Cheatsheets:LinuxCLI/FileOps: Difference between revisions

Latest revision as of 12:14, 2 September 2026

Languages: English · français · Esperanto

Quick reference for smart people — part of our Linux cheatsheets collection.

The everyday Linux command-line file operations: navigate, create, copy, move, inspect, search, and archive files.

New to the command line? Every command here has a built-in manual — try man ls or tldr for short examples; explainshell.com annotates any command line.

Navigation & listing

pwd              # print the current directory
cd dir           # enter dir
cd ..            # go up one level
cd ~             # home directory
cd -             # previous directory

ls — the workhorse:

ls [options] [dir]   # list directory contents
ls *.txt             # glob: only matching names

Option	Effect
`-l`	long format: permissions, owner, size, date
`-a`	hidden files (dotfiles)
`-h`	human-readable sizes (combine with -l)
`-R`	recurse into subdirectories
`-t`	sort by modification time, newest first

Creating & removing

mkdir newdir                      # create a directory
mkdir -p dir/subdir/newsubsubdir  # create parents as needed
mkdir -m 755 dir                  # set permissions at creation
rmdir emptydir                    # remove an EMPTY directory
rmdir -p dir/subdir/emptysubdir   # ... and its empty parents
touch file.txt                    # create an empty file (or refresh its mtime)
touch -t 202501011200 f           # set a specific timestamp
rm file.txt                       # delete a file
rm -r dir/                        # delete a directory and its contents
rm -f file                        # never prompt, ignore missing files
rm -i file                        # prompt before each delete

rm -rf is irreversible — there is no trash bin. Double-check the path (and your current directory with pwd) before deleting recursively.

Copying & moving

cp file.txt backup.txt      # copy a file
cp -r src/ dest/            # copy a directory recursively
cp -a src/ dest/            # recursive + preserve attributes
cp -i file.txt dest/        # prompt before overwriting
cp file1 file2 dest/        # several files into a directory
mv file.txt newname.txt     # rename
mv file.txt dir/            # move into a directory
mv -i file.txt dir/         # prompt before overwriting
mv src/ dest/               # move/rename a directory

Viewing files

cat file.txt            # print the whole file
cat file1 file2         # concatenate, in order
cat -n file.txt         # with line numbers
less file.txt           # scrollable pager: /search, g/G jump,f/b move forward/backward one page, q quit
head -n 5 file.txt      # show first 5 lines of a file. If no -n parsed, default is 10
tail -n 50 app.log      # show last 50 lines of a file. If no -n parsed, default is 10
tail -f app.log         # follow a growing log, Ctrl+C to stop

Searching

grep — search inside files:

grep [options] pattern file   # print matching lines

Option	Effect
`-i`	case-insensitive
`-n`	include line numbers
`-v`	invert: lines that do NOT match
`-r`	search a directory recursively
`-l`	only filenames with matches
`-c`	count matching lines
`-E`	extended regex, e.g. `grep -E 'a\|b' file`

find — search by name, type, age, size:

find . -name "*.py"                  # by name (glob)
find /home -type f                   # only files
find /home -type d                   # only directories
find . -mtime -7                     # modified in the last 7 days
find . -size +100M                   # larger than 100 MB
find . -name "*.log" -delete         # delete matches
find . -name "*.tmp" -exec <some-commands> {} \;  # run a command on each match

locate — faster index-based search:

locate file.txt    # instant filename lookup
locate folder/subfolder    # instant partial path lookup
sudo updatedb           # refresh the index (requires root permission)

locate searches an index refreshed by updatedb — files created since the last update are invisible to it. find is slower but always up to date.

Permissions & ownership

chmod — change file/folder permissions.

chmod 755 script.sh     # rwxr-xr-x
chmod 644 file.txt      # rw-r--r--
chmod u+x script.sh     # add execute for the owner
chmod -R 755 dir/       # recursive
chown alice file.txt        # change owner
chown alice:dev file.txt    # owner and group
chown -R alice:dev dir/
chgrp dev file.txt       # change group only
stat file.txt            # perms, size, timestamps, inode
file archive.tar         # detect the file type

On GNU Linux, there are two notations for file permissions.

Numeric: three digits representing permissions for owner|owner's group|others
Each digit is an arithmetic sum of all accorded permissions: 4:read, 2:write, 1:execute

Symbolic: mostly used to add/remove a single permission.
r:read, w:write, x:execute

u=owner, g=owner's group, o=others, a=all

+add, - to remove, = to set permission to exactly the given value.

Below are some common file permissions:

Numeric Symbolic

7 rwx

6 rw-

5 r-x

4 r--

You can read permissions of a given directory and its content with ls -l.
The command output will be something like this:
drwxr-xr-x  2 rongzhou rongzhou    4096 Aŭg 20 16:19 9781474271325.sdr
-rw-rw-r--  1 rongzhou rongzhou    1758 Mar 10 15:26 antaurigardo.html
drwxrwxr-x  4 rongzhou rongzhou    4096 Mar 13 14:28 Arduino
drwxr-xr-x  4 rongzhou rongzhou    4096 Jul 22 12:20 Bildujo
The first column is permission.

