mailparser 0.5.0.beta2 → 0.5.0.beta3

data/lib/mailparser/rfc2047.rb

@@ -21,9 +21,10 @@ module MailParser::RFC2047
     end
     charset_converter ||= MailParser::ConvCharset.method(:conv_charset)
     last_charset = nil
-    ret = ""
-    ret.force_encoding(charset) if String.method_defined? :force_encoding and charset
-    split_decode(str) do |s, cs, raw|
+    words = []
+    mime_word = false
+    str.split(/(\s+)/).each do |s|
+      s, cs, raw = decode_word(s)
       if charset
         begin
           s = charset_converter.call(cs || charset, charset, s)
@@ -32,32 +33,37 @@ module MailParser::RFC2047
           cs = nil
         end
       end
-      ret << " " if last_charset.nil? or cs.nil?
-      ret << s
-      last_charset = cs
+      if cs
+        words.pop if mime_word and words.last =~ /\A\s*\z/
+        mime_word = true
+      elsif s !~ /\A\s*\z/
+        mime_word = false
+      end
+      words.push s
+    end
+    begin
+      ret = words.join
+    rescue
+      ret = words.map{|s| s.force_encoding('binary')}.join
     end
-    return ret.strip
+    ret
   end
 
-  def split_decode(str)
-    while str =~ /\=\?([^\(\)\<\>\@\,\;\:\"\/\[\]\?\.\=]+)\?([QB])\?([^\? ]+)\?\=/i do
-      raw = $&
-      pre, charset, encoding, enc_text, after = $`, $1.downcase, $2.downcase, $3, $'
-      s = pre.strip
-      yield s, nil, raw unless s.empty?
-      s = encoding == "q" ? q_decode(enc_text) : b_decode(enc_text)
+  def decode_word(str)
+    charset = nil
+    if str =~ /\=\?([^\(\)\<\>\@\,\;\:\"\/\[\]\?\.\=]+)\?([QB])\?([^\? ]+)\?\=/i
+      charset, encoding, enc_text = $1.downcase, $2.downcase, $3
+      raw = str
+      str = encoding == "q" ? q_decode(enc_text) : b_decode(enc_text)
       if String.method_defined? :force_encoding
         begin
-          s.force_encoding(charset)
+          str.force_encoding(charset)
         rescue
-          s.force_encoding('ascii-8bit')
+          str.force_encoding('ascii-8bit')
         end
       end
-      yield s, charset, raw
-      str = after
     end
-    s = str.strip
-    yield s, nil, raw unless s.empty?
+    [str, charset, raw]
   end
 
   def q_decode(str)

data/lib/mailparser.rb

@@ -78,7 +78,7 @@ module MailParser
           @parsed = Loose.parse(@name, @raw, @opt)
         end
       else
-        r = @raw.chomp.gsub(/\s+/, " ")
+        r = @raw.chomp.gsub(/\r?\n/, '').gsub(/\t/, ' ')
         if @opt[:decode_mime_header] then
           @parsed = RFC2047.decode(r, @opt)
         else

data/test/test_mailparser.rb

@@ -257,13 +257,21 @@ EOS
   def test_subject_multi_line()
     msg = StringIO.new(<<EOS)
 Subject: This is a pen.
-  Is this a pen?
+ Is this a pen?
 
 EOS
     m = MailParser::Message.new(msg)
     assert_equal("This is a pen. Is this a pen?", m.subject)
   end
 
+  def test_subject_keep_multi_space
+    msg = StringIO.new(<<EOS)
+Subject:   This   is   a   pen.  
+EOS
+    m = MailParser::Message.new(msg)
+    assert_equal("This   is   a   pen.  ", m.subject)
+  end
+
   def test_subject_multi_header()
     msg = StringIO.new(<<EOS)
 Subject: This is a pen.
@@ -292,6 +300,22 @@ EOS
     assert_equal("This is a pen.", m.subject)
   end
 
