Skip to content


Subversion checkout URL

You can clone with
Download ZIP
Fetching contributors…
Cannot retrieve contributors at this time
791 lines (620 sloc) 26 KB
\input texinfo.tex @c -*-texinfo-*-
@c %**start of header
@settitle mu-guile user manual
@documentencoding utf-8
@c %**end of header
@dircategory The Algorithmic Language Scheme
* mu-guile manual: (mu-guile). Guile bindings for the @t{mu} e-mail indexer/searcher.
@end direntry
Copyright @copyright{} 2012 Dirk-Jan C. Binnema
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License, Version 1.3 or
any later version published by the Free Software Foundation; with no
Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts. A
copy of the license is included in the section entitled ``GNU Free
Documentation License.''
@end quotation
@end copying
@node Top
@top mu4e Manual
Welcome to @t{mu-guile}!
@t{mu-guile} is a binding of the @t{mu} email search engine for the @t{guile}
programming language. That means that you can write simple (and not so simple)
programs to data-mine your e-mail database.
* Introduction::
* Getting started::
* Initializing mu-guile::
* Messages::
* Contacts::
* Attachments and other parts::
* Statistics::
* Plotting data::
* GNU Free Documentation License:: The license of this manual.
@end menu
@node Introduction
@chapter Introduction
@t{mu} is a program for indexing / searching e-mails stored in Maildirs.
@t{guile} is the @emph{GNU Ubiquitous Intelligent Language for Extensions} - a
version of the @emph{Scheme} programming language and the official GNU
extension language.
@t{mu-guile} connects @t{mu} and @t{guile}, and allows you to easily write
programs for your specific needs.
@node Getting started
@chapter Getting started
* Installation::
* First steps::
@end menu
This chapter walks you through the installation and some basic steps to ensure
things work correctly.
@node Installation
@section Installation
@t{mu-guile} is part of @t{mu} - by installing the latter, the former will be
installed as well, provided that you have @t{guile} version 2.0 installed.
At the time of writing, there are no distribution packages for @t{mu-guile},
so we are assuming installation from source packages.
Installation follows the normal sequence of:
$ tar xvfz mu-<version>.tar.gz # use the specific version
$ cd mu-<version>
@end example
The output of @t{./configure} should end with a little text describing the
detected versions of various libraries @t{mu} depends on. In particular, it
should mention the @t{guile} version, e.g.
Guile version :
@end example
If you don't see any line referring to @t{guile}, please install it, and run
@t{configure} again. Note once more, @t{mu-guile} requires @t{guile} version
After a succesfull @t{./configure}, we can make and install the package:
$ make && sudo make install
@end example
After this, @t{mu} and @t{mu-guile} should be installed. Note that the above
instructions will normally install things under @t{/usr/local}; you may need
to update @t{guile}'s @t{%load-path} to find it there.
You can check the current load-path with the following:
guile -c '(display %load-path)(newline)'
@end example
If necessary, you can add the @t{%load-path} by adding something like the
following to your @file{~/.guile}:
(set! %load-path (cons "/usr/local/share/guile/site/2.0" %load-path))
@end lisp
Or, alternatively, you can set @t{GUILE_LOAD_PATH}:
export GUILE_LOAD_PATH="/usr/local/share/guile/site/2.0"
@end example
Note, in both cases the directory should be the directory that contains the
installed @t{mu.scm}; so if you installed @t{mu} under a different prefix, you
must change the load-path accordingly.
After this, you should be ready to go.
@node First steps
@section First steps
Assuming @t{mu-guile} has been installed correctly (@ref{Installation}), and
also assuming that you have already indexed your e-mail messages (if
necessary, see the @t{mu-index} man-page), we are ready to start @t{mu-guile};
a session may look something like this:
$ guile
GNU Guile
Copyright (C) 1995-2011 Free Software Foundation, Inc.
Guile comes with ABSOLUTELY NO WARRANTY; for details type `,show w'.
This program is free software, and you are welcome to redistribute it
under certain conditions; type `,show c' for details.
Enter `,help' for help.
