mail 2.6.6 → 2.8.1

Sign up to get free protection for your applications and to get access to all the features.
Files changed (180) hide show
  1. checksums.yaml +5 -5
  2. data/MIT-LICENSE +1 -1
  3. data/README.md +134 -119
  4. data/lib/mail/attachments_list.rb +10 -9
  5. data/lib/mail/body.rb +73 -84
  6. data/lib/mail/check_delivery_params.rb +28 -21
  7. data/lib/mail/configuration.rb +2 -0
  8. data/lib/mail/constants.rb +27 -5
  9. data/lib/mail/elements/address.rb +53 -47
  10. data/lib/mail/elements/address_list.rb +11 -19
  11. data/lib/mail/elements/content_disposition_element.rb +9 -16
  12. data/lib/mail/elements/content_location_element.rb +6 -11
  13. data/lib/mail/elements/content_transfer_encoding_element.rb +6 -11
  14. data/lib/mail/elements/content_type_element.rb +16 -23
  15. data/lib/mail/elements/date_time_element.rb +7 -15
  16. data/lib/mail/elements/envelope_from_element.rb +22 -23
  17. data/lib/mail/elements/message_ids_element.rb +18 -13
  18. data/lib/mail/elements/mime_version_element.rb +7 -15
  19. data/lib/mail/elements/phrase_list.rb +12 -10
  20. data/lib/mail/elements/received_element.rb +27 -19
  21. data/lib/mail/encodings/7bit.rb +9 -14
  22. data/lib/mail/encodings/8bit.rb +2 -21
  23. data/lib/mail/encodings/base64.rb +11 -12
  24. data/lib/mail/encodings/binary.rb +3 -22
  25. data/lib/mail/encodings/identity.rb +24 -0
  26. data/lib/mail/encodings/quoted_printable.rb +6 -6
  27. data/lib/mail/encodings/transfer_encoding.rb +38 -29
  28. data/lib/mail/encodings/unix_to_unix.rb +3 -1
  29. data/lib/mail/encodings.rb +81 -54
  30. data/lib/mail/envelope.rb +11 -14
  31. data/lib/mail/field.rb +119 -98
  32. data/lib/mail/field_list.rb +60 -7
  33. data/lib/mail/fields/bcc_field.rb +34 -52
  34. data/lib/mail/fields/cc_field.rb +28 -49
  35. data/lib/mail/fields/comments_field.rb +27 -37
  36. data/lib/mail/fields/common_address_field.rb +170 -0
  37. data/lib/mail/fields/common_date_field.rb +58 -0
  38. data/lib/mail/fields/common_field.rb +77 -0
  39. data/lib/mail/fields/common_message_id_field.rb +42 -0
  40. data/lib/mail/fields/content_description_field.rb +7 -14
  41. data/lib/mail/fields/content_disposition_field.rb +13 -38
  42. data/lib/mail/fields/content_id_field.rb +24 -51
  43. data/lib/mail/fields/content_location_field.rb +11 -25
  44. data/lib/mail/fields/content_transfer_encoding_field.rb +31 -31
  45. data/lib/mail/fields/content_type_field.rb +50 -80
  46. data/lib/mail/fields/date_field.rb +23 -52
  47. data/lib/mail/fields/from_field.rb +28 -49
  48. data/lib/mail/fields/in_reply_to_field.rb +38 -49
  49. data/lib/mail/fields/keywords_field.rb +18 -31
  50. data/lib/mail/fields/message_id_field.rb +25 -71
  51. data/lib/mail/fields/mime_version_field.rb +19 -30
  52. data/lib/mail/fields/named_structured_field.rb +11 -0
  53. data/lib/mail/fields/named_unstructured_field.rb +11 -0
  54. data/lib/mail/fields/optional_field.rb +9 -7
  55. data/lib/mail/fields/{common/parameter_hash.rb → parameter_hash.rb} +13 -11
  56. data/lib/mail/fields/received_field.rb +43 -57
