man Mail::Message () - general message object


Mail::Message - general message object


 Mail::Message has extra code in

   is a Mail::Reporter

 Mail::Message is extended by


 use Mail::Box::Manager;
 my $mgr    = Mail::Box::Manager->new;
 my $folder = $mgr->open(folder => 'InBox');
 my $msg    = $folder->message(2);    # $msg is a Mail::Message now

 my $subject = $msg->subject;         # The message's subject
 my @cc      = $msg->cc;              # List of Mail::Address'es

 my $msg       = Mail::Message->build(...);
 my $reply_msg = Mail::Message->reply(...);
 my $frwd_msg  = Mail::Message->forward(...);

 my Mail::Message::Head $head = $msg->head;
 my Mail::Message::Body $body = $msg->decoded;


A CWMail::Message object is a container for MIME-encoded message information, as defined by RFC2822. Everything what is not specificly related to storing the messages in mailboxes (folders) is implemented in this class. Methods which are related to folders is implemented in the Mail::Box::Message extension.

The main methods are get(), to get information from a message header field, and decoded() to get the intended content of a message. But there are many more which can assist your program.

Complex message handling, like construction of replies and forwards, are implemented in separate packages which are autoloaded into this class. This means you can simply use these methods as if they are part of this class. Those package add functionality to all kinds of message objects.



$obj->clone(OPTIONS) Create a copy of this message. Returned is a CWMail::Message object. The head and body, the log and trace levels are taken. Labels are copied with the message, but the delete and modified flags are not. BE WARNED: the clone of any kind of message (or a message part) will always be a CWMail::Message object. For example, a Mail::Box::Message's clone is detached from the folder of its original. When you use Mail::Box::addMessage() with the cloned message at hand, then the clone will automatically be coerced into the right message type to be added. See also Mail::Box::Message::copyTo() and Mail::Box::Message::moveTo().

 Option        Defined in       Default
 shallow                        <false>
 shallow_body                   <false>
 shallow_head                   <false>
. shallow BOOLEAN When a shallow clone is made, the header and body of the message will not be cloned, but shared. This is quite dangerous: for instance in some folder types, the header fields are used to store folder flags. When one of both shallow clones change the flags, that will update the header and thereby be visible in both. There are situations where a shallow clone can be used safely. For instance, when Mail::Box::Message::moveTo() is used and you are sure that the original message cannot get undeleted after the move. . shallow_body BOOLEAN A rather safe bet, because you are not allowed to modify the body of a message: you may only set a new body with body(). . shallow_head BOOLEAN Only the head uses is reused, not the body. This is probably a bad choice, because the header fields can be updated, for instance when labels change. Example:
 $copy = $msg->clone;


 Option      Defined in       Default                                                       
 body                         undef                                                         
 body_type                    L<Mail::Message::Body::Lines|Mail::Message::Body::Lines>      
 deleted                      <false>                                                       
 field_type                   undef                                                         
 head                         undef                                                         
 head_type                    L<Mail::Message::Head::Complete|Mail::Message::Head::Complete>
 labels                       {}                                                            
 log         L<Mail::Reporter>  C<'WARNINGS'>                                                 
 messageId                    undef                                                         
 modified                     <false>                                                       
 trace       L<Mail::Reporter>  C<'WARNINGS'>                                                 
 trusted                      <false>
. body OBJECT Instantiate the message with a body which has been created somewhere before the message is constructed. The OBJECT must be a sub-class of Mail::Message::Body. See also body() and storeBody(). . body_type CLASS Default type of body to be created for readBody(). . deleted BOOLEAN Is the file deleted from the start? . field_type CLASS . head OBJECT Instantiate the message with a head which has been created somewhere before the message is constructed. The OBJECT must be a (sub-)class of Mail::Message::Head. See also head(). . head_type CLASS Default type of head to be created for readHead(). . labels ARRAY|HASH Initial values of the labels. In case of Mail::Box::Message's, this shall reflect the state the message is in. For newly constructed Mail::Message's, this may be anything you want, because coerce() will take care of the folder specifics once the message is added to one. . log LEVEL . messageId STRING The id on which this message can be recognized. If none specified and not defined in the header --but one is needed there will be one assigned to the message to be able to pass unique message-ids between objects. . modified BOOLEAN Flags this message as being modified from the beginning on. Usually, modification is auto-detected, but there may be reasons to be extra explicit. . trace LEVEL . trusted BOOLEAN Is this message from a trusted source? If not, the content must be checked before use. This checking will be performed when the body data is decoded or used for transmission.

