Kahibaro
Discord Login Register ☰
☰
Previous
Up
Next

7.4.1 Bash-based tools

Table of Contents

Why Bash Is Ideal for Small Linux Tools

Bash is available on almost every Linux system, starts instantly, and glues together existing commands. Bash-based tools shine when:

In this chapter we focus on building robust, reusable Bash-based tools rather than simple, one-off scripts.

Designing Bash Tools as Real Commands

A “tool” should behave like a normal Unix command:

Installing Your Tool on `$PATH`

For a single user, place executables in ~/bin or ~/.local/bin and ensure it’s in $PATH: 

mkdir -p ~/.local/bin
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.bashrc
. ~/.bashrc

Now any script you copy or symlink into ~/.local/bin can be run by name.

Script Skeleton for a Bash Tool

A solid starting template: 

#!/usr/bin/env bash
set -euo pipefail
# Default values
verbose=0
usage() {
    cat <<EOF
Usage: ${0##*/} [OPTIONS] ARG...
Description:
  Short description of what this tool does.
Options:
  -v, --verbose    Increase verbosity
  -h, --help       Show this help and exit
EOF
}
log() {
    # log to stderr
    echo "[$(date +'%F %T')] $*" >&2
}
die() {
    echo "Error: $*" >&2
    exit 1
}
main() {
    # Parse arguments
    local args=()
    while [[ $# -gt 0 ]]; do
        case "$1" in
            -v|--verbose)
                verbose=$((verbose+1))
                shift
                ;;
            -h|--help)
                usage
                exit 0
                ;;
            --) # end of options
                shift
                break
                ;;
            -*)
                die "Unknown option: $1"
                ;;
            *)
                args+=("$1")
                shift
                ;;
        esac
    done
    # Restore positional arguments
    set -- "${args[@]}" "$@"
    # Example: require at least one argument
    [[ $# -ge 1 ]] || { usage >&2; exit 1; }
    # Your core logic here
    if (( verbose > 0 )); then
        log "Running with $# arguments: $*"
    fi
    # ...
}
main "$@"

Key ideas:

Input, Output, and Exit Codes in Tools

Well-behaved tools treat:

Distinguishing stdout and stderr

list_big_files() {
    local dir=${1:-.}
    [[ -d "$dir" ]] || die "Not a directory: $dir"
    # Progress message on stderr
    echo "Scanning directory: $dir" >&2
    # Data to stdout
    find "$dir" -type f -size +100M -print
}

Usage: 

# See only results
mytool /var | sort > big_files.txt 2>/dev/null
# See only progress/errors
mytool /var >/dev/null

Returning Meaningful Exit Codes

Common conventions:

Example: 

if ! some_command; then
    die "some_command failed"  # exits with 1
fi

Or, for specific codes: 

readonly E_USAGE=2
readonly E_NOT_FOUND=3
[[ $# -ge 1 ]] || { usage >&2; exit "$E_USAGE"; }
if [[ ! -f "$1" ]]; then
    echo "Not found: $1" >&2
    exit "$E_NOT_FOUND"
fi

Robust Argument Parsing Patterns

For sophisticated tools, option parsing matters.

Manual Parsing with `while` Loop

The skeleton above used: 

while [[ $# -gt 0 ]]; do
    case "$1" in
        -o|--option)
            opt_value=$2
            shift 2
            ;;
        *)
            args+=("$1")
            shift
            ;;
    esac
done

Pattern for options that require values: 

        -o|--output)
            [[ $# -ge 2 ]] || die "Missing value for $1"
            output=$2
            shift 2
            ;;

Using `getopts` for Short Options

getopts is useful when you only need short options (like -a -b -c value): 

#!/usr/bin/env bash
set -euo pipefail
verbose=0
output=""
usage() {
    echo "Usage: ${0##*/} [-v] [-o FILE] ARG..."
}
while getopts ":vo:" opt; do
    case "$opt" in
        v) verbose=$((verbose+1)) ;;
        o) output=$OPTARG ;;
        \?) echo "Unknown option: -$OPTARG" >&2; usage >&2; exit 2 ;;
        :)  echo "Option -$OPTARG requires an argument." >&2; usage >&2; exit 2 ;;
    esac
done
shift $((OPTIND - 1))
# Now $@ contains non-option arguments

getopts does not handle --long-options; for that, the manual while loop is usually clearer.

Safe Data Handling and Defensive Practices

Writing tools means your script will be used in unpredictable environments. You need to handle:

Handling Filenames Safely

Always quote variables and prefer arrays: 

  rm $files
  for f in $files; do ...
  rm -- "${files[@]}"
  for f in "${files[@]}"; do ...

