emacs/doc/misc/sc.texi

\input texinfo  @comment -*-texinfo-*-
@comment 3.48
@comment %**start of header (This is for running Texinfo on a region.)
@setfilename ../../info/sc.info
@settitle Supercite User's Manual
@include docstyle.texi
@iftex
@finalout
@end iftex

@c @setchapternewpage odd               % For book style double sided manual.
@comment %**end of header (This is for running Texinfo on a region.)

@copying
This document describes Supercite, an Emacs package for citing and
attributing replies to mail and news messages.

Copyright @copyright{} 1993, 2001--2023 Free Software Foundation, Inc.

@quotation
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, with the Front-Cover Texts being ``A GNU Manual'',
and with the Back-Cover Texts as in (a) below.  A copy of the license
is included in the section entitled ``GNU Free Documentation License''.

(a) The FSF's Back-Cover Text is: ``You have the freedom to copy and
modify this GNU manual.''
@end quotation
@end copying

@c      @smallbook

@dircategory Emacs network features
@direntry
* SC: (sc).                     Supercite lets you cite parts of messages
                                  you're replying to, in flexible ways.
@end direntry

@titlepage
@title Supercite User's Manual
@subtitle cite and attribute mail and
@subtitle news, in flexible ways

@page
@vskip 0pt plus 1filll
@insertcopying
@end titlepage

@summarycontents
@contents

@ifnottex
@node Top
@top Supercite

@insertcopying

The manual is divided
into the following chapters.

@menu
* Introduction::
* Citations::
* Information Keys and the Info Alist::
* Reference Headers::
* Getting Connected::
* Replying and Yanking::
* Selecting an Attribution::
* Configuring the Citation Engine::
* Post-yank Formatting Commands::
* Hints to MUA Authors::
* Thanks and History::

* GNU Free Documentation License::
* Concept Index::
* Command Index::
* Key Index::
* Variable Index::
@end menu
@end ifnottex


@node  Introduction
@chapter Introduction

@cindex MUA
@cindex NUA
Supercite is a GNU Emacs package written entirely in Emacs Lisp.  It
interfaces to most of the commonly used Emacs mail user agents
(@dfn{MUAs}) and news user agents (@dfn{NUAs}), and provides
sophisticated facilities for the citing and attributing of message
replies.  Supercite has a very specific and limited role in the
process of composing replies to both USENET network news and
electronic mail.

The preferred way to spell Supercite is with a capital @samp{S},
lowercase @samp{upercite}.

@menu
* Usage Overview::
* What Supercite Does Not Do::
* What Supercite Does::
@end menu

@c FIXME: move it above the menu?  --xfq
Supercite is only useful in conjunction with MUAs and NUAs such as VM,
Gnus, RMAIL, MH-E, etc.  Supercite is typically called by the MUA after a
reply buffer has been setup.  Thereafter, Supercite's many commands and
formatting styles are available in that reply buffer until the reply is
sent.  Supercite is re-initialized in each new reply buffer.


@node Usage Overview
@section Usage Overview
@kindex r
@kindex f
@kindex C-c C-y
@cindex yank
@cindex cite, citing
@cindex attribute, attributing

Typical usage is as follows.  You want to reply or followup to a
message in your MUA@.  You will probably hit @kbd{r} (i.e., ``reply'')
or @kbd{f} (i.e., ``forward'') to begin composing the reply.  In
response, the MUA will create a reply buffer and initialize the
outgoing mail headers appropriately.  The body of the reply will
usually be empty at this point.  You now decide that you would like to
include part of the original message in your reply.  To do this, you
@dfn{yank} the original message into the reply buffer, typically with
a key stroke such as @kbd{C-c C-y}.  This sequence will invoke an
MUA-specific function which fills the body of the reply with the
original message and then @dfn{attributes} this text to its author.
This is called @dfn{citing} and its effect is to prefix every line
from the original message with a special text tag.  Most MUAs provide
some default style of citing; by using Supercite you gain a wider
flexibility in the look and style of citations.  Supercite's only job
is to cite the original message.

@node  What Supercite Does Not Do
@section What Supercite Doesn't Do

Because of this clear division of labor, there are useful features which
are the sole responsibility of the MUA, even though it might seem that
Supercite should provide them.  For example, many people would like to
be able to yank (and cite) only a portion of the original message.
Since Supercite only modifies the text it finds in the reply buffer as
set up by the MUA, it is the MUA's responsibility to do partial yanking.
@xref{Reply Buffer Initialization}.

@vindex mail-header-separator
Another potentially useful thing would be for Supercite to set up the
outgoing mail headers with information it gleans from the reply buffer.
But by previously agreed upon convention, any text above the
@code{mail-header-separator} which separates mail headers from message
bodies cannot be modified by Supercite.  Supercite, in fact, doesn't
know anything about the meaning of these headers, and never ventures
outside the designated region. @xref{Hints to MUA Authors}, for more
details.

@node  What Supercite Does
@section What Supercite Does
@findex sc-cite-original

Supercite is invoked for the first time on a reply buffer via your MUA's
reply or forward command.  This command will actually perform citations
by calling a hook variable to which Supercite's top-level function
@code{sc-cite-original} has been added.  When @code{sc-cite-original} is
executed, the original message must be set up in a very specific way,
but this is handled automatically by the MUA@.  @xref{Hints to MUA
Authors}.

@cindex info alist
The first thing Supercite does, via @code{sc-cite-original}, is to parse
through the original message's mail headers.  It saves this data in an
@dfn{information association list}, or @dfn{info alist}.  The information
in this list is used in a number of places throughout Supercite.
@xref{Information Keys and the Info Alist}.

@cindex nuking mail headers
@cindex reference header
After the mail header info is extracted, the headers are optionally
removed (@dfn{nuked}) from the reply.  Supercite then writes a
@dfn{reference header} into the buffer.  This reference header is a
string carrying details about the citation it is about to perform.

@cindex modeline
Next, Supercite visits each line in the reply, transforming the line
according to a customizable ``script''.  Lines which were not previously
cited in the original message are given a citation, while already cited
lines remain untouched, or are coerced to your preferred style.
Finally, Supercite installs a keymap into the reply buffer so that you
have access to Supercite's post-yank formatting and reciting commands as
you subsequently edit your reply.  You can tell that Supercite has been
installed into the reply buffer because that buffer's modeline will
display the minor mode string @samp{SC}.

@cindex filladapt
@cindex gin-mode
@vindex fill-prefix
@findex fill-paragraph
When the original message is cited by @code{sc-cite-original}, it will
(optionally) be filled by Supercite.  However, if you manually edit the
cited text and want to re-fill it, you must use an add-on package such
as @cite{filladapt} or @cite{gin-mode}.  These packages can recognize
Supercited text and will fill them appropriately.  Emacs's built-in
filling routines, e.g., @code{fill-paragraph}, do not recognize cited
text and will not re-fill them properly because it cannot guess the
@code{fill-prefix} being used.
@xref{Post-yank Formatting Commands}, for details.

As mentioned above, Supercite provides commands to recite or uncite
regions of text in the reply buffer, and commands to perform other
beautifications on the cited original text, maintaining consistent and
informative citations throughout.  Supercite tries to be as configurable
as possible to allow for a wide range of personalized citation styles,
but it is also immediately useful with the default configuration, once
it has been properly connected to your MUA@.  @xref{Getting Connected},
for more details.

@node  Citations
@chapter Citations
@cindex nested citations
@cindex citation

A @dfn{citation} is the acknowledgment of the original author of a mail
message in the body of the reply.  There are two basic citation styles
which Supercite supports.  The first, called @dfn{nested citations} is
an anonymous form of citation; in other words, an indication is made
that the cited line was written by someone @emph{other} that the current
message author (i.e., other than you, the person composing the reply),
but no reference is made as to the identity of the original author.
This style should look familiar since its use on the net is widespread.
Here's an example of what a message buffer would look like using nested
citations after multiple replies:

@example
>> John originally wrote this
>> and this as well
> Jane said that John didn't know
> what he was talking about
And that's what I think too.
@end example

@menu
* Citation Elements::
* Recognizing Citations::
@end menu

Note that multiple inclusions of the original messages result in a
nesting of the @samp{>} characters.  This can sometimes be quite
confusing when many levels of citations are included since it may be
difficult or impossible to figure out who actually participated in the
thread, and multiple nesting of @samp{>} characters can sometimes
make the message very difficult for the eye to scan.

@cindex non-nested citations
In @dfn{non-nested citations}, each cited line begins with an
informative string attributing that line to the original author.  Only
the first level of attribution will be shown; subsequent citations
don't nest the citation strings.  The above dialog might look like
this when non-nested citations are used:

@example
John> John originally wrote this
John> and this as well
Jane> Jane said that John didn't know
Jane> what he was talking about
And that's what I think too.
@end example

Notice here that my inclusion of Jane's inclusion of John's original
message did not result in a line cited with @samp{Jane>John>}.

