haz_cli/commands/

cache.rs

//! `haz cache` subcommands per `CACHE-021` / `CACHE-022` /
//! `AUX-017`..`AUX-027`.
//!
//! - [`clear`] backs `haz cache clear`: an unconditional reset of
//!   `<workspace_root>/.haz/cache`. Subsequent lookups against
//!   the same workspace are misses until new entries are stored.
//! - [`clean`] backs `haz cache clean [--soft] [--max-age <D>]
//!   [--max-size <B>] [--dry-run]`: composable reclamation of
//!   objectively-stale entries plus age- / size-bounded
//!   eviction (`CACHE-022`, `AUX-022`..`AUX-027`). The executor
//!   merges CLI flags with workspace `cache.clean.*` defaults
//!   (`AUX-025`) and enforces the "at least one mode" rule
//!   (`AUX-022`).
//! - [`info`] backs `haz cache info`: a read-only walk that
//!   classifies entries and reports counts, total bytes, and
//!   schema-prefix breakdown without mutating the cache root.
//!
//! All three subcommands first load the workspace via
//! [`crate::workspace_load::load_workspace`] to locate the
//! workspace root. The cache subcommand surface is independent of
//! the bearing-project concept (`EXEC-022`) and ignores cwd
//! beyond using it as the discovery starting point. A future
//! revision MAY expose a discovery-only fast path so the cache
//! commands succeed against a workspace with broken `haz.yml`
//! files; for v1, a parseable workspace is required.

use std::io::{self, Write};
use std::path::Path;
use std::time::{SystemTime, UNIX_EPOCH};

use haz_cache::cache::Cache;
use haz_cache::clean::{CleanOptions, CleanReport, EvictionMode};
use haz_cache::info::CacheInfoReport;
use haz_cache::manifest::HashFunctionLabel;
use haz_domain::settings::cache::HashAlgo;
use haz_domain::settings::cache_clean::CacheCleanSettings;
use haz_vfs::StdFilesystem;
use miette::{IntoDiagnostic, Result, WrapErr, miette};

use crate::cli::CacheCleanArgs;
use crate::workspace_load::load_workspace;

/// The hash algorithm haz-cli uses when probing manifests. MUST
/// match the algorithm used by `commands::run::execute` so the
/// schema-mismatch arm of `CACHE-022` only fires for genuinely
/// out-of-band manifests, never for a cache the same `haz`
/// invocation could itself produce.
const HASH_ALGO: HashAlgo = HashAlgo::Blake3;

/// Execute `haz cache clear` against `cwd`.
///
/// Returns 0 on success. Errors propagate as miette diagnostics
/// (workspace-load failure, filesystem failure during the
/// recursive removal).
///
/// # Errors
///
/// Returns a [`miette::Report`] for workspace-load or
/// invalidation failures.
pub fn clear(cwd: &Path) -> Result<i32> {
    let (workspace, _graph) = load_workspace(cwd)?;
    let cache = build_cache(workspace.root.as_path());
    cache
        .clear()
        .into_diagnostic()
        .wrap_err("cache clear failed")?;
    let mut out = io::stdout().lock();
    writeln!(out, "cache cleared.")
        .into_diagnostic()
        .wrap_err("failed to write confirmation")?;
    Ok(0)
}

/// Execute `haz cache clean` against `cwd` with the supplied CLI
/// args.
///
/// Resolves the effective mode set by combining `args` with
/// workspace `cache.clean.*` defaults per `AUX-025` (CLI overrides
/// workspace), enforces the `AUX-022` "at least one mode" rule,
/// runs the unified [`Cache::clean`], and emits the `AUX-024`
/// report.
///
/// # Errors
///
/// Returns a [`miette::Report`] on workspace-load failure, on a
/// no-effective-mode invocation, or on a filesystem failure
/// during the eviction walk.
pub fn clean(cwd: &Path, args: &CacheCleanArgs) -> Result<i32> {
    let (workspace, _graph) = load_workspace(cwd)?;
    let opts = resolve_clean_options(args, &workspace.settings.cache.clean, system_now_unix())?;
    let cache = build_cache(workspace.root.as_path());
    let report = cache
        .clean(&opts)
        .into_diagnostic()
        .wrap_err("cache clean failed")?;
    let mut out = io::stdout().lock();
    print_clean_report(&opts, &report, &mut out)
        .into_diagnostic()
        .wrap_err("failed to write clean report")?;
    Ok(0)
}

