logmv/

lib.rs

//! Logged atomic file move and trash with an append-only JSON-Lines audit trail.
//!
//! `logmv` renames a file, or moves it to a trash directory, and records the
//! operation as one compact JSON line per action so every change is auditable.
//! Moves are atomic renames only (never a copy fallback), never overwrite an
//! existing destination, and trash never unlinks. See [`run`] for the
//! orchestration and [`Op`] for the operations it performs.
//!
//! # Platform support
//!
//! logmv supports only macOS and Linux. The crate bakes in macOS/Linux-specific
//! assumptions: `EXDEV` == raw OS error 18, `'/'` path separators (see
//! `ends_with_sep`), and the `~/.Trash` move model (resolved in `main.rs`). Other
//! platforms differ on all three, so a non-macOS/Linux build is rejected at
//! compile time via `compile_error!` rather than silently misbehaving. The guard
//! lives here; the binary depends on the library, so it is guarded transitively.
//!
//! One further Linux-only divergence affects log fidelity, not the move itself.
//! The rename is byte-faithful: it uses the real `OsStr` bytes (see
//! `path_to_cstring`), so a non-UTF-8 filename is moved correctly. The log,
//! however, records `src` and `dst` via `to_string_lossy`, so on Linux each
//! invalid UTF-8 byte is written as U+FFFD (the Unicode replacement character);
//! JSON strings must be valid Unicode, so the raw bytes cannot be stored
//! verbatim. A log line whose `src` or `dst` contains U+FFFD is therefore not a
//! byte-exact record and is not reliably reversible. macOS (APFS/HFS+)
//! enforces UTF-8, so this case is unreachable there.

#![warn(missing_docs)]

#[cfg(not(any(target_os = "macos", target_os = "linux")))]
compile_error!("logmv supports only macOS and Linux");

use std::ffi::CString;
use std::ffi::OsStr;
use std::ffi::c_char;
use std::ffi::c_int;
use std::ffi::c_uint;
use std::fs;
use std::fs::OpenOptions;
use std::io;
use std::io::Write;
use std::path::Path;
use std::path::PathBuf;

use chrono::DateTime;
use chrono::FixedOffset;
use chrono::SecondsFormat;
use serde_json::Map;
use serde_json::Value;

/// EXDEV (cross-device link) is raw os error 18 on macOS and Linux.
const EXDEV: i32 = 18;

/// EEXIST (destination already exists) is raw os error 17 on macOS and Linux.
const EEXIST: i32 = 17;

/// The four keys `logmv` stamps itself; metadata pairs may never use them (AC13).
const CANONICAL_KEYS: [&str; 4] = ["ts", "act", "src", "dst"];

/// A file-system operation for logmv to perform.
pub enum Op {
    /// Move `src` to `dst` via an atomic rename, then log a `move` line.
    Move {
        /// Source path to move (canonicalized to absolute when logged).
        src: PathBuf,
        /// Destination path; an existing dir (or trailing `/`) means move *into* it.
        dst: PathBuf,
    },
    /// Move `path` into `trash_dir`, disambiguating on name collision, then log a `trash` line.
    Trash {
        /// Path to send to the trash.
        path: PathBuf,
        /// Trash directory to move `path` into (e.g. `~/.Trash`).
        trash_dir: PathBuf,
    },
}