Constructing a message

$obj->bounce([RG-OBJECT|OPTIONS]) See Constructing a message in Mail::Message::Construct::Bounce

Mail::Message->build([MESSAGE|PART|BODY], CONTENT) See Constructing a message in Mail::Message::Construct::Build

Mail::Message->buildFromBody(BODY, [HEAD], HEADERS) See Constructing a message in Mail::Message::Construct::Build

$obj->forward(OPTIONS) See Constructing a message in Mail::Message::Construct::Forward

$obj->forwardAttach(OPTIONS) See Constructing a message in Mail::Message::Construct::Forward

$obj->forwardEncapsulate(OPTIONS) See Constructing a message in Mail::Message::Construct::Forward

$obj->forwardInline(OPTIONS) See Constructing a message in Mail::Message::Construct::Forward

$obj->forwardNo(OPTIONS) See Constructing a message in Mail::Message::Construct::Forward

$obj->forwardPostlude See Constructing a message in Mail::Message::Construct::Forward

$obj->forwardPrelude See Constructing a message in Mail::Message::Construct::Forward

$obj->forwardSubject(STRING) See Constructing a message in Mail::Message::Construct::Forward

Mail::Message->read(FILEHANDLE|SCALAR|REF-SCALAR|ARRAY-OF-LINES, OPTIONS) See Constructing a message in Mail::Message::Construct::Read

$obj->rebuild(OPTIONS) See Constructing a message in Mail::Message::Construct::Rebuild

$obj->reply(OPTIONS) See Constructing a message in Mail::Message::Construct::Reply

$obj->replyPrelude([STRING|FIELD|ADDRESS|ARRAY-OF-THINGS]) See Constructing a message in Mail::Message::Construct::Reply


Mail::Message->replySubject(STRING) See Constructing a message in Mail::Message::Construct::Reply

The message

$obj->container If the message is a part of another message, CWcontainer returns the reference to the containing body. Example:

 my Mail::Message $msg = ...
 return unless $msg->body->isMultipart;
 my $part   = $msg->body->part(2);
 return unless $part->body->isMultipart;
 my $nested = $part->body->part(3);
 $nested->container;  # returns $msg->body
 $nested->toplevel;   # returns $msg
 $msg->container;     # returns undef
 $msg->toplevel;      # returns $msg
 $msg->isPart;        # returns false
 $part->isPart;       # returns true

$obj->isDummy Dummy messages are used to fill holes in linked-list and such, where only a message-id is known, but not the place of the header of body data. This method is also available for Mail::Message::Dummy objects, where this will return CWtrue. On any extension of CWMail::Message, this will return CWfalse.

$obj->isPart Returns true if the message is a part of another message. This is the case for Mail::Message::Part extensions of CWMail::Message.

$obj->messageId Retrieve the message's id. Every message has a unique message-id. This id is used mainly for recognizing discussion threads.

$obj->print([FILEHANDLE]) Print the message to the FILE-HANDLE, which defaults to the selected filehandle, without the encapsulation sometimes required by a folder type, like write() does. Example:

 $message->print(\*STDERR);  # to the error output
 $message->print;            # to the selected file
 my $out = IO::File->new('out', 'w');
 $message->print($out);      # no encapsulation: no folder
 $message->write($out);      # with encapsulation: is folder.