/// Merge CLI flags with workspace defaults per `AUX-025` and
/// enforce the `AUX-022` "at least one mode" rule.
fn resolve_clean_options(
    args: &CacheCleanArgs,
    defaults: &CacheCleanSettings,
    now_unix: u64,
) -> Result<CleanOptions> {
    let max_age = args.max_age.or(defaults.max_age);
    let max_size = args.max_size.or(defaults.max_size);
    if !args.soft && max_age.is_none() && max_size.is_none() {
        return Err(miette!(
            "haz cache clean requires at least one mode: --soft, --max-age <DURATION>, or --max-size <BYTES> (AUX-022; workspace cache.clean.max_age / max_size MAY supply a default per AUX-025)"
        ));
    }
    Ok(CleanOptions {
        soft: args.soft,
        max_age,
        max_size,
        dry_run: args.dry_run,
        now_unix,
    })
}

fn system_now_unix() -> u64 {
    SystemTime::now()
        .duration_since(UNIX_EPOCH)
        .map_or(0, |d| d.as_secs())
}

/// Render an `AUX-024` clean report to `out`.
///
/// The format is implementation-defined within the `AUX-024`
/// requirements: every required category appears as its own
/// labelled line, and the per-entry dry-run plan is emitted only
/// when [`CleanOptions::dry_run`] is set and the eviction set is
/// non-empty (so the section header never dangles).
fn print_clean_report(
    opts: &CleanOptions,
    report: &CleanReport,
    out: &mut impl Write,
) -> io::Result<()> {
    if opts.dry_run {
        writeln!(out, "dry-run: AUX-024 plan only, no entries removed.")?;
    }
    writeln!(out, "entries inspected: {}", report.inspected)?;
    let (soft_label, age_label, size_label) = if opts.dry_run {
        (
            "would evict --soft",
            "would evict --max-age",
            "would evict --max-size",
        )
    } else {
        ("evicted --soft", "evicted --max-age", "evicted --max-size")
    };
    writeln!(out, "{}: {}", soft_label, report.evicted_by_soft)?;
    writeln!(out, "{}: {}", age_label, report.evicted_by_max_age)?;
    writeln!(out, "{}: {}", size_label, report.evicted_by_max_size)?;
    let tmp_verb = if opts.dry_run { "noted" } else { "removed" };
    writeln!(
        out,
        "orphan tmp directories {tmp_verb}: {}",
        report.removed_tmp_dirs,
    )?;
    writeln!(
        out,
        "orphan restore directories {tmp_verb}: {}",
        report.removed_restore_dirs,
    )?;
    let bytes_label = if opts.dry_run {
        "bytes projected"
    } else {
        "bytes reclaimed"
    };
    writeln!(out, "{bytes_label}: {}", report.bytes_reclaimed)?;
    if opts.dry_run && !report.evicted_entries.is_empty() {
        writeln!(out, "dry-run plan:")?;
        for e in &report.evicted_entries {
            writeln!(
                out,
                "  {} created_at={} footprint={} mode={}",
                e.key_hex_prefix,
                e.created_at_unix,
                e.footprint,
                mode_label_name(e.matched_mode),
            )?;
        }
    }
    Ok(())
}

fn mode_label_name(m: EvictionMode) -> &'static str {
    match m {
        EvictionMode::Soft => "--soft",
        EvictionMode::MaxAge => "--max-age",
        EvictionMode::MaxSize => "--max-size",
    }
}

