shoes/tests/rule_allocation.rs

use std::io::Write;

use shoes::rule_set::{Behavior, Format, RuleSource, YAML_PARSER_SCRATCH_BYTES};
use shoes::rules::{Compiler, Expression};
use shoes_test_support::measure::CountingAllocator;

#[global_allocator]
static ALLOCATOR: CountingAllocator = CountingAllocator::new();

fn peak<T>(run: impl FnOnce() -> T) -> (T, usize) {
    let baseline = ALLOCATOR.snapshot().live_bytes;
    ALLOCATOR.reset_peak();
    let result = run();
    (
        result,
        ALLOCATOR.snapshot().peak_bytes.saturating_sub(baseline),
    )
}

fn load_peak(bytes: &[u8], format: Format, precharged: usize) -> (std::io::Result<()>, usize) {
    let dir = tempfile::tempdir().unwrap();
    let path = dir.path().join("source");
    std::fs::write(&path, bytes).unwrap();
    let source = RuleSource {
        url: None,
        download: None,
        update_interval: None,
        path,
        format: Some(format),
        behavior: (format == Format::MihomoYaml).then_some(Behavior::Domain),
        code: (format == Format::GeositeDat).then(|| "TEST".into()),
        attributes: vec![],
    };
    let mut compiler = Compiler::default();
    compiler.charge(precharged, 0).unwrap();
    peak(|| shoes::rule_set::load_source(&source, &mut compiler).map(|_| ()))
}

fn message(tag: u32, bytes: &[u8]) -> Vec<u8> {
    let mut output = Vec::new();
    prost::encoding::encode_key(tag, prost::encoding::WireType::LengthDelimited, &mut output);
    prost::encoding::encode_varint(bytes.len() as u64, &mut output);
    output.extend_from_slice(bytes);
    output
}

// A separate test executable keeps allocator measurements isolated from parallel
// library tests. Inputs are allocated before measuring compiler-owned memory.
#[test]
fn adversarial_inputs_stay_within_their_compilation_reserves() {
    for yaml in [
        format!("payload: [{}]", "[],".repeat(350_000)),
        format!("{{[{}]: a}}", "a,".repeat(350_000)),
        format!("[[{}]]", ":,".repeat(350_000)),
    ] {
        let (result, used) = load_peak(yaml.as_bytes(), Format::MihomoYaml, 0);
        assert!(result.is_err());
        assert!(
            used < YAML_PARSER_SCRATCH_BYTES + yaml.len() * 2,
            "YAML peak: {used}"
        );
        println!("YAML bytes={}, peak={used}", yaml.len());
    }
    let json = format!(r#"{{"{}\n":0}}"#, "a".repeat(1024 * 1024));
    let (result, used) = load_peak(json.as_bytes(), Format::SingboxJson, 0);
    assert!(result.is_err());
    assert!(used < json.len() * 4 + 4096, "JSON peak: {used}");
    println!("JSON bytes={}, peak={used}", json.len());

    let available = 16 * 1024 * 1024;
    for json in [
        format!("[{}]", vec!["[]"; 200_000].join(",")),
        format!("[{}]", vec![r#"{"a":0}"#; 200_000].join(",")),
        format!(
            r#"{{"version":3,"rules":[{{"domain":[{}]}}]}}"#,
            vec![r#""a.example""#; 200_000].join(",")
        ),
    ] {
        let (result, used) = load_peak(
            json.as_bytes(),
            Format::SingboxJson,
            512 * 1024 * 1024 - available,
        );
        let error = result.unwrap_err();
        assert!(error.to_string().contains("budget"), "{error}");
        assert!(
            used < available,
            "JSON peak: {used}, available: {available}"
        );
        println!(
            "JSON tree bytes={}, peak={used}, available={available}",
            json.len()
        );
    }

    let count = 100_000;
    let mut domain = message(2, b"example.com");
    domain.extend_from_slice(&[0x1a, 0].repeat(count));
    let mut category = message(1, b"TEST");
    category.extend(message(2, &domain));
    let dat = message(1, &category);
    let (result, used) = load_peak(&dat, Format::GeositeDat, 0);
    result.unwrap();
    assert!(
        used < dat.len() * 10 + count * 160 + 4096,
        "DAT peak: {used}"
    );
    println!("DAT attributes={count}, peak={used}");

    let patterns = vec![format!("(?i){}", "a".repeat(60_000)); 8];
    let bytes: usize = patterns.iter().map(String::len).sum();
    let (_, used) = peak(|| Compiler::default().compile(Expression::DomainRegex(patterns)));
    let reserve = bytes * 512 + 64 * 1024 * 1024;
    assert!(used < reserve, "regex peak: {used}, reserve: {reserve}");
    println!("regex bytes={bytes}, peak={used}, reserve={reserve}");

    let inflated = vec![0; 8 * 1024 * 1024 + 1];
    let mut zlib =
        flate2::write::ZlibEncoder::new(b"SRS\x02".to_vec(), flate2::Compression::fast());
    zlib.write_all(&inflated).unwrap();
    let zstd = ruzstd::encoding::compress_to_vec(
        inflated.as_slice(),
        ruzstd::encoding::CompressionLevel::Fastest,
    );
    for (format, input, available) in [
        (Format::Srs, zlib.finish().unwrap(), 20 * 1024 * 1024),
        (Format::Mrs, zstd, 44 * 1024 * 1024),
    ] {
        let (result, used) = load_peak(&input, format, 512 * 1024 * 1024 - available);
        let error = result.unwrap_err();
        assert!(error.to_string().contains("budget"), "{format:?}: {error}");
        assert!(
            used < available,
            "{format:?} peak: {used}, available: {available}"
        );
        println!("{format:?} growth rejection peak={used}, available={available}");
    }
}