mailparser 0.4.22a → 0.5.0.beta1

data/test/test_mailparser.rb

@@ -381,7 +381,11 @@ Content-Type: text/plain; charset=iso-2022-jp
 \e\$B\$\"\$\$\$\&\$\(\$\*\e(B
 EOS
     m = MailParser::Message.new(msg)
-    assert_equal("\e\$B\$\"\$\$\$\&\$\(\$\*\e(B\n", m.body)
+    if String.method_defined? :encode
+      assert_equal("あいうえお\n".encode('iso-2022-jp'), m.body)
+    else
+      assert_equal("\e\$B\$\"\$\$\$\&\$\(\$\*\e(B\n", m.body)
+    end
   end
 
   def test_body_iso2022jp_charset()
@@ -391,7 +395,7 @@ Content-Type: text/plain; charset=iso-2022-jp
 
 \e\$B\$\"\$\$\$\&\$\(\$\*\e(B
 EOS
-    m = MailParser::Message.new(msg, :output_charset=>"utf8")
+    m = MailParser::Message.new(msg, :output_charset=>"utf-8")
     assert_equal("あいうえお\n", m.body)
   end
 
@@ -402,7 +406,7 @@ Content-Type: application/octet-stream; charset=iso-2022-jp
 
 \e\$B\$\"\$\$\$\&\$\(\$\*\e(B
 EOS
-    m = MailParser::Message.new(msg, :output_charset=>"utf8")
+    m = MailParser::Message.new(msg, :output_charset=>"utf-8")
     assert_equal("\e\$B\$\"\$\$\$\&\$\(\$\*\e(B\n", m.body)
   end
 
@@ -413,40 +417,43 @@ Content-Type: text/plain
 
 abcdefg
 EOS
-    m = MailParser::Message.new(msg, :output_charset=>"utf8")
+    m = MailParser::Message.new(msg, :output_charset=>"utf-8")
     assert_equal("abcdefg\n", m.body)
   end
 
   def test_body_preconv
-    m = MailParser::Message.new(<<EOS, :output_charset=>"utf8")
+    m = MailParser::Message.new(<<EOS, :output_charset=>"utf-8")
 From: TOMITA Masahiro <tommy@tmtm.org>
 Content-Type: text/plain; charset=iso-2022-jp
 
 \e$B$3$l$OK\\J8$G$9\e(B
 EOS
     assert_equal "これは本文です\n", m.body
-    assert_equal "\e$B$3$l$OK\\J8$G$9\e(B\n", m.body_preconv
+    if String.method_defined? :encode
+      assert_equal "これは本文です\n".encode('iso-2022-jp'), m.body_preconv
+    else
+      assert_equal "\e$B$3$l$OK\\J8$G$9\e(B\n", m.body_preconv
+    end
   end
 
-  def test_body_charset_converter
-    m = MailParser::Message.new(<<EOS, :output_charset=>"utf8", :charset_converter=>proc{|f,t,s| "abcdefg"})
+  def test_body_preconv_with_unknown_charset
+    m = MailParser::Message.new(<<EOS)
 From: TOMITA Masahiro <tommy@tmtm.org>
-Content-Type: text/plain; charset=iso-2022-jp
+Content-Type: text/plain; charset=hogehoge
 
-\e$B$3$l$OK\\J8$G$9\e(B
+abc
 EOS
-    assert_equal "abcdefg", m.body
+    assert_equal "abc\n", m.body_preconv
   end
 
-  def test_body_use_file
-    m = MailParser::Message.new(<<EOS, :output_charset=>"utf8", :use_file=>1)
+  def test_body_charset_converter
+    m = MailParser::Message.new(<<EOS, :output_charset=>"utf-8", :charset_converter=>proc{|f,t,s| "abcdefg"})
 From: TOMITA Masahiro <tommy@tmtm.org>
 Content-Type: text/plain; charset=iso-2022-jp
 
