[中文] [한국어]

fd is a program to find entries in your filesystem. It is a simple, fast and user-friendly alternative to find. While it does not aim to support all of find's powerful functionality, it provides sensible (opinionated) defaults for a majority of use cases.

Quick links:

• How to use
• Installation
• Troubleshooting

• Intuitive syntax: fd PATTERN instead of find -iname '*PATTERN*'.
• Regular expression (default) and glob-based patterns.
• Very fast due to parallelized directory traversal.
• Uses colors to highlight different file types (same as ls).
• Supports parallel command execution
• Smart case: the search is case-insensitive by default. It switches to case-sensitive if the pattern contains an uppercase character*.
• Ignores hidden directories and files, by default.
• Ignores patterns from your .gitignore, by default.
• The command name is 50% shorter* than find :-).

First, to get an overview of all available command line options, you can either run fd -h for a concise help message or fd --help for a more detailed version.

fd is designed to find entries in your filesystem. The most basic search you can perform is to run fd with a single argument: the search pattern. For example, assume that you want to find an old script of yours (the name included netflix):

> fd netfl
Software/python/imdb-ratings/netflix-details.py

If called with just a single argument like this, fd searches the current directory recursively for any entries that contain the pattern netfl.

The search pattern is treated as a regular expression. Here, we search for entries that start with x and end with rc:

> cd /etc
> fd '^x.*rc$'
X11/xinit/xinitrc
X11/xinit/xserverrc

The regular expression syntax used by fd is documented here.

If we want to search a specific directory, it can be given as a second argument to fd:

> fd passwd /etc
/etc/default/passwd
/etc/pam.d/passwd
/etc/passwd

fd can be called with no arguments. This is very useful to get a quick overview of all entries in the current directory, recursively (similar to ls -R):

> cd fd/tests
> fd
testenv
testenv/mod.rs
tests.rs

If you want to use this functionality to list all files in a given directory, you have to use a catch-all pattern such as . or ^:

> fd . fd/tests/
testenv
testenv/mod.rs
tests.rs

Often, we are interested in all files of a particular type. This can be done with the -e (or --extension) option. Here, we search for all Markdown files in the fd repository:

> cd fd
> fd -e md
CONTRIBUTING.md
README.md

The -e option can be used in combination with a search pattern:

> fd -e rs mod
src/fshelper/mod.rs
src/lscolors/mod.rs
tests/testenv/mod.rs

To find files with exactly the provided search pattern, use the -g (or --glob) option:

> fd -g libc.so /usr
/usr/lib32/libc.so
/usr/lib/libc.so

By default, fd does not search hidden directories and does not show hidden files in the search results. To disable this behavior, we can use the -H (or --hidden) option:

> fd pre-commit
> fd -H pre-commit
.git/hooks/pre-commit.sample

If we work in a directory that is a Git repository (or includes Git repositories), fd does not search folders (and does not show files) that match one of the .gitignore patterns. To disable this behavior, we can use the -I (or --no-ignore) option:

> fd num_cpu
> fd -I num_cpu
target/debug/deps/libnum_cpus-f5ce7ef99006aa05.rlib

To really search all files and directories, simply combine the hidden and ignore features to show everything (-HI).

By default, fd only matches the filename of each file. However, using the --full-path or -p option, you can match against the full path.

> fd -p -g '**/.git/config'
> fd -p '.*/lesson-\d+/[a-z]+.(jpg|png)'

Instead of just showing the search results, you often want to do something with them. fd provides two ways to execute external commands for each of your search results:

• The -x/--exec option runs an external command for each of the search results (in parallel).
• The -X/--exec-batch option launches the external command once, with all search results as arguments.

Recursively find all zip archives and unpack them:

fd -e zip -x unzip

If there are two such files, file1.zip and backup/file2.zip, this would execute unzip file1.zip and unzip backup/file2.zip. The two unzip processes run in parallel (if the files are found fast enough).

Find all *.h and *.cpp files and auto-format them inplace with clang-format -i:

fd -e h -e cpp -x clang-format -i

Note how the -i option to clang-format can be passed as a separate argument. This is why we put the -x option last.

Find all test_*.py files and open them in your favorite editor:

fd -g 'test_*.py' -X vim

Note that we use capital -X here to open a single vim instance. If there are two such files, test_basic.py and lib/test_advanced.py, this will run vim test_basic.py lib/test_advanced.py.

To see details like file permissions, owners, file sizes etc., you can tell fd to show them by running ls for each result:

fd … -X ls -lhd --color=always

This pattern is so useful that fd provides a shortcut. You can use the -l/--list-details option to execute ls in this way: fd … -l.

