Command Reference

tar - manipulate tar files

• tar -czf <archive> <directory> - create a gzip-compressed archive containing the given directory
• tar -chzf <archive> <directory> - same as above, but dereference any symbolic links and archive the target file
• tar -cf <archive> -C <directory> <file1> <file2> - create an uncompressed archive containing two files relative to the given directory
• tar -xf <archive> -C <directory> <file1> <file2> - extract 2 files from the archive into the given directory
• tar -taf <archive> - list files in the archive and automatically determine the compression scheme

find - find files and directories matching criteria

• -L follow symlinks
• -type <f,d,l> - include only files, directories, or links respectively
• -name <glob> - filter file name by glob pattern
• -path <glob> - filter by glob pattern against entire path
• -print - print matching file paths
• -depth - process directory contents before directory itself
• -maxdepth <n> - limit depth of search
• -mindepth <n> - ignore shallow files
• -exec <command> \; - run a command for each result where \; marks the end of the command and {} optionally specifies where arguments are placed
• -print0 - output null-terminated paths for use with xargs
• find . -type f -exec grep "pattern" {} \; - find files that match pattern
• find <file1> -newer <file2> -print - print the newer of two files
• find tmp -maxdepth 1 -name *.mp3 -print0 | xargs -0 rm - delete MP3 files that are 1 level deep within tmp

xargs - treat input as arguments to a command

• -n <n> - limit args per instance of command. Often -n 1 is needed when a command accepts only one file
• -L <n> - limit lines of input per command. Often -L 1 when each line of input corresponds to an instance of the command
• -P <n> - limit number of parallel commands, or 0 for unlimited
• -I<symbol> - symbol to replace with argument values in command, usually {}. Otherwise, they are placed at the end of the command
• -0 - read input as null-terminated strings. Mainly used along with find -print0
• xargs -n 1 -P 1 -I{} ssh -tt {} 'top -n 1 -b | grep "Cpu(s)"' - ssh to hosts one at a time and run top on them
• find . -name '*.wav' -print0 | xargs -0 md5sum - find all WAV files and print their MD5 sums
• cat old-archive-ids | xargs -L1 -I{} ./delete.sh {} - run a script once for each line in a file

basename and dirname - remove prefix and suffix from file paths

• basename path/to/file.txt - yields file.txt
• basename path/to/file.txt .txt - yields file
• basename -a -s .txt file1.txt file2.txt - yields file1\nfile2
• dirname path/to/file.txt - yields path/to
• dirname path/to/ - yields path
• dirname path/to - yields path

sed - manipulate text line-wise

• sed -n <n>p - print only line n of the input
• sed -e '2,$d' - delete all but first line of input
• sed -e "/^FirstKeyword\$/,/^SecondKeyword\$/d" - delete lines once FirstKeyword is seen until after SecondKeyword is seen
• sed -e "/^FirstKeyword\$/,/^SecondKeyword\$/!d" - the opposite of the above; keep only lines between and including the two keywords
• sed 's/Counters=\(.*\)$/\1/' - keep only the value of Counters
• sed 's/,/\n/g' - replace all commas with newlines
• sed -r 's/(^|_)([a-z])/\U\2/g' - UpperCamelCase a string
• sed 's/^.*customerId.{"s":"\([A-Z0-9]*\)"}.*$/\1 &/g' - capture ID value and copy it to the front of the line
• find src -name '*.c' -exec sed -i 's/= OLD_MACRO(\(\w*\))\[\(.*\)\]/= NEW_MACRO(\1, \2)/' {} \; - in each C file, replace a particular macro usage of the form OLD_MACRO(foo)[bar] with NEW_MACRO(foo, bar)

tr - delete or replace characters

• tr '[A-Z]' '[a-z]' - replace all upper case letters with lower case
• tr -d '"' - remove all double-quotes
• tr -s '\t' ',' - replace tabs with commas

cmp - compare files byte-wise

• cmp -s <file1> <file2> - exits non-zero if files are not the same

pr and paste - join files column-wise

• pr -m -t <file1> <file2> - prints files side-by-side
• paste <file1> <file2> - prints files side-by-side

comm - some set operations on sorted files

• comm -12 <sorted_file_1> <sorted_file_2> - outputs lines common to both files
• comm -23 <sorted_file_1> <sorted_file_2> - outputs lines unique to the first file
• comm -3 <sorted_file_1> <sorted_file_2> - outputs lines that are unique to either file

