Doc/lib/libmailbox.tex

\section{\module{mailbox} ---
          Manipulate mailboxes in various formats}

\declaremodule{}{mailbox}
\moduleauthor{Gregory K.~Johnson}{gkj@gregorykjohnson.com}
\sectionauthor{Gregory K.~Johnson}{gkj@gregorykjohnson.com}
\modulesynopsis{Manipulate mailboxes in various formats}


This module defines two classes, \class{Mailbox} and \class{Message}, for
accessing and manipulating on-disk mailboxes and the messages they contain.
\class{Mailbox} offers a dictionary-like mapping from keys to messages.
\class{Message} extends the \module{email.Message} module's \class{Message}
class with format-specific state and behavior. Supported mailbox formats are
Maildir, mbox, MH, Babyl, and MMDF.

\begin{seealso}
    \seemodule{email}{Represent and manipulate messages.}
\end{seealso}

\subsection{\class{Mailbox} objects}
\label{mailbox-objects}

\begin{classdesc*}{Mailbox}
A mailbox, which may be inspected and modified.
\end{classdesc*}

The \class{Mailbox} class defines an interface and
is not intended to be instantiated.  Instead, format-specific
subclasses should inherit from \class{Mailbox} and your code
should instantiate a particular subclass.

The \class{Mailbox} interface is dictionary-like, with small keys
corresponding to messages. Keys are issued by the \class{Mailbox}
instance with which they will be used and are only meaningful to that
\class{Mailbox} instance. A key continues to identify a message even
if the corresponding message is modified, such as by replacing it with
another message.

Messages may be added to a \class{Mailbox} instance using the set-like
method \method{add()} and removed using a \code{del} statement or the
set-like methods \method{remove()} and \method{discard()}.

\class{Mailbox} interface semantics differ from dictionary semantics in some
noteworthy ways. Each time a message is requested, a new
representation (typically a \class{Message} instance) is generated
based upon the current state of the mailbox. Similarly, when a message
is added to a \class{Mailbox} instance, the provided message
representation's contents are copied. In neither case is a reference
to the message representation kept by the \class{Mailbox} instance.

The default \class{Mailbox} iterator iterates over message representations, not
keys as the default dictionary iterator does. Moreover, modification of a
mailbox during iteration is safe and well-defined. Messages added to the
mailbox after an iterator is created will not be seen by the iterator. Messages
removed from the mailbox before the iterator yields them will be silently
skipped, though using a key from an iterator may result in a
\exception{KeyError} exception if the corresponding message is subsequently
removed.

\begin{notice}[warning]
Be very cautious when modifying mailboxes that might be
simultaneously changed by some other process.  The safest mailbox
format to use for such tasks is Maildir; try to avoid using
single-file formats such as mbox for concurrent writing.  If you're
modifying a mailbox, you
\emph{must} lock it by calling the \method{lock()} and
\method{unlock()} methods \emph{before} reading any messages in the file
or making any changes by adding or deleting a message.  Failing to
lock the mailbox runs the risk of losing messages or corrupting the entire
mailbox.
\end{notice}

\class{Mailbox} instances have the following methods:

\begin{methoddesc}{add}{message}
Add \var{message} to the mailbox and return the key that has been assigned to
it.

Parameter \var{message} may be a \class{Message} instance, an
\class{email.Message.Message} instance, a string, or a file-like object (which
should be open in text mode). If \var{message} is an instance of the
appropriate format-specific \class{Message} subclass (e.g., if it's an
\class{mboxMessage} instance and this is an \class{mbox} instance), its
format-specific information is used. Otherwise, reasonable defaults for
format-specific information are used.
\end{methoddesc}

\begin{methoddesc}{remove}{key}
\methodline{__delitem__}{key}
\methodline{discard}{key}
Delete the message corresponding to \var{key} from the mailbox.

If no such message exists, a \exception{KeyError} exception is raised if the
method was called as \method{remove()} or \method{__delitem__()} but no
exception is raised if the method was called as \method{discard()}. The
behavior of \method{discard()} may be preferred if the underlying mailbox
format supports concurrent modification by other processes.
\end{methoddesc}

\begin{methoddesc}{__setitem__}{key, message}
Replace the message corresponding to \var{key} with \var{message}. Raise a
\exception{KeyError} exception if no message already corresponds to \var{key}.

As with \method{add()}, parameter \var{message} may be a \class{Message}
instance, an \class{email.Message.Message} instance, a string, or a file-like
object (which should be open in text mode). If \var{message} is an instance of
the appropriate format-specific \class{Message} subclass (e.g., if it's an
\class{mboxMessage} instance and this is an \class{mbox} instance), its
format-specific information is used. Otherwise, the format-specific information
of the message that currently corresponds to \var{key} is left unchanged. 
\end{methoddesc}

\begin{methoddesc}{iterkeys}{}
\methodline{keys}{}
Return an iterator over all keys if called as \method{iterkeys()} or return a
list of keys if called as \method{keys()}.
\end{methoddesc}

\begin{methoddesc}{itervalues}{}
\methodline{__iter__}{}
\methodline{values}{}
Return an iterator over representations of all messages if called as
\method{itervalues()} or \method{__iter__()} or return a list of such
representations if called as \method{values()}. The messages are represented as
instances of the appropriate format-specific \class{Message} subclass unless a
custom message factory was specified when the \class{Mailbox} instance was
initialized. \note{The behavior of \method{__iter__()} is unlike that of
dictionaries, which iterate over keys.}
\end{methoddesc}

\begin{methoddesc}{iteritems}{}
\methodline{items}{}
Return an iterator over (\var{key}, \var{message}) pairs, where \var{key} is a
key and \var{message} is a message representation, if called as
\method{iteritems()} or return a list of such pairs if called as
\method{items()}. The messages are represented as instances of the appropriate
format-specific \class{Message} subclass unless a custom message factory was
specified when the \class{Mailbox} instance was initialized.
\end{methoddesc}

\begin{methoddesc}{get}{key\optional{, default=None}}
\methodline{__getitem__}{key}
Return a representation of the message corresponding to \var{key}. If no such
message exists, \var{default} is returned if the method was called as
\method{get()} and a \exception{KeyError} exception is raised if the method was
called as \method{__getitem__()}. The message is represented as an instance of
the appropriate format-specific \class{Message} subclass unless a custom
message factory was specified when the \class{Mailbox} instance was
initialized.
\end{methoddesc}

\begin{methoddesc}{get_message}{key}
Return a representation of the message corresponding to \var{key} as an
instance of the appropriate format-specific \class{Message} subclass, or raise
a \exception{KeyError} exception if no such message exists.
\end{methoddesc}

\begin{methoddesc}{get_string}{key}
Return a string representation of the message corresponding to \var{key}, or
raise a \exception{KeyError} exception if no such message exists.
\end{methoddesc}

\begin{methoddesc}{get_file}{key}
Return a file-like representation of the message corresponding to \var{key},
or raise a \exception{KeyError} exception if no such message exists. The
file-like object behaves as if open in binary mode. This file should be closed
once it is no longer needed.

