;
 		}
 
-		$feeds = array();
-		foreach ( $opml->body->outline as $outline ) {
-			$role = (string) $outline['text'];
-			foreach ( $outline->outline as $friend ) {
-				$user_login = str_replace( ' ', '-', strtolower( sanitize_user( (string) $friend['text'] ) ) );
-				$feed = User_Feed::get_by_url( (string) $friend['xmlUrl'] );
-				if ( $feed instanceof User_Feed ) {
-					$friend_user = $feed->get_friend_user();
-					if ( ! isset( $feeds[ $friend_user->user_login ] ) ) {
-						$feeds[ $friend_user->user_login ] = array();
-					}
-					$feeds[ $friend_user->user_login ][] = $feed;
-					continue;
-				}
+		$username = (string) $friend['text'];
+		if ( ! $username ) {
+			$username = (string) $friend['title'];
+		}
+		if ( ! $username ) {
+			$username = (string) $friend['htmlUrl'];
+			$username = preg_replace( '/^https?:\/\//', '', $username );
+		}
+		if ( ! $username ) {
+			$username = (string) $xml_url;
+			$username = preg_replace( '/^https?:\/\//', '', $username );
+		}
 
-				$friend_user = User::get_by_username( $user_login );
-				if ( ! $friend_user || is_wp_error( $friend_user ) ) {
-					$friend_user = Subscription::create(
-						$user_login,
-						'subscription',
-						(string) $friend['htmlUrl'],
-						(string) $friend['text']
-					);
-				}
-				if ( ! $friend_user instanceof User ) {
-					continue;
-				}
-				$feed = $friend_user->save_feed(
-					(string) $friend['xmlUrl'],
-					array(
-						'active'    => true,
-						'mime-type' => 'atom' === (string) $friend['type'] ? 'application/atom+xml' : 'application/rss+xml',
-					)
-				);
-				if ( ! $feed instanceof User_Feed ) {
-					continue;
-				}
+		$user_login = User::sanitize_username( $username );
+		$feed = User_Feed::get_by_url( $xml_url );
+		if ( $feed instanceof User_Feed ) {
+			return $feed;
+		}
+
+		$friend_user = User::get_by_username( $user_login );
+		if ( ! $friend_user || is_wp_error( $friend_user ) ) {
+			$friend_user = Subscription::create(
+				$user_login,
+				'subscription',
+				(string) $friend['htmlUrl'],
+				(string) $friend['text']
+			);
+		}
+
+		if ( ! $friend_user instanceof User ) {
+			return null;
+		}
+
+		$feed = $friend_user->save_feed(
+			$xml_url,
+			array(
+				'active'    => true,
+				'mime-type' => 'atom' === (string) $friend['type'] ? 'application/atom+xml' : 'application/rss+xml',
+			)
+		);
+
+		if ( ! $feed instanceof User_Feed ) {
+			if ( is_wp_error( $feed ) && apply_filters( 'friends_debug', false ) ) {
+				wp_trigger_error( __FUNCTION__, $feed->get_error_message() );
+
+			}
+			return null;
+		}
+
+		return $feed;
+	}
+
+	private static function recurse_into_opml( $feeds, $friend ) {
+		$xml_url = (string) $friend['xmlUrl'];
+		if ( $xml_url ) {
+			$feed = self::get_feed_from_opml_node( $friend );
+			if ( $feed instanceof User_Feed ) {
+				$friend_user = $feed->get_friend_user();
 
 				if ( ! isset( $feeds[ $friend_user->user_login ] ) ) {
 					$feeds[ $friend_user->user_login ] = array();
 				}
 				$feeds[ $friend_user->user_login ][] = $feed;
 			}
+			return $feeds;
+		}
+
+		foreach ( $friend->outline as $child ) {
+			$feeds = self::recurse_into_opml( $feeds, $child );
+		}
+
+		return $feeds;
+	}
+
+	public static function opml( $opml ) {
+		$opml = simplexml_load_string( $opml );
+		if ( ! $opml ) {
+			return new \WP_Error( 'friends_import_opml_error', __( 'Failed to parse OPML.', 'friends' ) );
 		}
 
