tmail 1.2.3.1 → 1.2.6

data/README

@@ -6,6 +6,16 @@
     Trans            assitant developer
     Minero Aoki      original developer   
 
+== RUBY 1.9 COMPATIBILITY
+
+Note... as of 1.2.5, TMail is not compatible with Ruby 1.9.1.  We are now using rchardet
+to compare encodings and there is work to do for Ruby 1.9.1 due to it's encoding
+capability.
+
+For 1.9.1 + compatibility, I suggest you look at Mail (https://github.com/mikel/mail/tree)
+while this is a work in progress, it is what I am working on to replace TMail for Ruby
+1.9 and beyond.
+
 == DESCRIPTION:
 
 TMail is a mail handling library for Ruby.  It abstracts a mail message into a usable object allowing you to read, set, add and delete headers and the mail body.

data/ext/tmailscanner/tmail/tmailscanner.c

@@ -15,9 +15,27 @@
 # include <stdlib.h>
 #endif
 
+
 #include "ruby.h"
+#ifndef RSTRING_PTR
+#define RSTRING_PTR(obj) RSTRING(obj)->ptr
+#endif
+ 
+#ifndef RSTRING_LEN
+#define RSTRING_LEN(obj) RSTRING(obj)->len
+#endif
+
+#ifdef HAVE_RUBY_VM_H
+#include "ruby/re.h"
+#include "ruby/encoding.h"
+#else
 #include "re.h"
+#endif
 
+#ifdef HAVE_RUBY_VM_H
+const unsigned char *re_mbctab;
+#define ismbchar(c) re_mbctab[(unsigned char)(c)]
+#endif
 
 #define TMAIL_VERSION "1.2.3"
 
@@ -72,9 +90,9 @@ mails_s_new(klass, str, ident, cmt)
     sc = ALLOC_N(struct scanner, 1);
 
     StringValue(str);
-    sc->pbeg = RSTRING(str)->ptr;
+    sc->pbeg = RSTRING_PTR(str);
     sc->p    = sc->pbeg;
-    sc->pend = sc->p + RSTRING(str)->len;
+    sc->pend = sc->p + RSTRING_LEN(str);
 
     sc->flags = 0;
     Check_Type(ident, T_SYMBOL);
@@ -180,6 +198,18 @@ skip_iso2022jp_string(sc)
     }
 }
 
+#ifdef HAVE_RUBY_VM_H
+static void
+skip_japanese_string(sc)
+  struct scanner *sc;
+{
+  while(sc->p < sc->pend) {
+	if (! ismbchar(*sc->p)) return;
+	rb_encoding *enc = rb_enc_get(sc);
+	sc->p += mbclen(sc->p, sc->pend, enc);
+  }
+}  
+#else
 static void
 skip_japanese_string(sc)
     struct scanner *sc;
@@ -189,6 +219,7 @@ skip_japanese_string(sc)
         sc->p += mbclen(*sc->p);
     }
 }
+#endif
 
 
 #define scan_atom(sc) scan_word(sc, ATOM_SYMBOLS)
@@ -376,9 +407,9 @@ digit_p(str)
     char *p;
     int i;
 
-    p = RSTRING(str)->ptr;
-    for (i = 0; i < RSTRING(str)->len; i++) {
-        if (! IS_DIGIT(RSTRING(str)->ptr[i]))
+    p = RSTRING_PTR(str);
+    for (i = 0; i < RSTRING_LEN(str); i++) {
+        if (! IS_DIGIT(RSTRING_PTR(str)[i]))
             return 0;
     }
     return 1;
@@ -396,7 +427,7 @@ atomsym(sc, str)
         return tok_digit;
     }
     else if (RECV_MODE_P(sc)) {
-        char *p = RSTRING(str)->ptr;
+        char *p = RSTRING_PTR(str);
         if      (nccmp(p, "from")) return tok_from;
         else if (nccmp(p, "by"))   return tok_by;
         else if (nccmp(p, "via"))  return tok_via;
@@ -417,8 +448,8 @@ debug_print(sc, sym, val)
     s = rb_funcall(sym, rb_intern("inspect"), 0),
     printf("%7ld %-10s token=<%s>\n",
            (unsigned long)(sc->pend - sc->p),
-           RSTRING(s)->ptr,
-           RSTRING(val)->ptr);
+           RSTRING_PTR(s),
+           RSTRING_PTR(val));
 }
 
