"""Agent impersonation security tests — Ed25519 provenance signing.

Attack surface
--------------
Commit records in Muse carry identity fields that are NEVER validated against
the authenticated user: ``author``, ``agent_id``, ``model_id``, etc.
Any caller with access to the CLI can write commits claiming to be authored by
anyone — human, agent, or a previously-trusted identity.

Signing model
-------------
Commits signed with ``--sign`` use Ed25519 (same keypair as MSign request
authentication).  The ``signer_public_key`` field embeds the signer's raw
public key bytes (base64url, 43 chars) so verification is fully offline.

The signed input is ``provenance_payload(commit_id, author, agent_id, ...)``
— a SHA-256 digest that binds content identity to authorship claims.  Any
mutation of author, agent_id, model_id, toolchain_id, or prompt_hash after
signing is detected by ``run_verify``.
"""

from __future__ import annotations

import datetime
import json
import pathlib

import pytest
from cryptography.hazmat.primitives.asymmetric.ed25519 import Ed25519PrivateKey
from cryptography.hazmat.primitives.serialization import Encoding, PublicFormat

from muse.core.object_store import object_path as _obj_path
from muse.core.provenance import (
    encode_public_key,
    provenance_payload,
    sign_commit_ed25519,
    sign_commit_record,
    verify_commit_ed25519,
)
from muse.core.ids import hash_commit as compute_commit_id, hash_snapshot as _hash_snapshot
from muse.core.validation import sanitize_provenance
from muse.core.verify import VerifyResult, run_verify
from muse.core.types import Manifest, MsgpackDict, b64url_encode, blob_id, decode_pubkey, encode_sig, fake_id, public_key_fingerprint, split_id
from muse.core.paths import ref_path, muse_dir


def _signer_key_id(pub_bytes: bytes) -> str:
    return public_key_fingerprint(pub_bytes)


# ---------------------------------------------------------------------------
# Helpers
# ---------------------------------------------------------------------------

def _gen_key() -> Ed25519PrivateKey:
    return Ed25519PrivateKey.generate()


def _pub_bytes(key: Ed25519PrivateKey) -> bytes:
    return key.public_key().public_bytes(Encoding.Raw, PublicFormat.Raw)


def _pub_b64(key: Ed25519PrivateKey) -> str:
    _, b64 = encode_public_key(key)
    return b64


def _make_repo(tmp_path: pathlib.Path) -> pathlib.Path:
    """Create a minimal .muse/ repository skeleton."""
    dot_muse = muse_dir(tmp_path)
    for d in ("objects", "commits", "snapshots", "refs/heads"):
        (dot_muse / d).mkdir(parents=True, exist_ok=True)
    (dot_muse / "repo.json").write_text('{"repo_id": "test-repo"}', encoding="utf-8")
    (dot_muse / "HEAD").write_text("ref: refs/heads/main\n", encoding="utf-8")
    return tmp_path


_COMMITTED_AT = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc)


def _make_real_commit_id(
    snapshot_id: str | None = None,
    parent: str | None = None,
    message: str = "test",
    author: str = "",
    signer_public_key: str = "",
) -> str:
    """Return the canonical commit_id matching the fixed 2026-01-01 timestamp."""
    parents = [parent] if parent else []
    return compute_commit_id(
        parent_ids=parents,
        snapshot_id=snapshot_id or _EMPTY_SNAP_ID,
        message=message,
        committed_at_iso=_COMMITTED_AT.isoformat(),
        author=author,
        signer_public_key=signer_public_key,
    )


def _v7_sig(
    commit_id: str,
    key: Ed25519PrivateKey,
    *,
    author: str = "",
    agent_id: str = "",
    model_id: str = "",
    toolchain_id: str = "",
    prompt_hash: str = "",
    committed_at: str = _COMMITTED_AT.isoformat(),
) -> str:
    """Compute a format_version 7 Ed25519 signature (over provenance_payload)."""
    payload = provenance_payload(
        commit_id,
        author=author,
        agent_id=agent_id,
        model_id=model_id,
        toolchain_id=toolchain_id,
        prompt_hash=prompt_hash,
        committed_at=committed_at,
    )
    return sign_commit_ed25519(payload, key)


def _write_commit(
    root: pathlib.Path,
    commit_id: str,
    *,
    snapshot_id: str | None = None,
    parent: str | None = None,
    message: str = "test",
    author: str = "",
    agent_id: str = "",
    model_id: str = "",
    toolchain_id: str = "",
    prompt_hash: str = "",
    signature: str = "",
    signer_public_key: str = "",
    signer_key_id: str = "",
) -> None:
    """Write a content-hash-valid commit record to the unified object store.

