From b4041091ab5eaae15d2996edc1897e351252c7e5 Mon Sep 17 00:00:00 2001
From: Barnaby Walters <barnaby@waterpigs.co.uk>
Date: Wed, 23 Feb 2022 20:42:47 +0100
Subject: [PATCH] Fixed issue parsing elements with mixed valid and invalid mf2
 classnames on

Previously, an element with an invalid root classname such as  would
correctly not get parsed into a -type mf struct, but the h-12 would
prevent the element from being parsed for valid classnames. This is now fixed.
---
 Mf2/Parser.php           | 26 ++++++++++++++++----------
 tests/Mf2/ParserTest.php | 31 +++++++++++++++++++++++++++++++
 2 files changed, 47 insertions(+), 10 deletions(-)

diff --git a/Mf2/Parser.php b/Mf2/Parser.php
index e8e6dda..68c911a 100644
--- a/Mf2/Parser.php
+++ b/Mf2/Parser.php
@@ -144,6 +144,15 @@ function mfNamesFromClass($class, $prefix='h-') {
 	return $matches;
 }
 
+/**
+ * Registered with the XPath object and used within XPaths for finding root elements.
+ * @param string $class
+ * @return bool
+ */
+function classHasMf2RootClassname($class) {
+	return count(mfNamesFromClass($class, 'h-')) > 0;
+}
+
 /**
  * Get Nested µf Property Name From Class
  *
@@ -371,7 +380,10 @@ public function __construct($input, $url = null, $jsonMode = false) {
 			@$doc->loadHTML('');
 		}
 
+		// Create an XPath object and allow some PHP functions to be used within XPath queries.
 		$this->xpath = new DOMXPath($doc);
+		$this->xpath->registerNamespace('php', 'http://php.net/xpath');
+		$this->xpath->registerPhpFunctions('\\Mf2\\classHasMf2RootClassname');
 
 		$baseurl = $url;
 		foreach ($this->xpath->query('//base[@href]') as $base) {
@@ -1164,7 +1176,7 @@ public function parseH(\DOMElement $e, $is_backcompat = false, $has_nested_mf =
 			'type' => $mfTypes,
 			'properties' => $return
 		);
-		
+
 		if(trim($e->getAttribute('id')) !== '') {
 			$parsed['id'] = trim($e->getAttribute("id"));
 		}
@@ -1506,7 +1518,7 @@ public function parseFromId($id, $convertClassic=true) {
 	public function getRootMF(DOMElement $context = null) {
 		// start with mf2 root class name xpath
 		$xpaths = array(
-			'contains(concat(" ",normalize-space(@class)), " h-")'
+			'(php:function("\\Mf2\\classHasMf2RootClassname", normalize-space(@class)))'
 		);
 
 		// add mf1 root class names
@@ -1687,15 +1699,9 @@ public function addMfClasses(DOMElement $el, $classes) {
 	 */
 	public function hasRootMf2(\DOMElement $el) {
 		$class = str_replace(array("\t", "\n"), ' ', $el->getAttribute('class'));
-		$classes = array_filter(explode(' ', $class));
 
-		foreach ( $classes as $classname ) {
-			if ( strpos($classname, 'h-') === 0 ) {
-				return true;
-			}
-		}
-
-		return false;
+		// Check for valid mf2 root classnames, not just any classname with a h- prefix.
+		return count(mfNamesFromClass($class, 'h-')) > 0;
 	}
 
 	/**
diff --git a/tests/Mf2/ParserTest.php b/tests/Mf2/ParserTest.php
index e0f96cc..3338026 100644
--- a/tests/Mf2/ParserTest.php
+++ b/tests/Mf2/ParserTest.php
@@ -860,5 +860,36 @@ public function testInvalidOrdinalDate() {
 		$this->assertEquals('2016-12-31', Mf2\normalizeOrdinalDate('2016-366'));
 		$this->assertEquals('', Mf2\normalizeOrdinalDate('2016-367'));
 	}
+
+	/**
+	 * @see https://github.com/microformats/php-mf2/issues/230
+	 */
+	public function testPropertyWithInvalidHPrefixedRootClassParsed() {
+		$input = <<<EOF
+<div class="h-card">
+	<img class="u-photo w-32 h-32" alt="Jon Doe" src="/image.jpg"/>
+</div>
+EOF;
+
+		$output = Mf2\parse($input);
+		$this->assertEquals(array('value' => '/image.jpg', 'alt' => 'Jon Doe'), $output['items'][0]['properties']['photo'][0]);
+	}
+
+	public function testGetRootMfOnlyFindsValidElements() {
+		$input = <<<EOF
+<div class="h-entry>"> <a href="https://example.com" class="u-url">content</a></div>
+<div class="h-entry1>"> <a href="https://example.com" class="u-url">content</a></div>
+<div class="h-👍"> <a href="https://example.com" class="u-url">content</a></div>
+<div class="h-hentry_"> <a href="https://example.com" class="u-url">content</a></div>
+<div class="h-"> <a href="https://example.com" class="u-url">content</a></div>
+<div class="h-vendor123-name"><a href="https://example.com" class="u-url">content</a></div>
+EOF;
+
+		$p = new Mf2\Parser($input);
+		$rootEls = $p->getRootMF();
+
+		$this->assertEquals(1, count($rootEls));
+		$this->assertEquals('h-vendor123-name', $rootEls->item(0)->getAttribute('class'));
+	}
 }