% Copyright 2005-2017 Cisco Systems, Inc.
% 
% Licensed under the Apache License, Version 2.0 (the "License");
% you may not use this file except in compliance with the License.
% You may obtain a copy of the License at
% 
% http://www.apache.org/licenses/LICENSE-2.0
% 
% Unless required by applicable law or agreed to in writing, software
% distributed under the License is distributed on an "AS IS" BASIS,
% WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
% See the License for the specific language governing permissions and
% limitations under the License.
\chapter{Expression Editor\label{CHPTEXPEDITOR}}

When the expression editor (expeditor) is enabled as described in
Section~\ref{SECTUSEEXPEDITOR}, it allows the user to edit expressions
entered into the system and move backwards and forwards through
a history of entered expressions.
This chapter describes a set of parameters that may be used to
control various aspects of the expression editor's behavior
(Section~\ref{SECTEXPEDITORPARAMS}),
a procedure for binding key sequences to editing commands
(Section~\ref{SECTEXPEDITORKEYBINDING}),
the built-in editing commands
(Section~\ref{SECTEXPEDITOREDITCMDS}), and mechanisms for creating new
editing commands (Section~\ref{SECTEXPEDITORNEWCMDS}).

These mechanisms are available through the \scheme{expression-editor} module.

%----------------------------------------------------------------------------
\entryheader
\formdef{expression-editor}{\categorymodule}{expression-editor}
\listlibraries
\endentryheader

The \scheme{expression-editor} module exports a set of bindings for
parameters and other procedures that can be used to modify how the
expression editor interacts with the user, including the particular keys
used to invoke the various editing commands.

\medskip
Basic use of the expression editor is described in Section~\ref{SECTUSEEXPEDITOR}.

\xdef\cntl#1{\scheme{^#1}}

\section{Expression Editor Parameters\label{SECTEXPEDITORPARAMS}}

%----------------------------------------------------------------------------
\noskipentryheader
\formdef{ee-auto-indent}{\categoryglobalparameter}{ee-auto-indent}
\nolistlibraries
\endnoskipentryheader

The value of \scheme{ee-auto-indent} is a boolean value that determines
whether the expression editor indents expressions as they are entered.
Its default value is \scheme{#t}.


%----------------------------------------------------------------------------
\entryheader
\formdef{ee-standard-indent}{\categoryglobalparameter}{ee-standard-indent}
\nolistlibraries
\endentryheader

The value of \scheme{ee-standard-indent} is a nonnegative fixnum
value that determines the amount (in single spaces) by which each
expression is indented relative to the enclosing expression, if
not aligned otherwise by one of the indenter's other heuristics,
when \scheme{ee-auto-indent} is true or when one of the indentation
commands is invoked explicitly.
It's default value is \scheme{2}.


%----------------------------------------------------------------------------
\entryheader
\formdef{ee-auto-paren-balance}{\categoryglobalparameter}{ee-auto-paren-balance}
\nolistlibraries
\endentryheader

The value of \scheme{ee-auto-paren-balance} is a boolean value that determines
whether the expression editor automatically corrects a close
parenthesis or bracket, when typed, to match the corresponding open
parenthesis or bracket, if any.
Its default value is \scheme{#t}.


%----------------------------------------------------------------------------
\entryheader
\formdef{ee-flash-parens}{\categoryglobalparameter}{ee-flash-parens}
\nolistlibraries
\endentryheader

The value of \scheme{ee-flash-parens} is a boolean value that determines
whether the expression editor briefly moves the cursor when an open
or close parenthesis or bracket is typed to the
matching close or open parenthesis or bracket (if any).
Its default value is \scheme{#t}.


%----------------------------------------------------------------------------
\entryheader
\formdef{ee-paren-flash-delay}{\categoryglobalparameter}{ee-paren-flash-delay}
\nolistlibraries
\endentryheader

The value of \scheme{ee-paren-flash-delay} is a nonnegative fixnum
value that determines the amount of time (in milliseconds) that the
expression editor pauses when the cursor is moved to the matching
parenthesis or bracket, if any, when a parenthesis or bracket is
entered.
The value is ignored if the \scheme{ee-flash-parens} is false.
Its default value is \scheme{100}.