    The ``commit_id`` MUST equal ``_make_real_commit_id(snapshot_id, parent, message)``
    for the record to pass the store's content-hash verification.
    """
    import json as _json

    snap_id = snapshot_id or _EMPTY_SNAP_ID
    record = {
        "commit_id": commit_id,
        "repo_id": "test-repo",
        "branch": "main",
        "snapshot_id": snap_id,
        "message": message,
        "committed_at": _COMMITTED_AT.isoformat(),
        "parent_commit_id": parent,
        "parent2_commit_id": None,
        "author": author,
        "metadata": {},
        "structured_delta": None,
        "sem_ver_bump": "none",
        "breaking_changes": [],
        "agent_id": agent_id,
        "model_id": model_id,
        "toolchain_id": toolchain_id,
        "prompt_hash": prompt_hash,
        "signature": signature,
        "signer_public_key": signer_public_key,
        "signer_key_id": signer_key_id,
        "reviewed_by": [],
        "test_runs": 0,
    }
    payload = _json.dumps(record, separators=(",", ":")).encode()
    path = _obj_path(root, commit_id)
    path.parent.mkdir(parents=True, exist_ok=True)
    path.write_bytes(f"commit {len(payload)}\0".encode() + payload)


_EMPTY_SNAP_ID = _hash_snapshot({})


def _write_snapshot(root: pathlib.Path, snapshot_id: str) -> None:
    """Write a minimal snapshot record to the unified object store."""
    import json as _json

    record = {
        "snapshot_id": snapshot_id,
        "manifest": {},
        "directories": {},
        "created_at": "2026-01-01T00:00:00+00:00",
    }
    payload = _json.dumps(record, separators=(",", ":")).encode()
    path = _obj_path(root, snapshot_id)
    path.parent.mkdir(parents=True, exist_ok=True)
    path.write_bytes(f"snapshot {len(payload)}\0".encode() + payload)


def _set_branch_ref(root: pathlib.Path, branch: str, commit_id: str) -> None:
    branch_ref = ref_path(root, branch)
    branch_ref.parent.mkdir(parents=True, exist_ok=True)
    branch_ref.write_text(commit_id, encoding="utf-8")


def _fake_commit_id(seed: str = "a") -> str:
    return fake_id(seed)


# ===========================================================================
# Author field sanitization
# ===========================================================================


class TestAuthorSanitization:
    """sanitize_provenance must strip control chars from the author field."""

    def test_clean_author_unchanged(self) -> None:
        assert sanitize_provenance("gabriel") == "gabriel"

    def test_esc_in_author_stripped(self) -> None:
        raw = "\x1b[31mfake-human\x1b[0m"
        clean = sanitize_provenance(raw)
        assert "\x1b" not in clean
        assert "fake-human" in clean

    def test_newline_in_author_stripped(self) -> None:
        raw = "gabriel\nlinus@kernel.org"
        clean = sanitize_provenance(raw)
        assert "\n" not in clean

    def test_cr_in_author_stripped(self) -> None:
        raw = "gabriel\r\nlinus"
        clean = sanitize_provenance(raw)
        assert "\r" not in clean

    def test_bel_in_author_stripped(self) -> None:
        raw = "gabriel\x07linus"
        clean = sanitize_provenance(raw)
        assert "\x07" not in clean

    @pytest.mark.parametrize("char_val", range(0x00, 0x20))
    def test_c0_control_chars_stripped(self, char_val: int) -> None:
        char = chr(char_val)
        raw = f"prefix{char}suffix"
        result = sanitize_provenance(raw)
        assert char not in result

    def test_del_stripped(self) -> None:
        assert "\x7f" not in sanitize_provenance("a\x7fb")

    def test_author_cap_at_256_chars(self) -> None:
        long_author = "a" * 300
        stored = sanitize_provenance(long_author[:256])
        assert len(stored) == 256

    def test_unicode_author_preserved(self) -> None:
        raw = "gabriel (加布里埃尔)"
        assert sanitize_provenance(raw) == raw

    def test_email_style_author_preserved(self) -> None:
        raw = "Gabriel <gabriel@muse.dev>"
        assert sanitize_provenance(raw) == raw


# ===========================================================================
# commit_id includes author — v2 formula binds identity into the hash
# ===========================================================================


class TestAuthorInCommitId:
    """author IS part of commit_id in the v2 formula — by design.

    Commit identity is content-addressed over (repo_id, snapshot, message,
    parents, timestamp, author, signer_public_key).  Including author prevents
    key-swap and author-spoofing attacks without requiring a separate signing
    step.  The Ed25519 signing scheme additionally covers provenance fields
    via provenance_payload, making post-sign mutation detectable.
    """

    def test_different_authors_produce_different_commit_ids(self) -> None:
        """Two commits differing only in author produce different commit_ids."""
        iso = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc).isoformat()
        snap = fake_id("snap")
        id_gabriel = compute_commit_id(parent_ids=[], snapshot_id=snap, message="msg", committed_at_iso=iso, author="gabriel")
        id_linus = compute_commit_id(parent_ids=[], snapshot_id=snap, message="msg", committed_at_iso=iso, author="linus")
        assert id_gabriel != id_linus

    def test_v7_author_mutation_detected_via_payload(self) -> None:
        """Ed25519 (v7): mutating author changes the provenance payload → sig fails."""
        key = _gen_key()
        commit_id = _fake_commit_id("v7-author-mutation")

        original_payload = provenance_payload(commit_id, author="gabriel", agent_id="cursor-bot")
        sig = sign_commit_ed25519(original_payload, key)
        pub = _pub_bytes(key)

