From 908801ec7e08497d6fd45f06f2f3f90aa7d03688 Mon Sep 17 00:00:00 2001
From: Renato Alves <19148962+renatonascalves@users.noreply.github.com>
Date: Tue, 10 Sep 2024 12:51:31 -0300
Subject: [PATCH 1/3] WIP: issue-1157


From 1af588276369028daa81191ba4cd224b75b49ffb Mon Sep 17 00:00:00 2001
From: Renato Alves <19148962+renatonascalves@users.noreply.github.com>
Date: Tue, 10 Sep 2024 23:38:36 -0300
Subject: [PATCH 2/3] Adjust base URL to reflect the value set in the `home`
 option field

---
 admin/apple-actions/index/class-export.php    |   2 +-
 .../apple-exporter/class-exporter-content.php |   3 +-
 includes/apple-exporter/class-parser.php      |   6 +-
 .../components/test-class-cover.php           |  12 +-
 .../test-class-exporter-content.php           |  40 +++++-
 tests/apple-exporter/test-class-parser.php    | 124 ++++++++++++++++--
 6 files changed, 160 insertions(+), 27 deletions(-)

diff --git a/admin/apple-actions/index/class-export.php b/admin/apple-actions/index/class-export.php
index 7c602de4..b12a9175 100644
--- a/admin/apple-actions/index/class-export.php
+++ b/admin/apple-actions/index/class-export.php
@@ -141,7 +141,7 @@ public function fetch_exporter() {
 			if ( ! empty( $post_thumb_url ) ) {
 				// If the post thumb URL is root-relative, convert it to fully-qualified.
 				if ( str_starts_with( $post_thumb_url, '/' ) ) {
-					$post_thumb_url = site_url( $post_thumb_url );
+					$post_thumb_url = home_url( $post_thumb_url );
 				}
 
 				// Compile the post_thumb object using the URL and caption from the featured image.
diff --git a/includes/apple-exporter/class-exporter-content.php b/includes/apple-exporter/class-exporter-content.php
index 988ce69a..6d2bde9c 100644
--- a/includes/apple-exporter/class-exporter-content.php
+++ b/includes/apple-exporter/class-exporter-content.php
@@ -103,14 +103,13 @@ class Exporter_Content {
 	 *
 	 * @param string $url The URL to format.
 	 *
-	 * @access protected
 	 * @return string The formatted URL on success, or a blank string on failure.
 	 */
 	public static function format_src_url( $url ) {
 
 		// If this is a root-relative path, make absolute.
 		if ( 0 === strpos( $url, '/' ) ) {
-			$url = site_url( $url );
+			$url = home_url( $url );
 		}
 
 		// Decode the HTML entities since the URL is from the src attribute.
diff --git a/includes/apple-exporter/class-parser.php b/includes/apple-exporter/class-parser.php
index 9ed4c9a6..df29e44f 100644
--- a/includes/apple-exporter/class-parser.php
+++ b/includes/apple-exporter/class-parser.php
@@ -68,9 +68,9 @@ public function parse( $html ): string {
 		// Fork for format.
 		if ( 'html' === $this->format ) {
 			return $this->parse_html( $html );
-		} else {
-			return $this->parse_markdown( $html );
 		}
+
+		return $this->parse_markdown( $html );
 	}
 
 	/**
@@ -204,7 +204,7 @@ private function remove_empty_a_tags( string $html ): string {
 	private function handle_root_relative_urls( string $html ): string {
 		return preg_replace_callback(
 			'/(<a[^>]+href=(["\'])\/[^\/].*?\2[^>]*>)/m',
-			fn( $matches ) => str_replace( 'href="/', 'href="' . get_site_url() . '/', $matches[0] ),
+			fn( $matches ) => str_replace( 'href="/', 'href="' . get_home_url() . '/', $matches[0] ),
 			$html
 		);
 	}
diff --git a/tests/apple-exporter/components/test-class-cover.php b/tests/apple-exporter/components/test-class-cover.php
index 6828589b..4972b370 100644
--- a/tests/apple-exporter/components/test-class-cover.php
+++ b/tests/apple-exporter/components/test-class-cover.php
@@ -35,7 +35,7 @@ public function filter_apple_news_cover_json( $json ) {
 	 * @return string The filtered URL, made root-relative.
 	 */
 	public function filter_wp_get_attachment_url( $url ) {
-		return str_replace( site_url(), '', $url );
+		return str_replace( home_url(), '', $url );
 	}
 
