fixes #80 improve internal URL detection

leonstafford · leonstafford · commit 6fc3070b4f79 · 2020-06-10T03:21:52.000+09:30
diff --git a/src/HTMLProcessor.php b/src/HTMLProcessor.php
@@ -654,6 +654,14 @@ public function isInternalLink( string $link ) : bool {
             return $link_host === parse_url( $domain, PHP_URL_HOST );
         }
 
+        $extension = pathinfo( $link, PATHINFO_EXTENSION );
+
+        if ( ! $extension ) {
+            if ( substr( $link, -1 ) !== '/' ) {
+                return false;
+            }
+        }
+
         // match anything without a colon, comma or space ie favicon.ico, not mailto:, viewport
         if (
             ( strpos( $link, ':' ) === false ) &&
diff --git a/tests/HTMLProcessorTest.php b/tests/HTMLProcessorTest.php
@@ -489,6 +489,20 @@ public function processHTMLProvider() {
                 $this->loadTestHTML( 'input_inline_style_processing' ),
                 $this->loadTestHTML( 'output_inline_style_processing' ),
             ],
+            'preserves non-URI meta content values' => [
+                false, // $remove_conditional_head_comments = false
+                false, // $remove_html_comments = false
+                false, // $remove_wp_links = false
+                false, // $remove_wp_meta = false
+                '', // $rewrite_rules = ''
+                'https://mynewdomain.com', // $base_url
+                '', // $selected_deployment_option = 'folder'
+                'http://localhost:4444', // $wp_site_url
+                '/tmp/', // $wp_uploads_path - temp write file during test while refactoring
+                'http://localhost:4444/',
+                $this->loadTestHTML( 'input_meta_contents' ),
+                $this->loadTestHTML( 'output_meta_contents' ),
+            ],
         ];
     }
 }
diff --git a/tests/data/HTMLProcessorTest/input_meta_contents.html b/tests/data/HTMLProcessorTest/input_meta_contents.html
@@ -0,0 +1,15 @@
+<!DOCTYPE html>
+<html class="no-js" lang="en-US">
+ <head>
+    <meta charset="UTF-8">
+    <meta name="twitter:card" content="summary">
+    <title>Static HTML Output Testing &#8211; test html processer</title>
+  </head>
+
+  <body>
+    <p>Ohai there!</p>
+
+    <p>Meta content like "summery" shouldn't be detected as a URL</p>
+ </body>
+</html>
+
diff --git a/tests/data/HTMLProcessorTest/output_meta_contents.html b/tests/data/HTMLProcessorTest/output_meta_contents.html
@@ -0,0 +1,14 @@
+<!DOCTYPE html>
+<html class="no-js" lang="en-US">
+ <head>
+    <meta charset="UTF-8">
+    <meta name="twitter:card" content="summary">
+    <title>Static HTML Output Testing &ndash; test html processer</title>
+  </head>
+
+  <body>
+    <p>Ohai there!</p>
+
+    <p>Meta content like "summery" shouldn't be detected as a URL</p>
+ </body>
+</html>