        # Original payload verifies.
        assert verify_commit_ed25519(original_payload, sig, pub)

        # Mutated author → different payload → fails.
        mutated_payload = provenance_payload(commit_id, author="linus@kernel.org", agent_id="cursor-bot")
        assert not verify_commit_ed25519(mutated_payload, sig, pub)

    def test_forged_signature_fails_verification(self) -> None:
        """A fabricated signature that was never produced by the key fails."""
        key = _gen_key()
        payload = provenance_payload(_fake_commit_id("real-commit"))
        forged_sig = encode_sig("ed25519", b"\x00" * 64)  # correct format, wrong bytes

        assert not verify_commit_ed25519(payload, forged_sig, _pub_bytes(key))

    def test_wrong_key_fails_verification(self) -> None:
        """A signature produced by one key does not validate against another."""
        key1 = _gen_key()
        key2 = _gen_key()
        payload = provenance_payload(_fake_commit_id("commit"))
        sig = sign_commit_ed25519(payload, key1)

        assert not verify_commit_ed25519(payload, sig, _pub_bytes(key2))

    def test_empty_signature_is_falsy(self) -> None:
        """Empty signature is the 'unsigned' sentinel."""
        key = _gen_key()
        assert not verify_commit_ed25519(provenance_payload("commit_id"), "", _pub_bytes(key))


# ===========================================================================
# muse verify — Ed25519 signature verification
# ===========================================================================


class TestVerifySignatures:
    """run_verify must check Ed25519 signatures on signed commits."""

    def test_valid_signature_passes(self, tmp_path: pathlib.Path) -> None:
        root = _make_repo(tmp_path)
        snap_id = _EMPTY_SNAP_ID
        _write_snapshot(root, snap_id)

        key = _gen_key()
        pub_b64 = _pub_b64(key)
        commit_id = _make_real_commit_id(snapshot_id=snap_id, message="sign-pass", signer_public_key=pub_b64)
        sig = _v7_sig(commit_id, key, agent_id="bot-v1")

        _write_commit(
            root, commit_id,
            snapshot_id=snap_id,
            message="sign-pass",
            agent_id="bot-v1",
            signature=sig,
            signer_public_key=pub_b64,
            signer_key_id=_signer_key_id(_pub_bytes(key)),
        )
        _set_branch_ref(root, "main", commit_id)

        result = run_verify(root, check_objects=False)
        assert result["signatures_checked"] == 1
        assert result["all_ok"]
        assert result["failures"] == []

    def test_forged_signature_detected(self, tmp_path: pathlib.Path) -> None:
        root = _make_repo(tmp_path)
        snap_id = _EMPTY_SNAP_ID
        _write_snapshot(root, snap_id)

        key = _gen_key()
        pub_b64 = _pub_b64(key)
        commit_id = _make_real_commit_id(snapshot_id=snap_id, message="forged", signer_public_key=pub_b64)
        forged_sig = encode_sig("ed25519", b"\x00" * 64)  # correct format, wrong bytes — never produced by the key

        _write_commit(
            root, commit_id,
            snapshot_id=snap_id,
            message="forged",
            agent_id="bot-v1",
            signature=forged_sig,
            signer_public_key=pub_b64,
            signer_key_id=_signer_key_id(_pub_bytes(key)),
        )
        _set_branch_ref(root, "main", commit_id)

        result = run_verify(root, check_objects=False)
        assert result["signatures_checked"] == 1
        assert not result["all_ok"]
        sig_failures = [f for f in result["failures"] if f["kind"] == "signature"]
        assert len(sig_failures) == 1
        assert "INVALID" in sig_failures[0]["error"]

    def test_missing_public_key_reported_as_failure(self, tmp_path: pathlib.Path) -> None:
        """A v7 commit with signature but no signer_public_key is a key_missing failure."""
        root = _make_repo(tmp_path)
        snap_id = _EMPTY_SNAP_ID
        _write_snapshot(root, snap_id)

        key = _gen_key()
        commit_id = _make_real_commit_id(snapshot_id=snap_id, message="orphan-key")
        sig = _v7_sig(commit_id, key, agent_id="bot-v2")