 	/**
@@ -64,10 +64,10 @@ public function test_cover_role() {
 		// Create test post.
 		$post_id = self::factory()->post->create();
 
-		// Create a new attachment and assign it as the featured image for the cover component. 
-		set_post_thumbnail( 
-			$post_id, 
-			$this->get_new_attachment( 0 ) 
+		// Create a new attachment and assign it as the featured image for the cover component.
+		set_post_thumbnail(
+			$post_id,
+			$this->get_new_attachment( 0 )
 		);
 
 		// Check that the cover component's default role is 'photo'.
@@ -120,7 +120,7 @@ public function test_relative_url() {
 
 		// Create a post with an image with a root-relative src property and ensure its URL is expanded fully when converted to a Cover component.
 		$image_id_1 = $this->get_new_attachment( 0, 'Test Caption', 'Test alt text.' );
-		$post_id_1  = self::factory()->post->create( [ 'post_content' => str_replace( site_url(), '', $this->get_image_with_caption( $image_id_1 ) ) ] );
+		$post_id_1  = self::factory()->post->create( [ 'post_content' => str_replace( home_url(), '', $this->get_image_with_caption( $image_id_1 ) ) ] );
 		$json_1     = $this->get_json_for_post( $post_id_1 );
 		$this->assertEquals( wp_get_attachment_image_url( $image_id_1, 'full' ), $json_1['components'][0]['components'][0]['URL'] );
 
diff --git a/tests/apple-exporter/test-class-exporter-content.php b/tests/apple-exporter/test-class-exporter-content.php
index 32d8ecd7..2aadd0ea 100644
--- a/tests/apple-exporter/test-class-exporter-content.php
+++ b/tests/apple-exporter/test-class-exporter-content.php
@@ -67,10 +67,48 @@ public function test_complete_content_with_cover_config() {
 	/**
 	 * Ensure we decode the HTML entities in URLs extracted from HTML attributes.[type]
 	 */
-	public function test_format_src_url() {
+	public function test_format_src_url(): void {
 		$this->assertEquals(
 			'https://www.example.org/some.mp3?one=two&query=arg',
 			Exporter_Content::format_src_url( 'https://www.example.org/some.mp3?one=two&amp;query=arg' )
 		);
+
+		// Change the home URL.
+		update_option( 'home', 'https://www.custom-example.org' );
+
+		$this->assertEquals(
+			'https://www.example.org/some.mp3?one=two&query=arg',
+			Exporter_Content::format_src_url( 'https://www.example.org/some.mp3?one=two&amp;query=arg' )
+		);
+
+		$this->assertEquals(
+			'https://www.custom-example.org/',
+			Exporter_Content::format_src_url( '/' )
+		);
+
+		$this->assertEquals(
+			'https://www.custom-example.org',
+			Exporter_Content::format_src_url( 'https://www.custom-example.org' )
+		);
+
+		$this->assertNotEquals(
+			'https://www.custom-example.org',
+			Exporter_Content::format_src_url( '/' ),
+			'Root URL is missing a trailing slash.'
+		);
+
+		$this->assertEquals(
+			'https://www.example.org',
+			Exporter_Content::format_src_url( 'https://www.example.org' )
+		);
+
+		$this->assertNotEquals(
+			'https://www.example.org/', // The / here is intentional.
+			Exporter_Content::format_src_url( 'https://www.example.org' ),
+			'Root URL has a trailing slash.'
+		);
+
+		// Reset the home URL.
+		update_option( 'home', 'https://www.example.org' );
 	}
 }
diff --git a/tests/apple-exporter/test-class-parser.php b/tests/apple-exporter/test-class-parser.php
index 3ff5db44..97147ad8 100644
--- a/tests/apple-exporter/test-class-parser.php
+++ b/tests/apple-exporter/test-class-parser.php
@@ -23,11 +23,11 @@ class Apple_News_Parser_Test extends Apple_News_Testcase {
 	 */
 	public function test_parse_markdown(): void {
 		// Create a basic HTML post.
-		$post = '<html><body><h2>A heading</h2><p><strong>This is strong.</strong><br><a href="https://www.apple.com">This is a link</a></p></body></html>';
+		$html = '<html><body><h2>A heading</h2><p><strong>This is strong.</strong><br><a href="https://www.apple.com">This is a link</a></p></body></html>';
 
