🌸 A command-line fuzzy finder
Go to file
Junegunn Choi 20915529b7 Add link to examples wiki page 2014-05-02 23:38:36 +09:00
ext Fix gemspec 2014-03-14 18:25:55 +09:00
plugin [Vim] Allow vertical split of tmux window 2014-04-12 19:53:39 +09:00
test Fix test code indentation 2014-04-03 14:53:47 +09:00
.gitignore Add .DS_Store to .gitignore 2014-03-19 13:14:20 +01:00
README.md Add link to examples wiki page 2014-05-02 23:38:36 +09:00
Rakefile Extended mode 2013-11-15 20:40:57 +09:00
fzf Fix ranking when multiple regions overlap 2014-04-03 14:51:01 +09:00
fzf-completion.bash Make host name completion require trigger sequence (#13) 2013-12-23 23:16:07 +09:00
fzf-completion.zsh Implement bash auto-completion with fzf 2013-11-20 01:29:36 +09:00
fzf.gemspec 0.8.3 2014-04-05 12:56:10 +09:00
install [fish] Remove temporary file after use 2014-05-02 16:35:36 +09:00

README.md

fzf - Fuzzy finder for your shell

fzf is a general-purpose fuzzy finder for your shell.

It was heavily inspired by ctrlp.vim and the likes.

Requirements

fzf requires Ruby (>= 1.8.5).

Installation

Clone this repository and run install script.

git clone https://github.com/junegunn/fzf.git ~/.fzf
~/.fzf/install

In case you don't have git installed:

mkdir -p ~/.fzf
curl -L https://github.com/junegunn/fzf/archive/master.tar.gz |
    tar xz --strip-components 1 -C ~/.fzf
~/.fzf/install

The script will setup:

  • fzf function (bash, zsh, fish)
  • Key bindings (CTRL-T, CTRL-R, and ALT-C) (bash, zsh, fish)
  • Fuzzy auto-completion (bash)

If you don't use any of the aforementioned shells, you have to manually place fzf executable in a directory included in $PATH. Key bindings and auto-completion will not be available in that case.

Install as Vim plugin

Once you have cloned the repository, add the following line to your .vimrc.

set rtp+=~/.fzf

Or you may use any Vim plugin manager, such as vim-plug.

Usage

usage: fzf [options]

  Search
    -x, --extended       Extended-search mode
    -e, --extended-exact Extended-search mode (exact match)
    -i                   Case-insensitive match (default: smart-case match)
    +i                   Case-sensitive match
    -n, --nth=[-]N[,..]  Comma-separated list of field indexes for limiting
                         search scope (positive or negative integers)
    -d, --delimiter=STR  Field delimiter regex for --nth (default: AWK-style)

  Search result
    -s, --sort=MAX       Maximum number of matched items to sort (default: 1000)
    +s, --no-sort        Do not sort the result. Keep the sequence unchanged.

  Interface
    -m, --multi          Enable multi-select with tab/shift-tab
        --no-mouse       Disable mouse
    +c, --no-color       Disable colors
    +2, --no-256         Disable 256-color
        --black          Use black background

  Scripting
    -q, --query=STR      Start the finder with the given query
    -1, --select-1       Automatically select the only match
    -0, --exit-0         Exit immediately when there's no match
    -f, --filter=STR     Filter mode. Do not start interactive finder.

  Environment variables
    FZF_DEFAULT_COMMAND  Default command to use when input is tty
    FZF_DEFAULT_OPTS     Defaults options. (e.g. "-x -m --sort 10000")

fzf will launch curses-based finder, read the list from STDIN, and write the selected item to STDOUT.

find * -type f | fzf > selected

Without STDIN pipe, fzf will use find command to fetch the list of files excluding hidden ones. (You can override the default command with FZF_DEFAULT_COMMAND)

vim $(fzf)

If you want to preserve the exact sequence of the input, provide --no-sort (or +s) option.

history | fzf +s

Keys

Use CTRL-J and CTRL-K (or CTRL-N and CTRL-P) to change the selection, press enter key to select the item. CTRL-C, CTRL-G, or ESC will terminate the finder.

The following readline key bindings should also work as expected.

  • CTRL-A / CTRL-E
  • CTRL-B / CTRL-F
  • CTRL-W / CTRL-U / CTRL-Y
  • ALT-B / ALT-F

If you enable multi-select mode with -m option, you can select multiple items with TAB or Shift-TAB key.

You can also use mouse. Double-click on an item to select it or shift-click (or ctrl-click) to select multiple items. Use mouse wheel to move the cursor up and down.

Extended-search mode

With -x or --extended option, fzf will start in "extended-search mode".