/// Typed errors for logmv operations.
#[derive(Debug, thiserror::Error)]
pub enum Error {
    /// The resolved destination already exists; the move was refused (never overwrite).
    #[error("destination already exists: {0}")]
    DestinationExists(PathBuf),
    /// The rename crossed filesystems (`EXDEV`); logmv never falls back to a copy.
    #[error("cross-volume rename not supported (EXDEV)")]
    CrossVolume,
    /// The `rename` syscall failed for a reason other than a cross-volume move.
    #[error("rename failed: {0}")]
    Rename(#[source] io::Error),
    /// The move succeeded but the log append failed: filesystem and log may have drifted.
    #[error("move succeeded but log append failed: filesystem and log may have drifted: {0}")]
    DriftAfterMove(#[source] io::Error),
    /// A `--mkdir` directory creation failed; any partially created directories are unlogged.
    #[error("directory creation failed; any partially created directories are unlogged: {0}")]
    MkdirCreate(#[source] io::Error),
    /// A `--mkdir` directory was created but its log append failed: filesystem and log may have drifted.
    #[error("directories created but log append failed: filesystem and log may have drifted: {0}")]
    DriftAfterMkdir(#[source] io::Error),
    /// A `--rmdir` directory removal's log append failed: filesystem and log may have drifted.
    #[error(
        "directory removed but its log append failed: filesystem and log may have drifted: {0}"
    )]
    DriftAfterRmdir(#[source] io::Error),
    /// Resolving a path to its canonical absolute form failed.
    #[error("canonicalize failed: {0}")]
    Canonicalize(#[source] io::Error),
    /// A metadata key collided with a canonical key (`ts`/`act`/`src`/`dst`).
    #[error("metadata key collides with canonical key: {0}")]
    MetadataKeyCollision(String),
}

/// Assemble one compact JSON-Lines entry from the given fields.
///
/// `ts` is injected (not read from a clock) so this function is pure and
/// deterministically testable. `src`/`dst` must already be absolute strings.
/// For trash ops, `dst` is the canonical landing path under the trash dir
/// (the disambiguated target), just like move.
/// Refuses any pair whose key is a canonical key (ts/act/src/dst) via
/// `Error::MetadataKeyCollision` (AC13). Canonical four are written first,
/// then pairs in the given order, all values as JSON strings.
fn build_entry(
    ts: DateTime<FixedOffset>,
    act: &str,
    src: &str,
    dst: &str,
    pairs: &[(&str, &str)],
) -> Result<String, Error> {
    // Refuse before assembling anything: a colliding key must yield no line
    // (and, via `run`, no move and no log).
    if let Some(key) = first_colliding_key(pairs) {
        return Err(Error::MetadataKeyCollision(key.to_string()));
    }

    let mut map = Map::new();
    map.insert(
        "ts".to_string(),
        Value::String(ts.to_rfc3339_opts(SecondsFormat::Secs, false)),
    );
    map.insert("act".to_string(), Value::String(act.to_string()));
    map.insert("src".to_string(), Value::String(src.to_string()));
    map.insert("dst".to_string(), Value::String(dst.to_string()));
    // Pairs follow the canonical four in given order (serde_json `preserve_order`),
    // every value a JSON string; serde owns all key/value escaping (AC6).
    for (k, v) in pairs {
        map.insert((*k).to_string(), Value::String((*v).to_string()));
    }

    Ok(Value::Object(map).to_string())
}

/// Return the first metadata key that collides with a canonical key, if any.
/// Single source of truth for the collision guard: `build_entry` runs it before
/// assembling a line, and `run` runs it early so a bad pair aborts before any
/// `--mkdir` mutation (AC11).
fn first_colliding_key<'a>(pairs: &'a [(&str, &str)]) -> Option<&'a str> {
    pairs
        .iter()
        .find(|(k, _)| CANONICAL_KEYS.contains(k))
        .map(|(k, _)| *k)
}

/// Map an `io::Error` from `fs::rename` to a typed `Error`.
///
/// EXDEV (raw os error 18) → `Error::CrossVolume`.
/// Anything else → `Error::Rename`.
fn classify_rename_err(err: io::Error) -> Error {
    if err.raw_os_error() == Some(EXDEV) {
        Error::CrossVolume
    } else {
        Error::Rename(err)
    }
}

/// `AT_FDCWD`: resolve a relative `renameat2` path against the current working
/// directory, matching the relative-path semantics `fs::rename` has today.
#[cfg(target_os = "linux")]
const AT_FDCWD: c_int = -100;

/// `RENAME_NOREPLACE`: fail with `EEXIST` rather than clobbering an existing
/// destination entry (name-level; a symlink is not followed).
#[cfg(target_os = "linux")]
const RENAME_NOREPLACE: c_uint = 1;

/// `RENAME_EXCL`: macOS exclusive-create flag from `<sys/stdio.h>`
/// (`RENAME_SECLUDE 0x1`, `RENAME_SWAP 0x2`, `RENAME_EXCL 0x4`); fail with
/// `EEXIST` rather than clobbering an existing destination entry.
#[cfg(target_os = "macos")]
const RENAME_EXCL: c_uint = 0x0000_0004;