dd - read and write files or devices byte-wise

• dd if=<file> ibs=1 skip=200 count=100 - read 100 bytes starting at the 200th byte
• dd if=/dev/sdf of=image.bin bs=8M - write the entire device /dev/sdf to a file with a block size of 8M
• dd if=<image> of=<device> bs=16M status=progress oflag=sync - recommended by Debian for writing an isohybrid image to USB

sudo - execute commands as a different user

• The -E flag will preserve your current environment when running the command
• Using --preserve-env=<list> will preserve only the listed environment variables

kill and pkill - terminate processes or send signals

• kill <pid>, kill -15 <pid>, or kill -TERM <pid> - ask a process to stop gracefully
• kill -9 <pid> or kill -KILL <pid> - force-kill a process by PID
• pkill -KILL -f <part_of_process_name> - force-kill processes with a name that includes the given string

Other signals

kill is most commonly used with the TERM and KILL signals, but there are others that a given process may or may not respond to:

• HUP - sometimes causes a daemon or service to reload its configuration files
• INT - equivalent to pressing Ctrl+C in a terminal
• ALRM - represents a timed event
• QUIT and ABRT - similar to INT but often used to generate a core dump
• STOP and CONT - used to pause and resume a process respectively

stat and readlink - print file information

The stat command is not compatible between Linux and BSD / macOS, unfortunately, so readlink may be a better option for the specific use case of resolving a link to its target.

• stat -f '%Y' <symbolic_link> (BSD) or readlink <symbolic_link> (Both) - print target of a symbolic link
• stat -f '%Lp' <file> (BSD) or stat -c '%a' <file> (Linux) - print permission mask in octal form

tcpdump - capture network traffic

• tcpdump -A -vvv <some.endpoint.host> - print TCP packets to and from the given host as ASCII with maximum verbosity

dig - DNS lookup

• dig <some.endpoint.host> - view DNS resolution for the given host
• dig -x <IP address> - reverse DNS lookup to determine hostname for a given IP address

nc - read and write TCP/UDP connections

• nc <some.host.name> <port> - open TCP connection
• nc -u <some.host.name> <port> - open UDP connection
• nc -l <port> - listen on a given port
• nc -z <some.host.name> <port_range_start>-<port_range_end> - attempt to connect on a range of ports and report which were open

tail - print and follow ends of files

• tail -f <file> - print last lines of a file and print new lines as they are appended
• tail -F <file> - same as -f but deals with the named file being renamed or replaced by a new file

mount - manage mounted filesystems

• mount - view currently mounted paths and filesystem types
• mount <device> <path> - mount a device while attempting to detect the filesystem automatically
• umount <path> - unmount the filesystem mounted at the given path

mkfs - family of commands for formatting storage

• mkfs.exfat <device> - format a block device with exFAT

watch - run a command repeatedly to observe changes

• watch -t <command> - run command repeatedly with the default 2 second interval and hide the header
• watch -n 60 <command> - run command repeatedly every minute
• watch -d <command> - run command repeatedly and highlight differences between consecutive runs
• watch --differences=permanent <command> - run command repeatedly and highlight differences from initial run

rsync - copy and mirror directory structures

• -delete - delete files that don't exist on sender (system)
• -v - verbose (-vv will provide more detailed information)
• -e ssh <options> - specify ssh with options for remote targets
• -a - archive mode which preserves permissions (owners, groups), times, symbolic links, and devices
• -r - recurse into directories
• -z - compress file data during transfer
• --exclude <directory> - excludes the given directory
• -P - show progress during the transfer
• -n - perform a dry-run without actually writing anything
• -i - show changes
• -W - transfer whole files; do not attempt a delta sync

If / is placed at the end of the source folder, rsync will copy only the content of the folder. Otherwise, it will copy the folder itself. Similarly, if / is placed at the end of the destination folder, rsync will paste the data directly inside the folder. Otherwise, it will create a folder with that name and copy into it.

