Close #212: Change Android escaping to preserve basic styling tags and anything inside CDATA.

3 files changed, 45 insertions, 8 deletions

diff --git a/README.md b/README.md
index a11bb49..8618a69 100644
--- a/README.md
+++ b/README.md
@@ -78,6 +78,7 @@ Twine currently supports the following output formats:
 
 * [iOS and OS X String Resources][applestrings] (format: apple)
 * [Android String Resources][androidstrings] (format: android)
+    * Supports [basic styling][androidstyling] with \<b\>, \<i\>, \<u\> and \<a\> links. These tags will *not* be escaped. Use [`getText()`](https://developer.android.com/reference/android/content/res/Resources.html#getText(int)) to read these strings. Also tags inside `<![CDATA[` won't be escaped. See [\#212](https://github.com/scelis/twine/issues/212) for details.
 * [Gettext PO Files][gettextpo] (format: gettext)
 * [jquery-localize Language Files][jquerylocalize] (format: jquery)
 * [Django PO Files][djangopo] (format: django)
@@ -212,9 +213,10 @@ Many thanks to all of the contributors to the Twine project, including:
 [INI]: http://en.wikipedia.org/wiki/INI_file
 [applestrings]: http://developer.apple.com/documentation/Cocoa/Conceptual/LoadingResources/Strings/Strings.html
 [androidstrings]: http://developer.android.com/guide/topics/resources/string-resource.html
+[androidstyling]: http://developer.android.com/guide/topics/resources/string-resource.html#FormattingAndStyling
 [gettextpo]: http://www.gnu.org/savannah-checkouts/gnu/gettext/manual/html_node/PO-Files.html
 [jquerylocalize]: https://github.com/coderifous/jquery-localize
 [djangopo]: https://docs.djangoproject.com/en/dev/topics/i18n/translation/
 [tizen]: https://developer.tizen.org/documentation/articles/localization
 [flash]: http://help.adobe.com/en_US/FlashPlatform/reference/actionscript/3/mx/resources/IResourceManager.html#getString()
-[printf]: https://en.wikipedia.org/wiki/Printf_format_string
+[printf]: https://en.wikipedia.org/wiki/Printf_format_string
\ No newline at end of file
diff --git a/lib/twine/formatters/android.rb b/lib/twine/formatters/android.rb
index b9dba3e..ced90b7 100644
--- a/lib/twine/formatters/android.rb
+++ b/lib/twine/formatters/android.rb
@@ -99,12 +99,28 @@ module Twine
         "\t<string name=\"%{key}\">%{value}</string>"
       end
 
+      def gsub_unless(text, pattern, replacement)
+        text.gsub(pattern) do |match|
+          match_start_position = Regexp.last_match.offset(0)[0]
+          yield(text[0, match_start_position]) ? match : replacement
+        end
+      end
+
+      # http://developer.android.com/guide/topics/resources/string-resource.html#FormattingAndStyling
       def escape_value(value)
-        # escape double and single quotes, & signs and tags
-        value = escape_quotes(value)
-        value.gsub!("'", "\\\\'")
-        value.gsub!(/&/, '&amp;')
-        value.gsub!('<', '&lt;')
+        inside_cdata = /<\!\[CDATA\[((?!\]\]>).)*$/       # opening CDATA tag ('<![CDATA[') not followed by a closing tag (']]>')
+        inside_opening_anchor_tag = /<a\s?((?!>).)*$/     # anchor tag start ('<a ') not followed by a '>'
+
+        # escape double and single quotes and & signs
+        value = gsub_unless(value, '"', '\\"') { |substring| substring =~ inside_cdata || substring =~ inside_opening_anchor_tag }
+        value = gsub_unless(value, "'", "\\'") { |substring| substring =~ inside_cdata }
+        value = gsub_unless(value, /&/, '&amp;') { |substring| substring =~ inside_cdata || substring =~ inside_opening_anchor_tag }
+
+        # escape opening angle brackes unless it's a supported styling tag
+        # https://github.com/scelis/twine/issues/212
+        # https://stackoverflow.com/questions/3235131/#18199543
+        angle_bracket = /<(?!(\/?(b|u|i|a|\!\[CDATA)))/   # matches all `<` but <b>, <u>, <i>, <a> and <![CDATA
+        value = gsub_unless(value, angle_bracket, '&lt;') { |substring| substring =~ inside_cdata }
 
         # 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>
diff --git a/test/test_formatters.rb b/test/test_formatters.rb
index a63b0a5..ae30787 100644
--- a/test/test_formatters.rb
+++ b/test/test_formatters.rb
@@ -45,8 +45,27 @@ class TestAndroidFormatter < FormatterTest
       '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>',
+
+      '<b>bold</b>'               => '<b>bold</b>',
+      '<i>italic</i>'             => '<i>italic</i>',
+      '<u>underline</u>'          => '<u>underline</u>',
+
+      '<span>inline</span>'       => '&lt;span>inline&lt;/span>',
+      '<p>paragraph</p>'          => '&lt;p>paragraph&lt;/p>',
+
+      '<a href="target">link</a>'     => '<a href="target">link</a>',
+      '<a href="target">"link"</a>'   => '<a href="target">\"link\"</a>',
+      '<a href="target"></a>"out"'    => '<a href="target"></a>\"out\"',
+      '<a href="http://url.com?param=1&param2=3&param3=%20">link</a>'   =>   '<a href="http://url.com?param=1&param2=3&param3=%20">link</a>',
+
+      '<p>escaped</p><![CDATA[]]>'    => '&lt;p>escaped&lt;/p><![CDATA[]]>',
+      '<![CDATA[]]><p>escaped</p>'    => '<![CDATA[]]>&lt;p>escaped&lt;/p>',
+      '<![CDATA[<p>unescaped</p>]]>'  => '<![CDATA[<p>unescaped</p>]]>',
+      '<![CDATA[]]><![CDATA[<p>unescaped</p>]]>'  => '<![CDATA[]]><![CDATA[<p>unescaped</p>]]>',
+
+      '<![CDATA[&]]>'  => '<![CDATA[&]]>',
+      '<![CDATA[\']]>' => '<![CDATA[\']]>',
+      '<![CDATA["]]>'  => '<![CDATA["]]>',
 
       '<xliff:g></xliff:g>' => '<xliff:g></xliff:g>',
       '<xliff:g>untouched</xliff:g>' => '<xliff:g>untouched</xliff:g>',