From ce96dcd4259bcc1badf5f6b690d12525231aca45 Mon Sep 17 00:00:00 2001
From: Scott Weber <scott@mobiata.com>
Date: Mon, 20 Feb 2012 14:43:33 -0500
Subject: [PATCH] android-specific improvements

- cross-reference between iOS/Android language/region codes
- support for default languages other than English
- proper escaping of ' and "
- translation between %@ and %s, including numbered parameters (e.g. %1$@)
- proper XML-escaping of < and &
---
 lib/twine/formatters/android.rb | 61 ++++++++++++++++++++++++---------
 lib/twine/stringsfile.rb        | 13 ++++++-
 2 files changed, 57 insertions(+), 17 deletions(-)

diff --git a/lib/twine/formatters/android.rb b/lib/twine/formatters/android.rb
index 633df1f..449490c 100644
--- a/lib/twine/formatters/android.rb
+++ b/lib/twine/formatters/android.rb
@@ -1,6 +1,5 @@
 # encoding: utf-8
 
-require 'cgi'
 require 'rexml/document'
 
 module Twine
@@ -9,6 +8,20 @@ module Twine
       FORMAT_NAME = 'android'
       EXTENSION = '.xml'
       DEFAULT_FILE_NAME = 'strings.xml'
+      LANG_CODES = Hash[
+        'zh' => 'zh-Hans',
+        'zh-rCN' => 'zh-Hans',
+        'zh-rHK' => 'zh-Hant',
+        'zh-rTW' => 'zh-TW',
+        'en-rGB' => 'en-UK',
+        'fr-rCA' => 'fr-CA',
+        'in' => 'id',
+        'nb' => 'no'
+        # TODO: spanish
+      ]
+      DEFAULT_LANG_CODES = Hash[
+        'zh-TW' => 'zh-Hant' # if we don't have a zh-TW translation, try zh-Hant before en
+      ]
 
       def self.can_handle_directory?(path)
         Dir.entries(path).any? { |item| /^values-.+$/.match(item) }
@@ -24,6 +37,7 @@ module Twine
           match = /^values-(.*)$/.match(segment)
           if match
             lang = match[1]
+            lang = LANG_CODES.fetch(lang, lang)
             lang.sub!('-r', '-')
             return lang
           end
@@ -42,14 +56,19 @@ module Twine
             value.gsub!('\\\'', '\'')
             value.gsub!('\\"', '"')
             value.gsub!(/\n/, '')
-            value.gsub!('%s', '%@')
+            value.gsub!(/%([0-9\$]*)s/, '%\1@')
+            value.gsub!('&lt;', '<')
+            value.gsub!('&amp;', '&')
             set_translation_for_key(key, lang, value)
           end
         end
       end
 
       def write_file(path, lang)
-        default_lang = @strings.language_codes[0]
+        default_lang = [@strings.language_codes[0]]
+        if DEFAULT_LANG_CODES.has_key?(lang)
+          default_lang.insert(0, DEFAULT_LANG_CODES[lang])
+        end
         File.open(path, 'w:UTF-8') do |f|
           f.puts "<?xml version=\"1.0\" encoding=\"utf-8\"?>\n<!-- Android Strings File -->\n<!-- Generated by Twine -->\n<!-- Language: #{lang} -->"
           f.write '<resources>'
@@ -67,22 +86,32 @@ module Twine
                 end
 
                 key = row.key
-                key = CGI.escapeHTML(key)
 
                 value = row.translated_string_for_lang(lang, default_lang)
-                value.gsub!('\'', '\\\\\'')
-                value.gsub!('%@', '%s')
-                value = CGI.escapeHTML(value)
-
-                comment = row.comment
-                if comment
-                  comment = comment.gsub('--', '—')
+                value = String.new(value) # use a copy to prevent modifying the original
+                if value
+                  # if values is nil, there was no default, so let Android handle the defaulting
+                  
+                  # Android enforces the following rules on the values
+                  #  1) apostrophes and quotes must be escaped with a backslash
+                  value.gsub!('\'', '\\\\\'')
+                  value.gsub!('"', '\\\\"')
+                  #  2) ampersand and less-than must be in XML-escaped form
+                  value.gsub!('&', '&amp;')
+                  value.gsub!('<', '&lt;')
+                  #  3) use "s" instead of "@" for substituting strings
+                  value.gsub!(/%([0-9\$]*)@/, '%\1s')
+  
+                  comment = row.comment
+                  if comment
+                    comment = comment.gsub('--', '—')
+                  end
+  
+                  if comment && comment.length > 0
+                    f.puts "\t<!-- #{comment} -->\n"
+                  end
+                  f.puts "\t<string name=\"#{key}\">#{value}</string>"
                 end
-
-                if comment && comment.length > 0
-                  f.puts "\t<!-- #{comment} -->\n"
-                end
-                f.puts "\t<string name=\"#{key}\">#{value}</string>"
               end
             end
           end
diff --git a/lib/twine/stringsfile.rb b/lib/twine/stringsfile.rb
index 47ff1be..3d82a15 100644
--- a/lib/twine/stringsfile.rb
+++ b/lib/twine/stringsfile.rb
@@ -41,7 +41,18 @@ module Twine
     end
     
     def translated_string_for_lang(lang, default_lang=nil)
-      @translations[lang] || @translations[default_lang]
+      if @translations[lang]
+        return @translations[lang]
+      elsif default_lang.respond_to?("each")
+        default_lang.each do |def_lang|
+          if @translations[def_lang]
+            return @translations[def_lang]
+          end
+        end
+        return nil
+      else
+        return @translations[default_lang]
+      end
     end
   end