@end verbatim
Now, we need to load some of the @t{mu-guile} modules:
scheme@(guile-user)> (use-modules (mu) (mu message))
@end verbatim
This will load the basic modules for dealing with messages. After we have
loaded the modules, we need to initialize the @t{mu-guile} system:
scheme@(guile-user)> (mu:initialize)
@end verbatim
When this is done, we can start querying the database. We discuss various
methods and functions later in this manual, but just to give an example, let's
get a list of the subjects of all messages that mention @emph{hello}:
scheme@(guile-user)> (for-each
(format #t "Subject: ~a\n" (mu:subject msg)))
(mu:message-list "hello"))
@end verbatim
Note, the multi-lines in the example are only for readability; since it can be
a bit uncomfortable to type long sequences at the 'REPL' (the Guile
command-line), we recommend using a tool like
@node Initializing mu-guile
@chapter Initializing mu-guile
It is of course possible to write separate programs with @t{mu-guile}, but for
now we'll do things @emph{interactively}, i.e., from the Guile-prompt
We start our @t{mu-guile} session by starting @t{guile}:
$ guile
GNU Guile
Copyright (C) 1995-2011 Free Software Foundation, Inc.
Guile comes with ABSOLUTELY NO WARRANTY; for details type `,show w'.
This program is free software, and you are welcome to redistribute it
under certain conditions; type `,show c' for details.
Enter `,help' for help.
@end verbatim
Now, the first thing we need to do is load the @t{mu-guile} modules;
currently, there are six available:
@item @code{mu} - initialization, functions to get messages, contacts
@item @code{mu message} - functions to deal with messages
@item @code{mu contact} - functions to deal with contacts
@item @code{mu part} - functions to deal with message-parts / attachments
@item @code{mu stats} - some functions for doing statistics on your messages
@item @code{mu plot} - functions to draw graphs from the statistics (requires @t{gnuplot}
@end itemize
Let's simply load all of them:
scheme@(guile-user)> (use-modules (mu) (mu message) (mu contact) (mu part)
(mu stats) (mu plot))
@end verbatim
Assuming you have installed everything correctly, the first time you do this,
@t{guile} will probably respond by showing some message about compiling the
modules, and then return to you with another prompt.
Before we can do anything with @t{mu guile}, we need to initialize the
system. This goes like this:
scheme@(guile-user)> (mu:initialize)
@end verbatim
Which will use the default location of @file{~/.mu}. Or, instead, if you keep
your @t{mu} data in a non-standard place:
scheme@(guile-user)> (mu:initialize "/path/to/my/mu/")
@end verbatim
If all worked up until here, we're ready to go with @t{mu-guile} - hurray! In
the next chapters we'll walk through all the modules.
@node Messages
@chapter Messages
In this chapter, we discuss how to find messages and how to do various things
with them.
* Finding messages::
* Message methods::
* Example - the longest subject::
@end menu
@node Finding messages
@section Finding messages
Now we are ready to retrieve some messages from the system. There are two main
functions to do this:
@item @code{(mu:message-list [<search-expression>])}
@item @code{(mu:for-each-message <function> [<search-expression>])}
@end itemize
The first function, @code{mu:message-list} returns a list of all messages
matching @t{<search-expression>}; if you leave @t{<search-expression>} out, it
returns @emph{all} messages.
For example, to get all messages with @emph{coffee} in the subject line, you
could do:
scheme@(guile-user)> (mu:message-list "subject:coffee")
$1 = (#<<mu:message> 9040640> #<<mu:message> 9040630>
#<<mu:message> 9040570>)
@end verbatim
So, since apparently we have three messages matching @t{subject:coffee}, we
get a list of three @t{<mu:message>} objects. Let's just use the
@code{mu:subject} function ('method') provided by @t{<mu:message>} objects to
retrieve the subject-field (more about methods in the next section).
For your convenience, @t{guile} has saved the result of our last query in a
variable called @t{$1}, so to get the subject of the first message in the
list, we can do:
scheme@(guile-user)> (mu:subject (car $1))
$2 = "Re: best coffee ever!"
