haz-cli 0.1.0

//! End-to-end tests for `haz cache clean` against per-scenario
//! fixtures under `tests/fixtures/cache_clean/`. Each fixture is
//! a minimal workspace; cache state under `.haz/cache/` is built
//! at runtime (via `Cache::store` for valid entries plus
//! hand-rolled scaffolding for the stale shapes) so the entries
//! remain valid as `CHAPTER_REVISION` evolves.
//!
//! The tests exercise the four AUX-022..AUX-027 branches the CLI
//! must surface:
//!
//! - AUX-022 rejection when no mode is supplied (neither CLI nor
//!   workspace) and the success-path under each individual mode.
//! - AUX-024 dry-run reporting + on-disk preservation.
//! - AUX-025 workspace-default fallback when the CLI is silent.
//! - The CFG-034 / CFG-035 grammar rejection at the CLI parser
//!   layer.

use std::collections::BTreeMap;
use std::fs;
use std::path::Path;
use std::process::Command;

use haz_cache::cache::Cache;
use haz_cache::key::CacheKey;
use haz_cache::key::prefix::CHAPTER_REVISION;
use haz_cache::layout;
use haz_cache::manifest::{HashFunctionLabel, Manifest};
use haz_cache::store::{StoreInputs, StoredOutput};
use haz_cli::EXIT_INTERNAL_OR_CONFIG_ERROR;
use haz_domain::settings::cache::HashAlgo;
use haz_vfs::StdFilesystem;
use tempfile::TempDir;

const HAZ_BIN: &str = env!("CARGO_BIN_EXE_haz");
const HASH_ALGO: HashAlgo = HashAlgo::Blake3;

fn fixture_tempdir(name: &str) -> TempDir {
    let src = Path::new(env!("CARGO_MANIFEST_DIR"))
        .join("tests/fixtures/cache_clean")
        .join(name);
    let dst = TempDir::new().expect("tempdir creation failed");
    copy_dir_recursive(&src, dst.path());
    dst
}

fn copy_dir_recursive(src: &Path, dst: &Path) {
    fs::create_dir_all(dst).unwrap();
    for entry in fs::read_dir(src).unwrap() {
        let entry = entry.unwrap();
        let src_path = entry.path();
        let dst_path = dst.join(entry.file_name());
        if entry.file_type().unwrap().is_dir() {
            copy_dir_recursive(&src_path, &dst_path);
        } else {
            fs::copy(&src_path, &dst_path).unwrap();
        }
    }
}

/// Run `haz cache clean <args>` against `workspace_root`. Returns
/// the captured `(stdout, stderr, exit_code)` tuple.
fn haz_cache_clean(workspace_root: &Path, args: &[&str]) -> (String, String, Option<i32>) {
    let mut cmd = Command::new(HAZ_BIN);
    cmd.args(["cache", "clean"]);
    cmd.args(args);
    cmd.current_dir(workspace_root);
    let output = cmd.output().expect("failed to spawn haz");
    (
        String::from_utf8(output.stdout).expect("haz stdout was not utf-8"),
        String::from_utf8(output.stderr).expect("haz stderr was not utf-8"),
        output.status.code(),
    )
}

fn build_cache(workspace_root: &Path) -> Cache<StdFilesystem> {
    Cache::new(StdFilesystem::new(), workspace_root, HASH_ALGO)
}

fn key_with_first_byte(first: u8) -> CacheKey {
    let mut bytes = [0u8; 32];
    bytes[0] = first;
    CacheKey::from_bytes(bytes)
}

/// Store a valid entry whose manifest carries `created_at_unix`.
/// Tests that exercise `--max-age` rely on `created_at = 0` so
/// the wall-clock `now` is guaranteed to lie past any reasonable
/// cutoff, evicting the entry under every `--max-age` value.
fn store_valid_entry_at(
    cache: &Cache<StdFilesystem>,
    key: &CacheKey,
    rel: &str,
    bytes: &[u8],
    created_at_unix: u64,
) {
    let workspace_root = cache.workspace_root().to_path_buf();
    let target = workspace_root.join(rel);
    let anchored = format!("/{rel}");
    fs::create_dir_all(target.parent().unwrap()).unwrap();
    fs::write(&target, bytes).unwrap();
    let outs = [StoredOutput {
        workspace_absolute_path: &anchored,
        on_disk_path: &target,
        mode: 0o644,
    }];
    cache
        .store(
            key,
            &StoreInputs {
                outputs: &outs,
                stdout: b"",
                stderr: b"",
                created_at_unix,
            },
        )
        .expect("store should succeed for a fresh key");
}

