mailparser 0.4.22a → 0.5.0.beta1

data/README.txt

@@ -87,9 +87,6 @@ opt は Hash オブジェクトで次の値を指定できる。
  :strict => true
    RFC違反時に ParseError 例外を発生する。デフォルトは false。
 
- :keep_raw => true
-   生メッセージ文字列を保持する。デフォルトは false。
-
  :use_file => bytes
    raw が指定したサイズを超えたら、メモリではなくファイルを使う。nil 指定時は無制限にメモリを使う。デフォルトは nil。
 
@@ -307,10 +304,9 @@ m.rawheader  # => "From: TOMITA Masahiro <tommy@tmtm.org>\nTo: foo@example.com\n
 
 === raw ===
 
-生メッセージ文字列を返す。:keep_raw=>true である必要がある。
-:keep_raw=>false の場合は空文字列が返る。
+生メッセージ文字列を返す。
 {{{
-m = MailParser::Message.new(StringIO.new(<<EOS), :keep_raw=>true)
+m = MailParser::Message.new(StringIO.new(<<EOS))
 From: TOMITA Masahiro <tommy@tmtm.org>
 To: foo@example.com
 Subject: subject

data/lib/mailparser/conv_charset.rb

@@ -1,27 +1,48 @@
-# Copyright (C) 2007-2010 TOMITA Masahiro
+# coding: ascii-8bit
+# Copyright (C) 2007-2011 TOMITA Masahiro
 # mailto:tommy@tmtm.org
 
-require "iconv"
-require "nkf"
+module MailParser
+end
 
 class MailParser::ConvCharset
-  CHARSET = {
-    "sjis"      => "cp932",
-    "x-sjis"    => "cp932",
-    "shift_jis" => "cp932",
-    "shift-jis" => "cp932",
-  }
-  def self.conv_charset(from, to, str)
-    from = CHARSET[from.downcase] || from.downcase
-    to = CHARSET[to.downcase] || to.downcase
-    s = str
-    if from == "iso-2022-jp" then
-      s = NKF.nkf("-m0Jxs", str)
-      from = "cp932"
+  if String.method_defined? :encode
+    CHARSET = {
+      "sjis"        => "cp932",
+      "x-sjis"      => "cp932",
+      "shift_jis"   => "cp932",
+      "shift-jis"   => "cp932",
+      "iso-2022-jp" => "cp50221",
+    }
+
+    def self.conv_charset(from, to, str)
+      from = CHARSET[from.downcase] || from.downcase
+      to = CHARSET[to.downcase] || to.downcase
+      str.encode(to, from, :invalid=>:replace, :undef=>:replace, :replace=>'?')
     end
-    if to == "iso-2022-jp" then
-      return NKF.nkf("-m0Sxj", Iconv.iconv("cp932", from, s)[0])
+  else
+    require "iconv"
+    require "nkf"
+
+    CHARSET = {
+      "sjis"        => "cp932",
+      "x-sjis"      => "cp932",
+      "shift_jis"   => "cp932",
+      "shift-jis"   => "cp932",
+    }
+
+    def self.conv_charset(from, to, str)
+      from = CHARSET[from.downcase] || from.downcase
+      to = CHARSET[to.downcase] || to.downcase
+      s = str
+      if from == "iso-2022-jp" then
+        s = NKF.nkf("-m0Jxs", str)
+        from = "cp932"
+      end
+      if to == "iso-2022-jp" then
+        return NKF.nkf("-m0Sxj", Iconv.iconv("cp932", from, s)[0])
+      end
+      return Iconv.iconv(to, from, s)[0]
     end
-    return Iconv.iconv(to, from, s)[0]
   end
 end

data/lib/mailparser/error.rb

@@ -1,3 +1,4 @@
+# coding: ascii-8bit
 # Copyright (C) 2006-2010 TOMITA Masahiro
 # mailto:tommy@tmtm.org
 

data/lib/mailparser/loose.rb

@@ -1,4 +1,4 @@
-# -*- coding: utf-8 -*-
+# coding: ascii-8bit
 # Copyright (C) 2007-2010 TOMITA Masahiro
 # mailto:tommy@tmtm.org
 