/// Execute `haz cache info` against `cwd`.
///
/// Loads the workspace to locate the cache root, walks the cache
/// in read-only mode via [`Cache::info`], and prints the
/// classification per `AUX-019`. Returns 0 on success.
///
/// # Errors
///
/// Returns a [`miette::Report`] for workspace-load or filesystem
/// read failures during the cache walk.
pub fn info(cwd: &Path) -> Result<i32> {
    let (workspace, _graph) = load_workspace(cwd)?;
    let cache = build_cache(workspace.root.as_path());
    let report = cache
        .info()
        .into_diagnostic()
        .wrap_err("cache info walk failed")?;
    let mut out = io::stdout().lock();
    print_info_report(cache.cache_root(), &report, &mut out)
        .into_diagnostic()
        .wrap_err("failed to write cache info report")?;
    Ok(0)
}

/// Render an [`AUX-019`] info report to `out`.
///
/// The format is implementation-defined; this v1 lists each
/// category on its own line. The `by schema` section is omitted
/// entirely when there are no well-formed entries to break down.
fn print_info_report(
    cache_root: &Path,
    report: &CacheInfoReport,
    out: &mut impl Write,
) -> io::Result<()> {
    writeln!(out, "cache root: {}", cache_root.display())?;
    writeln!(out, "well-formed entries: {}", report.well_formed_entries)?;
    writeln!(out, "corrupt entries: {}", report.corrupt_entries)?;
    writeln!(out, "orphan tmp directories: {}", report.orphan_tmp_dirs)?;
    writeln!(
        out,
        "orphan restore directories: {}",
        report.orphan_restore_dirs,
    )?;
    writeln!(out, "total bytes: {}", report.total_bytes)?;
    if !report.by_schema.is_empty() {
        writeln!(out, "by schema:")?;
        for ((rev, hash), count) in &report.by_schema {
            writeln!(
                out,
                "  (revision={rev}, hash={}): {count}",
                hash_label_name(*hash),
            )?;
        }
    }
    Ok(())
}

/// Map a [`HashFunctionLabel`] to the spec's canonical lowercase
/// string (`CACHE-002` registry).
fn hash_label_name(label: HashFunctionLabel) -> &'static str {
    match label {
        HashFunctionLabel::Blake3 => "blake3",
        HashFunctionLabel::Sha256 => "sha256",
    }
}

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

#[cfg(test)]
mod tests {
    use super::{
        clean, clear, hash_label_name, info, mode_label_name, print_clean_report,
        print_info_report, resolve_clean_options,
    };
    use crate::cli::CacheCleanArgs;
    use haz_cache::clean::{CleanOptions, CleanReport, EvictedEntry, EvictionMode};
    use haz_cache::info::CacheInfoReport;
    use haz_cache::manifest::HashFunctionLabel;
    use haz_domain::settings::cache_clean::CacheCleanSettings;
    use haz_domain::settings::cache_clean::max_age::MaxAge;
    use haz_domain::settings::cache_clean::max_size::MaxSize;
    use std::collections::BTreeMap;
    use std::fs;
    use std::path::{Path, PathBuf};

    fn write_file(path: &Path, contents: &str) {
        if let Some(parent) = path.parent() {
            fs::create_dir_all(parent).unwrap();
        }
        fs::write(path, contents).unwrap();
    }

    /// Build an explicit-mode workspace `load_workspace` accepts.
    ///
    /// Explicit mode (`.haz/workspace.yml`) rather than implicit
    /// mode (`.git` marker), because cache tests pre-populate
    /// `.haz/cache/...`; with implicit mode, discovery sees the
    /// `.haz/` directory and tries to read a `workspace.yml` that
    /// isn't there. Explicit mode keeps `.haz/` legitimately used
    /// by the workspace, so a sibling `.haz/cache/` subtree does
    /// not confuse discovery.
    fn write_workspace(root: &Path) {
        write_file(&root.join(".haz/workspace.yml"), "projects:\n  - /lib\n");
        write_file(
            &root.join("lib/haz.yml"),
            "name: lib\ntasks:\n  default:\n    command: [\"true\"]\n",
        );
    }