  57. data/lib/mail/fields/references_field.rb +35 -49
  58. data/lib/mail/fields/reply_to_field.rb +28 -49
  59. data/lib/mail/fields/resent_bcc_field.rb +28 -49
  60. data/lib/mail/fields/resent_cc_field.rb +28 -49
  61. data/lib/mail/fields/resent_date_field.rb +5 -30
  62. data/lib/mail/fields/resent_from_field.rb +28 -49
  63. data/lib/mail/fields/resent_message_id_field.rb +5 -29
  64. data/lib/mail/fields/resent_sender_field.rb +27 -56
  65. data/lib/mail/fields/resent_to_field.rb +28 -49
  66. data/lib/mail/fields/return_path_field.rb +50 -54
  67. data/lib/mail/fields/sender_field.rb +34 -55
  68. data/lib/mail/fields/structured_field.rb +3 -30
  69. data/lib/mail/fields/subject_field.rb +9 -11
  70. data/lib/mail/fields/to_field.rb +28 -49
  71. data/lib/mail/fields/unstructured_field.rb +32 -47
  72. data/lib/mail/header.rb +71 -110
  73. data/lib/mail/mail.rb +2 -10
  74. data/lib/mail/matchers/attachment_matchers.rb +15 -0
  75. data/lib/mail/matchers/has_sent_mail.rb +21 -1
  76. data/lib/mail/message.rb +113 -117
  77. data/lib/mail/multibyte/chars.rb +21 -178
  78. data/lib/mail/multibyte/unicode.rb +10 -10
  79. data/lib/mail/multibyte/utils.rb +26 -43
  80. data/lib/mail/multibyte.rb +55 -16
  81. data/lib/mail/network/delivery_methods/exim.rb +5 -4
  82. data/lib/mail/network/delivery_methods/file_delivery.rb +11 -10
  83. data/lib/mail/network/delivery_methods/logger_delivery.rb +34 -0
  84. data/lib/mail/network/delivery_methods/sendmail.rb +62 -21
  85. data/lib/mail/network/delivery_methods/smtp.rb +75 -50
  86. data/lib/mail/network/delivery_methods/smtp_connection.rb +3 -4
  87. data/lib/mail/network/delivery_methods/test_mailer.rb +4 -2
  88. data/lib/mail/network/retriever_methods/base.rb +8 -8
  89. data/lib/mail/network/retriever_methods/imap.rb +20 -7
  90. data/lib/mail/network/retriever_methods/pop3.rb +5 -3
  91. data/lib/mail/network/retriever_methods/test_retriever.rb +2 -1
  92. data/lib/mail/network.rb +1 -0
  93. data/lib/mail/parser_tools.rb +15 -0
  94. data/lib/mail/parsers/address_lists_parser.rb +33225 -116
  95. data/lib/mail/parsers/address_lists_parser.rl +179 -0
  96. data/lib/mail/parsers/content_disposition_parser.rb +882 -49
  97. data/lib/mail/parsers/content_disposition_parser.rl +89 -0
  98. data/lib/mail/parsers/content_location_parser.rb +809 -23
  99. data/lib/mail/parsers/content_location_parser.rl +78 -0
  100. data/lib/mail/parsers/content_transfer_encoding_parser.rb +509 -21
  101. data/lib/mail/parsers/content_transfer_encoding_parser.rl +71 -0
  102. data/lib/mail/parsers/content_type_parser.rb +1037 -56
  103. data/lib/mail/parsers/content_type_parser.rl +90 -0
  104. data/lib/mail/parsers/date_time_parser.rb +877 -25
  105. data/lib/mail/parsers/date_time_parser.rl +69 -0
  106. data/lib/mail/parsers/envelope_from_parser.rb +3669 -40
  107. data/lib/mail/parsers/envelope_from_parser.rl +89 -0
  108. data/lib/mail/parsers/message_ids_parser.rb +5146 -25
  109. data/lib/mail/parsers/message_ids_parser.rl +93 -0