\note{Unlike other representations of messages, file-like representations are
not necessarily independent of the \class{Mailbox} instance that created them
or of the underlying mailbox. More specific documentation is provided by each
subclass.}
\end{methoddesc}

\begin{methoddesc}{has_key}{key}
\methodline{__contains__}{key}
Return \code{True} if \var{key} corresponds to a message, \code{False}
otherwise.
\end{methoddesc}

\begin{methoddesc}{__len__}{}
Return a count of messages in the mailbox.
\end{methoddesc}

\begin{methoddesc}{clear}{}
Delete all messages from the mailbox.
\end{methoddesc}

\begin{methoddesc}{pop}{key\optional{, default}}
Return a representation of the message corresponding to \var{key} and delete
the message. If no such message exists, return \var{default} if it was supplied
or else raise a \exception{KeyError} exception. The message is represented as
an instance of the appropriate format-specific \class{Message} subclass unless
a custom message factory was specified when the \class{Mailbox} instance was
initialized.
\end{methoddesc}

\begin{methoddesc}{popitem}{}
Return an arbitrary (\var{key}, \var{message}) pair, where \var{key} is a key
and \var{message} is a message representation, and delete the corresponding
message. If the mailbox is empty, raise a \exception{KeyError} exception. The
message is represented as an instance of the appropriate format-specific
\class{Message} subclass unless a custom message factory was specified when the
\class{Mailbox} instance was initialized.
\end{methoddesc}

\begin{methoddesc}{update}{arg}
Parameter \var{arg} should be a \var{key}-to-\var{message} mapping or an
iterable of (\var{key}, \var{message}) pairs. Updates the mailbox so that, for
each given \var{key} and \var{message}, the message corresponding to \var{key}
is set to \var{message} as if by using \method{__setitem__()}. As with
\method{__setitem__()}, each \var{key} must already correspond to a message in
the mailbox or else a \exception{KeyError} exception will be raised, so in
general it is incorrect for \var{arg} to be a \class{Mailbox} instance.
\note{Unlike with dictionaries, keyword arguments are not supported.}
\end{methoddesc}

\begin{methoddesc}{flush}{}
Write any pending changes to the filesystem. For some \class{Mailbox}
subclasses, changes are always written immediately and \method{flush()} does
nothing, but you should still make a habit of calling this method.
\end{methoddesc}

\begin{methoddesc}{lock}{}
Acquire an exclusive advisory lock on the mailbox so that other processes know
not to modify it. An \exception{ExternalClashError} is raised if the lock is
not available. The particular locking mechanisms used depend upon the mailbox
format.  You should \emph{always} lock the mailbox before making any 
modifications to its contents.
\end{methoddesc}

\begin{methoddesc}{unlock}{}
Release the lock on the mailbox, if any.
\end{methoddesc}

\begin{methoddesc}{close}{}
Flush the mailbox, unlock it if necessary, and close any open files. For some
\class{Mailbox} subclasses, this method does nothing.
\end{methoddesc}


\subsubsection{\class{Maildir}}
\label{mailbox-maildir}

\begin{classdesc}{Maildir}{dirname\optional{, factory=rfc822.Message\optional{,
create=True}}}
A subclass of \class{Mailbox} for mailboxes in Maildir format. Parameter
\var{factory} is a callable object that accepts a file-like message
representation (which behaves as if opened in binary mode) and returns a custom
representation. If \var{factory} is \code{None}, \class{MaildirMessage} is used
as the default message representation. If \var{create} is \code{True}, the
mailbox is created if it does not exist.

It is for historical reasons that \var{factory} defaults to
\class{rfc822.Message} and that \var{dirname} is named as such rather than
\var{path}. For a \class{Maildir} instance that behaves like instances of other
\class{Mailbox} subclasses, set \var{factory} to \code{None}.
\end{classdesc}

Maildir is a directory-based mailbox format invented for the qmail mail
transfer agent and now widely supported by other programs. Messages in a
Maildir mailbox are stored in separate files within a common directory
structure. This design allows Maildir mailboxes to be accessed and modified by
multiple unrelated programs without data corruption, so file locking is
unnecessary.

Maildir mailboxes contain three subdirectories, namely: \file{tmp}, \file{new},
and \file{cur}. Messages are created momentarily in the \file{tmp} subdirectory
and then moved to the \file{new} subdirectory to finalize delivery. A mail user
agent may subsequently move the message to the \file{cur} subdirectory and
store information about the state of the message in a special "info" section
appended to its file name.

Folders of the style introduced by the Courier mail transfer agent are also
supported. Any subdirectory of the main mailbox is considered a folder if
\character{.} is the first character in its name. Folder names are represented
by \class{Maildir} without the leading \character{.}. Each folder is itself a
Maildir mailbox but should not contain other folders. Instead, a logical
nesting is indicated using \character{.} to delimit levels, e.g.,
"Archived.2005.07".

\begin{notice}
The Maildir specification requires the use of a colon (\character{:}) in
certain message file names. However, some operating systems do not permit this
character in file names, If you wish to use a Maildir-like format on such an
operating system, you should specify another character to use instead. The
exclamation point (\character{!}) is a popular choice. For example:
\begin{verbatim}
import mailbox
mailbox.Maildir.colon = '!'
\end{verbatim}
The \member{colon} attribute may also be set on a per-instance basis.
\end{notice}

\class{Maildir} instances have all of the methods of \class{Mailbox} in
addition to the following:

\begin{methoddesc}{list_folders}{}
Return a list of the names of all folders.
\end{methoddesc}

\begin{methoddesc}{get_folder}{folder}
Return a \class{Maildir} instance representing the folder whose name is
\var{folder}. A \exception{NoSuchMailboxError} exception is raised if the
folder does not exist.
\end{methoddesc}

\begin{methoddesc}{add_folder}{folder}
Create a folder whose name is \var{folder} and return a \class{Maildir}
instance representing it.
\end{methoddesc}

\begin{methoddesc}{remove_folder}{folder}
Delete the folder whose name is \var{folder}. If the folder contains any
messages, a \exception{NotEmptyError} exception will be raised and the folder
will not be deleted.
\end{methoddesc}

\begin{methoddesc}{clean}{}
Delete temporary files from the mailbox that have not been accessed in the
last 36 hours. The Maildir specification says that mail-reading programs
should do this occasionally.
\end{methoddesc}

Some \class{Mailbox} methods implemented by \class{Maildir} deserve special
remarks:

\begin{methoddesc}{add}{message}
\methodline[Maildir]{__setitem__}{key, message}
\methodline[Maildir]{update}{arg}
\warning{These methods generate unique file names based upon the current
process ID. When using multiple threads, undetected name clashes may occur and
cause corruption of the mailbox unless threads are coordinated to avoid using
these methods to manipulate the same mailbox simultaneously.}
\end{methoddesc}

\begin{methoddesc}{flush}{}
All changes to Maildir mailboxes are immediately applied, so this method does
nothing.
\end{methoddesc}

\begin{methoddesc}{lock}{}
\methodline{unlock}{}
Maildir mailboxes do not support (or require) locking, so these methods do
nothing. 
\end{methoddesc}

\begin{methoddesc}{close}{}
\class{Maildir} instances do not keep any open files and the underlying
mailboxes do not support locking, so this method does nothing.
\end{methoddesc}

\begin{methoddesc}{get_file}{key}
Depending upon the host platform, it may not be possible to modify or remove
the underlying message while the returned file remains open.
\end{methoddesc}

\begin{seealso}
    \seelink{http://www.qmail.org/man/man5/maildir.html}{maildir man page from
    qmail}{The original specification of the format.}
    \seelink{http://cr.yp.to/proto/maildir.html}{Using maildir format}{Notes
    on Maildir by its inventor. Includes an updated name-creation scheme and
    details on "info" semantics.}
    \seelink{http://www.courier-mta.org/?maildir.html}{maildir man page from
    Courier}{Another specification of the format. Describes a common extension
    for supporting folders.}
\end{seealso}