@end verbatim
The second function we mentioned, @code{mu:for-each-message}, executes some
function for each message matched by the search expression (or @emph{all}
messages if the search expression is omitted).
scheme@(guile-user)> (mu:for-each-message
(display (subject msg))
Re: best coffee ever!
best coffee ever!
Coffee beans
@end verbatim
Using @code{mu:message-list} and/or
@code{mu:for-each-message}@footnote{Implementation node:
@code{mu:message-list} is implemented in terms of @code{mu:for-each-message},
not the other way around. Due to the way @t{mu} works,
@code{mu:for-each-message} is rather more efficient than a combination of
@code{for-each} and @code{mu:message-list}} and a couple of @t{<mu:message>}
methods, together with what Guile/Scheme provides, should allow for many
interesting programs.
@node Message methods
@section Message methods
Now that we've seen how to retrieve lists of message objects
(@code{<mu:message>}), let's see what we can do with such an object.
@code{<mu:message>} defines the following methods that all take a single
@code{<mu:message>} object as a parameter. We won't go into the exact meanings
for all of these functions here - for the details about various flags /
properties, please refer to the @t{mu-find} man-page.
@item @code{(mu:bcc msg)}: the @t{Bcc} field of the message, or @t{#f} if there is none
@item @code{(mu:body-html msg)}: : the html body of the message, or @t{#f} if there is none
@item @code{(mu:body-txt msg)}: the plain-text body of the message, or @t{#f} if there is none
@item @code{(mu:cc msg)}: the @t{Bcc} field of the message, or @t{#f} if there is none
@item @code{(mu:date msg)}: the @t{Date} field of the message, or 0 if there is none
@item @code{(mu:flags msg)}: list of message-flags for this message
@item @code{(mu:from msg)}: the @t{From} field of the message, or @t{#f} if there is none
@item @code{(mu:maildir msg)}: the maildir this message lives in, or @t{#f} if there is none
@item @code{(mu:message-id msg)}: the @t{Message-Id} field of the message, or @t{#f} if there is none
@item @code{(mu:path msg)}: the file system path for this message
@item @code{(mu:priority msg)}: the priority of this message (either @t{mu:prio:low}, @t{mu:prio:normal} or @t{mu:prio:high}
@item @code{(mu:references msg)}: the list of messages (message-ids) this message
refers to in(mu: the @t{References:} header
@item @code{(mu:size msg)}: size of the message in bytes
@item @code{(mu:subject msg)}: the @t{Subject} field of the message, or @t{#f} if there is none.
@item @code{(mu:tags msg)}: list of tags for this message
@item @code{(mu:to msg)}: the sender of the message, or @t{#f} if there is none.
@end itemize
With these methods, we can query messages for their properties; for example:
scheme@(guile-user)> (define msg (car (mu:message-list "snow")))
scheme@(guile-user)> (mu:subject msg)
$1 = "Re: Running in the snow is beautiful"
scheme@(guile-user)> (mu:flags msg)
$2 = (mu:flag:replied mu:flag:seen)
scheme@(guile-user)> (strftime "%F" (localtime (mu:date msg)))
$3 = "2011-01-15"
@end verbatim
There are a couple more methods:
@item @code{(mu:header msg "<header-name>")} returns an arbitrary message
header (or @t{#f} if not found) -- e.g. @code{(header msg "User-Agent")}
@item If you include the @t{mu contact} module, the @code{(mu:contacts
msg [contact-type])} method (to get a list of contacts) is
added. @xref{Contacts}.
@item If you include the @t{mu part} module, the @code{((mu:parts msg)} and
@code{(mu:attachments msg)} methods are added. @xref{Attachments and other parts}.