  110. data/lib/mail/parsers/mime_version_parser.rb +497 -26
  111. data/lib/mail/parsers/mime_version_parser.rl +68 -0
  112. data/lib/mail/parsers/phrase_lists_parser.rb +870 -22
  113. data/lib/mail/parsers/phrase_lists_parser.rl +90 -0
  114. data/lib/mail/parsers/received_parser.rb +8776 -43
  115. data/lib/mail/parsers/received_parser.rl +91 -0
  116. data/lib/mail/parsers/rfc2045_content_transfer_encoding.rl +13 -0
  117. data/lib/mail/parsers/rfc2045_content_type.rl +25 -0
  118. data/lib/mail/parsers/rfc2045_mime.rl +16 -0
  119. data/lib/mail/parsers/rfc2183_content_disposition.rl +15 -0
  120. data/lib/mail/parsers/rfc3629_utf8.rl +19 -0
  121. data/lib/mail/parsers/rfc5234_abnf_core_rules.rl +22 -0
  122. data/lib/mail/parsers/rfc5322.rl +74 -0
  123. data/lib/mail/parsers/rfc5322_address.rl +72 -0
  124. data/lib/mail/parsers/{ragel/date_time.rl → rfc5322_date_time.rl} +8 -1
  125. data/lib/mail/parsers/rfc5322_lexical_tokens.rl +60 -0
  126. data/lib/mail/parsers.rb +11 -25
  127. data/lib/mail/part.rb +6 -10
  128. data/lib/mail/parts_list.rb +62 -6
  129. data/lib/mail/smtp_envelope.rb +57 -0
  130. data/lib/mail/utilities.rb +343 -74
  131. data/lib/mail/version.rb +2 -2
  132. data/lib/mail/yaml.rb +30 -0
  133. data/lib/mail.rb +5 -35
  134. metadata +111 -66
  135. data/CHANGELOG.rdoc +0 -803
  136. data/CONTRIBUTING.md +0 -60
  137. data/Dependencies.txt +0 -2
  138. data/Gemfile +0 -14
  139. data/Rakefile +0 -29
  140. data/TODO.rdoc +0 -9
  141. data/lib/mail/core_extensions/smtp.rb +0 -25
  142. data/lib/mail/core_extensions/string/access.rb +0 -146
  143. data/lib/mail/core_extensions/string/multibyte.rb +0 -79
  144. data/lib/mail/core_extensions/string.rb +0 -21
  145. data/lib/mail/fields/common/address_container.rb +0 -17
  146. data/lib/mail/fields/common/common_address.rb +0 -136
  147. data/lib/mail/fields/common/common_date.rb +0 -36
  148. data/lib/mail/fields/common/common_field.rb +0 -61
  149. data/lib/mail/fields/common/common_message_id.rb +0 -49
  150. data/lib/mail/multibyte/exceptions.rb +0 -9
  151. data/lib/mail/parsers/ragel/common.rl +0 -185
  152. data/lib/mail/parsers/ragel/parser_info.rb +0 -61
  153. data/lib/mail/parsers/ragel/ruby/machines/address_lists_machine.rb +0 -14864
  154. data/lib/mail/parsers/ragel/ruby/machines/address_lists_machine.rb.rl +0 -37
  155. data/lib/mail/parsers/ragel/ruby/machines/content_disposition_machine.rb +0 -751
  156. data/lib/mail/parsers/ragel/ruby/machines/content_disposition_machine.rb.rl +0 -37
  157. data/lib/mail/parsers/ragel/ruby/machines/content_location_machine.rb +0 -614
  158. data/lib/mail/parsers/ragel/ruby/machines/content_location_machine.rb.rl +0 -37
  159. data/lib/mail/parsers/ragel/ruby/machines/content_transfer_encoding_machine.rb +0 -447
  160. data/lib/mail/parsers/ragel/ruby/machines/content_transfer_encoding_machine.rb.rl +0 -37
  161. data/lib/mail/parsers/ragel/ruby/machines/content_type_machine.rb +0 -825
  162. data/lib/mail/parsers/ragel/ruby/machines/content_type_machine.rb.rl +0 -37
