Merge pull request #191 from gRegorLove/issue190

Fix implied photo parsing

Author: aaronpk

Mf2/Parser.php

@@ -1078,7 +1078,7 @@ public function parseH(\DOMElement $e, $is_backcompat = false, $has_nested_mf =
 			$photo = $this->parseImpliedPhoto($e);
 
 			if ($photo !== false) {
-				$return['photo'][] = $this->resolveUrl($photo);
+				$return['photo'][] = $photo;
 			}
 
 		}
@@ -1155,38 +1155,37 @@ public function parseH(\DOMElement $e, $is_backcompat = false, $has_nested_mf =
 	 */
 	public function parseImpliedPhoto(\DOMElement $e) {
 
+		// img.h-x[src]
 		if ($e->tagName == 'img') {
-			return $e->getAttribute('src');
+			return $this->resolveUrl($e->getAttribute('src'));
 		}
 
+		// object.h-x[data]
 		if ($e->tagName == 'object' && $e->hasAttribute('data')) {
-			return $e->getAttribute('data');
+			return $this->resolveUrl($e->getAttribute('data'));
 		}
 
 		$xpaths = array(
-			'./img',
-			'./object',
-			'./*[not(contains(concat(" ", @class), " h-"))]/img[count(preceding-sibling::img)+count(following-sibling::img)=0]',
-			'./*[not(contains(concat(" ", @class), " h-"))]/object[count(preceding-sibling::object)+count(following-sibling::object)=0]',
+			// .h-x>img[src]:only-of-type:not[.h-*]
+			'./img[not(contains(concat(" ", @class), " h-")) and count(../img) = 1 and @src]',
+			// .h-x>object[data]:only-of-type:not[.h-*]
+			'./object[not(contains(concat(" ", @class), " h-")) and count(../object) = 1 and @data]',
+			// .h-x>:only-child:not[.h-*]>img[src]:only-of-type:not[.h-*]
+			'./*[not(contains(concat(" ", @class), " h-")) and count(../*) = 1 and count(img) = 1]/img[not(contains(concat(" ", @class), " h-")) and @src]',
+			// .h-x>:only-child:not[.h-*]>object[data]:only-of-type:not[.h-*]
+			'./*[not(contains(concat(" ", @class), " h-")) and count(../*) = 1 and count(object) = 1]/object[not(contains(concat(" ", @class), " h-")) and @data]',
 		);
 
 		foreach ($xpaths as $path) {
 			$els = $this->xpath->query($path, $e);
 
-			if ($els->length == 1) {
+			if ($els !== false && $els->length === 1) {
 				$el = $els->item(0);
-				$hClasses = mfNamesFromElement($el, 'h-');
-
-				// no nested h-
-				if (empty($hClasses)) {
-
-					if ($el->tagName == 'img') {
-						return $el->getAttribute('src');
-					} else if ($el->tagName == 'object' && $el->hasAttribute('data')) {
-						return $el->getAttribute('data');
-					}
-
-				} // no nested h-
+				if ($el->tagName == 'img') {
+					return $this->resolveUrl($el->getAttribute('src'));
+				} else if ($el->tagName == 'object') {
+					return $this->resolveUrl($el->getAttribute('data'));
+				}
 			}
 		}
 

tests/Mf2/ParseImpliedTest.php

@@ -293,6 +293,26 @@ public function testIgnoredPhotoObjectInNestedH() {
 		$this->assertArrayNotHasKey('photo', $result['items'][0]['properties']);
 	}
 
+	/**
+	 * @see https://github.com/indieweb/php-mf2/issues/190
+	 */
+	public function testIgnoredMultiChildrenWithNestedPhotoImg() {
+		$input = '<div class="h-card"> <a href="https://example.com"><img src="https://example.com/photo.jpg"></a> <span class="p-name"><a href="/User:Example.com">Max Mustermann</a></span> </div>';
+		$result = Mf2\parse($input);
+
+		$this->assertArrayNotHasKey('photo', $result['items'][0]['properties']);
+	}
+
+	/**
+	 * @see https://github.com/indieweb/php-mf2/issues/190
+	 */
+	public function testIgnoredMultiChildrenWithNestedPhotoObject() {
+		$input = '<div class="h-card"> <a href="https://example.com"><object data="https://example.com/photo.jpg"></object></a> <span class="p-name"><a href="/User:Example.com">Max Mustermann</a></span> </div>';
+		$result = Mf2\parse($input);
+
+		$this->assertArrayNotHasKey('photo', $result['items'][0]['properties']);
+	}
+
 	/**
 	 * Imply properties only on explicit h-x class name root microformat element (no backcompat roots)
 	 * @see http://microformats.org/wiki/microformats2-parsing#parsing_for_implied_properties