 		// Convert to Markdown.
 		$parser   = new Parser( 'markdown' );
-		$markdown = $parser->parse( $post );
+		$markdown = $parser->parse( $html );
 
 		// Verify.
 		$this->assertEquals( $markdown, "## A heading\n**This is strong.**\n[This is a link](https://www.apple.com)\n\n" );
@@ -38,10 +38,10 @@ public function test_parse_markdown(): void {
 	 */
 	public function test_parse_html(): void {
 		// Create a basic HTML post.
-		$post = '<h2 id="heading-target" class="someClass">A heading</h2><p><strong>This is strong.</strong><br><a href="https://www.apple.com" target="_blank">This is a link</a></p><div>The div tags will disappear.</div>';
+		$html_post = '<h2 id="heading-target" class="someClass">A heading</h2><p><strong>This is strong.</strong><br><a href="https://www.apple.com" target="_blank">This is a link</a></p><div>The div tags will disappear.</div>';
 
 		// Parse only HTML that's valid for Apple News.
-		$html = ( new Parser( 'html' ) )->parse( $post );
+		$html = ( new Parser( 'html' ) )->parse( $html_post );
 
 		// Verify.
 		$this->assertEquals( $html, 'A heading<p><strong>This is strong.</strong><br><a href="https://www.apple.com">This is a link</a></p>The div tags will disappear.' );
@@ -53,8 +53,6 @@ public function test_parse_html(): void {
 	 * @see \Apple_Exporter\Parser::parse
 	 */
 	public function test_clean_html_markdown(): void {
-		// Create a post.
-		global $post;
 		$post_content = <<<HTML
 <a href="https://www.google.com">Absolute link</a>
 
@@ -70,18 +68,17 @@ public function test_clean_html_markdown(): void {
 
 <a href="thisisntarealurl">Not a real URL</a>
 HTML;
-		$post         = $this->factory->post->create_and_get( // phpcs:ignore WordPress.WP.GlobalVariablesOverride.Prohibited
+		$article      = $this->factory->post->create_and_get(
 			[
 				'post_type'    => 'article',
 				'post_title'   => 'Test Article',
 				'post_content' => $post_content,
 			]
 		);
-		$permalink    = get_permalink( $post );
 
 		// Convert to Markdown.
 		$parser   = new Parser( 'markdown' );
-		$markdown = $parser->parse( apply_filters( 'the_content', $post->post_content ) ); // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound
+		$markdown = $parser->parse( apply_filters( 'the_content', $article->post_content ) ); // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound
 
 		// Verify.
 		$this->assertEquals(
@@ -96,14 +93,63 @@ public function test_clean_html_markdown(): void {
 		);
 	}
 
+	/**
+	 * Test the anchor cleaning functions of the parser for Markdown.
+	 *
+	 * @see \Apple_Exporter\Parser::parse
+	 */
+	public function test_clean_html_markdown_with_diffent_home_url(): void {
+		update_option( 'home', 'https://www.custom-example.org' );
+
+		$post_content = <<<HTML
+<a href="https://www.google.com">Absolute link</a>
+
+<a href="/2018/05/03/an-92-test">Root-relative link</a>
+
+<a id="testanchor">Test Anchor</a>
+
+<a href="#testanchor">Anchor Link</a>
+
+<a>Legit empty link</a>
+
+<a href=" ">Link that trims to empty</a>
+
+<a href="thisisntarealurl">Not a real URL</a>
+HTML;
+		$article      = $this->factory->post->create_and_get(
+			[
+				'post_type'    => 'article',
+				'post_title'   => 'Test Article',
+				'post_content' => $post_content,
+			]
+		);
+
+		// Convert to Markdown.
+		$parser   = new Parser( 'markdown' );
+		$markdown = $parser->parse( apply_filters( 'the_content', $article->post_content ) ); // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound
+
+		// Verify.
+		$this->assertEquals(
+			'[Absolute link](https://www.google.com)'
+			. '[Root-relative link](https://www.custom-example.org/2018/05/03/an-92-test)'
+			. 'Test Anchor'
+			. '[Anchor Link](#testanchor)'
+			. 'Legit empty link'
+			. 'Link that trims to empty'
+			. 'Not a real URL',
+			str_replace( "\n", '', $markdown )
+		);
+
+		// Reset the home URL.
+		update_option( 'home', 'https://www.example.org' );
+	}
+
 	/**
 	 * Test the anchor cleaning functions of the parser for HTML.
 	 *
 	 * @see \Apple_Exporter\Parser::parse
 	 */
 	public function test_clean_html(): void {
-		// Create a post.
-		global $post;
 		$post_content = <<<HTML
 <a href="https://www.google.com">Absolute link</a>
 