  163. data/lib/mail/parsers/ragel/ruby/machines/date_time_machine.rb +0 -817
  164. data/lib/mail/parsers/ragel/ruby/machines/date_time_machine.rb.rl +0 -37
  165. data/lib/mail/parsers/ragel/ruby/machines/envelope_from_machine.rb +0 -2149
  166. data/lib/mail/parsers/ragel/ruby/machines/envelope_from_machine.rb.rl +0 -37
  167. data/lib/mail/parsers/ragel/ruby/machines/message_ids_machine.rb +0 -1570
  168. data/lib/mail/parsers/ragel/ruby/machines/message_ids_machine.rb.rl +0 -37
  169. data/lib/mail/parsers/ragel/ruby/machines/mime_version_machine.rb +0 -440
  170. data/lib/mail/parsers/ragel/ruby/machines/mime_version_machine.rb.rl +0 -37
  171. data/lib/mail/parsers/ragel/ruby/machines/phrase_lists_machine.rb +0 -564
  172. data/lib/mail/parsers/ragel/ruby/machines/phrase_lists_machine.rb.rl +0 -37
  173. data/lib/mail/parsers/ragel/ruby/machines/rb_actions.rl +0 -51
  174. data/lib/mail/parsers/ragel/ruby/machines/received_machine.rb +0 -5144
  175. data/lib/mail/parsers/ragel/ruby/machines/received_machine.rb.rl +0 -37
  176. data/lib/mail/parsers/ragel/ruby/parser.rb.rl.erb +0 -37
  177. data/lib/mail/parsers/ragel/ruby.rb +0 -40
  178. data/lib/mail/parsers/ragel.rb +0 -18
  179. data/lib/mail/version_specific/ruby_1_8.rb +0 -126
  180. data/lib/mail/version_specific/ruby_1_9.rb +0 -226
@@ -1,37 +0,0 @@
1
- %%{
2
- machine received;
3
-
4
- include rb_actions "rb_actions.rl";
5
- include common "../../common.rl";
6
-
7
- getkey data_unpacked[p];
8
-
9
- main := received;
10
- }%%
11
-
12
- module Mail
13
- module Parsers
14
- module Ragel
15
- module ReceivedMachine
16
- %%write data;
17
-
18
- def self.parse(data)
19
- p = 0
20
- eof = data.length
21
- stack = []
22
-
23
- actions = []
24
- data_unpacked = data.bytes.to_a
25
- %%write init;
26
- %%write exec;
27
-
28
- if p == eof && cs >= %%{ write first_final; }%%
29
- return actions, nil
30
- else
31
- return [], "Only able to parse up to #{data[0..p]}"
32
- end
33
- end
34
- end
35
- end
36
- end
37
- end
@@ -1,37 +0,0 @@
1
- %%{
2
- machine <%= parser_name %>;
3
-
4
- include rb_actions "rb_actions.rl";
5
- include common "../../common.rl";
6
-
7
- getkey data_unpacked[p];
8
-
9
- main := <%= parser_name %>;
10
- }%%
11
-
12
- module Mail
13
- module Parsers
14
- module Ragel
15
- module <%= parser_name_cc %>
16
- %%write data;
17
-
18
- def self.parse(data)
19
- p = 0
20
- eof = data.length
21
- stack = []
22
-
23
- actions = []
24
- data_unpacked = data.bytes.to_a
25
- %%write init;
26
- %%write exec;
27
-
28
- if p == eof && cs >= %%{ write first_final; }%%
29
- return actions, nil
30
- else
31
- return [], "Only able to parse up to #{data[0..p]}"
32
- end
33
- end
34
- end
35
- end
36
- end
37
- end
@@ -1,40 +0,0 @@
1
- module Mail
2
- module Parsers
3
- module Ragel
4
- module Ruby
5
- def self.silence_warnings
6
- old, $VERBOSE = $VERBOSE, nil
7
- yield
8
- ensure
9
- $VERBOSE = old
10
- end
11
-
12
- # Ragel-generated parsers give a lot of warnings
13
- # and may cause logs to balloon in size
14
- silence_warnings do
15
