No commit message

Makefile

@@ -1,11 +1,17 @@
 all:
 	cd rfc2822; make
 	cd rfc2822obs; make
+	cd rfc2045; make
+	cd rfc2183; make
 
 clean:
 	cd rfc2822; make clean
 	cd rfc2822obs; make clean
+	cd rfc2045; make clean
+	cd rfc2183; make clean
 
 test:
 	ruby -w test_rfc2822.rb
 	ruby -w test_rfc2822obs.rb
+	ruby -w test_rfc2045.rb
+	ruby -w test_rfc2183.rb

mailparser.rb

@@ -27,6 +27,11 @@ class MailParser
     "resent-message-id" => RFC2822obs,
     "return-path"       => RFC2822obs,
     "received"          => RFC2822obs,
+    "content-type"      => RFC2045,
+    "content-description" => RFC2045,
+    "content-transfer-encoding" => RFC2045,
+    "content-id"        => RFC2045,
+    "mime-version"      => RFC2045,
   }
 
   class Header

rfc2045.rb

@@ -0,0 +1,53 @@
+#
+# $Id$
+#
+# Copyright (C) 2006 TOMITA Masahiro
+# mailto:[email protected]
+
+require "rfc2822"
+require "rfc2045/parser"
+
+module RFC2045
+
+  class ParseError < StandardError
+  end
+
+  HEADER_TYPE = {
+    "content-type"              => :CONTENT_TYPE,
+    "content-description"       => :UNSTRUCTURED,
+    "content-transfer-encoding" => :CONTENT_TRANSFER_ENCODING,
+    "content-id"                => [RFC2822, :MSG_ID],
+    "mime-version"              => :MIME_VERSION,
+  }
+
+  class ContentType
+    def initialize(type, subtype, params)
+      @type, @subtype, @params = type.downcase, subtype.downcase, params
+    end
+
+    attr_reader :type, :subtype, :params
+  end
+
+  class ContentTransferEncoding
+    def initialize(mechanism)
+      @mechanism = mechanism.downcase
+    end
+
+    attr_reader :mechanism
+  end
+
+  module_function
+  def parse(name, value)
+    htype = HEADER_TYPE[name.downcase] || :UNSTRUCTURED
+    if htype == :UNSTRUCTURED then
+      return value.chomp
+    end
+    if htype.is_a? Array then
+      parser = htype[0]::Parser.new
+      parser.parse(htype[1], value)
+    else
+      parser = Parser.new
+      parser.parse(htype, value)
+    end
+  end
+end

rfc2045/Makefile

@@ -0,0 +1,7 @@
+all: parser.rb
+
+clean:
+	rm -rf parser.rb
+
+parser.rb: parser.y
+	racc -v -o parser.rb parser.y

rfc2045/parser.y

@@ -0,0 +1,86 @@
+#
+# $Id$
+#
+# Copyright (C) 2006 TOMITA Masahiro
+# mailto:[email protected]
+
+class RFC2045::Parser
+
+  options no_result_var
+
+rule
+
+all             : CONTENT_TYPE content_type {val[1]}
+#               | CONTENT_DESCRIPTION
+                | CONTENT_TRANSFER_ENCODING content_transfer_encoding {val[1]}
+#               | CONTENT_ID content_id {val[1]}
+                | MIME_VERSION mime_version {val[1]}
+
+content_type    : type '/' subtype parameter_list
+                  {
+                    ContentType.new(val[0], val[2], val[3])
+                  }
+
+content_transfer_encoding: mechanism
+                  {
+                    ContentTransferEncoding.new(val[0])
+                  }
+
+#content_id      : msg_id
+
+mime_version    : DIGIT '.' DIGIT
+                  {
+                    val.join
+                  }
+
+mechanism       : TOKEN
+
+type            : TOKEN
+
+subtype         : TOKEN
+
+parameter_list  : /* empty */
+                  {
+                    {}
+                  }
+                | parameter_list ';' parameter
+                  {
+                    pn, pv = val[2]
+                    pv = $1 if pv =~ /\A\"(.*)\"\Z/
+                    val[0][pn] = pv
+                    val[0]
+                  }
+
+parameter       : attribute '=' value
+                  {
+                    [val[0].downcase, val[2]]
+                  }
+
+attribute       : TOKEN
+
+value           : TOKEN
+                | QUOTED_STRING
+
+---- inner
+
+require "rfc2045/scanner"
+
+def parse(header_type, value)
+  @header_type = header_type
+  @value = value
+  @scanner = RFC2045::Scanner.new(header_type, value)
+  ret = yyparse(self, :parse_sub)
+  @comments = @scanner.comments
+  ret
+end
+
+def parse_sub(&block)
+  yield @header_type, @value
+  @scanner.scan(&block)
+end
+
+def on_error(t, val, vstack)
+#  p t, val, vstack
+#  p racc_token2str(t)
+  raise ParseError, val
+end