The -X option is also useful when combining fd with ripgrep (rg) in order to search within a certain class of files, like all C++ source files:

fd -e cpp -e cxx -e h -e hpp -X rg 'std::cout'

Convert all *.jpg files to *.png files:

fd -e jpg -x convert {} {.}.png

Here, {} is a placeholder for the search result. {.} is the same, without the file extension. See below for more details on the placeholder syntax.

The terminal output of commands run from parallel threads using -x will not be interlaced or garbled, so fd -x can be used to rudimentarily parallelize a task run over many files. An example of this is calculating the checksum of each individual file within a directory.

fd -tf -x md5sum > file_checksums.txt

The -x and -X options take a command template as a series of arguments (instead of a single string). If you want to add additional options to fd after the command template, you can terminate it with a \;.

The syntax for generating commands is similar to that of GNU Parallel:

• {}: A placeholder token that will be replaced with the path of the search result (documents/images/party.jpg).
• {.}: Like {}, but without the file extension (documents/images/party).
• {/}: A placeholder that will be replaced by the basename of the search result (party.jpg).
• {//}: The parent of the discovered path (documents/images).
• {/.}: The basename, with the extension removed (party).

If you do not include a placeholder, fd automatically adds a {} at the end.

For -x/--exec, you can control the number of parallel jobs by using the -j/--threads option. Use --threads=1 for serial execution.

Sometimes we want to ignore search results from a specific subdirectory. For example, we might want to search all hidden files and directories (-H) but exclude all matches from .git directories. We can use the -E (or --exclude) option for this. It takes an arbitrary glob pattern as an argument:

> fd -H -E .git …

We can also use this to skip mounted directories:

> fd -E /mnt/external-drive …

.. or to skip certain file types:

> fd -E '*.bak' …

To make exclude-patterns like these permanent, you can create a .fdignore file. They work like .gitignore files, but are specific to fd. For example:

> cat ~/.fdignore
/mnt/external-drive
*.bak

Note: fd also supports .ignore files that are used by other programs such as rg or ag.

If you want fd to ignore these patterns globally, you can put them in fd's global ignore file. This is usually located in ~/.config/fd/ignore in macOS or Linux, and %APPDATA%\fd\ignore in Windows.

You can use fd to remove all files and directories that are matched by your search pattern. If you only want to remove files, you can use the --exec-batch/-X option to call rm. For example, to recursively remove all .DS_Store files, run:

> fd -H '^\.DS_Store$' -tf -X rm

If you are unsure, always call fd without -X rm first. Alternatively, use rms "interactive" option:

> fd -H '^\.DS_Store$' -tf -X rm -i

If you also want to remove a certain class of directories, you can use the same technique. You will have to use rms --recursive/-r flag to remove directories.

Note: there are scenarios where using fd … -X rm -r can cause race conditions: if you have a path like …/foo/bar/foo/… and want to remove all directories named foo, you can end up in a situation where the outer foo directory is removed first, leading to (harmless) "'foo/bar/foo': No such file or directory" errors in the rm call.

This is the output of fd -h. To see the full set of command-line options, use fd --help which also includes a much more detailed help text.

Usage: fd [OPTIONS] [pattern] [path]...

Arguments:
  [pattern]  the search pattern (a regular expression, unless '--glob' is used; optional)
  [path]...  the root directories for the filesystem search (optional)

Options:
  -H, --hidden                     Search hidden files and directories
  -I, --no-ignore                  Do not respect .(git|fd)ignore files
  -s, --case-sensitive             Case-sensitive search (default: smart case)
  -i, --ignore-case                Case-insensitive search (default: smart case)
  -g, --glob                       Glob-based search (default: regular expression)
  -a, --absolute-path              Show absolute instead of relative paths
  -l, --list-details               Use a long listing format with file metadata
  -L, --follow                     Follow symbolic links
  -p, --full-path                  Search full abs. path (default: filename only)
  -d, --max-depth <depth>          Set maximum search depth (default: none)
  -E, --exclude <pattern>          Exclude entries that match the given glob pattern
  -t, --type <filetype>            Filter by type: file (f), directory (d), symlink (l),
                                   executable (x), empty (e), socket (s), pipe (p)
  -e, --extension <ext>            Filter by file extension
  -S, --size <size>                Limit results based on the size of files
      --changed-within <date|dur>  Filter by file modification time (newer than)
      --changed-before <date|dur>  Filter by file modification time (older than)
  -o, --owner <user:group>         Filter by owning user and/or group
  -x, --exec <cmd>...              Execute a command for each search result
  -X, --exec-batch <cmd>...        Execute a command with all search results at once
  -c, --color <when>               When to use colors [default: auto] [possible values: auto,
                                   always, never]
  -h, --help                       Print help (see more with '--help')
  -V, --version                    Print version

Let's search my home folder for files that end in [0-9].jpg. It contains ~190.000 subdirectories and about a million files. For averaging and statistical analysis, I'm using hyperfine. The following benchmarks are performed with a "warm"/pre-filled disk-cache (results for a "cold" disk-cache show the same trends).