@vindex sc-nested-citation-p
@vindex nested-citation-p @r{(sc-)}
Supercite supports both styles of citation, and the variable
@code{sc-nested-citation-p} controls which style it will use when
citing previously uncited text.  When this variable is @code{nil} (the
default), non-nested citations are used.  When non-@code{nil}, nested
citations are used.


@node  Citation Elements
@section Citation Elements
@cindex citation string

@dfn{Citation strings} are composed of one or more elements.
Non-nested citations are composed of four elements, three of which are
directly user definable.  The elements are concatenated together, in
this order:

@cindex citation leader
@vindex citation-leader @r{(sc-)}
@vindex sc-citation-leader
@enumerate
@item
The @dfn{citation leader}.  The citation leader is contained in the
variable @code{sc-citation-leader}, and has the default value of a
string containing four spaces.

@cindex attribution string
@item
The @dfn{attribution string}.  This element is supplied automatically by
Supercite, based on your preferences and the original message's mail
headers, though you may be asked to confirm Supercite's choice.
@xref{Selecting an Attribution}, for more details.

@cindex citation delimiter
@vindex sc-citation-delimiter
@vindex citation-delimiter @r{(sc-)}
@item
The @dfn{citation delimiter}.  This string, contained in the variable
@code{sc-citation-delimiter} visually separates the citation from the
text of the line.  This variable has a default value of @code{">"} and
for best results, the string should consist of only a single character.

@cindex citation separator
@vindex citation-separator @r{(sc-)}
@vindex sc-citation-separator
@item
The @dfn{citation separator}.  The citation separator is contained in
the variable @code{sc-citation-separator}, and has the default value of
a string containing a single space.
@end enumerate

For example, suppose you were using the default values for the above
variables, and Supercite provided the attribution string @samp{Jane}.
In this case, the composed, non-nested citation string used might be
something like
@code{@asis{"    Jane> "}}.
This citation string will be inserted in front of
every line in the original message that is not already cited.

Nested citations, being simpler than non-nested citations, are composed
of the same elements, sans the attribution string.  Supercite is smart
enough to not put additional spaces between citation delimiters for
multi-level nested citations.

@node  Recognizing Citations
@section Recognizing Citations

Supercite also recognizes citations in the original article, and can
transform these already cited lines in a number of ways.  This is how
Supercite suppresses the multiple citing of non-nested citations.
Recognition of cited lines is controlled by variables analogous to
those that make up the citation string as mentioned previously.

@vindex sc-citation-leader-regexp
@vindex citation-leader-regexp @r{(sc-)}
@vindex sc-citation-delimiter-regexp
@vindex citation-delimiter-regexp @r{(sc-)}
@vindex sc-citation-separator-regexp
@vindex citation-separator-regexp @r{(sc-)}
@vindex sc-citation-root-regexp
@vindex citation-root-regexp @r{(sc-)}
@vindex sc-citation-nonnested-root-regexp
@vindex citation-nonnested-root-regexp @r{(sc-)}

The variable @code{sc-citation-leader-regexp} describes how citation
leaders can look, by default it matches any number of spaces or tabs.
Note that since the lisp function @code{looking-at} is used to do the
matching, if you change this variable it need not start with a leading
@code{"^"}.

Similarly, the variables @code{sc-citation-delimiter-regexp} and
@code{sc-citation-separator-regexp} respectively describe how citation
delimiters and separators can look.  They follow the same rule as
@code{sc-citation-leader-regexp} above.

When Supercite composes a citation string, it provides the attribution
automatically.  The analogous variable which handles recognition of the
attribution part of citation strings is @code{sc-citation-root-regexp}.
This variable describes the attribution root for both nested and
non-nested citations.  By default it can match zero-to-many alphanumeric
characters (also ``.'', ``-'', and ``_'').  But in some situations,
Supercite has to determine whether it is looking at a nested or
non-nested citation.  Thus the variable
@code{sc-citation-nonnested-root-regexp} is used to describe only
non-nested citation roots.  It is important to remember that if you
change @code{sc-citation-root-regexp} you should always also change
@code{sc-citation-nonnested-root-regexp}.

@node  Information Keys and the Info Alist
@chapter Information Keys and the Info Alist
@cindex information keys
@cindex Info Alist
@cindex information extracted from mail fields
@findex sc-mail-field
@findex mail-field @r{(sc-)}

@dfn{Mail header information keys} are nuggets of information that
Supercite extracts from the various mail headers of the original
message, placed in the reply buffer by the MUA@.  Information is kept
in the @dfn{Info Alist} as key-value pairs, and can be retrieved for
use in various places within Supercite, such as in header rewrite
functions and attribution selection.  Other bits of data, composed and
created by Supercite, are also kept as key-value pairs in this alist.
In the case of mail fields, the key is the name of the field, omitting
the trailing colon.  Info keys are always case insensitive (as are
mail headers), and the value for a corresponding key can be retrieved
from the alist with the @code{sc-mail-field} function.  Thus, if the
following fields were present in the original article:

@example
Date:@: 08 April 1991, 17:32:09 EST
Subject:@: Better get out your asbestos suit
@end example

@vindex sc-mumble
@vindex mumble @r{(sc-)}
@noindent
then, the following lisp constructs return:

@example
(sc-mail-field "date")
==> "08 April 1991, 17:32:09 EST"

(sc-mail-field "subject")
==> "Better get out your asbestos suit"
@end example

Since the argument to @code{sc-mail-field} can be any string, it is
possible that the mail field will not be present on the info alist
(possibly because the mail header was not present in the original
message).  In this case, @code{sc-mail-field} will return the value of
the variable @code{sc-mumble}.

Supercite always places all mail fields found in the yanked original
article into the info alist.  If possible, Supercite will also places
the following keys into the info alist:

@table @code
@cindex sc-attribution info field
@cindex attribution info field (sc-)
@item "sc-attribution"
the selected attribution string.

@cindex sc-citation info field
@cindex citation info field (sc-)
@item "sc-citation"
the non-nested citation string.

@cindex sc-from-address info field
@cindex from-address info field (sc-)
@item "sc-from-address"
email address extracted from the @samp{From:@:} field.

@cindex sc-reply-address info field
@cindex reply-address info field (sc-)
@item "sc-reply-address"
email address extracted from the @samp{Reply-To:@:} field.

@cindex sc-sender-address info field
@cindex sender-address info field (sc-)
@item "sc-sender-address"
email address extracted from the @samp{Sender:@:} field.

@cindex sc-emailname info field
@cindex emailname info field (sc-)
@item "sc-emailname"
email terminus extracted from the @samp{From:@:} field.

@cindex sc-initials info field
@cindex initials info field (sc-)
@item "sc-initials"
the author's initials.

@cindex sc-author info field
@cindex author info field (sc-)
@item "sc-author"
the author's full name.

@cindex sc-firstname info field
@cindex firstname info field (sc-)
@item "sc-firstname"
the author's first name.

@cindex sc-lastname info field
@cindex lastname info field (sc-)
@item "sc-lastname"
the author's last name.

@cindex sc-middlename-1 info field
@cindex middlename-1 info field (sc-)
@item "sc-middlename-1"
the author's first middle name.
@end table

If the author's name has more than one middle name, they will appear as
info keys with the appropriate index (e.g., @code{"sc-middlename-2"},
@dots{}).  @xref{Selecting an Attribution}.

@node  Reference Headers
@chapter Reference Headers
@cindex reference headers

Supercite will insert an informative @dfn{reference header} at the
beginning of the cited body of text, which display more detail about the
original article and provides the mapping between the attribution and
the original author in non-nested citations.  Whereas the citation
string usually only contains a portion of the original author's name,
the reference header can contain such information as the author's full
name, email address, the original article's subject, etc.  In fact any
information contained in the info alist can be inserted into a reference
header.

@menu
* The Built-in Header Rewrite Functions::
* Electric References::
@end menu

@cindex header rewrite functions
@vindex sc-rewrite-header-list
@vindex rewrite-header-list @r{(sc-)}
There are a number of built-in @dfn{header rewrite functions} supplied
by Supercite, but you can write your own custom header rewrite
functions (perhaps using the built-in ones as examples).  The variable
@code{sc-rewrite-header-list} contains the list of such header rewrite
functions.  This list is consulted both when inserting the initial
reference header, and when displaying @dfn{electric references}.
@xref{Electric References}.

@vindex sc-preferred-header-style
@vindex preferred-header-style @r{(sc-)}
When Supercite is initially run on a reply buffer (via
@code{sc-cite-original}), it will automatically call one of these
functions.  The one it uses is defined in the variable
@code{sc-preferred-header-style}.  The value of this variable is an
integer which is an index into the @code{sc-rewrite-header-list},
beginning at zero.

