From 6701504876d6c9242eb310b35f41d40d9785ab4e Mon Sep 17 00:00:00 2001
From: Barnaby Walters <barnaby@waterpigs.co.uk>
Date: Sun, 12 Jul 2015 15:09:26 +0100
Subject: [PATCH] Corrected handling of value and html keys of property-nested
 microformat structures

Fixes #64. As per https://github.com/indieweb/php-mf2/issues/64, microformats nested under e-* properties dropped their
html keys. When looking into fixing this, I found several other nested cases where nested mf structure value keys
were being incorrectly assigned. This commit fixes the handling of microformats nested under p-, e- and u- properties,
with associated tests.
---
 Mf2/Parser.php                         | 30 ++++++++++++++++++++-----
 tests/Mf2/CombinedMicroformatsTest.php | 31 +++++++++++++++++++++++++-
 tests/Mf2/ParserTest.php               |  4 ++--
 3 files changed, 56 insertions(+), 9 deletions(-)

diff --git a/Mf2/Parser.php b/Mf2/Parser.php
index a7e9ef6..b8a954f 100644
--- a/Mf2/Parser.php
+++ b/Mf2/Parser.php
@@ -153,12 +153,17 @@ function nestedMfPropertyNamesFromClass($class) {
 	$class = str_replace(array(' ', '	', "\n"), ' ', $class);
 	foreach (explode(' ', $class) as $classname) {
 		foreach ($prefixes as $prefix) {
-			$compare_classname = strtolower(' ' . $classname);
-			if (stristr($compare_classname, $prefix) && ($compare_classname != $prefix)) {
-				$propertyNames = array_merge($propertyNames, mfNamesFromClass($classname, ltrim($prefix)));
+			// Check if $classname is a valid property classname for $prefix.
+			if (mb_substr($classname, 0, mb_strlen($prefix)) == $prefix && $classname != $prefix) {
+				$propertyName = mb_substr($classname, mb_strlen($prefix));
+				$propertyNames[$propertyName][] = $prefix;
 			}
 		}
 	}
+	
+	foreach ($propertyNames as $property => $prefixes) {
+		$propertyNames[$property] = array_unique($prefixes);
+	}
 
 	return $propertyNames;
 }
@@ -663,7 +668,9 @@ public function parseH(\DOMElement $e) {
 			// If result was already parsed, skip it
 			if (null === $result)
 				continue;
-
+			
+			// In most cases, the value attribute of the nested microformat should be the p- parsed value of the elemnt.
+			// The only times this is different is when the microformat is nested under certain prefixes, which are handled below.
 			$result['value'] = $this->parseP($subMF);
 
 			// Does this µf have any property names other than h-*?
@@ -671,8 +678,19 @@ public function parseH(\DOMElement $e) {
 
 			if (!empty($properties)) {
 				// Yes! It’s a nested property µf
-				foreach ($properties as $property) {
-					$return[$property][] = $result;
+				foreach ($properties as $property => $prefixes) {
+					// Note: handling microformat nesting under multiple conflicting prefixes is not currently specified by the mf2 parsing spec.
+					$prefixSpecificResult = $result;
+					if (in_array('p-', $prefixes)) {
+						$prefixSpecificResult['value'] = $prefixSpecificResult['properties']['name'][0];
+					} elseif (in_array('e-', $prefixes)) {
+						$eParsedResult = $this->parseE($subMF);
+						$prefixSpecificResult['html'] = $eParsedResult['html'];
+						$prefixSpecificResult['value'] = $eParsedResult['value'];
+					} elseif (in_array('u-', $prefixes)) {
+						$prefixSpecificResult['value'] = $this->parseU($subMF);
+					}
+					$return[$property][] = $prefixSpecificResult;
 				}
 			} else {
 				// No, it’s a child µf
diff --git a/tests/Mf2/CombinedMicroformatsTest.php b/tests/Mf2/CombinedMicroformatsTest.php
index a50f54d..de63fe1 100644
--- a/tests/Mf2/CombinedMicroformatsTest.php
+++ b/tests/Mf2/CombinedMicroformatsTest.php
@@ -45,7 +45,7 @@ public function testHEventLocationHCard() {
 			"start": ["2012-06-30"],
 			"end": ["2012-07-01"],
 			"location": [{
-				"value": "Geoloqi, 920 SW 3rd Ave. Suite 400, Portland, OR",
+				"value": "Geoloqi",
 				"type": ["h-card"],
 				"properties": {
 					"name": ["Geoloqi"],
@@ -220,4 +220,33 @@ public function testNestedMicroformatUnderMultipleProperties() {
 		$this->assertCount(1, $mf['items'][0]['properties']['like-of']);
 		$this->assertCount(1, $mf['items'][0]['properties']['in-reply-to']);
 	}
+	
+	/**
+	 * Test microformats nested under e-* property classnames retain html: key in structure
+	 * 
+	 * @see https://github.com/indieweb/php-mf2/issues/64
+	 */
+	public function testMicroformatsNestedUnderEPropertyClassnamesRetainHtmlKey() {
+		$input = '<div class="h-entry"><div class="h-card e-content"><p>Hello</p></div></div>';
+		$mf = Mf2\parse($input);
+		
+		$this->assertEquals($mf['items'][0]['properties']['content'][0]['html'], '<p>Hello</p>');
+	}
+	
+	/**
+	 * Test microformats nested under u-* property classnames derive value: key from parsing as u-*
+	 */
+	public function testMicroformatsNestedUnderUPropertyClassnamesDeriveValueCorrectly() {
+		$input = '<div class="h-entry"><img class="u-url h-card" alt="This should not be the value" src="This should be the value" /></div>';
+		$mf = Mf2\parse($input);
+		
+		$this->assertEquals($mf['items'][0]['properties']['url'][0]['value'], 'This should be the value');
+	}
+	
+	public function testMicroformatsNestedUnderPPropertyClassnamesDeriveValueFromFirstPName() {
+		$input = '<div class="h-entry"><div class="p-author h-card">This post was written by <span class="p-name">Zoe</span>.</div></div>';
+		$mf = Mf2\parse($input);
+		
+		$this->assertEquals($mf['items'][0]['properties']['author'][0]['value'], 'Zoe');
+	}
 }
diff --git a/tests/Mf2/ParserTest.php b/tests/Mf2/ParserTest.php
index 094f5d8..54a229d 100644
--- a/tests/Mf2/ParserTest.php
+++ b/tests/Mf2/ParserTest.php
@@ -48,8 +48,8 @@ public function testMicroformatStripsPrefixFromPropertyClassname() {
 	}
 
 	public function testNestedMicroformatPropertyNameWorks() {
-		$expected = array('location', 'author');
-		$test = 'someclass p-location someotherclass u-author';
+		$expected = array('location' => array('p-'), 'author' => array('u-', 'p-'));
+		$test = 'someclass p-location someotherclass u-author p-author';
 		$actual = Mf2\nestedMfPropertyNamesFromClass($test);
 
 		$this->assertEquals($expected, $actual);