From 60b0eb2adf71f0896b7cc19a826e0c1941734c92 Mon Sep 17 00:00:00 2001
From: Sebastian Ludwig <sebastian@lurado.de>
Date: Tue, 3 May 2016 22:49:43 +0200
Subject: [PATCH] Allowing xliff tags in android values and escaping special
 characters as recommended by Android docs.

---
 lib/twine/formatters/android.rb | 45 +++++++++++++++++++++++++--------
 test/test_formatters.rb         | 22 +++++++++++++---
 2 files changed, 53 insertions(+), 14 deletions(-)

diff --git a/lib/twine/formatters/android.rb b/lib/twine/formatters/android.rb
index acc5da8..24c28aa 100644
--- a/lib/twine/formatters/android.rb
+++ b/lib/twine/formatters/android.rb
@@ -112,19 +112,44 @@ module Twine
         "\t<string name=\"%{key}\">%{value}</string>"
       end
 
-      def format_value(value)
-        # Android enforces the following rules on the values
-        #  1) apostrophes and quotes must be escaped with a backslash
+      def escape_value(value)
+        # escape double and single quotes, & signs and tags
         value = escape_quotes(value)
         value.gsub!("'", "\\\\'")
-        #  2) HTML escape the string
-        value = CGI.escapeHTML(value)
-        #  3) convert placeholders (e.g. %@ -> %s)
-        value = convert_placeholders_from_twine_to_android(value)
-        #  4) escape non resource identifier @ signs (http://developer.android.com/guide/topics/resources/accessing-resources.html#ResourcesFromXml)
+        value.gsub!(/&/, '&amp;')
+        value.gsub!('<', '&lt;')
+
+        # escape non resource identifier @ signs (http://developer.android.com/guide/topics/resources/accessing-resources.html#ResourcesFromXml)
         resource_identifier_regex = /@(?!([a-z\.]+:)?[a-z+]+\/[a-zA-Z_]+)/   # @[<package_name>:]<resource_type>/<resource_name>
-        value.gsub!(resource_identifier_regex, '\@')
-        #  5) replace beginning and end spaces with \u0020. Otherwise Android strips them.
+        value.gsub(resource_identifier_regex, '\@')
+      end
+
+      # see http://developer.android.com/guide/topics/resources/string-resource.html#FormattingAndStyling
+      # however unescaped HTML markup like in "Welcome to <b>Android</b>!" is stripped when retrieved with getString() (http://stackoverflow.com/questions/9891996/)
+      def format_value(value)
+        value = value.dup
+
+        # capture xliff tags and replace them with a placeholder
+        xliff_tags = []
+        value.gsub! /<xliff:g.+?<\/xliff:g>/ do
+          xliff_tags << $&
+          'TWINE_XLIFF_TAG_PLACEHOLDER'
+        end
+
+        # escape everything outside xliff tags
+        value = escape_value(value)
+
+        # put xliff tags back into place
+        xliff_tags.each do |xliff_tag|
+          # escape content of xliff tags
+          xliff_tag.gsub! /(<xliff:g.*?>)(.*)(<\/xliff:g>)/ do "#{$1}#{escape_value($2)}#{$3}" end
+          value.sub! 'TWINE_XLIFF_TAG_PLACEHOLDER', xliff_tag
+        end
+        
+        # convert placeholders (e.g. %@ -> %s)
+        value = convert_placeholders_from_twine_to_android(value)
+        
+        # replace beginning and end spaces with \u0020. Otherwise Android strips them.
         value.gsub(/\A *| *\z/) { |spaces| '\u0020' * spaces.length }
       end
 
diff --git a/test/test_formatters.rb b/test/test_formatters.rb
index c40a12a..b8f0001 100644
--- a/test/test_formatters.rb
+++ b/test/test_formatters.rb
@@ -101,10 +101,24 @@ class TestAndroidFormatter < FormatterTest
     assert_equal "value\\u0020", @formatter.format_value('value ')
   end
 
-  def test_format_value_escapes_single_quotes
-    skip 'not working with ruby 2.0'
-    # http://stackoverflow.com/questions/18735608/cgiescapehtml-is-escaping-single-quote
-    assert_equal "not \\'so\\' easy", @formatter.format_value("not 'so' easy")
+  def test_format_value_escaping
+    values = {
+      'this & that'               => 'this &amp; that',
+      'this < that'               => 'this &lt; that',
+      "it's complicated"          => "it\\'s complicated",
+      'a "good" way'              => 'a \"good\" way',
+      '<b>bold</b>'               => '&lt;b>bold&lt;/b>',
+      '<a href="target">link</a>' => '&lt;a href=\"target\">link&lt;/a>',
+
+      '<xliff:g></xliff:g>' => '<xliff:g></xliff:g>',
+      '<xliff:g>untouched</xliff:g>' => '<xliff:g>untouched</xliff:g>',
+      '<xliff:g id="42">untouched</xliff:g>' => '<xliff:g id="42">untouched</xliff:g>',
+      '<xliff:g id="1">first</xliff:g> inbetween <xliff:g id="2">second</xliff:g>' => '<xliff:g id="1">first</xliff:g> inbetween <xliff:g id="2">second</xliff:g>'
+    }
+    
+    values.each do |input, expected|
+      assert_equal expected, @formatter.format_value(input)
+    end
   end
 
   def test_format_value_escapes_non_resource_identifier_at_signs