$obj->send([MAILER], OPTIONS) Transmit the message to anything outside this Perl program. MAILER is a Mail::Transport::Send object. When the MAILER is not specified, one will be created, and kept as default for the next messages as well. The OPTIONS are mailer specific, and a mixture of what is usable for the creation of the mailer object and the sending itself. Therefore, see for possible options Mail::Transport::Send::new() and Mail::Transport::Send::send(). Example:

is short (but little less flexibile) for
 my $mailer = Mail::Transport::SMTP->new(@smtpopts);
 $mailer->send($message, @sendopts);
See examples/ in the distribution of Mail::Box. Example:
 $message->send(via => 'sendmail')

$obj->size Returns an estimated size of the whole message in bytes. In many occasions, the functions which process the message further, for instance send() or print() will need to add/change header lines or add CR characters, so the size is only an estimate with a few percent margin of the real result. The computation assumes that each line ending is represented by one character (like UNIX, MacOS, and sometimes Cygwin), and not two characters (like Windows and sometimes Cygwin). If you write the message to file on a system which uses CR and LF to end a single line (all Windows versions), the result in that file will be at least nrLines() larger than this method returns.

$obj->toplevel Returns a reference to the main message, which will be the current message if the message is not part of another message.

$obj->write([FILEHANDLE]) Write the message to the FILE-HANDLE, which defaults to the selected FILEHANDLE, with all surrounding information which is needed to put it correctly in a folder file. In most cases, the result of CWwrite will be the same as with print(). The main exception is for Mbox folder messages, which will get printed with their leading 'From ' line and a trailing blank. Each line of their body which starts with 'From ' will have an '>' added in front.

The header

$obj->bcc Returns the addresses which are specified on the CWBcc header line (or lines) A list of Mail::Address objects is returned. CWBcc stands for Blind Carbon Copy: destinations of the message which are not listed in the messages actually sent. So, this field will be empty for received messages, but may be present in messages you construct yourself.

$obj->cc Returns the addresses which are specified on the CWCc header line (or lines) A list of Mail::Address objects is returned. CWCc stands for Carbon Copy; the people addressed on this line receive the message informational, and are usually not expected to reply on its content.

$obj->date Method has been removed for reasons of consistency. Use timestamp() or CW$msg-head->get('Date')>.

$obj->destinations Returns a list of Mail::Address objects which contains the combined info of active CWTo, CWCc, and CWBcc addresses. Double addresses are removed if detectable.

$obj->from Returns the addresses from the senders. It is possible to have more than one address specified in the CWFrom field of the message, according to the specification. Therefore a list of Mail::Address objects is returned, which usually has length 1. If you need only one address from a sender, for instance to create a original message by line in constructed forwarded message body, then use sender(). Example: using from() to get all sender addresses

 my @from = $message->from;

$obj->get(FIELDNAME) Returns the value which is stored in the header field with the specified name. The FIELDNAME is case insensitive. The unfolded body of the field is returned, stripped from any attributes. See Mail::Message::Field::body(). If the field has multiple appearances in the header, only the last instance is returned. If you need more complex handing of fields, then call Mail::Message::Head::get() yourself. See study() when you want to be smart, doing the better (but slower) job. Example: the get() short-cut for header fields

 print $msg->get('Content-Type'), "\n";
Is equivalent to:
 print $msg->head->get('Content-Type')->body, "\n";

$obj->guessTimestamp Return an estimate on the time this message was sent. The data is derived from the header, where it can be derived from the CWdate and CWreceived lines. For MBox-like folders you may get the date from the from-line as well. This method may return CWundef if the header is not parsed or only partially known. If you require a time, then use the timestamp() method, described below. Example: using guessTimestamp() to get a transmission date

 print "Receipt ", ($message->timestamp || 'unknown'), "\n";

$obj->head([HEAD]) Return (optionally after setting) the HEAD of this message. The head must be an (sub-)class of Mail::Message::Head. When the head is added, status information is taken from it and transformed into labels. More labels can be added by the LABELS hash. They are added later. Example:

 $msg->head(Mail::Message::Head->new);  # set
 my $head = $msg->head;                 # get

$obj->nrLines Returns the number of lines used for the whole message.