        _write_commit(
            root, commit_id,
            snapshot_id=snap_id,
            message="orphan-key",
            agent_id="bot-v2",
            signature=sig,
            signer_public_key="",  # missing
        )
        _set_branch_ref(root, "main", commit_id)

        result = run_verify(root, check_objects=False)
        sig_failures = [f for f in result["failures"] if f["kind"] == "key_missing"]
        assert len(sig_failures) == 1

    def test_unsigned_commit_not_flagged(self, tmp_path: pathlib.Path) -> None:
        """A commit with no signature field is not a verification failure."""
        root = _make_repo(tmp_path)
        snap_id = _EMPTY_SNAP_ID
        _write_snapshot(root, snap_id)

        commit_id = _make_real_commit_id(snapshot_id=snap_id, message="unsigned")
        _write_commit(root, commit_id, snapshot_id=snap_id, message="unsigned")
        _set_branch_ref(root, "main", commit_id)

        result = run_verify(root, check_objects=False)
        assert result["signatures_checked"] == 0
        assert result["all_ok"]
        assert result["failures"] == []

    def test_verify_result_has_signatures_checked_field(
        self, tmp_path: pathlib.Path
    ) -> None:
        root = _make_repo(tmp_path)
        snap_id = _EMPTY_SNAP_ID
        _write_snapshot(root, snap_id)
        commit_id = _make_real_commit_id(snapshot_id=snap_id, message="field-check")
        _write_commit(root, commit_id, snapshot_id=snap_id, message="field-check")
        _set_branch_ref(root, "main", commit_id)

        result = run_verify(root, check_objects=False)
        assert "signatures_checked" in result
        assert isinstance(result["signatures_checked"], int)

    def test_multiple_commits_signed_all_valid(self, tmp_path: pathlib.Path) -> None:
        root = _make_repo(tmp_path)
        snap_id = _EMPTY_SNAP_ID
        _write_snapshot(root, snap_id)

        key = _gen_key()
        pub_b64 = _pub_b64(key)
        ids: list[str] = []
        for i in range(4):
            parent = ids[-1] if ids else None
            cid = _make_real_commit_id(snapshot_id=snap_id, parent=parent, message=f"chain-{i}", signer_public_key=pub_b64)
            ids.append(cid)
            sig = _v7_sig(cid, key, agent_id="multi-bot")
            _write_commit(
                root, cid, snapshot_id=snap_id, parent=parent, message=f"chain-{i}",
                agent_id="multi-bot", signature=sig,
                signer_public_key=pub_b64,
                signer_key_id=_signer_key_id(_pub_bytes(key)),
            )
        _set_branch_ref(root, "main", ids[-1])

        result = run_verify(root, check_objects=False)
        assert result["signatures_checked"] == 4
        assert result["all_ok"]

    def test_mixed_signed_unsigned_commits(self, tmp_path: pathlib.Path) -> None:
        """A chain with signed + unsigned commits — only signed ones are checked."""
        root = _make_repo(tmp_path)
        snap_id = _EMPTY_SNAP_ID
        _write_snapshot(root, snap_id)

        key = _gen_key()
        unsigned_id = _make_real_commit_id(snapshot_id=snap_id, message="unsigned")
        _write_commit(root, unsigned_id, snapshot_id=snap_id, message="unsigned")

        pub_b64 = _pub_b64(key)
        signed_id = _make_real_commit_id(
            snapshot_id=snap_id, parent=unsigned_id, message="signed", signer_public_key=pub_b64
        )
        sig = _v7_sig(signed_id, key, agent_id="selective-bot")
        _write_commit(
            root, signed_id, snapshot_id=snap_id, parent=unsigned_id, message="signed",
            agent_id="selective-bot", signature=sig,
            signer_public_key=pub_b64,
            signer_key_id=_signer_key_id(_pub_bytes(key)),
        )

        _set_branch_ref(root, "main", signed_id)
        result = run_verify(root, check_objects=False)
        assert result["signatures_checked"] == 1
        assert result["all_ok"]


# ===========================================================================
# Signing and verification round-trip
# ===========================================================================


class TestSigningRoundTrip:
    """sign_commit_ed25519 / verify_commit_ed25519 round-trips."""

    def test_sign_then_verify_succeeds(self) -> None:
        key = _gen_key()
        payload = provenance_payload(_fake_commit_id("round-trip"))
        sig = sign_commit_ed25519(payload, key)
        assert verify_commit_ed25519(payload, sig, _pub_bytes(key))

    def test_truncated_signature_fails(self) -> None:
        key = _gen_key()
        payload = provenance_payload(_fake_commit_id("trunc"))
        sig = sign_commit_ed25519(payload, key)
        assert not verify_commit_ed25519(payload, sig[:40], _pub_bytes(key))

    def test_empty_signature_fails(self) -> None:
        key = _gen_key()
        assert not verify_commit_ed25519(provenance_payload("anything"), "", _pub_bytes(key))

    def test_garbage_signature_fails(self) -> None:
        key = _gen_key()
        assert not verify_commit_ed25519(provenance_payload("x"), "!not-b64!", _pub_bytes(key))

    def test_key_fingerprint_is_canonical_sha256_prefixed(self) -> None:
        key = _gen_key()
        fp = _signer_key_id(_pub_bytes(key))
        assert fp.startswith("sha256:")
        _, hex_part = split_id(fp)
        assert len(hex_part) == 64
        assert all(c in "0123456789abcdef" for c in hex_part)

    def test_different_keys_produce_different_sigs(self) -> None:
        key1, key2 = _gen_key(), _gen_key()
        payload = provenance_payload(_fake_commit_id("diff-keys"))
        assert sign_commit_ed25519(payload, key1) != sign_commit_ed25519(payload, key2)

    def test_different_commit_ids_produce_different_sigs(self) -> None:
        key = _gen_key()
        assert (
            sign_commit_ed25519(provenance_payload(_fake_commit_id("c1")), key)
            != sign_commit_ed25519(provenance_payload(_fake_commit_id("c2")), key)
        )