@node  The Built-in Header Rewrite Functions
@section The Built-in Header Rewrite Functions
@cindex header rewrite functions, built-in

Below are examples of the various built-in header rewrite functions.
Please note the following: first, the text which appears in the
examples below as @var{infokey} indicates that the corresponding value
of the info key from the info alist will be inserted there.
(@pxref{Information Keys and the Info Alist}).  For example, in @code{sc-header-on-said}
below, @var{date} and @var{from} correspond to the values of the
@samp{Date:@:} and @samp{From:@:} mail headers respectively.

@vindex sc-reference-tag-string
@vindex reference-tag-string @r{(sc-)}
Also, the string @code{">>>>>"} below is really the value of the
variable @code{sc-reference-tag-string}.  This variable is used in all
built-in header rewrite functions, and you can customize its value to
change the tag string globally.

Finally, the references headers actually written may omit certain parts
of the header if the info key associated with @var{infokey} is not
present in the info alist.  In fact, for all built-in headers, if the
@samp{From:@:} field is not present in the mail headers, the entire
reference header will be omitted (but this usually signals a serious
problem either in your MUA or in Supercite's installation).

@table @code
@findex sc-no-header
@findex no-header @r{(sc-)}
@item sc-no-header
This function produces no header.  It should be used instead of
@code{nil} to produce a blank header.  This header can possibly
contain a blank line after the @code{mail-header-separator} line.

@item sc-no-blank-line-or-header
@findex sc-no-blank-line-or-header
@findex no-blank-line-or-header @r{(sc-)}
This function is similar to @code{sc-no-header} except that any blank
line after the @code{mail-header-separator} line will be removed.

@item sc-header-on-said
@findex sc-header-on-said
@findex header-on-said @r{(sc-)}
@code{>>>>> On @var{date}, @var{from} said:}

@item sc-header-inarticle-writes
@findex sc-header-inarticle-writes
@findex header-inarticle-writes @r{(sc-)}
@code{>>>>> In article @var{message-id}, @var{from} writes:}

@item sc-header-regarding-adds
@findex sc-header-regarding-adds
@findex header-regarding-adds @r{(sc-)}
@code{>>>>> Regarding @var{subject}; @var{from} adds:}

@item sc-header-attributed-writes
@findex sc-header-attributed-writes
@findex header-attributed-writes @r{(sc-)}
@code{>>>>> "@var{sc-attribution}" == @var{sc-author} <@var{sc-reply-address}> writes:}

@item sc-header-author-writes
@findex sc-header-author-writes
@findex header-author-writes @r{(sc-)}
@code{>>>>> @var{sc-author} writes:}

@item sc-header-verbose
@findex sc-header-verbose
@findex header-verbose @r{(sc-)}
@code{>>>>> On @var{date},}@*
@code{>>>>> @var{sc-author}}@*
@code{>>>>> from the organization of @var{organization}}@*
@code{>>>>> who can be reached at:@: @var{sc-reply-address}}@*
@code{>>>>> (whose comments are cited below with:@: "@var{sc-cite}")}@*
@code{>>>>> had this to say in article @var{message-id}}@*
@code{>>>>> in newsgroups @var{newsgroups}}@*
@code{>>>>> concerning the subject of @var{subject}}@*
@code{>>>>> see @var{references} for more details}
@end table

@node  Electric References
@section Electric References
@cindex electric references

By default, when Supercite cites the original message for the first
time, it just goes ahead and inserts the reference header indexed by
@code{sc-preferred-header-style}.  However, you may want to select
different reference headers based on the type of reply or forwarding
you are doing.  You may also want to preview the reference header
before deciding whether to insert it into the reply buffer or
not. Supercite provides an optional @dfn{electric reference} mode
which you can drop into to give you this functionality.

@vindex sc-electric-references-p
@vindex electric-references-p @r{(sc-)}
If the variable @code{sc-electric-references-p} is non-@code{nil},
Supercite will bring up an electric reference mode buffer and place you
into a recursive edit.  The electric reference buffer is read-only, so
you cannot directly modify the reference text until you exit electric
references and insert the text into the reply buffer.  But you can cycle
through all the reference header rewrite functions in your
@code{sc-rewrite-header-list}.

You can also set a new preferred header style, jump to any header, or
jump to the preferred header.  The header will be shown in the electric
reference buffer and the header index and function name will appear in
the echo area.

The following commands are available while in electric reference mode
(shown here with their default key bindings):

@table @asis
@item @code{sc-eref-next} (@kbd{n})
@findex sc-eref-next
@findex eref-next @r{(sc-)}
@kindex n
@vindex sc-electric-circular-p
@vindex electric-circular-p @r{(sc-)}
Displays the next reference header in the electric reference buffer.  If
the variable @code{sc-electric-circular-p} is non-@code{nil}, invoking
@code{sc-eref-next} while viewing the last reference header in the list
will wrap around to the first header.

@item @code{sc-eref-prev} (@kbd{p})
@findex sc-eref-prev
@findex eref-prev @r{(sc-)}
@kindex p
Displays the previous reference header in the electric reference buffer.
If the variable @code{sc-electric-circular-p} is non-@code{nil},
invoking @code{sc-eref-prev} will wrap around to the last header.

@item @code{sc-eref-goto} (@kbd{g})
@findex sc-eref-goto
@findex eref-goto @r{(sc-)}
@kindex g
Goes to a specified reference header.  The index (into the
@code{sc-rewrite-header-list}) can be specified as a numeric argument to
the command.  Otherwise, Supercite will query you for the index in the
minibuffer.

@item @code{sc-eref-jump} (@kbd{j})
@findex sc-eref-jump
@findex eref-jump @r{(sc-)}
@kindex j
Display the preferred reference header, i.e., the one indexed by the current
value of @code{sc-preferred-header-style}.

@item @code{sc-eref-setn} (@kbd{s})
@findex sc-eref-setn
@findex eref-setn @r{(sc-)}
@kindex s
Set the preferred reference header (i.e.,
@code{sc-preferred-header-style}) to the currently displayed header.

@item @code{sc-eref-exit} (@kbd{C-j}, @key{RET}, and @kbd{@key{ESC} C-c})
@kindex RET
@kindex C-j
@kindex q
@findex sc-eref-exit
@findex eref-exit @r{(sc-)}
Exit from electric reference mode and insert the current header into the
reply buffer.

@item @code{sc-eref-abort} (@kbd{q}, @kbd{x})
@findex sc-eref-abort
@findex eref-abort @r{(sc-)}
@kindex x
Exit from electric reference mode without inserting the current header.
@end table

@vindex sc-electric-mode-hook
@vindex electric-mode-hook @r{(sc-)}
@noindent
Supercite will execute the hook @code{sc-electric-mode-hook} before
entering electric reference mode.

@node  Getting Connected
@chapter Getting Connected
@cindex citation interface specification

@vindex mail-citation-hook
@cindex .emacs file
In most cases, all that is necessary to begin using Supercite is to add
the following to @file{~.emacs}:

@example
(add-hook 'mail-citation-hook 'sc-cite-original)
@end example

@noindent For more details of the process, read on@dots{}

Hitting @kbd{C-c C-y} in your MUA's reply buffer yanks and cites the
original message into the reply buffer.  In reality, the citation of the
original message is performed via a call through a configurable hook
variable.  The name of this variable has been agreed to in advance as
part of the @dfn{citation interface specification}.  By default this
hook variable has a @code{nil} value, which the MUA recognizes to mean,
``use your default citation function.''  When you add Supercite's
citation function to the hook, thereby giving the variable a
non-@code{nil} value, it tells the MUA to run the hook via
@code{run-hooks} instead of using the default citation.

Early in Supercite's development, the Supercite author, a few MUA
authors, and some early Supercite users got together and agreed upon a
standard interface between MUAs and citation packages (of which
Supercite is currently the only known add-on @t{:-)}.  Supercite can
probably be used with most Emacs MUAs, with a greater or lesser degree
of effort.

To learn exactly how to connect Supercite to the software systems you
are using, read the appropriate following sections.  For details on the
interface specifications, or if you are writing or maintaining an MUA,
@pxref{Hints to MUA Authors}.

@cindex autoload
@cindex .emacs file
@findex sc-cite-original
@findex cite-original @r{(sc-)}
The first thing that everyone should do, regardless of the MUA you are
using is to set up Emacs so it will load Supercite at the appropriate
time.  This happens automatically if Supercite is distributed with your
Emacs version.  If not, you can set up an @dfn{autoload} for Supercite.