@end itemize
@node Example - the longest subject
@section Example - the longest subject
Now, let's write a little example -- let's find out what is the @emph{longest
subject} of any e-mail messages we received in the year 2011. You can try
this if you put the following in a separate file, make it executable, and run
it like any program.
exec guile -s $0 $@
(use-modules (mu) (mu message))
(use-modules (srfi srfi-1))
;; note: (subject msg) => #f if there is no subject
(define list-of-subjects
(map (lambda (msg)
(or (mu:subject msg) "")) (mu:message-list "date:2011..2011")))
;; see the mu-find manpage for the date syntax
(define longest-subject
(fold (lambda (subj1 subj2)
(if (> (string-length subj1) (string-length subj2))
subj1 subj2))
"" list-of-subjects))
(format #t "Longest subject: ~s" longest-subject)
@end verbatim
There are many other ways to solve the same problem, for example by using an
iterative approach with @code{mu:for-each-message}, but it should show how one
can easily write little programs to answer specific questions about an e-mail
@node Contacts
@chapter Contacts
We can retrieve the sender and recipients of an e-mail message using methods
like @code{mu:from}, @code{mu:to} etc.; @xref{Message methods}. These
functions return the list of recipients as a single string; however, often it
is more useful to deal with recipients as separate objects.
@t{mu-guile} offers some functionality for this in the @code{(mu contact)}
module. Also, it adds some contact-related methods for @code{<mu:message>}
* Contact functions and objects::
* All contacts::
* Utility functions::
* Example - mutt export::
@end menu
@node Contact functions and objects
@section Contact functions and objects
(use-modules (mu contact))
@end verbatim
After loading the @code{(mu contact)}, message objects (@pxref{Messages}) gain
the the @t{contacts}-methods:
@code{(mu:contacts <message-object> [<contact-type>])}
The @t{<contact-type>} is a symbol, one of @code{mu:to}, @code{mu:from},
@code{mu:cc} or @code{mu:bcc}; this will then get the contact objects for the
contacts of the corresponding type. If you leave out the contact-type (or
specify @t{#t} for it, you will get a list of @emph{all} contact objects for
the message.
A contact object (@code{<mu:contact>}) has two methods:
@item @code{mu:name} returns the name of the contact, or #f if there is none
@item @code{mu:email} returns the e-mail address of the contact, or #f if there is none
@end itemize
Let's get a list of all names and e-mail addresses in the 'To:' field, of
messages matching 'book':
(use-modules (mu) (mu message) (mu contact))
(lambda (msg)
(lambda (contact)
(format #t "~a => ~a\n"
(or (mu:email contact) "") (or (mu:name contact) "no-name")))
(mu:contacts msg mu:field:to)))
@end lisp
This shows what the methods do, but for many uses, it would be more useful to
have each of the contacts only show up @emph{once} - for that, please refer to
@xref{All contacts}.
@node All contacts
@section All contacts
Sometimes it may also be useful to look at @emph{all} the different contacts
in the @t{mu} database -- that is, all the different contacts. This is useful,
for example, when exporting contacts to some external format that can then be
important in an e-mail program.
To enable this, there is the function @code{mu:for-each-contact}, defined as
@code{(mu:for-each-contact function [search-expression])}.
This will aggregate the unique contacts from @emph{all} messages matching
@t{<search-expression>} (when it is left empty, it will match all messages in
the database), and execute @t{function} for each of these contacts.
The @t{function} receives an object of the type @t{<mu:contact-with-stats>},
which is a @emph{subclass} of the @t{<mu:contact>} class discussed in
@xref{Contact functions and objects}. @t{<mu:contact-with-stats>} objects
expose the following additional methods:
@item @code{(mu:frequency <contact>)}: returns the @emph{number of times} this contact occured in
one of the address fields
@item @code{(mu:last-seen <contact>)}: returns the @emph{most recent time} the contact was
seen in one of the address fields, as a @t{time_t} value
@end itemize
The method assumes an e-mail address is unique for a certain contact; if a
certain e-mail address occurs with different names, it uses the most recent
non-empty name.
@node Utility functions
@section Utility functions
To make dealing with contacts even easier, there are a number of utility
functions that can save you a bit of typing.
For converting contacts to some textual form, there is @code{(mu:contact->string
<mu:contact> format)}, which takes a contact and returns a text string with
the given format. Currently supported formats are @t{"org-contact}, @t{"mutt-alias"},
@t{"mutt-ab"}, @t{"wanderlust"} and @t{"plain"}.