• rsync -v -e ssh /home/localuser/testfile.txt remoteuser@X.X.X.X:/home/remoteuser/transfer - copy a file from local to remote
• rsync -r -a -v -e ssh --delete /home/localuser/testfolder remoteuser@X.X.X.X:/home/remoteuser/testfolder - sync a local folder to remote with archival flags set and deleting files that don't exist locally

git - the only source control anyone cares about anymore

• git init - initialize the current directory as a git repository
• git remote add origin <uri> - add an upstream remote repository
• git remote set-url origin <uri> - change the upstream URL of an existing remote
• git add <path> - stage modifications and new files under the given path
• git add --update <path> - stage modifications but ignore new files
• git fetch - update information from the origin
• git checkout -t origin/<branch> - checkout a local copy of an upstream branch
• git push origin :<branch> - delete a remote branch
• git checkout -b <local_name> --track origin/<branch_name> - create a local branch tracking a remote one with a different name
• git checkout -b <local_name> - create a new local copy of the current branch
• git push origin <change_id>:<branch> - push local changes only up to a particular commit
• git push -u origin <branch> - regular push of local changes to remote branch
• git clean -f -d - remove local untracked files and directories
• git reset --merge - abort a merge that isn't going well
• git reset --hard origin/<branch> - discard committed changes that haven't yet been pushed
• git branch -d <branch> - delete a local branch
• git diff HEAD~2 HEAD -- <paths...> - compare paths between current and 2 revisions prior
• git diff <commit> -- <paths...> - compare paths between current and a specific revision
• git submodule update --init --recursive - pull down and register missing submodules
• git submodule sync - synchronize all submodule URL configurations to the value specified in .gitmodules
• git switch <branch> - switch to a different, existing local branch

go - for the very limited golang exposure I have

• go mod init <name> - initialize a go module workspace
• go run <go_file> - run a go file containing a main function
• go test -v - run functions starting with "Test" found in file paths ending with "_test.go"

cmake - the closest thing to a standard C/C++ build system

• cmake -DCMAKE_BUILD_TYPE=Debug -DCMAKE_TOOLCHAIN_FILE=path/to/vcpkg/scripts/buildsystems/vcpkg.cmake -S./source/directory -B./build/directory -G Ninja - configure a vcpkg build
• cmake -DCMAKE_BUILD_TYPE=Debug -S./ -B./build - configure a build with mostly default options
• cmake -DCMAKE_CXX_INCLUDE_WHAT_YOU_USE=include-what-you-use -DCMAKE_BUILD_TYPE=Debug -S./ -B./build - to add include-what-you-use support
• cmake --build build/directory - execute a previously configured build

winget - Windows has a package manager, too

• winget search <term> - search packages by name, ID, tags
• winget add <package> - install package via an unambiguous name or ID

npm - because I can't avoid it these days

• npm init -y - initialize the directory as an NPM package
• npm install <package> or npm i <package> - install a package in this workspace

docker - work with containers

• docker build -t <image_name> . - From a directory with a Dockerfile, build an image with the given name
• docker build -t <image_name> -f <dockerfile> . - Same, but with a custom Dockerfile path
• docker build --no-cache -t <image_name> . - Same, but redo all cached build steps
• docker ps - See currently running containers
• docker ps -a - See all containers
• docker attach <container_id> - Reattach to a running container
• docker commit <container_id> <image_name> - Commit the state of a container to a new image
• docker stop <container_id> - Stop a running container
• docker start <container_id> - Restart a stopped container
• docker cp <path/on/host> <container_id>:<path/in/container> - Copy files between host and container
• docker volume create <volume_name> - Create a new volume
• docker volume ls - List volumes
• docker run -it <image_name> - Start a container in TTY interactive mode
• docker run --rm ... - auto-remove the container when it exits
• docker run -it -v /tmp/.X11-unix:/tmp/.X11-unix -e DISPLAY=docker.for.mac.host.internal:0 <image_name> - Start a container while making XQuartz available to it
• docker run -it -v <volume_name>:<path/in/container> <image_name> - Start a container with a volume attached
• docker run -it -v <path/in/container> <image_name> - Start a container with a new, unnamed volume
• docker run -it -v <path/on/host>:<path/in/container> <image_name> - Start a container with a host directory mounted
• docker container prune - remove stopped containers
• docker rm <container> - delete a container
• docker compose up -d - create a container from a docker-compose.yml file in the current directory
• docker exec -it <container> <command> - run a command within an existing, running container