To do the latter, put the following in your @file{.emacs} file:

@example
(autoload 'sc-cite-original "supercite" nil t)
@end example

@cindex point
@cindex mark
The function @code{sc-cite-original} is the top-level Supercite function
designed to be run from the citation hook.  It expects
@samp{point} and @samp{mark} to be set around the region to cite, and it
expects the original article's mail headers to be present within this
region.  Note that Supercite @emph{never} touches any text outside this
region.  Note further that the region need not be active
for @code{sc-cite-original} to do its job.
@xref{Hints to MUA Authors}.

The other step in the getting connected process is to make sure your
MUA calls @code{sc-cite-original} at the right time.  As mentioned
above, some MUAs handle this differently.  Read the sections that follow
pertaining to the MUAs you are using.

@node  Replying and Yanking
@chapter Replying and Yanking

This chapter explains what happens when you reply and yank an original
message from an MUA.

@menu
* Reply Buffer Initialization::
* Filling Cited Text::
@end menu

@node  Reply Buffer Initialization
@section Reply Buffer Initialization
@findex sc-cite-original
@findex cite-original @r{(sc-)}

Executing @code{sc-cite-original} performs the following steps as it
initializes the reply buffer:

@enumerate
@item
@vindex sc-pre-hook
@vindex pre-hook @r{(sc-)}
@emph{Runs @code{sc-pre-hook}.}
This hook variable is run before @code{sc-cite-original} does any other
work.  You could conceivably use this hook to set certain Supercite
variables based on the reply buffer's mode or name (i.e., to do
something different based on whether you are replying or following up to
an article).

@item
@emph{Inserts Supercite's keymap.}
@vindex sc-mode-map-prefix
@vindex mode-map-prefix @r{(sc-)}
@kindex C-c C-p
@cindex keymap prefix
Supercite provides a number of commands for performing post-yank
modifications to the reply buffer.  These commands are installed on
Supercite's top-level keymap.  Since Supercite has to interface with a
wide variety of MUAs, it does not install all of its commands directly
into the reply buffer's keymap.  Instead, it puts its commands on a
keymap prefix, then installs this prefix onto the buffer's keymap.  What
this means is that you typically have to type more characters to invoke
a Supercite command, but Supercite's key bindings can be made much more
consistent across MUAs.

You can control what key Supercite uses as its keymap prefix by changing
the variable @code{sc-mode-map-prefix}.  By default, this variable is
set to @code{C-c C-p}; a finger twister perhaps, but unfortunately the
best default due to the scarcity of available key bindings in many MUAs.

@item
@emph{Turns on Supercite minor mode.}
@cindex modeline
The modeline of the reply buffer should indicate that Supercite is
active in that buffer by displaying the string @samp{SC}.

@item
@emph{Sets the ``Undo Boundary.''}
@cindex undo boundary
Supercite sets an undo boundary before it begins to modify the original
yanked text.  This allows you to easily undo Supercite's changes to
affect alternative citing styles.

@item
@emph{Processes the mail headers.}
@vindex sc-confirm-always-p
@vindex confirm-always-p @r{(sc-)}
@vindex sc-mail-warn-if-non-rfc822-p
@vindex mail-warn-if-non-rfc822-p @r{(sc-)}
All previously retrieved info key-value pairs are deleted from the info
alist, then the mail headers in the body of the yanked message are
scanned.  Info key-value pairs are created for each header found.  Also,
such useful information as the author's name and email address are
extracted.  If the variable @code{sc-mail-warn-if-non-rfc822-p} is
non-@code{nil}, then Supercite will warn you if it finds a mail header
that does not conform to RFC 822 (or later).
This is rare and indicates a problem
either with your MUA or the original author's MUA, or some MTA (mail
transport agent) along the way.

@vindex sc-nuke-mail-headers
@vindex sc-nuke-mail-header-list
@vindex nuke-mail-headers @r{(sc-)}
@vindex nuke-mail-header-list @r{(sc-)}
Once the info keys have been extracted from the mail headers, the
headers are nuked from the reply buffer.  You can control exactly which
headers are removed or kept, but by default, all headers are removed.

There are two variables which control mail header nuking.  The variable
@code{sc-nuke-mail-headers} controls the overall behavior of the header
nuking routines.  By setting this variable to @code{'all}, you
automatically nuke all mail headers.  Likewise, setting this variable to
@code{'none} inhibits nuking of any mail headers.  In between these
extremes, you can tell Supercite to nuke only a specified list of mail
headers by setting this variable to @code{'specified}, or to keep only a
specified list of headers by setting it to @code{'keep}.

If @code{sc-nuke-mail-headers} is set to @code{'specified} or
@code{'keep}, then the variable @code{sc-nuke-mail-header-list} is
consulted for the list of headers to nuke or keep.  This variable
contains a list of regular expressions.  If the mail header line matches
a regular expression in this list, the header will be nuked or kept.
The line is matched against the regexp using @code{looking-at} rooted at
the beginning of the line.

@vindex sc-blank-lines-after-headers
@vindex blank-lines-after-headers @r{(sc-)}
If the variable @code{sc-blank-lines-after-headers} is non-@code{nil},
it contains the number of blank lines remaining in the buffer after mail
headers are nuked.  By default, only one blank line is left in the buffer.

@item
@emph{Selects the attribution and citation strings.}
Once the mail headers have been processed, Supercite selects a
attribution string and a citation string which it will use to cite the
original message.  @xref{Selecting an Attribution}, for details.

@item
@emph{Cites the message body.}
@vindex sc-cite-region-limit
@vindex cite-region-limit @r{(sc-)}
After the selection of the attribution and citation strings, Supercite
cites the original message by inserting the citation string prefix in
front of every uncited line.  You may not want Supercite to
automatically cite very long messages however.  For example, some email
could contain a smaller header section followed by a huge uuencoded
message.  It wouldn't make sense to cite the uuencoded message part when
responding to the original author's short preface.  For this reason,
Supercite provides a variable which limits the automatic citation of
long messages to a certain maximum number of lines.  The variable is
called @code{sc-cite-region-limit}.  If this variable contains an
integer, messages with more lines that this will not be cited at all,
and a warning message will be displayed.  Supercite has performed
everything necessary, though, for you to manually cite only the small
portion of the original message that you want to use.

If @code{sc-cite-region-limit} contains a non-@code{nil} value, the
original message will always be cited, regardless of its size.  If the
variable contains the value @code{nil}, the region will never be cited
automatically.  Use this if you always want to be able to edit and cite
the message manually.

@vindex sc-cite-blank-lines-p
@vindex cite-blank-lines-p @r{(sc-)}
The variable @code{sc-cite-blank-lines-p} controls whether blank lines
in the original message should be cited or not.  If this variable is
non-@code{nil}, blank lines will be cited just like non-blank lines.
Otherwise, blank lines will be treated as paragraph separators.

Citing of the original message is highly configurable.  Supercite's
default setup does a pretty good job of citing many common forms of
previously cited messages.  But there are as many citation styles out
there as people on the net, or just about!  It would be impossible for
Supercite to anticipate every style in existence, and you probably
wouldn't encounter them all anyway.  But you can configure Supercite to
recognize those styles you see often.
@xref{Configuring the Citation Engine}, for details.

@item
@emph{Runs @code{sc-post-hook}.}
@vindex sc-post-hook
@vindex post-hook @r{(sc-)}
This variable is very similar to @code{sc-pre-hook}, except that it runs
after @code{sc-cite-original} is finished.  This hook is provided mostly
for completeness and backward compatibility.  Perhaps it could be used to
reset certain variables set in @code{sc-pre-hook}.
@end enumerate

@node  Filling Cited Text
@section Filling Cited Text
@cindex filling paragraphs
@vindex sc-auto-fill-region-p
@vindex auto-fill-region-p @r{(sc-)}
@cindex filladapt
@cindex gin-mode
@findex sc-setup-filladapt
@findex setup-filladapt @r{(sc-)}

Supercite will automatically fill newly cited text from the original
message unless the variable @code{sc-auto-fill-region-p} has a
@code{nil} value.  Supercite will also re-fill paragraphs when you
manually cite or re-cite text.

However, during normal editing, Supercite itself cannot be used to fill
paragraphs.  This is a change from version 2.  There are other add-on
lisp packages which do filling much better than Supercite ever did.  The
two best known are @dfn{filladapt} and @dfn{gin-mode}.  Both work well
with Supercite and both are available at the normal Emacs Lisp archive
sites.  @dfn{gin-mode} works pretty well out of the box, but if you use
@dfn{filladapt}, you may want to run the function
@code{sc-setup-filladapt} after loading @file{supercite}
(e.g., using @code{with-eval-after-load}).  This simply
makes @dfn{filladapt} a little more Supercite savvy than its default
setup.