 \e$B$3$l$OK\\J8$G$9\e(B
 EOS
-    assert_equal "これは本文です\n", m.body
-    assert_equal "\e$B$3$l$OK\\J8$G$9\e(B\n", m.body_preconv
+    assert_equal "abcdefg", m.body
   end
 
   def test_filename()
@@ -568,63 +575,36 @@ EOS
     assert_equal("aaaa bbb", m.filename)
   end
 
-  def test_extract_message_type()
+  def test_multipart()
     msg = StringIO.new(<<EOS)
 From: from1@example.com
 Content-Type: multipart/mixed; boundary="xxxx"
 
+preamble
 --xxxx
 Content-Type: text/plain
 
 body1
 
 --xxxx
-Content-Type: message/rfc822
-
-From: from2@example.com
-Content-Type: text/plain
+Content-Type: application/octet-stream
 
 body2
 --xxxx--
+epilogue
 EOS
-    m = MailParser::Message.new(msg, :extract_message_type=>true)
-    assert_equal("<from1@example.com>", m.from.to_s)
-    assert_equal(2, m.part.size)
-    assert_equal("text", m.part[0].type)
-    assert_equal("body1\n", m.part[0].body)
-    assert_equal("message", m.part[1].type)
-    assert_equal("<from2@example.com>", m.part[1].message.from.to_s)
-    assert_equal("body2", m.part[1].message.body)
-  end
-
-  def test_extract_message_type_header_only
-    msg = StringIO.new(<<EOS)
-From: from1@example.com
-Content-Type: multipart/mixed; boundary="xxxx"
-
---xxxx
-Content-Type: text/plain
-
-body1
-
---xxxx
-Content-Type: message/rfc822
-
-From: from2@example.com
-Content-Type: text/plain
---xxxx--
-EOS
-    m = MailParser::Message.new(msg, :extract_message_type=>true)
+    m = MailParser::Message.new(msg)
     assert_equal("<from1@example.com>", m.from.to_s)
     assert_equal(2, m.part.size)
     assert_equal("text", m.part[0].type)
+    assert_equal("plain", m.part[0].subtype)
     assert_equal("body1\n", m.part[0].body)
-    assert_equal("message", m.part[1].type)
-    assert_equal("<from2@example.com>", m.part[1].message.from.to_s)
-    assert_equal("", m.part[1].message.body)
+    assert_equal("application", m.part[1].type)
+    assert_equal("octet-stream", m.part[1].subtype)
+    assert_equal("body2", m.part[1].body)
   end
 
-  def test_extract_message_type_skip_body()
+  def test_extract_message_type()
     msg = StringIO.new(<<EOS)
 From: from1@example.com
 Content-Type: multipart/mixed; boundary="xxxx"
@@ -643,17 +623,17 @@ Content-Type: text/plain
 body2
 --xxxx--
 EOS
-    m = MailParser::Message.new(msg, :extract_message_type=>true, :skip_body=>true)
+    m = MailParser::Message.new(msg, :extract_message_type=>true)
     assert_equal("<from1@example.com>", m.from.to_s)
     assert_equal(2, m.part.size)
     assert_equal("text", m.part[0].type)
-    assert_equal("", m.part[0].body)
+    assert_equal("body1\n", m.part[0].body)
     assert_equal("message", m.part[1].type)
     assert_equal("<from2@example.com>", m.part[1].message.from.to_s)
-    assert_equal("", m.part[1].message.body)
+    assert_equal("body2", m.part[1].message.body)
   end
 
-  def test_extract_message_type_text_body_only()
+  def test_extract_message_type_header_only
     msg = StringIO.new(<<EOS)
 From: from1@example.com
 Content-Type: multipart/mixed; boundary="xxxx"
@@ -668,19 +648,16 @@ Content-Type: message/rfc822
 