There are always 10 chars.

The first char is always d for directories and - for files.

char 2-4 are permissions for owner in symbolic notation

char 5-7 are permissions for owner's group

char 8-10 are permissions for others

Disk usage

du -sh dir/              # total size of a directory, human-readable
du -sh *                 # size of each item
du -h --max-depth=1 dir/ # per subdirectory
df -h                    # free space per mounted filesystem
df -h .                  # ... only for the filesystem holding the current dir

Links

ln target hardlink          # hard link: same file, another name
ln -s target symlink        # symbolic link: points to a path
ln -s /usr/local/bin/tool ~/bin/tool

Hard links cannot span filesystems or point at directories. ls -l shows a symlink's target with ->.

Text utilities

sort file.txt            # sort lines alphabetically
sort -n file.txt         # numeric sort
sort -r file.txt         # reverse
sort -u file.txt         # unique lines
sort -k2 file.txt        # sort by the 2nd field
diff file1 file2         # line-by-line differences
diff -u file1 file2      # unified diff (patch format)
diff -r dir1 dir2        # compare directories
cmp file1 file2          # byte compare: report first difference
echo "hello"             # print text
echo -n "no newline"     # omit the trailing newline
echo "hi" > file.txt     # overwrite a file with output
echo "hi" >> file.txt    # append to a file

> overwrites, >> appends — one character decides whether an existing file survives.

rg — ripgrep: a very fast recursive text search, the modern alternative to grep -r when you don't know the exact file (install: apt install ripgrep, brew install ripgrep):

rg pattern [path]           # recursive search; path defaults to the current directory
rg pattern file.txt         # search a single file
rg -i 'error|warning' src/  # -i: case-insensitive
rg -n 'TODO' app/           # -n: include line numbers (on by default on a terminal)
rg -v 'DEBUG' app/          # -v: lines that do NOT match
rg -w 'foo' app/            # -w: whole-word matches only
rg -F 'a+b' app/            # -F: literal string, no regex metacharacters
rg -l 'TODO' app/           # -l: only filenames with matches
rg -c 'error' app/          # -c: count matching lines per file
rg -C 3 'panic' app/        # -C N: N lines of context around each match
rg -t py 'import' app/      # -t TYPE: only files of a type (py, js, rs, ...)
rg -g '*.log' 'error' .     # -g GLOB: only files matching the glob
rg -g '!*.min.js' 'var' .   # -g '!glob': exclude files matching the glob
rg --hidden 'secret' .      # --hidden: also search hidden dotfiles
rg --no-ignore 'tmp' .      # --no-ignore: ignore .gitignore / .ignore rules
rg -uu 'pattern' .          # -uu: also hidden files; -uuu also binary (each -u disables more filtering)

rg only prints — it never edits files. By default it skips hidden files, binary files and anything matched by .gitignore; add -uu when results look incomplete. -r / --replace rewrites the matched text in the output only — use it to preview what a replacement would look like, then apply the real edit with sd below. </syntaxhighlight>

sd — search & displace: an intuitive regex find-and-replace, a friendlier sed (install: cargo install sd, brew install sd):

sd old_text new_text file.txt            # replace every occurrence of old_text, in place
sd old_text new_text file1.txt file2.txt # edit several files at once
sd old_text new_text **/*.md             # glob for all .md files in the directory and its subdirectories [1]
sd 'foo\s+bar' 'baz' file.txt            # regex find & replace (JS/Python-style syntax)
sd '(\w+) (\w+)' '$2 $1' file.txt        # regex capture groups: the example swap two words
sd -F 'a+b' 'axb' file.txt               # -F: literal text, '+' is not regex "one or more"
cat file.txt | sd -A '\n' ','            # -A: pattern may match across line breaks (join lines)
sd -p old_text new_text file.txt         # -p: preview the result, do not write
echo 'some_text' | sd old_text new_text   # no file given: read stdin, write stdout
sd old_text -- '-new_text' file.txt      # --: declare end of flags so new_text may start with a dash

sd replaces every occurrence and edits files in place, which is non-reversible.Dry-run with -p and keep files under version control.

[1]: You must enable globstar for ** to be interpreted correctly.
Add to ~/.bashrc:
shopt -s globstar

File format conversion

See Cheatsheets:Pandoc

Option	Effect
`-c`	create an archive
`-x`	extract
`-t`	list contents
`-f`	the archive file name — always followed by the archive
`-z`	gzip compression → `.tar.gz`
`-j`	bzip2 compression → `.tar.bz2`
`-v`	verbose: list each file as it is processed
`-C`	change to a directory before extracting