\subsubsection{\class{mbox}}
\label{mailbox-mbox}

\begin{classdesc}{mbox}{path\optional{, factory=None\optional{, create=True}}}
A subclass of \class{Mailbox} for mailboxes in mbox format. Parameter
\var{factory} is a callable object that accepts a file-like message
representation (which behaves as if opened in binary mode) and returns a custom
representation. If \var{factory} is \code{None}, \class{mboxMessage} is used as
the default message representation. If \var{create} is \code{True}, the mailbox
is created if it does not exist.
\end{classdesc}

The mbox format is the classic format for storing mail on \UNIX{} systems. All
messages in an mbox mailbox are stored in a single file with the beginning of
each message indicated by a line whose first five characters are "From~".

Several variations of the mbox format exist to address perceived shortcomings
in the original. In the interest of compatibility, \class{mbox} implements the
original format, which is sometimes referred to as \dfn{mboxo}. This means that
the \mailheader{Content-Length} header, if present, is ignored and that any
occurrences of "From~" at the beginning of a line in a message body are
transformed to ">From~" when storing the message, although occurences of
">From~" are not transformed to "From~" when reading the message.

Some \class{Mailbox} methods implemented by \class{mbox} deserve special
remarks:

\begin{methoddesc}{get_file}{key}
Using the file after calling \method{flush()} or \method{close()} on the
\class{mbox} instance may yield unpredictable results or raise an exception.
\end{methoddesc}

\begin{methoddesc}{lock}{}
\methodline{unlock}{}
Three locking mechanisms are used---dot locking and, if available, the
\cfunction{flock()} and \cfunction{lockf()} system calls.
\end{methoddesc}

\begin{seealso}
    \seelink{http://www.qmail.org/man/man5/mbox.html}{mbox man page from
    qmail}{A specification of the format and its variations.}
    \seelink{http://www.tin.org/bin/man.cgi?section=5\&topic=mbox}{mbox man
    page from tin}{Another specification of the format, with details on
    locking.}
    \seelink{http://home.netscape.com/eng/mozilla/2.0/relnotes/demo/content-length.html}
    {Configuring Netscape Mail on \UNIX{}: Why The Content-Length Format is
    Bad}{An argument for using the original mbox format rather than a
    variation.}
    \seelink{http://homepages.tesco.net./\tilde{}J.deBoynePollard/FGA/mail-mbox-formats.html}
    {"mbox" is a family of several mutually incompatible mailbox formats}{A
    history of mbox variations.}
\end{seealso}

\subsubsection{\class{MH}}
\label{mailbox-mh}

\begin{classdesc}{MH}{path\optional{, factory=None\optional{, create=True}}}
A subclass of \class{Mailbox} for mailboxes in MH format. Parameter
\var{factory} is a callable object that accepts a file-like message
representation (which behaves as if opened in binary mode) and returns a custom
representation. If \var{factory} is \code{None}, \class{MHMessage} is used as
the default message representation. If \var{create} is \code{True}, the mailbox
is created if it does not exist.
\end{classdesc}

MH is a directory-based mailbox format invented for the MH Message Handling
System, a mail user agent. Each message in an MH mailbox resides in its own
file. An MH mailbox may contain other MH mailboxes (called \dfn{folders}) in
addition to messages. Folders may be nested indefinitely. MH mailboxes also
support \dfn{sequences}, which are named lists used to logically group messages
without moving them to sub-folders. Sequences are defined in a file called
\file{.mh_sequences} in each folder.

The \class{MH} class manipulates MH mailboxes, but it does not attempt to
emulate all of \program{mh}'s behaviors. In particular, it does not modify and
is not affected by the \file{context} or \file{.mh_profile} files that are used
by \program{mh} to store its state and configuration.

\class{MH} instances have all of the methods of \class{Mailbox} in addition to
the following:

\begin{methoddesc}{list_folders}{}
Return a list of the names of all folders.
\end{methoddesc}

\begin{methoddesc}{get_folder}{folder}
Return an \class{MH} instance representing the folder whose name is
\var{folder}. A \exception{NoSuchMailboxError} exception is raised if the
folder does not exist.
\end{methoddesc}

\begin{methoddesc}{add_folder}{folder}
Create a folder whose name is \var{folder} and return an \class{MH} instance
representing it.
\end{methoddesc}

\begin{methoddesc}{remove_folder}{folder}
Delete the folder whose name is \var{folder}. If the folder contains any
messages, a \exception{NotEmptyError} exception will be raised and the folder
will not be deleted.
\end{methoddesc}

\begin{methoddesc}{get_sequences}{}
Return a dictionary of sequence names mapped to key lists. If there are no
sequences, the empty dictionary is returned.
\end{methoddesc}

\begin{methoddesc}{set_sequences}{sequences}
Re-define the sequences that exist in the mailbox based upon \var{sequences}, a
dictionary of names mapped to key lists, like returned by
\method{get_sequences()}.
\end{methoddesc}

\begin{methoddesc}{pack}{}
Rename messages in the mailbox as necessary to eliminate gaps in numbering.
Entries in the sequences list are updated correspondingly. \note{Already-issued
keys are invalidated by this operation and should not be subsequently used.}
\end{methoddesc}

Some \class{Mailbox} methods implemented by \class{MH} deserve special remarks:

\begin{methoddesc}{remove}{key}
\methodline{__delitem__}{key}
\methodline{discard}{key}
These methods immediately delete the message. The MH convention of marking a
message for deletion by prepending a comma to its name is not used.
\end{methoddesc}

\begin{methoddesc}{lock}{}
\methodline{unlock}{}
Three locking mechanisms are used---dot locking and, if available, the
\cfunction{flock()} and \cfunction{lockf()} system calls. For MH mailboxes,
locking the mailbox means locking the \file{.mh_sequences} file and, only for
the duration of any operations that affect them, locking individual message
files.
\end{methoddesc}

\begin{methoddesc}{get_file}{key}
Depending upon the host platform, it may not be possible to remove the
underlying message while the returned file remains open.
\end{methoddesc}

\begin{methoddesc}{flush}{}
All changes to MH mailboxes are immediately applied, so this method does
nothing.
\end{methoddesc}

\begin{methoddesc}{close}{}
\class{MH} instances do not keep any open files, so this method is equivelant
to \method{unlock()}.
\end{methoddesc}

\begin{seealso}
\seelink{http://www.nongnu.org/nmh/}{nmh - Message Handling System}{Home page
of \program{nmh}, an updated version of the original \program{mh}.}
\seelink{http://www.ics.uci.edu/\tilde{}mh/book/}{MH \& nmh: Email for Users \&
Programmers}{A GPL-licensed book on \program{mh} and \program{nmh}, with some
information on the mailbox format.}
\end{seealso}