/// Convert a path to a NUL-terminated C string for the `renameat2`/`renamex_np`
/// FFI, via raw bytes (never a lossy `String`, so non-UTF-8 paths pass through
/// unchanged). An interior NUL yields a clean `InvalidInput` error rather than
/// an `unwrap`/`expect` on the write path.
fn path_to_cstring(p: &Path) -> io::Result<CString> {
    use std::os::unix::ffi::OsStrExt;
    CString::new(p.as_os_str().as_bytes())
        .map_err(|e| io::Error::new(io::ErrorKind::InvalidInput, e))
}

/// Atomically rename `from` to `to`, refusing to clobber an existing destination
/// entry (name-level; a symlink is not followed).
///
/// Returns `Ok(())` on success, or `Err(io::Error::last_os_error())` carrying the
/// raw `errno` so callers can test `raw_os_error() == Some(EEXIST)` and still route
/// `EXDEV` and others through [`classify_rename_err`]. Classifies nothing itself,
/// mirroring today's `fs::rename(...).map_err(classify_rename_err)` split of
/// mechanism from policy.
#[cfg(target_os = "linux")]
fn rename_noclobber(from: &Path, to: &Path) -> io::Result<()> {
    // NOTE: libc renameat2 wrapper (glibc >=2.28); drop to a raw syscall() only
    // if a target without the wrapper (e.g. an old musl) is ever added to CI.
    unsafe extern "C" {
        fn renameat2(
            olddirfd: c_int,
            oldpath: *const c_char,
            newdirfd: c_int,
            newpath: *const c_char,
            flags: c_uint,
        ) -> c_int;
    }

    let from_c = path_to_cstring(from)?;
    let to_c = path_to_cstring(to)?;

    // SAFETY:
    // - `from_c`/`to_c` are `CString` locals that outlive the call; `.as_ptr()`
    //   yields non-null pointers to valid, readable, NUL-terminated bytes.
    // - `renameat2` only reads (never retains) the two `*const c_char` pointers;
    //   no Rust aliasing/mutability invariant is exposed (only `*const` passed).
    // - `AT_FDCWD` and `RENAME_NOREPLACE` are the documented kernel sentinel/flag.
    // - the return value is checked immediately and `errno` is read via
    //   `io::Error::last_os_error()` with no intervening libc call between.
    let rc = unsafe {
        renameat2(
            AT_FDCWD,
            from_c.as_ptr(),
            AT_FDCWD,
            to_c.as_ptr(),
            RENAME_NOREPLACE,
        )
    };
    if rc < 0 {
        return Err(io::Error::last_os_error());
    }
    Ok(())
}

/// Atomically rename `from` to `to`, refusing to clobber an existing destination
/// entry (name-level; a symlink is not followed).
///
/// Returns `Ok(())` on success, or `Err(io::Error::last_os_error())` carrying the
/// raw `errno` so callers can test `raw_os_error() == Some(EEXIST)` and still route
/// `EXDEV` and others through [`classify_rename_err`]. Classifies nothing itself,
/// mirroring today's `fs::rename(...).map_err(classify_rename_err)` split of
/// mechanism from policy.
#[cfg(target_os = "macos")]
fn rename_noclobber(from: &Path, to: &Path) -> io::Result<()> {
    unsafe extern "C" {
        fn renamex_np(from: *const c_char, to: *const c_char, flags: c_uint) -> c_int;
    }

    let from_c = path_to_cstring(from)?;
    let to_c = path_to_cstring(to)?;

    // SAFETY:
    // - `from_c`/`to_c` are `CString` locals that outlive the call; `.as_ptr()`
    //   yields non-null pointers to valid, readable, NUL-terminated bytes.
    // - `renamex_np` only reads (never retains) the two `*const c_char` pointers;
    //   no Rust aliasing/mutability invariant is exposed (only `*const` passed).
    // - `RENAME_EXCL` is the documented exclusive-create flag from `<sys/stdio.h>`.
    // - the return value is checked immediately and `errno` is read via
    //   `io::Error::last_os_error()` with no intervening libc call between.
    let rc = unsafe { renamex_np(from_c.as_ptr(), to_c.as_ptr(), RENAME_EXCL) };
    if rc < 0 {
        return Err(io::Error::last_os_error());
    }
    Ok(())
}