- Mail::Parsers::Ragel::FIELD_PARSERS.each do |field_parser|
16
- require "mail/parsers/ragel/ruby/machines/#{field_parser}_machine"
17
- end
18
- end
19
-
20
- MACHINE_LIST = {
21
- :address_lists => AddressListsMachine,
22
- :phrase_lists => PhraseListsMachine,
23
- :date_time => DateTimeMachine,
24
- :received => ReceivedMachine,
25
- :message_ids => MessageIdsMachine,
26
- :envelope_from => EnvelopeFromMachine,
27
- :mime_version => MimeVersionMachine,
28
- :content_type => ContentTypeMachine,
29
- :content_disposition => ContentDispositionMachine,
30
- :content_transfer_encoding => ContentTransferEncodingMachine,
31
- :content_location => ContentLocationMachine
32
- }
33
-
34
- def self.parse(machine, string)
35
- MACHINE_LIST[machine].parse(string)
36
- end
37
- end
38
- end
39
- end
40
- end
@@ -1,18 +0,0 @@
1
- # frozen_string_literal: true
2
- module Mail
3
- module Parsers
4
- module Ragel
5
- require 'mail/parsers/ragel/parser_info'
6
- require "mail/parsers/ragel/ruby"
7
-
8
- def self.parse(machine, string)
9
- @machine_module ||= Ruby
10
- @machine_module.parse(machine, string)
11
- end
12
-
13
- def self.machine_module=(m)
14
- @machine_module = m
15
- end
16
- end
17
- end
18
- end
@@ -1,126 +0,0 @@
1
- # encoding: utf-8
2
- # frozen_string_literal: true
3
-
4
- module Mail
5
- class Ruby18
6
- require 'base64'
7
- require 'iconv'
8
-
9
- # Escapes any parenthesis in a string that are unescaped. This can't
10
- # use the Ruby 1.9.1 regexp feature of negative look behind so we have
11
- # to do two replacement, first unescape everything, then re-escape it
12
- def Ruby18.escape_paren( str )
13
- re = /\\\)/
14
- str = str.gsub(re) { |s| ')'}
15
- re = /\\\(/
16
- str = str.gsub(re) { |s| '('}
17
- re = /([\(\)])/ # Only match unescaped parens
18
- str.gsub(re) { |s| '\\' + s }
19
- end
20
-
21
- def Ruby18.paren( str )
22
- str = $1 if str =~ /^\((.*)?\)$/
23
- str = escape_paren( str )
24
- '(' + str + ')'
25
- end
26
-
27
- def Ruby18.escape_bracket( str )
28
- re = /\\\>/
29
- str = str.gsub(re) { |s| '>'}
30
- re = /\\\</
31
- str = str.gsub(re) { |s| '<'}
32
- re = /([\<\>])/ # Only match unescaped parens
33
- str.gsub(re) { |s| '\\' + s }
34
- end
35
-
36
- def Ruby18.bracket( str )
37
- str = $1 if str =~ /^\<(.*)?\>$/
38
- str = escape_bracket( str )
39
- '<' + str + '>'
40
- end
41
-
42
- def Ruby18.decode_base64(str)
43
- Base64.decode64(str) if str
44
- end
45
-
46
- def Ruby18.encode_base64(str)
47
- Base64.encode64(str)
48
- end
49
-
50
- def Ruby18.has_constant?(klass, string)
51
- klass.constants.include?( string )
52
- end
53
-
54
- def Ruby18.get_constant(klass, string)
55
- klass.const_get( string )
56
- end
57
-
58
- def Ruby18.transcode_charset(str, from_encoding, to_encoding = 'UTF-8')
59
- Iconv.conv("#{normalize_iconv_charset_encoding(to_encoding)}//IGNORE", normalize_iconv_charset_encoding(from_encoding), str)
60
- end
61
-
62
- def Ruby18.b_value_encode(str, encoding)
63
- # Ruby 1.8 requires an encoding to work
64
- raise ArgumentError, "Must supply an encoding" if encoding.nil?
