runbuf.vim makes it easy to run the contents of a buffer in psql, python, bash, etc. and display the output.

This little demo video is probably clearer than any text description.

The original use case I wrote this for is for running SQL queries with PostgreSQL; you can use \e in psql to load $EDITOR, but it's a bit cumbersome. After running :Runbuf psql dbname you can write your queries and send it to psql with <C-s>. This will load the results in a split and keeps the query you've written.

I realized that this will also work well for python or other shells that accept stdin, so I renamed it from sql.vim to runbuf.vim :-) It's a pretty simple plugin, but sped up my PostgreSQL testing workflow by quite a bit.

Usage

Start it with :Runbuf <cmd>, for example :Runbuf psql mydb or :Runbuf python. This will load a file in ~/.cache/runbuf, which is the same file for every command. This is useful since Vim's undo history will give you a kind of edit history.

The file directory is determined with g:runbuf_dir, and the filename's extension based on the first word in the command from the g:runbuf_extensions dict.

If the command contains a %s then this will be replaced with the path of the input file; for most things you don't need this as many interpreters will read from stdin, but some compilers (e.g. go run) don't.

Hit <C-s> in normal or insert mode to run the command and get the output.

Note: many terminals eat <C-s> and will stop the terminal output (<C-q> to resume); you probably want to disable this in your shell config (setopt noflowcontrol in zsh; or stty -ixon quit undef).

Some examples:

:Runbuf psql dbname
:Runbuf python
:Runbuf sqlite3 file.sqlite

" Any shell command will do, so you can pass output options and the like.
:Runbuf psql --expanded -h somehost.example.com dbname

Settings

All settings with their defaults values.

" Key to map for running to command, mapped in normal and insert mode.
let g:runbuf_map = '<C-s>'

" Shortcuts for commands; useful if you have longer connection strings or
" the like.
let g:runbuf_commands = #{
    \ go: 'go run %s',
\ }

" List of commands to run when creating the output window.
let g:runbuf_output = ['below new', 'resize 15', 'setl nowrap']

" Directory to place the input files. Will be created.
let g:runbuf_dir = expand($XDG_CACHE_HOME isnot# '' ? $XDG_CACHE_HOME : '~/.cache/runbuf')

" Extensions for the input files, so filetypes work correctly and your edit
" history for Python and psql are separate.
" The filename will be <command>.<ext>
let g:runbuf_extensions = #{
    \ psql:    'sql',
    \ sqlite3: 'sql',
    \ mysql:   'sql',
    \ go:      'go',
    \ python:  'py',
    \ ruby:    'rb'
\ }

" Resize output buffer.
let g:runbuf_resize = get(g:, 'runbuf_resize', 1)

Protips

You can scroll the output buffer without leaving the input window with some additional mappings, for example to have <C-j> and <C-k> scroll by 3 lines:

augroup my-runbuf
    au!
    au Filetype *.runbuf
        \  nnoremap <buffer> <C-j> :echo win_execute(win_getid(bufnr(b:output)), "normal! 3\<lt>C-e>")<CR>
        \| nnoremap <buffer> <C-k> :echo win_execute(win_getid(bufnr(b:output)), "normal! 3\<lt>C-y>")<CR>
augroup end

.runbuf is appended to the filetype of the input buffer, so you can hook in other things here as well.

---

Undo works in the output window, in case you want to view the previous output. The last output will be restored if you leave the window.

---

You can use +Runbuf to start things a bit faster; for example:

$ vim '+Runbuf gc'

Where gc is a shortcut to connect to my PostgreSQL database defined in g:runbuf_commands.

---

Some PostgreSQL specific-things:

• If you're using PostgreSQL performance testing on indexes and the like then you can use something like:

  begin;
      drop index maybe_bad;
      create index maybe_better [..];
  
      explain analyze
          select [..]
  rollback;
  

  This won't work for MariaDB (as for as I know), but PostgreSQL is cool with it, and it's an easy way to test out some stuff while keeping your base tables intact.

• Since stuff is run through the psql CLI, special commands such as \x and \d will work just fine. You can also configure psql from ~/.psqlrc this is what I have:

  \set QUIET
  \pset linestyle unicode
  \pset footer off
  \pset null 'NULL'
  \timing on
  

  There are also flags for psql to set this (and other things). See psql(1) for docs on all of this.