notagrep

"Никакого кофе в мире не хватит, чтобы ускорить grep"

A fast grep alternative written in C, optimized for speed through algorithmic efficiency and SIMD vectorization.

Warning

This is alpha freaking ware and not ready for real usage.... yet

Installation

make

Quickstart

# Basic literal string search
notagrep "hello" file.txt

# Fixed string search (no regex)
notagrep -F "hello world" file.txt

# Regex search (default)
notagrep "error|warning" file.txt

# Case insensitive search
notagrep -i "TODO" .

# Count matches
notagrep -c "function" src/

# List files with matches
notagrep -l "import" .

Regex Support

Supported Features

• Alternation: a|b|c
• Character classes: [a-z], [^0-9]
• Quantifiers: *, +, ?
• Anchors: ^ (start of line), $ (end of line)
• Dot: . (any character)
• Grouping: (...)

NOT Supported

• Backreferences (\1, \2, etc.)
• Lookahead/lookbehind ((?=...), (?<=...))
• Named groups ((?P<name>...))
• Unicode character classes (\p{...})

Key Optimizations

• SIMD vectorization (NEON on ARM64, AVX2 on x86-64)
• Aho-Corasick automaton for multi-pattern matching
• Boyer-Moore-Horspool with unrolled skip loop
• Inner literal extraction for regex prefiltering
• Rare byte frequency optimization
• Parallel file processing with thread pool

Algorithm References

Clean-room implementations based on:

• Aho-Corasick: "Efficient String Matching: An Aid to Bibliographic Search" (1975) by Alfred Aho and Margaret Corasick
• Boyer-Moore: "A Fast String Searching Algorithm" (1977) by Robert Boyer and J Strother Moore
• Commentz-Walter hybrid: combines Aho-Corasick trie with Boyer-Moore skipping

License

See the source files for licensing information.