+		$feeds = self::recurse_into_opml( array(), $opml->body );
+
 		return $feeds;
 	}
 }
diff --git a/includes/class-user.php b/includes/class-user.php
@@ -388,7 +388,7 @@ public function save_feeds( $feeds = array() ) {
 		$errors = new \WP_Error();
 		foreach ( $feeds as $feed_url => $options ) {
 			if ( ! is_string( $feed_url ) || ! Friends::check_url( $feed_url ) ) {
-				$errors->add( 'invalid-url', 'An invalid URL was provided' );
+				$errors->add( 'invalid-url', 'An invalid URL was provided', $feed_url );
 				unset( $feeds[ $feed_url ] );
 				continue;
 			}
@@ -406,7 +406,8 @@ public function save_feeds( $feeds = array() ) {
 
 		$all_urls = array();
 		foreach ( wp_get_object_terms( $this->get_object_id(), User_Feed::TAXONOMY ) as $term ) {
-			$all_urls[ $term->name ] = $term->term_id;
+			$url = str_replace( '&amp;', '&', $term->name );
+			$all_urls[ $url ] = $term->term_id;
 		}
 
 		$user_feeds = wp_set_object_terms( $this->get_object_id(), array_keys( array_merge( $all_urls, $feeds ) ), User_Feed::TAXONOMY );
@@ -415,7 +416,8 @@ public function save_feeds( $feeds = array() ) {
 		}
 
 		foreach ( wp_get_object_terms( $this->get_object_id(), User_Feed::TAXONOMY ) as $term ) {
-			$all_urls[ $term->name ] = $term->term_id;
+			$url = str_replace( '&amp;', '&', $term->name );
+			$all_urls[ $url ] = $term->term_id;
 		}
 