$obj->sender Returns exactly one address, which is the originator of this message. The returned Mail::Address object is taken from the CWSender header field, unless that field does not exists, in which case the first address from the CWFrom field is taken. If none of both provide an address, CWundef is returned. Example: using sender() to get exactly one sender address

 my $sender = $message->sender;
 print "Reply to: ", $sender->format, "\n" if defined $sender;

$obj->study(FIELDNAME) Study the content of a field, like get() does, with as main difference that a Mail::Message::Field::Full object is returned. These objects stringify to an utf8 decoded representation of the data contained in the field, where get() does not decode. See Mail::Message::Field::study(). Example: the study() short-cut for header fields

 print $msg->study('to'), "\n";
Is equivalent to:
 print $msg->head->study('to'), "\n";       # and
 print $msg->head->get('to')->study, "\n";

$obj->subject Returns the message's subject, or the empty string. Example: using subject() to get the message's subject

 print $msg->subject;

$obj->timestamp Get a good timestamp for the message, doesn't matter how much work it is. The value returned is compatible with the platform dependent result of function time(). In these days, the timestamp as supplied by the message (in the CWDate field) is not trustable at all: many spammers produce illegal or unreal dates to influence their location in the displayed folder. To start, the received headers are tried for a date (see Mail::Message::Head::Complete::recvstamp()) and only then the CWDate field. In very rare cases, only with some locally produced messages, no stamp can be found.

$obj->to Returns the addresses which are specified on the CWTo header line (or lines). A list of Mail::Address objects is returned. The people addressed here are the targets of the content, and should read it contents carefully. Example: using to() to get all primar destination addresses

 my @to = $message->to;

The body

$obj->body([BODY]) Return the body of this message. BE WARNED that this returns you an object which may be encoded: use decoded() to get a body with usable data. With options, a new BODY is set for this message. This is not for normal use unless you understand the consequences: you change the message content without changing the message-ID. The right way to go is via

 $message = Mail::Message->buildFromBody($body);  # or
 $message = Mail::Message->build($body);          # or
 $message = $origmsg->forward(body => $body);
The BODY must be an (sub-)class of Mail::Message::Body. In this case, information from the specified body will be copied into the header. The body object will be encoded if needed, because messages written to file or transmitted shall not contain binary data. The converted body is returned. When BODY is CWundef, the current message body will be dissected from the message. All relation will be cut. The body is returned, and can be connected to a different message. Example:
 my $body      = $msg->body;
 my @encoded   = $msg->body->lines;
 my $new       = Mail::Message::Body->new(mime_type => 'text/html');
 my $converted = $msg->body($new);

$obj->decoded(OPTIONS) Decodes the body of this message, and returns it as a body object. If there was no encoding, the body object as read from file is passed on, however, some more work will be needed when a serious encoding is encountered. The OPTIONS control how the conversion takes place.

 Option       Defined in       Default       
 keep                          <false>       
 result_type                   <type of body>
. keep BOOLEAN Controls whether the decoded result will be kept. If not, the decoding may be performed more than once. However, it will consume extra resources... . result_type BODYTYPE Specifies which kind of body should be used for the final result, and eventual intermediate conversion stages. It is not sure that this will be the type of the body returned. BODYTYPE extends Mail::Message::Body. Example:
 my $dec = $message->body($message->decoded);
 my $dec = $message->decoded(keep => 1);   # same

$obj->encode(OPTIONS) Encode the message to a certain format. Read the details in the dedicated manual page Mail::Message::Body::Encode. The OPTIONS which can be specified here are those of the Mail::Message::Body::encode() method.

$obj->isMultipart Check whether this message is a multipart message (has attachments). To find this out, we need at least the header of the message; there is no need to read the body of the message to detect this.

$obj->isNested Returns CWtrue for CWmessage/rfc822 messages and message parts.

