Handle double quotes, backslashes, CRs, NLs and tabs

.gitignore

@@ -0,0 +1,29 @@
+*.gem
+*.rbc
+.bundle
+.config
+.yardoc
+Gemfile.lock
+InstalledFiles
+_yardoc
+vendor
+coverage
+doc/
+lib/bundler/man
+pkg
+rdoc
+spec/reports
+test/tmp
+test/version_tmp
+tmp
+
+# JetBrains
+.idea
+
+# mac
+.DS_Store
+__MACOSX
+
+# emacs turds
+(.*/)?\#[^/]*\#$
+

lib/apfel.rb

@@ -7,10 +7,12 @@ module Apfel
   # strings object
   def self.parse(file)
     file = read(file)
+    # confirmed that read does remove the first comment in utf
     DotStringsParser.new(file).parse_file
   end
 
   def self.read(file)
-   Reader.read(file)
+    # confirmed that read does remove the first comment in utf
+    Reader.read(file)
   end
 end

lib/apfel/line.rb

@@ -61,12 +61,47 @@ def key
 
     def value
       if key_value_pair?
-        cleaned_content.partition(/"\s*=\s*"/)[2].gsub!(/(^"|"$)/, "")
+        unescape_value cleaned_content.partition(/"\s*=\s*"/)[2].gsub!(/(^"|"$)/, "")
       end
     end
 
     def is_comment?
      whole_comment? || open_comment? || close_comment? || in_comment
     end
+
+    private
+
+    # http://developer.apple.com/library/mac/#documentation/cocoa/conceptual/LoadingResources/Strings/Strings.html
+    def unescape_value(string)
+      state = :normal
+      out = ''
+      string.each_char do |c|
+        case state
+        when :normal
+          if c == '\\'
+            state = :escape
+          else
+            out += c
+          end
+        when :escape
+          state = :normal
+          case c
+          when '\\'
+            out += '\\'
+          when '"'
+            out += '"'
+          when 'r'
+            out += "\r"
+          when 'n'
+            out += "\n"
+          when 't'
+            out += "\t"
+          else
+            out += '\\' + c # Do nothing, however in the future handling unicode escapes could be good
+          end
+        end
+      end
+      out
+    end
   end
 end

lib/apfel/reader.rb

@@ -4,9 +4,12 @@ class Reader
     # Reads in a file and returns an array consisting of each line of input
     # cleaned of new line characters
     def self.read(file)
-      File.open(file, "r") do |f|
-      content_array=[]
-      content = f.read
+      File.open(file, 'r') do |f|
+        content_array=[]
+        # http://stackoverflow.com/questions/5011504/is-there-a-way-to-remove-the-bom-from-a-utf-8-encoded-file
+        # problem is the BOM that can be found at char 0 in strings files
+        content = f.read.force_encoding('UTF-8')
+        content.sub!("\xEF\xBB\xBF".force_encoding("UTF-8"), '')
         content.each_line do |line|
           line.gsub!("\n","")
           content_array.push(line)

spec/apfel_escapes_spec.rb

@@ -0,0 +1,32 @@
+require 'spec_helper'
+require 'apfel'
+
+describe Apfel do
+  describe '::parse_file' do
+    context 'when given DotStrings file with escapes'do
+      let(:parsed_file_hash) do
+        Apfel.parse('./spec/files/escapes.strings').to_hash(:with_comments => false)
+      end
+
+      it 'should parse nl' do
+        parsed_file_hash['multiline'].should eq "line 1\nline 2"
+      end
+
+      it 'should parse cr' do
+        parsed_file_hash['mac'].should eq "before cr\rafter cr"
+      end
+
+      it 'should parse tabs' do
+        parsed_file_hash['tabs'].should eq "two spaces \t equals tab"
+      end
+
+      it 'should parse double quotes' do
+        parsed_file_hash['dq'].should eq "\"someone said this\""
+      end
+
+      it 'should parse backslashes' do
+        parsed_file_hash['backslash'].should eq "\\not a forward slash"
+      end
+    end
+  end
+end

spec/apfel_parse_ascii_file_spec.rb

@@ -0,0 +1,50 @@
+require 'spec_helper'
+require 'apfel'
+require 'apfel/parsed_dot_strings'
+
+describe Apfel do
+  describe '::parse_file' do
+    context 'when given a ASCII DotStrings file'do
+
+      it 'the file should be ascii' do
+        res = `file  -I ./spec/files/ascii.strings`
+        encoding = res.split(/=/).last.gsub!("\n",'')
+        encoding.should == 'us-ascii'
+      end
+
+      let(:parsed_file) do
+        Apfel.parse('./spec/files/ascii.strings')
+      end
+
+      it 'returns a ParsedDotStrings object' do
+        parsed_file.should be_a(Apfel::ParsedDotStrings)
+      end
+
+      #it 'should have the correct keys' do
+      #  parsed_file.keys.should include 'key_number_one'
+      #  parsed_file.keys.should include 'key_number_two'
+      #  parsed_file.keys.should include 'key_number_three'
+      #end
+      #
+      #it 'should have the correct values' do
+      #  parsed_file.values.should include 'value number one'
+      #  parsed_file.values.should include 'value number two'
+      #  parsed_file.values.should include 'value number three'
+      #end
+      #
+      describe 'should have the correct comments' do
+        it 'should have the correct comment for avoided social event' do
+          parsed_file.comments['avoided social event'].should == 'No comment provided by engineer.'
+        end
+
+        it 'should have the correct comment for binged' do
+          parsed_file.comments['binged'].should == 'No comment provided by engineer.'
+        end
+
+        it 'should have the correct comment for called a friend' do
+          parsed_file.comments['called a friend'].should == 'No comment provided by engineer.'
+        end
+      end
+    end
+  end
+end