    #[test]
    fn clear_on_fresh_workspace_is_noop_and_exits_zero() {
        let tempdir = tempfile::TempDir::new().unwrap();
        let root = tempdir.path();
        write_workspace(root);
        let code = clear(root).unwrap();
        assert_eq!(code, 0);
        // Cache root was never created; idempotent no-op behaviour
        // of `Cache::clear` is what makes this OK.
        assert!(!root.join(".haz/cache").exists());
    }

    #[test]
    fn clear_removes_an_existing_cache_root() {
        let tempdir = tempfile::TempDir::new().unwrap();
        let root = tempdir.path();
        write_workspace(root);
        // Pre-populate the cache root with a sentinel file; clear
        // MUST take the whole subtree out.
        let sentinel = root.join(".haz/cache/ab/sentinel");
        write_file(&sentinel, "anything");
        assert!(sentinel.exists());
        let code = clear(root).unwrap();
        assert_eq!(code, 0);
        assert!(!root.join(".haz/cache").exists());
    }

    #[test]
    fn clear_does_not_touch_files_outside_cache_root() {
        let tempdir = tempfile::TempDir::new().unwrap();
        let root = tempdir.path();
        write_workspace(root);
        write_file(&root.join("docs/keep.md"), "important");
        write_file(&root.join(".haz/cache/ab/c"), "stale");
        let code = clear(root).unwrap();
        assert_eq!(code, 0);
        assert!(!root.join(".haz/cache").exists());
        assert_eq!(
            fs::read_to_string(root.join("docs/keep.md")).unwrap(),
            "important"
        );
    }

    fn soft_only_args() -> CacheCleanArgs {
        CacheCleanArgs {
            soft: true,
            ..CacheCleanArgs::default()
        }
    }

    #[test]
    fn clean_soft_on_fresh_workspace_reports_all_zero() {
        let tempdir = tempfile::TempDir::new().unwrap();
        let root = tempdir.path();
        write_workspace(root);
        let code = clean(root, &soft_only_args()).unwrap();
        assert_eq!(code, 0);
    }

    #[test]
    fn clean_soft_reclaims_a_restore_staging_directory() {
        // A `.restore-<key>-<rand>` directory under the cache root
        // is the simplest "objectively stale" artefact for a
        // black-box test: it doesn't require constructing a valid
        // manifest, only creating the staging directory shape.
        let tempdir = tempfile::TempDir::new().unwrap();
        let root = tempdir.path();
        write_workspace(root);
        let staging = root.join(".haz/cache/.restore-abcdef-1234");
        write_file(&staging.join("leftover.bin"), "scratch");
        assert!(staging.exists());
        let code = clean(root, &soft_only_args()).unwrap();
        assert_eq!(code, 0);
        assert!(!staging.exists());
    }

    #[test]
    fn clean_rejects_invocation_with_no_effective_mode() {
        // AUX-022: a bare `haz cache clean` against a workspace
        // with no `cache.clean.*` defaults MUST be rejected with
        // a typed CLI error.
        let tempdir = tempfile::TempDir::new().unwrap();
        let root = tempdir.path();
        write_workspace(root);
        let err = clean(root, &CacheCleanArgs::default()).unwrap_err();
        let rendered = format!("{err}");
        assert!(
            rendered.contains("at least one mode"),
            "AUX-022 rejection diagnostic must surface the requirement: {rendered}",
        );
    }

    #[test]
    fn clean_dry_run_under_soft_leaves_restore_staging_in_place() {
        // AUX-024: `--dry-run` MUST compute the eviction set but
        // make no on-disk changes. Soft pass would normally
        // reclaim the staging directory; under dry-run it stays.
        let tempdir = tempfile::TempDir::new().unwrap();
        let root = tempdir.path();
        write_workspace(root);
        let staging = root.join(".haz/cache/.restore-abcdef-1234");
        write_file(&staging.join("leftover.bin"), "scratch");

        let args = CacheCleanArgs {
            soft: true,
            dry_run: true,
            ..CacheCleanArgs::default()
        };
        let code = clean(root, &args).unwrap();
        assert_eq!(code, 0);
        assert!(
            staging.exists(),
            "dry-run must not remove the restore staging directory",
        );
    }