$obj->parts(['ALL'|'ACTIVE'|'DELETED'|'RECURSE'|FILTER]) Returns the parts of this message. Usually, the term part is used with multipart messages: messages which are encapsulated in the body of a message. To abstract this concept: this method will return you all header-body combinations which are stored within this message except the multipart and message/rfc822 wrappers. Objects returned are CWMail::Message's and Mail::Message::Part's. The option default to 'ALL', which will return the message itself for single-parts, the nested content of a message/rfc822 object, respectively the parts of a multipart without recursion. In case of 'RECURSE', the parts of multiparts will be collected recursively. This option cannot be combined with the other options, which you may want: it that case you have to test yourself. 'ACTIVE' and 'DELETED' check for the deleted flag on messages and message parts. The FILTER is a code reference, which is called for each part of the messagei; each part as CWRECURSE would return. Example:

 my @parts = $msg->parts;           # $msg not multipart: returns ($msg)
 my $parts = $msg->parts('ACTIVE'); # returns ($msg)
 my @parts = $msg->parts;           # returns ($msg)
 my $parts = $msg->parts('ACTIVE'); # returns ()


$obj->delete Flag the message to be deleted, which is a shortcut for CW$msg->label(deleted => time); The real deletion only takes place on a synchronization of the folder. See deleted() as well. The time stamp of the moment of deletion is stored as value, but that is not always preserved in the folder (depends on the implementation). When the same message is deleted more than once, the first time stamp will stay. Example:

 $message->deleted(1);  # exactly the same
 $message->label(deleted => 1);
 delete $message;

$obj->deleted([BOOLEAN]) Set the delete flag for this message. Without argument, the method returns the same as isDeleted(), which is prefered. When a true value is given, delete() is called. Example:

 $message->deleted(1);          # delete
 $message->delete;              # delete (prefered)
 $message->deleted(0);          # undelete
 if($message->deleted) {...}    # check
 if($message->isDeleted) {...}  # check (prefered)

$obj->isDeleted Short-cut for CW$msg->label('deleted') For some folder types, you will get the time of deletion in return. This depends on the implementation. Example:

 next if $message->isDeleted;
 if(my $when = $message->isDeleted) {
    print scalar localtime $when;

$obj->isModified Returns whether this message is flagged as being modified. Modifications are changes in header lines, when a new body is set to the message (dangerous), or when labels change.

$obj->label(LABEL|PAIRS) Return the value of the LABEL, optionally after setting some values. In case of setting values, you specify key-value PAIRS. Labels are used to store knowledge about handling of the message within the folder. Flags about whether a message was read, replied to, or scheduled for deletion. Some labels are taken from the header's CWStatus and CWX-Status lines, however folder types like MH define a separate label file. Example:

 print $message->label('seen');
 if($message->label('seen')) {...};
 $message->label(seen => 1);
 $message->label(deleted => 1);  # same as $message->delete

$obj->labels Returns all known labels. In SCALAR context, it returns the knowledge as reference to a hash. This is a reference to the original data, but you shall *not* change that data directly: call CWlabel for changes! In LIST context, you get a list of names which are defined. Be warned that they will not all evaluate to true, although most of them will.

$obj->labelsToStatus When the labels were changed, that may effect the CWStatus and/or CWX-Status header lines of mbox messages. Read about the relation between these fields and the labels in the DETAILS chapter. The method will carefully only affect the result of modified() when there is a real change of flags, so not for each call to label().

$obj->modified([BOOLEAN]) Returns (optionally after setting) whether this message is flagged as being modified. See isModified().

$obj->statusToLabels Update the labels according the status lines in the header. See the description in the DETAILS chapter.

The whole message as text

$obj->file See The whole message as text in Mail::Message::Construct::Text

$obj->lines See The whole message as text in Mail::Message::Construct::Text

$obj->printStructure([FILEHANDLE|undef],[INDENT]) See The whole message as text in Mail::Message::Construct::Text

$obj->string See The whole message as text in Mail::Message::Construct::Text


$obj->clonedFrom Returns the MESSAGE which is the source of this message, which was created by a clone() operation.