Let's start with find:

Benchmark #1: find ~ -iregex '.*[0-9]\.jpg$'

  Time (mean ± σ):      7.236 s ±  0.090 s

  Range (min … max):    7.133 s …  7.385 s

find is much faster if it does not need to perform a regular-expression search:

Benchmark #2: find ~ -iname '*[0-9].jpg'

  Time (mean ± σ):      3.914 s ±  0.027 s

  Range (min … max):    3.876 s …  3.964 s

Now let's try the same for fd. Note that fd always performs a regular expression search. The options --hidden and --no-ignore are needed for a fair comparison, otherwise fd does not have to traverse hidden folders and ignored paths (see below):

Benchmark #3: fd -HI '.*[0-9]\.jpg$' ~

  Time (mean ± σ):     811.6 ms ±  26.9 ms

  Range (min … max):   786.0 ms … 870.7 ms

For this particular example, fd is approximately nine times faster than find -iregex and about five times faster than find -iname. By the way, both tools found the exact same 20880 files 😄.

Finally, let's run fd without --hidden and --no-ignore (this can lead to different search results, of course). If fd does not have to traverse the hidden and git-ignored folders, it is almost an order of magnitude faster:

Benchmark #4: fd '[0-9]\.jpg$' ~

  Time (mean ± σ):     123.7 ms ±   6.0 ms

  Range (min … max):   118.8 ms … 140.0 ms

Note: This is one particular benchmark on one particular machine. While I have performed quite a lot of different tests (and found consistent results), things might be different for you! I encourage everyone to try it out on their own. See this repository for all necessary scripts.

Concerning fd's speed, the main credit goes to the regex and ignore crates that are also used in ripgrep (check it out!).

fd can colorize files by extension, just like ls. In order for this to work, the environment variable LS_COLORS has to be set. Typically, the value of this variable is set by the dircolors command which provides a convenient configuration format to define colors for different file formats. On most distributions, LS_COLORS should be set already. If you are on Windows or if you are looking for alternative, more complete (or more colorful) variants, see here, here or here.

fd also honors the NO_COLOR environment variable.

Remember that fd ignores hidden directories and files by default. It also ignores patterns from .gitignore files. If you want to make sure to find absolutely every possible file, always use the options -H and -I to disable these two features:

> fd -HI …

A lot of special regex characters (like [], ^, $, ..) are also special characters in your shell. If in doubt, always make sure to put single quotes around the regex pattern:

> fd '^[A-Z][0-9]+$'

If your pattern starts with a dash, you have to add -- to signal the end of command line options. Otherwise, the pattern will be interpreted as a command-line option. Alternatively, use a character class with a single hyphen character:

> fd -- '-pattern'
> fd '[-]pattern'

Shell aliases and shell functions can not be used for command execution via fd -x or fd -X. In zsh, you can make the alias global via alias -g myalias="…". In bash, you can use export -f my_function to make available to child processes. You would still need to call fd -x bash -c 'my_function "$1"' bash. For other use cases or shells, use a (temporary) shell script.

You can use fd to generate input for the command-line fuzzy finder fzf:

export FZF_DEFAULT_COMMAND='fd --type file'
export FZF_CTRL_T_COMMAND="$FZF_DEFAULT_COMMAND"

Then, you can type vim <Ctrl-T> on your terminal to open fzf and search through the fd-results.

Alternatively, you might like to follow symbolic links and include hidden files (but exclude .git folders):

export FZF_DEFAULT_COMMAND='fd --type file --follow --hidden --exclude .git'

You can even use fd's colored output inside fzf by setting:

export FZF_DEFAULT_COMMAND="fd --type file --color=always"
export FZF_DEFAULT_OPTS="--ansi"

For more details, see the Tips section of the fzf README.

rofi is a graphical launch menu application that is able to create menus by reading from stdin. Piping fd output into rofis -dmenu mode creates fuzzy-searchable lists of files and directories.

