From 27beed6a16c7b188f3c2db5498d0bc8542a7e32b Mon Sep 17 00:00:00 2001
From: Joe Haines <hello@joehaines.co.uk>
Date: Mon, 11 Mar 2024 12:45:48 +0000
Subject: [PATCH] Fix Unicode encoding issues with detailed_message

---
 CHANGELOG.md          |  5 +++
 lib/bugsnag/report.rb |  9 ++++-
 spec/report_spec.rb   | 92 ++++++++++++++++++++++++++++++++++++-------
 3 files changed, 90 insertions(+), 16 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 68d4473f..746122f7 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,6 +1,11 @@
 Changelog
 =========
 
+## TBD
+
+* Fix Unicode encoding issues when using `Exception#detailed_message` (Ruby 3.2+)
+  | [#817](https://github.com/bugsnag/bugsnag-ruby/pull/817)
+
 ## v6.26.3 (24 January 2024)
 
 * Handle mailto links in `Cleaner#clean_url`
diff --git a/lib/bugsnag/report.rb b/lib/bugsnag/report.rb
index c2958d39..92d1b56c 100644
--- a/lib/bugsnag/report.rb
+++ b/lib/bugsnag/report.rb
@@ -466,8 +466,15 @@ def error_message(exception, class_name)
           exception.detailed_message
         end
 
+      # the string returned by 'detailed_message' defaults to 'ASCII_8BIT' but
+      # is actually UTF-8 encoded; we can't convert the encoding normally as its
+      # internal encoding doesn't match its actual encoding
+      message.force_encoding(::Encoding::UTF_8) if message.encoding == ::Encoding::ASCII_8BIT
+
       # remove the class name to be consistent with Exception#message
-      message.sub(" (#{class_name})", '')
+      message.sub!(" (#{class_name})".encode(message.encoding), "") rescue nil
+
+      message
     end
 
     def generate_raw_exceptions(exception)
diff --git a/spec/report_spec.rb b/spec/report_spec.rb
index 3756a741..a6c3bd5a 100644
--- a/spec/report_spec.rb
+++ b/spec/report_spec.rb
@@ -40,6 +40,17 @@ def detailed_message
   end
 end
 
+class ExceptionWithDetailedMessageReturningEncodedString < Exception
+  def initialize(message, encoding)
+    super(message)
+    @encoding = encoding
+  end
+
+  def detailed_message
+    "abc #{self} xyz".encode(@encoding)
+  end
+end
+
 shared_examples "Report or Event tests" do |class_to_test|
   context "metadata" do
     include_examples(
@@ -1465,24 +1476,75 @@ def detailed_message
     }
   end
 
-  it "uses Exception#detailed_message if available" do
-    Bugsnag.notify(ExceptionWithDetailedMessage.new("some message"))
+  context "#detailed_message" do
+    it "uses Exception#detailed_message if available" do
+      Bugsnag.notify(ExceptionWithDetailedMessage.new("some message"))
 
-    expect(Bugsnag).to have_sent_notification{ |payload, headers|
-      exception = get_exception_from_payload(payload)
-      expect(exception["errorClass"]).to eq("ExceptionWithDetailedMessage")
-      expect(exception["message"]).to eq("some message with some extra detail")
-    }
-  end
+      expect(Bugsnag).to have_sent_notification{ |payload, headers|
+        exception = get_exception_from_payload(payload)
+        expect(exception["errorClass"]).to eq("ExceptionWithDetailedMessage")
+        expect(exception["message"]).to eq("some message with some extra detail")
+      }
+    end
 
-  it "handles implementations of Exception#detailed_message with no 'highlight' parameter" do
-    Bugsnag.notify(ExceptionWithDetailedMessageButNoHighlight.new("some message"))
+    it "handles implementations of Exception#detailed_message with no 'highlight' parameter" do
+      Bugsnag.notify(ExceptionWithDetailedMessageButNoHighlight.new("some message"))
 
-    expect(Bugsnag).to have_sent_notification{ |payload, headers|
-      exception = get_exception_from_payload(payload)
-      expect(exception["errorClass"]).to eq("ExceptionWithDetailedMessageButNoHighlight")
-      expect(exception["message"]).to eq("detail about 'some message'")
-    }
+      expect(Bugsnag).to have_sent_notification{ |payload, headers|
+        exception = get_exception_from_payload(payload)
+        expect(exception["errorClass"]).to eq("ExceptionWithDetailedMessageButNoHighlight")
+        expect(exception["message"]).to eq("detail about 'some message'")
+      }
+    end
+
+    it "converts ASCII_8BIT encoding to UTF-8" do
+      Bugsnag.notify(Exception.new("大好き\n大好き"))
+
+      expect(Bugsnag).to have_sent_notification{ |payload, headers|
+        exception = get_exception_from_payload(payload)
+        expect(exception["message"]).to eq("大好き\n大好き")
+      }
+    end
+
+    it "leaves UTF-8 strings as-is" do
+      exception = ExceptionWithDetailedMessageButNoHighlight.new("Обичам те\n大好き")
+      expect(exception.detailed_message.encoding).to be(Encoding::UTF_8)
+
+      Bugsnag.notify(exception)
+
+      expect(Bugsnag).to have_sent_notification{ |payload, headers|
+        exception = get_exception_from_payload(payload)
+        expect(exception["message"]).to eq("detail about 'Обичам те\n大好き'")
+      }
+    end
+
+    it "handles UTF-16 strings" do
+      exception = ExceptionWithDetailedMessageReturningEncodedString.new("Обичам те\n大好き", Encoding::UTF_16)
+      expect(exception.detailed_message.encoding).to be(Encoding::UTF_16)
+
+      Bugsnag.notify(exception)
+
+      expect(Bugsnag).to have_sent_notification{ |payload, headers|
+        exception = get_exception_from_payload(payload)
+
+        # the exception message is converted to UTF-8 by the Cleaner
+        expect(exception["message"]).to eq("abc Обичам те\n大好き xyz")
+      }
+    end
+
+    it "handles Shift JIS strings" do
+      exception = ExceptionWithDetailedMessageReturningEncodedString.new("大好き\n大好き", Encoding::Shift_JIS)
+      expect(exception.detailed_message.encoding).to be(Encoding::Shift_JIS)
+
+      Bugsnag.notify(exception)
+
+      expect(Bugsnag).to have_sent_notification{ |payload, headers|
+        exception = get_exception_from_payload(payload)
+
+        # the exception message is converted to UTF-8 by the Cleaner
+        expect(exception["message"]).to eq("abc 大好き\n大好き xyz")
+      }
+    end
   end
 
   it "supports unix-style paths in backtraces" do