From a93f81e20beedb38d7d726d303ce5034b598da7a Mon Sep 17 00:00:00 2001
From: Ryan Lue <hello@ryanlue.com>
Date: Thu, 9 Aug 2018 12:05:09 +0800
Subject: [PATCH] Make String#utf8_encode more robust (fixes #2176)

---
 lib/core_ext/string.rb           | 12 +++++++++---
 spec/lib/core_ext/string_spec.rb | 12 ++++++++++--
 2 files changed, 19 insertions(+), 5 deletions(-)

diff --git a/lib/core_ext/string.rb b/lib/core_ext/string.rb
index dd522da16..f8699cd52 100644
--- a/lib/core_ext/string.rb
+++ b/lib/core_ext/string.rb
@@ -474,10 +474,15 @@ class String
   end
 
   def utf8_encode!(**options)
-    return self if (encoding == Encoding::UTF_8) && valid_encoding?
+    return force_encoding('utf-8') if dup.force_encoding('utf-8').valid_encoding?
 
-    input_encoding = viable_encodings(try_first: options[:from]).first
-    return encode!('utf-8', input_encoding) if input_encoding.present?
+    viable_encodings(try_first: options[:from]).each do |e|
+      begin
+        return encode!('utf-8', e)
+      rescue Encoding::UndefinedConversionError
+        nil
+      end
+    end
 
     case options[:fallback]
     when :output_to_binary
@@ -501,6 +506,7 @@ class String
     [provided, original, detected]
       .compact
       .reject { |e| Encoding.find(e) == Encoding::ASCII_8BIT }
+      .reject { |e| Encoding.find(e) == Encoding::UTF_8 }
       .select { |e| force_encoding(e).valid_encoding? }
       .tap { force_encoding(original) } # clean up changes from previous line
 
diff --git a/spec/lib/core_ext/string_spec.rb b/spec/lib/core_ext/string_spec.rb
index 52bd0ba8f..3981db7c6 100644
--- a/spec/lib/core_ext/string_spec.rb
+++ b/spec/lib/core_ext/string_spec.rb
@@ -2,7 +2,7 @@ require 'rails_helper'
 
 RSpec.describe String do
   describe '#utf8_encode' do
-    context 'for valid, UTF-8-encoded strings' do
+    context 'on valid, UTF-8-encoded strings' do
       let(:subject) { 'hello' }
 
       it 'returns an identical copy' do
@@ -10,9 +10,17 @@ RSpec.describe String do
         expect(subject.utf8_encode.encoding).to be(subject.encoding)
         expect(subject.utf8_encode).not_to be(subject)
       end
+
+      context 'which are incorrectly set to other, technically valid encodings' do
+        let(:subject) { 'ö'.force_encoding('tis-620') }
+
+        it 'sets input encoding to UTF-8 instead of attempting conversion' do
+          expect(subject.utf8_encode).to eq(subject.force_encoding('utf-8'))
+        end
+      end
     end
 
-    context 'for strings in other encodings' do
+    context 'on strings in other encodings' do
       let(:subject) { original_string.encode(input_encoding) }
 
       context 'with no from: option' do