From bbe6aba7ff9562844d97b6c6eecec8f43c2a5f2c Mon Sep 17 00:00:00 2001
From: Aline Marques <alinem_oliveira@yahoo.com>
Date: Thu, 29 Oct 2026 17:53:54 -0300
Subject: [PATCH 1/3] AO3-7637 Blank out invalid hrefs that break eBook
 downloads

---
 config/config.yml                                    |  2 +-
 .../gem-plugin_config/sanitizer_config.rb            |  9 +++++++++
 lib/html_cleaner.rb                                  |  3 ++-
 spec/lib/html_cleaner_spec.rb                        | 12 ++++++++++++
 4 files changed, 24 insertions(+), 2 deletions(-)

diff --git a/config/config.yml b/config/config.yml
index 4cbdc38edfe..9d565b6481c 100644
--- a/config/config.yml
+++ b/config/config.yml
@@ -334,7 +334,7 @@ COMMENT_MODERATION_THRESHOLD: 10
 EDITED_COMMENT_SPAM_CHECK_THRESHOLD: 10
 
 # SANITIZER VERSION
-SANITIZER_VERSION: 4
+SANITIZER_VERSION: 5
 
 # parameters that must be natural integers and their default values
 NONZERO_INTEGER_PARAMETERS:
diff --git a/config/initializers/gem-plugin_config/sanitizer_config.rb b/config/initializers/gem-plugin_config/sanitizer_config.rb
index 65e8be78caa..9fc6b2f4463 100644
--- a/config/initializers/gem-plugin_config/sanitizer_config.rb
+++ b/config/initializers/gem-plugin_config/sanitizer_config.rb
@@ -67,5 +67,14 @@ module Config
     rescue URI::InvalidURIError
       # do nothing, the sanitizer will handle it
     end
+
+    # On a elements, blank out hrefs that aren't valid URLs
+    INVALID_HREF_TRANSFORMER = lambda do |env|
+      return unless env[:node_name] == "a" && env[:node]["href"]
+      
+      URI.parse(env[:node]["href"])
+    rescue URI::InvalidURIError
+      env[:node]["href"] = ""
+    end
   end
 end
diff --git a/lib/html_cleaner.rb b/lib/html_cleaner.rb
index 249a0775dc2..20962acf26c 100644
--- a/lib/html_cleaner.rb
+++ b/lib/html_cleaner.rb
@@ -59,7 +59,8 @@ def sanitize_value(field, value)
       # We're allowing users to use HTML in this field
       transformers = [
         Sanitize::Config::OPEN_ATTRIBUTE_TRANSFORMER,
-        Sanitize::Config::RELATIVE_IMAGE_PATH_TRANSFORMER
+        Sanitize::Config::RELATIVE_IMAGE_PATH_TRANSFORMER,
+        Sanitize::Config::INVALID_HREF_TRANSFORMER
       ]
       if ArchiveConfig.FIELDS_ALLOWING_MEDIA_EMBEDS.include?(field.to_s)
         transformers << OtwSanitize::EmbedSanitizer.transformer
diff --git a/spec/lib/html_cleaner_spec.rb b/spec/lib/html_cleaner_spec.rb
index 4aec90672c1..0a674cda6c7 100644
--- a/spec/lib/html_cleaner_spec.rb
+++ b/spec/lib/html_cleaner_spec.rb
@@ -444,6 +444,18 @@ def one_cell_table(content)
           end
         end
 
+        context "when given an <a> tag with an invalid href" do
+          it "should clean up an invalid href while preserving the link text" do
+            result = sanitize_value(field, "<a href='https://[https://test.archiveofourown.org/users/testy]' rel='nofollow'>testy</a>")
+            expect(result).to eq("<p><a href=\"\" rel=\"nofollow\">testy</a></p>")
+          end
+
+          it "should not clean up a valid href" do 
+            result = sanitize_value(field, "<a href='https://test.archiveofourown.org/users/testy' rel='nofollow'>testy</a>")
+            expect(result).to eq("<p><a href=\"https://test.archiveofourown.org/users/testy\" rel=\"nofollow\">testy</a></p>")
+          end
+        end
+
         # These are from https://github.com/rgrove/sanitize/commit/a11498de9e283cd457b35ee252983662f7452aa9
         it 'should not preserve the content of removed `math` elements' do
           content = sanitize_value(field, '<math>hello! <script>alert(0)</script></math>')