@vindex sc-fixup-whitespace-p
@vindex fixup-whitespace-p @r{(sc-)}
Also, Supercite will collapse leading whitespace between the citation
string and the text on a line when the variable
@code{sc-fixup-whitespace-p} is non-@code{nil}.  The default value for
this variable is @code{nil}.

@vindex fill-prefix
Its important to understand that Supercite's automatic filling (during
the initial citation of the reply) is very fragile.  That is because
figuring out the @code{fill-prefix} for a particular paragraph is a
really hard thing to do automatically.  This is especially the case when
the original message contains code or some other text where leading
whitespace is important to preserve.  For this reason, many Supercite
users typically run with @code{sc-auto-fill-region-p} (and possibly also
@code{sc-fixup-whitespace-p}) set to @code{nil}.  They then manually
fill each cited paragraph in the reply buffer.

I usually run with both these variables containing their default values.
When Supercite's automatic filling breaks on a particular message, I
will use Emacs's undo feature to undo back before the citation was
applied to the original message.  Then I'll toggle the variables and
manually cite those paragraphs that I don't want to fill or collapse
whitespace on.  @xref{Variable Toggling Shortcuts}.

@kindex C-c C-p C-p
If you find that Supercite's automatic filling is just too fragile for
your tastes, you might consider one of these alternate approaches.
Also, to make life easier, a shortcut function to toggle the state of
both of these variables is provided on the key binding
@kbd{C-c C-p C-p} (with the default value of @code{sc-mode-map-prefix};
@pxref{Post-yank Formatting Commands}).

You will noticed that the minor mode string will
show the state of these variables as qualifier characters.  When both
variables are @code{nil}, the Supercite minor mode string will display
@samp{SC}.  When just @code{sc-auto-fill-region-p} is non-@code{nil}, the
string will display @samp{SC:f}, and when just
@code{sc-fixup-whitespace-p} is non-@code{nil}, the string will display
@samp{SC:w}.  When both variables are non-@code{nil}, the string will
display @samp{SC:fw}.  Note that the qualifiers chosen are mnemonics for
the default bindings of the toggling function for each respective
variable.
@xref{Variable Toggling Shortcuts}.

Why are these variables not set to @code{nil} by default?  It is because
many users won't manually fill paragraphs that are Supercited, and there
have been widespread complaints on the net about mail and news messages
containing lines greater than about 72 characters.  So the default is to
fill cited text.

@node  Selecting an Attribution
@chapter Selecting an Attribution
@cindex attribution list
@vindex sc-preferred-attribution-list
@vindex preferred-attribution-list @r{(sc-)}

As you know, the attribution string is the part of the author's name
that will be used to composed a non-nested citation string.  Supercite
scans the various mail headers present in the original article and uses
a number of heuristics to extract strings which it puts into the
@dfn{attribution association list} or @dfn{attribution alist}.  This is
analogous, but different from, the info alist previously mentioned.  Each
element in the attribution alist is a key-value pair containing such
information as the author's first name, middle names, and last name, the
author's initials, and the author's email terminus.

@menu
* Attribution Preferences::
* Anonymous Attributions::
* Author Names::
@end menu

@node  Attribution Preferences
@section Attribution Preferences

When you cite an original message, you can tell Supercite which part of
the author's name you would prefer it to use as the attribution.  The
variable @code{sc-preferred-attribution-list} controls this; it contains
keys which are matched against the attribution alist in the given order.
The first value of a key that produces a non-@code{nil}, non-empty
string match is used as the attribution string, and if no keys match, a
secondary mechanism is used to generate the attribution.
@xref{Anonymous Attributions}.

The following preferences are always available in the attribution alist
(barring error):

@table @code
@item "emailname"
the author's email terminus.

@item "initials"
the author's initials.

@item "firstname"
the author's first name.

@item "lastname"
the author's last name.

@item "middlename-1"
the author's first middle name.

@item "sc-lastchoice"
the last attribution string you have selected.  This is useful when you
recite paragraphs in the reply.

@item "sc-consult"
@vindex sc-attrib-selection-list
@vindex attrib-selection-list @r{(sc-)}
consults the customizable list @code{sc-attrib-selection-list} which can
be used to select special attributions based on the value of any info
key.  See below for details.

@item "x-attribution"
the original author's suggestion for attribution string choice.  See below
for details.
@end table

Middle name indexes can be any positive integer greater than zero,
though it is unlikely that many authors will have more than one middle
name, if that many.

At this point, let me digress into a discussion of etiquette.  It is my
belief that while the style of the citations is a reflection of the
personal tastes of the replier (i.e., you), the attribution selection is
ultimately the personal choice of the original author.  In a sense it is
his or her ``net nickname'', and therefore the author should have some
say in the selection of attribution string.  Imagine how you would feel
if someone gave you a nickname that you didn't like?

For this reason, Supercite recognizes a special mail header,
@samp{X-Attribution:}, which if present, tells Supercite the attribution
string preferred by the original author.  It is the value of this header
that is associated with the @code{"x-attribution"} key in the
attribution alist.  Currently, you can override the preference of this
key by changing @code{sc-preferred-attribution-list}, but that isn't
polite, and in the future Supercite may hard-code this.  For now, it is
suggested that if you change the order of the keys in this list, that
@code{"x-attribution"} always be first, or possible second behind only
@code{"sc-lastchoice"}.  This latter is the default.

@vindex sc-attrib-selection-list
@vindex attrib-selection-list @r{(sc-)}
The value @code{"sc-consult"} in @code{sc-preferred-attribution-list}
has a special meaning during attribution selection.  When Supercite
encounters this preference, it begins processing a customizable list of
attributions, contained in the variable @code{sc-attrib-selection-list}.
Each element in this list contains lists of the following form:

@example
@group
(@var{infokey} ((@var{regexp} . @var{attribution})
         (@var{regexp} . @var{attribution})
         (@dots{})))
@end group
@end example

@noindent
@findex sc-mail-field
@findex mail-field @r{(sc-)}
where @var{infokey} is a key for @code{sc-mail-field} and @var{regexp}
is a regular expression to match against the @var{infokey}'s value.  If
@var{regexp} matches the @var{infokey}'s value, the @var{attribution} is
used as the attribution string.  Actually, @var{attribution} can be a
string or a list; if it is a list, it is @code{eval}uated and the return
value (which must be a string), is used as the attribution.

This can be very useful for when you are replying to net acquaintances
who do not use the @samp{X-Attribution:@:} mail header.  You may know
what nickname they would prefer to use, and you can set up this list to
match against a specific mail field, e.g., @samp{From:@:}, allowing you
to cite your friend's message with the appropriate attribution.

@node  Anonymous Attributions
@section Anonymous Attributions
@vindex sc-default-author-name
@vindex default-author-name @r{(sc-)}
@vindex sc-default-attribution
@vindex default-attribution @r{(sc-)}

When the author's name cannot be found in the @samp{From:@:} mail
header, a fallback author name and attribution string must be supplied.
The fallback author name is contained in the variable
@code{sc-default-author-name} and the fallback attribution string is
contained in the variable @code{sc-default-attribution}.  Default values
for these variables are @code{"Anonymous"} and @code{"Anon"},
respectively.  Note that in most circumstances, getting the default
author name or attribution is a sign that something is set up
incorrectly.

@vindex sc-use-only-preference-p
@vindex use-only-preference-p @r{(sc-)}
Also, if the preferred attribution, which you specified in your
@code{sc-preferred-attribution-list} variable cannot be found, a
secondary method can be employed to find a valid attribution string.  The
variable @code{sc-use-only-preference-p} controls what happens in this
case.  If the variable's value is non-@code{nil}, then
@code{sc-default-author-name} and @code{sc-default-attribution} are
used, otherwise, the following steps are taken to find a valid
attribution string, and the first step to return a non-@code{nil},
non-empty string becomes the attribution:

@enumerate
@item
Use the last selected attribution, if there is one.

@item
Use the value of the @code{"x-attribution"} key.

@item
Use the author's first name.

@item
Use the author's last name.

@item
Use the author's initials.

@item
Find the first non-@code{nil}, non-empty attribution string in the
attribution alist.

@item
@code{sc-default-attribution} is used.
@end enumerate

@vindex sc-confirm-always-p
@vindex confirm-always-p @r{(sc-)}
Once the attribution string has been automatically selected, a number of
things can happen.  If the variable @code{sc-confirm-always-p} is
non-@code{nil}, you are queried for confirmation of the chosen
attribution string.  The possible values for completion are those strings
in the attribution alist, however you are not limited to these choices.
You can type any arbitrary string at the confirmation prompt.  The string
you enter becomes the value associated with the @code{"sc-lastchoice"}
key in the attribution alist.

@vindex sc-downcase-p
@vindex downcase-p @r{(sc-)}
Once an attribution string has been selected, Supercite will force the
string to lower case if the variable @code{sc-downcase-p} is
non-@code{nil}.