Mail::Message->coerce(MESSAGE, OPTIONS) Coerce a MESSAGE into a Mail::Message. In some occasions, for instance where you add a message to a folder, this coercion is automatically called to ensure that the correct message type is stored. The coerced message is returned on success, otherwise CWundef. The coerced message may be a reblessed version of the original message or a new object. In case the message has to be specialized, for instance from a general Mail::Message into a Mail::Box::Mbox::Message, no copy is needed. However, to coerce a Mail::Internet object into a Mail::Message, a lot of copying and converting will take place. Valid MESSAGEs which can be coerced into Mail::Message objects are of type

* Any type of Mail::Box::Message
* MIME::Entity's, using Mail::Message::Convert::MimeEntity
* Mail::Internet's, using Mail::Message::Convert::MailInternet
Mail::Message::Part's, which are extensions of CWMail::Message's, can also be coerced directly from a Mail::Message::Body. Example:
 my $folder  = Mail::Box::Mbox->new;
 my $message = Mail::Message->build(...);
 my $coerced = Mail::Box::Mbox::Message->coerce($message);
Simpler replacement for the previous two lines:
 my $coerced = $folder->addMessage($message);

$obj->isDelayed Check whether the message is delayed (not yet read from file). Returns true or false, dependent on the body type.

$obj->readBody(PARSER, HEAD [, BODYTYPE]) Read a body of a message. The PARSER is the access to the folder's file, and the HEAD is already read. Information from the HEAD is used to create expectations about the message's length, but also to determine the mime-type and encodings of the body data. The BODYTYPE determines which kind of body will be made and defaults to the value specified by new(body_type). BODYTYPE may be the name of a body class, or a reference to a routine which returns the body's class when passed the HEAD as only argument.

$obj->readFromParser(PARSER, [BODYTYPE]) Read one message from file. The PARSER is opened on the file. First readHead() is called, and the head is stored in the message. Then readBody() is called, to produce a body. Also the body is added to the message without decodings being done. The optional BODYTYPE may be a body class or a reference to a code which returns a body-class based on the header.

$obj->readHead(PARSER [,CLASS]) Read a head into an object of the specified CLASS. The CLASS defaults to new(head_type). The PARSER is the access to the folder's file.

$obj->recursiveRebuildPart(PART, OPTIONS) See Internals in Mail::Message::Construct::Rebuild

$obj->storeBody(BODY) Where the body() method can be used to set and get a body, with all the necessary checks, this method is bluntly adding the specified body to the message. No conversions, not checking.

$obj->takeMessageId([STRING]) Take the message-id from the STRING, or create one when the CWundef is specified. If not STRING nor CWundef is given, the current header of the message is requested for the value of the CW'Message-ID' field. Angles (if present) are removed from the id.

Error handling

$obj->AUTOLOAD See METHODS in Mail::Message::Construct

$obj->addReport(OBJECT) See Error handling in Mail::Reporter


Mail::Message->defaultTrace([LEVEL]|[LOGLEVEL, TRACELEVEL]|[LEVEL, CALLBACK]) See Error handling in Mail::Reporter

$obj->errors See Error handling in Mail::Reporter

$obj->log([LEVEL [,STRINGS]])

Mail::Message->log([LEVEL [,STRINGS]]) See Error handling in Mail::Reporter


Mail::Message->logPriority(LEVEL) See Error handling in Mail::Reporter

$obj->logSettings See Error handling in Mail::Reporter

$obj->notImplemented See Error handling in Mail::Reporter

$obj->report([LEVEL]) See Error handling in Mail::Reporter

$obj->reportAll([LEVEL]) See Error handling in Mail::Reporter


Mail::Message->shortSize([VALUE]) Represent an integer VALUE representing the size of file or memory, (which can be large) into a short string using M and K (Megabytes and Kilobytes). Without VALUE, the size of the message head is used.

$obj->shortString Convert the message header to a short string (without trailing newline), representing the most important facts (for debugging purposes only). For now, it only reports size and subject.