%----------------------------------------------------------------------------
\entryheader
\formdef{ee-default-repeat}{\categoryglobalparameter}{ee-default-repeat}
\nolistlibraries
\endentryheader

The value of \scheme{ee-default-repeat} is a nonnegative fixnum
value that determines the number of times the next command is
repeated after the \scheme{ee-command-repeat} editing command
(bound to \scheme{Esc-^U} by default) is used and \emph{not}
followed by a sequence of digits.
It's default value is \scheme{4}.


%----------------------------------------------------------------------------
\entryheader
\formdef{ee-noisy}{\categoryglobalparameter}{ee-noisy}
\nolistlibraries
\endentryheader

The value of \scheme{ee-noisy} is a boolean value that determines
whether the expression editor emits a beep (bell) when an error
occurs, such as an attempt to find the matching delimiter for a
non-delimiter character.
Its default value is \scheme{#f}.


%----------------------------------------------------------------------------
\entryheader
\formdef{ee-history-limit}{\categoryglobalparameter}{ee-history-limit}
\nolistlibraries
\endentryheader

The value of \scheme{ee-history-limit} is a nonnegative fixnum value
that determines the number of history entries retained by the
expression editor during and across sessions.
Only the last \scheme{(ee-history-limit)} entries are retained.
% Its default value is \scheme{100}.


%----------------------------------------------------------------------------
\entryheader
\formdef{ee-common-identifiers}{\categoryglobalparameter}{ee-common-identifiers}
\nolistlibraries
\endentryheader

The value of \scheme{ee-common-identifiers} is list of symbols that
are considered common enough that they should appear early when
one of the incremental identifier-completion editing commands is
invoked.
Its default value contains a few dozen entries.
They are all more than a few characters long (under the theory that
users will most likely type short ones out fully) and all would
appear later than they likely should when incremental
identifier-completion is used.


\section{Key Binding\label{SECTEXPEDITORKEYBINDING}}

Key bindings are established via \scheme{ee-bind-key}.
The default key bindings are described in Section~\ref{SECTEXPEDITOREDITCMDS}.

%----------------------------------------------------------------------------
\entryheader
\formdef{ee-bind-key}{\categoryprocedure}{(ee-bind-key \var{key} \var{procedure})}
\nolistlibraries
\returns unspecified
\endentryheader

The \scheme{ee-bind-key} procedure is used to add to or change the
set of key bindings recognized by the expression editor.

The \var{key} must be a character or string; if it is a string, it
must have the following form.

\begin{grammar}
\ang{key-string}\longis \scheme{"}\ang{key-char}\kplus\scheme{"}
\end{grammar}

where

\begin{grammar}
\ang{key-char}\longis \scheme{\\e} (specifying an escape character)\\
  \orbar \scheme{^\var{x}} (specifying control-\var{x})\\
  \orbar \scheme{\\^} (specifying caret)\\
  \orbar \scheme{\\\\} (specifying back slash)\\
  \orbar \scheme{plain char} (any character other than \scheme{\} or \scheme{^})
\end{grammar}

Note that each double-backslash in the syntax actually denotes just
one backslash in the string.

For example, the \var{key} \scheme{"\\eX"} represents the two-character
sequence Escape-x, i.e., the ``escape'' key followed by the (capital)
``X'' key.
Similarly, they \var{key} \scheme{"\\e^X"} represents the two-character
sequence Escape-Control-x, i.e., the ``escape'' key followed by
Control-X.

Character keys and string keys consisting of a single plain character
always represent a single keystroke.

The \var{procedure} argument should normally be one of the built-in editing
commands described below.
It is also possible to define new editing commands with
\index{\scheme{ee-string-macro}}\scheme{ee-string-macro}
and \index{\scheme{ee-compose}}\scheme{ee-compose}.