To read lines safely (including spaces): 

mapfile -t lines < <(command producing lines)
for line in "${lines[@]}"; do
    printf '%s\n' "$line"
done

Or using while: 

while IFS= read -r line; do
    printf '%s\n' "$line"
done < <(command)

Making Pipelines Fail on Error

With set -o pipefail, your tool will fail the whole pipeline if any command fails: 

set -euo pipefail
# If grep fails (e.g. no matches), the whole pipe fails:
grep -r "pattern" /some/dir | sed 's/pattern/FOUND/g'

You might intentionally tolerate failures; in that case, explicitly handle them: 

if ! grep -q "pattern" file; then
    echo "pattern not found, continuing..." >&2
fi

Temporary Files and Cleanup

Use mktemp and trap: 

tmpfile=$(mktemp)
cleanup() {
    rm -f "$tmpfile"
}
trap cleanup EXIT
# Use "$tmpfile"

For multiple resources: 

tmpdir=$(mktemp -d)
cleanup() {
    rm -rf "$tmpdir"
}
trap cleanup EXIT INT TERM

Structuring Bash Tools as Libraries + Frontends

For more complex tools, split reusable logic (a library) from the CLI “frontend”.

Simple Library Pattern

libfs.sh: 

#!/usr/bin/env bash
fs_list_big() {
    local dir=${1:-.}
    find "$dir" -type f -size +100M -print
}

bigfiles tool: 

#!/usr/bin/env bash
set -euo pipefail
# shellcheck source=libfs.sh
. "/path/to/libfs.sh"
usage() {
    echo "Usage: ${0##*/} [DIR]"
}
main() {
    local dir=${1:-.}
    fs_list_big "$dir"
}
main "$@"

This allows multiple tools to reuse the same functions.

Namespacing Functions

To avoid naming collisions in larger toolsets, use prefixes:

Example: 

backup_create_archive() { ... }
backup_list_archives() { ... }
backup_restore_archive() { ... }

Building Multi-Command “Tool Suites” (Subcommands)

Many serious Bash tools emulate git/docker style: 

mytool add ...
mytool list ...
mytool remove ...

Simple Subcommand Dispatcher

#!/usr/bin/env bash
set -euo pipefail
usage() {
    cat <<EOF
Usage: ${0##*/} <command> [options]
Commands:
  add       Add a new item
  list      List items
  remove    Remove an item
EOF
}
cmd_add() {
    # implementation of "add"
    echo "Adding: $*"
}
cmd_list() {
    # implementation of "list"
    echo "Listing items"
}
cmd_remove() {
    # implementation of "remove"
    echo "Removing: $*"
}
main() {
    local cmd=${1-}
    shift || true
    case "$cmd" in
        add)    cmd_add "$@" ;;
        list)   cmd_list "$@" ;;
        remove) cmd_remove "$@" ;;
        -h|--help|"")
            usage
            ;;
        *)
            echo "Unknown command: $cmd" >&2
            usage >&2
            exit 2
            ;;
    esac
}
main "$@"

This approach scales well and keeps each command’s code localized.

Interfacing with Other Tools

The strength of Bash-based tools lies in composing existing commands.

Wrapping External Commands

Sometimes your “tool” is just a strict, opinionated wrapper: 

#!/usr/bin/env bash
set -euo pipefail
# Example: safe-rm that moves files to a trash directory
TRASH_DIR=${TRASH_DIR:-"$HOME/.local/share/trash"}
mkdir -p "$TRASH_DIR"
usage() {
    echo "Usage: ${0##*/} FILE..."
}
main() {
    [[ $# -ge 1 ]] || { usage >&2; exit 2; }
    local f dest
    for f in "$@"; do
        [[ -e "$f" ]] || { echo "No such file: $f" >&2; continue; }
        dest="$TRASH_DIR/$(date +%s)_${f##*/}"
        mv -- "$f" "$dest"
        echo "Moved $f -> $dest"
    done
}
main "$@"

Using `grep`, `awk`, `sed` in a Maintainable Way

Keep transformations isolated and clear: 

# Bad: complex one-liner, hard to maintain
ps aux | grep myproc | grep -v grep | awk '{print $2}' | xargs kill
# Better: separated stages and clear logic
pids=$(ps -C myproc -o pid=)  # if possible, use targeted options
if [[ -n "$pids" ]]; then
    echo "$pids" | xargs kill
fi

When you must use pipelines: 

ps aux |
    awk '$11 ~ /myproc/ {print $2}' |
    xargs -r kill

Note the use of xargs -r to avoid running kill with no arguments.

Configuration Handling for Bash Tools

Non-trivial tools often need user configuration.