$obj->trace([LEVEL]) See Error handling in Mail::Reporter

$obj->warnings See Error handling in Mail::Reporter


$obj->DESTROY When a message is to accessible anymore by any user's reference, Perl will call DESTROY for final clean-up. In this case, the head and body are released, and de-registered for the folder. You shall not call this yourself!

$obj->destruct Remove the information contained in the message object. This will be ignored when more than one reference to the same message object exists, because the method has the same effect as assigning CWundef to the variable which contains the reference. Normal garbage collection will call DESTROY() when possible. This method is only provided to hide differences with messages which are located in folders: their Mail::Box::Message::destruct() works quite differently. Example: of Mail::Message destruct

 my $msg = Mail::Message->read;
 $msg = undef;    # same

$obj->inGlobalDestruction See Cleanup in Mail::Reporter


Error: Cannot include forward source as CW$include.

Unknown alternative for the forward(include). Valid choices are CWNO, CWINLINE, CWATTACH, and CWENCAPSULATE.

Error: Cannot include reply source as CW$include.

Unknown alternative for the CWinclude option of reply(). Valid choices are CWNO, CWINLINE, and CWATTACH.

Error: No address to create forwarded to.

If a forward message is created, a destination address must be specified.

Error: No default mailer found to send message.

The message send() mechanism had not enough information to automatically find a mail transfer agent to sent this message. Specify a mailer explicitly using the CWvia options.

Error: Only build() Mail::Message's; they are not in a folder yet

You may wish to construct a message to be stored in a some kind of folder, but you need to do that in two steps. First, create a normal Mail::Message, and then add it to the folder. During this Mail::Box::addMessage() process, the message will get coerce()-d into the right message type, adding storage information and the like.

Error: Package CW$package does not implement CW$method.

Fatal error: the specific package (or one of its superclasses) does not implement this method where it should. This message means that some other related classes do implement this method however the class at hand does not. Probably you should investigate this and probably inform the author of the package.

Error: bounce requires To, Cc, or Bcc

The message bounce() method forwards a received message off to someone else without modification; you must specified it's new destination. If you have the urge not to specify any destination, you probably are looking for reply(). When you wish to modify the content, use forward().

Error: forwardAttach requires a preamble object

Error: forwardEncapsulate requires a preamble object

Error: no rebuild rule CW$name defined.


Structure of a Message

A MIME-compliant message is build upon two parts: the header and the body.

The header

The header is a list of fields, some spanning more than one line (folded) each telling something about the message. Information stored in here are for instance the sender of the message, the receivers of the message, when it was transported, how it was transported, etc. Headers can grow quite large.

In MailBox, each message object manages exactly one header object (a Mail::Message::Head) and one body object (a Mail::Message::Body). The header contains a list of header fields, which are represented by Mail::Message::Field objects.

The body

The body contains the payload: the data to be transfered. The data can be encoded, only accessible with a specific application, and may use some weird character-set, like Vietnamese; the MailBox distribution tries to assist you with handling these e-mails without the need to know all the details. This additional information (meta-information) about the body data is stored in the header. The header contains more information, for instance about the message transport and relations to other messages.

Message object implementation

The general idea about the structure of a message is

  |  |
  |  `-has-one--Mail::Message::Body

However: there are about 7 kinds of body objects, 3 kinds of headers and 3 kinds of fields. You will usually not see too much of these kinds, because they are merely created for performance reasons and can be used all the same, with the exception of the multipart bodies.

A multipart body is either a Mail::Message::Body::Multipart (mime type CWmultipart/*) or a Mail::Message::Body::Nested (mime type CWmessage/rfc822). These bodies are more complex:

               |  |
               |  `-has-one--Mail::Message::Body

Before you try to reconstruct multiparts or nested messages yourself, you can better take a look at Mail::Message::Construct::Rebuild.

Message class implementation

The class structure of messages is very close to that of folders. For instance, a Mail::Box::File::Message relates to a Mail::Box::File folder.

As extra level of inheritance, it has a Mail::Message, which is a message without location. And there is a special case of message: Mail::Message::Part is a message encapsulated in a multipart body.

