lean_ctx/core/patterns/

mod.rs

/// Pattern engine version. Bump when output format changes to maintain
/// shell-output determinism guarantees. Consumers (proofs, benchmarks)
/// can embed this version to detect pattern-breaking updates.
///
/// v2 (#936): `json_schema::compress` now prefers the lossless `json_crush`
/// form over the schema outline for redundant array-of-object payloads.
pub const PATTERN_ENGINE_VERSION: u32 = 2;

pub mod alembic;
pub mod ansible;
pub mod argocd;
pub mod artisan;
pub mod aws;
pub mod bazel;
pub mod buf;
pub mod bun;
pub mod cargo;
pub mod clang;
pub mod cmake;
pub mod composer;
pub mod cosign;
pub mod curl;
pub mod dbt;
pub mod deno;
pub mod deploy;
pub mod deps_cmd;
pub mod docker;
pub mod dotnet;
pub mod env_filter;
pub mod eslint;
pub mod fd;
pub mod find;
pub mod flutter;
pub mod flyway;
pub mod gem;
pub mod gh;
pub mod git;
pub mod glab;
pub mod golang;
pub mod grep;
pub mod grype;
pub mod helm;
pub mod jj;
pub mod json_schema;
pub mod just;
pub mod kubectl;
pub mod linkerd;
pub mod log_dedup;
pub mod ls;
pub mod make;
pub mod maven;
pub mod mise;
pub mod mix;
pub mod mlflow;
pub mod mypy;
pub mod mysql;
pub mod next_build;
pub mod ninja;
pub mod npm;
pub mod ollama;
pub mod php;
pub mod pip;
pub mod playwright;
pub mod pnpm;
pub mod poetry;
pub mod prettier;
pub mod prisma;
pub mod psql;
pub mod pulumi;
pub mod pytest;
pub mod ruby;
pub mod ruff;
pub mod semgrep;
pub mod spark;
pub mod swift;
pub mod swiftlint;
pub mod syft;
pub mod sysinfo;
pub mod systemd;
pub mod terraform;
pub mod test;
pub mod trivy;
pub mod typescript;
pub mod wget;
pub mod zig;

use crate::core::tokens::count_tokens;

pub fn compress_output(command: &str, output: &str) -> Option<String> {
    // Policy gate: protected commands (Passthrough + Verbatim) must
    // never be pattern-compressed. The caller (compress_if_beneficial
    // or ctx_shell::handle) should have already checked, but we
    // defend in depth here.
    let policy = crate::shell::output_policy::classify(command, &[]);
    if policy.is_protected() {
        return None;
    }

    let stripped;
    let clean_output = {
        let s = crate::core::compressor::strip_ansi(output);
        if s.len() < output.len() {
            stripped = s;
            &stripped
        } else {
            output
        }
    };

    if let Some(engine) = crate::core::filters::FilterEngine::load()
        && let Some(filtered) = engine.apply(command, clean_output)
    {
        return shorter_only(filtered, output);
    }

    // VCS history (git/jj/gh/glab/hg) is owned by its dedicated compressor. Its
    // lines look log-ish (one per commit) but are NOT application logs, so the
    // generic json/log/test fallbacks would mis-summarize them — e.g. truncating
    // an explicit `git log --oneline -40` to "last 15" or reading a commit
    // subject like "fix: pending_errors" as an error line. Return the dedicated
    // compressor's result directly — even when it is not shorter, so an already
    // compact oneline log is preserved verbatim (full history intact) instead of
    // being reshaped by a generic heuristic.
    if has_vcs_owner(command) {
        // The VCS compressor is authoritative and is kept even when it is not
        // strictly *shorter*, so a compact `git log --oneline` is preserved
        // verbatim (full history intact). It must still never return *more*
        // tokens than its input, though — a tiny adversarial `git status` body
        // could otherwise reshape into a one-token-larger summary and break the
        // never-inflate invariant. Allow equal (verbatim), reject only growth.
        return try_specific_pattern(command, clean_output)
            .filter(|c| !c.trim().is_empty())
            .filter(|c| !inflates_tokens(c, output));
    }

    if let Some(compressed) = try_specific_pattern(command, clean_output)
        && let Some(r) = shorter_only(compressed, output)
    {
        return Some(r);
    }

    if let Some(r) = json_schema::compress(clean_output)
        && let Some(r) = shorter_only(r, output)
    {
        return Some(r);
    }

    if let Some(r) = log_dedup::compress(clean_output)
        && let Some(r) = shorter_only(r, output)
    {
        return Some(r);
    }

    if let Some(r) = test::compress(clean_output)
        && let Some(r) = shorter_only(r, output)
    {
        return Some(r);
    }