 From: from2@example.com
 Content-Type: text/plain
-
-body2
-
 --xxxx--
 EOS
-    m = MailParser::Message.new(msg, :extract_message_type=>true, :text_body_only=>true)
+    m = MailParser::Message.new(msg, :extract_message_type=>true)
     assert_equal("<from1@example.com>", m.from.to_s)
     assert_equal(2, m.part.size)
     assert_equal("text", m.part[0].type)
     assert_equal("body1\n", m.part[0].body)
     assert_equal("message", m.part[1].type)
     assert_equal("<from2@example.com>", m.part[1].message.from.to_s)
-    assert_equal("body2\n", m.part[1].message.body)
+    assert_equal("", m.part[1].message.body)
   end
 
   def test_extract_multipart_alternative_attach()
@@ -714,8 +691,13 @@ EOS
     m = MailParser::Message.new(msg)
     assert_equal(2, m.part.size)
     assert_equal(2, m.part[0].part.size)
-    assert_equal("hoge\n", m.part[0].part[0].body)
-    assert_equal("fuga html\n", m.part[0].part[1].body)
+    if String.method_defined? :encode
+      assert_equal("hoge\n".encode('iso-2022-jp'), m.part[0].part[0].body)
+      assert_equal("fuga html\n".encode('iso-2022-jp'), m.part[0].part[1].body)
+    else
+      assert_equal("hoge\n", m.part[0].part[0].body)
+      assert_equal("fuga html\n", m.part[0].part[1].body)
+    end
     assert_equal("attached file", m.part[1].body)
   end
 
@@ -783,7 +765,7 @@ hogehoge
 EOS
     m = MailParser::Message.new msg
     assert_equal "hoge", m.subject
-    assert_equal "hogehoge\n", m.body
+    assert_equal "", m.body
   end
 
   def test_parse_header_only_part()
@@ -843,6 +825,39 @@ EOS
     assert_equal "hoge\n", m.part[1].body
   end
 
+  def test_parse_invalid_format_header
+    msg = StringIO.new <<EOS
+From: hoge@example.com
+invalid-header
+Subject: hoge
+
+body
+EOS
+    m = MailParser::Message.new msg
+    assert_equal 'hoge', m.subject
+    assert_equal "body\n", m.body
+  end
+
+  def test_parse_invalid_format_header2
+    msg = StringIO.new <<EOS.chomp
+From: hoge@example.com
+invalid-header
+EOS
+    m = MailParser::Message.new msg
+    assert_equal "From: hoge@example.com\ninvalid-header", m.rawheader
+  end
+
+  def test_rawheader
+    msg = StringIO.new(<<EOS)
+From: from@example.com\r
+Content-Type: text/plain\r
+\r
+hogehoge\r
+EOS
+    m = MailParser::Message.new msg
+    assert_equal "From: from@example.com\r\nContent-Type: text/plain\r\n", m.rawheader
+  end
+
   def test_raw_single_part
     msg = StringIO.new(<<EOS)
 From: from@example.com
@@ -852,7 +867,7 @@ hogehoge
 
 fugafuga
 EOS
-    m = MailParser::Message.new msg, :keep_raw=>true
+    m = MailParser::Message.new msg
     assert_equal msg.string, m.raw
   end
 
@@ -882,15 +897,15 @@ Content-Type: text/plain
 fuga
 --xxx--
 EOS
-    m = MailParser::Message.new msg, :keep_raw=>true
+    m = MailParser::Message.new msg
     assert_equal msg.string, m.raw
-    assert_equal <<EOS, m.part[0].part[0].raw
+    assert_equal <<EOS.chomp, m.part[0].part[0].raw
 Content-Type: text/plain
 
 hoge
 hoge
 EOS
-    assert_equal <<EOS, m.part[1].raw
+    assert_equal <<EOS.chomp, m.part[1].raw
 Content-Type: text/plain
 
 fuga
@@ -916,15 +931,14 @@ Content-Type: text/plain
 body2
 --xxxx--
 EOS
-    m = MailParser::Message.new msg, :keep_raw=>true
+    m = MailParser::Message.new msg
     assert_equal msg.string, m.raw
     assert_equal <<EOS, m.part[0].raw
 Content-Type: text/plain
 
 body1
-
 EOS
-    assert_equal <<EOS, m.part[1].raw
+    assert_equal <<EOS.chomp, m.part[1].raw
 Content-Type: message/rfc822
 
