rgrep 🦀

Copyright (c) 2026 Francesco Tinti francesco.tinti@activemind.it AI-assisted port — Architect: Claude Opus 4.7 (Anthropic) · Implementer: Gemini Antigravity (Google)

rgrep is a modern, modular, and blazingly fast clone of the original GNU grep, completely rewritten in Rust.

This project started as an educational experiment to translate a minimal 30-line C implementation of grep into Rust, and iteratively evolved into a 100% compliant, feature-rich text search engine capable of matching almost all major GNU grep functionalities.

🚀 Features

rgrep uses Rust's heavily optimized regex and globset crates to provide lightning-fast text processing, alongside modern features like memory-mapping and robust file-system traversal.

Supported GNU Grep Flags:

Pattern & Regex Control

• -e, --regexp: Specify multiple patterns (OR logic).
• -f, --file: Read patterns line by line from a file.
• -F, --fixed-strings: Interpret patterns as literal strings instead of Regex.
• -i, --ignore-case: Ignore case distinctions.
• --no-ignore-case: Do not ignore case distinctions.
• -w, --word-regexp: Match whole words only.
• -x, --line-regexp: Match exactly the whole line.
• -E, -G, -P: Compatibility stubs for Extended, Basic, and Perl regex.

Context Control

• -A, --after-context=NUM: Print NUM lines of trailing context.
• -B, --before-context=NUM: Print NUM lines of leading context.
• -C, --context=NUM: Print NUM lines of output context.
• --group-separator=SEP: Use custom separator between context blocks (default: --).
• --no-group-separator: Do not use a group separator.

File & Directory Traversal

• -r, --recursive: Read all files under each directory, recursively.
• -R, --dereference-recursive: Recursive search, following symbolic links.
• -d, --directories=ACTION: Handle directories (read, skip).
• -D, --devices=ACTION: Handle devices (read, skip).
• --include=GLOB: Search only files matching GLOB.
• --exclude=GLOB: Skip files matching GLOB.
• --exclude-dir=GLOB: Exclude entire directories matching GLOB.
• --exclude-from=FILE: Read exclude globs from a file.

Output Formatting

• -n, --line-number: Prefix each line with its 1-based line number.
• -b, --byte-offset: Print the 0-based byte offset of each line.
• -H, --with-filename: Print the file name for each match.
• -h, --no-filename: Suppress the prefixing of file names.
• -l, --files-with-matches: Print only names of FILEs containing matches.
• -L, --files-without-match: Print only names of FILEs containing no matches.
• -o, --only-matching: Print only the matched parts of a line.
• -c, --count: Print a count of matching lines for each file.
• -T, --initial-tab: Line up tabs for easier reading.
• --color: Highlight matches and file names using ANSI colors.

Low Level & Binary

• -z, --null-data: Read lines separated by a zero byte (NULL) instead of newline.
• -Z, --null: Output a zero byte after the file name instead of a colon.
• -a, --text: Process a binary file as if it were text.
• -U, --binary: Treat the files as binary.
• --binary-files=TYPE: Treat binary files as TYPE.

Optimizations & Limits

• -m, --max-count=NUM: Stop reading a file after NUM matching lines.
• --mmap: Use low-level OS memory-mapped I/O to read input files (can improve performance on large files).
• --line-buffered: Flush output on every line.
• -q, --quiet, --silent: Exit immediately with zero status if any match is found.
• -s, --no-messages: Suppress error messages about nonexistent or unreadable files.
• --label=LABEL: Use LABEL as the standard input file name prefix.

📦 Installation

To build rgrep from source, ensure you have the Rust toolchain installed.

# Clone the repository
git clone https://github.com/francescotinti/rgrep.git
cd rgrep

# Build for release
cargo build --release

# The compiled binary will be at:
# target/release/rgrep

You can install it directly into your local cargo path:

cargo install --path .

🛠️ Usage Example

Search recursively in the current directory for the exact word "TODO", ignoring case, printing line numbers, showing 2 lines of context, and highlighting matches in color:

rgrep -r -w -i -n -C 2 --color "TODO" .

Search using memory-mapping for maximum speed and stop after 5 matches:

rgrep --mmap -m 5 "foo" large_file.log

Verification status (2026-09-30)

The current suite has 137 manifest cases and 46 default / 48 PCRE2 Rust tests. GNU grep 3.12 and BSD grep were exercised locally on macOS. GNU runs perform 3072 generated comparisons each; BSD runs explicitly skip those properties. Two known GNU differences remain, both in -T alignment; the binary-match diagnostic follows GNU >= 3.5 (stderr) since Step BD. See GNU_VERIFICATION.md for pass/skip counts and commands. The first CI run passed both Linux builds and macOS with PCRE2. The default macOS job exposed a Bash 3 argument-handling issue, corrected in the workflow; the second run (36766955016) passed all four jobs. See publication status.

Ondata 7's delimiter optimization reduced elapsed time by roughly 9–12% on three amplified count workloads in alternating pre/post measurements; small CLI workloads were mostly flat. See PERF_REPORT.md for raw evidence, uncertainty and the small fixed-quiet regression.

Experiment history

The experiment documentation is split into what remains to do (TODO.md), what was delivered (DONE.md) and the architecture with its evolution (ARCHITECTURE.md), plus the full historical specifications, audit log, diaries and archived commit scripts under history/. See its publication notes for provenance.

📜 License

This project is open-source and built for educational and practical usage.