spec/apfel_parse_utf8_file_spec.rb

@@ -0,0 +1,53 @@
+require 'spec_helper'
+require 'apfel'
+require 'apfel/parsed_dot_strings'
+
+describe Apfel do
+  describe '::parse_file' do
+    context 'when given a UTF8 DotStrings file'do
+
+      it 'the file should be utf-8' do
+        File.open('./spec/files/utf8.strings', 'r') do |f|
+          f.external_encoding.name.should == 'UTF-8'
+          content = f.read
+          content.encoding.name.should == 'UTF-8'
+        end
+      end
+
+      let(:parsed_file) do
+        Apfel.parse('./spec/files/utf8.strings')
+      end
+
+      it 'returns a ParsedDotStrings object' do
+        parsed_file.should be_a(Apfel::ParsedDotStrings)
+      end
+
+      #it 'should have the correct keys' do
+      #  parsed_file.keys.should include 'key_number_one'
+      #  parsed_file.keys.should include 'key_number_two'
+      #  parsed_file.keys.should include 'key_number_three'
+      #end
+      #
+      #it 'should have the correct values' do
+      #  parsed_file.values.should include 'value number one'
+      #  parsed_file.values.should include 'value number two'
+      #  parsed_file.values.should include 'value number three'
+      #end
+      #
+      describe 'should have the correct comments' do
+        it 'should have the correct comment for anger' do
+          parsed_file.comments['anger'].should == 'No comment provided by engineer.'
+        end
+
+        it 'should have the correct comment for anxiety' do
+          parsed_file.comments['anxiety'].should == 'No comment provided by engineer.'
+        end
+
+        it 'should have the correct comment for boredom' do
+          parsed_file.comments['boredom'].should == 'No comment provided by engineer.'
+        end
+      end
+
+    end
+  end
+end

spec/apfel_spec.rb

@@ -4,7 +4,7 @@
 
 describe Apfel do
   describe '::parse_file' do
-    context 'when given a valid DotStrings file'do
+    context 'when given an ASCII DotStrings file'do
 
       let(:parsed_file) do
         Apfel.parse(valid_file)
@@ -13,16 +13,43 @@
       it 'returns a ParsedDotStrings object' do
         parsed_file.should be_a(Apfel::ParsedDotStrings)
       end
+
+      it 'should have the correct keys' do
+        parsed_file.keys.should include 'key_number_one'
+        parsed_file.keys.should include 'key_number_two'
+        parsed_file.keys.should include 'key_number_three'
+      end
+
+      it 'should have the correct values' do
+        parsed_file.values.should include 'value number one'
+        parsed_file.values.should include 'value number two'
+        parsed_file.values.should include 'value number three'
+      end
+
+      describe 'should have the correct comments' do
+        it 'should have the correct comment for first' do
+          parsed_file.comments(with_keys: false).should include 'This is the first comment'
+        end
+
+        it 'should have the correct comment for second' do
+          parsed_file.comments['key_number_two'].should == 'This is a multiline comment'
+        end
+
+
+        it 'should have the correct comment for third' do
+          parsed_file.comments(with_keys: false).should include 'This is comment number 3'
+        end
+      end
     end
 
     context 'when given an invalid strings file' do
       context 'missing a semicolon' do
 
         let(:invalid_file_semicolon) do
-          create_temp_file(  <<-EOS
+          create_temp_file('ascii',  <<-EOS
 /* This is the first comment */
 "key_number_one" = "value number one"
-            EOS
+          EOS
           )
         end
 
@@ -35,7 +62,7 @@
 
       context 'not closed comment' do
         let(:invalid_file_comment) do
-          create_temp_file(<<-EOS
+          create_temp_file('ascii', <<-EOS
 /* This is the first comment
 "key_number_one" = "value number one";
 

spec/files/ascii.strings

@@ -0,0 +1,9 @@
+/* No comment provided by engineer. */
+"avoided social event" = "avoided social event";
+
+/* No comment provided by engineer. */
+"binged" = "binged";
+
+/* No comment provided by engineer. */
+"called a friend" = "called a friend";
+

spec/files/escapes.strings

@@ -0,0 +1,5 @@
+"multiline" = "line 1\nline 2";
+"mac" = "before cr\rafter cr";
+"tabs" = "two spaces \t equals tab";
+"dq" = "\"someone said this\"";
+"backslash" = "\\not a forward slash";

spec/files/utf8.strings

@@ -0,0 +1,8 @@
+/* No comment provided by engineer. */
+"anger" = "anger";
+
+/* No comment provided by engineer. */
+"anxiety" = "anxiety";
+
+/* No comment provided by engineer. */
+"boredom" = "boredom";

spec/reader_spec.rb

@@ -5,7 +5,7 @@ module Apfel
   describe Reader do
     describe '#read' do
       let(:temp_file) do
-        create_temp_file(<<-EOS
+        create_temp_file('ascii', <<-EOS
 This is a file with some lines.
 Roses are red, violets are blue.
 This text is really boring,