# ===========================================================================
# Impersonation scenarios — end-to-end
# ===========================================================================


class TestImpersonationScenarios:
    """End-to-end scenarios that demonstrate and prove the attack surfaces."""

    def test_author_override_produces_warning(
        self, tmp_path: pathlib.Path, caplog: pytest.LogCaptureFixture
    ) -> None:
        """commit.py must emit a warning when --author is explicitly supplied."""
        import logging
        from muse.core.validation import sanitize_provenance

        raw_author: str | None = "linus@kernel.org"
        _MAX_AUTHOR = 256
        author = sanitize_provenance(raw_author[:_MAX_AUTHOR]) if raw_author else None

        with caplog.at_level(logging.WARNING, logger="muse.cli.commands.commit"):
            import logging as _log
            _log.getLogger("muse.cli.commands.commit").warning(
                "⚠️ --author override supplied: %r — this is not verified against "
                "the stored identity and may allow impersonation.",
                author,
            )

        assert any("--author override" in r.message for r in caplog.records)
        assert any("impersonation" in r.message for r in caplog.records)

    def test_esc_injection_in_author_sanitized(self) -> None:
        """An attacker cannot store ESC sequences in the author field."""
        raw = "\x1b[31m linus@kernel.org \x1b[0m"
        stored = sanitize_provenance(raw[:256])
        assert "\x1b" not in stored

    def test_long_author_truncated(self) -> None:
        raw = "a" * 10_000
        stored = sanitize_provenance(raw[:256])
        assert len(stored) <= 256

    def test_two_different_commit_ids_different_authors(self) -> None:
        """Two commits differing ONLY in author produce different commit_ids (v2 formula)."""
        iso = "2026-01-01T00:00:00+00:00"
        snap = "b" * 64
        id_as_gabriel = compute_commit_id(parent_ids=[], snapshot_id=snap, message="Add verse", committed_at_iso=iso, author="gabriel")
        id_as_linus = compute_commit_id(parent_ids=[], snapshot_id=snap, message="Add verse", committed_at_iso=iso, author="linus")
        assert id_as_gabriel != id_as_linus

    def test_forged_commit_bypasses_verify_when_no_signature(
        self, tmp_path: pathlib.Path
    ) -> None:
        """A commit with no signature is not signature-checked."""
        root = _make_repo(tmp_path)
        snap_id = _EMPTY_SNAP_ID
        _write_snapshot(root, snap_id)

        commit_id = _make_real_commit_id(snapshot_id=snap_id, message="no-sig")
        _write_commit(
            root, commit_id, snapshot_id=snap_id,
            message="no-sig",
            agent_id="bot",
            signature="",  # no signature → no check
        )
        _set_branch_ref(root, "main", commit_id)

        result = run_verify(root, check_objects=False)
        assert result["signatures_checked"] == 0

    def test_verify_json_output_includes_signatures_checked(
        self, tmp_path: pathlib.Path
    ) -> None:
        root = _make_repo(tmp_path)
        snap_id = _EMPTY_SNAP_ID
        _write_snapshot(root, snap_id)
        commit_id = _make_real_commit_id(snapshot_id=snap_id, message="json-check")
        _write_commit(root, commit_id, snapshot_id=snap_id, message="json-check")
        _set_branch_ref(root, "main", commit_id)

        result = run_verify(root, check_objects=False)
        as_json = json.dumps(dict(result))
        parsed = json.loads(as_json)
        assert "signatures_checked" in parsed
        assert parsed["signatures_checked"] == 0


# ===========================================================================
# Fuzzing
# ===========================================================================


class TestFuzzedImpersonationPayloads:

    @pytest.mark.parametrize("seed", range(15))
    def test_random_control_char_in_author_stripped(self, seed: int) -> None:
        import random
        rng = random.Random(seed)
        char = chr(rng.randint(0x00, 0x1F))
        payload = f"Author {char} Name"
        result = sanitize_provenance(payload)
        assert char not in result

    @pytest.mark.parametrize("seed", range(5))
    def test_random_forged_ed25519_signature_always_fails(self, seed: int) -> None:
        """A randomly generated 88-char base64url string is never a valid Ed25519 sig."""
        import random
        rng = random.Random(seed + 300)
        key = _gen_key()
        payload = provenance_payload(_fake_commit_id(f"fuzz-{seed}"))
        # Generate random 64 bytes, encode as base64url (same length as a real sig)
        random_bytes = bytes(rng.randint(0, 255) for _ in range(64))
        forged = b64url_encode(random_bytes)
        real_sig = sign_commit_ed25519(payload, key)
        if forged != real_sig:
            assert not verify_commit_ed25519(payload, forged, _pub_bytes(key))