    None
}

/// True for version-control commands whose output is authoritative under their
/// own compressor and must not be reinterpreted by the generic log/test
/// fallbacks (commit history is not application log output).
pub(crate) fn has_vcs_owner(command: &str) -> bool {
    let c = command.trim_start().to_ascii_lowercase();
    c.starts_with("git ")
        || c.starts_with("jj ")
        || c.starts_with("gh ")
        || c.starts_with("glab ")
        || c.starts_with("hg ")
}

/// Collapse whitespace into single spaces so comparisons align with logical word tokens.
fn normalize_shell_tokens(text: &str) -> String {
    text.split_whitespace().collect::<Vec<_>>().join(" ")
}

/// True when `compressed` carries *more* tokens than the raw `original` — i.e.
/// the transform inflated rather than compressed. Compared against the raw
/// (pre-ANSI-strip) output so the guard matches the public `compress_output`
/// invariant exactly: the returned text never tokenizes larger than its input.
fn inflates_tokens(compressed: &str, original: &str) -> bool {
    count_tokens(compressed) > count_tokens(original)
}

fn shorter_only(compressed: String, original: &str) -> Option<String> {
    let orig_n = normalize_shell_tokens(original);
    let comp_n = normalize_shell_tokens(&compressed);
    let ct_c = count_tokens(&comp_n);
    let ct_o = count_tokens(&orig_n);
    if ct_c < ct_o || (ct_c == ct_o && comp_n.len() < orig_n.len()) {
        Some(compressed)
    } else {
        None
    }
}

type PatternMatcher = fn(&str) -> bool;
type PatternHandler = fn(&str, &str) -> Option<String>;