\subsubsection{\class{Babyl}}
\label{mailbox-babyl}

\begin{classdesc}{Babyl}{path\optional{, factory=None\optional{, create=True}}}
A subclass of \class{Mailbox} for mailboxes in Babyl format. Parameter
\var{factory} is a callable object that accepts a file-like message
representation (which behaves as if opened in binary mode) and returns a custom
representation. If \var{factory} is \code{None}, \class{BabylMessage} is used
as the default message representation. If \var{create} is \code{True}, the
mailbox is created if it does not exist.
\end{classdesc}

Babyl is a single-file mailbox format used by the Rmail mail user agent
included with Emacs. The beginning of a message is indicated by a line
containing the two characters Control-Underscore
(\character{\textbackslash037}) and Control-L (\character{\textbackslash014}).
The end of a message is indicated by the start of the next message or, in the
case of the last message, a line containing a Control-Underscore
(\character{\textbackslash037}) character.

Messages in a Babyl mailbox have two sets of headers, original headers and
so-called visible headers. Visible headers are typically a subset of the
original headers that have been reformatted or abridged to be more attractive.
Each message in a Babyl mailbox also has an accompanying list of \dfn{labels},
or short strings that record extra information about the message, and a list of
all user-defined labels found in the mailbox is kept in the Babyl options
section.

\class{Babyl} instances have all of the methods of \class{Mailbox} in addition
to the following:

\begin{methoddesc}{get_labels}{}
Return a list of the names of all user-defined labels used in the mailbox.
\note{The actual messages are inspected to determine which labels exist in the
mailbox rather than consulting the list of labels in the Babyl options section,
but the Babyl section is updated whenever the mailbox is modified.}
\end{methoddesc}

Some \class{Mailbox} methods implemented by \class{Babyl} deserve special
remarks:

\begin{methoddesc}{get_file}{key}
In Babyl mailboxes, the headers of a message are not stored contiguously with
the body of the message. To generate a file-like representation, the headers
and body are copied together into a \class{StringIO} instance (from the
\module{StringIO} module), which has an API identical to that of a file. As a
result, the file-like object is truly independent of the underlying mailbox but
does not save memory compared to a string representation.
\end{methoddesc}

\begin{methoddesc}{lock}{}
\methodline{unlock}{}
Three locking mechanisms are used---dot locking and, if available, the
\cfunction{flock()} and \cfunction{lockf()} system calls.
\end{methoddesc}

\begin{seealso}
\seelink{http://quimby.gnus.org/notes/BABYL}{Format of Version 5 Babyl Files}{A
specification of the Babyl format.}
\seelink{http://www.gnu.org/software/emacs/manual/html_node/Rmail.html}{Reading
Mail with Rmail}{The Rmail manual, with some information on Babyl semantics.}
\end{seealso}

\subsubsection{\class{MMDF}}
\label{mailbox-mmdf}

\begin{classdesc}{MMDF}{path\optional{, factory=None\optional{, create=True}}}
A subclass of \class{Mailbox} for mailboxes in MMDF format. Parameter
\var{factory} is a callable object that accepts a file-like message
representation (which behaves as if opened in binary mode) and returns a custom
representation. If \var{factory} is \code{None}, \class{MMDFMessage} is used as
the default message representation. If \var{create} is \code{True}, the mailbox
is created if it does not exist.
\end{classdesc}

MMDF is a single-file mailbox format invented for the Multichannel Memorandum
Distribution Facility, a mail transfer agent. Each message is in the same form
as an mbox message but is bracketed before and after by lines containing four
Control-A (\character{\textbackslash001}) characters. As with the mbox format,
the beginning of each message is indicated by a line whose first five
characters are "From~", but additional occurrences of "From~" are not
transformed to ">From~" when storing messages because the extra message
separator lines prevent mistaking such occurrences for the starts of subsequent
messages.

Some \class{Mailbox} methods implemented by \class{MMDF} deserve special
remarks:

\begin{methoddesc}{get_file}{key}
Using the file after calling \method{flush()} or \method{close()} on the
\class{MMDF} instance may yield unpredictable results or raise an exception.
\end{methoddesc}

\begin{methoddesc}{lock}{}
\methodline{unlock}{}
Three locking mechanisms are used---dot locking and, if available, the
\cfunction{flock()} and \cfunction{lockf()} system calls.
\end{methoddesc}

\begin{seealso}
\seelink{http://www.tin.org/bin/man.cgi?section=5\&topic=mmdf}{mmdf man page
from tin}{A specification of MMDF format from the documentation of tin, a
newsreader.}
\seelink{http://en.wikipedia.org/wiki/MMDF}{MMDF}{A Wikipedia article
describing the Multichannel Memorandum Distribution Facility.}
\end{seealso}

\subsection{\class{Message} objects}
\label{mailbox-message-objects}

\begin{classdesc}{Message}{\optional{message}}
A subclass of the \module{email.Message} module's \class{Message}. Subclasses
of \class{mailbox.Message} add mailbox-format-specific state and behavior.

If \var{message} is omitted, the new instance is created in a default, empty
state. If \var{message} is an \class{email.Message.Message} instance, its
contents are copied; furthermore, any format-specific information is converted
insofar as possible if \var{message} is a \class{Message} instance. If
\var{message} is a string or a file, it should contain an \rfc{2822}-compliant
message, which is read and parsed.
\end{classdesc}

The format-specific state and behaviors offered by subclasses vary, but in
general it is only the properties that are not specific to a particular mailbox
that are supported (although presumably the properties are specific to a
particular mailbox format). For example, file offsets for single-file mailbox
formats and file names for directory-based mailbox formats are not retained,
because they are only applicable to the original mailbox. But state such as
whether a message has been read by the user or marked as important is retained,
because it applies to the message itself.

There is no requirement that \class{Message} instances be used to represent
messages retrieved using \class{Mailbox} instances. In some situations, the
time and memory required to generate \class{Message} representations might not
not acceptable. For such situations, \class{Mailbox} instances also offer
string and file-like representations, and a custom message factory may be
specified when a \class{Mailbox} instance is initialized. 