rfc2045/scanner.rb

@@ -0,0 +1,57 @@
+#
+# $Id$
+#
+# Copyright (C) 2006 TOMITA Masahiro
+# mailto:[email protected]
+
+require "rfc2822"
+
+class RFC2045::Scanner < RFC2822::Scanner
+  TOKEN_RE = '\x21\x23-\x27\x2a\x2b\x2d\x2e\x30-\x39\x41-\x5a\x5e-\x7f'
+
+  def scan(&block)
+    case @header_type
+    when :MIME_VERSION
+      scan_mime_version(&block)
+    else
+      scan_structured(&block)
+    end
+  end
+
+  def scan_structured()
+    until @ss.eos?
+      case
+      when s = @ss.scan(/\s*\(/nmo)
+        s << cfws(@ss)
+        next
+      when s = @ss.scan(/\s+/nmo)
+        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
+      end
+    end
+    yield nil
+  end
+
+  def scan_mime_version()
+    until @ss.eos?
+      case
+      when s = @ss.scan(/\s*\(/nmo)
+        s << cfws(@ss)
+        next
+      when s = @ss.scan(/\s+/nmo)
+        next
+      when s = @ss.scan(/\d+/no)
+        yield :DIGIT, s
+      when s = @ss.scan(/./no)
+        yield s, s
+      end
+    end
+    yield nil
+  end
+
+end

rfc2183.rb

@@ -0,0 +1,40 @@
+#
+# $Id$
+#
+# Copyright (C) 2006 TOMITA Masahiro
+# mailto:[email protected]
+
+require "rfc2183/parser"
+
+module RFC2183
+
+  class ParseError < StandardError
+  end
+
+  HEADER_TYPE = {
+    "content-disposition"      => :CONTENT_DISPOSITION,
+  }
+
+  class ContentDisposition
+    def initialize(type, params)
+      @type, @params = type.downcase, params
+    end
+
+    attr_reader :type, :params
+  end
+
+  module_function
+  def parse(name, value)
+    htype = HEADER_TYPE[name.downcase] || :UNSTRUCTURED
+    if htype == :UNSTRUCTURED then
+      return value.chomp
+    end
+    if htype.is_a? Array then
+      parser = htype[0]::Parser.new
+      parser.parse(htype[1], value)
+    else
+      parser = Parser.new
+      parser.parse(htype, value)
+    end
+  end
+end

rfc2183/Makefile

@@ -0,0 +1,7 @@
+all: parser.rb
+
+clean:
+	rm -rf parser.rb
+
+parser.rb: parser.y
+	racc -v -o parser.rb parser.y

rfc2183/parser.y

@@ -0,0 +1,66 @@
+#
+# $Id$
+#
+# Copyright (C) 2006 TOMITA Masahiro
+# mailto:[email protected]
+
+class RFC2183::Parser
+
+  options no_result_var
+
+rule
+
+all             : CONTENT_DISPOSITION content_disposition {val[1]}
+
+content_disposition : type parameter_list
+                  {
+                    ContentDisposition.new(val[0], val[1])
+                  }
+
+type            : TOKEN
+
+parameter_list  : /* empty */
+                  {
+                    {}
+                  }
+                | parameter_list ';' parameter
+                  {
+                    pn, pv = val[2]
+                    pv = $1 if pv =~ /\A\"(.*)\"\Z/
+                    val[0][pn] = pv
+                    val[0]
+                  }
+
+parameter       : attribute '=' value
+                  {
+                    [val[0].downcase, val[2]]
+                  }
+
+attribute       : TOKEN
+
+value           : TOKEN
+                | QUOTED_STRING
+
+---- inner
+
+require "rfc2183/scanner"
+
+def parse(header_type, value)
+  @header_type = header_type
+  @value = value
+  @scanner = RFC2183::Scanner.new(header_type, value)
+  ret = yyparse(self, :parse_sub)
+  @comments = @scanner.comments
+  ret
+end
+
+def parse_sub(&block)
+  yield @header_type, @value
+  @scanner.scan(&block)
+end
+
+def on_error(t, val, vstack)
+#  p t, val, vstack
+#  p racc_token2str(t)
+  raise ParseError, val
+end

rfc2183/scanner.rb

@@ -0,0 +1,10 @@
+#
+# $Id$
+#
+# Copyright (C) 2006 TOMITA Masahiro
+# mailto:[email protected]
+
+require "rfc2045"
+
+class RFC2183::Scanner < RFC2045::Scanner
+end

rfc2822.rb

@@ -1,4 +1,8 @@
+#
 # $Id$
+#
+# Copyright (C) 2006 TOMITA Masahiro
+# mailto:[email protected]
 
 require "rfc2822/parser"
 
@@ -136,7 +140,6 @@ def time()
     end
   end
 
-
   module_function
   def parse(name, value)
     htype = HEADER_TYPE[name.downcase] || :UNSTRUCTURED