    #[test]
    fn clean_uses_workspace_default_max_age_when_cli_silent() {
        // AUX-025: workspace `cache.clean.max_age` MAY supply the
        // mode that satisfies AUX-022 when the CLI is silent.
        let tempdir = tempfile::TempDir::new().unwrap();
        let root = tempdir.path();
        write_file(
            &root.join(".haz/workspace.yml"),
            "projects:\n  - /lib\ncache:\n  clean:\n    max_age: 30d\n",
        );
        write_file(
            &root.join("lib/haz.yml"),
            "name: lib\ntasks:\n  default:\n    command: [\"true\"]\n",
        );
        // No CLI mode flags set; the workspace default supplies it.
        let code = clean(root, &CacheCleanArgs::default()).unwrap();
        assert_eq!(code, 0);
    }

    // ---- resolve_clean_options ----

    fn defaults_with(max_age: Option<&str>, max_size: Option<&str>) -> CacheCleanSettings {
        CacheCleanSettings {
            max_age: max_age.map(|s| MaxAge::parse(s).unwrap()),
            max_size: max_size.map(|s| MaxSize::parse(s).unwrap()),
        }
    }

    #[test]
    fn resolve_clean_options_rejects_no_effective_mode() {
        let err = resolve_clean_options(
            &CacheCleanArgs::default(),
            &CacheCleanSettings::default(),
            0,
        )
        .unwrap_err();
        let rendered = format!("{err}");
        assert!(
            rendered.contains("at least one mode"),
            "AUX-022 rejection diagnostic must surface the requirement: {rendered}",
        );
    }

    #[test]
    fn resolve_clean_options_falls_back_to_workspace_when_cli_silent() {
        let defaults = defaults_with(Some("30d"), Some("1GB"));
        let opts = resolve_clean_options(&CacheCleanArgs::default(), &defaults, 12345).unwrap();
        assert_eq!(opts.max_age, defaults.max_age);
        assert_eq!(opts.max_size, defaults.max_size);
        assert!(!opts.soft);
        assert!(!opts.dry_run);
        assert_eq!(opts.now_unix, 12345);
    }

    #[test]
    fn resolve_clean_options_cli_overrides_workspace_defaults() {
        // AUX-025: an explicit CLI flag overrides the workspace
        // default for that mode (the OTHER mode still inherits
        // its default when CLI is silent for it).
        let defaults = defaults_with(Some("30d"), Some("1GB"));
        let cli_age = MaxAge::parse("7d").unwrap();
        let args = CacheCleanArgs {
            max_age: Some(cli_age),
            ..CacheCleanArgs::default()
        };
        let opts = resolve_clean_options(&args, &defaults, 0).unwrap();
        assert_eq!(opts.max_age, Some(cli_age));
        assert_eq!(opts.max_size, defaults.max_size);
    }

    #[test]
    fn resolve_clean_options_soft_alone_satisfies_aux_022() {
        let opts =
            resolve_clean_options(&soft_only_args(), &CacheCleanSettings::default(), 0).unwrap();
        assert!(opts.soft);
        assert!(opts.max_age.is_none());
        assert!(opts.max_size.is_none());
    }

    // ---- print_clean_report ----

    fn sample_report() -> CleanReport {
        CleanReport {
            inspected: 7,
            evicted_by_soft: 2,
            evicted_by_max_age: 3,
            evicted_by_max_size: 1,
            removed_tmp_dirs: 1,
            removed_restore_dirs: 2,
            bytes_reclaimed: 4096,
            evicted_entries: vec![
                EvictedEntry {
                    key_hex_prefix: "deadbeef".to_owned(),
                    created_at_unix: 100,
                    footprint: 512,
                    matched_mode: EvictionMode::Soft,
                },
                EvictedEntry {
                    key_hex_prefix: "cafef00d".to_owned(),
                    created_at_unix: 200,
                    footprint: 1024,
                    matched_mode: EvictionMode::MaxAge,
                },
            ],
        }
    }

    fn render(opts: &CleanOptions, report: &CleanReport) -> String {
        let mut out = Vec::<u8>::new();
        print_clean_report(opts, report, &mut out).unwrap();
        String::from_utf8(out).unwrap()
    }

