Merge pull request #162 from Zegnat/rel-parsing

Improve rel parsing
microformats · Mar 25, 2018 · 2c6677d · 2c6677d
2 parents 743d979 + 64e7966
commit 2c6677d
Show file tree

Hide file tree

Showing 2 changed files with 74 additions and 14 deletions.
diff --git a/Mf2/Parser.php b/Mf2/Parser.php
@@ -1250,16 +1250,15 @@ public function parseRelsAndAlternates() {
 
 		// Iterate through all a, area and link elements with rel attributes
 		foreach ($this->xpath->query('//a[@rel and @href] | //link[@rel and @href] | //area[@rel and @href]') as $hyperlink) {
-			if ($hyperlink->getAttribute('rel') == '') {
+			// Parse the set of rels for the current link
+			$linkRels = array_unique(array_filter(preg_split('/[\t\n\f\r ]/', $hyperlink->getAttribute('rel'))));
+			if (count($linkRels) === 0) {
 				continue;
 			}
 
 			// Resolve the href
 			$href = $this->resolveUrl($hyperlink->getAttribute('href'));
 
-			// Split up the rel into space-separated values
-			$linkRels = array_filter(explode(' ', $hyperlink->getAttribute('rel')));
-
 			$rel_attributes = array();
 
 			if ($hyperlink->hasAttribute('media')) {
@@ -1278,8 +1277,8 @@ public function parseRelsAndAlternates() {
 				$rel_attributes['type'] = $hyperlink->getAttribute('type');
 			}
 
-			if ($hyperlink->nodeValue) {
-				$rel_attributes['text'] = $hyperlink->nodeValue;
+			if (strlen($hyperlink->textContent) > 0) {
+				$rel_attributes['text'] = $hyperlink->textContent;
 			}
 
 			if ($this->enableAlternates) {
@@ -1296,16 +1295,34 @@ public function parseRelsAndAlternates() {
 			}
 
 			foreach ($linkRels as $rel) {
-				$rels[$rel][] = $href;
+				if (!array_key_exists($rel, $rels)) {
+					$rels[$rel] = array($href);
+				} elseif (!in_array($href, $rels[$rel])) {
+					$rels[$rel][] = $href;
+				}
 			}
 
-			if (!in_array($href, $rel_urls)) {
-				$rel_urls[$href] = array_merge(
-					$rel_attributes, 
-					array('rels' => $linkRels)
-				);
+			if (!array_key_exists($href, $rel_urls)) {
+				$rel_urls[$href] = array('rels' => array());
 			}
 
+			// Add the attributes collected only if they were not already set
+			$rel_urls[$href] = array_merge(
+				$rel_attributes,
+				$rel_urls[$href]
+			);
+
+			// Merge current rels with those already set
+			$rel_urls[$href]['rels'] = array_merge(
+				$rel_urls[$href]['rels'],
+				$linkRels
+			);
+		}
+
+		// Alphabetically sort the rels arrays after removing duplicates
+		foreach ($rel_urls as $href => $object) {
+			$rel_urls[$href]['rels'] = array_unique($rel_urls[$href]['rels']);
+			sort($rel_urls[$href]['rels']);
 		}
 
 		if (empty($rels) and $this->jsonMode) {
@@ -1314,8 +1331,8 @@ public function parseRelsAndAlternates() {
 
 		if (empty($rel_urls) and $this->jsonMode) {
 			$rel_urls = new stdClass();
-		}		
-		
+		}
+
 		return array($rels, $rel_urls, $alternates);
 	}
 

diff --git a/tests/Mf2/RelTest.php b/tests/Mf2/RelTest.php
@@ -176,4 +176,47 @@ public function testRelURLs() {
     $this->assertArrayHasKey('rels', $output['rel-urls']['http://example.com/articles.atom']);
   }
 
+  /**
+   * @see https://github.com/microformats/microformats2-parsing/issues/29
+   * @see https://github.com/microformats/microformats2-parsing/issues/30
+   */
+  public function testRelURLsRelsUniqueAndSorted() {
+    $input = '<a href="#" rel="me bookmark"></a>
+<a href="#" rel="bookmark archived"></a>';
+    $parser = new Parser($input);
+    $output = $parser->parse();
+    $this->assertEquals($output['rel-urls']['#']['rels'], array('archived', 'bookmark', 'me'));
+  }
+
+  public function testRelURLsInfoMergesCorrectly() {
+    $input = '<a href="#" rel="a">This nodeValue</a>
+<a href="#" rel="a" hreflang="en">Not this nodeValue</a>';
+    $parser = new Parser($input);
+    $output = $parser->parse();
+    $this->assertEquals($output['rel-urls']['#']['hreflang'], 'en');
+    $this->assertArrayNotHasKey('media', $output['rel-urls']['#']);
+    $this->assertArrayNotHasKey('title', $output['rel-urls']['#']);
+    $this->assertArrayNotHasKey('type', $output['rel-urls']['#']);
+    $this->assertEquals($output['rel-urls']['#']['text'], 'This nodeValue');
+  }
+
+  public function testRelURLsNoDuplicates() {
+    $input = '<a href="#a" rel="a"></a>
+<a href="#b" rel="a"></a>
+<a href="#a" rel="a"></a>';
+    $parser = new Parser($input);
+    $output = $parser->parse();
+    $this->assertEquals($output['rels']['a'], array('#a', '#b'));
+  }
+
+  public function testRelURLsFalsyTextVSEmpty() {
+    $input = '<a href="#a" rel="a">0</a>
+<a href="#b" rel="b"></a>';
+    $parser = new Parser($input);
+    $output = $parser->parse();
+    $this->assertArrayHasKey('text', $output['rel-urls']['#a']);
+    $this->assertEquals($output['rel-urls']['#a']['text'], '0');
+    $this->assertArrayNotHasKey('text', $output['rel-urls']['#b']);
+  }
+
 }