Mapping fast keycodes in terminal Vim

Tip 1272 Printable Monobook Previous Next

created 2006 · complexity intermediate · author Gerald Lai · version 6.0

---

In the GUI version of Vim (gvim), the mapping of keys seem to work for the most part. For instance, it is easy to map Ctrl-Shift-F2 to a keystroke:

"delete all lines in the current buffer
:nmap <C-S-F2> ggdG

For terminal versions of Vim (such as xterm, rxvt, win32, etc), mapping something like Ctrl-Shift-F2 needs some extra work. It may seem daunting to deal with archaic terminal keycodes at first, but once you understand what is going on in the right context, it is quite simple.

1a. Keycodes

There are two types of keycodes: terminal keycodes and Vim keycodes. Terminal keycodes look something like

^[[1;2A

These are actual bytes that are sent to Vim by the terminal when we type Shift-Up, for example (on an xterm).

Vim keycodes look like

<S-Up>

Vim needs its own representation of keycodes because it runs on a variety of platforms. Vim can act upon its own keycodes, and leave the assignment of terminal keycodes to autodetection (or to manual setup by the user, which is what this guide is for).

1b. List of usable keycodes

To obtain a list of terminal keycodes, look it up in the manual for that terminal.

For a list of Vim's internal keycodes, see :help t_ku. Enter :set termcap to list the terminal keycodes (see :help terminal-options).

In addition to the Vim keycodes listed in the help section above, the keycodes for the function keys actually go up to F37. This means, we have

<F1> to <F37>
<S-F1> to <S-F37>

We can use this to our advantage to create more "responsive" mappings. More on this later in section 2b.

We can also set these Vim keycodes:

<C-Home>, <C-End>
<S-Home>, <S-End>
<S-a> to <S-z>
<C-a> to <C-z>
<A-a> to <A-z>

Be careful when setting these keycodes. See section 3a.

1c. Setting Vim keycodes

Vim keycodes look a lot like the keystrokes we would define in a mapping. They are easy to read, quite intuitive, and are almost treated the same by Vim. The only difference is that we can :set internal keycodes. For example, to manually set the keycode for Shift-Down_arrow:

:set <S-Down>=^[[1;2B

where ^[ is a literal ESC special character.

The following is not allowed because the keycode is not a listed Vim keycode:

:set <C-S-Down>=^[[1;6B

Once the correct terminal keycode is assigned to the appropriate Vim keycode, the keystroke should work in terminal Vim. Shift-Down_arrow should now work, but the Ctrl-Shift-Down_arrow keycode can not be set in the same way. For this we need mappings (see section 2b).

1d. Entering literal characters & terminal codes

Either in the :command line, or in insert mode:

To enter a literal character in Vim, first type Ctrl-v, followed by a single keystroke. Hence, typing Ctrl-v then Esc will produce "^[".

To enter a terminal keycode in Vim, first type Ctrl-v, followed by the keystroke we want to obtain the term keycode from. Hence, typing Ctrl-v + Shift-Down_arrow will produce "^[[1;2B".

This means that there are 2 ways to set the Vim keycode as shown in section 1c:

"literally enter the keycode
:set <S-Down>={C-v}{Esc}[1;2B

or

"let the terminal send its keycode to Vim
:set <S-Down>={C-v}{S-Down}

where curly brackets {..} denote the action of typing the keystroke. For instance, when you see {C-v}, hold down Ctrl and hit "v". Do not type it literally as "{C-v}".

We shall use the curly brackets as nomenclature from now on.

Also make use of the Normal mode command 'ga' to inspect the resulting keycode. Place the cursor over a special character, and type 'ga'. This will provide you with ascii information of the character under the cursor.

See :help i_ctrl-v, :help ga.

2a. Mappings

The function of mappings in Vim is to map a set of keystrokes to another set of keystrokes. For more information, see :help key-mapping, :help map-modes.

2b. Mapping "fast" keycodes

As mentioned in section 1c, the Ctrl-Shift-Down_arrow keycode cannot be set in the same way as the Shift-Down_arrow keycode because it is not a listed Vim keycode. One way to get around that is to do the following:

:map <Esc>[1;6B <C-S-Down>

This maps the literal terminal keycode to a Vim keystroke (remember, not a Vim keycode).

The disadvantage of this is that Vim will "wait" everytime after the Esc key pressed for a potential "[" + "1" + ";" + "6" + "B" keystroke combination, just in case you meant to do <C-S-Down> instead of a simple <Esc> to Normal mode.

If you typed "{Esc}[1;6B" fast enough, you would do a Ctrl-Shift-Down_arrow. After all, you are mapping a keystroke, not a keycode.

A better way of mapping keycodes is to first assign the terminal keycode to an unused Vim keycode, and then map the newly used Vim keycode. This way, we can set 'ttimeoutlen' to a small value to ensure that the terminal keycode can only be entered into Vim as fast as the terminal can, but will be humanly-impossible to do it manually.

This is how it is done:

:set timeout timeoutlen=1000 ttimeoutlen=100
:set <F13>=^[[1;6B
:map <F13> <C-S-Down>
:map! <F13> <C-S-Down>

Now, Ctrl-Shift-Down_arrow will work, and Esc will not pause.