 From: from2@example.com
@@ -951,15 +965,14 @@ From: from2@example.com
 Content-Type: text/plain
 --xxxx--
 EOS
-    m = MailParser::Message.new msg, :keep_raw=>true
+    m = MailParser::Message.new msg
     assert_equal msg.string, m.raw
     assert_equal <<EOS, m.part[0].raw
 Content-Type: text/plain
 
 body1
-
 EOS
-    assert_equal <<EOS, m.part[1].raw
+    assert_equal <<EOS.chomp, m.part[1].raw
 Content-Type: message/rfc822
 
 From: from2@example.com
@@ -975,7 +988,7 @@ Subject: hogehoge
 
 body1
 EOS
-    m = MailParser::Message.new msg, :keep_raw=>true
+    m = MailParser::Message.new msg
     assert_equal <<EOS, m.raw
 From: from1@example.com
 Subject: hogehoge
@@ -985,97 +998,6 @@ body1
 EOS
   end
 
-  def test_raw_and_use_file
-    msg = StringIO.new(<<EOS)
-From: from@example.com
-Content-Type: text/plain
-
-hogehoge
-
-fugafuga
-EOS
-    m = MailParser::Message.new msg, :keep_raw=>true, :use_file=>1
-    assert_equal msg.string, m.raw
-  end
-end
-
-class TC_DelimIO < Test::Unit::TestCase
-  include MailParser
-  def setup()
-  end
-  def teardown()
-  end
-
-  def test_gets
-    s = StringIO.new <<EOS
-aaaa
-bbbb
-cccc
-dddd
-EOS
-    dio = DelimIO.new(DelimIO.new(s), ["cccc"])
-    assert_equal "aaaa\n", dio.gets
-    assert_equal "bbbb\n", dio.gets
-    assert_equal nil, dio.gets
-    assert_equal true, dio.eof?
-    dio.ungets
-    assert_equal false, dio.eof?
-    assert_equal "bbbb\n", dio.gets
-    assert_equal nil, dio.gets
-    assert_equal true, dio.eof?
-  end
-
-  def test_each_line
-    s = StringIO.new <<EOS
-aaaa
-bbbb
-cccc
-dddd
-EOS
-    dio = DelimIO.new(DelimIO.new(s))
-    ret = []
-    dio.each_line do |line|
-      ret << line
-    end
-    assert_equal ["aaaa\n","bbbb\n","cccc\n","dddd\n"], ret
-    assert_equal true, dio.eof?
-  end
-
-  def test_each_line_delim
-    s = StringIO.new <<EOS
-aaaa
-bbbb
-cccc
-dddd
-EOS
-    dio = DelimIO.new(DelimIO.new(s), ["cccc"])
-    ret = []
-    dio.each_line do |line|
-      ret << line
-    end
-    assert_equal ["aaaa\n","bbbb\n"], ret
-    assert_equal true, dio.eof?
-  end
-
-  def test_ungets
-    s = StringIO.new <<EOS
-aaaa
-bbbb
-cccc
-dddd
-EOS
-    dio = DelimIO.new(DelimIO.new(s), ["cccc"])
-    ret = []
-    dio.each_line do |line|
-      ret << line
-    end
-    assert_equal ["aaaa\n","bbbb\n"], ret
-    assert_equal true, dio.eof?
-    dio.ungets
-    assert_equal false, dio.eof?
-    assert_equal "bbbb\n", dio.gets
-  end
-
   def test_base64_body
     msg = <<EOS
 Content-Transfer-Encoding: base64
@@ -1111,20 +1033,3 @@ EOS
   end
 