\subsubsection{\class{MaildirMessage}}
\label{mailbox-maildirmessage}

\begin{classdesc}{MaildirMessage}{\optional{message}}
A message with Maildir-specific behaviors. Parameter \var{message}
has the same meaning as with the \class{Message} constructor.
\end{classdesc}

Typically, a mail user agent application moves all of the messages in the
\file{new} subdirectory to the \file{cur} subdirectory after the first time the
user opens and closes the mailbox, recording that the messages are old whether
or not they've actually been read. Each message in \file{cur} has an "info"
section added to its file name to store information about its state. (Some mail
readers may also add an "info" section to messages in \file{new}.) The "info"
section may take one of two forms: it may contain "2," followed by a list of
standardized flags (e.g., "2,FR") or it may contain "1," followed by so-called
experimental information. Standard flags for Maildir messages are as follows:

\begin{tableiii}{l|l|l}{textrm}{Flag}{Meaning}{Explanation}
\lineiii{D}{Draft}{Under composition}
\lineiii{F}{Flagged}{Marked as important}
\lineiii{P}{Passed}{Forwarded, resent, or bounced}
\lineiii{R}{Replied}{Replied to}
\lineiii{S}{Seen}{Read}
\lineiii{T}{Trashed}{Marked for subsequent deletion}
\end{tableiii}

\class{MaildirMessage} instances offer the following methods:

\begin{methoddesc}{get_subdir}{}
Return either "new" (if the message should be stored in the \file{new}
subdirectory) or "cur" (if the message should be stored in the \file{cur}
subdirectory). \note{A message is typically moved from \file{new} to \file{cur}
after its mailbox has been accessed, whether or not the message is has been
read. A message \code{msg} has been read if \code{"S" not in msg.get_flags()}
is \code{True}.}
\end{methoddesc}

\begin{methoddesc}{set_subdir}{subdir}
Set the subdirectory the message should be stored in. Parameter \var{subdir}
must be either "new" or "cur".
\end{methoddesc}

\begin{methoddesc}{get_flags}{}
Return a string specifying the flags that are currently set. If the message
complies with the standard Maildir format, the result is the concatenation in
alphabetical order of zero or one occurrence of each of \character{D},
\character{F}, \character{P}, \character{R}, \character{S}, and \character{T}.
The empty string is returned if no flags are set or if "info" contains
experimental semantics.
\end{methoddesc}

\begin{methoddesc}{set_flags}{flags}
Set the flags specified by \var{flags} and unset all others.
\end{methoddesc}

\begin{methoddesc}{add_flag}{flag}
Set the flag(s) specified by \var{flag} without changing other flags. To add
more than one flag at a time, \var{flag} may be a string of more than one
character. The current "info" is overwritten whether or not it contains
experimental information rather than
flags.
\end{methoddesc}

\begin{methoddesc}{remove_flag}{flag}
Unset the flag(s) specified by \var{flag} without changing other flags. To
remove more than one flag at a time, \var{flag} maybe a string of more than one
character. If "info" contains experimental information rather than flags, the
current "info" is not modified.
\end{methoddesc}

\begin{methoddesc}{get_date}{}
Return the delivery date of the message as a floating-point number representing
seconds since the epoch.
\end{methoddesc}

\begin{methoddesc}{set_date}{date}
Set the delivery date of the message to \var{date}, a floating-point number
representing seconds since the epoch.
\end{methoddesc}

\begin{methoddesc}{get_info}{}
Return a string containing the "info" for a message. This is useful for
accessing and modifying "info" that is experimental (i.e., not a list of
flags).
\end{methoddesc}

\begin{methoddesc}{set_info}{info}
Set "info" to \var{info}, which should be a string.
\end{methoddesc}

When a \class{MaildirMessage} instance is created based upon an
\class{mboxMessage} or \class{MMDFMessage} instance, the \mailheader{Status}
and \mailheader{X-Status} headers are omitted and the following conversions
take place:

\begin{tableii}{l|l}{textrm}
    {Resulting state}{\class{mboxMessage} or \class{MMDFMessage} state}
\lineii{"cur" subdirectory}{O flag}
\lineii{F flag}{F flag}
\lineii{R flag}{A flag}
\lineii{S flag}{R flag}
\lineii{T flag}{D flag}
\end{tableii}

When a \class{MaildirMessage} instance is created based upon an
\class{MHMessage} instance, the following conversions take place:

\begin{tableii}{l|l}{textrm}
    {Resulting state}{\class{MHMessage} state}
\lineii{"cur" subdirectory}{"unseen" sequence}
\lineii{"cur" subdirectory and S flag}{no "unseen" sequence}
\lineii{F flag}{"flagged" sequence}
\lineii{R flag}{"replied" sequence}
\end{tableii}

When a \class{MaildirMessage} instance is created based upon a
\class{BabylMessage} instance, the following conversions take place:

\begin{tableii}{l|l}{textrm}
    {Resulting state}{\class{BabylMessage} state}
\lineii{"cur" subdirectory}{"unseen" label}
\lineii{"cur" subdirectory and S flag}{no "unseen" label}
\lineii{P flag}{"forwarded" or "resent" label}
\lineii{R flag}{"answered" label}
\lineii{T flag}{"deleted" label}
\end{tableii}

\subsubsection{\class{mboxMessage}}
\label{mailbox-mboxmessage}

\begin{classdesc}{mboxMessage}{\optional{message}}
A message with mbox-specific behaviors. Parameter \var{message} has the same
meaning as with the \class{Message} constructor.
\end{classdesc}

Messages in an mbox mailbox are stored together in a single file. The sender's
envelope address and the time of delivery are typically stored in a line
beginning with "From~" that is used to indicate the start of a message, though
there is considerable variation in the exact format of this data among mbox
implementations. Flags that indicate the state of the message, such as whether
it has been read or marked as important, are typically stored in
\mailheader{Status} and \mailheader{X-Status} headers.

Conventional flags for mbox messages are as follows:

\begin{tableiii}{l|l|l}{textrm}{Flag}{Meaning}{Explanation}
\lineiii{R}{Read}{Read}
\lineiii{O}{Old}{Previously detected by MUA}
\lineiii{D}{Deleted}{Marked for subsequent deletion}
\lineiii{F}{Flagged}{Marked as important}
\lineiii{A}{Answered}{Replied to}
\end{tableiii}

The "R" and "O" flags are stored in the \mailheader{Status} header, and the
"D", "F", and "A" flags are stored in the \mailheader{X-Status} header. The
flags and headers typically appear in the order mentioned.

\class{mboxMessage} instances offer the following methods:

\begin{methoddesc}{get_from}{}
Return a string representing the "From~" line that marks the start of the
message in an mbox mailbox. The leading "From~" and the trailing newline are
excluded.
\end{methoddesc}

\begin{methoddesc}{set_from}{from_\optional{, time_=None}}
Set the "From~" line to \var{from_}, which should be specified without a
leading "From~" or trailing newline. For convenience, \var{time_} may be
specified and will be formatted appropriately and appended to \var{from_}. If
\var{time_} is specified, it should be a \class{struct_time} instance, a tuple
suitable for passing to \method{time.strftime()}, or \code{True} (to use
\method{time.gmtime()}).
\end{methoddesc}

\begin{methoddesc}{get_flags}{}
Return a string specifying the flags that are currently set. If the message
complies with the conventional format, the result is the concatenation in the
following order of zero or one occurrence of each of \character{R},
\character{O}, \character{D}, \character{F}, and \character{A}.
\end{methoddesc}

\begin{methoddesc}{set_flags}{flags}
Set the flags specified by \var{flags} and unset all others. Parameter
\var{flags} should be the concatenation in any order of zero or more
occurrences of each of \character{R}, \character{O}, \character{D},
\character{F}, and \character{A}.
\end{methoddesc}

\begin{methoddesc}{add_flag}{flag}
Set the flag(s) specified by \var{flag} without changing other flags. To add
more than one flag at a time, \var{flag} may be a string of more than one
character.
\end{methoddesc}

\begin{methoddesc}{remove_flag}{flag}
Unset the flag(s) specified by \var{flag} without changing other flags. To
remove more than one flag at a time, \var{flag} maybe a string of more than one
character.
\end{methoddesc}