/// Ordered prefix → compressor dispatch table (#660 CC reduction): first
/// matching entry wins, mirroring the previous `if c.starts_with(..) { return
/// ...; }` cascade exactly — including that a matched entry's `None` is
/// final, never falls through to a later entry. Registering a new CLI tool
/// is now a one-line addition here instead of a new branch in
/// `try_specific_pattern`.
const PATTERNS: &[(PatternMatcher, PatternHandler)] = &[
    (
        |c| c.starts_with("git "),
        |c, output| git::compress(c, output),
    ),
    (
        |c| c.starts_with("gh "),
        |c, output| gh::compress(c, output),
    ),
    (
        |c| c.starts_with("glab "),
        |c, output| glab::try_glab_pattern(c, output),
    ),
    (
        |c| c == "terraform" || c.starts_with("terraform "),
        |c, output| terraform::compress(c, output),
    ),
    (
        |c| c == "make" || c.starts_with("make "),
        |c, output| make::compress(c, output),
    ),
    (
        |c| c == "just" || c.starts_with("just "),
        |c, output| just::compress(c, output),
    ),
    (
        |c| {
            c.starts_with("mvn ")
                || c.starts_with("./mvnw ")
                || c.starts_with("mvnw ")
                || c.starts_with("gradle ")
                || c.starts_with("./gradlew ")
                || c.starts_with("gradlew ")
        },
        |c, output| maven::compress(c, output),
    ),
    (
        |c| c.starts_with("kubectl ") || c.starts_with("k "),
        |c, output| kubectl::compress(c, output),
    ),
    (
        |c| c.starts_with("helm "),
        |c, output| helm::compress(c, output),
    ),
    (
        |c| c.starts_with("pnpm "),
        |c, output| pnpm::compress(c, output),
    ),
    (
        |c| c.starts_with("bun ") || c.starts_with("bunx "),
        |c, output| bun::compress(c, output),
    ),
    (
        |c| c.starts_with("deno "),
        |c, output| deno::compress(c, output),
    ),
    (
        |c| c.starts_with("npm ") || c.starts_with("yarn "),
        |c, output| npm::compress(c, output),
    ),
    (
        |c| c.starts_with("cargo "),
        |c, output| cargo::compress(c, output),
    ),
    (
        |c| c.starts_with("docker ") || c.starts_with("docker-compose "),
        |c, output| docker::compress(c, output),
    ),
    (
        |c| c.starts_with("pip ") || c.starts_with("pip3 ") || c.starts_with("python -m pip"),
        |c, output| pip::compress(c, output),
    ),
    (
        |c| c.starts_with("mypy") || c.starts_with("python -m mypy") || c.starts_with("dmypy "),
        |c, output| mypy::compress(c, output),
    ),
    (
        |c| c.starts_with("pytest") || c.starts_with("python -m pytest"),
        |c, output| pytest::compress(c, output).or_else(|| test::compress(output)),
    ),
    (
        |c| c.starts_with("ruff "),
        |c, output| ruff::compress(c, output),
    ),
    (
        |c| {
            c.starts_with("eslint")
                || c.starts_with("npx eslint")
                || c.starts_with("biome ")
                || c.starts_with("stylelint")
        },
        |c, output| eslint::compress(c, output),
    ),
    (
        |c| c.starts_with("prettier") || c.starts_with("npx prettier"),
        |_c, output| prettier::compress(output),
    ),
    (
        |c| c.starts_with("go ") || c.starts_with("golangci-lint") || c.starts_with("golint"),
        |c, output| golang::compress(c, output),
    ),
    (
        |c| {
            c.starts_with("playwright")
                || c.starts_with("npx playwright")
                || c.starts_with("cypress")
                || c.starts_with("npx cypress")
        },
        |c, output| playwright::compress(c, output),
    ),
    (
        |c| c.starts_with("vitest") || c.starts_with("npx vitest") || c.starts_with("pnpm vitest"),
        |_c, output| test::compress(output),
    ),
    (
        |c| {
            c.starts_with("next ")
                || c.starts_with("npx next")
                || c.starts_with("vite ")
                || c.starts_with("npx vite")
                || c.starts_with("vp ")
                || c.starts_with("vite-plus ")
        },
        |c, output| next_build::compress(c, output),
    ),
    (
        |c| c.starts_with("tsc") || c.contains("typescript"),
        |_c, output| typescript::compress(output),
    ),
    (
        |c| {
            c.starts_with("rubocop")
                || c.starts_with("bundle ")
                || c.starts_with("rake ")
                || c.starts_with("rails test")
                || c.starts_with("rspec")
        },
        |c, output| ruby::compress(c, output),
    ),
    (|c| is_grep_family(c), |_c, output| grep::compress(output)),
    (
        |c| c.starts_with("find "),
        |_c, output| find::compress(output),
    ),
    (
        |c| c.starts_with("fd ") || c.starts_with("fdfind "),
        |_c, output| fd::compress(output),
    ),
    (
        |c| c.starts_with("ls ") || c == "ls",
        |_c, output| ls::compress(output),
    ),
    (
        |c| c.starts_with("curl "),
        |c, output| curl::compress_with_cmd(c, output),
    ),
    (
        |c| c.starts_with("wget "),
        |_c, output| wget::compress(output),
    ),
    (
        |c| c == "env" || c.starts_with("env ") || c.starts_with("printenv"),
        |_c, output| env_filter::compress(output),
    ),
    (
        |c| c.starts_with("dotnet "),
        |c, output| dotnet::compress(c, output),
    ),
    (
        |c| {
            c.starts_with("flutter ")
                || (c.starts_with("dart ") && (c.contains(" analyze") || c.ends_with(" analyze")))
        },
        |c, output| flutter::compress(c, output),
    ),
    (
        |c| {
            c.starts_with("poetry ")
                || c.starts_with("uv ")
                || c.starts_with("conda ")
                || c.starts_with("mamba ")
                || c.starts_with("pipx ")
        },
        |c, output| poetry::compress(c, output),
    ),
    (
        |c| c.starts_with("aws "),
        |c, output| aws::compress(c, output),
    ),
    (
        |c| c.starts_with("psql ") || c.starts_with("pg_"),
        |c, output| psql::compress(c, output),
    ),
    (
        |c| c.starts_with("mysql ") || c.starts_with("mariadb "),
        |c, output| mysql::compress(c, output),