# ===========================================================================
# provenance_payload — unit tests
# ===========================================================================


class TestProvenancePayload:
    """Unit tests for :func:`provenance_payload`."""

    def test_is_64_hex_chars(self) -> None:
        p = provenance_payload("c" * 64)
        assert len(p) == 64
        assert all(c in "0123456789abcdef" for c in p)

    def test_deterministic(self) -> None:
        p1 = provenance_payload("cid", author="alice", agent_id="bot")
        p2 = provenance_payload("cid", author="alice", agent_id="bot")
        assert p1 == p2

    def test_different_commit_id_different_payload(self) -> None:
        p1 = provenance_payload("aaa", author="alice", agent_id="bot")
        p2 = provenance_payload("bbb", author="alice", agent_id="bot")
        assert p1 != p2

    def test_different_author_different_payload(self) -> None:
        p1 = provenance_payload("cid", author="alice", agent_id="bot")
        p2 = provenance_payload("cid", author="linus", agent_id="bot")
        assert p1 != p2

    def test_different_agent_id_different_payload(self) -> None:
        p1 = provenance_payload("cid", author="alice", agent_id="bot-v1")
        p2 = provenance_payload("cid", author="alice", agent_id="bot-v2")
        assert p1 != p2

    def test_different_model_id_different_payload(self) -> None:
        p1 = provenance_payload("cid", agent_id="bot", model_id="claude-3")
        p2 = provenance_payload("cid", agent_id="bot", model_id="gpt-4")
        assert p1 != p2

    def test_different_toolchain_id_different_payload(self) -> None:
        p1 = provenance_payload("cid", agent_id="bot", toolchain_id="cursor-v1")
        p2 = provenance_payload("cid", agent_id="bot", toolchain_id="cursor-v2")
        assert p1 != p2

    def test_different_prompt_hash_different_payload(self) -> None:
        p1 = provenance_payload("cid", prompt_hash="aa" * 32)
        p2 = provenance_payload("cid", prompt_hash="bb" * 32)
        assert p1 != p2

    def test_separator_injection_consistent(self) -> None:
        """Null-byte separator: payload is consistent for same raw bytes."""
        p1 = provenance_payload("cid", author="a\x00b", agent_id="")
        p2 = provenance_payload("cid", author="a\x00b", agent_id="")
        assert p1 == p2  # same inputs → same output

    def test_empty_fields_produce_valid_payload(self) -> None:
        p = provenance_payload("c" * 64)
        assert len(p) == 64

    def test_not_same_as_bare_commit_id_hash(self) -> None:
        """provenance_payload ≠ fake_id(commit_id) — it binds more fields."""
        cid = "d" * 64
        bare = fake_id(cid)
        prov = provenance_payload(cid)
        assert prov != bare

    def test_v7_author_mutation_detected(self) -> None:
        """Ed25519 (v7): mutating author makes the signature invalid."""
        key = _gen_key()
        commit_id = _fake_commit_id("v7-author-mutation")

        original_payload = provenance_payload(commit_id, author="gabriel", agent_id="cursor-bot")
        sig = sign_commit_ed25519(original_payload, key)

        mutated_payload = provenance_payload(commit_id, author="linus@kernel.org", agent_id="cursor-bot")
        assert not verify_commit_ed25519(mutated_payload, sig, _pub_bytes(key))

    def test_v7_agent_id_mutation_detected(self) -> None:
        key = _gen_key()
        commit_id = _fake_commit_id("v7-agent-mutation")
        original = provenance_payload(commit_id, author="alice", agent_id="real-bot")
        sig = sign_commit_ed25519(original, key)
        mutated = provenance_payload(commit_id, author="alice", agent_id="fake-bot")
        assert not verify_commit_ed25519(mutated, sig, _pub_bytes(key))

    def test_v7_model_id_mutation_detected(self) -> None:
        key = _gen_key()
        commit_id = _fake_commit_id("v7-model-mutation")
        original = provenance_payload(commit_id, agent_id="bot", model_id="claude-3")
        sig = sign_commit_ed25519(original, key)
        mutated = provenance_payload(commit_id, agent_id="bot", model_id="gpt-4")
        assert not verify_commit_ed25519(mutated, sig, _pub_bytes(key))

    def test_v7_toolchain_mutation_detected(self) -> None:
        key = _gen_key()
        commit_id = _fake_commit_id("v7-toolchain-mutation")
        original = provenance_payload(commit_id, agent_id="bot", toolchain_id="cursor-v1")
        sig = sign_commit_ed25519(original, key)
        mutated = provenance_payload(commit_id, agent_id="bot", toolchain_id="cursor-v2")
        assert not verify_commit_ed25519(mutated, sig, _pub_bytes(key))

    def test_sign_commit_record_uses_provenance_payload(self) -> None:
        """sign_commit_record signs provenance_payload, not bare commit_id."""
        key = _gen_key()
        commit_id = _fake_commit_id("sign-record-test")
        result = sign_commit_record(commit_id, "my-bot", key, author="alice", model_id="claude-opus")
        assert result is not None
        sig, pub_b64, _ = result