 end
-
-class TC_DataBuffer < Test::Unit::TestCase
-  def test_string
-    b = MailParser::DataBuffer.new(nil)
-    assert_equal '', b.str
-    b << 'hogehoge'
-    b << 'fugafuga'
-    assert_equal 'hogehogefugafuga', b.str
-  end
-  def test_file
-    b = MailParser::DataBuffer.new(1)
-    assert_equal '', b.str
-    b << 'hogehoge'
-    b << 'fugafuga'
-    assert_equal 'hogehogefugafuga', b.str
-  end
-end

data/test/test_rfc2045.rb

@@ -3,7 +3,7 @@
 
 require "mailparser/rfc2045"
 
-unless ARGV.empty?
+if $0 == __FILE__ and not ARGV.empty?
   ARGV.each do |fname|
     File.open(fname) do |f|
       header = []

data/test/test_rfc2047.rb

@@ -1,4 +1,4 @@
-# -*- coding: utf-8 -*-
+# coding: ascii-8bit
 # Copyright (C) 2006-2010 TOMITA Masahiro
 # mailto:tommy@tmtm.org
 
@@ -39,19 +39,19 @@ class TC_RFC2047 < Test::Unit::TestCase
   end
 
   def test_split_decode_q_ascii()
-    s = MailParser::RFC2047.split_decode("=?us-ascii?q?hoge?=")
-    assert_equal(1, s.size)
-    assert_equal("hoge", s[0])
-    assert_equal("us-ascii", s[0].charset)
-    assert_equal("=?us-ascii?q?hoge?=", s[0].raw)
+    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
   end
 
   def test_split_decode_q_ascii_upcase()
-    s = MailParser::RFC2047.split_decode("=?US-ASCII?Q?hoge?=")
-    assert_equal(1, s.size)
-    assert_equal("hoge", s[0])
-    assert_equal("us-ascii", s[0].charset)
-    assert_equal("=?US-ASCII?Q?hoge?=", s[0].raw)
+    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
   end
 
   def test_decode_q_ascii()
@@ -69,6 +69,24 @@ class TC_RFC2047 < Test::Unit::TestCase
     assert_equal("abcdefg", s)
   end
 
+  def test_decode_with_output_charset
+    s = MailParser::RFC2047.decode('abcdefg', :output_charset=>'utf-8')
+    assert_equal Encoding::UTF_8, s.encoding if defined? Encoding
+    assert_equal('abcdefg', s)
+  end
+
+  def test_decode_raw_utf8
+    s = MailParser::RFC2047.decode('あいう')
+    assert_equal('あいう', s)
+  end
+
+  if defined? Encoding
+    def test_decode_raw_utf8_with_output_charset
+      s = MailParser::RFC2047.decode('あいう', :output_charset=>'utf-8')
+      assert_equal('あいう'.force_encoding('utf-8'), s)
+    end
+  end
+
   def test_decode_encode_plain()
     s = MailParser::RFC2047.decode("012345 =?us-ascii?q?hoge?= abcdefg")
     assert_equal("012345 hoge abcdefg", s)
@@ -101,12 +119,16 @@ class TC_RFC2047 < Test::Unit::TestCase
 
   def test_decode_sjis()
     s = MailParser::RFC2047.decode("=?sjis?b?h0A=?=", "UTF-8")
-    assert_equal("\xe2\x91\xa0", s)
+    if String.method_defined? :force_encoding
+      assert_equal("\xe2\x91\xa0".force_encoding('utf-8'), s)
+    end
   end
 
   def test_decode_iso2022jp()
     s = MailParser::RFC2047.decode("=?iso-2022-jp?b?GyRCLSEbKEI=?=", "UTF-8")
-    assert_equal("\xe2\x91\xa0", s)
+    if String.method_defined? :force_encoding
+      assert_equal("\xe2\x91\xa0".force_encoding('utf-8'), s)
+    end
   end
 
   def test_decode_charset_converter()

data/test/test_rfc2183.rb

@@ -3,7 +3,7 @@
 
 require "mailparser/rfc2183"
 
