From 4440d11fb662c57428a2aba622209d6d1ddc0a59 Mon Sep 17 00:00:00 2001
From: Louise Crow <louise.crow@gmail.com>
Date: Tue, 9 Jun 2015 16:33:03 +0100
Subject: Round trip through utf-16 to clean utf-8 string

As noted in the ruby docs (http://ruby-doc.org/core-1.9.3/String.html#method-i-encode),
any conversion from an encoding to the same encoding is a no-op, covert it first to utf-16.
---
 lib/normalize_string.rb | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)

(limited to 'lib/normalize_string.rb')

diff --git a/lib/normalize_string.rb b/lib/normalize_string.rb
index d850d7e05..8b54c080c 100644
--- a/lib/normalize_string.rb
+++ b/lib/normalize_string.rb
@@ -77,16 +77,20 @@ def convert_string_to_utf8(s, suggested_character_encoding=nil)
     begin
         result = normalize_string_to_utf8 s, suggested_character_encoding
     rescue EncodingNormalizationError
-        result = s
-        if String.method_defined?(:encode)
-            result = s.force_encoding("utf-8").encode("utf-8", :invalid => :replace,
-                                                               :undef => :replace,
-                                                               :replace => "")
-        end
+        result = scrub(s)
     end
     result
 end
 
+def scrub(string)
+    if String.method_defined?(:encode)
+        string = string.force_encoding("utf-8")
+        string.valid_encoding? ? string : string.encode("utf-16le", :invalid => :replace, :replace => "").encode("utf-8")
+    else
+        Iconv.conv('UTF-8//IGNORE', 'UTF-8', string)
+    end
+end
+
 def log_text_details(message, text)
     if String.method_defined?(:encode)
         STDERR.puts "#{message}, we have text: #{text}, of class #{text.class} and encoding #{text.encoding}"
-- 
cgit v1.2.3


From 3d8f0cc6b70b55aad20ab4d86642f0e6d605c921 Mon Sep 17 00:00:00 2001
From: Louise Crow <louise.crow@gmail.com>
Date: Thu, 11 Jun 2015 17:11:26 +0100
Subject: convert_string_to_utf8 returns struct of string and scrubbing status.

---
 lib/normalize_string.rb | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

(limited to 'lib/normalize_string.rb')

diff --git a/lib/normalize_string.rb b/lib/normalize_string.rb
index 8b54c080c..69853fd6e 100644
--- a/lib/normalize_string.rb
+++ b/lib/normalize_string.rb
@@ -73,13 +73,18 @@ def convert_string_to_utf8_or_binary(s, suggested_character_encoding=nil)
     result
 end
 
+class StringConversionResult < Struct.new(:string, :scrubbed)
+  alias_method :scrubbed?, :scrubbed
+end
+
 def convert_string_to_utf8(s, suggested_character_encoding=nil)
     begin
         result = normalize_string_to_utf8 s, suggested_character_encoding
+        StringConversionResult.new(result, false)
     rescue EncodingNormalizationError
         result = scrub(s)
+        StringConversionResult.new(result, true)
     end
-    result
 end
 
 def scrub(string)
-- 
cgit v1.2.3