@node Example - mutt export
@section Example - mutt export
Let's see how we could export the addresses in the @t{mu} database to the
addressbook format of the venerable
@t{mutt}@footnote{@url{}} e-mail client.
The addressbook format that @t{mutt} uses is a sequence of lines that look
something like:
alias <nick> [<name>] "<" <email> ">"
@end verbatim
Anyway, there is the function @code{(mu:contact->string <mu:contact> format)}
that we can use to do the conversion.
We may want to focus on people with whom we have frequent correspondence; so
we may want to limit ourselves to people we have seen at least 10 times in the
last year.
It is a bit hard to @emph{guess} the nick name for e-mail contacts, but
@code{mu:contact->string} tries something based on the name. You can always
adjust them later by hand, obviously.
exec guile -s $0 $@
(use-modules (mu) (mu message) (mu contact))
;; Get a list of contacts that were seen at least 20 times since 2010
(define (selected-contacts)
(let ((addrs '())
(start (car (mktime (car (strptime "%F" "2010-01-01")))))
(minfreq 20))
(lambda (contact)
(if (and (mu:email contact)
(>= (mu:frequency contact) minfreq)
(>= (mu:last-seen contact) start))
(set! addrs (cons contact addrs)))))
(lambda (contact)
(format #t "~a\n" (mu:contact->string contact "mutt-alias")))
@end lisp
This simple program could be improved in many ways; this is left as an
excercise to the reader.
@node Attachments and other parts
@chapter Attachments and other parts
To deal with @emph{attachments}, or, more in general @emph{MIME-parts}, there
is the @t{mu part} module.
* Parts and their methods::
* Attachment example::
@end menu
@node Parts and their methods
@section Parts and their methods
The module defines the @code{<mu-part>} class, and adds two methods to
@code{<mu:message>} objects:
@item @code{(mu:parts msg)} - returns a list @code{<mu-part>} objects, one for
each MIME-parts in the message.
@item @code{(mu:attachments msg)} - like @code{parts}, but only list those MIME-parts
that look like proper attachments.
@end itemize
A @code{<mu:part>} object exposes a few methods to get information about the
@item @code{(mu:name <part>)} - returns the file name of the mime-part, or @code{#f} if
there is none.
@item @code{(mu:mime-type <part>)} - returns the mime-type of the mime-part, or @code{#f}
if there is none.
@item @code{(mu:size <part>)} - returns the size in bytes of the mime-part
@end itemize
Then, we may want to save the part to a file; this can be done using either:
@item @code{(mu:save part <part>)} - save a part to a temporary file, return the file
name@footnote{the temporary filename is a predictable function of (user-id,
msg-path, part-index)}
@item @code{(mu:save-as <part> <path>)} - save part to file at path
@end itemize
@node Attachment example
@section Attachment example
Let's look at some small example. Let's get a list of the biggest attachments
in messages about Luxemburg:
exec guile -s $0 $@
(use-modules (mu) (mu message) (mu part))
(define (all-attachments expr)
"Return a list of (name . size) for all attachments in messages
matching EXPR."
(let ((pairs '()))
(lambda (msg)
(lambda (att) ;; add (filename . size) to the list
(set! pairs (cons (cons (mu:name att) (or (mu:size att) 0)) pairs)))
(mu:attachments msg)))
(lambda (att)
(format #t "~a: ~,1fKb\n"
(car att) (exact->inexact (/ (cdr att) 1024))))
(sort (all-attachments "Luxemburg")
(lambda (att1 att2)
(< (cdr att1) (cdr att2)))))
@end lisp
As an exercise for the reader, you might want to re-rewrite the
@code{all-attachments} in terms of @code{mu:message-list}, which would
probably be a bit more elegant.
@node Statistics
@chapter Statistics
@t{mu-guile} offers some convenience functions to determine various statistics
about the messages in the database.
@code{(mu:tabulate-messages <function> [<search-expr>])} applies
@t{<function>} to each message matching @t{<search-expr>} (leave empty to
match @emph{all} messages), and returns a associative list (a list of pairs)
with each of the different results of @t{<function>} and their frequencies.