/// Hand-write a manifest whose `chapter_revision` does NOT match
/// the current value. The resulting entry is "objectively stale"
/// per CACHE-022 and counts under `--soft` only.
fn write_schema_stale_entry(cache: &Cache<StdFilesystem>, key: &CacheKey) {
    let entry_dir = layout::entry_dir(cache.cache_root(), key);
    fs::create_dir_all(&entry_dir).unwrap();
    let manifest = Manifest {
        chapter_revision: CHAPTER_REVISION.saturating_add(1),
        hash_function: HashFunctionLabel::Blake3,
        key: *key,
        outputs: vec![],
        stdout_len: 0,
        stderr_len: 0,
        stdout_hash: [0u8; 32],
        stderr_hash: [0u8; 32],
        exit_status: 0,
        created_at_unix: 0,
    };
    fs::write(
        layout::manifest_path(cache.cache_root(), key),
        manifest.to_json_bytes(),
    )
    .unwrap();
}

/// Create an orphan `<shard>/.tmp-<key>-<random>` directory (the
/// incomplete two-phase-store shape from CACHE-017).
fn write_orphan_tmp_dir(cache: &Cache<StdFilesystem>, key: &CacheKey, random: &str) {
    let tmp = layout::tmp_entry_dir(cache.cache_root(), key, random);
    fs::create_dir_all(&tmp).unwrap();
    fs::write(tmp.join("partial.bin"), b"scratch").unwrap();
}

/// Create an orphan `<cache_root>/.restore-<key>-<random>`
/// staging directory (the interrupted-restore shape).
fn write_orphan_restore_dir(cache: &Cache<StdFilesystem>, key: &CacheKey, random: &str) {
    let staging = layout::restore_staging_dir(cache.cache_root(), key, random);
    fs::create_dir_all(&staging).unwrap();
    fs::write(staging.join("leftover.bin"), b"scratch").unwrap();
}

/// Pre-populate `<workspace_root>/.haz/cache/` with:
///
/// - 2 well-formed entries at `created_at_unix = 0`.
/// - 1 schema-mismatched entry.
/// - 1 orphan `.tmp-` directory.
/// - 1 orphan `.restore-` directory.
fn populate_mixed_cache(workspace_root: &Path) {
    let cache = build_cache(workspace_root);
    let key_a = key_with_first_byte(0xAA);
    store_valid_entry_at(&cache, &key_a, "lib/out_a", b"hello", 0);
    let key_b = key_with_first_byte(0xBB);
    store_valid_entry_at(&cache, &key_b, "lib/out_b", b"world!", 0);
    let key_c = key_with_first_byte(0xCC);
    write_schema_stale_entry(&cache, &key_c);
    let key_d = key_with_first_byte(0xDD);
    write_orphan_tmp_dir(&cache, &key_d, "tmpsuffix");
    let key_e = key_with_first_byte(0xEE);
    write_orphan_restore_dir(&cache, &key_e, "rsfxsuffix");
}

/// Recursively snapshot every regular file under `root`,
/// returning a `BTreeMap<relative_path, bytes>`. Symlinks and
/// other non-file kinds are skipped.
fn snapshot_dir(root: &Path) -> BTreeMap<String, Vec<u8>> {
    let mut out = BTreeMap::new();
    snapshot_into(root, root, &mut out);
    out
}

fn snapshot_into(base: &Path, dir: &Path, out: &mut BTreeMap<String, Vec<u8>>) {
    let Ok(entries) = fs::read_dir(dir) else {
        return;
    };
    for entry in entries.flatten() {
        let path = entry.path();
        let Ok(ft) = entry.file_type() else { continue };
        if ft.is_file() {
            let rel = path
                .strip_prefix(base)
                .map_or_else(|_| path.display().to_string(), |p| p.display().to_string());
            let bytes = fs::read(&path).unwrap_or_default();
            out.insert(rel, bytes);
        } else if ft.is_dir() {
            snapshot_into(base, &path, out);
        }
    }
}

// --- AUX-022: no-mode rejection ---

#[test]
fn aux_022_no_mode_rejects_with_diagnostic() {
    // Bare `haz cache clean` against a workspace lacking
    // `cache.clean.*` defaults MUST be rejected per AUX-022.
    let tmp = fixture_tempdir("fresh_workspace");
    let (_, stderr, code) = haz_cache_clean(tmp.path(), &[]);
    assert_ne!(
        code,
        Some(0),
        "no-mode invocation must exit non-zero; stderr: {stderr}",
    );
    assert!(
        stderr.contains("at least one mode"),
        "AUX-022 diagnostic must surface the requirement: {stderr}",
    );
}