@vindex sc-attribs-preselect-hook
@vindex attribs-preselect-hook @r{(sc-)}
@vindex sc-attribs-postselect-hook
@vindex attribs-postselect-hook @r{(sc-)}

Two hook variables provide even greater control of the attribution
selection process.  The hook @code{sc-attribs-preselect-hook} is run
before any attribution is selected.  Likewise, the hook
@code{sc-attribs-postselect-hook} is run after the attribution is
selected (and the corresponding citation string is built), but before
these values are committed for use by Supercite.  During the
post-selection hook, the local variables @code{attribution} and
@code{citation} are bound to the appropriate strings.  By changing these
variables in your hook functions, you change the attribution and
citation strings used by Supercite.  One possible use of this would be
to override any automatically derived attribution string when it is only
one character long; e.g., you prefer to use @code{"initials"} but the
author only has one name.

@node  Author Names
@section Author Names
@cindex author names

Supercite employs a number of heuristics to decipher the author's name
based on value of the @samp{From:@:} mail field of the original message.
Supercite can recognize almost all of the common @samp{From:@:} field
formats in use.  If you encounter a @samp{From:@:} field that Supercite
cannot parse, please report this bug using @kbd{M-x report-emacs-bug}.

@vindex sc-titlecue-regexp
@vindex titlecue-regexp @r{(sc-)}
There are a number of Supercite variables that control how author names
are extracted from the @samp{From:@:} header.  Some headers may contain a
descriptive title as in:

@example
From:@: computer!speedy!doe (John Xavier-Doe -- Decent Hacker)
@end example

Supercite knows which part of the @samp{From:@:} header is email address
and which part is author name, but in this case the string @code{"Decent
Hacker"} is not part of the author's name.  You can tell Supercite to
ignore the title, while still recognizing hyphenated names through the
use of a regular expression in the variable @code{sc-titlecue-regexp}.
This variable has the default value of @code{"\\\\s +-+\\\\s +"}.  Any
text after this regexp is encountered is ignored as noise.

@vindex sc-name-filter-alist
@vindex name-filter-alist @r{(sc-)}
Some @samp{From:@:} headers may contain extra titles in the name fields
not separated by a title cue, but which are nonetheless not part of the
author's name proper.  Examples include the titles ``Dr.'', ``Mr.'',
``Ms.'', ``Jr.'', ``Sr.'', and ``III'' (e.g., Thurston Howe, the Third).
Also, some companies prepend or append the name of the division,
organization, or project on the author's name.  All of these titles are
noise which should be ignored.  The variable @code{sc-name-filter-alist}
is used for this purpose.  As implied by its name, this variable is an
association list, where each element is a cons cell of the form:

@example
(@var{regexp} . @var{position})
@end example

@noindent
where @var{regexp} is a regular expression that is matched (using
@code{string-match}) against each element of the @samp{From:@:} field's
author name.  @var{position} is a position indicator, starting at zero.
Thus to strip out all titles of ``Dr.'', ``Mr.'', etc.  from the name,
@code{sc-name-filter-alist} would have an entry such as:

@example
("^\\(Mr\\|Mrs\\|Ms\\|Dr\\)[.]?$" . 0)
@end example

@noindent
which only removes them if they appear as the first word in the name.
The position indicator is an integer, or one of the two special symbols
@code{last} or @code{any}.  @code{last} always matches against the last
word in the name field, while @code{any} matches against every word in
the name field.

@node  Configuring the Citation Engine
@chapter Configuring the Citation Engine
@cindex Regi
@cindex frames (Regi)
@cindex entries (Regi)

At the heart of Supercite is a regular expression interpreting engine
called @dfn{Regi}.  Regi operates by interpreting a data structure
called a Regi-frame (or just @dfn{frame}), which is a list of
Regi-entries (or just @dfn{entry}).  Each entry contains a predicate,
typically a regular expression, which is matched against a line of text
in the current buffer.  If the predicate matches true, an associated
expression is @code{eval}uated.  In this way, an entire region of text
can be transformed in an @emph{awk}-like manner.  Regi is used
throughout Supercite, from mail header information extraction, to header
nuking, to citing text.

@menu
* Using Regi::
* Frames You Can Customize::
@end menu

While the details of Regi are discussed below (@pxref{Using Regi}), only
those who wish to customize certain aspects of Supercite need concern
themselves with it.  It is important to understand though, that any
conceivable citation style that can be described by a regular expression
can be recognized by Supercite.  This leads to some interesting
applications.  For example, if you regularly receive email from a
co-worker that uses an uncommon citation style (say one that employs a
@samp{|} or @samp{@}} character at the front of the line), it is
possible for Supercite to recognize this and @emph{coerce} the citation
to your preferred style, for consistency.  In theory, it is possible for
Supercite to recognize such things as uuencoded messages or C code and
cite or fill those differently from normal text.  None of this is
currently part of Supercite, but contributions are welcome!

@node  Using Regi
@section Using Regi
@findex regi-interpret
@findex eval
@findex looking-at

Regi works by interpreting frames with the function
@code{regi-interpret}.  A frame is a list of arbitrary size where each
element is an entry of the following form:

@example
(@var{pred} @var{func} [@var{negate-p} [@var{case-fold-search}]])
@end example

Regi starts with the first entry in a frame, evaluating the @var{pred}
of that entry against the beginning of the line that @samp{point} is on.
If the @var{pred} evaluates to true (or false if the optional
@var{negate-p} is non-@code{nil}), then the @var{func} for that entry is
@code{eval}uated.  How processing continues is determined by the return
value for @var{func}, and is described below.  If @var{pred} was false
the next entry in the frame is checked until all entries have been
matched against the current line.  If no entry matches, @samp{point} is
moved forward one line and the frame is reset to the first entry.

@var{pred} can be a string, a variable, a list or one of the following
symbols: @code{t}, @code{begin}, @code{end}, or @code{every}.  If
@var{pred} is a string, or a variable or list that @code{eval}uates to a
string, it is interpreted as a regular expression.  This regexp is
matched against the current line, from the beginning, using
@code{looking-at}.  This match folds case if the optional
@var{case-fold-search} is non-@code{nil}.  If @var{pred} is not a
string, or does not @code{eval}uate to a string, it is interpreted as a
binary value (@code{nil} or non-@code{nil}).

The four special symbol values for @var{pred} are recognized:

@table @code
@item t
Always produces a true outcome.
@item begin
Always executed before the frame is interpreted.  This can be used to
initialize some global variables for example.
@item end
Always executed after frame interpreting is completed.  This can be used
to perform any necessary post-processing.
@item every
Executes whenever the frame is reset, usually after the entire frame has
been matched against the current line.
@end table

Note that @var{negate-p} and @var{case-fold-search} are ignored if
@var{pred} is one of these special symbols.  Only the first occurrence of
each symbol in a frame is used; any duplicates are ignored.  Also
note that for performance reasons, the entries associated with these
symbols are removed from the frame during the main interpreting loop.

Your @var{func} can return certain values which control continued Regi
processing.  By default, if your @var{func} returns @code{nil} (as it
should be careful to do explicitly), Regi will reset the frame to the
first entry, and advance @samp{point} to the beginning of the next line.
If a list is returned from your function, it can contain any combination
of the following elements:

@table @asis
@item the symbol @code{continue}
This tells Regi to continue processing entries after a match, instead of
resetting the frame and moving @samp{point}.  In this way, lines of text
can have multiple matches, but you have to be careful to avoid entering
infinite loops.

@item the symbol @code{abort}
This tells Regi to terminate frame processing.  However, any @code{end}
entry is still processed.

@item the list @code{(frame . @var{newframe})}
This tells Regi to substitute @var{newframe} as the frame it is
interpreting.  In other words, your @var{func} can modify the Regi frame
on the fly.  @var{newframe} can be a variable containing a frame, or it
can be the frame in-lined.

@item the list @code{(step . @var{step})}
Tells Regi to move @var{step} number of lines forward as it continues
processing.  By default, Regi moves forward one line.  @var{step} can be
zero or negative of course, but watch out for infinite loops.
@end table

During execution of your @var{func}, the following variables will be
temporarily bound to some useful information:

@table @code
@item curline
The current line in the buffer that Regi is @code{looking-at}, as a string.
@item curframe
The current frame being interpreted.
@item curentry
The current frame entry being interpreted.
@end table

@node  Frames You Can Customize
@section Frames You Can Customize
@vindex sc-nuke-mail-header

As mentioned earlier, Supercite uses various frames to perform
certain jobs such as mail header information extraction and mail header
nuking.  However, these frames are not available for you to customize,
except through abstract interfaces such as @code{sc-nuke-mail-header},
et al.