python and pip

• python -m venv ./venv - create a virtual environment

pacman - Arch linux package manager

• pacman -Q - list installed packages
• pacman -Qi - list installed packages with detail
• pacman -Qe - list packages explicitly installed by the user
• pacman -Qei - list packages explicitly installed by the user with detail
• pacman -Qdtq - list installed, orphaned packages
• pacman -Ql <package_name> - list files provided by a given installed package
• pacman -Rs <package_name> - remove package
• pacman -Rs $(pacman -Qdtq) - remove orphaned packages
• pacman -Ss <search_string> - search available packages
• pacman -S <package_name> - install package
• pacman -Syu - full system update
• pacman -Scc - clean package cache

apt - More common package manager

• apt update - synchronize local index with repositories
• apt upgrade - install all newer versions of installed packages

ruby - using Ruby for one-liners

Ruby has command-line options that make it useful for writing inline scripts as an alternative to sed or awk.

• -F <pattern> - specifies the input field separator ($;)
• -n - causes Ruby to add while gets around your script which makes it iterate over file name arguments somewhat like sed -n or awk
• -p - mostly the same as -n, but prints the value of $_ at the each end of each loop
• -a - turns on auto-split mode when used with -n or -p. In auto-split mode, Ruby executes $F = $_.split at the beginning of each loop
• -d or --debug - turns on debug mode; $DEBUG will be set to true
• -i <extension> - specifies in-place-edit mode with an optional extension to be used for a backup copy
• -l- enables automatic line-ending processing, which sets $\ to the value of $/ and chops every line read using chop!
• -e <ruby_statements> - specifies script from command-line and causes Ruby not to search the rest of arguments for a script file name
• -r <library> - load the specified library using require. It is useful when using -n or -p
• ruby -p -e ’$_.tr! "a-z", "A-Z"’ - equivalent to tr '[a-z]' '[A-Z]'

ffmpeg - media encoding

• ffmpeg -framerate 30 -i "%d.bmp" -r 30 -pix_fmt yuv420p video.mp4 - create a video from a sequence of image files
• ffmpeg -i original.jpg -qmin 1 -q:v 1 -vf "scale=iw/2:ih/2" reduced.jpg - re-encode an image at half the resolution
• ffmpeg -i video.mp4 -c:v null -c:a copy audio.m4a - extract only the audio from an MP4 video
• ffmpeg -i input.avi -vf 'yadif=1:1,drawbox=y=ih-h:w=0:h=11:t=max,hqdn3d=6' -acodec aac -vcodec libx264 -pix_fmt yuv420p -preset veryslow -crf 20 -aspect 4:3 output.mp4 - re-encode an interlaced, 4:3 AVI as an MP4 after masking overscan noise and denoising the image
• ffmpeg -i input.avi -acodec copy -vcodec ffv1 -coder 1 -g 1 -context 1 -pix_fmt yuv410p output.avi - re-encode as lossless FFV1
• for f in **/*.wav; ./ffmpeg.exe -i "$f" flac/"$(basename $f .wav)".flac - encode all WAV files in under the current path as FLAC and put the files under a "flac" directory
• ffmpeg -hide_banner -i input.mp4 -vf "scale='if(lt(iw,ih),min(720,iw),-2)':'if(lt(iw,ih),-2,min(720,ih))'" -c:v libx265 -crf 28 -preset veryslow -c:a copy output.mp4 - re-encode as HEVC and scale to a maximum width or height of 720
• ffmpeg -i input.gif -filter_complex "[0:v]crop=300:270:100:0,split[a][b];[a]palettegen=max_colors=32[p];[b][p]paletteuse=dither=none" output.gif - crop and re-encode gif at a reduced quality
• ffmpeg -i input.mp4 -filter_complex "[0:v] fps=8,scale=300:-1:flags=lanczos,crop=in_w:in_h:0:0,split[a][b];[a]palettegen=reserve_transparent=on[p];[b][p]paletteuse=dither=bayer:bayer_scale=5" output.gif - convert video to gif
• ffmpeg -i input.png -vf "scale=64:64:force_original_aspect_ratio=decrease,pad=64:64:(ow-iw)/2:(oh-ih)/2:color=black@0" out.ico - convert an image to a 64x64 ICO file
• ffmpeg -i input.mp4 -vf "scale=iw/2:ih/2" -c:v libx265 -crf 28 -preset slow -c:a copy output.mp4 - encode video at half the input resolution
• ffmpeg -i input.mp4 -vf "scale=-2:720" -c:v libx265 -crf 28 -preset slow -c:a copy output.mp4 - preserve input ratio while scaling to a height of 720.
• ffmpeg -i input.mp4 -vf "scale='if(gt(max(iw,ih)/2,720),-2,iw)':'if(gt(max(iw,ih)/2,720),-2,ih)'" -c:v libx265 -crf 28 -preset slow -c:a copy output.mp4 - scale video to a height of 720 but only if it is currently larger than that
• ffprobe -v error -select_streams v:0 -show_entries stream=width,height -of csv=s=x:p=0 input.mp4 - print the dimensions of a video
• ffmpeg -i input.mp4 -vf "scale='if(lt(iw,ih),min(720,iw),-2)':'if(lt(iw,ih),-2,min(720,ih))'" -c:v libx265 -crf 28 -preset slow -c:a copy output.mp4 - scale the shorter dimension of a video to 720 (for handling horizontal and vertical video)