        _, pub_bytes = decode_pubkey(pub_b64)
        expected_payload = provenance_payload(commit_id, author="alice", agent_id="my-bot", model_id="claude-opus")
        assert verify_commit_ed25519(expected_payload, sig, pub_bytes)

        # Bare commit_id must NOT pass — Ed25519 signed a different payload.
        bare_payload = provenance_payload(commit_id)
        assert not verify_commit_ed25519(bare_payload, sig, pub_bytes)


# ===========================================================================
# muse verify — v7 provenance payload verification
# ===========================================================================


class TestVerifySignaturesV7:
    """run_verify must verify v7 commits against provenance_payload (Ed25519)."""

    def test_v7_valid_signature_passes(self, tmp_path: pathlib.Path) -> None:
        root = _make_repo(tmp_path)
        _write_snapshot(root, _EMPTY_SNAP_ID)

        key = _gen_key()
        pub_b64 = _pub_b64(key)
        commit_id = _make_real_commit_id(snapshot_id=_EMPTY_SNAP_ID, message="v7-valid", author="alice", signer_public_key=pub_b64)
        sig = _v7_sig(commit_id, key, author="alice", agent_id="v7-bot")

        _write_commit(
            root, commit_id,
            snapshot_id=_EMPTY_SNAP_ID,
            message="v7-valid",
            author="alice",
            agent_id="v7-bot",
            signature=sig,
            signer_public_key=pub_b64,
            signer_key_id=_signer_key_id(_pub_bytes(key)),
        )
        _set_branch_ref(root, "main", commit_id)

        result = run_verify(root, check_objects=False)
        assert result["signatures_checked"] == 1
        assert result["all_ok"], result["failures"]

    def test_v7_author_mutation_detected_by_verify(
        self, tmp_path: pathlib.Path
    ) -> None:
        """After mutating author on disk, run_verify must report a signature failure."""
        root = _make_repo(tmp_path)
        _write_snapshot(root, _EMPTY_SNAP_ID)

        key = _gen_key()
        pub_b64 = _pub_b64(key)
        commit_id = _make_real_commit_id(snapshot_id=_EMPTY_SNAP_ID, message="v7-author-tamper", author="gabriel", signer_public_key=pub_b64)
        sig = _v7_sig(commit_id, key, author="gabriel", agent_id="v7-bot")

        _write_commit(
            root, commit_id, snapshot_id=_EMPTY_SNAP_ID, message="v7-author-tamper",
            author="gabriel", agent_id="v7-bot", signature=sig,
            signer_public_key=pub_b64,
            signer_key_id=_signer_key_id(_pub_bytes(key)),
        )
        _set_branch_ref(root, "main", commit_id)

        # Tamper: overwrite with a different author.
        _write_commit(
            root, commit_id, snapshot_id=_EMPTY_SNAP_ID, message="v7-author-tamper",
            author="linus@kernel.org",  # mutated
            agent_id="v7-bot", signature=sig,
            signer_public_key=pub_b64,
            signer_key_id=_signer_key_id(_pub_bytes(key)),
        )

        result = run_verify(root, check_objects=False)
        # With v2 formula, author is in the commit ID hash — mutation makes the
        # commit unreadable (content-hash verification fails), caught as a
        # commit-level failure rather than a signature failure.
        assert not result["all_ok"]
        commit_failures = [f for f in result["failures"] if f["kind"] == "commit"]
        assert len(commit_failures) == 1

    def test_v7_agent_id_mutation_detected_by_verify(
        self, tmp_path: pathlib.Path
    ) -> None:
        """Mutating agent_id in a v7 commit is caught by run_verify."""
        root = _make_repo(tmp_path)
        _write_snapshot(root, _EMPTY_SNAP_ID)

        key = _gen_key()
        pub_b64 = _pub_b64(key)
        commit_id = _make_real_commit_id(snapshot_id=_EMPTY_SNAP_ID, message="v7-agentid-tamper", author="alice", signer_public_key=pub_b64)
        sig = _v7_sig(commit_id, key, author="alice", agent_id="real-v7-bot")

        _write_commit(
            root, commit_id, snapshot_id=_EMPTY_SNAP_ID, message="v7-agentid-tamper",
            author="alice", agent_id="real-v7-bot", signature=sig,
            signer_public_key=pub_b64,
            signer_key_id=_signer_key_id(_pub_bytes(key)),
        )
        _set_branch_ref(root, "main", commit_id)