 #define D(expr) do {\

data/lib/tmail.rb

@@ -3,3 +3,4 @@ require 'tmail/mail'
 require 'tmail/mailbox'
 require 'tmail/core_extensions'
 require 'tmail/net'
+require 'tmail/vendor/rchardet-1.3/lib/rchardet'

data/lib/tmail/address.rb

@@ -38,7 +38,7 @@ module TMail
   # = Class Address
   # 
   # Provides a complete handling library for email addresses. Can parse a string of an
-  # address directly or take in preformatted addresses themseleves.  Allows you to add
+  # address directly or take in preformatted addresses themselves.  Allows you to add
   # and remove phrases from the front of the address and provides a compare function for
   # email addresses.
   # 
@@ -46,7 +46,7 @@ module TMail
   # 
   # Just pass the email address in as a string to Address.parse:
   # 
-  #  email = TMail::Address.parse('Mikel Lindsaar <mikel@lindsaar.net>)
+  #  email = TMail::Address.parse('Mikel Lindsaar <mikel@lindsaar.net>')
   #  #=> #<TMail::Address mikel@lindsaar.net>
   #  email.address
   #  #=> "mikel@lindsaar.net"
@@ -63,7 +63,7 @@ module TMail
   # Address.parse and catch any SyntaxError:
   # 
   #  begin
-  #    TMail::Mail.parse("mikel   2@@@@@ me .com")
+  #    TMail::Address.parse("mikel   2@@@@@ me .com")
   #  rescue TMail::SyntaxError
   #    puts("Invalid Email Address Detected")
   #  else
@@ -81,41 +81,7 @@ module TMail
     # 
     # Raises a TMail::SyntaxError on invalid email format
     def Address.parse( str )
-      Parser.parse :ADDRESS, special_quote_address(str)
-    end
-    
-    def Address.special_quote_address(str) #:nodoc:
-      # Takes a string which is an address and adds quotation marks to special
-      # edge case methods that the RACC parser can not handle.
-      #
-      # Right now just handles two edge cases:
-      #
-      # Full stop as the last character of the display name:
-      #   Mikel L. <mikel@me.com>
-      # Returns:
-      #   "Mikel L." <mikel@me.com>
-      #
-      # Unquoted @ symbol in the display name:
-      #   mikel@me.com <mikel@me.com>
-      # Returns:
-      #   "mikel@me.com" <mikel@me.com>
-      #
-      # Any other address not matching these patterns just gets returned as is. 
-      case
-      # This handles the missing "" in an older version of Apple Mail.app
-      # around the display name when the display name contains a '@'
-      # like 'mikel@me.com <mikel@me.com>'
-      # Just quotes it to: '"mikel@me.com" <mikel@me.com>'
-      when str =~ /\A([^"].+@.+[^"])\s(<.*?>)\Z/
-        return "\"#{$1}\" #{$2}"
-      # This handles cases where 'Mikel A. <mikel@me.com>' which is a trailing
-      # full stop before the address section.  Just quotes it to
-      # '"Mikel A. <mikel@me.com>"
-      when str =~ /\A(.*?\.)\s(<.*?>)\Z/
-        return "\"#{$1}\" #{$2}"
-      else
-        str
-      end
+      Parser.parse :ADDRESS, str
     end
 
     def address_group? #:nodoc:
@@ -143,7 +109,7 @@ module TMail
       
       # This is to catch an unquoted "@" symbol in the local part of the
       # address.  Handles addresses like <"@"@me.com> and makes sure they
-      # stay like <"@"@me.com> (previously were becomming <@@me.com>)
+      # stay like <"@"@me.com> (previously were becoming <@@me.com>)
       if local && (local.join == '@' || local.join =~ /\A[^"].*?@.*?[^"]\Z/)
         @local = "\"#{local.join}\""
       else
@@ -412,7 +378,7 @@ module TMail
         if first
           first = false
         else
-          strategy.meta ','
+          strategy.puts_meta ','
         end
         strategy.space
         mbox.accept strategy

data/lib/tmail/attachments.rb

@@ -4,43 +4,62 @@
 
 =end
 
+require 'kconv'
 require 'stringio'
 
 module TMail
   class Attachment < StringIO
     attr_accessor :original_filename, :content_type
+    alias quoted_filename original_filename
   end
 
   class Mail
     def has_attachments?
-      multipart? && parts.any? { |part| attachment?(part) }
+      attachment?(self) || multipart? && parts.any? { |part| attachment?(part) }
     end
 
