How to Use ripgrep (rg) Instead of grep

ripgrep (rg) is a line-oriented search tool that recursively searches directories for a regex pattern. It is faster than grep in almost every real-world benchmark, respects .gitignore rules by default, and has sensible defaults that cut down on flag-typing. If you spend any time searching codebases or log files from the command line, switching to rg is one of the easiest quality-of-life upgrades you can make.

Why ripgrep Is Faster

ripgrep is written in Rust and uses the regex crate, which compiles patterns to finite automata and avoids backtracking on most inputs. Beyond the regex engine, three things explain the speed advantage over GNU grep:

• Automatic parallelism: directory traversal runs across multiple threads by default.
• Smart file skipping: hidden files, binary files, and anything matched by a .gitignore, .ignore, or .rgignore file are skipped without you asking.
• Memory-mapped I/O: for large files, rg can read directly from mapped memory rather than buffering through the kernel.

On a typical Node.js or Python project directory, rg is 5–10× faster than grep -r because it never enters node_modules or .git — those are covered by .gitignore.

Installation

Debian / Ubuntu

sudo apt update && sudo apt install ripgrep

Fedora / RHEL 9+ / Rocky Linux

sudo dnf install ripgrep

On RHEL 8 and older Rocky 8, enable the EPEL repository first: sudo dnf install epel-release.

Arch Linux

sudo pacman -S ripgrep

Verify the install

rg --version

Output will look similar to: ripgrep 14.1.0 (rev ...). The exact version varies by distro.

Basic Search Syntax

The simplest form mirrors grep: a pattern and an optional path. Without a path, rg searches the current directory recursively.

# Search current directory recursively
rg 'TODO'

# Search a specific file
rg 'error' /var/log/syslog

# Search a specific directory
rg 'def main' ~/projects/myapp

Output includes the filename, line number, and a highlighted match. Colours are on by default when writing to a terminal.

Regex Syntax

ripgrep uses Rust regex syntax, which is close to PCRE but with some differences: lookaheads and backreferences are not supported by default. For most searches this never matters. Use -P (PCRE2 mode) if you need them — provided rg was compiled with PCRE2 support, which most distro packages include.

Common patterns

# Case-insensitive match
rg -i 'warning'

# Match whole words only
rg -w 'log'

# Fixed string (no regex interpretation — faster for literals)
rg -F 'user.name'

# Extended regex — already the default, but explicit
rg -e 'foo|bar'

# PCRE2 for lookahead
rg -P 'foo(?=bar)'

Anchors and character classes

# Lines starting with a hash (comments)
rg '^#' config.ini

# IP-address-like pattern
rg '\b\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}\b' access.log

# Match lines that end with a semicolon
rg ';$' src/main.c

gitignore Support and Controlling What Gets Searched

By default rg reads and respects these ignore files in order: .gitignore, .git/info/exclude, a global gitignore, .ignore, and .rgignore. This is the single biggest practical advantage over plain grep -r.

# Override gitignore — search everything including ignored paths
rg --no-ignore 'secret'

# Include hidden files and directories (e.g. dotfiles)
rg --hidden 'alias'

# Both: search absolutely everything
rg --no-ignore --hidden 'password'

# Add your own ignore file per-project
echo 'dist/' >> .rgignore

Useful Flags

Filtering by file type

ripgrep has a built-in type system. Use --type (-t) to restrict searches to known file extensions, or --type-not (-T) to exclude them.

# Only Python files
rg -t py 'import os'

# Exclude test files (JavaScript)
rg -T js 'console.log'

# List all known types
rg --type-list

Context lines

# 3 lines before and after each match
rg -C 3 'panic'

# Only lines before
rg -B 2 'ERROR'

# Only lines after
rg -A 5 'BEGIN CERTIFICATE'

Counting, filenames, and line numbers

# Count matches per file
rg -c 'TODO'

# Print only filenames that contain a match
rg -l 'FIXME'

# Suppress line numbers (they're on by default with color output)
rg --no-line-number 'debug'

Multiline mode

By default rg matches line by line. Enable multiline mode with -U to let . match newlines.

rg -U 'start.*?end' --multiline-dotall src/

Replace (non-destructive preview)

rg does not edit files in place, but the -r flag prints what a substitution would look like — useful before piping into sed.

# Preview replacement output
rg 'foo' -r 'bar' src/config.py

Output formatting and piping

# Null-delimited filenames for xargs
rg -l0 'TODO' | xargs -0 vim

# JSON output for scripting
rg --json 'error' app.log | jq '.data.lines.text'

Practical Recipes

# Find all functions named 'handle_*' across a Go project
rg 'func handle_\w+' --type go

# Search compressed logs (requires rg compiled with the 'search-zip' feature)
rg -z 'Out of memory' /var/log/

# Find TODO comments with the author name that follows
rg 'TODO\(\w+\)'

# Search only files modified in the last day (combine with find)
find . -mtime -1 | xargs rg 'migration'

Verification

Clone a small public repo (or use any local project) and compare timing against GNU grep to confirm rg is working correctly and faster:

time grep -r 'TODO' ~/projects/myapp --include='*.py' > /dev/null
time rg 'TODO' -t py ~/projects/myapp > /dev/null

Both should return the same results. rg should be measurably faster on any directory with more than a few hundred files.

Troubleshooting

rg skips files you expect to be searched

Run with --debug to see exactly which files are being skipped and why:

rg --debug 'pattern' 2>&1 | grep 'ignore\|skip'

The most common cause is an overly broad .gitignore entry. Use --no-ignore to bypass all ignore files as a diagnostic step.

No matches but you are sure the text is there

Check whether the file is binary. rg skips binary files by default. Add -a (--text) to force text mode:

rg -a 'pattern' suspicious_file

Pattern works in grep but not rg

GNU grep defaults to BRE/ERE; rg uses Rust regex syntax. Character class shortcuts like \w work in both, but POSIX classes like [:alpha:] need a different form. Try -P for PCRE2 mode if you are porting a complex grep command.