When an \class{mboxMessage} instance is created based upon a
\class{MaildirMessage} instance, a "From~" line is generated based upon the
\class{MaildirMessage} instance's delivery date, and the following conversions
take place:

\begin{tableii}{l|l}{textrm}
    {Resulting state}{\class{MaildirMessage} state}
\lineii{R flag}{S flag}
\lineii{O flag}{"cur" subdirectory}
\lineii{D flag}{T flag}
\lineii{F flag}{F flag}
\lineii{A flag}{R flag}
\end{tableii}

When an \class{mboxMessage} instance is created based upon an \class{MHMessage}
instance, the following conversions take place:

\begin{tableii}{l|l}{textrm}
    {Resulting state}{\class{MHMessage} state}
\lineii{R flag and O flag}{no "unseen" sequence}
\lineii{O flag}{"unseen" sequence}
\lineii{F flag}{"flagged" sequence}
\lineii{A flag}{"replied" sequence}
\end{tableii}

When an \class{mboxMessage} instance is created based upon a
\class{BabylMessage} instance, the following conversions take place:

\begin{tableii}{l|l}{textrm}
    {Resulting state}{\class{BabylMessage} state}
\lineii{R flag and O flag}{no "unseen" label}
\lineii{O flag}{"unseen" label}
\lineii{D flag}{"deleted" label}
\lineii{A flag}{"answered" label}
\end{tableii}

When a \class{Message} instance is created based upon an \class{MMDFMessage}
instance, the "From~" line is copied and all flags directly correspond:

\begin{tableii}{l|l}{textrm}
    {Resulting state}{\class{MMDFMessage} state}
\lineii{R flag}{R flag}
\lineii{O flag}{O flag}
\lineii{D flag}{D flag}
\lineii{F flag}{F flag}
\lineii{A flag}{A flag}
\end{tableii}

\subsubsection{\class{MHMessage}}
\label{mailbox-mhmessage}

\begin{classdesc}{MHMessage}{\optional{message}}
A message with MH-specific behaviors. Parameter \var{message} has the same
meaning as with the \class{Message} constructor.
\end{classdesc}

MH messages do not support marks or flags in the traditional sense, but they do
support sequences, which are logical groupings of arbitrary messages. Some mail
reading programs (although not the standard \program{mh} and \program{nmh}) use
sequences in much the same way flags are used with other formats, as follows:

\begin{tableii}{l|l}{textrm}{Sequence}{Explanation}
\lineii{unseen}{Not read, but previously detected by MUA}
\lineii{replied}{Replied to}
\lineii{flagged}{Marked as important}
\end{tableii}

\class{MHMessage} instances offer the following methods:

\begin{methoddesc}{get_sequences}{}
Return a list of the names of sequences that include this message.
\end{methoddesc}

\begin{methoddesc}{set_sequences}{sequences}
Set the list of sequences that include this message.
\end{methoddesc}

\begin{methoddesc}{add_sequence}{sequence}
Add \var{sequence} to the list of sequences that include this message.
\end{methoddesc}

\begin{methoddesc}{remove_sequence}{sequence}
Remove \var{sequence} from the list of sequences that include this message.
\end{methoddesc}

When an \class{MHMessage} instance is created based upon a
\class{MaildirMessage} instance, the following conversions take place:

\begin{tableii}{l|l}{textrm}
    {Resulting state}{\class{MaildirMessage} state}
\lineii{"unseen" sequence}{no S flag}
\lineii{"replied" sequence}{R flag}
\lineii{"flagged" sequence}{F flag}
\end{tableii}

When an \class{MHMessage} instance is created based upon an \class{mboxMessage}
or \class{MMDFMessage} instance, the \mailheader{Status} and
\mailheader{X-Status} headers are omitted and the following conversions take
place:

\begin{tableii}{l|l}{textrm}
    {Resulting state}{\class{mboxMessage} or \class{MMDFMessage} state}
\lineii{"unseen" sequence}{no R flag}
\lineii{"replied" sequence}{A flag}
\lineii{"flagged" sequence}{F flag}
\end{tableii}

When an \class{MHMessage} instance is created based upon a \class{BabylMessage}
instance, the following conversions take place:

\begin{tableii}{l|l}{textrm}
    {Resulting state}{\class{BabylMessage} state}
\lineii{"unseen" sequence}{"unseen" label}
\lineii{"replied" sequence}{"answered" label}
\end{tableii}

\subsubsection{\class{BabylMessage}}
\label{mailbox-babylmessage}

\begin{classdesc}{BabylMessage}{\optional{message}}
A message with Babyl-specific behaviors. Parameter \var{message} has the same
meaning as with the \class{Message} constructor.
\end{classdesc}

Certain message labels, called \dfn{attributes}, are defined by convention to
have special meanings. The attributes are as follows:

\begin{tableii}{l|l}{textrm}{Label}{Explanation}
\lineii{unseen}{Not read, but previously detected by MUA}
\lineii{deleted}{Marked for subsequent deletion}
\lineii{filed}{Copied to another file or mailbox}
\lineii{answered}{Replied to}
\lineii{forwarded}{Forwarded}
\lineii{edited}{Modified by the user}
\lineii{resent}{Resent}
\end{tableii}

By default, Rmail displays only
visible headers. The \class{BabylMessage} class, though, uses the original
headers because they are more complete. Visible headers may be accessed
explicitly if desired.

\class{BabylMessage} instances offer the following methods:

\begin{methoddesc}{get_labels}{}
Return a list of labels on the message.
\end{methoddesc}

\begin{methoddesc}{set_labels}{labels}
Set the list of labels on the message to \var{labels}.
\end{methoddesc}

\begin{methoddesc}{add_label}{label}
Add \var{label} to the list of labels on the message.
\end{methoddesc}

\begin{methoddesc}{remove_label}{label}
Remove \var{label} from the list of labels on the message.
\end{methoddesc}

\begin{methoddesc}{get_visible}{}
Return an \class{Message} instance whose headers are the message's visible
headers and whose body is empty.
\end{methoddesc}

\begin{methoddesc}{set_visible}{visible}
Set the message's visible headers to be the same as the headers in
\var{message}. Parameter \var{visible} should be a \class{Message} instance, an
\class{email.Message.Message} instance, a string, or a file-like object (which
should be open in text mode).
\end{methoddesc}

\begin{methoddesc}{update_visible}{}
When a \class{BabylMessage} instance's original headers are modified, the
visible headers are not automatically modified to correspond. This method
updates the visible headers as follows: each visible header with a
corresponding original header is set to the value of the original header, each
visible header without a corresponding original header is removed, and any of
\mailheader{Date}, \mailheader{From}, \mailheader{Reply-To}, \mailheader{To},
\mailheader{CC}, and \mailheader{Subject} that are present in the original
headers but not the visible headers are added to the visible headers.
\end{methoddesc}