Config Files with Defaults

Pattern:

  1. Set hard-coded defaults
  2. Optionally load a config file if present
  3. Allow command-line options to override config

Example: 

#!/usr/bin/env bash
set -euo pipefail
# 1. Hard-coded defaults
log_level="info"
endpoint="https://api.example.com"
# 2. Optional config
config_file="${XDG_CONFIG_HOME:-$HOME/.config}/mytool/config"
if [[ -f "$config_file" ]]; then
    # shellcheck source=/dev/null
    . "$config_file"
fi
# 3. Command-line overrides
while [[ $# -gt 0 ]]; do
    case "$1" in
        --log-level)
            log_level=$2; shift 2 ;;
        --endpoint)
            endpoint=$2; shift 2 ;;
        # ...
        *)
            break ;;
    esac
done
# Now use $log_level, $endpoint

Config file example (~/.config/mytool/config): 

log_level="debug"
endpoint="https://staging-api.example.com"

Testing and Linting Bash Tools

Robust tools need at least basic automated checks.

Using `shellcheck`

shellcheck is a static analyzer for shell scripts: 

shellcheck mytool

It catches:

You can run it in CI or as a pre-commit hook.

Simple Test Patterns

For small tools, basic tests can be simple scripts: 

#!/usr/bin/env bash
set -euo pipefail
# Example regression test
output=$(./mytool --version)
if [[ "$output" != "mytool 1.0"* ]]; then
    echo "Version output mismatch: $output" >&2
    exit 1
fi

Or use bats (Bash Automated Testing System) if you want a more structured test framework.

Performance Considerations for Bash Tools

Bash isn’t for heavy computation, but you can keep tools snappy:

Prefer Built-ins When It’s Clear

Examples: 

  (( count++ ))
  len=${#var}
  if [[ $var == foo* ]]; then
      ...
  fi

Avoid Forks in Loops

Bad: 

for f in *; do
    size=$(stat -c '%s' "$f")
    echo "$f $size"
done

Better, where possible: 

stat -c '%n %s' *   # single stat invocation

Or at least limit commands in the inner loop.

Packaging and Distribution of Bash Tools

Once a Bash tool is stable, you may want to distribute it.

Simple: Single Script Download

Host your script somewhere, then users can: 

curl -fsSL https://example.com/mytool -o ~/.local/bin/mytool
chmod +x ~/.local/bin/mytool

Optionally add a --version option and a self-update mechanism (carefully).

Slightly More Advanced: Tarball or Git Repo

Example install.sh: 

#!/usr/bin/env bash
set -euo pipefail
prefix=${1:-/usr/local}
install -Dm755 bin/mytool "$prefix/bin/mytool"
echo "Installed to $prefix/bin/mytool"

Distribution as proper system packages (deb, rpm) is usually done with other tooling and is beyond the scope of this chapter.

Example: A Realistic Bash-Based Tool

To tie it together, here’s a simplified log-filtering tool.

Requirements: 

#!/usr/bin/env bash
set -euo pipefail
level=""
input_file=""
usage() {
    cat <<EOF
Usage: ${0##*/} [OPTIONS] [FILE]
Filter log lines by level.
Options:
  -l, --level LEVEL   Log level to match (e.g., INFO, WARN, ERROR)
  -h, --help          Show this help and exit
If FILE is omitted, read from standard input.
EOF
}
die() {
    echo "Error: $*" >&2
    exit 1
}
main() {
    # Parse options
    while [[ $# -gt 0 ]]; do
        case "$1" in
            -l|--level)
                [[ $# -ge 2 ]] || die "Missing value for $1"
                level=$2
                shift 2
                ;;
            -h|--help)
                usage
                exit 0
                ;;
            --)
                shift
                break
                ;;
            -*)
                die "Unknown option: $1"
                ;;
            *)
                input_file=$1
                shift
                ;;
        esac
    done
    [[ -n "$level" ]] || die "Log level is required (use -l LEVEL)"
    # Choose input source
    if [[ -n "$input_file" ]]; then
        [[ -r "$input_file" ]] || die "Cannot read: $input_file"
        input=(<"$input_file")
    else
        input=(</dev/stdin)
    fi
    # Use grep with a precise pattern
    # Example log format: "2025-01-01 12:00:00 [LEVEL] message"
    grep -- "\[$level\]" "${input[@]}"
}
main "$@"

This tool:

You can extend it with subcommands, configuration, or more complex filtering as needed.

Bash-based tools are at their best when they:

The patterns in this chapter are directly reusable for building your own serious Bash command-line tools.

Previous
Up
Next

Views: 124

Comments

Please login to add a comment.

Don't have an account? Register now!

© 2025 Kahibaro

Visitors: 5322 / 24h

Impressum / Imprint

Datenschutz / Data Policy