Create a case-insensitive searchable multi-select list of PDF files under your $HOME directory and open the selection with your configured PDF viewer. To list all file types, drop the -e pdf argument.

fd --type f -e pdf . $HOME | rofi -keep-right -dmenu -i -p FILES -multi-select | xargs -I {} xdg-open {}

To modify the list that is presented by rofi, add arguments to the fd command. To modify the search behaviour of rofi, add arguments to the rofi command.

The emacs package find-file-in-project can use fd to find files.

After installing find-file-in-project, add the line (setq ffip-use-rust-fd t) to your ~/.emacs or ~/.emacs.d/init.el file.

In emacs, run M-x find-file-in-project-by-selected to find matching files. Alternatively, run M-x find-file-in-project to list all available files in the project.

To format the output of fd similar to the tree command, install as-tree and pipe the output of fd to as-tree:

fd | as-tree

This can be more useful than running tree by itself because tree does not ignore any files by default, nor does it support as rich a set of options as fd does to control what to print:

❯ fd --extension rs | as-tree
.
├── build.rs
└── src
    ├── app.rs
    └── error.rs

For more information about as-tree, see the as-tree README.

Note that fd has a builtin feature for command execution with its -x/--exec and -X/--exec-batch options. If you prefer, you can still use it in combination with xargs:

> fd -0 -e rs | xargs -0 wc -l

Here, the -0 option tells fd to separate search results by the NULL character (instead of newlines). In the same way, the -0 option of xargs tells it to read the input in this way.

... and other Debian-based Linux distributions.

If you run Ubuntu 19.04 (Disco Dingo) or newer, you can install the officially maintained package:

sudo apt install fd-find

Note that the binary is called fdfind as the binary name fd is already used by another package. It is recommended that after installation, you add a link to fd by executing command ln -s $(which fdfind) ~/.local/bin/fd, in order to use fd in the same way as in this documentation. Make sure that $HOME/.local/bin is in your $PATH.

If you use an older version of Ubuntu, you can download the latest .deb package from the release page and install it via:

sudo dpkg -i fd_8.7.0_amd64.deb  # adapt version number and architecture

If you run Debian Buster or newer, you can install the officially maintained Debian package:

sudo apt-get install fd-find

Note that the binary is called fdfind as the binary name fd is already used by another package. It is recommended that after installation, you add a link to fd by executing command ln -s $(which fdfind) ~/.local/bin/fd, in order to use fd in the same way as in this documentation. Make sure that $HOME/.local/bin is in your $PATH.

Starting with Fedora 28, you can install fd from the official package sources:

dnf install fd-find

You can install the fd package from the official sources, provided you have the appropriate repository enabled:

apk add fd

You can install the fd package from the official repos:

pacman -S fd

You can use the fd ebuild from the official repo:

emerge -av fd

You can install the fd package from the official repo:

zypper in fd

You can install fd via xbps-install:

xbps-install -S fd

Get the latest fd-v*-x86_64-unknown-linux-gnu.tar.gz file from sharkdp on github

tar xf fd-v*-x86_64-unknown-linux-gnu.tar.gz
chown -R root:root fd-v*-x86_64-unknown-linux-gnu
cd fd-v*-x86_64-unknown-linux-gnu
sudo cp fd /bin
gzip fd.1
chown root:root fd.1.gz
sudo cp fd.1.gz /usr/share/man/man1
sudo cp autocomplete/fd.bash /usr/share/bash-completion/completions/fd
source /usr/share/bash-completion/completions/fd
fd

You can install fd with Homebrew:

brew install fd

… or with MacPorts:

sudo port install fd

You can download pre-built binaries from the release page.

Alternatively, you can install fd via Scoop:

scoop install fd

Or via Chocolatey:

choco install fd

Or via Winget:

winget install sharkdp.fd

You can install the fd package from the official repo:

guix install fd

You can use the Nix package manager to install fd:

nix-env -i fd

You can install the fd-find package from the official repo:

pkg install fd-find

On linux and macOS, you can install the fd-find package:

npm install -g fd-find

With Rust's package manager cargo, you can install fd via:

cargo install fd-find

Note that rust version 1.64.0 or later is required.

make is also needed for the build.

The release page includes precompiled binaries for Linux, macOS and Windows. Statically-linked binaries are also available: look for archives with musl in the file name.

git clone https://github.com/sharkdp/fd

# Build
cd fd
cargo build

# Run unit tests and integration tests
cargo test

# Install
cargo install --path .

• sharkdp
• tmccombs
• tavianator

Copyright (c) 2017-2021 The fd developers

fd is distributed under the terms of both the MIT License and the Apache License 2.0.

See the LICENSE-APACHE and LICENSE-MIT files for license details.