/// Canonicalize a destination path that must NOT yet exist: canonicalize the
/// parent directory (which does exist) and rejoin the file name.
fn canonicalize_new(dst: &Path) -> Result<PathBuf, Error> {
    let file_name = dst.file_name().ok_or_else(|| {
        Error::Canonicalize(io::Error::new(
            io::ErrorKind::InvalidInput,
            "destination has no file name",
        ))
    })?;
    let parent = match dst.parent() {
        Some(p) if !p.as_os_str().is_empty() => p,
        _ => Path::new("."),
    };
    let parent_abs = fs::canonicalize(parent).map_err(Error::Canonicalize)?;
    Ok(parent_abs.join(file_name))
}

/// Atomically move `path` into `trash_dir`, never clobbering an existing entry:
/// try `name` as-is, then `<stem>-<n>[.<ext>]` for n = 1,2,3,…, bumping `n` only
/// on `EEXIST` and returning the winning path on the first successful rename.
/// Any other errno is classified via [`classify_rename_err`]. The disambiguation
/// is driven by the atomic rename result (not a pre-scan), so the returned path is
/// exactly the candidate that won the rename.
fn rename_into_trash(path: &Path, trash_dir: &Path, name: &OsStr) -> Result<PathBuf, Error> {
    let first = trash_dir.join(name);
    match rename_noclobber(path, &first) {
        Ok(()) => return Ok(first),
        Err(e) if e.raw_os_error() == Some(EEXIST) => {}
        Err(e) => return Err(classify_rename_err(e)),
    }

    let as_path = Path::new(name);
    let stem = as_path
        .file_stem()
        .unwrap_or(name)
        .to_string_lossy()
        .into_owned();
    let ext = as_path
        .extension()
        .map(|e| e.to_string_lossy().into_owned());

    let mut n = 1u32;
    loop {
        let candidate_name = match &ext {
            Some(ext) => format!("{stem}-{n}.{ext}"),
            None => format!("{stem}-{n}"),
        };
        let candidate = trash_dir.join(candidate_name);
        match rename_noclobber(path, &candidate) {
            Ok(()) => return Ok(candidate),
            Err(e) if e.raw_os_error() == Some(EEXIST) => {}
            Err(e) => return Err(classify_rename_err(e)),
        }
        n += 1;
    }
}

/// Write `line` plus one trailing newline to `w` in a single `write` call, so a
/// log entry can never be split across two syscalls (which two concurrent runs
/// could interleave). Generic over `Write` so it is unit-testable.
fn write_log_line<W: Write>(w: &mut W, line: &str) -> io::Result<()> {
    w.write_all(format!("{line}\n").as_bytes())
}

/// Append exactly one line (with trailing newline) to `log`, creating it if absent,
/// then flush it to disk so the record is at least as durable as the event it logs.
fn append_log(log: &Path, line: &str) -> io::Result<()> {
    let mut file = OpenOptions::new().create(true).append(true).open(log)?;
    write_log_line(&mut file, line)?;
    // NOTE: unconditional sync_data; add --no-sync only if a batch caller measurably needs it.
    file.sync_data()
}

/// Does the path end with a directory separator (explicit directory intent)?
// NOTE: macOS/Linux `'/'` only, consistent with the hardcoded EXDEV const.
fn ends_with_sep(p: &Path) -> bool {
    p.to_string_lossy().ends_with('/')
}

/// Resolve the final move target. When `dst` is an existing directory (or ends
/// with a separator, signalling directory intent), move `src` *into* it keeping
/// its basename: `dst/basename(src)`. Otherwise `dst` is the full target path
/// (today's behavior).
fn resolve_move_target(src: &Path, dst: &Path) -> PathBuf {
    match (ends_with_sep(dst) || dst.is_dir(), src.file_name()) {
        (true, Some(name)) => dst.join(name),
        _ => dst.to_path_buf(),
    }
}

