From dab54a884315c3bd19a775ba1f62a2277e9534d3 Mon Sep 17 00:00:00 2001
From: David Herrera <david@alley.com>
Date: Mon, 18 Nov 2024 05:10:36 -0500
Subject: [PATCH] Allow content matching selectors to be excluded

---
 admin/apple-actions/index/class-export.php    | 47 +++++++++++++++++++
 ...-admin-apple-settings-section-advanced.php | 14 ++++++
 .../class-admin-apple-settings-section.php    |  1 +
 includes/apple-exporter/class-settings.php    |  1 +
 .../apple-actions/index/test-class-export.php | 28 +++++++++++
 5 files changed, 91 insertions(+)

diff --git a/admin/apple-actions/index/class-export.php b/admin/apple-actions/index/class-export.php
index b12a9175..867cd94e 100644
--- a/admin/apple-actions/index/class-export.php
+++ b/admin/apple-actions/index/class-export.php
@@ -606,6 +606,7 @@ private function get_content( $post ) {
 		$content = apply_filters( 'the_content', $content ); // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound
 
 		// Clean up the HTML a little.
+		$content = $this->exclude_selectors( $content );
 		$content = $this->remove_tags( $content );
 		$content = $this->remove_entities( $content );
 
@@ -643,6 +644,52 @@ private function remove_entities( $content ) {
 		return str_replace( '&amp;', '&', $content );
 	}
 
+	/**
+	 * Remove excluded selectors from the content.
+	 *
+	 * @param string $content The content to be filtered.
+	 * @return string
+	 */
+	private function exclude_selectors( $content ) {
+		$raw_selectors = $this->get_setting( 'excluded_selectors' );
+
+		$selectors = explode( ',', $raw_selectors );
+		$selectors = array_map( 'trim', $selectors );
+		$selectors = array_filter( $selectors );
+
+		if ( count( $selectors ) === 0 ) {
+			return $content;
+		}
+
+		libxml_use_internal_errors( true );
+		$dom = new \DOMDocument();
+		$dom->loadHTML( '<?xml encoding="utf-8" ?>'. $content, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD );
+		$xpath = new \DOMXPath( $dom );
+		libxml_clear_errors();
+
+		foreach ( $selectors as $selector ) {
+			$nodes = [];
+
+			if ( str_starts_with( $selector, '#' ) ) {
+				$nodes = $xpath->query( '//*[@id="' . substr( $selector, 1 ) . '"]' );
+			}
+
+			if ( str_starts_with( $selector, '.' ) ) {
+				$nodes = $xpath->query( '//*[contains(concat(" ", normalize-space(@class), " "), " ' . substr( $selector, 1 ) . ' ")]' );
+			}
+
+			if ( is_iterable( $nodes ) ) {
+				foreach ( $nodes as $node ) {
+					$node->parentNode->removeChild( $node );
+				}
+			}
+		}
+
+		$content = $dom->saveHTML();
+
+		return $content;
+	}
+
 	/**
 	 * Loads settings for the Exporter_Content from the WordPress post metadata.
 	 *
diff --git a/admin/settings/class-admin-apple-settings-section-advanced.php b/admin/settings/class-admin-apple-settings-section-advanced.php
index fda8cd0b..17bb48b8 100644
--- a/admin/settings/class-admin-apple-settings-section-advanced.php
+++ b/admin/settings/class-admin-apple-settings-section-advanced.php
@@ -71,6 +71,16 @@ public function __construct( $page ) {
 				'description' => __( 'Enter a CSS class name that will be used to generate the Aside component. Do not prefix with a period.', 'apple-news' ),
 				'required'    => false,
 			],
+			'excluded_selectors'		=> [
+				'label'       => __( 'Selectors', 'apple-news' ),
+				'type'        => 'text',
+				'size'        => 150,
+				'description' => sprintf(
+					__( 'Enter a comma-separated list of CSS class or ID selectors, like %s. Elements in post content matching these selectors will be removed from the content published to Apple News.', 'apple-news' ),
+					'<code>.my-class, #my-id</code>',
+				),
+				'required'    => false,
+			],
 		];
 
 		// Add the groups.
@@ -91,6 +101,10 @@ public function __construct( $page ) {
 				'label'    => __( 'Aside Component', 'apple-news' ),
 				'settings' => [ 'aside_component_class' ],
 			],
+			'selectors' => [
+				'label'    => __( 'Excluded Elements', 'apple-news' ),
+				'settings' => [ 'excluded_selectors' ],
+			],
 		];
 
 		parent::__construct( $page );
diff --git a/admin/settings/class-admin-apple-settings-section.php b/admin/settings/class-admin-apple-settings-section.php
index a96aab80..2954abbd 100644
--- a/admin/settings/class-admin-apple-settings-section.php
+++ b/admin/settings/class-admin-apple-settings-section.php
@@ -135,6 +135,7 @@ class Admin_Apple_Settings_Section extends Apple_News {
 		'b'        => [],
 		'strong'   => [],
 		'i'        => [],
+		'code'     => [],
 		'em'       => [],
 		'a'        => [
 			'href'   => [],
diff --git a/includes/apple-exporter/class-settings.php b/includes/apple-exporter/class-settings.php
index fefb2220..dc1d8c7e 100644
--- a/includes/apple-exporter/class-settings.php
+++ b/includes/apple-exporter/class-settings.php
@@ -49,6 +49,7 @@ class Settings {
 		'post_types'                  => [ 'post' ],
 		'show_metabox'                => 'yes',
 		'use_remote_images'           => 'yes',
+		'excluded_selectors'          => '',
 	];
 
 	/**
diff --git a/tests/admin/apple-actions/index/test-class-export.php b/tests/admin/apple-actions/index/test-class-export.php
index 7ffa9ac6..64a12402 100644
--- a/tests/admin/apple-actions/index/test-class-export.php
+++ b/tests/admin/apple-actions/index/test-class-export.php
@@ -259,6 +259,34 @@ public function test_remove_entities() {
 		);
 	}
 
+	/**
+	 * Tests the ability to remove selectors from the content.
+	 */
+	public function test_remove_selectors() {
+		$post_id = $this->factory->post->create(
+			[
+				'post_content' => <<<HTML
+<p class="foo">First paragraph</p>
+<p class="bar">Second paragraph</p>
+<p class="foo bar">Third paragraph</p>
+<p class="foobar">Fourth paragraph</p>
+<p id="foo">Fifth paragraph</p>
+<p id="bar">Sixth paragraph</p>
+HTML,
+			]
+		);
+
+		$this->settings->excluded_selectors = '.foo, #foo';
+
+		$export   = new Export( $this->settings, $post_id );
+		$exporter = $export->fetch_exporter();
+		$content  = $exporter->get_content()->content();
+
+		$this->assertStringNotContainsString( 'First paragraph', $content );
+		$this->assertStringNotContainsString( 'Third paragraph', $content );
+		$this->assertStringNotContainsString( 'Fifth paragraph', $content );
+	}
+
 	/**
 	 * Tests the behavior of the apple_news_is_exporting() function.
 	 */