+    # Returns true if this part's content main type is text, else returns false.
+    # By main type is meant "text/plain" is text.  "text/html" is text
+    def text_content_type?
+      self.header['content-type'] && (self.header['content-type'].main_type == 'text')
+    end
+  
+    def inline_attachment?(part)
+      part['content-id'] || (part['content-disposition'] && part['content-disposition'].disposition == 'inline' && !part.text_content_type?)
+    end
+  
     def attachment?(part)
-      part.disposition_is_attachment? || part.content_type_is_text?
+      part.disposition_is_attachment? || (!part.content_type.nil? && !part.text_content_type?) unless part.multipart?
     end
-
+  
     def attachments
       if multipart?
-        parts.collect { |part| 
-          if part.multipart?
-            part.attachments
-          elsif attachment?(part)
-            content   = part.body # unquoted automatically by TMail#body
-            file_name = (part['content-location'] &&
-                          part['content-location'].body) ||
-                        part.sub_header("content-type", "name") ||
-                        part.sub_header("content-disposition", "filename")
-            
-            next if file_name.blank? || content.blank?
-            
-            attachment = Attachment.new(content)
-            attachment.original_filename = file_name.strip
-            attachment.content_type = part.content_type
-            attachment
-          end
-        }.flatten.compact
-      end      
+        parts.collect { |part| attachment(part) }.flatten.compact
+      elsif attachment?(self)
+        [attachment(self)]
+      end
     end
+  
+    private
+    
+    def attachment(part)
+      if part.multipart?
+        part.attachments
+      elsif attachment?(part)
+        content   = part.body # unquoted automatically by TMail#body
+        file_name = (part['content-location'] && part['content-location'].body) ||
+                    part.sub_header('content-type', 'name') ||
+                    part.sub_header('content-disposition', 'filename') ||
+                    'noname'
+
+        return if content.blank?
+
+        attachment = TMail::Attachment.new(content)
+        attachment.original_filename = file_name.strip unless file_name.blank?
+        attachment.content_type = part.content_type
+        attachment
+      end
+    end
+
   end
 end

data/lib/tmail/encode.rb

@@ -113,6 +113,7 @@ module TMail
 
     encoded = '=\?(?:iso-2022-jp|euc-jp|shift_jis)\?[QB]\?[a-z0-9+/=]+\?='
     ENCODED_WORDS = /#{encoded}(?:\s+#{encoded})*/i
+    SPACER       = "\t"
 
     OUTPUT_ENCODING = {
       'EUC'  => 'e',
@@ -165,6 +166,10 @@ module TMail
       @f << str
     end
 
+    def puts_meta( str )
+      @f << str
+    end
+
     def text( str )
       @f << decode(str)
     end
@@ -301,6 +306,10 @@ module TMail
       add_text str
     end
 
+    def puts_meta( str )
+      add_text str + @eol + SPACER
+    end
+
     def text( str )
       scanadd normalize_encoding(str)
     end

data/lib/tmail/header.rb

@@ -59,7 +59,7 @@ module TMail
       #
       # This is because a mailbox doesn't have the : after the From that designates the
       # beginning of the envelope sender (which can be different to the from address of 
-      # the emial)
+      # the email)
       #
       # Other fields can be passed as normal, "Reply-To", "Received" etc.
       #
@@ -243,6 +243,8 @@ module TMail
 
     def do_parse
       quote_boundary
+      quote_unquoted_name
+      quote_unquoted_bencode
       obj = Parser.parse(self.class::PARSE_TYPE, @body, @comments)
       set obj if obj
     end
@@ -314,7 +316,7 @@ module TMail
         if first
           first = false
         else
-          strategy.meta ','
+          strategy.puts_meta ','
           strategy.space
         end
         a.accept strategy
@@ -817,7 +819,7 @@ module TMail
         if v
           strategy.meta ';'
           strategy.space
-          strategy.kv_pair k, v
+          strategy.kv_pair k, unquote(v)
         end
       end
     end

data/lib/tmail/interface.rb

@@ -42,7 +42,7 @@ module TMail
     # Allows you to query the mail object with a string to get the contents
     # of the field you want.
     # 
-    # Returns a string of the exact contnts of the field
+    # Returns a string of the exact contents of the field
     # 
     #  mail.from = "mikel <mikel@lindsaar.net>"
     #  mail.header_string("From") #=> "mikel <mikel@lindsaar.net>"
@@ -591,16 +591,43 @@ module TMail
     end
 