 		foreach ( $feeds as $url => $feed_options ) {

diff --git a/tests/data/feedland.opml b/tests/data/feedland.opml
@@ -0,0 +1,44 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<opml version="2.0">
+	<head>
+		<title>bph/wordpress</title>
+		<description>bph's subscription list, wordpress category. List created by feedlandDatabase v0.7.55.</description>
+		<dateModified>Thu, 28 Nov 2024 16:14:43 GMT</dateModified>
+		</head>
+	<body>
+		<outline type="rss" text="WP Mayor" xmlUrl="https://wpmayor.com/feed/" htmlUrl="https://wpmayor.com/" category="wordpress"/>
+		<outline type="rss" text="Block themes — WordPress.org" xmlUrl="https://wordpress.org/themes/tags/full-site-editing/feed/" htmlUrl="https://wordpress.org/themes" category="wordpress"/>
+		<outline type="rss" text="All themes — WordPress.org" xmlUrl="https://wordpress.org/themes/feed/" htmlUrl="https://wordpress.org/themes" category="wordpress"/>
+		<outline type="rss" text="WP BizDev" xmlUrl="https://wpbiz.dev/feed/" htmlUrl="https://wpbiz.dev/" category="wordpress"/>
+		<outline type="rss" text="WordPress Guides, Tutorials &amp; Tips - Kinsta® Blog" xmlUrl="https://kinsta.com/blog/feed/" htmlUrl="https://kinsta.com/blog/" category="wordpress"/>
+		<outline type="rss" text="WebTNG" xmlUrl="https://www.webtng.com/feed/" htmlUrl="https://www.webtng.com/" category="wordpress"/>
+		<outline type="rss" text="Automattic Design" xmlUrl="https://automattic.design/feed/" htmlUrl="https://automattic.design/" category="wordpress"/>
+		<outline type="rss" text="WP-CONTENT.CO" xmlUrl="https://wp-content.co/feed" htmlUrl="https://wp-content.co/" category="wordpress"/>
+		<outline type="rss" text="Brian Coords" xmlUrl="https://www.briancoords.com/feed" htmlUrl="https://www.briancoords.com/" category="wordpress"/>
+		<outline type="rss" text="WP Plugins RSS" xmlUrl="https://wp-plugins-rss.edavis.workers.dev/feed/" htmlUrl="https://wordpress.org/plugins/browse/new/" category="wordpress"/>
+		<outline type="rss" text="editortips" xmlUrl="https://editortips.com/feed/" htmlUrl="https://editortips.com/" category="wordpress"/>
+		<outline type="rss" text="The WooCommerce Developer Blog" xmlUrl="https://developer.woo.com/feed/" htmlUrl="https://developer.woocommerce.com/" category="wordpress"/>
+		<outline type="rss" text="DevotePress" xmlUrl="https://devotepress.com/feed/" htmlUrl="https://devotepress.com/" category="wordpress"/>
+		<outline type="rss" text="Rich Tabor" xmlUrl="https://richtabor.com/feed/" htmlUrl="https://rich.blog/" category="wordpress"/>
+		<outline type="rss" text="Post Status" xmlUrl="https://poststatus.com/feed/" htmlUrl="https://poststatus.com/" category="wordpress"/>
+		<outline type="rss" text="WPShout" xmlUrl="https://www.codeinwp.com/feed/" htmlUrl="https://wpshout.com/" category="wordpress"/>
+		<outline type="rss" text="Matt Mullenweg" xmlUrl="https://ma.tt/feed/" htmlUrl="https://ma.tt/" category="wordpress"/>
+		<outline type="rss" text="WordPress News" xmlUrl="https://wordpress.org/news/feed/" htmlUrl="https://wordpress.org/news" category="wordpress"/>
+		<outline type="rss" text="Human Made Blog" xmlUrl="https://humanmade.com/blog/feed/" htmlUrl="https://humanmade.com/blog/" category="wordpress"/>
+		<outline type="rss" text="Builders" xmlUrl="https://wpengine.com/builders/feed/" htmlUrl="https://wpengine.com/builders/" category="wordpress"/>
+		<outline type="rss" text="Tom McFarlin" xmlUrl="https://tommcfarlin.com/feed/" htmlUrl="https://tommcfarlin.com/" category="wordpress"/>
+		<outline type="rss" text="Tutorials | Learn WordPress" xmlUrl="https://learn.wordpress.org/tutorials/feed/" htmlUrl="https://learn.wordpress.org/" category="wordpress"/>