    #[test]
    fn print_clean_report_renders_required_aux_024_categories() {
        let opts = CleanOptions {
            soft: true,
            ..CleanOptions::default()
        };
        let rendered = render(&opts, &sample_report());
        assert!(
            rendered.contains("entries inspected: 7"),
            "missing inspected line: {rendered}",
        );
        assert!(
            rendered.contains("evicted --soft: 2"),
            "missing soft line: {rendered}",
        );
        assert!(
            rendered.contains("evicted --max-age: 3"),
            "missing max-age line: {rendered}",
        );
        assert!(
            rendered.contains("evicted --max-size: 1"),
            "missing max-size line: {rendered}",
        );
        assert!(
            rendered.contains("orphan tmp directories removed: 1"),
            "missing tmp line: {rendered}",
        );
        assert!(
            rendered.contains("orphan restore directories removed: 2"),
            "missing restore line: {rendered}",
        );
        assert!(
            rendered.contains("bytes reclaimed: 4096"),
            "missing bytes line: {rendered}",
        );
        assert!(
            !rendered.contains("dry-run"),
            "non-dry-run output must not mention dry-run: {rendered}",
        );
    }

    #[test]
    fn print_clean_report_lists_per_entry_detail_under_dry_run() {
        let opts = CleanOptions {
            soft: true,
            dry_run: true,
            ..CleanOptions::default()
        };
        let rendered = render(&opts, &sample_report());
        assert!(
            rendered.contains("would evict --soft: 2"),
            "dry-run soft label: {rendered}",
        );
        assert!(
            rendered.contains("would evict --max-age: 3"),
            "dry-run max-age label: {rendered}",
        );
        assert!(
            rendered.contains("bytes projected: 4096"),
            "dry-run bytes label: {rendered}",
        );
        assert!(
            rendered.contains("dry-run plan:"),
            "dry-run plan header: {rendered}",
        );
        assert!(
            rendered.contains("deadbeef created_at=100 footprint=512 mode=--soft"),
            "dry-run plan soft entry: {rendered}",
        );
        assert!(
            rendered.contains("cafef00d created_at=200 footprint=1024 mode=--max-age"),
            "dry-run plan max-age entry: {rendered}",
        );
    }

    #[test]
    fn print_clean_report_omits_dry_run_plan_when_evicted_entries_empty() {
        // The "dry-run plan:" header MUST NOT appear when there
        // is nothing to list; otherwise it would dangle.
        let opts = CleanOptions {
            soft: true,
            dry_run: true,
            ..CleanOptions::default()
        };
        let empty = CleanReport::default();
        let rendered = render(&opts, &empty);
        assert!(
            !rendered.contains("dry-run plan:"),
            "empty dry-run plan must omit the header: {rendered}",
        );
    }

    #[test]
    fn mode_label_name_maps_each_variant() {
        assert_eq!(mode_label_name(EvictionMode::Soft), "--soft");
        assert_eq!(mode_label_name(EvictionMode::MaxAge), "--max-age");
        assert_eq!(mode_label_name(EvictionMode::MaxSize), "--max-size");
    }

    // ---- info ----

    #[test]
    fn aux_021_info_on_fresh_workspace_exits_zero() {
        let tempdir = tempfile::TempDir::new().unwrap();
        let root = tempdir.path();
        write_workspace(root);
        let code = info(root).unwrap();
        assert_eq!(code, 0);
        // No cache root was created during info; AUX-018 read-only.
        assert!(!root.join(".haz/cache").exists());
    }