    ),
    (
        |c| c.starts_with("prisma ") || c.starts_with("npx prisma"),
        |c, output| prisma::compress(c, output),
    ),
    (
        |c| c.starts_with("swift "),
        |c, output| swift::compress(c, output),
    ),
    (
        |c| c.starts_with("zig "),
        |c, output| zig::compress(c, output),
    ),
    (
        |c| c.starts_with("cmake ") || c.starts_with("ctest"),
        |c, output| cmake::compress(c, output),
    ),
    (
        |c| c.starts_with("ninja"),
        |c, output| ninja::compress(c, output),
    ),
    (
        |c| c.starts_with("ansible") || c.starts_with("ansible-playbook"),
        |c, output| ansible::compress(c, output),
    ),
    (
        |c| c.starts_with("composer "),
        |c, output| composer::compress(c, output),
    ),
    (
        |c| c.starts_with("php artisan") || c.starts_with("artisan "),
        |c, output| artisan::compress(c, output),
    ),
    (
        |c| c.starts_with("./vendor/bin/pest") || c.starts_with("pest "),
        |_c, output| artisan::compress("php artisan test", output),
    ),
    (
        |c| c.starts_with("mix ") || c.starts_with("iex "),
        |c, output| mix::compress(c, output),
    ),
    (
        |c| c.starts_with("bazel ") || c.starts_with("blaze "),
        |c, output| bazel::compress(c, output),
    ),
    (
        |c| c.starts_with("systemctl ") || c.starts_with("journalctl"),
        |c, output| systemd::compress(c, output),
    ),
    (
        |c| c.starts_with("jest") || c.starts_with("npx jest") || c.starts_with("pnpm jest"),
        |_c, output| test::compress(output),
    ),
    (
        |c| c.starts_with("mocha") || c.starts_with("npx mocha"),
        |_c, output| test::compress(output),
    ),
    (
        |c| c.starts_with("tofu "),
        |c, output| terraform::compress(c, output),
    ),
    (
        |c| c.starts_with("ps ") || c == "ps",
        |_c, output| sysinfo::compress_ps(output),
    ),
    (
        |c| c.starts_with("df ") || c == "df",
        |_c, output| sysinfo::compress_df(output),
    ),
    (
        |c| c.starts_with("du ") || c == "du",
        |_c, output| sysinfo::compress_du(output),
    ),
    (
        |c| c.starts_with("ping "),
        |_c, output| sysinfo::compress_ping(output),
    ),
    (
        |c| c.starts_with("jq ") || c == "jq",
        |_c, output| json_schema::compress(output),
    ),
    (
        |c| c.starts_with("hadolint"),
        |c, output| eslint::compress(c, output),
    ),
    (
        |c| c.starts_with("yamllint") || c.starts_with("npx yamllint"),
        |c, output| eslint::compress(c, output),
    ),
    (
        |c| c.starts_with("markdownlint") || c.starts_with("npx markdownlint"),
        |c, output| eslint::compress(c, output),
    ),
    (
        |c| c.starts_with("oxlint") || c.starts_with("npx oxlint"),
        |c, output| eslint::compress(c, output),
    ),
    (
        |c| c.starts_with("pyright") || c.starts_with("basedpyright"),
        |c, output| mypy::compress(c, output),
    ),
    (
        |c| c.starts_with("turbo ") || c.starts_with("npx turbo"),
        |c, output| npm::compress(c, output),
    ),
    (
        |c| c.starts_with("nx ") || c.starts_with("npx nx"),
        |c, output| npm::compress(c, output),
    ),
    (
        |c| c.starts_with("clang++ ") || c.starts_with("clang "),
        |c, output| clang::compress(c, output),
    ),
    (
        |c| {
            c.starts_with("gcc ")
                || c.starts_with("g++ ")
                || c.starts_with("cc ")
                || c.starts_with("c++ ")
        },
        |c, output| cmake::compress(c, output),
    ),
    // --- data domain (#657) ---
    (
        |c| c == "dbt" || c.starts_with("dbt "),
        |c, output| dbt::compress(c, output),
    ),
    (
        |c| c == "alembic" || c.starts_with("alembic "),
        |c, output| alembic::compress(c, output),
    ),
    (
        |c| c == "flyway" || c.starts_with("flyway "),
        |c, output| flyway::compress(c, output),
    ),
    (
        |c| c.starts_with("spark-submit") || c.starts_with("spark-sql") || c.starts_with("pyspark"),
        |c, output| spark::compress(c, output),
    ),
    // --- ai domain (#658) ---
    (
        |c| c == "ollama" || c.starts_with("ollama "),
        |c, output| ollama::compress(c, output),
    ),
    (
        |c| c.starts_with("mlflow "),
        |c, output| mlflow::compress(c, output),
    ),
    // --- security / supply-chain (#659) ---
    (
        |c| c.starts_with("semgrep "),
        |c, output| semgrep::compress(c, output),
    ),
    (
        |c| c.starts_with("trivy "),
        |c, output| trivy::compress(c, output),
    ),
    (
        |c| c.starts_with("grype "),
        |c, output| grype::compress(c, output),
    ),
    (
        |c| c.starts_with("syft "),
        |c, output| syft::compress(c, output),
    ),
    (
        |c| c.starts_with("cosign "),
        |c, output| cosign::compress(c, output),
    ),
    (
        |c| c.starts_with("swiftlint"),
        |c, output| swiftlint::compress(c, output),
    ),
    // --- vcs / toolchain (#660) ---
    (
        |c| c == "jj" || c.starts_with("jj "),
        |c, output| jj::compress(c, output),
    ),
    (
        |c| c == "mise" || c.starts_with("mise "),
        |c, output| mise::compress(c, output),
    ),
    (
        |c| c == "buf" || c.starts_with("buf "),
        |c, output| buf::compress(c, output),
    ),
    (
        |c| c.starts_with("gem "),
        |c, output| gem::compress(c, output),
    ),
    // --- edge / infra (#661) ---
    (
        |c| c == "pulumi" || c.starts_with("pulumi "),
        |c, output| pulumi::compress(c, output),
    ),
    (
        |c| c.starts_with("linkerd "),
        |c, output| linkerd::compress(c, output),
    ),
    (
        |c| c.starts_with("argocd "),
        |c, output| argocd::compress(c, output),
    ),
    (
        |c| {
            c == "vercel"
                || c.starts_with("vercel ")
                || c == "fly"
                || c.starts_with("fly ")
                || c.starts_with("flyctl ")
                || c.starts_with("wrangler ")
                || c.starts_with("skaffold ")
                || c.starts_with("supabase ")
        },
        |c, output| deploy::compress(c, output),
    ),
];