@vindex sc-default-cite-frame
However, the citation frames Supercite uses provide a lot of customizing
power and are thus available to you to change to suit your needs.  The
workhorse of citation is the frame contained in the variable
@code{sc-default-cite-frame}.  This frame recognizes many situations,
such as blank lines, which it interprets as paragraph separators.  It
also recognizes previously cited nested and non-nested citations in the
original message.  By default it will coerce non-nested citations into
your preferred citation style, and it will add a level of citation to
nested citations.  It will also simply cite uncited lines in your
preferred style.

@cindex unciting
@cindex reciting
@vindex sc-default-uncite-frame
@vindex sc-default-recite-frame
In a similar vein, there are default frames for @dfn{unciting} and
@dfn{reciting}, contained in the variables
@code{sc-default-uncite-frame} and @code{sc-default-recite-frame}
respectively.

As mentioned earlier (@pxref{Recognizing Citations}), citations are
recognized through the values of the regular expressions
@code{sc-citation-root-regexp}, et al.  To recognize odd styles, you
could modify these variables, or you could modify the default citing
frame.  Alternatively, you could set up association lists of frames for
recognizing specific alternative forms.

@vindex sc-cite-frame-alist
@vindex sc-uncite-frame-alist
@vindex sc-recite-frame-alist
For each of the actions---citing, unciting, and reciting---an alist is
consulted to find the frame to use (@code{sc-cite-frame-alist},
@code{sc-uncite-frame-alist}, and @code{sc-recite-frame-alist}
respectively).  These frames can contain alists of the form:

@example
((@var{infokey} (@var{regexp} . @var{frame}) (@var{regexp} . @var{frame}) @dots{})
 (@var{infokey} (@var{regexp} . @var{frame}) (@var{regexp} . @var{frame}) @dots{})
 (@dots{}))
@end example

@vindex sc-mail-field
@findex string-match
Where @var{infokey} is a key suitable for @code{sc-mail-field},
@var{regexp} is a regular expression which is @code{string-match}'d
against the value of the @code{sc-mail-field} key, and @var{frame} is
the frame to use if a match occurred.  @var{frame} can be a variable
containing a frame or a frame in-lined.

When Supercite is about to cite, uncite, or recite a region, it consults
the appropriate alist and attempts to find a frame to use.  If one
is not found from the alist, then the appropriate default frame is used.

@node  Post-yank Formatting Commands
@chapter Post-yank Formatting Commands
@vindex sc-mode-map-prefix
@vindex mode-map-prefix @r{(sc-)}
@kindex C-c C-p

Once the original message has been yanked into the reply buffer, and
@code{sc-cite-original} has had a chance to do its thing, a number of
useful Supercite commands will be available to you.  Since there is wide
variety in the keymaps that MUAs set up in their reply buffers, it is
next to impossible for Supercite to properly sprinkle its commands into
the existing keymap.  For this reason Supercite places its commands on a
separate keymap, putting this keymap onto a prefix key in the reply
buffer.  You can customize the prefix key Supercite uses by changing the
variable @code{sc-mode-map-prefix}.  By default, the
@code{sc-mode-map-prefix} is @kbd{C-c C-p}; granted, not a great choice,
but unfortunately the best general solution so far.  In the rest of this
chapter, we'll assume you've installed Supercite's keymap on the default
prefix.

@menu
* Citing Commands::
* Insertion Commands::
* Variable Toggling Shortcuts::
* Mail Field Commands::
* Miscellaneous Commands::
@end menu

@node   Citing Commands
@section Commands to Manually Cite, Recite, and Uncite
@vindex sc-cite-region-limit

Probably the three most common post-yank formatting operations that you
will perform will be the manual citing, reciting, and unciting of
regions of text in the reply buffer.  Often you may want to recite a
paragraph to use a nickname, or manually cite a message when setting
@code{sc-cite-region-limit} to @code{nil}.  The following commands
perform these functions on the region of text between @samp{point} and
@samp{mark}.  Each of them sets the @dfn{undo boundary} before modifying
the region so that the command can be undone in the standard Emacs
way.

Here is the list of Supercite citing commands:

@table @asis
@findex sc-cite-region
@findex cite-region @r{(sc-)}
@kindex C-c C-p c
@vindex sc-pre-cite-hook
@vindex pre-cite-hook @r{(sc-)}
@vindex sc-confirm-always-p
@vindex confirm-always-p
@kindex C-u
@item @code{sc-cite-region} (@kbd{C-c C-p c})
This command cites each line in the region of text by interpreting the
selected frame from @code{sc-cite-frame-alist}, or the default citing
frame @code{sc-default-cite-frame}.  It runs the hook
@code{sc-pre-cite-hook} before interpreting the frame.  With an optional
universal argument (@kbd{C-u}), it temporarily sets
@code{sc-confirm-always-p} to @code{t} so you can confirm the
attribution string for a single manual citing.
@xref{Configuring the Citation Engine}.

@findex sc-uncite-region
@findex uncite-region @r{(sc-)}
@kindex C-c C-p u
@item @code{sc-uncite-region} (@kbd{C-c C-p u})
This command removes any citation strings from the beginning of each
cited line in the region by interpreting the selected frame from
@code{sc-uncite-frame-alist}, or the default unciting frame
@code{sc-default-uncite-frame}.  It runs the hook
@code{sc-pre-uncite-hook} before interpreting the frame.
@xref{Configuring the Citation Engine}.

@findex sc-recite-region
@findex recite-region @r{(sc-)}
@kindex C-c C-p r
@item @code{sc-recite-region} (@kbd{C-c C-p r})
This command recites each line the region by interpreting the selected
frame from @code{sc-recite-frame-alist}, or the default reciting frame
@code{sc-default-recite-frame}.  It runs the hook
@code{sc-pre-recite-hook} before interpreting the frame.
@xref{Configuring the Citation Engine}.

@vindex sc-confirm-always-p
@vindex confirm-always-p @r{(sc-)}
Supercite will always ask you to confirm the attribution when reciting a
region, regardless of the value of @code{sc-confirm-always-p}.
@end table

@node  Insertion Commands
@section Insertion Commands

These two functions insert various strings into the reply buffer.

@table @asis
@findex sc-insert-reference
@findex insert-reference @r{(sc-)}
@kindex C-c C-p w
@item @code{sc-insert-reference} (@kbd{C-c C-p w})
@vindex sc-preferred-header-style
@vindex preferred-header-style @r{(sc-)}
Inserts a reference header into the reply buffer at @samp{point}.  With
no arguments, the header indexed by @code{sc-preferred-header-style} is
inserted.  An optional numeric argument is the index into
@code{sc-rewrite-header-list} indicating which reference header to
write.

With just the universal argument (@kbd{C-u}), electric reference mode is
entered, regardless of the value of @code{sc-electric-references-p}.

@findex sc-insert-citation
@findex insert-citation @r{(sc-)}
@kindex C-c C-p i
@item @code{sc-insert-citation} (@kbd{C-c C-p i})
Inserts the current citation string at the beginning of the line that
@samp{point} is on.  If the line is already cited, Supercite will issue
an error and will not cite the line.
@end table

@node  Variable Toggling Shortcuts
@section Variable Toggling Shortcuts
@cindex toggling variables

Supercite defines a number of commands that make it easier for you to
toggle and set various Supercite variables as you are editing the reply
buffer.  For example, you may want to turn off filling or whitespace
cleanup, but only temporarily.  These toggling shortcut commands make
this easy to do.

@kindex C-c C-p C-t
Like Supercite commands in general, the toggling commands are placed on
a keymap prefix within the greater Supercite keymap.  For the default
value of @code{sc-mode-map-prefix}, this will be
@kbd{C-c C-p C-t}.

The following commands toggle the value of certain Supercite variables
which take only a binary value:

@table @kbd
@item C-c C-p C-t b
Toggles the variable @code{sc-mail-nuke-blank-lines-p}.

@item C-c C-p C-t c
Toggles the variable @code{sc-confirm-always-p}.

@item C-c C-p C-t d
Toggles the variable @code{sc-downcase-p}.

@item C-c C-p C-t e
Toggles the variable @code{sc-electric-references-p}.

@item C-c C-p C-t f
Toggles the variable @code{sc-auto-fill-region-p}.

@item C-c C-p C-t o
Toggles the variable @code{sc-electric-circular-p}.

@item C-c C-p C-t s
Toggles the variable @code{sc-nested-citation-p}.

@item C-c C-p C-t u
Toggles the variable @code{sc-use-only-preferences-p}.

@item C-c C-p C-t w
Toggles the variable @code{sc-fixup-whitespace-p}.
@end table

@findex set-variable
The following commands let you set the value of multi-value variables,
in the same way that Emacs's @code{set-variable} does:

@table @kbd
@item C-c C-p C-t a
Sets the value of the variable @code{sc-preferred-attribution-list}.