+		<outline type="rss" text="Full Site Editing" xmlUrl="https://fullsiteediting.com/feed/" htmlUrl="https://fullsiteediting.com/" category="wordpress"/>
+		<outline type="rss" text="Blog – WP Fieldwork" xmlUrl="https://wpfieldwork.com/blog/feed/" htmlUrl="https://wpfieldwork.com/" category="wordpress"/>
+		<outline type="rss" text="Gutenberg Hub" xmlUrl="https://gutenberghub.com/feed" htmlUrl="https://gutenberghub.com/" category="wordpress"/>
+		<outline type="rss" text="The WP Minute" xmlUrl="https://thewpminute.com/feed/" htmlUrl="https://thewpminute.com/" category="wordpress"/>
+		<outline type="rss" text="WordPress Developer Blog" xmlUrl="https://developer.wordpress.org/news/feed/" htmlUrl="https://developer.wordpress.org/news" category="wordpress"/>
+		<outline type="rss" text="Ollie WordPress Block Theme" xmlUrl="https://olliewp.com/feed/" htmlUrl="https://olliewp.com/" category="wordpress"/>
+		<outline type="rss" text="WordPress.tv" xmlUrl="https://wordpress.tv/feed/" htmlUrl="https://wordpress.tv/" category="wordpress"/>
+		<outline type="rss" text="Torque" xmlUrl="https://torquemag.io/feed/" htmlUrl="https://torquemag.io/" category="wordpress"/>
+		<outline type="rss" text="WordPress – CSS-Tricks" xmlUrl="https://css-tricks.com/tag/wordpress/feed/" htmlUrl="https://css-tricks.com/" category="wordpress"/>
+		<outline type="rss" text="Blog - Bill Erickson" xmlUrl="https://www.billerickson.net/blog/feed/" htmlUrl="https://www.billerickson.net/blog/" category="wordpress"/>
+		<outline type="rss" text="WP Builds" xmlUrl="https://wpbuilds.com/feed/podcast/" htmlUrl="https://wpbuilds.com/series/wp-builds/" category="wordpress"/>
+		<outline type="rss" text="WP Tavern" xmlUrl="https://wptavern.com/feed" htmlUrl="https://wptavern.com/" category="wordpress"/>
+		</body>
+	</opml>
diff --git a/tests/data/friends.opml b/tests/data/friends.opml
@@ -0,0 +1,37 @@
+<?xml version="1.0" encoding="utf-8"?><opml version="2.0">
+	<head>
+		<title>My Friends</title>
+		<dateCreated>Fri, 29 Nov 2024 06:47:29 +0000</dateCreated>
+		<ownerName>adm</ownerName>
+	</head>
+	<body>
+	<outline text="Friends">
+		<outline text="alex" htmlUrl="https://akirk.blog/friends/alex.mu.blog/" title="alex" type="rss" xmlUrl="https://alex.mu.blog/feed/?me=akirk.blog&#038;until=1764398848&#038;auth=$2y$12$H6eft9cdHzvnNopyVzC4C.TS.yuNGL1Qox3nu/qREMI1HqamMTQ7u"/>
+		<outline text="alex" htmlUrl="https://akirk.blog/friends/alex.mu.blog/" title="alex" type="rss" xmlUrl="https://mastodon.local/@alex"/>
+		<outline text="Matt on Not-WordPress" htmlUrl="https://akirk.blog/friends/matt.blog/" title="Matt on Not-WordPress" type="rss" xmlUrl="https://matt.blog/feed/?me=akirk.blog&#038;until=1764398848&#038;auth=$2y$12$hwATj8IIKltWpk3GJVPCnOEsgK8vbmk7mgReOzCaO8yh1kPWryaxa"/>
+	</outline>
+	<outline text="Subscriptions">
+		<outline text="Aaron Jorbin" htmlUrl="https://akirk.blog/friends/aaron-jorbin/" title="Aaron Jorbin" type="rss" xmlUrl="https://aaron.jorb.in/feed/"/>
+		<outline text="Adam Zielinski" htmlUrl="https://akirk.blog/friends/adam-zielinski/" title="Adam Zielinski" type="rss" xmlUrl="https://adamadam.blog/feed/"/>
+		<outline text="alex" htmlUrl="https://akirk.blog/friends/mastodon.local-users-alex/" title="alex" type="rss" xmlUrl="https://wpfriends.at/feed/"/>
+		<outline text="Alex Kirk" htmlUrl="https://akirk.blog/friends/alex-kirk/" title="Alex Kirk" type="rss" xmlUrl="https://alex.kirk.at/feed/"/>
+		<outline text="Andreas Klinger" htmlUrl="https://akirk.blog/friends/andreas-klinger/" title="Andreas Klinger" type="rss" xmlUrl="https://klinger.io/rss.xml"/>
+		<outline text="Barnaby Walters" htmlUrl="https://akirk.blog/friends/barnaby-walters/" title="Barnaby Walters" type="rss" xmlUrl="https://akirk.blog/friends/barnaby-walters/feed/"/>