/// A grep-family search command, bare or with arguments (#980, #985).
///
/// The bare form matters: `proxy::compress::infer_command` synthesises `grep`
/// (no arguments) from tool names like `search_files`. The old `starts_with("grep ")`
/// required a trailing space and never matched, so those results fell through to the
/// terse pipeline which corrupted source identifiers.
fn is_grep_family(cmd: &str) -> bool {
    let first = cmd.split_whitespace().next().unwrap_or("");
    let base = first.rsplit('/').next().unwrap_or(first);
    matches!(
        base,
        "grep" | "egrep" | "fgrep" | "rg" | "ag" | "ack" | "ugrep" | "sift"
    )
}

pub fn try_specific_pattern(cmd: &str, output: &str) -> Option<String> {
    let cl = cmd.to_ascii_lowercase();
    let c = cl.as_str();

    PATTERNS
        .iter()
        .find(|(matches, _)| matches(c))
        .and_then(|(_, handle)| handle(c, output))
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn routes_git_commands() {
        let output = "On branch main\nnothing to commit";
        assert!(compress_output("git status", output).is_some());
    }

    #[test]
    fn vcs_path_never_inflates_tokens() {
        // Regression for the `compress_output_never_inflates_tokens` property
        // (Windows CI): the VCS branch returned the git compressor's reshaped
        // output without the `shorter_only` guard the other paths use, so this
        // tiny adversarial `git status` body grew 10 -> 11 tokens. The result
        // must now never tokenize larger than its input (None == use original).
        let output = " Abu a\nAa aa_A00A\n\n-";
        if let Some(compressed) = compress_output("git status", output) {
            assert!(
                count_tokens(&compressed) <= count_tokens(output),
                "VCS compress inflated: {} > {}",
                count_tokens(&compressed),
                count_tokens(output),
            );
        }
    }

    #[test]
    fn routes_cargo_commands() {
        let output = "   Compiling lean-ctx v2.1.1\n    Finished `release` profile [optimized] target(s) in 30.5s";
        assert!(compress_output("cargo build --release", output).is_some());
    }

    #[test]
    fn routes_npm_commands() {
        let output = "added 150 packages, and audited 151 packages in 5s\n\n25 packages are looking for funding\n  run `npm fund` for details\n\nfound 0 vulnerabilities";
        assert!(compress_output("npm install", output).is_some());
    }

    #[test]
    fn routes_docker_commands() {
        let output = "CONTAINER ID   IMAGE     COMMAND   CREATED   STATUS    PORTS     NAMES";
        // docker ps is Verbatim (via is_container_listing), so compress_output
        // correctly returns None (policy gate). docker build should still compress.
        assert!(compress_output("docker ps", output).is_none());
        let build_output =
            "Step 1/5 : FROM node:18\n ---> abc123\nStep 2/5 : COPY . .\nSuccessfully built def456";
        assert!(compress_output("docker build .", build_output).is_some());
    }

    #[test]
    fn routes_mypy_commands() {
        let output = "src/main.py:10: error: Missing return  [return]\nFound 1 error in 1 file (checked 3 source files)";
        assert!(compress_output("mypy .", output).is_some());
        assert!(compress_output("python -m mypy src/", output).is_some());
    }

    #[test]
    fn routes_pytest_commands() {
        let output = "===== test session starts =====\ncollected 5 items\ntest_main.py ..... [100%]\n===== 5 passed in 0.5s =====";
        assert!(compress_output("pytest", output).is_some());
        assert!(compress_output("python -m pytest tests/", output).is_some());
    }

    #[test]
    fn routes_data_domain() {
        let dbt =
            "20:14:02  Found 12 models\n20:14:20  Done. PASS=11 WARN=0 ERROR=1 SKIP=0 TOTAL=12";
        assert!(
            compress_output("dbt run", dbt).is_some(),
            "dbt routed+compressible"
        );
        let alembic = "INFO  [alembic.runtime.migration] Context impl PostgresqlImpl.\nINFO  [alembic.runtime.migration] Running upgrade  -> a1b2c3, create users table\nINFO  [alembic.runtime.migration] Running upgrade a1b2c3 -> d4e5f6, add email index";
        assert!(compress_output("alembic upgrade head", alembic).is_some());
        let flyway = "Flyway Community Edition 9.22.0 by Redgate\nDatabase: jdbc:postgresql://localhost/db\nMigrating schema \"public\" to version \"5 - add orders\"\nSuccessfully applied 1 migration to schema \"public\", now at version v5";
        assert!(compress_output("flyway migrate", flyway).is_some());
        let spark = "23/01/01 12:00:00 INFO SparkContext: Running Spark version 3.4.0\n23/01/01 12:00:01 INFO ResourceUtils: none configured\n23/01/01 12:00:10 INFO DAGScheduler: Job 0 finished: collect, took 5.1 s\n23/01/01 12:00:15 ERROR Executor: Exception in task 0.0";
        assert!(compress_output("spark-submit app.py", spark).is_some());
    }