This can best be demonstrated with a little example. Suppose we want to know
how many messages we receive per weekday:
exec guile -s $0 $@
(use-modules (mu) (mu message) (mu stats) (mu plot))
;; create a list like (("Sun" . 13) ("Mon" . 23) ...)
(define weekday-table
(lambda (msg)
(tm:wday (localtime (mu:date msg)))))
(lambda (a b) (< (car a) (car b))))))
(lambda (elm)
(format #t "~a: ~a\n" (car elm) (cdr elm)))
@end lisp
The function @code{weekday-table} uses @code{mu:tabulate-message} to get the
frequencies per hour -- this returns a list of pairs:
((5 . 2339) (0 . 2278) (4 . 2800) (2 . 3184) (6 . 1856) (3 . 2833) (1 . 2993))
@end verbatim
We sort these pairs by the day number, and then apply
@code{mu:weekday-numbers->names}, which takes the list, and returns a list
where the day numbers are replace by there abbreviated name (in the current
locale). Note, there is also @code{mu:month-numbers->names}.
The script then outputs these numbers in the following form:
Sun: 2278
Mon: 2993
Tue: 3184
Wed: 2833
Thu: 2800
Fri: 2339
Sat: 1856
@end verbatim
Clearly, Saturday is a slow day for e-mail...
@node Plotting data
@chapter Plotting data
You can plot the results in the format produced by @code{mu:tabulate} with the
@t{(mu plot)} module, an experimental module that requires the
@t{gnuplot}@footnote{@url{}} program to be installed
on your system.
The @code{mu:plot} function takes the following arguments:
@code{(mu:plot <data> <title> <x-label> <y-label> [<want-ascii>])}
Here, @code{<data>} is a table of data in the format that @code{mu:tabulate}
produces. @code{<title>}, @code{<x-label>} and @code{<y-lablel>} are,
respectively, the title of the graph, and the labels for X- and
Y-axis. Finally, if you pass @t{#t} for the final @code{<want-ascii>}
parameter, a plain-text rendering of the graph will be produced; otherwise, a
graphical window will be shown.
An example should clarify how this works in practice; let's plot the number of
message per hour:
exec guile -s $0 $@
(use-modules (mu) (mu message) (mu contact) (mu stats) (mu plot))
(define (mail-per-hour-table)
(lambda (msg)
(tm:hour (localtime (mu:date msg)))))
(lambda (x y) (< (car x) (car y)))))
(mu:plot (mail-per-hour-table) "Mail per hour" "Hour" "Frequency" #t)
@end lisp
Mail per hour
1200 ++--+--+--+--+-+--+--+--+--+-+--+--+--+-+--+--+--+--+-+--+--+--+--++
|+ + + + + + + "/tmp/fileHz7D2u" using 2:xticlabels(1) ********
1100 ++ *** +*
**** * * *
1000 *+ * **** * +*
* * ****** **** * ** * *
900 *+ * * ** **** * **** ** * +*
* * * ** * * ********* * ** ** * *
800 *+ * **** ** * * * * ** * * ** ** * +*
700 *+ *** **** * ** * * * * ** **** * ** ** * +*
* * * **** * * ** * * * * ** * **** ** ** * *
600 *+ * **** * * * * ** * * * * ** * * * ** ** * +*
* * ** * * * * * ** * * * * ** * * * ** ** * *
500 *+ * ** * * * * * ** * * * * ** * * * ** ** * +*
* * ** **** *** * * * ** * * * * ** * * * ** ** * *
400 *+ * ** ** **** * * * * * ** * * * * ** * * * ** ** * +*
*+ *+**+**+* +*******+* +* +*+ *+**+* +*+ *+ *+**+* +*+ *+**+**+* +*
300 ********************************************************************
0 1 2 3 4 5 6 7 8 910 11 12 1314 15 16 17 1819 20 21 22 23
@end verbatim
@node GNU Free Documentation License
@appendix GNU Free Documentation License
@include fdl.texi
Jump to Line
Something went wrong with that request. Please try again.