/// `--mkdir`: create the move destination's missing parent chain (`mkdir -p`) and
/// append one `mkdir` line per directory actually created, parent → child, before
/// the move line. A fully-present chain creates nothing and logs nothing (AC6).
/// `mkdir` line carries `src:"-"`, `dst:<created dir>` (Q1 directional sentinel).
fn mkdir_chain(parent: &Path, log: &Path, ts: DateTime<FixedOffset>) -> Result<(), Error> {
    // Missing ancestors, collected child → parent then reversed to parent → child.
    let mut missing: Vec<PathBuf> = Vec::new();
    let mut cur = Some(parent);
    while let Some(dir) = cur {
        if dir.as_os_str().is_empty() || dir.exists() {
            break;
        }
        missing.push(dir.to_path_buf());
        cur = dir.parent();
    }
    if missing.is_empty() {
        return Ok(());
    }
    missing.reverse();

    // create_dir_all is idempotent and never clobbers an existing dir (AC4). A
    // failure may leave partially-created, unlogged dirs, but the requested chain
    // is not confirmed present, so this is not post-create drift.
    fs::create_dir_all(parent).map_err(Error::MkdirCreate)?;

    for dir in &missing {
        // The dirs now exist but are not yet logged; a canonicalize failure here is
        // created-but-not-logged drift, not a generic path-resolution failure.
        let abs = fs::canonicalize(dir).map_err(Error::DriftAfterMkdir)?;
        let line = build_entry(ts, "mkdir", "-", &abs.to_string_lossy(), &[])?;
        // Logged only after the dir exists; an append failure here is real drift.
        append_log(log, &line).map_err(Error::DriftAfterMkdir)?;
    }
    Ok(())
}

/// `--rmdir`: after a successful, logged move/trash, remove the source's now-empty
/// parent and cascade upward, removing each now-empty ancestor and stopping at the
/// first non-empty one (`rmdir -p`). Truly-empty-only: a dir holding `.DS_Store`
/// (or anything) is left in place (AC9). `start` paths come from the canonicalized
/// `abs_src`, so the cascade walks real (non-symlink) ancestors only.
/// `rmdir` line carries `src:<removed dir>`, `dst:"-"` (Q1 directional sentinel).
fn rmdir_cascade(start: Option<&Path>, log: &Path, ts: DateTime<FixedOffset>) -> Result<(), Error> {
    let mut cur = start.map(Path::to_path_buf);
    while let Some(dir) = cur {
        if dir.as_os_str().is_empty() {
            break;
        }
        // Cascade boundary gate: stop at a non-empty (or unreadable) directory.
        match fs::read_dir(&dir) {
            Ok(mut entries) => {
                if entries.next().is_some() {
                    break;
                }
            }
            Err(_) => break,
        }
        // remove_dir is the atomic safety guard: it removes only empty dirs.
        match fs::remove_dir(&dir) {
            Ok(()) => {
                let line = build_entry(ts, "rmdir", &dir.to_string_lossy(), "-", &[])?;
                // Logged only after removal; an append failure here is real drift.
                append_log(log, &line).map_err(Error::DriftAfterRmdir)?;
                cur = dir.parent().map(Path::to_path_buf);
            }
            // NOTE: a remove_dir failure (race repopulation, EACCES) stops the
            // cascade silently; rmdir is best-effort cleanup after a logged move and
            // the dir simply remaining is safe. Upgrade to a stderr warning if
            // cleanup visibility ever matters.
            Err(_) => break,
        }
    }
    Ok(())
}

