aboutsummaryrefslogtreecommitdiffstats
path: root/lib/normalize_string.rb
diff options
context:
space:
mode:
authorLouise Crow <louise.crow@gmail.com>2015-06-23 10:01:49 +0100
committerLouise Crow <louise.crow@gmail.com>2015-06-23 10:01:49 +0100
commitac59a76cc380af792a09e5dd82b8402dc966af42 (patch)
tree8c0640c44a11ac6f9e42fdb7fc7e6618da2ed6fe /lib/normalize_string.rb
parent994a13b08be94220512961b28de8fb95c8d181a5 (diff)
parent7983e37506f3fa16211255ef802bf50735adbdb8 (diff)
Merge branch 'binary-or-utf8-attachment-bodies' into develop
Diffstat (limited to 'lib/normalize_string.rb')
-rw-r--r--lib/normalize_string.rb23
1 files changed, 16 insertions, 7 deletions
diff --git a/lib/normalize_string.rb b/lib/normalize_string.rb
index d850d7e05..69853fd6e 100644
--- a/lib/normalize_string.rb
+++ b/lib/normalize_string.rb
@@ -73,18 +73,27 @@ 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 = s
- if String.method_defined?(:encode)
- result = s.force_encoding("utf-8").encode("utf-8", :invalid => :replace,
- :undef => :replace,
- :replace => "")
- end
+ result = scrub(s)
+ StringConversionResult.new(result, true)
+ end
+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
- result
end
def log_text_details(message, text)