Improve handling of escaped and invalid URLs

jgarber623 · Aug 21, 2022 · b0d6c75 · b0d6c75
1 parent a6197ab
commit b0d6c75
Show file tree

Hide file tree

Showing 2 changed files with 11 additions and 13 deletions.
diff --git a/lib/nokogiri/html_ext/document.rb b/lib/nokogiri/html_ext/document.rb
@@ -69,13 +69,15 @@ def base_href=(url)
       #
       # @return [String]
       def resolve_relative_url(url)
-        unescape(
-          uri_parser.join(*[document.url.strip, base_href, unescape(url)].compact.map { |u| escape(u) })
+        url_str = url.to_s
+
+        uri_parser.unescape(
+          uri_parser.join(*[document.url.strip, base_href, url_str].compact.map { |u| uri_parser.escape(u) })
                     .normalize
                     .to_s
         )
       rescue URI::InvalidComponentError, URI::InvalidURIError
-        unescape(url)
+        url
       end
 
       # Convert the document's relative URLs to absolute URLs.
@@ -99,10 +101,6 @@ def resolve_relative_urls!
 
       private
 
-      def escape(url)
-        uri_parser.escape(url.to_s)
-      end
-
       def resolve_relative_urls_for(attributes_map)
         attributes_map.each do |attribute, names|
           xpaths = names.map { |name| "//#{name}[@#{attribute}]" }
@@ -113,10 +111,6 @@ def resolve_relative_urls_for(attributes_map)
         end
       end
 
-      def unescape(url)
-        uri_parser.unescape(url.to_s)
-      end
-
       def uri_parser
         @uri_parser ||= URI::DEFAULT_PARSER
       end

diff --git a/spec/lib/nokogiri/html_ext/document_spec.rb b/spec/lib/nokogiri/html_ext/document_spec.rb
@@ -63,6 +63,8 @@
         <body>
           <a href="/home">Home</a>
           <img srcset="../foo.png 480w, ../bar.png 720w, /biz.jpg">
+          <img src="/commons/thumb/9/96/H%C3%A5kon-Wium-Lie-2009-03.jpg/215px-H%C3%A5kon-Wium-Lie-2009-03.jpg">
+          <a href="/foo%2epdf">Relative escaped PDF</a>
           <a href="mailto:email%40jgarber%2eexample">Valid escaped electronic mail</a>
           <a href="mailto:email_at_jgarber%2eexample">Invalid electronic mail</a>
         </body>
@@ -81,8 +83,10 @@
         <body>
           <a href="https://jgarber.example/home">Home</a>
           <img srcset="https://jgarber.example/foo/foo.png 480w, https://jgarber.example/foo/bar.png 720w, https://jgarber.example/biz.jpg">
-          <a href="mailto:email@jgarber.example">Valid escaped electronic mail</a>
-          <a href="mailto:email_at_jgarber.example">Invalid electronic mail</a>
+          <img src="https://jgarber.example/commons/thumb/9/96/H%C3%A5kon-Wium-Lie-2009-03.jpg/215px-H%C3%A5kon-Wium-Lie-2009-03.jpg">
+          <a href="https://jgarber.example/foo%2epdf">Relative escaped PDF</a>
+          <a href="mailto:email%40jgarber%2eexample">Valid escaped electronic mail</a>
+          <a href="mailto:email_at_jgarber%2eexample">Invalid electronic mail</a>
         </body>
         </html>
       HTML