In this mode, you can specify multiple patterns delimited by spaces, such as: ^music .mp3$ sbtrkt !rmx

Token Description Match type
^music Items that start with music prefix-exact-match
.mp3$ Items that end with .mp3 suffix-exact-match
sbtrkt Items that match sbtrkt fuzzy-match
!rmx Items that do not match rmx inverse-fuzzy-match
'wild Items that include wild exact-match (quoted)
!'fire Items that do not include fire inverse-exact-match

If you don't need fuzzy matching and do not wish to "quote" every word, start fzf with -e or --extended-exact option.

Useful examples

# fe [FUZZY PATTERN] - Open the selected file with the default editor
#   - Bypass fuzzy finder if there's only one match (--select-1)
#   - Exit if there's no match (--exit-0)
fe() {
  local file
  file=$(fzf --query="$1" --select-1 --exit-0)
  [ -n "$file" ] && ${EDITOR:-vim} "$file"
}

# fd - cd to selected directory
fd() {
  local dir
  dir=$(find ${1:-*} -path '*/\.*' -prune \
                  -o -type d -print 2> /dev/null | fzf +m) &&
  cd "$dir"
}

# fda - including hidden directories
fda() {
  local dir
  dir=$(find ${1:-.} -type d 2> /dev/null | fzf +m) && cd "$dir"
}

# fh - repeat history
fh() {
  eval $(([ -n "$ZSH_NAME" ] && fc -l 1 || history) | fzf +s | sed 's/ *[0-9]* *//')
}

# fkill - kill process
fkill() {
  ps -ef | sed 1d | fzf -m | awk '{print $2}' | xargs kill -${1:-9}
}

# fbr - checkout git branch
fbr() {
  local branches branch
  branches=$(git branch) &&
  branch=$(echo "$branches" | fzf +s +m) &&
  git checkout $(echo "$branch" | sed "s/.* //")
}

# fco - checkout git commit
fco() {
  local commits commit
  commits=$(git log --pretty=oneline --abbrev-commit --reverse) &&
  commit=$(echo "$commits" | fzf +s +m -e) &&
  git checkout $(echo "$commit" | sed "s/ .*//")
}

# ftags - search ctags
ftags() {
  local line
  [ -e tags ] &&
  line=$(
    awk 'BEGIN { FS="\t" } !/^!/ {print toupper($4)"\t"$1"\t"$2"\t"$3}' tags |
    cut -c1-80 | fzf --nth=1,2
  ) && $EDITOR $(cut -f3 <<< "$line") -c "set nocst" \
                                      -c "silent tag $(cut -f2 <<< "$line")"
}

For more examples, see the wiki page.

Key bindings for command line

The install script will setup the following key bindings.

bash/zsh

  • CTRL-T - Paste the selected file path(s) into the command line
  • CTRL-R - Paste the selected command from history into the command line
  • ALT-C - cd into the selected directory

If you're on a tmux session, CTRL-T will launch fzf in a new split-window. You may disable this tmux integration by setting FZF_TMUX to 0, or change the height of the window with FZF_TMUX_HEIGHT (e.g. 20, 50%).

The source code can be found in ~/.fzf.bash and in ~/.fzf.zsh.

Auto-completion

Disclaimer: Auto-completion feature is currently experimental, it can change over time

bash

Files and directories

Fuzzy completion for files and directories can be triggered if the word before the cursor ends with the trigger sequence which is by default **.

  • COMMAND [DIRECTORY/][FUZZY_PATTERN]**<TAB>
# Files under current directory
# - You can select multiple items with TAB key
vim **<TAB>

# Files under parent directory
vim ../**<TAB>

# Files under parent directory that match `fzf`
vim ../fzf**<TAB>

# Files under your home directory
vim ~/**<TAB>


# Directories under current directory (single-selection)
cd **<TAB>

# Directories under ~/github that match `fzf`
cd ~/github/fzf**<TAB>

Process IDs

Fuzzy completion for PIDs is provided for kill command. In this case there is no trigger sequence, just press tab key after kill command.

# Can select multiple processes with <TAB> or <Shift-TAB> keys
kill -9 <TAB>

Host names

For ssh and telnet commands, fuzzy completion for host names is provided. The names are extracted from /etc/hosts and ~/.ssh/config.

ssh **<TAB>
telnet **<TAB>

Settings

# Use ~~ as the trigger sequence instead of the default **
export FZF_COMPLETION_TRIGGER='~~'

# Options to fzf command
export FZF_COMPLETION_OPTS='+c -x'

zsh

TODO 😃

(Pull requests are appreciated.)

Usage as Vim plugin

(fzf is a command-line utility, naturally it is only accessible in terminal Vim)

:FZF[!]

If you have set up fzf for Vim, :FZF command will be added.

" Look for files under current directory
:FZF

" Look for files under your home directory
:FZF ~

" With options
:FZF --no-sort -m /tmp

Note that the environment variables FZF_DEFAULT_COMMAND and FZF_DEFAULT_OPTS also apply here.