/// Orchestrate: resolve final dst → never-overwrite + pair-collision check →
/// `--mkdir` (create + log each) → atomic rename → log move/trash → `--rmdir`
/// (remove + log each, cascading).
///
/// `log` is the path to the JSON-Lines file to append to (created if absent).
/// `pairs` are free K/V metadata pairs inserted into the line after the canonical four.
/// `mkdir`/`rmdir` gate the directory-creation / cascading-removal behaviors.
///
/// # Errors
///
/// - [`Error::MetadataKeyCollision`] if any pair key is a canonical key
///   (`ts`/`act`/`src`/`dst`), checked before any mutation, so nothing moves.
/// - [`Error::DestinationExists`] if the resolved target already exists (never overwrite).
/// - [`Error::Canonicalize`] if a source or destination path cannot be resolved.
/// - [`Error::MkdirCreate`] if `--mkdir` fails to create the destination's parent
///   chain; any partially created directories are unlogged.
/// - [`Error::CrossVolume`] or [`Error::Rename`] if the atomic rename fails.
/// - [`Error::DriftAfterMove`], [`Error::DriftAfterMkdir`], or [`Error::DriftAfterRmdir`]
///   if the rename succeeded but a subsequent log append or directory operation
///   failed: the filesystem and the log have drifted, and the error is loud.
///
/// # Examples
///
/// ```no_run
/// use std::path::{Path, PathBuf};
/// use logmv::{run, Op};
///
/// let op = Op::Move {
///     src: PathBuf::from("report.txt"),
///     dst: PathBuf::from("archive.txt"),
/// };
/// run(op, Path::new("ops.log"), &[("by", "cc")], false, false)?;
/// # Ok::<(), logmv::Error>(())
/// ```
pub fn run(
    op: Op,
    log: &Path,
    pairs: &[(&str, &str)],
    mkdir: bool,
    rmdir: bool,
) -> Result<(), Error> {
    let ts = chrono::Local::now().fixed_offset();

    // Early pair-collision check: a colliding key must abort before any --mkdir
    // mutation and before the move (AC11). build_entry re-checks (idempotent).
    if let Some(key) = first_colliding_key(pairs) {
        return Err(Error::MetadataKeyCollision(key.to_string()));
    }

    // Each arm performs its own atomic no-clobber rename (move: once; trash: in a
    // retry loop), then yields the line to append and the canonical source path.
    let (line, abs_src) = match op {
        Op::Move { src, dst } => {
            let abs_src = fs::canonicalize(&src).map_err(Error::Canonicalize)?;
            // Resolve the final target: into an existing / trailing-slash directory.
            let target = resolve_move_target(&src, &dst);
            // --mkdir: create the destination's missing parent chain, logging each.
            // A colliding destination implies its parent exists, so mkdir_chain is a
            // no-op on collision and a refused move creates no orphan dirs (AC11).
            if mkdir {
                if let Some(parent) = target.parent() {
                    mkdir_chain(parent, log, ts)?;
                }
            }
            let abs_dst = canonicalize_new(&target)?;
            let line = build_entry(
                ts,
                "move",
                &abs_src.to_string_lossy(),
                &abs_dst.to_string_lossy(),
                pairs,
            )?;
            // Atomic no-clobber rename: EEXIST names the colliding target (never
            // overwrite); EXDEV and others keep CrossVolume/Rename. No log on refusal.
            rename_noclobber(&src, &target).map_err(|e| {
                if e.raw_os_error() == Some(EEXIST) {
                    Error::DestinationExists(target.clone())
                } else {
                    classify_rename_err(e)
                }
            })?;
            (line, abs_src)
        }
        Op::Trash { path, trash_dir } => {
            let abs_src = fs::canonicalize(&path).map_err(Error::Canonicalize)?;
            let name = path.file_name().ok_or_else(|| {
                Error::Canonicalize(io::Error::new(
                    io::ErrorKind::InvalidInput,
                    "trash source has no file name",
                ))
            })?;
            // never-overwrite for trash: disambiguate rather than clobber. The
            // winning candidate is chosen by the atomic rename, then logged.
            let winner = rename_into_trash(&path, &trash_dir, name)?;
            let abs_dst = canonicalize_new(&winner)?;
            let line = build_entry(
                ts,
                "trash",
                &abs_src.to_string_lossy(),
                &abs_dst.to_string_lossy(),
                pairs,
            )?;
            (line, abs_src)
        }
    };

    // Rename happened: an append failure now is real drift, reported loudly (AC9).
    append_log(log, &line).map_err(Error::DriftAfterMove)?;

    // --rmdir runs only after a successful move AND its log append (AC11), for both
    // move and trash; it walks the now-empty source parent upward, cascading.
    if rmdir {
        rmdir_cascade(abs_src.parent(), log, ts)?;
    }
    Ok(())
}