// --- AUX-024: --soft happy path on a populated cache ---

#[test]
fn clean_soft_reports_aux_024_categories_on_populated_cache() {
    let tmp = fixture_tempdir("populated_with_mixed_entries");
    populate_mixed_cache(tmp.path());

    let (stdout, stderr, code) = haz_cache_clean(tmp.path(), &["--soft"]);
    assert_eq!(code, Some(0), "expected exit 0; stderr: {stderr}");
    // Fixture contributes: 1 schema-stale entry, 1 orphan tmp,
    // 1 orphan restore. The two well-formed entries survive.
    assert!(
        stdout.contains("entries inspected: 3"),
        "expected inspected=3 (2 well-formed + 1 corrupt): {stdout}",
    );
    assert!(
        stdout.contains("evicted --soft: 1"),
        "expected one soft-evicted entry (the schema-stale one): {stdout}",
    );
    assert!(
        stdout.contains("evicted --max-age: 0"),
        "max-age inactive must read zero: {stdout}",
    );
    assert!(
        stdout.contains("evicted --max-size: 0"),
        "max-size inactive must read zero: {stdout}",
    );
    assert!(
        stdout.contains("orphan tmp directories removed: 1"),
        "expected one orphan tmp removed: {stdout}",
    );
    assert!(
        stdout.contains("orphan restore directories removed: 1"),
        "expected one orphan restore removed: {stdout}",
    );

    // The well-formed entries survive on disk.
    let cache_root = tmp.path().join(".haz/cache");
    let key_a = key_with_first_byte(0xAA);
    assert!(layout::entry_dir(&cache_root, &key_a).exists());
}

// --- AUX-022: --max-age happy path with created_at=0 ---

#[test]
fn clean_max_age_evicts_well_formed_entries_strictly_older_than_cutoff() {
    let tmp = fixture_tempdir("populated_with_mixed_entries");
    populate_mixed_cache(tmp.path());

    // 1s cutoff; every well-formed entry has created_at=0 and is
    // well past `now - 1s`, so both get evicted.
    let (stdout, stderr, code) = haz_cache_clean(tmp.path(), &["--max-age", "1s"]);
    assert_eq!(code, Some(0), "expected exit 0; stderr: {stderr}");
    assert!(
        stdout.contains("evicted --max-age: 2"),
        "expected both well-formed entries evicted by max-age: {stdout}",
    );
    // Corrupt entry stays (AUX-023: --max-age ignores corrupt entries).
    let cache_root = tmp.path().join(".haz/cache");
    let key_c = key_with_first_byte(0xCC);
    assert!(
        layout::entry_dir(&cache_root, &key_c).exists(),
        "schema-stale entry must NOT be evicted by --max-age alone",
    );
    // Orphan tmp / restore dirs stay (only --soft reclaims them).
    let key_d = key_with_first_byte(0xDD);
    let key_e = key_with_first_byte(0xEE);
    assert!(layout::tmp_entry_dir(&cache_root, &key_d, "tmpsuffix").exists());
    assert!(layout::restore_staging_dir(&cache_root, &key_e, "rsfxsuffix").exists());
}

// --- AUX-024: dry-run preserves disk state and prints a plan ---

#[test]
fn aux_024_dry_run_does_not_modify_disk_and_emits_plan() {
    let tmp = fixture_tempdir("populated_with_mixed_entries");
    populate_mixed_cache(tmp.path());

    let cache_root = tmp.path().join(".haz/cache");
    let before = snapshot_dir(&cache_root);

    let (stdout, stderr, code) =
        haz_cache_clean(tmp.path(), &["--soft", "--max-age", "1s", "--dry-run"]);
    assert_eq!(code, Some(0), "expected exit 0; stderr: {stderr}");
    assert!(
        stdout.contains("dry-run: AUX-024 plan only"),
        "dry-run banner missing: {stdout}",
    );
    assert!(
        stdout.contains("would evict --soft:"),
        "dry-run mode labels: {stdout}",
    );
    assert!(
        stdout.contains("bytes projected:"),
        "dry-run bytes label: {stdout}",
    );
    assert!(
        stdout.contains("dry-run plan:"),
        "per-entry plan header: {stdout}",
    );
    // The per-entry plan must list each evicted entry's mode.
    assert!(
        stdout.contains("mode=--soft"),
        "dry-run plan must include soft mode entries: {stdout}",
    );
    assert!(
        stdout.contains("mode=--max-age"),
        "dry-run plan must include max-age mode entries: {stdout}",
    );

    let after = snapshot_dir(&cache_root);
    assert_eq!(
        before, after,
        "dry-run must leave the cache root byte-identical",
    );
}