    #[test]
    fn aux_018_info_does_not_mutate_an_existing_cache_root() {
        // Pre-populate a mix of cache shapes (an orphan restore
        // dir + a corrupt entry without a manifest). Snapshot the
        // (path, bytes) state of `.haz/cache/` before and after
        // info; the two snapshots MUST be byte-identical.
        let tempdir = tempfile::TempDir::new().unwrap();
        let root = tempdir.path();
        write_workspace(root);
        write_file(
            &root.join(".haz/cache/.restore-abcdef-1234/leftover.bin"),
            "scratch",
        );
        // Corrupt entry: shard dir + entry dir, no manifest.
        fs::create_dir_all(root.join(".haz/cache/ab/0000")).unwrap();
        write_file(&root.join(".haz/cache/ab/0000/stdout"), "");
        let before = snapshot_dir(&root.join(".haz/cache"));
        let code = info(root).unwrap();
        assert_eq!(code, 0);
        let after = snapshot_dir(&root.join(".haz/cache"));
        assert_eq!(
            before, after,
            "cache root state must not change under info()",
        );
    }

    #[test]
    fn aux_019_info_output_lists_every_required_category() {
        // The format is implementation-defined; this test pins it
        // by checking that every AUX-019 category appears as its
        // own labelled line. Constructing a CacheInfoReport
        // directly bypasses the FS walk so the test is purely a
        // format check.
        let report = CacheInfoReport {
            well_formed_entries: 5,
            corrupt_entries: 2,
            orphan_tmp_dirs: 1,
            orphan_restore_dirs: 3,
            total_bytes: 12345,
            by_schema: {
                let mut m: BTreeMap<(u8, HashFunctionLabel), u64> = BTreeMap::new();
                m.insert((0, HashFunctionLabel::Blake3), 5);
                m
            },
        };
        let mut out = Vec::<u8>::new();
        let cache_root = PathBuf::from("/some/workspace/.haz/cache");
        print_info_report(&cache_root, &report, &mut out).unwrap();
        let rendered = String::from_utf8(out).unwrap();
        assert!(
            rendered.contains("cache root: /some/workspace/.haz/cache"),
            "missing cache root line: {rendered}",
        );
        assert!(
            rendered.contains("well-formed entries: 5"),
            "missing well-formed line: {rendered}",
        );
        assert!(
            rendered.contains("corrupt entries: 2"),
            "missing corrupt line: {rendered}",
        );
        assert!(
            rendered.contains("orphan tmp directories: 1"),
            "missing orphan tmp line: {rendered}",
        );
        assert!(
            rendered.contains("orphan restore directories: 3"),
            "missing orphan restore line: {rendered}",
        );
        assert!(
            rendered.contains("total bytes: 12345"),
            "missing total bytes line: {rendered}",
        );
        assert!(
            rendered.contains("by schema:"),
            "missing schema breakdown header: {rendered}",
        );
        assert!(
            rendered.contains("(revision=0, hash=blake3): 5"),
            "missing schema breakdown entry: {rendered}",
        );
    }

    #[test]
    fn aux_019_info_output_omits_schema_section_when_no_well_formed_entries() {
        // The `by schema:` section MUST NOT appear when there are
        // no well-formed entries to enumerate; otherwise the
        // output would carry a dangling header.
        let report = CacheInfoReport {
            well_formed_entries: 0,
            corrupt_entries: 2,
            orphan_tmp_dirs: 0,
            orphan_restore_dirs: 0,
            total_bytes: 0,
            by_schema: BTreeMap::new(),
        };
        let mut out = Vec::<u8>::new();
        print_info_report(Path::new("/ws/.haz/cache"), &report, &mut out).unwrap();
        let rendered = String::from_utf8(out).unwrap();
        assert!(
            !rendered.contains("by schema:"),
            "schema section should be omitted: {rendered}",
        );
    }

    #[test]
    fn hash_label_name_maps_spec_canonical_strings() {
        // CACHE-002 registry: the canonical labels are `blake3`
        // and `sha256` (lowercase).
        assert_eq!(hash_label_name(HashFunctionLabel::Blake3), "blake3");
        assert_eq!(hash_label_name(HashFunctionLabel::Sha256), "sha256");
    }

    /// Recursively snapshot every regular file under `root`,
    /// returning a `BTreeMap<relative_path, bytes>`. Symlinks and
    /// other non-file kinds are skipped; nested directories
    /// recurse.
    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);
            }
        }
    }
}