+		<outline text="bengreeley.com" htmlUrl="https://akirk.blog/friends/bengreeley.com/" title="bengreeley.com" type="rss" xmlUrl="https://www.bengreeley.com/feed/"/>
+		<outline text="Chris Coyier" htmlUrl="https://akirk.blog/friends/chriscoyier.net/" title="Chris Coyier" type="rss" xmlUrl="https://chriscoyier.net/feed/"/>
+		<outline text="David Bisset" htmlUrl="https://akirk.blog/friends/davidbisset-phpc.social/" title="David Bisset" type="rss" xmlUrl="https://phpc.social/@davidbisset"/>
+		<outline text="Ingrid Brodnig" htmlUrl="https://akirk.blog/friends/mastodon.social-users-brodnig/" title="Ingrid Brodnig" type="rss" xmlUrl="https://mastodon.social/users/brodnig.rss"/>
+		<outline text="Jim Nielsen" htmlUrl="https://akirk.blog/friends/jim-nielsen.com/" title="Jim Nielsen" type="rss" xmlUrl="https://blog.jim-nielsen.com/feed.xml"/>
+		<outline text="Matt Mullenweg" htmlUrl="https://akirk.blog/friends/ma.tt/" title="Matt Mullenweg" type="rss" xmlUrl="https://ma.tt/feed/"/>
+		<outline text="mkln.org" htmlUrl="https://akirk.blog/friends/mkln.org/" title="mkln.org" type="rss" xmlUrl="https://mkln.org/feed"/>
+		<outline text="notiz.blog" htmlUrl="https://akirk.blog/friends/notiz.blog/" title="notiz.blog" type="rss" xmlUrl="https://notiz.blog/feed/"/>
+		<outline text="paolo.blog" htmlUrl="https://akirk.blog/friends/paolo.blog/" title="paolo.blog" type="rss" xmlUrl="https://paolo.blog/feed/"/>
+		<outline text="pfefferle" htmlUrl="https://akirk.blog/friends/pfefferle-mastodon-social/" title="pfefferle" type="rss" xmlUrl="https://mastodon-social/@pfefferle"/>
+		<outline text="restofworld.org" htmlUrl="https://akirk.blog/friends/restofworld.org/" title="restofworld.org" type="rss" xmlUrl="https://restofworld.org/feed/latest"/>
+		<outline text="simonwillison.net" htmlUrl="https://akirk.blog/friends/simonwillison.net/" title="simonwillison.net" type="atom" xmlUrl="https://simonwillison.net/atom/everything/"/>
+		<outline text="stratechery.com" htmlUrl="https://akirk.blog/friends/stratechery.com/" title="stratechery.com" type="rss" xmlUrl="https://stratechery.com/feed/"/>
+		<outline text="wptavern.com" htmlUrl="https://akirk.blog/friends/wptavern.com/" title="wptavern.com" type="rss" xmlUrl="https://wptavern.com/feed/"/>
+		<outline text="xkcd" htmlUrl="https://akirk.blog/friends/xkcd/" title="xkcd" type="rss" xmlUrl="https://xkcd.com/rss.xml"/>
+	</outline>
+	</body>
+</opml>
diff --git a/tests/test-feed.php b/tests/test-feed.php
@@ -678,4 +678,32 @@ public function test_podcast() {
 		$this->assertEquals( 'https://podcast.local/2022/10/episode-1/', $post->guid );
 		$this->assertStringContainsString( 'first-episode.mp3', $post->post_content );
 	}
+
+	public function test_import_feedland_opml() {
+		add_filter( 'friends_pre_check_url', '__return_true' );
+
+		$opml = file_get_contents( __DIR__ . '/data/feedland.opml' );
+		$feeds = Import::opml( $opml );
+		$users_created = count( $feeds );
+		$feeds_imported = 0;
+		foreach ( $feeds as $user => $user_feeds ) {
+			$feeds_imported += count( $user_feeds );
+		}
+		$this->assertEquals( 34, $users_created );
+		$this->assertEquals( 34, $feeds_imported );
+	}
+
+	public function test_import_friends_opml() {
+		add_filter( 'friends_pre_check_url', '__return_true' );
+
+		$opml = file_get_contents( __DIR__ . '/data/friends.opml' );
+		$feeds = Import::opml( $opml );
+		$users_created = count( $feeds );
+		$feeds_imported = 0;
+		foreach ( $feeds as $user => $user_feeds ) {
+			$feeds_imported += count( $user_feeds );
+		}
+		$this->assertEquals( 22, $users_created );
+		$this->assertEquals( 24, $feeds_imported );
+	}
 }