    #[test]
    fn routes_ai_domain() {
        let ollama = "NAME              ID              SIZE      MODIFIED\nllama3.2:latest   a80c4f17acd5    2.0 GB    3 days ago\nqwen2.5-coder:7b  2b0496514337    4.7 GB    2 weeks ago";
        assert!(compress_output("ollama list", ollama).is_some());
        let mlflow = "2024/01/01 12:00:01 INFO mlflow.projects.backend.local: === Running command 'python train.py' ===\nCollecting numpy==1.26.0\nDownloading numpy-1.26.0.whl (18.2 MB)\n2024/01/01 12:00:30 INFO mlflow.projects: === Run (ID 'abc123def456') succeeded ===";
        assert!(compress_output("mlflow run .", mlflow).is_some());
    }

    #[test]
    fn routes_security_domain() {
        let trivy = "2024-01-01T12:00:00.000Z\tINFO\tscanning\nnginx:latest (debian 12.1)\n=====\nTotal: 45 (LOW: 20, HIGH: 8, CRITICAL: 2)";
        assert!(compress_output("trivy image nginx", trivy).is_some());
        let grype = "NAME       INSTALLED  FIXED-IN  TYPE  VULNERABILITY   SEVERITY\nlibssl1.1  1.1.1n     1.1.1w    deb   CVE-2023-1234   Critical\nzlib1g     1.2.11     1.2.13    deb   CVE-2022-5678   High";
        assert!(compress_output("grype nginx", grype).is_some());
        let syft = "NAME       VERSION    TYPE\nadduser    3.118      deb\napt        2.6.1      deb\nlodash     4.17.21    npm";
        assert!(compress_output("syft nginx", syft).is_some());
        let semgrep = "Scanning 120 files.\n  src/app.py\n     python.security.dangerous-subprocess-use\n        Detected subprocess.\n         42┆ subprocess.call(x)\nRan 450 rules on 120 files: 1 findings.";
        assert!(compress_output("semgrep scan", semgrep).is_some());
        let swiftlint = "Linting Swift files in current working directory\nLinting 'A.swift' (1/3)\nLinting 'B.swift' (2/3)\nLinting 'C.swift' (3/3)\n/path/A.swift:10:5: warning: Line Length Violation: Line should be 120 chars or less (line_length)\n/path/A.swift:22:1: warning: Trailing Whitespace Violation: no trailing whitespace (trailing_whitespace)\n/path/B.swift:5:1: error: Force Cast Violation: avoid force casts (force_cast)\n/path/C.swift:8:3: warning: Todo Violation: resolve TODOs (todo)\nDone linting! Found 4 violations, 1 serious in 3 files.";
        assert!(compress_output("swiftlint", swiftlint).is_some());
    }

    #[test]
    fn routes_vcs_toolchain_domain() {
        let jj = "@  qpvuntsm user@host.com 2024-01-01 12:00:00 1234abcd\n│  add feature x\n○  zzzzmmmm user@host.com 2024-01-01 11:00:00 main 5678efab\n│  initial commit\n~";
        assert!(compress_output("jj log", jj).is_some());
        let mise = "node    20.10.0  ~/.config/mise/config.toml\npython  3.12.0   ~/.tool-versions\nrust    1.75.0   ~/.config/mise/config.toml";
        assert!(compress_output("mise ls", mise).is_some());
        let buf_lines: Vec<String> = (0..30)
            .map(|i| format!("proto/f{i}.proto:{i}:1:Field name should be lower_snake_case here."))
            .collect();
        let buf = buf_lines.join("\n");
        assert!(compress_output("buf lint", &buf).is_some());
        let gem = "Fetching rails-7.1.0.gem\nSuccessfully installed activesupport-7.1.0\nSuccessfully installed rails-7.1.0\nParsing documentation for rails-7.1.0\nInstalling ri documentation for rails-7.1.0\nDone installing documentation for rails after 3 seconds\n2 gems installed";
        assert!(compress_output("gem install rails", gem).is_some());
        let uv = "Resolved 42 packages in 120ms\nDownloading numpy (18.2MiB)\n 100%|████████| 18.2M/18.2M [00:01<00:00, 15.3MiB/s]\nDownloading pandas (12.1MiB)\n 100%|████████| 12.1M/12.1M [00:00<00:00, 14.1MiB/s]\nPrepared 5 packages in 1.2s\nInstalled 5 packages in 30ms\n + numpy==1.26.0\n + pandas==2.1.0";
        assert!(compress_output("uv add pandas", uv).is_some());
    }