From c843b77c9b400f22aeb374aa8a04b0f8a2215f59 Mon Sep 17 00:00:00 2001
From: Aline Marques <alinem_oliveira@yahoo.com>
Date: Thu, 29 Oct 2026 18:13:13 -0300
Subject: [PATCH 2/3] AO3-7637 Only blank out hrefs with unclosed brackets in
 the host

---
 config/initializers/gem-plugin_config/sanitizer_config.rb | 8 ++++----
 spec/lib/html_cleaner_spec.rb                             | 7 ++++++-
 2 files changed, 10 insertions(+), 5 deletions(-)

diff --git a/config/initializers/gem-plugin_config/sanitizer_config.rb b/config/initializers/gem-plugin_config/sanitizer_config.rb
index 9fc6b2f4463..7b0b21a5ad2 100644
--- a/config/initializers/gem-plugin_config/sanitizer_config.rb
+++ b/config/initializers/gem-plugin_config/sanitizer_config.rb
@@ -71,10 +71,10 @@ module Config
     # On a elements, blank out hrefs that aren't valid URLs
     INVALID_HREF_TRANSFORMER = lambda do |env|
       return unless env[:node_name] == "a" && env[:node]["href"]
-      
-      URI.parse(env[:node]["href"])
-    rescue URI::InvalidURIError
-      env[:node]["href"] = ""
+
+      host = Addressable::URI.parse(env[:node]["href"]).host
+
+      env[:node]["href"] = "" if host&.start_with?("[") && !host&.end_with?("]")
     end
   end
 end
diff --git a/spec/lib/html_cleaner_spec.rb b/spec/lib/html_cleaner_spec.rb
index 0a674cda6c7..9ff53f1c402 100644
--- a/spec/lib/html_cleaner_spec.rb
+++ b/spec/lib/html_cleaner_spec.rb
@@ -450,10 +450,15 @@ def one_cell_table(content)
             expect(result).to eq("<p><a href=\"\" rel=\"nofollow\">testy</a></p>")
           end
 
-          it "should not clean up a valid href" do 
+          it "should not clean up a valid href" do
             result = sanitize_value(field, "<a href='https://test.archiveofourown.org/users/testy' rel='nofollow'>testy</a>")
             expect(result).to eq("<p><a href=\"https://test.archiveofourown.org/users/testy\" rel=\"nofollow\">testy</a></p>")
           end
+
+          it "should not clean up a valid href with non-ASCII characters" do
+            result = sanitize_value(field, "<a href='https://test.archiveofourown.org/users/café' rel='nofollow'>testy</a>")
+            expect(result).to eq("<p><a href=\"https://test.archiveofourown.org/users/café\" rel=\"nofollow\">testy</a></p>")
+          end
         end
 
         # These are from https://github.com/rgrove/sanitize/commit/a11498de9e283cd457b35ee252983662f7452aa9

From 39172b7f6676cfce00dd75d060e2bc5f5f9b050b Mon Sep 17 00:00:00 2001
From: Aline Marques <alinem_oliveira@yahoo.com>
Date: Thu, 29 Oct 2026 18:20:56 -0300
Subject: [PATCH 3/3] AO3-7637 Handle hrefs that Addressable can't parse

---
 config/initializers/gem-plugin_config/sanitizer_config.rb | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/config/initializers/gem-plugin_config/sanitizer_config.rb b/config/initializers/gem-plugin_config/sanitizer_config.rb
index 7b0b21a5ad2..c0b12c76b18 100644
--- a/config/initializers/gem-plugin_config/sanitizer_config.rb
+++ b/config/initializers/gem-plugin_config/sanitizer_config.rb
@@ -75,6 +75,8 @@ module Config
       host = Addressable::URI.parse(env[:node]["href"]).host
 
       env[:node]["href"] = "" if host&.start_with?("[") && !host&.end_with?("]")
+    rescue Addressable::URI::InvalidURIError
+      env[:node]["href"] = ""
     end
   end
 end