// ---------------------------------------------------------------------------
// Unit tests: T1, T2, T3, T_pairs, T_collide_u, T4
// ---------------------------------------------------------------------------

#[cfg(test)]
mod tests {
    use super::*;
    use chrono::TimeZone;
    use std::io;

    // Shared fixed timestamp for U tests: 2024-01-15T10:30:00+00:00
    fn fixed_ts() -> DateTime<FixedOffset> {
        FixedOffset::east_opt(0)
            .unwrap()
            .with_ymd_and_hms(2024, 1, 15, 10, 30, 0)
            .unwrap()
    }

    // T1: AC1, AC5
    // build_entry(move, empty pairs) → exact compact JSON line with ONLY the four
    // canonical keys ts/act/src/dst in that order, act="move", absolute src/dst,
    // ts ISO-8601 second-precision with offset. Parses as valid JSON.
    #[test]
    fn t1_build_entry_move_exact_line() {
        let ts = fixed_ts();
        let line = build_entry(ts, "move", "/abs/src/file.txt", "/abs/dst/file.txt", &[])
            .expect("build_entry must succeed with no pairs");

        // Must parse as valid JSON.
        let v: serde_json::Value =
            serde_json::from_str(&line).expect("build_entry output must be valid JSON");

        // Field values.
        assert_eq!(v["ts"], "2024-01-15T10:30:00+00:00");
        assert_eq!(v["act"], "move");
        assert_eq!(v["src"], "/abs/src/file.txt");
        assert_eq!(v["dst"], "/abs/dst/file.txt");

        // Exact compact line: only 4 keys, in order, no placeholders.
        let expected = r#"{"ts":"2024-01-15T10:30:00+00:00","act":"move","src":"/abs/src/file.txt","dst":"/abs/dst/file.txt"}"#;
        assert_eq!(line, expected);
    }

    // AC2: non-UTF-8 path converted the way run() converts it
    // (Path::to_string_lossy) is logged with each invalid byte as U+FFFD, and
    // the line stays valid JSON. Pins that a faithful undo of such a line is
    // impossible: the raw 0xFF byte is not recoverable from the log.
    #[cfg(unix)]
    #[test]
    fn t_non_utf8_path_logs_replacement_char() {
        use std::os::unix::ffi::OsStrExt;

        let ts = fixed_ts();
        let lossy = Path::new(OsStr::from_bytes(b"file\xFF.txt")).to_string_lossy();

        let line = build_entry(ts, "move", &lossy, "/abs/dst", &[])
            .expect("build_entry must succeed with a lossily-converted src");

        // Must parse as valid JSON.
        let v: serde_json::Value =
            serde_json::from_str(&line).expect("build_entry output must be valid JSON");

        // The invalid byte is recorded as U+FFFD, not the original byte.
        assert_eq!(v["src"], "file\u{FFFD}.txt");
    }

    // T3: AC6  [no-JSON-corruption invariant]
    // build_entry with a pair whose KEY AND VALUE each contain quote + backslash +
    // unicode → line is still valid JSON and both key and value round-trip
    // byte-exact (serde_json escapes object keys too).
    #[test]
    fn t3_build_entry_special_chars_in_key_and_value_round_trips() {
        let ts = fixed_ts();
        // Key and value each contain a double-quote, backslash, and unicode snowman.
        let tricky_key = r#"k"ey\ ☃"#;
        let tricky_val = r#"say "hello" \ ☃"#;

        let line = build_entry(
            ts,
            "move",
            "/abs/src",
            "/abs/dst",
            &[(tricky_key, tricky_val)],
        )
        .expect("build_entry must succeed with special-char pair");

        // Must parse as valid JSON (not corrupted).
        let v: serde_json::Value = serde_json::from_str(&line)
            .expect("line with special chars in key and value must still be valid JSON");

        // Both key and value must round-trip byte-exact.
        let obj = v.as_object().unwrap();
        assert_eq!(
            obj.get(tricky_key)
                .and_then(|v| v.as_str())
                .expect("tricky key must be present in JSON object"),
            tricky_val
        );
    }

    // T_pairs: AC5
    // build_entry with multiple pairs → pairs appear AFTER the canonical four,
    // in the given order; every value is a JSON string (even a numeric-looking one).
    #[test]
    fn t_pairs_appear_after_canonical_in_order_as_strings() {
        let ts = fixed_ts();
        let pairs: &[(&str, &str)] = &[("by", "cc"), ("ac", "p"), ("num", "42")];

        let line = build_entry(ts, "move", "/abs/src", "/abs/dst", pairs)
            .expect("build_entry must succeed with pairs");

        let v: serde_json::Value =
            serde_json::from_str(&line).expect("line with pairs must be valid JSON");

        let obj = v.as_object().unwrap();
        let keys: Vec<&str> = obj.keys().map(|k| k.as_str()).collect();

        // First four must be the canonical keys in order.
        assert_eq!(
            &keys[..4],
            &["ts", "act", "src", "dst"],
            "first four keys must be canonical in order, got: {keys:?}"
        );
        // Pairs follow in the given order.
        assert_eq!(
            &keys[4..],
            &["by", "ac", "num"],
            "pairs must follow canonical keys in given order, got: {keys:?}"
        );

        // Every value is a JSON string (not number, bool, etc.)
        assert_eq!(v["by"], serde_json::Value::String("cc".into()));
        assert_eq!(v["ac"], serde_json::Value::String("p".into()));
        // "42" must stay a string, not be coerced to a number.
        assert_eq!(v["num"], serde_json::Value::String("42".into()));
    }

    // T_collide_u: AC13  [canonical-unspoofable invariant]
    // build_entry with a pair key in {ts,act,src,dst} → Err(MetadataKeyCollision),
    // no line produced.
    #[test]
    fn t_collide_u_build_entry_refuses_canonical_key() {
        let ts = fixed_ts();

        for &canonical in &["ts", "act", "src", "dst"] {
            let result = build_entry(
                ts,
                "move",
                "/abs/src",
                "/abs/dst",
                &[(canonical, "spoofed")],
            );
            assert!(
                matches!(&result, Err(Error::MetadataKeyCollision(k)) if k == canonical),
                "expected MetadataKeyCollision({canonical}), got: {result:?}"
            );
        }
    }

    // T4: AC8  [atomic-only / no-copy-fallback]
    // classify_rename_err maps raw os error 18 (EXDEV) to Error::CrossVolume;
    // any other error maps to Error::Rename.
    // REUSE: behavior identical to old contract; stays green, no edit.
    #[test]
    fn t4_classify_rename_err_exdev_and_other() {
        // EXDEV is raw os error 18 on macOS and Linux.
        let exdev = io::Error::from_raw_os_error(18);
        let result = classify_rename_err(exdev);
        assert!(
            matches!(result, Error::CrossVolume),
            "os error 18 must map to CrossVolume, got: {result:?}"
        );

        // A different os error (e.g. EACCES = 13) maps to Rename.
        let other = io::Error::from_raw_os_error(13);
        let result2 = classify_rename_err(other);
        assert!(
            matches!(result2, Error::Rename(_)),
            "non-EXDEV error must map to Rename, got: {result2:?}"
        );
    }

    // T_write_log_line_single_write_call: AC1 + AC3
    // write_log_line must emit line + newline in a SINGLE Write::write call
    // (locks out any reintroduction of a two-piece writeln!-style write).
    #[derive(Default)]
    struct CountingWriter {
        writes: usize,
        buf: Vec<u8>,
    }

    impl io::Write for CountingWriter {
        fn write(&mut self, data: &[u8]) -> io::Result<usize> {
            self.writes += 1;
            self.buf.extend_from_slice(data);
            Ok(data.len())
        }

        fn flush(&mut self) -> io::Result<()> {
            Ok(())
        }
    }

    #[test]
    fn t_write_log_line_single_write_call() {
        let mut w = CountingWriter::default();
        let line = "{\"ts\":\"x\",\"act\":\"move\"}";

        let r = write_log_line(&mut w, line);

        assert!(r.is_ok());
        assert_eq!(
            w.writes, 1,
            "line + newline must be a single write, not two"
        );
        assert_eq!(w.buf, format!("{line}\n").into_bytes());
    }
}