    #[test]
    fn routes_edge_infra_domain() {
        let pulumi = "Updating (dev):\n     Type   Name   Status\n +   pulumi:pulumi:Stack proj created\n +   aws:s3:Bucket b1 created\n +   aws:s3:Bucket b2 created\n +   aws:s3:Bucket b3 created\n +   aws:lambda:Function fn created\n\nOutputs:\n    url: \"https://x.example.com\"\n\nResources:\n    + 5 created\n    10 unchanged\n\nDuration: 35s";
        assert!(compress_output("pulumi up", pulumi).is_some());
        let linkerd = "kubernetes-api\n--------------\n√ can initialize the client\n√ can query the Kubernetes API\n√ is running the minimum kubectl version\n\nlinkerd-existence\n-----------------\n√ 'linkerd-config' config map exists\n× control plane pods are ready\n    some pods are not ready\n\nStatus check results are ×";
        assert!(compress_output("linkerd check", linkerd).is_some());
        let argocd = "Name:               argocd/myapp\nProject:            default\nSync Status:        Synced\nHealth Status:      Healthy\n\nGROUP  KIND  NAMESPACE  NAME  STATUS  HEALTH  HOOK  MESSAGE\n  Service ns s1 Synced Healthy\n  Service ns s2 Synced Healthy\n  Service ns s3 Synced Healthy\napps Deployment ns d1 OutOfSync Progressing";
        assert!(compress_output("argocd app get myapp", argocd).is_some());
        let vercel = "Vercel CLI 33.0.0\nInstalling dependencies...\nadded 420 packages in 12s\nBuilding...\nCompiling pages\nCollecting page data\nGenerating static pages\nProduction: https://my-app.vercel.app [45s]";
        assert!(compress_output("vercel deploy --prod", vercel).is_some());
        let wrangler = "wrangler 3.0.0\n-------------------\nyour worker has access to the following bindings:\n- KV Namespaces:\n  - CACHE: abc123\nTotal Upload: 1.2 MiB / gzip: 0.4 MiB\nUploaded my-worker (3.5 sec)\nPublished my-worker (1.2 sec)\n  https://my-worker.example.workers.dev";
        assert!(compress_output("wrangler deploy", wrangler).is_some());
    }

    #[test]
    fn unknown_command_returns_none() {
        assert!(compress_output("some-unknown-tool --version", "v1.0").is_none());
    }

    #[test]
    fn case_insensitive_routing() {
        let output = "On branch main\nnothing to commit";
        assert!(compress_output("Git Status", output).is_some());
        assert!(compress_output("GIT STATUS", output).is_some());
    }

    #[test]
    fn routes_vp_and_vite_plus() {
        let output = "  VITE v5.0.0  ready in 200 ms\n\n  -> Local:   http://localhost:5173/\n  -> Network: http://192.168.1.2:5173/";
        assert!(compress_output("vp build", output).is_some());
        assert!(compress_output("vite-plus build", output).is_some());
    }

    #[test]
    fn routes_bunx_commands() {
        let output = "1 pass tests\n0 fail tests\n3 skip tests\nDone 12ms\nsome extra line\nmore output here";
        let result = compress_output("bunx test", output);
        assert!(
            result.is_some(),
            "bunx should compress when output is large enough"
        );
        assert!(result.unwrap().contains("bun test: 1 passed"));
    }

    #[test]
    fn routes_deno_task() {
        let output = "Task dev deno run --allow-net server.ts\nListening on http://localhost:8000";
        assert!(try_specific_pattern("deno task dev", output).is_some());
    }

    #[test]
    fn routes_jest_commands() {
        let output = "PASS  tests/main.test.js\nTest Suites: 1 passed, 1 total\nTests:       5 passed, 5 total\nTime:        2.5 s";
        assert!(try_specific_pattern("jest", output).is_some());
        assert!(try_specific_pattern("npx jest --coverage", output).is_some());
    }

    #[test]
    fn routes_mocha_commands() {
        let output = "  3 passing (50ms)\n  1 failing\n\n  1) Array #indexOf():\n     Error: expected -1 to equal 0";
        assert!(try_specific_pattern("mocha", output).is_some());
        assert!(try_specific_pattern("npx mocha tests/", output).is_some());
    }

    #[test]
    fn routes_tofu_commands() {
        let output = "Initializing the backend...\nInitializing provider plugins...\nTerraform has been successfully initialized!";
        assert!(try_specific_pattern("tofu init", output).is_some());
    }

    #[test]
    fn routes_ps_commands() {
        let mut lines = vec!["USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND".to_string()];
        for i in 0..20 {
            lines.push(format!("user {i} 0.0 0.1 1234 123 ? S 10:00 0:00 proc_{i}"));
        }
        let output = lines.join("\n");
        assert!(try_specific_pattern("ps aux", &output).is_some());
    }