65
- encoding = encoding.to_s.upcase.gsub('_', '-')
66
- [Encodings::Base64.encode(str), normalize_iconv_charset_encoding(encoding)]
67
- end
68
-
69
- def Ruby18.b_value_decode(str)
70
- match = str.match(/\=\?(.+)?\?[Bb]\?(.*)\?\=/m)
71
- if match
72
- encoding = match[1]
73
- str = Ruby18.decode_base64(match[2])
74
- str = transcode_charset(str, encoding)
75
- end
76
- str
77
- end
78
-
79
- def Ruby18.q_value_encode(str, encoding)
80
- # Ruby 1.8 requires an encoding to work
81
- raise ArgumentError, "Must supply an encoding" if encoding.nil?
82
- encoding = encoding.to_s.upcase.gsub('_', '-')
83
- [Encodings::QuotedPrintable.encode(str), encoding]
84
- end
85
-
86
- def Ruby18.q_value_decode(str)
87
- match = str.match(/\=\?(.+)?\?[Qq]\?(.*)\?\=/m)
88
- if match
89
- encoding = match[1]
90
- string = match[2].gsub(/_/, '=20')
91
- # Remove trailing = if it exists in a Q encoding
92
- string = string.sub(/\=$/, '')
93
- str = Encodings::QuotedPrintable.decode(string)
94
- str = transcode_charset(str, encoding)
95
- end
96
- str
97
- end
98
-
99
- def Ruby18.param_decode(str, encoding)
100
- URI.unescape(str)
101
- end
102
-
103
- def Ruby18.param_encode(str)
104
- encoding = $KCODE.to_s.downcase
105
- language = Configuration.instance.param_encode_language
106
- "#{encoding}'#{language}'#{URI.escape(str)}"
107
- end
108
-
109
- private
110
-
111
- def Ruby18.normalize_iconv_charset_encoding(encoding)
112
- case encoding.upcase
113
- when 'UTF8', 'UTF_8'
114
- 'UTF-8'
115
- when 'UTF16', 'UTF-16'
116
- 'UTF-16BE'
117
- when 'UTF32', 'UTF-32'
118
- 'UTF-32BE'
119
- when 'KS_C_5601-1987'
120
- 'CP949'
121
- else
122
- encoding
123
- end
124
- end
125
- end
126
- end
@@ -1,226 +0,0 @@
1
- # encoding: utf-8
2
- # frozen_string_literal: true
3
-
4
- module Mail
5
- class Ruby19
6
- class StrictCharsetEncoder
7
- def encode(string, charset)
8
- string.force_encoding(Mail::Ruby19.pick_encoding(charset))
9
- end
10
- end
11
-
12
- class BestEffortCharsetEncoder
13
- def encode(string, charset)
14
- string.force_encoding(pick_encoding(charset))
15
- end
16
-
17
- private
18
-
19
- def pick_encoding(charset)
20
- charset = case charset
21
- when /ansi_x3.110-1983/
22
- 'ISO-8859-1'
23
- when /Windows-?1258/i # Windows-1258 is similar to 1252
24
- "Windows-1252"
25
- else
26
- charset
27
- end
28
- Mail::Ruby19.pick_encoding(charset)
29
- end
30
- end
31
-
32
- class << self
33
- attr_accessor :charset_encoder
34
- end
35
- self.charset_encoder = StrictCharsetEncoder.new
36
-
37
- # Escapes any parenthesis in a string that are unescaped this uses
38
- # a Ruby 1.9.1 regexp feature of negative look behind
39
- def Ruby19.escape_paren( str )