When a \class{BabylMessage} instance is created based upon a
\class{MaildirMessage} instance, the following conversions take place:

\begin{tableii}{l|l}{textrm}
    {Resulting state}{\class{MaildirMessage} state}
\lineii{"unseen" label}{no S flag}
\lineii{"deleted" label}{T flag}
\lineii{"answered" label}{R flag}
\lineii{"forwarded" label}{P flag}
\end{tableii}

When a \class{BabylMessage} instance is created based upon an
\class{mboxMessage} or \class{MMDFMessage} instance, the \mailheader{Status}
and \mailheader{X-Status} headers are omitted and the following conversions
take place:

\begin{tableii}{l|l}{textrm}
    {Resulting state}{\class{mboxMessage} or \class{MMDFMessage} state}
\lineii{"unseen" label}{no R flag}
\lineii{"deleted" label}{D flag}
\lineii{"answered" label}{A flag}
\end{tableii}

When a \class{BabylMessage} instance is created based upon an \class{MHMessage}
instance, the following conversions take place:

\begin{tableii}{l|l}{textrm}
    {Resulting state}{\class{MHMessage} state}
\lineii{"unseen" label}{"unseen" sequence}
\lineii{"answered" label}{"replied" sequence}
\end{tableii}

\subsubsection{\class{MMDFMessage}}
\label{mailbox-mmdfmessage}

\begin{classdesc}{MMDFMessage}{\optional{message}}
A message with MMDF-specific behaviors. Parameter \var{message} has the same
meaning as with the \class{Message} constructor.
\end{classdesc}

As with message in an mbox mailbox, MMDF messages are stored with the sender's
address and the delivery date in an initial line beginning with "From ".
Likewise, flags that indicate the state of the message are typically stored in
\mailheader{Status} and \mailheader{X-Status} headers.

Conventional flags for MMDF messages are identical to those of mbox message and
are as follows:

\begin{tableiii}{l|l|l}{textrm}{Flag}{Meaning}{Explanation}
\lineiii{R}{Read}{Read}
\lineiii{O}{Old}{Previously detected by MUA}
\lineiii{D}{Deleted}{Marked for subsequent deletion}
\lineiii{F}{Flagged}{Marked as important}
\lineiii{A}{Answered}{Replied to}
\end{tableiii}

The "R" and "O" flags are stored in the \mailheader{Status} header, and the
"D", "F", and "A" flags are stored in the \mailheader{X-Status} header. The
flags and headers typically appear in the order mentioned.

\class{MMDFMessage} instances offer the following methods, which are identical
to those offered by \class{mboxMessage}:

\begin{methoddesc}{get_from}{}
Return a string representing the "From~" line that marks the start of the
message in an mbox mailbox. The leading "From~" and the trailing newline are
excluded.
\end{methoddesc}

\begin{methoddesc}{set_from}{from_\optional{, time_=None}}
Set the "From~" line to \var{from_}, which should be specified without a
leading "From~" or trailing newline. For convenience, \var{time_} may be
specified and will be formatted appropriately and appended to \var{from_}. If
\var{time_} is specified, it should be a \class{struct_time} instance, a tuple
suitable for passing to \method{time.strftime()}, or \code{True} (to use
\method{time.gmtime()}).
\end{methoddesc}

\begin{methoddesc}{get_flags}{}
Return a string specifying the flags that are currently set. If the message
complies with the conventional format, the result is the concatenation in the
following order of zero or one occurrence of each of \character{R},
\character{O}, \character{D}, \character{F}, and \character{A}.
\end{methoddesc}

\begin{methoddesc}{set_flags}{flags}
Set the flags specified by \var{flags} and unset all others. Parameter
\var{flags} should be the concatenation in any order of zero or more
occurrences of each of \character{R}, \character{O}, \character{D},
\character{F}, and \character{A}.
\end{methoddesc}

\begin{methoddesc}{add_flag}{flag}
Set the flag(s) specified by \var{flag} without changing other flags. To add
more than one flag at a time, \var{flag} may be a string of more than one
character.
\end{methoddesc}

\begin{methoddesc}{remove_flag}{flag}
Unset the flag(s) specified by \var{flag} without changing other flags. To
remove more than one flag at a time, \var{flag} maybe a string of more than one
character.
\end{methoddesc}

When an \class{MMDFMessage} instance is created based upon a
\class{MaildirMessage} instance, a "From~" line is generated based upon the
\class{MaildirMessage} instance's delivery date, and the following conversions
take place:

\begin{tableii}{l|l}{textrm}
    {Resulting state}{\class{MaildirMessage} state}
\lineii{R flag}{S flag}
\lineii{O flag}{"cur" subdirectory}
\lineii{D flag}{T flag}
\lineii{F flag}{F flag}
\lineii{A flag}{R flag}
\end{tableii}

When an \class{MMDFMessage} instance is created based upon an \class{MHMessage}
instance, the following conversions take place:

\begin{tableii}{l|l}{textrm}
    {Resulting state}{\class{MHMessage} state}
\lineii{R flag and O flag}{no "unseen" sequence}
\lineii{O flag}{"unseen" sequence}
\lineii{F flag}{"flagged" sequence}
\lineii{A flag}{"replied" sequence}
\end{tableii}

When an \class{MMDFMessage} instance is created based upon a
\class{BabylMessage} instance, the following conversions take place:

\begin{tableii}{l|l}{textrm}
    {Resulting state}{\class{BabylMessage} state}
\lineii{R flag and O flag}{no "unseen" label}
\lineii{O flag}{"unseen" label}
\lineii{D flag}{"deleted" label}
\lineii{A flag}{"answered" label}
\end{tableii}

When an \class{MMDFMessage} instance is created based upon an
\class{mboxMessage} instance, the "From~" line is copied and all flags directly
correspond:

\begin{tableii}{l|l}{textrm}
    {Resulting state}{\class{mboxMessage} state}
\lineii{R flag}{R flag}
\lineii{O flag}{O flag}
\lineii{D flag}{D flag}
\lineii{F flag}{F flag}
\lineii{A flag}{A flag}
\end{tableii}

\subsection{Exceptions}

The following exception classes are defined in the \module{mailbox} module:

\begin{classdesc}{Error}{}
The based class for all other module-specific exceptions.
\end{classdesc}

\begin{classdesc}{NoSuchMailboxError}{}
Raised when a mailbox is expected but is not found, such as when instantiating
a \class{Mailbox} subclass with a path that does not exist (and with the
\var{create} parameter set to \code{False}), or when opening a folder that does
not exist.
\end{classdesc}

\begin{classdesc}{NotEmptyErrorError}{}
Raised when a mailbox is not empty but is expected to be, such as when deleting
a folder that contains messages.
\end{classdesc}

\begin{classdesc}{ExternalClashError}{}
Raised when some mailbox-related condition beyond the control of the program
causes it to be unable to proceed, such as when failing to acquire a lock that
another program already holds a lock, or when a uniquely-generated file name
already exists.
\end{classdesc}

\begin{classdesc}{FormatError}{}
Raised when the data in a file cannot be parsed, such as when an \class{MH}
instance attempts to read a corrupted \file{.mh_sequences} file.
\end{classdesc}

\subsection{Deprecated classes and methods}
\label{mailbox-deprecated}

Older versions of the \module{mailbox} module do not support modification of
mailboxes, such as adding or removing message, and do not provide classes to
represent format-specific message properties. For backward compatibility, the
older mailbox classes are still available, but the newer classes should be used
in preference to them.