@@ -119,18 +165,17 @@ public function test_clean_html(): void {
 
 <a href="thisisntarealurl">Not a real URL</a>
 HTML;
-		$post         = $this->factory->post->create_and_get( // phpcs:ignore WordPress.WP.GlobalVariablesOverride.Prohibited
+		$article      = $this->factory->post->create_and_get(
 			[
 				'post_type'    => 'article',
 				'post_title'   => 'Test Article',
 				'post_content' => $post_content,
 			]
 		);
-		$permalink    = get_permalink( $post );
 
 		// Parse the post with HTML content format.
 		$parser      = new Parser( 'html' );
-		$parsed_html = $parser->parse( apply_filters( 'the_content', $post->post_content ) ); // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound
+		$parsed_html = $parser->parse( apply_filters( 'the_content', $article->post_content ) ); // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound
 
 		// Verify.
 		$this->assertEquals(
@@ -144,4 +189,55 @@ public function test_clean_html(): void {
 			str_replace( "\n", '', $parsed_html )
 		);
 	}
+
+	/**
+	 * Test the clean HTML function with a different root URL.
+	 *
+	 * @see \Apple_Exporter\Parser::parse
+	 */
+	public function test_clean_html_with_diffent_home_url(): void {
+		update_option( 'home', 'https://www.custom-example.org' );
+
+		$post_content = <<<HTML
+<a href="https://www.google.com">Absolute link</a>
+
+<a href="/2018/05/03/an-92-test">Root-relative link</a>
+
+<a id="testanchor">Test Anchor</a>
+
+<a href="#testanchor">Anchor Link</a>
+
+<a>Legit empty link</a>
+
+<a href=" ">Link that trims to empty</a>
+
+<a href="thisisntarealurl">Not a real URL</a>
+HTML;
+		$article      = $this->factory->post->create_and_get(
+			[
+				'post_type'    => 'article',
+				'post_title'   => 'Test Article',
+				'post_content' => $post_content,
+			]
+		);
+
+		// Parse the post with HTML content format.
+		$parser      = new Parser( 'html' );
+		$parsed_html = $parser->parse( apply_filters( 'the_content', $article->post_content ) ); // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound
+
+		// Verify.
+		$this->assertEquals(
+			'<p><a href="https://www.google.com">Absolute link</a></p>'
+				. '<p><a href="https://www.custom-example.org/2018/05/03/an-92-test">Root-relative link</a></p>'
+				. '<p>Test Anchor</p>'
+				. '<p><a href="#testanchor">Anchor Link</a></p>'
+				. '<p>Legit empty link</p>'
+				. '<p>Link that trims to empty</p>'
+				. '<p>Not a real URL</p>',
+			str_replace( "\n", '', $parsed_html )
+		);
+
+		// Reset the home URL.
+		update_option( 'home', 'https://www.example.org' );
+	}
 }

From 47550f8c2066610300e9526c0ebbea335396611c Mon Sep 17 00:00:00 2001
From: Renato Alves <19148962+renatonascalves@users.noreply.github.com>
Date: Tue, 10 Sep 2024 23:38:59 -0300
Subject: [PATCH 3/3] Ready for review