40
- re = /(?<!\\)([\(\)])/ # Only match unescaped parens
41
- str.gsub(re) { |s| '\\' + s }
42
- end
43
-
44
- def Ruby19.paren( str )
45
- str = $1 if str =~ /^\((.*)?\)$/
46
- str = escape_paren( str )
47
- '(' + str + ')'
48
- end
49
-
50
- def Ruby19.escape_bracket( str )
51
- re = /(?<!\\)([\<\>])/ # Only match unescaped brackets
52
- str.gsub(re) { |s| '\\' + s }
53
- end
54
-
55
- def Ruby19.bracket( str )
56
- str = $1 if str =~ /^\<(.*)?\>$/
57
- str = escape_bracket( str )
58
- '<' + str + '>'
59
- end
60
-
61
- def Ruby19.decode_base64(str)
62
- if !str.end_with?("=") && str.length % 4 != 0
63
- str = str.ljust((str.length + 3) & ~3, "=")
64
- end
65
- str.unpack( 'm' ).first
66
- end
67
-
68
- def Ruby19.encode_base64(str)
69
- [str].pack( 'm' )
70
- end
71
-
72
- def Ruby19.has_constant?(klass, string)
73
- klass.const_defined?( string, false )
74
- end
75
-
76
- def Ruby19.get_constant(klass, string)
77
- klass.const_get( string )
78
- end
79
-
80
- def Ruby19.transcode_charset(str, from_encoding, to_encoding = Encoding::UTF_8)
81
- charset_encoder.encode(str.dup, from_encoding).encode(to_encoding, :undef => :replace, :invalid => :replace, :replace => '')
82
- end
83
-
84
- def Ruby19.b_value_encode(str, encoding = nil)
85
- encoding = str.encoding.to_s
86
- [Ruby19.encode_base64(str), encoding]
87
- end
88
-
89
- def Ruby19.b_value_decode(str)
90
- match = str.match(/\=\?(.+)?\?[Bb]\?(.*)\?\=/m)
91
- if match
92
- charset = match[1]
93
- str = Ruby19.decode_base64(match[2])
94
- str = charset_encoder.encode(str, charset)
95
- end
96
- decoded = str.encode(Encoding::UTF_8, :invalid => :replace, :replace => "")
97
- decoded.valid_encoding? ? decoded : decoded.encode(Encoding::UTF_16LE, :invalid => :replace, :replace => "").encode(Encoding::UTF_8)
98
- rescue Encoding::UndefinedConversionError, ArgumentError, Encoding::ConverterNotFoundError
99
- warn "Encoding conversion failed #{$!}"
100
- str.dup.force_encoding(Encoding::UTF_8)
101
- end
102
-
103
- def Ruby19.q_value_encode(str, encoding = nil)
104
- encoding = str.encoding.to_s
105
- [Encodings::QuotedPrintable.encode(str), encoding]
106
- end
107
-
108
- def Ruby19.q_value_decode(str)
109
- match = str.match(/\=\?(.+)?\?[Qq]\?(.*)\?\=/m)
110
- if match
111
- charset = match[1]
112
- string = match[2].gsub(/_/, '=20')
113
- # Remove trailing = if it exists in a Q encoding
114
- string = string.sub(/\=$/, '')
115
- str = Encodings::QuotedPrintable.decode(string)
116
- str = charset_encoder.encode(str, charset)
117
- # We assume that binary strings hold utf-8 directly to work around
118
- # jruby/jruby#829 which subtly changes String#encode semantics.