@item C-c C-p C-t l
Sets the value of the variable @code{sc-cite-region-limit}.

@item C-c C-p C-t n
Sets the value of the variable @code{sc-mail-nuke-mail-headers}.

@item C-c C-p C-t N
Sets the value of the variable @code{sc-mail-header-nuke-list}.

@item C-c C-p C-t p
Sets the value of the variable @code{sc-preferred-header-style}.
@end table

@kindex C-c C-p C-p
One special command is provided to toggle both
@code{sc-auto-fill-region-p} and @code{sc-fixup-whitespace-p} together.
This is because you typically want to run Supercite with either variable
as @code{nil} or non-@code{nil}.  The command to toggle these variables
together is bound on @kbd{C-c C-p C-p}.

Finally, the command @kbd{C-c C-p C-t h} (also @kbd{C-c C-p C-t ?})
brings up a Help message on the toggling keymap.


@node  Mail Field Commands
@section Mail Field Commands

These commands allow you to view, modify, add, and delete various bits
of information from the info alist.
@xref{Information Keys and the Info Alist}.

@table @asis
@kindex C-c C-p f
@findex sc-mail-field-query
@findex mail-field-query @r{(sc-)}
@kindex C-c C-p f
@item @code{sc-mail-field-query} (@kbd{C-c C-p f})
Allows you to interactively view, modify, add, and delete info alist
key-value pairs.  With no argument, you are prompted (with completion)
for an info key.  The value associated with that key is displayed in the
minibuffer.  With an argument, this command will first ask if you want
to view, modify, add, or delete an info key.  Viewing is identical to
running the command with no arguments.

If you want to modify the value of a key, Supercite will first prompt
you (with completion) for the key of the value you want to change.  It
will then put you in the minibuffer with the key's current value so you
can edit the value as you wish.  When you hit @key{RET}, the key's value
is changed.  Minibuffer history is kept for the values.

If you choose to delete a key-value pair, Supercite will prompt you (with
completion) for the key to delete.

If you choose to add a new key-value pair, Supercite firsts prompts you
for the key to add.  Note that completion is turned on for this prompt,
but you can type any key name here, even one that does not yet exist.
After entering the key, Supercite prompts you for the key's value.  It
is not an error to enter a key that already exists, but the new value
will override any old value.  It will not replace it though; if you
subsequently delete the key-value pair, the old value will reappear.

@findex sc-mail-process-headers
@findex mail-process-headers @r{(sc-)}
@kindex C-c C-p g
@item @code{sc-mail-process-headers} (@kbd{C-c C-p g})
This command lets you re-initialize Supercite's info alist from any set
of mail headers in the region between @samp{point} and @samp{mark}.
This function is especially useful for replying to digest messages where
Supercite will initially set up its information for the digest
originator, but you want to cite each component article with the real
message author.  Note that unless an error during processing occurs, any
old information is lost.
@end table

@node  Miscellaneous Commands
@section Miscellaneous Commands

@table @asis
@findex sc-open-line
@findex open-line @r{(sc-)}
@findex open-line
@kindex C-c C-p o
@item @code{sc-open-line} (@kbd{C-c C-p o})
Similar to Emacs's standard @code{open-line} commands, but inserts the
citation string in front of the new line.  As with @code{open-line},
an optional numeric argument inserts that many new lines.
@end table

@node  Hints to MUA Authors
@chapter Hints to MUA Authors

In June of 1989, some discussion was held between the various MUA
authors, the Supercite author, and other Supercite users.  These
discussions centered around the need for a standard interface between
MUAs and Supercite (or any future Supercite-like packages).  This
interface was formally proposed by Martin Neitzel on Fri, 23 Jun 89, in
a mail message to the Supercite mailing list:

@example
        Martin> Each news/mail-reader should provide a form of
        Martin> mail-yank-original that

        Martin> 1: inserts the original message incl. header into the
        Martin>    reply buffer; no indentation/prefixing is done, the header
        Martin>    tends to be a "full blown" version rather than to be
        Martin>    stripped down.

        Martin> 2: 'point' is at the start of the header, 'mark' at the
        Martin>    end of the message body.

        Martin> 3: (run-hooks 'mail-yank-hooks)

        Martin> [Supercite] should be run as such a hook and merely
        Martin> rewrite the message.  This way it isn't anymore
        Martin> [Supercite]'s job to gather the original from obscure
        Martin> sources. [@dots{}]
@end example

@vindex mail-citation-hook
@vindex mail-yank-hooks
@cindex sendmail.el
@findex mail-yank-original
@findex defvar
This specification was adopted, but underwent a slight modification with
the release of Emacs 19.  Instead of the variable
@code{mail-yank-hooks}, the hook variable that the MUA should provide is
@code{mail-citation-hook}.  Richard Stallman suggests that the MUAs
should @code{defvar} @code{mail-citation-hook} to @code{nil} and perform
some default citing when that is the case.

If you are writing a new MUA package, or maintaining an existing MUA
package, you should make it conform to this interface so that your users
will be able to link Supercite easily and seamlessly.  To do this, when
setting up a reply or forward buffer, your MUA should follow these
steps:

@enumerate
@item
Insert the original message, including the mail headers into the reply
buffer.  At this point you should not modify the raw text in any way
(except for any necessary decoding, e.g., of quoted-printable text), and
you should place all the original headers into the body of the reply.
This means that many of the mail headers will be duplicated, one copy
above the @code{mail-header-separator} line and one copy below, however
there will probably be more headers below this line.

@item
Set @samp{point} to the beginning of the line containing the first mail
header in the body of the reply.  Set @samp{mark} at the end of the
message text.  It is very important that the region be set around the
text Supercite is to modify and that the mail headers are within this
region.  Supercite will not venture outside the region for any reason,
and anything within the region is fair game, so don't put anything that
@strong{must} remain unchanged inside the region.

@item
Run the hook @code{mail-citation-hook}.  You will probably want to
provide some kind of default citation functions in cases where the user
does not have Supercite installed.  By default, your MUA should
@code{defvar} @code{mail-citation-hook} to @code{nil}, and in your
yanking function, check its value.  If it finds
@code{mail-citation-hook} to be @code{nil}, it should perform some
default citing behavior.  User who want to connect to Supercite then
need only add @code{sc-cite-original} to this list of hooks using
@code{add-hook}.
@end enumerate

If you do all this your MUA will join the ranks of those that conform to
this interface ``out of the box.''

@node Thanks and History
@chapter Thanks and History

The Supercite package was derived from its predecessor Superyank 1.11
which was inspired by various bits of code and ideas from Martin Neitzel
and Ashwin Ram.  They were the folks who came up with the idea of
non-nested citations and implemented some rough code to provide this
style.  Superyank and Supercite version 2 evolved to the point where much
of the attribution selection mechanism was automatic, and features have
been continuously added through the comments and suggestions of the
Supercite mailing list participants.

With version 3, Supercite underwent an almost complete rewrite,
benefiting in a number of ways, including vast improvements in the
speed of performance, a big reduction in size of the code and in the use
of Emacs resources, and a much cleaner and flexible internal
architecture.  Most of this work was internal and not of very great
importance to the casual user.  There were some changes at the
user-visible level, but for the most part, the Supercite configuration
variables from version 2 should still be relevant to version 3.
Hopefully Supercite version 3 is faster, smaller, and much more flexible
than its predecessors.

In the version 2 manual I thanked some specific people for their help in
developing Supercite 2.  You folks know who you are and your continued
support is greatly appreciated.  I wish to thank everyone on the
Supercite mailing list, especially the brave alpha testers, who helped
considerably in testing out the concepts and implementation of Supercite
version 3.  Special thanks go out to the MUA and Emacs authors Kyle
Jones, Stephen Gildea, Richard Stallman, and Jamie Zawinski for coming
to a quick agreement on the new @code{mail-citation-hook} interface, and
for adding the magic lisp to their code to support this.

All who have helped and contributed have been greatly appreciated.

Supercite was written by Barry Warsaw.

@node GNU Free Documentation License
@appendix GNU Free Documentation License
@include doclicense.texi

@node  Concept Index
@unnumbered Concept Index
@printindex cp

@node  Command Index
@unnumbered Command Index

Since all supercite commands are prepended with the string
``@code{sc-}'', each appears under its @code{sc-}@var{command} name and
its @var{command} name.
@iftex
@sp 2
@end iftex
@printindex fn

@node  Key Index
@unnumbered Key Index
@printindex ky

@node Variable Index
@unnumbered Variable Index

Since all supercite variables are prepended with the string
``@code{sc-}'', each appears under its @code{sc-}@var{variable} name and
its @var{variable} name.
@iftex
@sp 2
@end iftex
@printindex vr
@bye