From 706540a6330d9e9dc490aa50b02e76116a103d64 Mon Sep 17 00:00:00 2001
From: Jacopo <beschi.jacopo@gmail.com>
Date: Wed, 19 Jun 2024 09:52:59 +0200
Subject: [PATCH] Port last changes from HEY

---
 lib/homographic_spoofing/detector/idn.rb | 14 +++++++++++++-
 test/detector/email_address_test.rb      | 11 +++++++++--
 test/detector/idn_test.rb                |  3 ++-
 test/detector/local_test.rb              |  2 +-
 test/detector/quoted_string_test.rb      |  2 +-
 5 files changed, 26 insertions(+), 6 deletions(-)

diff --git a/lib/homographic_spoofing/detector/idn.rb b/lib/homographic_spoofing/detector/idn.rb
index eed39eb..f67e6bd 100644
--- a/lib/homographic_spoofing/detector/idn.rb
+++ b/lib/homographic_spoofing/detector/idn.rb
@@ -61,6 +61,18 @@ def contexts
     end
 
     def public_suffix
-      @public_suffix ||= PublicSuffix.parse(domain, ignore_private: true)
+      @public_suffix ||= icann_domain || non_icann_domain
+    end
+
+    def icann_domain
+      PublicSuffix.parse(domain, ignore_private: true) if PublicSuffix.valid?(domain)
+    end
+
+    def non_icann_domain
+      if PublicSuffix::List.default.find(domain, default: nil, ignore_private: true).present?
+        PublicSuffix::Domain.new(domain)
+      else
+        raise PublicSuffix::DomainInvalid
+      end
     end
 end
diff --git a/test/detector/email_address_test.rb b/test/detector/email_address_test.rb
index 9fba5d8..981b517 100644
--- a/test/detector/email_address_test.rb
+++ b/test/detector/email_address_test.rb
@@ -7,11 +7,11 @@ class HomographicSpoofing::Detector::EmailAddressTest < ActiveSupport::TestCase
   end
 
   test "detect local part" do
-    assert_attack "tᴡitter@twitter.com"
+    assert_attack "t\u{1D21}itter@twitter.com" # tᴡitter@twitter.com
   end
 
   test "detect idn" do
-    assert_attack "jacopo@tᴡitter.com"
+    assert_attack "jacopo@t\u{1D21}itter.com" # jacopo@tᴡitter.com
   end
 
   test "ignore nil email address parts" do
@@ -23,6 +23,13 @@ class HomographicSpoofing::Detector::EmailAddressTest < ActiveSupport::TestCase
     assert_safe "jacopo"
   end
 
+  test "private domains are allowed but idn attacks are detected" do
+    assert_safe "someone.from@notaires.fr"
+    assert_attack "someone.from@n\u{1D0F}taires.fr" # someone.from@nᴏtaires.fr
+    assert_safe "someone.from@blogspot.com"
+    assert_attack "someone.from@bl\u{1D0F}gspot.com" # someone.from@blᴏgspot.com
+  end
+
   private
     def assert_attack(email_address)
       assert HomographicSpoofing::Detector::EmailAddress.detected?(email_address), "No attack detected for #{email_address}."
diff --git a/test/detector/idn_test.rb b/test/detector/idn_test.rb
index 61b5352..6a8f14b 100644
--- a/test/detector/idn_test.rb
+++ b/test/detector/idn_test.rb
@@ -12,6 +12,7 @@ class HomographicSpoofing::Detector::IdnTest < ActiveSupport::TestCase
   test "Valid mixed scripts" do
     assert_safe("おaかbがcキdギeクf.co.jp") # Latin, Hiragana, Katakana
     assert_safe("aㄉbㄊcde夕f.com")         # Latin, Bopomofo, Han
+    assert_safe("example.বাংলা")            # Latin, Bengali
   end
 
   test "Invalid mixed scripts" do
@@ -265,8 +266,8 @@ class HomographicSpoofing::Detector::IdnTest < ActiveSupport::TestCase
   end
 
   test "PublixSuffix parsing errors" do
-    assert_attack("nic.bd", reason: "invalid_domain")
     assert_attack("  ", reason: "invalid_domain")
+    assert_attack("example|.com", reason: "disallowed_characters")
   end
 
   private
diff --git a/test/detector/local_test.rb b/test/detector/local_test.rb
index ade2b85..e8ebc90 100644
--- a/test/detector/local_test.rb
+++ b/test/detector/local_test.rb
@@ -37,7 +37,7 @@ class HomographicSpoofing::Detector::LocalTest < ActiveSupport::TestCase
 
   test "Unicode NFKC format" do
     # Invalid unicode
-    assert_attack("text".force_encoding("ISO-8859-1"), reason: "invalid_unicode")
+    assert_attack("text".encode("ISO-8859-1"), reason: "invalid_unicode")
     assert_attack("1𝟤3rf".unicode_normalize(:nfc), reason: "nfkc")
     assert_safe("scope")
   end
diff --git a/test/detector/quoted_string_test.rb b/test/detector/quoted_string_test.rb
index 8b1d016..cf98769 100644
--- a/test/detector/quoted_string_test.rb
+++ b/test/detector/quoted_string_test.rb
@@ -3,7 +3,7 @@
 class HomographicSpoofing::Detector::QuotedStringTest < ActiveSupport::TestCase
   test "Unicode NFC format" do
     # Invalid unicode
-    assert_attack("text".force_encoding("ISO-8859-1"), reason: "invalid_unicode")
+    assert_attack("text".dup.force_encoding("ISO-8859-1"), reason: "invalid_unicode")
     assert_attack("1A\u030A3rf", reason: "nfc")
     assert_safe("scope")
   end