If you're on a tmux session, :FZF will launch fzf in a new split-window whose height can be adjusted with g:fzf_tmux_height (default: '40%'). However, the bang version (:FZF!) will always start in fullscreen.

fzf#run([options])

For more advanced uses, you can call fzf#run() function which returns the list of the selected items.

fzf#run() may take an options-dictionary:

Option name Type Description
source string External command to generate input to fzf (e.g. find .)
source list Vim list as input to fzf
sink string Vim command to handle the selected item (e.g. e, tabe)
sink funcref Reference to function to process each selected item
options string Options to fzf
dir string Working directory
tmux_width number/string Use tmux vertical split with the given height (e.g. 20, 50%)
tmux_height number/string Use tmux horizontal split with the given height (e.g. 20, 50%)

Examples

If sink option is not given, fzf#run will simply return the list.

let items = fzf#run({ 'options': '-m +c', 'dir': '~', 'source': 'ls' })

But if sink is given as a string, the command will be executed for each selected item.

" Each selected item will be opened in a new tab
let items = fzf#run({ 'sink': 'tabe', 'options': '-m +c', 'dir': '~', 'source': 'ls' })

We can also use a Vim list as the source as follows:

" Choose a color scheme with fzf
nnoremap <silent> <Leader>C :call fzf#run({
\   'source':
\     map(split(globpath(&rtp, "colors/*.vim"), "\n"),
\         "substitute(fnamemodify(v:val, ':t'), '\\..\\{-}$', '', '')"),
\   'sink':       'colo',
\   'options':    '+m',
\   'tmux_width': 20
\ })<CR>

sink option can be a function reference. The following example creates a handy mapping that selects an open buffer.

" List of buffers
function! g:buflist()
  redir => ls
  silent ls
  redir END
  return split(ls, '\n')
endfunction

function! g:bufopen(e)
  execute 'buffer '. matchstr(a:e, '^[ 0-9]*')
endfunction

nnoremap <silent> <Leader><Enter> :call fzf#run({
\   'source':      reverse(g:buflist()),
\   'sink':        function('g:bufopen'),
\   'options':     '+m',
\   'tmux_height': '40%'
\ })<CR>

Articles

Tips

Rendering issues

If you have any rendering issues, check the followings:

  1. Make sure $TERM is correctly set. fzf will use 256-color only if it contains 256 (e.g. xterm-256color)
  2. If you're on screen or tmux, $TERM should be either screen or screen-256color
  3. Some terminal emulators (e.g. mintty) have problem displaying default background color and make some text unable to read. In that case, try --black option. And if it solves your problem, I recommend including it in FZF_DEFAULT_OPTS for further convenience.
  4. If you still have problem, try --no-256 option or even --no-color.
  5. Ruby 1.9 or above is required for correctly displaying unicode characters.

Ranking algorithm

fzf sorts the result first by the length of the matched substring, then by the length of the whole string. However it only does so when the number of matches is less than the limit which is by default 1000, in order to avoid the cost of sorting a large list and limit the response time of the query.

This limit can be adjusted with -s option, or with the environment variable FZF_DEFAULT_OPTS.

export FZF_DEFAULT_OPTS="--sort 20000"

Respecting .gitignore, .hgignore, and svn:ignore

ag or pt will do the filtering:

# Feed the output of ag into fzf
ag -l -g "" | fzf

# Setting ag as the default source for fzf
export FZF_DEFAULT_COMMAND='ag -l -g ""'

# Now fzf (w/o pipe) will use ag instead of find
fzf

git ls-tree for fast traversal

If you're running fzf in a large git repository, git ls-tree can boost up the speed of the traversal.

# Copy the original fzf function to __fzf
declare -f __fzf > /dev/null ||
  eval "$(echo "__fzf() {"; declare -f fzf | grep -v '^{' | tail -n +2)"

# Use git ls-tree when possible
fzf() {
  if [ -n "$(git rev-parse HEAD 2> /dev/null)" ]; then
    FZF_DEFAULT_COMMAND="git ls-tree -r --name-only HEAD" __fzf "$@"
  else
    __fzf "$@"
  fi
}

Fish shell

It's a known bug of fish that it doesn't allow reading from STDIN in command substitution, which means simple vim (fzf) won't work as expected. The workaround is to store the result of fzf to a temporary file.

function vimf
  if fzf > $TMPDIR/fzf.result
    vim (cat $TMPDIR/fzf.result)
  end
end

function fe
  set tmp $TMPDIR/fzf.result
  fzf --query="$argv[1]" --select-1 --exit-0 > $tmp
  if [ (cat $tmp | wc -l) -gt 0 ]
    vim (cat $tmp)
  end
end

Windows

fzf works on Cygwin and MSYS2. You may need to use --black option on MSYS2 to avoid rendering issues.

License

MIT

Author

Junegunn Choi