119
- str.force_encoding(Encoding::UTF_8) if str.encoding == Encoding::ASCII_8BIT
120
- end
121
- decoded = str.encode(Encoding::UTF_8, :invalid => :replace, :replace => "")
122
- decoded.valid_encoding? ? decoded : decoded.encode(Encoding::UTF_16LE, :invalid => :replace, :replace => "").encode(Encoding::UTF_8)
123
- rescue Encoding::UndefinedConversionError, ArgumentError, Encoding::ConverterNotFoundError
124
- warn "Encoding conversion failed #{$!}"
125
- str.dup.force_encoding(Encoding::UTF_8)
126
- end
127
-
128
- def Ruby19.param_decode(str, encoding)
129
- str = uri_parser.unescape(str)
130
- str = charset_encoder.encode(str, encoding) if encoding
131
- str
132
- end
133
-
134
- def Ruby19.param_encode(str)
135
- encoding = str.encoding.to_s.downcase
136
- language = Configuration.instance.param_encode_language
137
- "#{encoding}'#{language}'#{uri_parser.escape(str)}"
138
- end
139
-
140
- def Ruby19.uri_parser
141
- @uri_parser ||= URI::Parser.new
142
- end
143
-
144
- # Pick a Ruby encoding corresponding to the message charset. Most
145
- # charsets have a Ruby encoding, but some need manual aliasing here.
146
- #
147
- # TODO: add this as a test somewhere:
148
- # Encoding.list.map { |e| [e.to_s.upcase == pick_encoding(e.to_s.downcase.gsub("-", "")), e.to_s] }.select {|a,b| !b}
149
- # Encoding.list.map { |e| [e.to_s == pick_encoding(e.to_s), e.to_s] }.select {|a,b| !b}
150
- def Ruby19.pick_encoding(charset)
151
- charset = charset.to_s
152
- encoding = case charset.downcase
153
-
154
- # ISO-8859-8-I etc. http://en.wikipedia.org/wiki/ISO-8859-8-I
155
- when /^iso[-_]?8859-(\d+)(-i)?$/
156
- "ISO-8859-#{$1}"
157
-
158
- # ISO-8859-15, ISO-2022-JP and alike
159
- when /^iso[-_]?(\d{4})-?(\w{1,2})$/
160
- "ISO-#{$1}-#{$2}"
161
-
162
- # "ISO-2022-JP-KDDI" and alike
163
- when /^iso[-_]?(\d{4})-?(\w{1,2})-?(\w*)$/
164
- "ISO-#{$1}-#{$2}-#{$3}"
165
-
166
- # UTF-8, UTF-32BE and alike
167
- when /^utf[\-_]?(\d{1,2})?(\w{1,2})$/
168
- "UTF-#{$1}#{$2}".gsub(/\A(UTF-(?:16|32))\z/, '\\1BE')
169
-
170
- # Windows-1252 and alike
171
- when /^windows-?(.*)$/
172
- "Windows-#{$1}"
173
-
174
- when '8bit'
175
- Encoding::ASCII_8BIT
176
-
177
- # alternatives/misspellings of us-ascii seen in the wild
178
- when /^iso[-_]?646(-us)?$/, 'us=ascii'
179
- Encoding::ASCII
180
-
181
- # Microsoft-specific alias for MACROMAN
182
- when 'macintosh'
183
- Encoding::MACROMAN
184
-
185
- # Microsoft-specific alias for CP949 (Korean)
186
- when 'ks_c_5601-1987'
187
- Encoding::CP949
188
-
189
- # Wrongly written Shift_JIS (Japanese)
190
- when 'shift-jis'
191
- Encoding::Shift_JIS
192
-
193
- # GB2312 (Chinese charset) is a subset of GB18030 (its replacement)
194
- when 'gb2312'
195
- Encoding::GB18030
196
-
197
- when 'cp-850'
198
- Encoding::CP850
199
-
200
- when 'latin2'
201
- Encoding::ISO_8859_2
202
-
203
- else
204
- charset
205
- end
206
-
207
- convert_to_encoding(encoding)
208
- end
209
-
210
- class << self
211
- private
212
-
213
- def convert_to_encoding(encoding)
214
- if encoding.is_a?(Encoding)
215
- encoding
216
- else
217
- begin
218
- Encoding.find(encoding)
219
- rescue ArgumentError
220
- Encoding::BINARY
221
- end
222
- end
223
- end
224
- end
225
- end
226
- end