+  def test_subject_mime_decode_keep_multi_space
+    msg = StringIO.new(<<EOS)
+Subject:   This   is   a   pen.  
+EOS
+    m = MailParser::Message.new(msg, :decode_mime_header=>true)
+    assert_equal("This   is   a   pen.  ", m.subject)
+  end
+
+  def test_subject_mime_decode_keep_multi_space2
+    msg = StringIO.new(<<EOS)
+Subject: abcdefg   =?us-ascii?q?hoge?=   =?us-ascii?q?fuga?=  
+EOS
+    m = MailParser::Message.new(msg, :decode_mime_header=>true)
+    assert_equal("abcdefg   hogefuga  ", m.subject)
+  end
+
   def test_subject_mime_decode_charset()
     msg = StringIO.new(<<EOS)
 Subject: =?iso-2022-jp?b?GyRCJCIkJCQmJCgkKhsoQg==?=

data/test/test_rfc2047.rb

@@ -38,20 +38,18 @@ class TC_RFC2047 < Test::Unit::TestCase
     assert_equal("とみた", MailParser::RFC2047.b_decode("44Go 44 G/4 4Gf"))
   end
 
-  def test_split_decode_q_ascii()
-    MailParser::RFC2047.split_decode("=?us-ascii?q?hoge?=") do |s, cs, raw |
-      assert_equal 'hoge', s
-      assert_equal 'us-ascii', cs
-      assert_equal '=?us-ascii?q?hoge?=', raw
-    end
+  def test_decode_word()
+    s, cs, raw = MailParser::RFC2047.decode_word("=?us-ascii?q?hoge?=")
+    assert_equal 'hoge', s
+    assert_equal 'us-ascii', cs
+    assert_equal '=?us-ascii?q?hoge?=', raw
   end
 
-  def test_split_decode_q_ascii_upcase()
-    MailParser::RFC2047.split_decode("=?US-ASCII?Q?hoge?=") do |s, cs, raw|
-      assert_equal 'hoge', s
-      assert_equal 'us-ascii', cs
-      assert_equal '=?US-ASCII?Q?hoge?=', raw
-    end
+  def test_decode_word_upcase()
+    s, cs, raw = MailParser::RFC2047.decode_word("=?US-ASCII?Q?hoge?=")
+    assert_equal 'hoge', s
+    assert_equal 'us-ascii', cs
+    assert_equal '=?US-ASCII?Q?hoge?=', raw
   end
 
   def test_decode_q_ascii()
@@ -121,6 +119,8 @@ class TC_RFC2047 < Test::Unit::TestCase
     s = MailParser::RFC2047.decode("=?sjis?b?h0A=?=", "UTF-8")
     if String.method_defined? :force_encoding
       assert_equal("\xe2\x91\xa0".force_encoding('utf-8'), s)
+    else
+      assert_equal("\xe2\x91\xa0", s)
     end
   end
 
@@ -128,9 +128,16 @@ class TC_RFC2047 < Test::Unit::TestCase
     s = MailParser::RFC2047.decode("=?iso-2022-jp?b?GyRCLSEbKEI=?=", "UTF-8")
     if String.method_defined? :force_encoding
       assert_equal("\xe2\x91\xa0".force_encoding('utf-8'), s)
+    else
+      assert_equal("\xe2\x91\xa0", s)
     end
   end
 
+  def test_decode_different_charset
+    s = MailParser::RFC2047.decode("=?iso-2022-jp?b?GyRCJCIbKEI=?= =?utf-8?b?44GE?=")
+    assert_equal("\e$B$\"\e(B\xE3\x81\x84", s)
+  end
+
   def test_decode_charset_converter()
     proc = Proc.new{|f,t,s| s.gsub(/o/, "X")}
     s = MailParser::RFC2047.decode("=?us-ascii?q?hoge?=", :output_charset=>"utf-8", :charset_converter=>proc)

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: mailparser
 version: !ruby/object:Gem::Version
-  version: 0.5.0.beta2
+  version: 0.5.0.beta3
   prerelease: 6
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-09-18 00:00:00.000000000 Z
+date: 2012-09-23 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: mmapscanner