Better handling for link rel dns-prefetch and add preconnect support (#536)

preconnect is only DNS checked.

This is allowed even in the Resource Hints Editor's Draft
https://w3c.github.io/resource-hints/#preconnect

linkcheck/htmlutil/linkparse.py

@@ -157,7 +157,8 @@ class LinkFinder:
                 base = self.base_ref
             # note: value can be None
             value = attrs.get(attr)
-            if tag == 'link' and attrs.get('rel') == 'dns-prefetch':
+            rel = attrs.get('rel', '').lower()
+            if tag == 'link' and ('dns-prefetch' in rel or 'preconnect' in rel):
                 if ':' in value:
                     value = value.split(':', 1)[1]
                 value = 'dns:' + value.rstrip('/')

tests/test_linkparser.py

@@ -60,6 +60,30 @@ class TestLinkparser(unittest.TestCase):
         url = " alink "
         self._test_one_link(content % url, url)
 
+    def test_rel_parsing(self):
+        # Test <link rel> parsing.
+        content = '<link rel="%s" href="%s">'
+        rel = "dns-prefetch"
+        url = "https://alink"
+        expected = "dns://alink"
+        self._test_one_link(content % (rel, url), expected)
+        url = "//alink/"
+        self._test_one_link(content % (rel, url), expected)
+        rel = "preconnect"
+        url = "https://alink"
+        self._test_one_link(content % (rel, url), expected)
+        rel = "dns-prefetch preconnect"
+        url = "https://alink"
+        self._test_one_link(content % (rel, url), expected)
+
+    def test_link_without_rel_parsing(self):
+        # <link> tags without rel attr should not raise TypeError.
+        content = '<link href="%s">'
+        url = "https://alink"
+        expected = "https://alink"
+        # Dummy test, we just have to make sure no error was raised.
+        self._test_one_link(content % url, expected)
+
     def test_img_srcset_parsing(self):
         content = '<img srcset="%s 1x">'
         url = "imagesmall.jpg"