    #[test]
    fn routes_ping_commands() {
        let output = "PING google.com (1.2.3.4): 56 data bytes\n64 bytes from 1.2.3.4: icmp_seq=0 ttl=116 time=12ms\n3 packets transmitted, 3 packets received, 0.0% packet loss\nrtt min/avg/max/stddev = 11/12/13/1 ms";
        assert!(try_specific_pattern("ping -c 3 google.com", output).is_some());
    }

    #[test]
    fn routes_jq_to_json_schema() {
        let output = "{\"name\": \"test\", \"version\": \"1.0\", \"items\": [{\"id\": 1}, {\"id\": 2}, {\"id\": 3}, {\"id\": 4}, {\"id\": 5}, {\"id\": 6}, {\"id\": 7}, {\"id\": 8}, {\"id\": 9}, {\"id\": 10}]}";
        assert!(try_specific_pattern("jq '.items' data.json", output).is_some());
    }

    #[test]
    fn routes_linting_tools() {
        let lint_output = "src/main.py:10: error: Missing return\nsrc/main.py:20: error: Unused var\nFound 2 errors";
        assert!(try_specific_pattern("hadolint Dockerfile", lint_output).is_some());
        assert!(try_specific_pattern("oxlint src/", lint_output).is_some());
        assert!(try_specific_pattern("pyright src/", lint_output).is_some());
        assert!(try_specific_pattern("basedpyright src/", lint_output).is_some());
    }

    #[test]
    fn routes_fd_commands() {
        let output = "src/main.rs\nsrc/lib.rs\nsrc/util/helpers.rs\nsrc/util/math.rs\ntests/integration.rs\n";
        assert!(try_specific_pattern("fd --extension rs", output).is_some());
        assert!(try_specific_pattern("fdfind .rs", output).is_some());
    }

    #[test]
    fn routes_just_commands() {
        let output = "Available recipes:\n    build\n    test\n    lint\n";
        assert!(try_specific_pattern("just --list", output).is_some());
        assert!(try_specific_pattern("just build", output).is_some());
    }

    #[test]
    fn routes_ninja_commands() {
        let output = "[1/10] Compiling foo.c\n[10/10] Linking app\n";
        assert!(try_specific_pattern("ninja", output).is_some());
        assert!(try_specific_pattern("ninja -j4", output).is_some());
    }

    #[test]
    fn routes_clang_commands() {
        let output =
            "src/main.c:10:5: error: use of undeclared identifier 'foo'\n1 error generated.\n";
        assert!(try_specific_pattern("clang src/main.c", output).is_some());
        assert!(try_specific_pattern("clang++ -std=c++17 main.cpp", output).is_some());
    }

    #[test]
    fn routes_cargo_run() {
        let output = "   Compiling foo v0.1.0\n    Finished `dev` profile\nHello, world!";
        assert!(try_specific_pattern("cargo run", output).is_some());
    }

    #[test]
    fn routes_cargo_bench() {
        let output = "   Compiling foo v0.1.0\ntest bench_parse ... bench: 1234 ns/iter";
        assert!(try_specific_pattern("cargo bench", output).is_some());
    }

    #[test]
    fn routes_build_tools() {
        let build_output = "   Compiling foo v0.1.0\n    Finished release [optimized]";
        assert!(try_specific_pattern("gcc -o main main.c", build_output).is_some());
        assert!(try_specific_pattern("g++ -o main main.cpp", build_output).is_some());
    }

    #[test]
    fn routes_monorepo_tools() {
        let output = "npm warn deprecated inflight@1.0.6\nnpm warn deprecated rimraf@3.0.2\nadded 150 packages, and audited 151 packages in 5s\n\n25 packages are looking for funding\n  run `npm fund` for details\n\nfound 0 vulnerabilities";
        assert!(try_specific_pattern("turbo install", output).is_some());
        assert!(try_specific_pattern("nx install", output).is_some());
    }

    #[test]
    fn gh_api_passthrough_never_compresses() {
        let huge = "line\n".repeat(5000);
        assert!(
            compress_output("gh api repos/owner/repo/actions/jobs/123/logs", &huge).is_none(),
            "gh api must never be compressed, even for large output"
        );
        assert!(compress_output("gh api repos/owner/repo/actions/runs/123/logs", &huge).is_none());
    }

    #[test]
    fn gh_log_flags_passthrough() {
        let huge = "line\n".repeat(5000);
        assert!(compress_output("gh run view 123 --log-failed", &huge).is_none());
        assert!(compress_output("gh run view 123 --log", &huge).is_none());
    }

    #[test]
    fn gh_structured_commands_still_compress() {
        let output = "On branch main\nnothing to commit";
        assert!(try_specific_pattern("gh pr list", output).is_some());
        assert!(try_specific_pattern("gh run list", output).is_some());
    }
}