0xNaN/ngrep

grep plus word embeddings

What's novel

grep plus word embeddings

Code Analysis

15 files read · 4 rounds

A CLI tool that extends regular expressions with a semantic `~` operator, using word embeddings and cosine similarity to match text by meaning rather than literal string matching, built on a fork of fancy-regex with a custom binary embedding storage format.

Strengths

Clean, surgical integration of neural matching into a proven regex engine via trait-based abstractions (NeuralMatcher/NeuralMatcherFactory), with a well-designed mmap-backed binary storage format and proper separation between the regex engine, embedding model, and CLI layers.

Weaknesses

Limited test coverage for the neural matching path (no integration tests for `~` operator end-to-end), the `max_codepoints` hard limit of 32 is arbitrary and undocumented, and the per-token embedding lookup approach means multi-word semantic matching is not truly supported despite the `+` quantifier syntax.

Score Breakdown

Innovation

5 (25%)

Craft

68 (35%)

Traction

15 (15%)

Scope

78 (25%)

Signal breakdown

Innovation

Not Fork+1

Code Novelty+1

Concept Novelty+2

Craft

Ci-3

Tests+8

Polish+1

Releases+3

Has License+0

Code Quality+22

Readme Quality+15

Recent Activity+7

Structure Quality+5

Commit Consistency+5

Has Dependency Mgmt+5

Traction

Forks+0

Stars+12

Hn Points+0

Watchers+3

Early Traction+0

Devto Reactions+0

Community Contribs+0

Scope

Commits+8

Languages+3

Subsystems+13

Bloat Penalty+0

Completeness+6

Contributors+5

Authored Files+12

Readme Code Match+3

Architecture Depth+7

Implementation Depth+8

Evidence

Commits

Contributors

Files

Active weeks

TestsCI/CDREADMELicenseContributing

Repository

Language

Rust

Stars

Forks

License

—