The message types are:

 Mail::Box::Mbox::Message            Mail::Box::POP3::Message
 |  Mail::Box::Dbx::Message      Mail::Box::IMAP4::Message  |
 |  |                                                    |  |
 Mail::Box::File::Message             Mail::Box::Net::Message
         |                                      |
         |       Mail::Box::Maildir::Message    |
         |       |   Mail::Box::MH::Message     |
         |       |   |                          |
         |       Mail::Box::Dir::Message        |
         |                |                     |
         `------------.   |   .-----------------'
                      |   |   |
                   Mail::Box::Message    Mail::Message::Part
                          |                     |
                          |       .-------------'
                          |       |
                    Mail::Reporter (general base class)

By far most folder features are implemented in Mail::Box, so available to all folder types. Sometimes, features which appear in only some of the folder types are simulated for folders that miss them, like sub-folder support for MBOX.

Two strange other message types are defined: the Mail::Message::Dummy, which fills holes in Mail::Box::Thread::Node lists, and a Mail::Box::Message::Destructed, this is an on purpose demolished message to reduce memory consumption.


Labels (also named Flags) are used to indicate some special condition on the message, primary targeted on organizational issues: which messages are already read or should be deleted. There is a very strong user relation to labels.

The main complication is that each folder type has its own way of storing labels. To give an indication: MBOX folders use CWStatus and CWX-Status header fields, MH uses a file, MAILDIR encodes the flags in the message's filename, and IMAP has flags as part of the protocol.

Besides, some folder types can store labels with user defined names, where other lack that feature. Some folders have case-insensitive labels, other don't. Read all about the specifics in the manual page of the message type you actually have.

Predefined labels

To standardize the folder types, MailBox has defined the following labels, which can be used with the label() and labels() methods on all kinds of messages:

* deleted
This message is flagged to be deleted once the folder closes. Be very careful about the concept of 'delete' in a folder context : it is only a flag, and does not involve immediate action! This means, for instance, that the memory which is used by Perl to store the message is not released immediately (see destruct() if you need to). The methods delete(), deleted(), and isDeleted() are only short-cuts for managing the CWdelete label (as of MailBox 2.052).
* draft
The user has prepared this message, but is has not been send (yet). This flag is not automatically added to a message by MailBox, and has only a meaning in user applications.
* flagged
Messages can be flagged for some purpose, for instance as result of a search for spam in a folder. The Mail::Box::messages() method can be used to collect all these flagged messages from the folder. Probably it is more useful to use an understandable name (like CWspam) for these selections, however these self-defined labels can not stored in all folder types.
* old
The message was already in the folder when it was opened the last time, so was not recently added to the folder. This flag will never automatically be set by MailBox, because it would probably conflict with the user's idea of what is old.
* passed
Not often used or kept, this flag indicates that the message was bounced or forwarded to someone else.
* replied
The user (or application) has sent a message back to the sender of the message, as response of this one. This flag is automatically set if you use reply(), but not with forward() or bounce().
* seen
When this flag is set, the receiver of the message has consumed the message. A mail user agent (MUA) will set this flag when the user has opened the message once.

Status and X-Status fields

Mbox folders have no special means of storing information about messages (except the message separator line), and therefore have to revert to adding fields to the message header when something special comes up. This feature is also enabled for POP3, although whether that works depends on the POP server.

All applications which can handle mbox folders support the CWStatus and CWX-Status field convensions. The following encoding is used:

 Flag   Field       Label
 R      Status   => seen    (Read)
 O      Status   => old     (not recent)
 A      X-Status => replied (Answered)
 F      X-Status => flagged

There is no special flag for CWdeleted, which most other folders support: messages flagged to be deleted will never be written to a folder file when it is closed.


See the MailBox website at <> for more details.


Distribution version 2.063. Written by Mark Overmeer ( See the ChangeLog for other contributors.

Copyright (c) 2001-2003 by the author(s). All rights reserved. This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.