From 0fd8341c777a847d13073b96d7f3850c65150465 Mon Sep 17 00:00:00 2001
From: Dimitris Massaad <dmassaad@admin.grnet.gr>
Date: Sun, 29 Jun 2025 20:36:26 +0300
Subject: [PATCH] wp:author entries can be multiline

---
 phpunit/data/valid-wxr-1.1.xml         |  8 ++++++++
 phpunit/tests/parser.php               | 12 ++++++++++++
 src/parsers/class-wxr-parser-regex.php | 23 +++++++++++++----------
 3 files changed, 33 insertions(+), 10 deletions(-)

diff --git a/phpunit/data/valid-wxr-1.1.xml b/phpunit/data/valid-wxr-1.1.xml
index f389741f..12c962fe 100644
--- a/phpunit/data/valid-wxr-1.1.xml
+++ b/phpunit/data/valid-wxr-1.1.xml
@@ -36,6 +36,14 @@
 	<wp:base_blog_url>http://localhost/</wp:base_blog_url>
 
 	<wp:author><wp:author_id>2</wp:author_id><wp:author_login>john</wp:author_login><wp:author_email>johndoe@example.org</wp:author_email><wp:author_display_name><![CDATA[John Doe]]></wp:author_display_name><wp:author_first_name><![CDATA[John]]></wp:author_first_name><wp:author_last_name><![CDATA[Doe]]></wp:author_last_name></wp:author>
+	<wp:author>
+		<wp:author_id>3</wp:author_id>
+		<wp:author_login>jane</wp:author_login>
+		<wp:author_email>janedoe@example.org</wp:author_email>
+		<wp:author_display_name><![CDATA[Jane Doe]]></wp:author_display_name>
+		<wp:author_first_name><![CDATA[Jane]]></wp:author_first_name>
+		<wp:author_last_name><![CDATA[Doe]]></wp:author_last_name>
+	</wp:author>
 
 	<wp:category><wp:term_id>3</wp:term_id><wp:category_nicename>alpha</wp:category_nicename><wp:category_parent></wp:category_parent><wp:cat_name><![CDATA[alpha]]></wp:cat_name><wp:category_description><![CDATA[The alpha category]]></wp:category_description></wp:category>
 	<wp:tag><wp:term_id>22</wp:term_id><wp:tag_slug>clippable</wp:tag_slug><wp:tag_name><![CDATA[Clippable]]></wp:tag_name><wp:tag_description><![CDATA[The Clippable post_tag]]></wp:tag_description></wp:tag>
diff --git a/phpunit/tests/parser.php b/phpunit/tests/parser.php
index 3464a5e3..d55ca250 100644
--- a/phpunit/tests/parser.php
+++ b/phpunit/tests/parser.php
@@ -80,6 +80,18 @@ public function test_wxr_version_1_1() {
 				$result['authors']['john'],
 				$message
 			);
+			$this->assertEqualSetsWithIndex(
+				array(
+					'author_id'           => 3,
+					'author_login'        => 'jane',
+					'author_email'        => 'janedoe@example.org',
+					'author_display_name' => 'Jane Doe',
+					'author_first_name'   => 'Jane',
+					'author_last_name'    => 'Doe',
+				),
+				$result['authors']['jane'],
+				$message
+			);
 			$this->assertEqualSetsWithIndex(
 				array(
 					'term_id'              => 3,
diff --git a/src/parsers/class-wxr-parser-regex.php b/src/parsers/class-wxr-parser-regex.php
index 4ac1a96d..de930070 100644
--- a/src/parsers/class-wxr-parser-regex.php
+++ b/src/parsers/class-wxr-parser-regex.php
@@ -34,6 +34,7 @@ public function parse( $file ) {
 			'wp:category' => array( 'categories', array( $this, 'process_category' ) ),
 			'wp:tag'      => array( 'tags', array( $this, 'process_tag' ) ),
 			'wp:term'     => array( 'terms', array( $this, 'process_term' ) ),
+			'wp:author'   => array( 'authors', array( $this, 'process_author' ) ),
 		);
 
 		$fp = $this->fopen( $file, 'r' );
@@ -60,20 +61,17 @@ public function parse( $file ) {
 					$this->base_blog_url = $this->base_url;
 				}
 
-				if ( false !== strpos( $importline, '<wp:author>' ) ) {
-					preg_match( '|<wp:author>(.*?)</wp:author>|is', $importline, $author );
-					$a                                   = $this->process_author( $author[1] );
-					$this->authors[ $a['author_login'] ] = $a;
-					continue;
-				}
-
 				foreach ( $multiline_tags as $tag => $handler ) {
 					// Handle multi-line tags on a singular line
 					$pos         = strpos( $importline, "<$tag>" );
 					$pos_closing = strpos( $importline, "</$tag>" );
 					if ( preg_match( '|<' . $tag . '>(.*?)</' . $tag . '>|is', $importline, $matches ) ) {
-						$this->{$handler[0]}[] = call_user_func( $handler[1], $matches[1] );
-
+						$result = call_user_func( $handler[1], $matches[1] );
+						if ( 'wp:author' === $tag && isset( $result['author_login'] ) ) {
+							$this->authors[ $result['author_login'] ] = $result;
+						} else {
+							$this->{$handler[0]}[] = $result;
+						}
 					} elseif ( false !== $pos ) {
 						// Take note of any content after the opening tag
 						$multiline_content = trim( substr( $importline, $pos + strlen( $tag ) + 2 ) );
@@ -86,7 +84,12 @@ public function parse( $file ) {
 						$in_multiline       = false;
 						$multiline_content .= trim( substr( $importline, 0, $pos_closing ) );
 
-						$this->{$handler[0]}[] = call_user_func( $handler[1], $multiline_content );
+						$result = call_user_func( $handler[1], $multiline_content );
+						if ( 'wp:author' === $tag && isset( $result['author_login'] ) ) {
+							$this->authors[ $result['author_login'] ] = $result;
+						} else {
+							$this->{$handler[0]}[] = $result;
+						}
 					}
 				}