     # Destructively sets the message ID of the mail object instance to the passed in string
+    # 
+    # Invalid message IDs are ignored (silently, unless configured otherwise) and result in 
+    # a nil message ID.  Left and right angle brackets are required.
     #
+    # Be warned however, that calling mail.ready_to_send will overwrite whatever value you
+    # have in this field with an automatically generated unique value.
+    # 
+    # If you really want to set your own message ID and know what you are doing per the
+    # various RFCs, you can do so with the enforced_message_id= command
+    # 
     # Example:
     # 
     #  mail = TMail::Mail.new
+    #  mail.message_id = "<348F04F142D69C21-291E56D292BC@xxxx.net>"
+    #  mail.message_id #=> "<348F04F142D69C21-291E56D292BC@xxxx.net>"
     #  mail.message_id = "this_is_my_badly_formatted_message_id"
-    #  mail.message_id #=> "this_is_my_badly_formatted_message_id"
+    #  mail.message_id #=> nil
     def message_id=( str )
       set_string_attr 'Message-Id', str
     end
 
+    # Destructively sets the message ID of the mail object instance to the passed in string
+    # and also guarantees that calling #ready_to_send will not destroy what you set as the
+    # message_id
+    # 
+    # Example:
+    # 
+    #  mail = TMail::Mail.new
+    #  mail.message_id = "<348F04F142D69C21-291E56D292BC@xxxx.net>"
+    #  mail.message_id #=> "<348F04F142D69C21-291E56D292BC@xxxx.net>"
+    #  mail.ready_to_send
+    #  mail.message_id #=> "<348F04F142D69C21-291E56D292BC@xxxx.net>"
+    def enforced_message_id=( str )
+      @message_id_enforced = true
+      self.message_id = ( str )
+    end
+
     # Returns the "In-Reply-To:" field contents as an array of this mail instance if it exists
     # 
     # If the in_reply_to field does not exist, returns nil by default or you can pass in as
@@ -838,7 +865,17 @@ module TMail
       if h = @header['content-type']
         h['charset'] or default
       else
-        default
+        mime_version_charset || default
+      end
+    end
+
+    # some weird emails come with the charset specified in the mime-version header:
+    #
+    #  #<TMail::MimeVersionHeader "1.0\n charset=\"gb2312\"">
+    #
+    def mime_version_charset
+      if header['mime-version'].inspect =~ /charset=('|\\")?([^\\"']+)/
+        $2
       end
     end
 

data/lib/tmail/mail.rb

@@ -255,7 +255,7 @@ module TMail
     alias fetch []
 
     # Allows you to set or delete TMail header objects at will.
-    # Eamples:
+    # Examples:
     #  @mail = TMail::Mail.new
     #  @mail['to'].to_s       # => 'mikel@test.com.au'
     #  @mail['to'] = 'mikel@elsewhere.org'
@@ -265,7 +265,7 @@ module TMail
     #  @mail['to'].to_s       # => nil
     #  @mail.encoded          # => "\r\n"
     # 
-    # Note: setting mail[] = nil actualy deletes the header field in question from the object,
+    # Note: setting mail[] = nil actually deletes the header field in question from the object,
     # it does not just set the value of the hash to nil
     def []=( key, val )
       dkey = key.downcase
@@ -547,7 +547,7 @@ module TMail
     end
     
     def read_multipart( src )
-      bound = @header['content-type'].params['boundary']
+      bound = @header['content-type'].params['boundary'] || ::TMail.new_boundary
       is_sep = /\A--#{Regexp.quote bound}(?:--)?[ \t]*(?:\n|\r\n|\r)/
       lastbound = "--#{bound}--"