Potentially unused Vim keycodes that can be used include:

<F13> to <F37>
<S-F13> to <S-F37>
<xF1> to <xF4>
<S-xF1> to <S-xF4>

When setting any Vim keycode, exercise caution. Do some checks to see if the terminal code you're setting the Vim keycode to is set elsewhere. There must be no conflicts.

For instance, in a Win32 terminal, <F1> to <F4> will always be set the same as <xF1> to <xF4>, respectively. Therefore, you cannot use those keycodes.

See :help 'timeout', :help 'timeoutlen', :help 'ttimeoutlen'.

2c. Laying it out in vimrc

Here is an example of how to set up the keycodes in a vimrc file. It is highly recommended that you start Vim without any startup scripts that may interfere with keycode detection:

$ vim -Nu NONE .vimrc

set timeout timeoutlen=1000 ttimeoutlen=100
if !has("gui_running")
  if &term == "xterm"
    set <Home>=^[[H <End>=^[[F <BS>=^?
    set <S-Up>=^[[1;2A <S-Down>=^[[1;2B <S-Right>=^[[1;2C <S-Left>=^[[1;2D
    set <xF1>={C-v}{C-S-Up}^[[1;6A <xF2>=^[[1;6B <xF3>=^[[1;6C <xF4>=^[[1;6D
    map <xF1> <C-S-Up>
    map <xF2> <C-S-Down>
    map <xF3> <C-S-Right>
    map <xF4> <C-S-Left>
    map! <xF1> <C-S-Up>
    map! <xF2> <C-S-Down>
    map! <xF3> <C-S-Right>
    map! <xF4> <C-S-Left>
  elseif &term == "win32"
    ...
  else
    ...
  endif
endif

where ^[ is a literal ESC special character and ^? is a literal ascii 0x7f character.

See :help 'term', :help has-patch.

You can also use a function like the following to automatically assign fast keycodes:

" MapFastKeycode: helper for fast keycode mappings
" makes use of unused vim keycodes <[S-]F15> to <[S-]F37>
function! <SID>MapFastKeycode(key, keycode)
    if s:fast_i == 46
        echohl WarningMsg
        echomsg "Unable to map ".a:key.": out of spare keycodes"
        echohl None
        return
    endif
    let vkeycode = '<'.(s:fast_i/23==0 ? '' : 'S-').'F'.(15+s:fast_i%23).'>'
    exec 'set '.vkeycode.'='.a:keycode
    exec 'map '.vkeycode.' '.a:key
    let s:fast_i += 1
endfunction
let s:fast_i = 0

It uses Vim keycodes <F15> and above. If you don't have keys for <F13> and <F14> on your keyboard, you can also make use of those, for a total of 50 keys you can map this way.

Once defined, you can use the function to map keys like so: (example for rxvt)

call <SID>MapFastKeycode('<C-F3>', "\e[13^")
call <SID>MapFastKeycode('<C-F4>', "\e[14^")
call <SID>MapFastKeycode('<M-j>', "\ej")
call <SID>MapFastKeycode('<M-k>', "\ek")
" ...

3a. Troubleshooting I

You only need to employ unused Vim keycodes either when
(a) there are no available Vim keycodes that match the terminal keycode, and
(b) the terminal keycode is longer than a single keystroke
or
(c) setting the Vim keycode that matches the terminal keycode causes weird behavior.

Example where (a) is not true:

:set <F13>=^[[1;2B
:map <F13> <S-Down>

Setting the unused <F13> is redundant, because <S-Down> is already available as a Vim keycode. Do this instead:

:set <S-Down>=^[[1;2B

Example where (b) is not true:

For a Win32 terminal, mapping Ctrl-Backspace is done by

:map <C-{C-v}{BS}> <C-BS>

Since the terminal keycode can be represented by <C-^?>, which is a single keystroke, there is no need to employ an unused Vim keycode.

Example where (c) is true:

For an rxvt terminal, merely doing

:set <A-j>=^[j <A-u>=^[u <A-6>=^[6

will not make Alt-j, Alt-u and Alt-6 work. Reason unknown. Need to assign to unused Vim keycodes.

3b. Troubleshooting II

Sometimes, performing

:set <xF1>={C-v}{A-S-F1}

produces

:set <xF1>=^[<S-F1>

where ^[ is a literal ESC special character and "<S-F1>" is literal text.

This means that you may have already set Vim's keycode <S-F1>, and that it was part of the Alt-Shift-F1 terminal keycode that was sent to Vim. Vim processes the entire terminal keycode and noticed that it could replace part of it with its own internal representation.

To fix this and enter the full terminal keycode, backspace up to "<" and do {C-v}{S-F1}. The full action would be:

:set <xF1>={C-v}{A-S-F1}{BS 6 times}{C-v}{S-F1}

to produce

:set <xF1>=^[^[[23~

To give some perspective, here are the actual terminal keycodes for rxvt:

<S-F1> = ^[[23~
<A-S-F1> = ^[^[[23~