@@ -6,6 +6,8 @@ require "time"
 require "strscan"
 require "mailparser/rfc2822"
 require "mailparser/rfc2045"
+require "mailparser/rfc2047"
+require "mailparser/rfc2183"
 
 module MailParser
   module Loose
@@ -109,7 +111,7 @@ module MailParser
       v = ""
       unless a.empty?
         while i < a[0].length do
-          if a[0][i] =~ /\A[a-z0-9]+\z/ino then
+          if a[0][i] =~ /\A[a-z0-9]+\z/i then
             v = a[0][i+1]
             name_val[a[0][i].downcase] = v
             i += 1
@@ -125,10 +127,10 @@ module MailParser
     # Content-Type ヘッダをパースして RFC2045::ContentType を返す
     def parse_content_type(str, opt={})
       token = split_by(Tokenizer.token(str), ";")
-      type, subtype = token.shift.to_s.split("/", 2)
+      type, subtype = token.empty? ? nil : token.shift.join.split("/", 2)
       params = {}
-      token.map do |param|
-        pn, pv = param.to_s.split(/=/, 2)
+      token.each do |param|
+        pn, pv = param.join.split(/=/, 2)
         params[pn.to_s] = pv.to_s.gsub(/\A"|"\z/,"")
       end
       type = "text" if type.nil? or type.empty?
@@ -151,10 +153,10 @@ module MailParser
     # Content-Disposition ヘッダをパースして RFC2183::ContentDisposition を返す
     def parse_content_disposition(str, opt={})
       token = split_by(Tokenizer.token(str), ";")
-      type = token.shift.to_s
+      type = token.empty? ? '' : token.shift.join
       params = {}
-      token.map do |param|
-        pn, pv = param.to_s.split(/=/, 2)
+      token.each do |param|
+        pn, pv = param.join.split(/=/, 2)
         params[pn.to_s] = pv.to_s.gsub(/\A"|"\z/,"")
       end
       RFC2183::ContentDisposition.new(type, params)
@@ -185,11 +187,11 @@ module MailParser
           if opt[:decode_mime_header] then
             display_name = RFC2047.decode(display_name, opt)
           end
-          mailaddr = m[a1+1..a2-1].to_s
+          mailaddr = m[a1+1..a2-1].join
           local_part, domain = mailaddr.split(/@/, 2)
           ret << RFC2822::Mailbox.new(RFC2822::AddrSpec.new(local_part, domain), display_name)
         else
-          local_part, domain = m.to_s.split(/@/, 2)
+          local_part, domain = m.join.split(/@/, 2)
           ret << RFC2822::Mailbox.new(RFC2822::AddrSpec.new(local_part, domain))
         end
       end
@@ -233,9 +235,9 @@ module MailParser
       def token()
         token = []
         while @ss.rest? do
-          if s = @ss.scan(/\s+/nmo) then
+          if s = @ss.scan(/\s+/) then
             # ignore
-          elsif s = @ss.scan(/\(/nmo) then
+          elsif s = @ss.scan(/\(/) then
             begin
               pos = @ss.pos
               cfws(@ss)
@@ -243,12 +245,12 @@ module MailParser
               @ss.pos = pos
               token << s
             end
-          elsif s = @ss.scan(/\"(\s*(\\[#{TEXT_RE}]|[#{QTEXT_RE}]))*\s*\"/nmo) ||
-              @ss.scan(/\[(\s*(\\[#{TEXT_RE}]|[#{DTEXT_RE}]))*\s*\]/nmo) ||
-              @ss.scan(/[#{ATEXT_RE}]+/no)
+          elsif s = @ss.scan(/\"(\s*(\\[#{TEXT_RE}]|[#{QTEXT_RE}]))*\s*\"/o) ||
+              @ss.scan(/\[(\s*(\\[#{TEXT_RE}]|[#{DTEXT_RE}]))*\s*\]/o) ||
+              @ss.scan(/[#{ATEXT_RE}]+/o)
             token << s
           else
-            token << @ss.scan(/./no)
+            token << @ss.scan(/./)
           end
         end
         return token
@@ -258,9 +260,9 @@ module MailParser
       def token_received()
         ret = []
         while @ss.rest? do
-          if s = @ss.scan(/[\s]+/nmo) then
+          if s = @ss.scan(/[\s]+/) then
             # ignore blank
-          elsif s = @ss.scan(/\(/nmo) then
+          elsif s = @ss.scan(/\(/) then
             begin
               pos = @ss.pos
               cfws(@ss)
@@ -268,12 +270,12 @@ module MailParser
               @ss.pos = pos
               ret.last << s unless ret.empty?
             end
-          elsif s = @ss.scan(/\"([\s]*(\\[#{TEXT_RE}]|[#{QTEXT_RE}]))*[\s]*\"/nmo)
+          elsif s = @ss.scan(/\"([\s]*(\\[#{TEXT_RE}]|[#{QTEXT_RE}]))*[\s]*\"/o)
             ret << s
           elsif s = @ss.scan(/;/)
             ret << s
           else
-            ret << @ss.scan(/[^\s\(\;]+/nmo)
+            ret << @ss.scan(/[^\s\(\;]+/o)
           end
         end
         return ret

data/lib/mailparser/rfc2045/parser.rb

@@ -1,19 +1,16 @@
+# coding: ascii-8bit
 #
 # DO NOT MODIFY!!!!
-# This file is automatically generated by racc 1.4.5
-# from racc grammer file "lib/mailparser/rfc2045/parser.y".
+# This file is automatically generated by Racc 1.4.7
+# from Racc grammer file "".
 #
 
-require 'racc/parser'
-
-
+require 'racc/parser.rb'
 module MailParser
-
   module RFC2045
-
     class Parser < Racc::Parser
 
-module_eval <<'..end lib/mailparser/rfc2045/parser.y modeval..idcdf01fc9ea', 'lib/mailparser/rfc2045/parser.y', 62
+module_eval(<<'...end parser.y/module_eval...', 'parser.y', 62)
 
 require "mailparser/rfc2045/scanner"
 
@@ -30,7 +27,7 @@ def parse(header_type, value)
 end
 
 def parse_sub(&block)
-  yield @header_type, nil
+  yield [@header_type, nil]
   @scanner.scan(&block)
 end
 
@@ -39,207 +36,210 @@ def on_error(t, val, vstack)
 #  p racc_token2str(t)
   raise MailParser::ParseError, val+@scanner.rest
 end
-..end lib/mailparser/rfc2045/parser.y modeval..idcdf01fc9ea
-
-##### racc 1.4.5 generates ###
-
-racc_reduce_table = [
- 0, 0, :racc_error,
- 2, 13, :_reduce_1,
- 2, 13, :_reduce_2,
- 2, 13, :_reduce_3,
- 4, 14, :_reduce_4,
- 1, 15, :_reduce_5,
- 3, 16, :_reduce_6,
- 1, 20, :_reduce_none,
- 1, 17, :_reduce_none,
- 1, 18, :_reduce_none,
- 0, 19, :_reduce_10,
- 3, 19, :_reduce_11,
- 3, 21, :_reduce_12,
- 1, 22, :_reduce_none,
- 1, 23, :_reduce_none,
- 1, 23, :_reduce_none ]
-
-racc_reduce_n = 16
-
-racc_shift_n = 29
+...end parser.y/module_eval...
+##### State transition tables begin ###
 
 racc_action_table = [
-    28,     8,    10,    26,     1,     3,     4,    13,    14,    15,
-     7,    18,    19,    21,    23,    25,    16 ]
+    27,     8,    11,    28,     2,     3,     4,    13,    14,    15,
+     5,    18,    19,    21,    24,    25,    16 ]
 
 racc_action_check = [
-    25,     2,     3,    25,     0,     0,     0,     4,     6,     8,
-     1,    14,    16,    20,    21,    22,    13 ]
+    25,     2,     3,    25,     0,     0,     0,     4,     5,     7,
+     1,    15,    16,    20,    21,    23,    13 ]
 
 racc_action_pointer = [
-     2,     2,     1,    -6,     1,   nil,     3,   nil,     9,   nil,
-   nil,   nil,   nil,     9,     3,   nil,     6,   nil,   nil,   nil,
-     4,     6,     5,   nil,   nil,    -8,   nil,   nil,   nil ]
+     2,    10,    -7,    -6,     1,     8,   nil,     4,   nil,   nil,
+   nil,   nil,   nil,     9,   nil,     3,     6,   nil,   nil,   nil,
+     4,     6,   nil,     5,   nil,    -8,   nil,   nil,   nil ]
 
 racc_action_default = [
-   -16,   -16,   -16,   -16,   -16,    -1,   -16,    -8,   -16,    -2,
-    -7,    -5,    -3,   -16,   -16,    29,   -16,   -10,    -9,    -6,
-    -4,   -16,   -16,   -13,   -11,   -16,   -15,   -12,   -14 ]
+   -16,   -16,   -16,   -16,   -16,   -16,    -1,   -16,    -8,    -2,
+    -5,    -7,    -3,   -16,    29,   -16,   -16,   -10,    -9,    -6,
+    -4,   -16,   -11,   -16,   -13,   -16,   -12,   -14,   -15 ]
 
 racc_goto_table = [
-     2,     5,     9,    12,     6,    17,    20,    11,    24,    22,
-    27 ]
+     1,     6,     9,    12,     7,    17,    20,    10,    22,    23,
+    26 ]
 
 racc_goto_check = [
      1,     2,     3,     4,     5,     6,     7,     8,     9,    10,
     11 ]
 
 racc_goto_pointer = [
-   nil,     0,     0,    -1,    -1,     3,    -9,   -11,     4,   -13,
+   nil,     0,    -1,    -1,    -1,     2,   -10,   -11,     4,   -13,
    -12,   -15 ]
 
 racc_goto_default = [
    nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,
    nil,   nil ]
 
-racc_token_table = {
- false => 0,
- Object.new => 1,
- :CONTENT_TYPE => 2,
- :CONTENT_TRANSFER_ENCODING => 3,
- :MIME_VERSION => 4,
- "/" => 5,
- :DIGIT => 6,
- "." => 7,
- :TOKEN => 8,
- ";" => 9,
- "=" => 10,
- :QUOTED_STRING => 11 }
+racc_reduce_table = [
+  0, 0, :racc_error,
+  2, 13, :_reduce_1,
+  2, 13, :_reduce_2,
+  2, 13, :_reduce_3,
+  4, 14, :_reduce_4,
+  1, 15, :_reduce_5,
+  3, 16, :_reduce_6,
+  1, 20, :_reduce_none,
+  1, 17, :_reduce_none,
+  1, 18, :_reduce_none,
+  0, 19, :_reduce_10,
+  3, 19, :_reduce_11,
+  3, 21, :_reduce_12,
+  1, 22, :_reduce_none,
+  1, 23, :_reduce_none,
+  1, 23, :_reduce_none ]
 
-racc_use_result_var = false
+racc_reduce_n = 16
+
+racc_shift_n = 29
+
+racc_token_table = {
+  false => 0,
+  :error => 1,
+  :CONTENT_TYPE => 2,
+  :CONTENT_TRANSFER_ENCODING => 3,
+  :MIME_VERSION => 4,
+  "/" => 5,
+  :DIGIT => 6,
+  "." => 7,
+  :TOKEN => 8,
+  ";" => 9,
+  "=" => 10,
+  :QUOTED_STRING => 11 }
 
 racc_nt_base = 12
 
+racc_use_result_var = false
+
 Racc_arg = [
- racc_action_table,
- racc_action_check,
- racc_action_default,
- racc_action_pointer,
- racc_goto_table,
- racc_goto_check,
- racc_goto_default,
- racc_goto_pointer,
- racc_nt_base,
- racc_reduce_table,
- racc_token_table,
- racc_shift_n,
- racc_reduce_n,
- racc_use_result_var ]
+  racc_action_table,
+  racc_action_check,
+  racc_action_default,
+  racc_action_pointer,
+  racc_goto_table,
+  racc_goto_check,
+  racc_goto_default,
+  racc_goto_pointer,
+  racc_nt_base,
+  racc_reduce_table,
+  racc_token_table,
+  racc_shift_n,
+  racc_reduce_n,
+  racc_use_result_var ]
 
 Racc_token_to_s_table = [
-'$end',
-'error',
-'CONTENT_TYPE',
-'CONTENT_TRANSFER_ENCODING',
-'MIME_VERSION',
-'"/"',
-'DIGIT',
-'"."',
-'TOKEN',
-'";"',
-'"="',
-'QUOTED_STRING',
-'$start',
-'all',
-'content_type',
-'content_transfer_encoding',
-'mime_version',
-'type',
-'subtype',
-'parameter_list',
-'mechanism',
-'parameter',
-'attribute',
-'value']
+  "$end",
+  "error",
+  "CONTENT_TYPE",
+  "CONTENT_TRANSFER_ENCODING",
+  "MIME_VERSION",
+  "\"/\"",
+  "DIGIT",
+  "\".\"",
+  "TOKEN",
+  "\";\"",
+  "\"=\"",
+  "QUOTED_STRING",
+  "$start",
+  "all",
+  "content_type",
+  "content_transfer_encoding",
+  "mime_version",
+  "type",
+  "subtype",
+  "parameter_list",
+  "mechanism",
+  "parameter",
+  "attribute",
+  "value" ]
 
 Racc_debug_parser = false
 
-##### racc system variables end #####
+##### State transition tables end #####
 
- # reduce 0 omitted
+# reduce 0 omitted
 
-module_eval <<'.,.,', 'lib/mailparser/rfc2045/parser.y', 9
-  def _reduce_1( val, _values)
-val[1]
+module_eval(<<'.,.,', 'parser.y', 9)
+  def _reduce_1(val, _values)
+    val[1]
   end
 .,.,
 
-module_eval <<'.,.,', 'lib/mailparser/rfc2045/parser.y', 11
-  def _reduce_2( val, _values)
-val[1]
+module_eval(<<'.,.,', 'parser.y', 11)
+  def _reduce_2(val, _values)
+    val[1]
   end
 .,.,
 
-module_eval <<'.,.,', 'lib/mailparser/rfc2045/parser.y', 13
-  def _reduce_3( val, _values)
-val[1]
+module_eval(<<'.,.,', 'parser.y', 13)
+  def _reduce_3(val, _values)
+    val[1]
   end
 .,.,
 
-module_eval <<'.,.,', 'lib/mailparser/rfc2045/parser.y', 19
-  def _reduce_4( val, _values)
-                    ContentType.new(val[0], val[2], val[3])
+module_eval(<<'.,.,', 'parser.y', 17)
+  def _reduce_4(val, _values)
+                        ContentType.new(val[0], val[2], val[3])
+                  
   end
 .,.,
 
-module_eval <<'.,.,', 'lib/mailparser/rfc2045/parser.y', 24
-  def _reduce_5( val, _values)
-                    ContentTransferEncoding.new(val[0])
+module_eval(<<'.,.,', 'parser.y', 22)
+  def _reduce_5(val, _values)
+                        ContentTransferEncoding.new(val[0])
+                  
   end
 .,.,
 
-module_eval <<'.,.,', 'lib/mailparser/rfc2045/parser.y', 31
-  def _reduce_6( val, _values)
-                    val.join
+module_eval(<<'.,.,', 'parser.y', 29)
+  def _reduce_6(val, _values)
+                        val.join
+                  
   end
 .,.,
 
- # reduce 7 omitted
+# reduce 7 omitted
 
- # reduce 8 omitted
+# reduce 8 omitted
 
- # reduce 9 omitted
+# reduce 9 omitted
 
-module_eval <<'.,.,', 'lib/mailparser/rfc2045/parser.y', 42
-  def _reduce_10( val, _values)
-                    {}
+module_eval(<<'.,.,', 'parser.y', 40)
+  def _reduce_10(val, _values)
+                        {}
+                  
   end
 .,.,
 