robocopy - somewhat like rsync for Windows

• robocopy <source> <destination> /E /COPY:DAT /DCOPY:DAT - deep copy files while preserving timestamps, including empty subdirectories

aws - AWS CLI commands I need to remember

• aws codecommit create-repository --region <region> --repository-name <name> - create a new repository
• aws codecommit list-repositories --region <region> - list existing repositories for a region
• aws codecommit get-repository --region <region> --repository-name <name> - get details of a repository such as URL for cloning

jq - transform and filter JSON

• jq '.[0:5]' - show the first 5 elements in a list
• jq '.[0:5] | .[].title' - show a particular property of the first 5 elements in a list
• jq '.[0:5][] | .title' - equivalent to the above
• jq 'length' - show the number of items in a list
• jq 'sort_by(.time) | reverse | .[0:5][] | .title'- sort by a time property in reverse and then get the first 5 titles
• jq '.[0:5][] | [.time, .title, .titleUrl] | @tsv' - build a tab-separated table from selected properties
• jq 'map(select((.titleUrl // "") | contains("/watch"))) | .[0:5]' - select only entries with a titleUrl containing "/watch"
• jq 'map(select((.titleUrl // "") | contains("/post") | not)) | .[0:5]' - reject entries with a titleUrl containing "/post"

column - print tabular data

• column -t -s$'\t' | less -S - format data separated by tabs and show via less with horizontal scrolling
• column -t -s, | less -S - format data separated by commas and show via less with horizontal scrolling

wget - download files over HTTP

• wget -O - <url> - fetch a URL and write to standard output rather than a file
• wget -qO - <url> - fetch a URL and write to standard output but suppress the usual progress bar

curl - download or make various requests over HTTP

• curl <url> - fetch a URL and write the content to standard output
• curl -I <url> - fetch only the headers for the URL
• curl -IL <url> - fetch only the headers for the URL while following any redirects

socat - cat for sockets

• socat -v TCP-LISTEN:11434,fork,reuseaddr TCP:127.0.0.1:11435 - listen for TCP traffic on a given port, print it, and forward to another destination

ss - investigate socket usage

• ss -tulpn - display processes using and listening to TCP and UDP sockets that are listening, use numbers instead of human-readable alternatives

xsel - access X11 clipboard from terminal

• xsel --clipboard --output - write clipboard content to stdout

Miscellaneous

• cal - display calendars
• at - schedule a task at a given time
• fortune - print random messages from a collection
• diff - compare files or directory structures
• patch - apply changes specified by a diff
• tsort - topologically sort a graph
• od - display files in hexadecimal or other forms
• nohup - prevent a program from stopping after logout
• nl - add line numbers to output
• join - join lines in files by shared field values
• bc - do calculations
• file - heuristically determine file content type
• fc - deal with command history
• csplit - split a file into multiple based on context lines
• split - split a file into multiple files of a fixed size
• fmt, fold - apply word wrapping to text
• banner, figlet, and cowsay - print fancy text
• pexec and parallel - run commands concurrently
• javap -classpath some_jar.jar some.package.SomeClass - list method signatures of a class within a JAR file
• tree <file> - write input to both standard output and the given file
• wget -O <output_path> <some.resource/address> - download a given resource to a given path
• chsh -s <shell_path> <user> - change a user's login shell
• uuencode and uudecode - binary to ASCII format
• recutils - relational database operations on plain text files
• apropos - search manual pages