-unless ARGV.empty?
+if $0 == __FILE__ and not ARGV.empty?
   ARGV.each do |fname|
     File.open(fname) do |f|
       header = []

data/test/test_rfc2822.rb

@@ -5,7 +5,7 @@
 require "timeout"
 require "mailparser/rfc2822"
 
-unless ARGV.empty?
+if $0 == __FILE__ and not ARGV.empty?
   ARGV.each do |fname|
     File.open(fname) do |f|
       header = []
@@ -89,7 +89,11 @@ class TC_RFC2822Parser < Test::Unit::TestCase
     assert_equal("a@b.c", m.addr_spec.to_s)
 
     m = MailParser::RFC2822::Parser.new(:decode_mime_header=>true).parse(:MAILBOX, "=?euc-jp?q?=A4=A2=A4=A4?= <a@b.c>")
-    assert_equal("\xA4\xA2\xA4\xA4", m.display_name.to_s)
+    if String.method_defined? :encode
+      assert_equal("あい".encode('euc-jp'), m.display_name.to_s)
+    else
+      assert_equal("\xA4\xA2\xA4\xA4", m.display_name.to_s)
+    end
     assert_equal("a@b.c", m.addr_spec.to_s)
 
     m = MailParser::RFC2822::Parser.new(:decode_mime_header=>true, :output_charset=>"utf-8").parse(:MAILBOX, "=?euc-jp?q?=A4=A2=A4=A4?= <a@b.c>")

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: mailparser
 version: !ruby/object:Gem::Version
-  version: 0.4.22a
+  version: 0.5.0.beta1
   prerelease: 6
 platform: ruby
 authors:
@@ -9,16 +9,31 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2011-12-27 00:00:00.000000000 Z
-dependencies: []
-description: 
+date: 2012-09-17 00:00:00.000000000 Z
+dependencies:
+- !ruby/object:Gem::Dependency
+  name: mmapscanner
+  requirement: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+description: MailParser is a parser for mail message
 email: tommy@tmtm.org
 executables: []
 extensions: []
 extra_rdoc_files: []
 files:
 - README.txt
-- HISTORY
 - lib/mailparser.rb
 - lib/mailparser/rfc2183/scanner.rb
 - lib/mailparser/rfc2183/parser.rb
@@ -33,7 +48,6 @@ files:
 - lib/mailparser/rfc2822.rb
 - lib/mailparser/rfc2822/scanner.rb
 - lib/mailparser/rfc2822/parser.rb
-- lib/mailparser/obsolete.rb
 - lib/mailparser/rfc2231.rb
 - test.rb
 - test/test_loose.rb
@@ -41,7 +55,6 @@ files:
 - test/test_rfc2183.rb
 - test/test_rfc2047.rb
 - test/test_rfc2822.rb
-- test/test_obsolete.rb
 - test/test_mailparser.rb
 - test/test_rfc2045.rb
 homepage: http://github.com/tmtm/mailparser
@@ -65,7 +78,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: 1.3.1
 requirements: []
 rubyforge_project: 
-rubygems_version: 1.8.11
+rubygems_version: 1.8.23
 signing_key: 
 specification_version: 3
 summary: Mail Parser
@@ -76,6 +89,6 @@ test_files:
 - test/test_rfc2183.rb
 - test/test_rfc2047.rb
 - test/test_rfc2822.rb
-- test/test_obsolete.rb
 - test/test_mailparser.rb
 - test/test_rfc2045.rb
+has_rdoc: