rims-rfc822 0.1.0 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 3721bb4e1b0e4865de80a3756bc313a910180b207c76bbe4e25de30c2bb14f0b
-  data.tar.gz: f1d6f3b4b1bc84d32644beeb89628c814be2b4674bc3e5c334ce2adcc8889a11
+  metadata.gz: 51ec6ae41467b2b68bd8d1bb23916ad42f27fb6b688f8769f64e61db5e8afd98
+  data.tar.gz: f48b88893f67da1da2824336bfe274b033131527420de1926bea05faa7036618
 SHA512:
-  metadata.gz: cb1a1894bed18cfc81da85627620de46d1754b8dc50145cc40de4af0c2ed9d5a75ea7cff508b889442b06c6a513c2d607824be15ec470e3718d203d150092c61
-  data.tar.gz: 42a2832fd4e246525c3e8932a7b7a1890b258862bb7adfd4077610cc8cf85fc05e277dfea02ecb93dea20f0b7894f31757a6ab5808fe626832f208099cefd453
+  metadata.gz: 77e6b92ea9e32a7a94972d493103e03face8eaf6872400a324242431b3c8ecd5211bf0bd4c77959c47e634d820bd26089f250cd49a5e54ccfdf496cbbdaf7e08
+  data.tar.gz: a94818fedb229ad9df8683d027aa9c0cacf57a9b5296ae279794d1132e310b0a93a7158bc2c25f72279117ce29c90090b945a2e79f552fab07b1b3a1467dedb2

data/README.md

@@ -41,10 +41,16 @@ p msg.body.raw_source
 # type attributes
 p msg.media_main_type
 p msg.media_sub_type
+p msg.media_subtype # alias of media_sub_type
 p msg.content_type
-p msg.content_type_parameters
+p msg.content_type_parameter(name)
+p msg.content_type_parameter_list
 p msg.charset
 p msg.boundary
+p msg.content_disposition
+p msg.content_disposition_parameter(name)
+p msg.content_disposition_parameter_list
+p msg.content_language
 
 # headear attributes
 p msg.date

data/lib/rims/rfc822.rb

@@ -1,4 +1,4 @@
-# -*- coding: utf-8 -*-
+# -*- coding: utf-8; frozen_string_literal: true -*-
 
 require 'rims/rfc822/version'
 require 'time'
@@ -7,8 +7,14 @@ module RIMS
   module RFC822
     def split_message(msg_txt)
       header_txt, body_txt = msg_txt.lstrip.split(/\r?\n\r?\n/, 2)
-      header_txt << $& if $&
-      [ header_txt, body_txt ]
+      if ($&) then
+        header_txt << $& if $&
+      else
+        body_txt = header_txt
+        header_txt = nil
+      end
+
+      [ header_txt.freeze, body_txt.freeze ].freeze
     end
     module_function :split_message
 
@@ -24,7 +30,7 @@ module RIMS
         )
       }x)
 
-      for _name, value in field_pair_list
+      for name, value in field_pair_list
         value.strip!
         name.freeze
         value.freeze
@@ -34,59 +40,12 @@ module RIMS
     end
     module_function :parse_header
 
-    def parse_content_type(content_type_txt)
-      src_txt = content_type_txt.dup
-      if (src_txt.sub!(%r"\A \s* (?<main_type>\S+?) \s* / \s* (?<sub_type>\S+?) \s* (?:;|\Z)"x, '')) then
-        main_type = $~[:main_type]
-        sub_type = $~[:sub_type]
-
-        params = {}
-        src_txt.scan(%r'(?<name>\S+?) \s* = \s* (?: (?<quoted_string>".*?") | (?<token>\S+?) ) \s* (?:;|\Z)'x) do
-          name = $~[:name]
-          if ($~[:quoted_string]) then
-            quoted_value = $~[:quoted_string]
-            value = unquote_phrase(quoted_value)
-          else
-            value = $~[:token]
-          end
-          params[name.downcase] = [ name.freeze, value.freeze ].freeze
-        end
-
-        [ main_type.freeze, sub_type.freeze, params.freeze ].freeze
-      else
-        [ 'application'.freeze, 'octet-stream'.freeze, {}.freeze ].freeze
-      end
-    end
-    module_function :parse_content_type
-
-    def parse_multipart_body(boundary, body_txt)
-      delim = '--' + boundary
-      term = delim + '--'
-      body_txt2, _body_epilogue_txt = body_txt.split(term, 2)
-      if (body_txt2) then
-        _body_preamble_txt, body_parts_txt = body_txt2.split(delim, 2)
-        if (body_parts_txt) then
-          part_list = body_parts_txt.split(delim, -1)
-          for part_txt in part_list
-            part_txt.lstrip!
-            part_txt.chomp!("\n")
-            part_txt.chomp!("\r")
-            part_txt.freeze
-          end
-          return part_list.freeze
-        end
-      end
-
-      [].freeze
-    end
-    module_function :parse_multipart_body
-
     def unquote_phrase(phrase_txt)
       state = :raw
       src_txt = phrase_txt.dup
       dst_txt = ''.encode(phrase_txt.encoding)
 
-      while (src_txt.sub!(/\A(:? " | \( | \) | \\ | [^"\(\)\\]+ )/x, ''))
+      while (src_txt.sub!(/\A (?: " | \( | \) | \\ | [^"\(\)\\]+ )/x, ''))
         match_txt = $&
         case (state)
         when :raw
@@ -127,6 +86,111 @@ module RIMS
     end
     module_function :unquote_phrase
 
+    def parse_parameters(parameters_txt)
+      params = {}
+      parameters_txt.scan(%r'(?<name>\S+?) \s* = \s* (?: (?<quoted_string>".*?") | (?<token>\S+?) ) \s* (?:;|\Z)'x) do
+        name = $~[:name]
+        if ($~[:quoted_string]) then
+          quoted_value = $~[:quoted_string]
+          value = unquote_phrase(quoted_value)
+        else
+          value = $~[:token]
+        end
+        params[name.downcase.freeze] = [ name.freeze, value.freeze ].freeze
+      end
+
+      params.freeze
+    end
+    module_function :parse_parameters
+
+    def split_parameters(type_params_txt)
+      type, params_txt = type_params_txt.split(';', 2)
+      if (type) then
+        type.strip!
+        type.freeze
+        if (params_txt) then
+          params = parse_parameters(params_txt)
+        else
+          params = {}.freeze
+        end
+        [ type, params ].freeze
+      else
+        [ nil, {}.freeze ].freeze
+      end
+    end
+    module_function :split_parameters
+
+    def parse_content_type(type_txt)
+      media_type_txt, params = split_parameters(type_txt)
+      if (media_type_txt) then
+        main_type, sub_type = media_type_txt.split('/', 2)
+        if (main_type) then
+          main_type.strip!
+          main_type.freeze
+          if (sub_type) then
+            sub_type.strip!
+            sub_type.freeze
+            if (! main_type.empty? && ! sub_type.empty?) then
+              return [ main_type, sub_type, params ].freeze
+            end
+          end
+        end
+      end
+
+      [ 'application'.dup.force_encoding(type_txt.encoding).freeze,
+        'octet-stream'.dup.force_encoding(type_txt.encoding).freeze,
+        params
+      ].freeze
+    end
+    module_function :parse_content_type
+
+    def parse_content_disposition(disposition_txt)
+      split_parameters(disposition_txt)
+    end
+    module_function :parse_content_disposition
+
+    def parse_content_language(language_tags_txt)
+      tag_list = language_tags_txt.split(',')
+      for tag in tag_list
+        tag.strip!
+        tag.freeze
+      end
+      tag_list.reject!(&:empty?)
+
+      tag_list.freeze
+    end
+    module_function :parse_content_language
+
+    def parse_multipart_body(boundary, body_txt)
+      delim = '--' + boundary
+      term = delim + '--'
+      body_txt2, _body_epilogue_txt = body_txt.split(term, 2)
+      if (body_txt2) then
+        _body_preamble_txt, body_parts_txt = body_txt2.split(delim, 2)
+        if (body_parts_txt) then
+          part_list = body_parts_txt.split(delim, -1)
+          for part_txt in part_list
+            part_txt.lstrip!
+            part_txt.chomp!("\n")
+            part_txt.chomp!("\r")
+            part_txt.freeze
+          end
+          return part_list.freeze
+        end
+      end
+
+      [].freeze
+    end
+    module_function :parse_multipart_body
+
+    Address = Struct.new(:display_name, :route, :local_part, :domain)
+    class Address
+      # compatible for Net::MAP::Address
+      alias name display_name
+      alias mailbox local_part
+      alias host domain
+    end
+
     def parse_mail_address_list(address_list_txt)
       addr_list = []
       src_txt = address_list_txt.dup
@@ -142,9 +206,9 @@ module RIMS
         then
           display_name = $~[:display_name]
           group_list = $~[:group_list]
-          addr_list << [ nil, nil, unquote_phrase(display_name), nil ].freeze
+          addr_list << Address.new( nil, nil, unquote_phrase(display_name), nil).freeze
           addr_list.concat(parse_mail_address_list(group_list))
-          addr_list << [ nil, nil, nil, nil ].freeze
+          addr_list << Address.new(nil, nil, nil, nil).freeze
         elsif (src_txt.sub!(%r{
                  \A
                  \s*
@@ -153,7 +217,7 @@ module RIMS
                  ,?
                }x, ''))
         then
-          addr_list << [ nil, nil, $~[:local_part].freeze, $~[:domain].freeze ].freeze
+          addr_list << Address.new(nil, nil, $~[:local_part].freeze, $~[:domain].freeze).freeze
         elsif (src_txt.sub!(%r{
                  \A
                  \s*
@@ -178,7 +242,7 @@ module RIMS
           route = $~[:route]
           local_part = $~[:local_part]
           domain = $~[:domain]
-          addr_list << [ unquote_phrase(display_name), route.freeze, local_part.freeze, domain.freeze ].freeze
+          addr_list << Address.new(unquote_phrase(display_name), route.freeze, local_part.freeze, domain.freeze).freeze
         else
           break
         end
@@ -192,7 +256,7 @@ module RIMS
       include Enumerable
 
       def initialize(header_txt)
-        @raw_source = header_txt.freeze
+        @raw_source = header_txt
         @field_list = nil
         @field_map = nil
       end
@@ -255,7 +319,7 @@ module RIMS
 
     class Body
       def initialize(body_txt)
-        @raw_source = body_txt.freeze
+        @raw_source = body_txt
       end
 
       attr_reader :raw_source
@@ -263,13 +327,13 @@ module RIMS
 
     class Message
       def initialize(msg_txt)
-        @raw_source = msg_txt.freeze
+        @raw_source = msg_txt.dup.freeze
         @header = nil
         @body = nil
         @content_type = nil
-        @is_multipart = nil
+        @content_disposition = nil
+        @content_language = nil
         @parts = nil
-        @is_message = nil
         @message = nil
         @date = nil
         @from = nil
@@ -304,61 +368,122 @@ module RIMS
 
       def setup_content_type
         if (@content_type.nil?) then
-          @content_type = RFC822.parse_content_type(header['content-type'] || '')
+          @content_type = RFC822.parse_content_type(header['Content-Type'] || '')
           self
         end
       end
       private :setup_content_type
 
-      def to_upper(text_or_nil)
-        text_or_nil.upcase if text_or_nil
-      end
-      private :to_upper
-
       def media_main_type
         setup_content_type
         @content_type[0]
       end
 
-      def media_main_type_upcase
-        setup_content_type
-        to_upper(@content_type[0])
-      end
-
       def media_sub_type
         setup_content_type
         @content_type[1]
       end
 
-      def media_sub_type_upcase
-        setup_content_type
-        to_upper(@content_type[1])
-      end
+      alias media_subtype media_sub_type
 
       def content_type
         "#{media_main_type}/#{media_sub_type}"
       end
 
-      def content_type_upcase
-        to_upper("#{media_main_type}/#{media_sub_type}")
+      def media_main_type_upcase
+        # not return `nil'
+        media_main_type.upcase
       end
 
-      def content_type_parameters
-        setup_content_type
-        @content_type[2].each_value.map{|name, value| [ name, value ] }
+      def media_sub_type_upcase
+        # not return `nil'
+        media_sub_type.upcase
       end
 
-      def charset
+      alias media_subtype_upcase media_sub_type_upcase
+
+      def content_type_upcase
+        content_type.upcase
+      end
+
+      def content_type_parameter(name)
         setup_content_type
-        if (name_value_pair = @content_type[2]['charset']) then
+        if (name_value_pair = @content_type[2][name.downcase]) then
           name_value_pair[1]
         end
       end
 
+      def content_type_parameter_list
+        setup_content_type
+        @content_type[2].values
+      end
+
+      alias content_type_parameters content_type_parameter_list
+
+      def charset
+        content_type_parameter('charset')
+      end
+
       def boundary
+        content_type_parameter('boundary')
+      end
+
+      def setup_content_disposition
+        if (header.key? 'Content-Disposition') then
+          if (@content_disposition.nil?) then
+            @content_disposition = RFC822.parse_content_disposition(header['Content-Disposition'])
+            self
+          end
+        end
+      end
+      private :setup_content_type
+
+      def content_disposition
+        setup_content_disposition
+        @content_disposition && @content_disposition[0]
+      end
+
+      def content_disposition_upcase
+        if (type = content_disposition) then
+          type.upcase
+        end
+      end
+
+      def content_disposition_parameter(name)
+        setup_content_disposition
+        if (@content_disposition) then
+          if (name_value_pair = @content_disposition[1][name.downcase]) then
+            name_value_pair[1]
+          end
+        end
+      end
+
+      def content_disposition_parameter_list
         setup_content_type
-        if (name_value_pair = @content_type[2]['boundary']) then
-          name_value_pair[1]
+        @content_disposition && @content_disposition[1].values
+      end
+
+      alias content_disposition_parameters content_disposition_parameter_list
+
+      def setup_content_language
+        if (header.key? 'Content-Language') then
+          if (@content_language.nil?) then
+            @content_language = header.field_value_list('Content-Language').map{|tags_txt| RFC822.parse_content_language(tags_txt) }.inject(:+)
+            @content_language.freeze
+            self
+          end
+        end
+      end
+      private :setup_content_language
+
+      def content_language
+        setup_content_language
+        @content_language
+      end
+
+      def content_language_upcase
+        if (tag_list = content_language) then
+          tag_list.map{|tag| tag.upcase }
         end
       end
 
@@ -367,10 +492,7 @@ module RIMS
       end
 
       def multipart?
-        if (@is_multipart.nil?) then
-          @is_multipart = (media_main_type_upcase == 'MULTIPART')
-        end
-        @is_multipart
+        media_main_type_upcase == 'MULTIPART'
       end
 
       def parts
@@ -378,20 +500,19 @@ module RIMS
           if (@parts.nil?) then
             if (boundary = self.boundary) then
               part_list = RFC822.parse_multipart_body(boundary, body.raw_source)
-              @parts = part_list.map{|msg_txt| Message.new(msg_txt) }.freeze
+              @parts = part_list.map{|msg_txt| Message.new(msg_txt) }
             else
-              @parts = [].freeze
+              @parts = []
             end
+            @parts.freeze
           end
+
           @parts
         end
       end
 
       def message?
-        if (@is_message.nil?) then
-          @is_message = (media_main_type_upcase == 'MESSAGE')
-        end
-        @is_message
+        media_main_type_upcase == 'MESSAGE'
       end
 
       def message
@@ -399,6 +520,7 @@ module RIMS
           if (@message.nil?) then
             @message = Message.new(body.raw_source)
           end
+
           @message
         end
       end
@@ -411,9 +533,10 @@ module RIMS
             rescue ArgumentError
               @date = Time.at(0)
             end
+            @date.freeze
           end
 
-          @date.freeze
+          @date
         end
       end
 
@@ -422,10 +545,11 @@ module RIMS
           ivar_name = '@' + field_name.downcase.gsub('-', '_')
           addr_list = instance_variable_get(ivar_name)
           if (addr_list.nil?) then
-            addr_list = header.field_value_list(field_name).map{|addr_list_str| RFC822.parse_mail_address_list(addr_list_str) }.inject(:+)
+            addr_list = header.field_value_list(field_name).map{|addr_list_txt| RFC822.parse_mail_address_list(addr_list_txt) }.inject(:+)
             addr_list.freeze
             instance_variable_set(ivar_name, addr_list)
           end
+
           addr_list
         end
       end

data/lib/rims/rfc822/version.rb

@@ -2,7 +2,7 @@
 
 module RIMS
   module RFC822
-    VERSION = '0.1.0'
+    VERSION = '0.2.0'
   end
 end
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rims-rfc822
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 0.2.0
 platform: ruby
 authors:
 - TOKI Yoshinori
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2019-06-13 00:00:00.000000000 Z
+date: 2019-06-20 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler