Merge pull request #820 from gjtorikian/regexpiry

Remove unnecessary regexp
gjtorikian · May 7, 2024 · f9ca010 · f9ca010
1 parent 1c8b2c1
commit f9ca010
Show file tree

Hide file tree

Showing 3 changed files with 50 additions and 31 deletions.
diff --git a/lib/html_proofer/attribute/url.rb b/lib/html_proofer/attribute/url.rb
@@ -220,11 +220,24 @@ def without_hash
         @url.to_s.sub(/##{hash}/, "")
       end
 
-      # catch any obvious issues, like strings in port numbers
+      # catch any obvious issues
       private def clean_url!
-        return if @url =~ /^([!#{Regexp.last_match(0)}-;=?-\[\]_a-z~]|%[0-9a-fA-F]{2})+$/
-
-        @url = Addressable::URI.parse(@url).normalize.to_s
+        parsed_url = Addressable::URI.parse(@url)
+        url = if parsed_url.scheme.nil?
+          parsed_url
+        else
+          parsed_url.normalize
+        end.to_s
+
+        # normalize strips this off, which causes issues with cache
+        @url = if @url.end_with?("/") && !url.end_with?("/")
+          "#{url}/"
+        elsif !@url.end_with?("/") && url.end_with?("/")
+          url.chop
+        else
+          url
+        end
+      rescue Addressable::URI::InvalidURIError # rubocop:disable Lint/SuppressedException; error will be reported at check time
       end
 
       private def swap_urls!

diff --git a/spec/html-proofer/fixtures/links/unicode_domain.html b/spec/html-proofer/fixtures/links/unicode_domain.html
@@ -2,6 +2,6 @@
 
 <body>
 
-<p><a href="https://αναπτυξηεφαρμογων.gr">A real link</a></p>
+<p><a href="https://granö.fi">A real link</a></p>
 
 </html>
diff --git a/...l-proofer/fixtures/vcr_cassettes/links/unicode_domain_html_log_level_error_type_file_.yml b/...l-proofer/fixtures/vcr_cassettes/links/unicode_domain_html_log_level_error_type_file_.yml