-module_eval <<'.,.,', 'lib/mailparser/rfc2045/parser.y', 49
-  def _reduce_11( val, _values)
-                    pn, pv = val[2]
+module_eval(<<'.,.,', 'parser.y', 44)
+  def _reduce_11(val, _values)
+                        pn, pv = val[2]
                     pv = $1 if pv =~ /\A\"(.*)\"\Z/m
                     val[0][pn] = pv.gsub(/\s*\n\s*/, " ")
                     val[0]
+                  
   end
 .,.,
 
-module_eval <<'.,.,', 'lib/mailparser/rfc2045/parser.y', 54
-  def _reduce_12( val, _values)
-                    [val[0].downcase, val[2]]
+module_eval(<<'.,.,', 'parser.y', 52)
+  def _reduce_12(val, _values)
+                        [val[0].downcase, val[2]]
+                  
   end
 .,.,
 
- # reduce 13 omitted
+# reduce 13 omitted
 
- # reduce 14 omitted
+# reduce 14 omitted
 
- # reduce 15 omitted
+# reduce 15 omitted
 
- def _reduce_none( val, _values)
+def _reduce_none(val, _values)
   val[0]
- end
+end
 
     end   # class Parser
-
-  end   # module RFC2045
-
-end   # module MailParser
+    end   # module RFC2045
+  end   # module MailParser

data/lib/mailparser/rfc2045/scanner.rb

@@ -1,3 +1,4 @@
+# coding: ascii-8bit
 # Copyright (C) 2006-2010 TOMITA Masahiro
 # mailto:tommy@tmtm.org
 
@@ -18,17 +19,17 @@ class MailParser::RFC2045::Scanner < MailParser::RFC2822::Scanner
   def scan_structured()
     until @ss.eos?
       case
-      when s = @ss.scan(/\s*\(/nmo)
+      when s = @ss.scan(/\s*\(/)
         s << cfws(@ss)
         next
-      when s = @ss.scan(/\s+/nmo)
+      when s = @ss.scan(/\s+/)
         next
-      when s = @ss.scan(/\"(\s*(\\[#{TEXT_RE}]|[#{QTEXT_RE}]))*\s*\"/nmo)
-        yield :QUOTED_STRING, s
-      when s = @ss.scan(/[#{TOKEN_RE}]+/no)
-        yield :TOKEN, s
-      when s = @ss.scan(/./no)
-        yield s, s
+      when s = @ss.scan(/\"(\s*(\\[#{TEXT_RE}]|[#{QTEXT_RE}]))*\s*\"/o)
+        yield [:QUOTED_STRING, s]
+      when s = @ss.scan(/[#{TOKEN_RE}]+/o)
+        yield [:TOKEN, s]
+      when s = @ss.scan(/./)
+        yield [s, s]
       end
     end
     yield nil
@@ -37,15 +38,15 @@ class MailParser::RFC2045::Scanner < MailParser::RFC2822::Scanner
   def scan_mime_version()
     until @ss.eos?
       case
-      when s = @ss.scan(/\s*\(/nmo)
+      when s = @ss.scan(/\s*\(/)
         s << cfws(@ss)
         next
-      when s = @ss.scan(/\s+/nmo)
+      when s = @ss.scan(/\s+/)
         next
-      when s = @ss.scan(/\d+/no)
-        yield :DIGIT, s
-      when s = @ss.scan(/./no)
-        yield s, s
+      when s = @ss.scan(/\d+/)
+        yield [:DIGIT, s]
+      when s = @ss.scan(/./)
+        yield [s, s]
       end
     end
     yield nil

data/lib/mailparser/rfc2045.rb

@@ -1,3 +1,4 @@
+# coding: ascii-8bit
 # Copyright (C) 2006-2010 TOMITA Masahiro
 # mailto:tommy@tmtm.org