\section{Editing Commands\label{SECTEXPEDITOREDITCMDS}}

\xdef\editproc#1#2{\medskip {\bf command:} \index{\scheme{#1}}\scheme{#1}\\
                            {\bf key(s):} #2\par{}}
\xdef\endeditproc{\par{}}

\def\ECgroup#1 {\medskip\noindent{\bf #1}\par}

The editing commands are grouped into sections according to usage.
Each is listed along with the default character sequence or sequences by
which it may be invoked.

\ECgroup{Insertion commands}

\editproc{ee-insert-self}{most printing characters}
Inserts the entered character into the entry.
\endeditproc

\editproc{ee-insert-paren}{\scheme{(}, \scheme{)}, \scheme{[}, \scheme{]}}
Inserts the entered parenthesis or bracket into the entry.

If the parameter
\index{\scheme{ee-auto-paren-balance}}\scheme{ee-auto-paren-balance} is
true, the editor corrects close delimiters if necessary to balance
existing open delimiters, when a matching open delimiter can be found.

If the parameter \index{\scheme{ee-flash-parens}}\scheme{ee-flash-parens}
is true, the editor briefly moves the cursor to the matching delimiter, if
one can be found, pausing for an amount of time controlled by the
parameter \index{\scheme{ee-paren-flash-delay}}\scheme{ee-paren-flash-delay}.
If the matching delimiter is not presently displayed, the cursor is flashed
to the upper-left or lower-left corner of the displayed portion of the
entry, as appropriate.

The behavior of this command is undefined if used for something other
than a parenthesis or bracket.
\endeditproc

\editproc{ee-newline}{none}
Inserts a newline at the cursor position, moves to the next line, and
indents that line if the parameter
\index{\scheme{ee-auto-indent}}\scheme{ee-auto-indent} is true.
Does nothing if the entry is empty.
See also \scheme{ee-newline/accept}.
\endeditproc

\editproc{ee-open-line}{\cntl{O}}
Inserts a newline at the cursor position and indents the next line,
but does not move to the next line.
\endeditproc

\editproc{ee-yank-kill-buffer}{\cntl{Y}}
Inserts the contents of the kill buffer, which is set by the deletion
commands described below.
\endeditproc

\editproc{ee-yank-selection}{\cntl{V}}
Inserts the contents of the window system's current selection or paste
buffer.
When running in a shell window under X Windows, this command requires that
the DISPLAY environment variable be set to the appropriate display.
\endeditproc

% The following is no longer true, since auto-indent is disabled when
% the characters are coming fast enough that they are likely happening
% as the result of a paste operation.
%
% While it may be possible to insert the contents of the current selection
% or paste buffer via some other means, e.g., via the mouse or some
% OS-specific key combination, this command is preferable whenever the
% parameter \index{\scheme{ee-auto-indent}}\scheme{ee-auto-indent} is true
% for multi-line input that is already indented.

\ECgroup{Cursor movement commands}

\editproc{ee-backward-char}{leftarrow, \cntl{B}}
Moves the cursor left one character.
\endeditproc

\editproc{ee-forward-char}{rightarrow, \cntl{F}}
Moves the cursor right one character.
\endeditproc

\editproc{ee-next-line}{downarrow, \cntl{N}}
Moves the cursor down one line (and to the left if necessary so that
the cursor does not sit beyond the last possible position).
If the cursor is at the end of the current entry, and the current
entry has not been modified, this command behaves like
\index{\scheme{ee-history-fwd}}\scheme{ee-history-fwd}.
\endeditproc

\editproc{ee-previous-line}{uparrow, \cntl{P}}
Moves the cursor up one line (and to the left if necessary so that
the cursor does not sit beyond the last possible position).
If the cursor is at the top of the current entry, and the current
entry has not been modified, this command behaves like
\index{\scheme{ee-history-bwd}}\scheme{ee-history-bwd}.
\endeditproc

\editproc{ee-beginning-of-line}{home, \cntl{A}}
Moves the cursor to the first character of the current line.
\endeditproc

\editproc{ee-end-of-line}{end, \cntl{E}}
Moves the cursor to the right of the last character of the current line.
\endeditproc

\editproc{ee-beginning-of-entry}{escape-\scheme{<}}
Moves the cursor to the first character of the entry.
\endeditproc