// --- AUX-025: workspace default satisfies AUX-022 ---

#[test]
fn aux_025_workspace_default_max_age_satisfies_aux_022() {
    // The fixture's `.haz/workspace.yml` carries
    // `cache.clean.max_age: 1s`. A bare `haz cache clean`
    // succeeds because AUX-025 lets the workspace default
    // supply the mode AUX-022 requires.
    let tmp = fixture_tempdir("with_workspace_max_age_default");
    let cache = build_cache(tmp.path());
    let key = key_with_first_byte(0xAA);
    store_valid_entry_at(&cache, &key, "lib/out", b"x", 0);

    let (stdout, stderr, code) = haz_cache_clean(tmp.path(), &[]);
    assert_eq!(code, Some(0), "expected exit 0; stderr: {stderr}");
    assert!(
        stdout.contains("evicted --max-age: 1"),
        "expected the entry to be evicted under the workspace default: {stdout}",
    );
    // The entry is gone from disk.
    let cache_root = tmp.path().join(".haz/cache");
    assert!(!layout::entry_dir(&cache_root, &key).exists());
}

// --- CFG-034: malformed CLI duration rejected at parse time ---

#[test]
fn cfg_034_bare_integer_max_age_rejected_at_cli_parse_time() {
    let tmp = fixture_tempdir("fresh_workspace");
    let (_, stderr, code) = haz_cache_clean(tmp.path(), &["--max-age", "90"]);
    // CFG-034 / AUX-027: malformed-duration rejection MUST land in
    // the EXEC-021 internal / configuration-error class. The
    // task-failure class (`1`) MUST NOT be returned.
    assert_eq!(
        code,
        Some(EXIT_INTERNAL_OR_CONFIG_ERROR),
        "malformed --max-age must use the config-error class; stderr: {stderr}",
    );
    // The parser does not run the workspace load before failing;
    // the diagnostic is bpaf's, anchored at the flag value.
    assert!(
        !stderr.is_empty(),
        "missing diagnostic for malformed duration: {stderr}",
    );
}

// --- AUX-027: cache clean exit-code class mapping ---

#[test]
fn aux_027_success_exits_zero_and_does_not_use_task_failure_class() {
    // AUX-027: `haz cache clean` MUST exit 0 on a successful run,
    // including when the eviction set is empty. The task-failure
    // class (1) MUST NOT be returned.
    const EXIT_TASK_FAILURE: i32 = 1;

    let tmp = fixture_tempdir("fresh_workspace");
    let (_, stderr, code) = haz_cache_clean(tmp.path(), &["--soft"]);
    assert_eq!(
        code,
        Some(0),
        "AUX-027: --soft on a fresh workspace exits 0; stderr: {stderr}",
    );
    assert_ne!(code, Some(EXIT_TASK_FAILURE));
}

#[test]
fn aux_027_no_mode_rejection_uses_config_error_class() {
    // AUX-027: a bare `haz cache clean` (no mode flag, no
    // workspace default) MUST be rejected per AUX-022. The exit
    // code MUST be in the EXEC-021 internal / configuration-error
    // class.
    let tmp = fixture_tempdir("fresh_workspace");
    let (_, stderr, code) = haz_cache_clean(tmp.path(), &[]);
    assert_eq!(
        code,
        Some(EXIT_INTERNAL_OR_CONFIG_ERROR),
        "AUX-027: no-mode rejection MUST use the config-error class; stderr: {stderr}",
    );
}

#[test]
fn aux_027_malformed_duration_uses_config_error_class() {
    // AUX-027 / CFG-034: a malformed `--max-age` value MUST be
    // rejected with a typed CLI error and the exit code MUST land
    // in the EXEC-021 internal / configuration-error class. This
    // is the test that originally surfaced the bpaf-induced
    // mis-mapping (parse failures exited with `1`, the
    // task-failure class); regression-locks the
    // `apps/haz-cli/src/main.rs` `ParseFailure` mapping.
    let tmp = fixture_tempdir("fresh_workspace");
    let (_, stderr, code) = haz_cache_clean(tmp.path(), &["--max-age", "90"]);
    assert_eq!(
        code,
        Some(EXIT_INTERNAL_OR_CONFIG_ERROR),
        "AUX-027: malformed duration MUST use the config-error class; stderr: {stderr}",
    );
}