Older mailbox objects support only iteration and provide a single public
method:

\begin{methoddesc}[oldmailbox]{next}{}
Return the next message in the mailbox, created with the optional \var{factory}
argument passed into the mailbox object's constructor. By default this is an
\class{rfc822.Message} object (see the \refmodule{rfc822} module).  Depending
on the mailbox implementation the \var{fp} attribute of this object may be a
true file object or a class instance simulating a file object, taking care of
things like message boundaries if multiple mail messages are contained in a
single file, etc.  If no more messages are available, this method returns
\code{None}.
\end{methoddesc}

Most of the older mailbox classes have names that differ from the current
mailbox class names, except for \class{Maildir}. For this reason, the new
\class{Maildir} class defines a \method{next()} method and its constructor
differs slightly from those of the other new mailbox classes.

The older mailbox classes whose names are not the same as their newer
counterparts are as follows:

\begin{classdesc}{UnixMailbox}{fp\optional{, factory}}
Access to a classic \UNIX-style mailbox, where all messages are
contained in a single file and separated by \samp{From }
(a.k.a.\ \samp{From_}) lines.  The file object \var{fp} points to the
mailbox file.  The optional \var{factory} parameter is a callable that
should create new message objects.  \var{factory} is called with one
argument, \var{fp} by the \method{next()} method of the mailbox
object.  The default is the \class{rfc822.Message} class (see the
\refmodule{rfc822} module -- and the note below).

\begin{notice}
  For reasons of this module's internal implementation, you will
  probably want to open the \var{fp} object in binary mode.  This is
  especially important on Windows.
\end{notice}

For maximum portability, messages in a \UNIX-style mailbox are
separated by any line that begins exactly with the string \code{'From
'} (note the trailing space) if preceded by exactly two newlines.
Because of the wide-range of variations in practice, nothing else on
the \samp{From_} line should be considered.  However, the current
implementation doesn't check for the leading two newlines.  This is
usually fine for most applications.

The \class{UnixMailbox} class implements a more strict version of
\samp{From_} line checking, using a regular expression that usually correctly
matched \samp{From_} delimiters.  It considers delimiter line to be separated
by \samp{From \var{name} \var{time}} lines.  For maximum portability,
use the \class{PortableUnixMailbox} class instead.  This class is
identical to \class{UnixMailbox} except that individual messages are
separated by only \samp{From } lines.

For more information, see
\citetitle[http://home.netscape.com/eng/mozilla/2.0/relnotes/demo/content-length.html]{Configuring
Netscape Mail on \UNIX: Why the Content-Length Format is Bad}.
\end{classdesc}

\begin{classdesc}{PortableUnixMailbox}{fp\optional{, factory}}
A less-strict version of \class{UnixMailbox}, which considers only the
\samp{From } at the beginning of the line separating messages.  The
``\var{name} \var{time}'' portion of the From line is ignored, to
protect against some variations that are observed in practice.  This
works since lines in the message which begin with \code{'From '} are
quoted by mail handling software at delivery-time.
\end{classdesc}

\begin{classdesc}{MmdfMailbox}{fp\optional{, factory}}
Access an MMDF-style mailbox, where all messages are contained
in a single file and separated by lines consisting of 4 control-A
characters.  The file object \var{fp} points to the mailbox file.
Optional \var{factory} is as with the \class{UnixMailbox} class.
\end{classdesc}

\begin{classdesc}{MHMailbox}{dirname\optional{, factory}}
Access an MH mailbox, a directory with each message in a separate
file with a numeric name.
The name of the mailbox directory is passed in \var{dirname}.
\var{factory} is as with the \class{UnixMailbox} class.
\end{classdesc}

\begin{classdesc}{BabylMailbox}{fp\optional{, factory}}
Access a Babyl mailbox, which is similar to an MMDF mailbox.  In
Babyl format, each message has two sets of headers, the
\emph{original} headers and the \emph{visible} headers.  The original
headers appear before a line containing only \code{'*** EOOH ***'}
(End-Of-Original-Headers) and the visible headers appear after the
\code{EOOH} line.  Babyl-compliant mail readers will show you only the
visible headers, and \class{BabylMailbox} objects will return messages
containing only the visible headers.  You'll have to do your own
parsing of the mailbox file to get at the original headers.  Mail
messages start with the EOOH line and end with a line containing only
\code{'\e{}037\e{}014'}.  \var{factory} is as with the
\class{UnixMailbox} class.
\end{classdesc}

If you wish to use the older mailbox classes with the \module{email} module
rather than the deprecated \module{rfc822} module, you can do so as follows:

\begin{verbatim}
import email
import email.Errors
import mailbox

def msgfactory(fp):
    try:
        return email.message_from_file(fp)
    except email.Errors.MessageParseError:
        # Don't return None since that will
        # stop the mailbox iterator
        return ''

mbox = mailbox.UnixMailbox(fp, msgfactory)
\end{verbatim}

Alternatively, if you know your mailbox contains only well-formed MIME
messages, you can simplify this to:

\begin{verbatim}
import email
import mailbox

mbox = mailbox.UnixMailbox(fp, email.message_from_file)
\end{verbatim}

\subsection{Examples}
\label{mailbox-examples}

A simple example of printing the subjects of all messages in a mailbox that
seem interesting:

\begin{verbatim}
import mailbox
for message in mailbox.mbox('~/mbox'):
    subject = message['subject']       # Could possibly be None.
    if subject and 'python' in subject.lower():
        print subject
\end{verbatim}

To copy all mail from a Babyl mailbox to an MH mailbox, converting all
of the format-specific information that can be converted:

\begin{verbatim}
import mailbox
destination = mailbox.MH('~/Mail')
destination.lock()
for message in mailbox.Babyl('~/RMAIL'):
    destination.add(MHMessage(message))
destination.flush()
destination.unlock()
\end{verbatim}

This example sorts mail from several mailing lists into different
mailboxes, being careful to avoid mail corruption due to concurrent
modification by other programs, mail loss due to interruption of the
program, or premature termination due to malformed messages in the
mailbox:

\begin{verbatim}
import mailbox
import email.Errors

list_names = ('python-list', 'python-dev', 'python-bugs')

boxes = dict((name, mailbox.mbox('~/email/%s' % name)) for name in list_names)
inbox = mailbox.Maildir('~/Maildir', factory=None)

for key in inbox.iterkeys():
    try:
        message = inbox[key]
    except email.Errors.MessageParseError:
        continue                # The message is malformed. Just leave it.

    for name in list_names:
        list_id = message['list-id']
        if list_id and name in list_id:
            # Get mailbox to use
            box = boxes[name]

            # Write copy to disk before removing original.
            # If there's a crash, you might duplicate a message, but
            # that's better than losing a message completely.
            box.lock()
            box.add(message)
            box.flush()         
            box.unlock()

            # Remove original message
            inbox.lock()
            inbox.discard(key)
            inbox.flush()
            inbox.unlock()
            break               # Found destination, so stop looking.

for box in boxes.itervalues():
    box.close()
\end{verbatim}