Sensitive-rs

English | 简体中文

A high-performance Rust crate for multi-pattern string matching, validation, filtering, and replacement.

Features

• Find all sensitive words: find_all
• First match with metadata (Match): find_first_match
• Validate text contains sensitive words: validate
• Remove sensitive words: filter
• Replace sensitive words with a character: replace
• Multi-algorithm engine: Aho-Corasick, Wu-Manber, Regex
• Noise removal via configurable regex
• Variant detection (拼音、形似字) — pinyin plus a 50+ group shape-confusable map
• Parallel search with optional rayon support (parallel feature, enabled by default)
• LRU cache for hot queries
• Batch processing: find_all_batch
• Layered matching: find_all_layered
• Streaming processing: find_all_streaming
• Criterion benchmarks and runnable examples for release validation

Algorithm selection

The engine auto-selects based on vocabulary size:

Override with Filter::with_algorithm(...) or --algorithm on the CLI.

Platform Support

WASM

[dependencies]
sensitive-rs = { version = "1.3", default-features = false, features = ["wasm"] }

import init, { WasmFilter } from 'sensitive-rs';
await init();
const filter = new WasmFilter();
filter.addWord('赌博');
filter.findAll('含有赌博内容'); // ['赌博']
filter.loadWords('色情\n诈骗'); // bulk-load from in-memory text

no_std (embedded)

[dependencies]
sensitive-rs = { version = "1.3", default-features = false }

Core find_all / find_in / replace / filter work without std. Pinyin/shape variant detection, the LRU cache, and the file/network loaders require the std feature (on by default).

Async

[dependencies]
sensitive-rs = { version = "1.3", features = ["async-io"] }

#[tokio::main]
async fn main() -> std::io::Result<()> {
    use sensitive_rs::Filter;
    let mut filter = Filter::new();
    filter.load_word_dict_async("dict/dict.txt").await?;
    Ok(())
}

Installation

Add to your Cargo.toml:

[dependencies]
sensitive-rs = "1.3.0"

For environments that should avoid rayon (for example WASM or embedded targets), disable default features:

[dependencies]
sensitive-rs = { version = "1.3.0", default-features = false }

Quick Start

use sensitive_rs::Filter;

fn main() {
    let mut filter = Filter::new();
    filter.add_words(&["rust", "filter", "敏感词"]);

    let text = "hello rust, this is a filter demo 包含敏感词";
    let found = filter.find_all(text);
    println!("Found: {:?}", found);

    let cleaned = filter.replace(text, '*');
    println!("Cleaned: {}", cleaned);
}

Advanced Usage

Batch processing:

let texts = vec!["text1", "text2"];
let results = filter.find_all_batch(&texts);

Layered matching:

let layered = filter.find_all_layered("some long text");

Streaming large files:

use std::fs::File;
use std::io::BufReader;

let reader = BufReader::new(File::open("large.txt")?);
let stream_results = filter.find_all_streaming(reader)?;

CLI Usage

Install with the cli feature:

[dependencies]
sensitive-rs = { version = "1.3.0", features = ["cli"] }

Or install directly:

cargo install sensitive-rs --features cli

Both sensitive and sensitive-rs commands are available after installation.

Commands

# Find sensitive words
sensitive check "含有赌博和色情内容"

# Validate (exit 1 if sensitive words found)
sensitive validate "clean text"

# Replace sensitive words
sensitive replace '*' "含有赌博内容"

# Remove sensitive words
sensitive filter "含有赌博内容"

# Read from file
sensitive check --file input.txt

# Pipe from stdin
echo "text" | sensitive check

Options

• --dict <path> — custom dictionary file
• --dict-all — use extended dictionary (27k words)
• --algorithm <algo> — force algorithm: aho-corasick, wumanber, regex
• --variant — enable pinyin and shape variant detection
• --noise-pattern <regex> — custom noise removal regex
• --json — JSON output format
• --color — force colored output

Examples and Benchmarks

cargo run --example basic
cargo run --example batch
cargo run --example custom_dict
cargo run --example variant
cargo bench

Documentation

For detailed documentation, please refer to Documentation.

License

Licensed under either of

• Apache License, Version 2.0, LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0
• MIT license LICENSE-MIT or http://opensource.org/licenses/MIT

at your option.

Contribution

Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 or MIT license, shall be dual licensed as above, without any additional terms or conditions.