\editproc{ee-end-of-entry}{escape-\scheme{>}}
Moves the cursor to the right of the last character of the entry.
\endeditproc

\editproc{ee-goto-matching-delimiter}{escape-\scheme{]}}
Moves the cursor to the matching delimiter.
Has no effect if the character under the cursor is not a parenthesis
or bracket or if no matching delimiter can be found.
\endeditproc

\editproc{ee-flash-matching-delimiter}{\cntl{]}}
Moves the cursor briefly to the matching delimiter, if
one can be found, pausing for an amount of time controlled by the
parameter \index{\scheme{ee-paren-flash-delay}}\scheme{ee-paren-flash-delay}.
If the matching delimiter is not presently displayed, the cursor is flashed
to the upper-left or lower-left corner of the displayed portion of the
entry, as appropriate.
\endeditproc

\editproc{ee-exchange-point-and-mark}{\cntl{X}-\cntl{X}}
Moves the cursor to the mark and leaves the mark at the old cursor
position.
(The mark can be set with \scheme{ee-set-mark}.)
\endeditproc

\editproc{ee-forward-sexp}{escape-\cntl{F}}
Moves the cursor to the start of the next expression.
\endeditproc

\editproc{ee-backward-sexp}{escape-\cntl{B}}
Moves the cursor to the start of the preceding expression.
\endeditproc

\editproc{ee-forward-word}{escape-f, escape-\scheme{F}}
Moves the cursor to the end of the next word.
\endeditproc

\editproc{ee-backward-word}{escape-b, escape-\scheme{B}}
Moves the cursor to the start of the preceding word.
\endeditproc

\editproc{ee-forward-page}{pagedown, \cntl{X}-\scheme{]}}
Moves the cursor down one screen page.
\endeditproc

\editproc{ee-backward-page}{pageup, \cntl{X}-\scheme{[}}
Moves the cursor up one screen page.
\endeditproc


\ECgroup{Deletion commands}

\editproc{ee-delete-char}{delete}
Deletes the character under the cursor.

See also \scheme{ee-eof/delete-char}.
\endeditproc

\editproc{ee-backward-delete-char}{backspace (rubout), \cntl{H}}
Deletes the character to the left of the cursor.
\endeditproc

\editproc{ee-delete-line}{\cntl{U}}
Deletes the contents of the current line, leaving behind an empty line.
When used on the first line of a multiline entry of which only the first line
is displayed, i.e., immediately after history movement, \scheme{ee-delete-line}
deletes the contents of the entire entry, like \scheme{ee-delete-entry}
(described below).
\endeditproc

\editproc{ee-delete-to-eol}{\cntl{K}, escape-\scheme{K}}
If the cursor is at the end of a line, joins the line with the next
line, otherwise deletes from the cursor position to the end of the line.
\endeditproc

\editproc{ee-delete-between-point-and-mark}{\cntl{W}}
Deletes text between the current cursor position and the mark.
(The mark can be set with \scheme{ee-set-mark}.)
\endeditproc

\editproc{ee-delete-entry}{\cntl{G}}
Deletes the contents of the current entry.
\endeditproc

\editproc{ee-reset-entry}{\cntl{C}}
Deletes the contents of the current entry and moves to the end of the
history.
\endeditproc

\editproc{ee-delete-sexp}{escape-\cntl{K}, escape-delete}
Deletes the expression that starts under the cursor, or if
no expression starts under the cursor, deletes up to the next
expression.
\endeditproc

\editproc{ee-backward-delete-sexp}{escape-backspace (escape-rubout), escape-\cntl{H}}
Deletes the expression to the left of the cursor.
\endeditproc

\ECgroup{Identifier/filename completion commands}

These commands perform either identifier or filename completion.
Identifier completion is performed outside of a string constant, and filename
completion is performed within a string constant.
(In determining whether the cursor is within a string constant, the
expression editor looks only at the current line and so can be fooled
by string constants that span multiple lines.)

\editproc{ee-id-completion}{none}
Inserts the common prefix of possible completions of the identifier or
filename immediately to the left of the cursor.
Identifier completion is based on the identifiers
defined in the interaction environment.
When there is exactly one possible completion, the common prefix is the
completion.
This command has no effect if no filename or identifier prefix is
immediately the left of the cursor or if the possible completions have
no common prefix.
If run twice in succession, a list of possible completions is displayed.

See also
\index{\scheme{ee-id-completion/indent}}\scheme{ee-id-completion/indent}.
\endeditproc

\editproc{ee-next-id-completion}{\cntl{R}}
Inserts one of the possible completions of the identifier or filename
immediately to the left of the cursor.
Identifier completion is based on the identifiers
defined in the interaction environment.
If run twice or more in succession, this command cycles through all of
the possible completions.
The order is determined by the following heuristics: appearing first
are identifiers whose names appear in the list value of the parameter
\index{\scheme{ee-common-identifiers}}\scheme{ee-common-identifiers};
appearing second are identifiers bound in the interaction environment
but not bound in the scheme-environment (i.e., identifiers defined by
the user), and appearing last are those in the scheme environment.
Within the set of matches appearing in the \scheme{ee-common-identifiers}
list, those listed earliest are shown first; the order is alphabetical
within the other two sets.

See also
\index{\scheme{ee-next-id-completion/indent}}\scheme{ee-next-id-completion/indent}.
\endeditproc

\ECgroup{History movement commands}

\index{\scheme{--eehistory} command-line-option}%
The expression editor maintains a history of entries during each session.
It also saves the history across sessions unless this behavior is
disabled via the command-line argument ``\scheme{--eehistory off}.''

When moving from one history entry to another, only the first line of each
multi-line entry is displayed.
The redisplay command (which \cntl{L} is bound to by default) can be used
to display the entire entry.
It is also possible to move down one line at a time to expose just part of
the rest of the entry.

\editproc{ee-history-bwd}{escape-uparrow, escape-\cntl{P}}
Moves to the preceding history entry
if the current entry is empty or has not been modified;
otherwise, has no effect.

See also \scheme{ee-previous-line}.
\endeditproc

\editproc{ee-history-fwd}{escape-downarrow, escape-\cntl{N}}
Moves to the next history entry
if the current entry is empty or has not been modified;
otherwise, has no effect.

See also \scheme{ee-next-line}.
\endeditproc

\editproc{ee-history-bwd-prefix}{escape-\scheme{p}}
Moves to the closest previous history entry, if any, that starts with
the sequence of characters that makes up the current entry.
May be used multiple times to search for same prefix.
\endeditproc

\editproc{ee-history-fwd-prefix}{escape-\scheme{n}}
Moves to the closest following history entry, if any, that starts with
the sequence of characters that makes up the current entry.
May be used multiple times to search for same prefix.
\endeditproc

\editproc{ee-history-bwd-contains}{escape-\scheme{P}}
Moves to the closest previous history entry, if any, that contains within
it the sequence of characters that makes up the current entry.
May be used multiple times to search for same content.
\endeditproc

\editproc{ee-history-fwd-contains}{escape-\scheme{N}}
Moves to the closest following history entry, if any, that contains within
it the sequence of characters that makes up the current entry.
May be used multiple times to search for same content.
\endeditproc


\ECgroup{Indentation commands}

\editproc{ee-indent}{escape-tab}
Re-indents the current line.

See also \scheme{ee-next-id-completion/indent}.
\endeditproc

\editproc{ee-indent-all}{escape-\scheme{q}, escape-\scheme{Q}, escape-\cntl{Q}}
Re-indents each line of the entire entry.
\endeditproc


\ECgroup{Miscellaneous commands}

\editproc{ee-accept}{\cntl{J}}
Causes the expression editor to invoke the Scheme reader on the contents
of the entry.
If the read is successful, the expression is returned to the waiter;
otherwise, an error message is printed, the entry redisplayed, and the
cursor left (if possible) at the start of the invalid subform.

See also \scheme{ee-newline/accept}.
\endeditproc

\editproc{ee-eof}{none}
Causes end-of-file to be returned from the expression editor,
which in turn causes the waiter to exit.
Ignored unless entry is empty.

See also \scheme{ee-eof/delete-char}.
\endeditproc

\editproc{ee-redisplay}{\cntl{L}}
Redisplays the current expression.
If run twice in succession, clears the screen and redisplays the
expression at the top of the screen.
\endeditproc

\editproc{ee-suspend-process}{\cntl{Z}}
Suspends the current process in shells that support job control.
\endeditproc

\editproc{ee-set-mark}{\cntl{@}, \cntl{}space}
Sets the mark to the current cursor position.
\endeditproc

\editproc{ee-command-repeat}{escape-\cntl{U}}
Repeats the next command $n$ times.
If the next character typed is a digit, $n$ is determined by reading
up the sequence of the digits typed and treating it as a decimal
number.
Otherwise, $n$ is the value of the parameter
\index{\scheme{ee-default-repeat}}\scheme{ee-default-repeat}.
\endeditproc


\ECgroup{Combination commands}

\editproc{ee-newline/accept}{enter, \cntl{M}}
Behaves like \scheme{ee-accept} if run at the end (not including
whitespace) of an entry that starts with a balanced expression;
otherwise, behaves like \scheme{ee-newline}.
\endeditproc

\editproc{ee-id-completion/indent}{tab}
Behaves like \scheme{ee-id-completion} if an identifier (outside
a string constant) or filename (within a string constant) appears
just to the left of the cursor and the last character of that
identifier or filename was just entered;
otherwise, behaves like \scheme{ee-indent}.

If an existing identifier or filename, i.e., not one just typed, appears to the left
of the cursor, the first use of this command behaves like
\scheme{ee-newline}, the second consecutive use behaves like
\scheme{ee-id-completion}, and the third behaves like a second consecutive
use of \scheme{ee-id-completion}.
\endeditproc

\editproc{ee-next-id-completion/indent}{none}
Behaves like \scheme{ee-next-id-completion} if an identifier (outside
a string constant) or filename (within a string constant)
appears just to the left of the cursor and the last character of that
identifier or identifier was just entered;
otherwise, behaves like \scheme{ee-indent}.
\endeditproc

\editproc{ee-eof/delete-char}{\cntl{D}}
Behaves like \scheme{ee-delete-char} if the entry is nonempty;
otherwise, behaves like \scheme{ee-eof}.
If the entry is nonempty and this command is run twice or more in
succession, it does nothing once the entry becomes empty.
This is to prevent accidental exit from the waiter in cases where
the command is run repeatedly (perhaps with the help of a keyboard's
auto-repeat feature) to delete all of the characters in an entry.
\endeditproc

\section{Creating New Editing Commands\label{SECTEXPEDITORNEWCMDS}}

%----------------------------------------------------------------------------
\noskipentryheader
\formdef{ee-string-macro}{\categoryprocedure}{(ee-string-macro \var{string})}
\nolistlibraries
\returns a new editing command
\endnoskipentryheader

The new editing command produced inserts \var{string} before the current
cursor position.

Two string macros are predefined:

\begin{tabular}{ll}
\scheme{(ee-string-macro "(define ")} & ~~escape-d\\
\scheme{(ee-string-macro "(lambda ")} & ~~escape-l
\end{tabular}

%----------------------------------------------------------------------------
\entryheader
\formdef{ee-compose}{\categoryprocedure}{(ee-compose \var{ecmd} \dots)}
\nolistlibraries
\returns a new editing command
\endentryheader

Each \var{ecmd} must be an editing command.

The new editing command runs each of the editing commands
\scheme{\var{ecmd} \dots} in sequence.

% this is of limited utility until we provide some
% inspection procedures, like point-pos and end-of-line?

For example, the following expression binds \cntl{X}-p to an editing
command that behaves like \scheme{ee-history-bwd-prefix} but leaves the
cursor at the end of the expression rather than at the end of the first
line, causing the entire entry to be displayed.

\schemedisplay
(let ()
  (import expression-editor)
  (ee-bind-key "^Xp"
    (ee-compose ee-history-bwd ee-end-of-entry)))
\endschemedisplay

A command such as \scheme{ee-id-completion} that performs a different
action when run twice in succession will not recognize that it has been
run twice in succession if run as part of a composite command.