fix: convert to_utf8 correctly, without hidding errors

https://jira.railsc.ru/browse/BPC-22244

lib/string_tools/core_ext/string.rb

@@ -163,32 +163,21 @@ def detect_encoding
     e
   end
 
-  def to_utf8!
-    self.replace(self.to_utf8)
-  end
-
-  def to_utf8
-    # и так utf
-    return self if is_utf8?
+  def to_utf8(inplace = false)
+    return self if valid_encoding? && is_utf8?
 
-    enc = detect_encoding
+    source_enc = detect_encoding
+    return '' unless source_enc
 
-    # если utf или английские буквы, то тоже ок
-    return self if ['utf-8', 'ascii'].include?(enc)
-
-    # если неизвестная каша, то возвращаем пустую строку
-    return '' if enc.nil?
+    encode_utf8(source_enc, inplace)
+  end
 
-    # иначе пытаемся перекодировать
-    encode 'utf-8', enc, :undef => :replace, :invalid => :replace
-  rescue
-    ''
+  def to_utf8!
+    to_utf8(true)
   end
 
   def to_cp1251
-    encode 'cp1251', :undef => :replace, :invalid => :replace
-  rescue
-    ''
+    encode 'cp1251', undef: :replace, invalid: :replace, replace: ''
   end
 
   def to_cp1251!
@@ -230,6 +219,12 @@ def remove_nonprintable!
 
   private
 
+  def encode_utf8(source_enc, inplace = false)
+    args = ['utf-8', source_enc, undef: :replace, invalid: :replace, replace: '']
+
+    inplace ? encode!(*args) : encode(*args)
+  end
+
   def surround_with_ansi(ascii_seq)
     "#{ascii_seq}#{protect_escape_of(ascii_seq)}#{ANSI_CLEAR}"
   end

spec/string_spec.rb

@@ -2,6 +2,30 @@
 require 'spec_helper'
 
 describe String do
+  describe '#to_utf8' do
+    let(:invalid_str) { "кирилиц\xD0".dup }
+
+    it do
+      expect(invalid_str).not_to be_valid_encoding
+      expect(invalid_str.to_utf8).to be_valid_encoding
+      expect(invalid_str.to_utf8.is_utf8?).to eq(true)
+      expect(invalid_str.to_utf8).to eq('кирилиц')
+    end
+  end
+
+  describe '#to_utf8!' do
+    let(:invalid_str) { "кирилиц\xD0".dup }
+
+    it do
+      expect(invalid_str).not_to be_valid_encoding
+      invalid_str.to_utf8!
+
+      expect(invalid_str).to be_valid_encoding
+      expect(invalid_str.is_utf8?).to eq(true)
+      expect(invalid_str).to eq('кирилиц')
+    end
+  end
+
   describe '#mb_downcase' do
     it { expect("Кириллица".mb_downcase).to eq("кириллица") }
   end