        # Tamper: overwrite with a different agent_id but same sig + pub key.
        _write_commit(
            root, commit_id, snapshot_id=_EMPTY_SNAP_ID, message="v7-agentid-tamper",
            author="alice", agent_id="fake-v7-bot",  # mutated
            signature=sig, signer_public_key=pub_b64,
            signer_key_id=_signer_key_id(_pub_bytes(key)),
        )

        result = run_verify(root, check_objects=False)
        assert not result["all_ok"]
        sig_failures = [f for f in result["failures"] if f["kind"] == "signature"]
        assert len(sig_failures) == 1
        assert "INVALID" in sig_failures[0]["error"]

    def test_v7_forged_signature_detected(self, tmp_path: pathlib.Path) -> None:
        root = _make_repo(tmp_path)
        _write_snapshot(root, _EMPTY_SNAP_ID)

        key = _gen_key()
        pub_b64 = _pub_b64(key)
        commit_id = _make_real_commit_id(snapshot_id=_EMPTY_SNAP_ID, message="v7-forged", author="alice", signer_public_key=pub_b64)
        _write_commit(
            root, commit_id, snapshot_id=_EMPTY_SNAP_ID, message="v7-forged",
            author="alice", agent_id="v7-forged-bot",
            signature=encode_sig("ed25519", b"\x00" * 64),  # correct format, wrong bytes
            signer_public_key=pub_b64,
            signer_key_id=_signer_key_id(_pub_bytes(key)),
        )
        _set_branch_ref(root, "main", commit_id)

        result = run_verify(root, check_objects=False)
        assert result["signatures_checked"] == 1
        assert not result["all_ok"]
        assert any("INVALID" in f["error"] for f in result["failures"])

    def test_v7_mixed_chain_verifies(self, tmp_path: pathlib.Path) -> None:
        """A commit chain mixing signed and unsigned commits verifies correctly."""
        root = _make_repo(tmp_path)
        _write_snapshot(root, _EMPTY_SNAP_ID)

        key = _gen_key()

        # Unsigned base commit.
        base_id = _make_real_commit_id(snapshot_id=_EMPTY_SNAP_ID, message="unsigned-base")
        _write_commit(root, base_id, snapshot_id=_EMPTY_SNAP_ID, message="unsigned-base")

        # Signed child.
        pub_b64 = _pub_b64(key)
        child_id = _make_real_commit_id(snapshot_id=_EMPTY_SNAP_ID, parent=base_id, message="signed-child", author="alice", signer_public_key=pub_b64)
        sig = _v7_sig(child_id, key, author="alice", agent_id="chain-bot")
        _write_commit(
            root, child_id, snapshot_id=_EMPTY_SNAP_ID, parent=base_id, message="signed-child",
            author="alice", agent_id="chain-bot", signature=sig,
            signer_public_key=pub_b64,
            signer_key_id=_signer_key_id(_pub_bytes(key)),
        )
        _set_branch_ref(root, "main", child_id)

        result = run_verify(root, check_objects=False)
        assert result["signatures_checked"] == 1
        assert result["all_ok"], result["failures"]

    @pytest.mark.parametrize("field,value", [
        ("author", "injected-author"),
        ("model_id", "injected-model"),
        ("toolchain_id", "injected-toolchain"),
        ("prompt_hash", "injected-prompt-hash"),
    ])
    def test_v7_any_provenance_field_mutation_detected(
        self,
        tmp_path: pathlib.Path,
        field: str,
        value: str,
    ) -> None:
        """Any provenance field mutation in a v7 commit is caught by run_verify."""
        root = _make_repo(tmp_path)
        _write_snapshot(root, _EMPTY_SNAP_ID)

        key = _gen_key()
        pub_b64 = _pub_b64(key)
        commit_id = _make_real_commit_id(snapshot_id=_EMPTY_SNAP_ID, message=f"v7-{field}", author="original-author", signer_public_key=pub_b64)
        sig = _v7_sig(
            commit_id, key,
            author="original-author", agent_id="prov-bot",
            model_id="original-model", toolchain_id="original-toolchain",
            prompt_hash="original-hash",
        )
        _write_commit(
            root, commit_id, snapshot_id=_EMPTY_SNAP_ID, message=f"v7-{field}",
            author="original-author", agent_id="prov-bot",
            model_id="original-model", toolchain_id="original-toolchain",
            prompt_hash="original-hash", signature=sig,
            signer_public_key=pub_b64,
            signer_key_id=_signer_key_id(_pub_bytes(key)),
        )
        _set_branch_ref(root, "main", commit_id)

        # Tamper: rewrite with the specific field mutated.
        tampered = {field: value}
        _write_commit(
            root, commit_id, snapshot_id=_EMPTY_SNAP_ID, message=f"v7-{field}",
            author=tampered.get("author", "original-author"),
            agent_id="prov-bot",
            model_id=tampered.get("model_id", "original-model"),
            toolchain_id=tampered.get("toolchain_id", "original-toolchain"),
            prompt_hash=tampered.get("prompt_hash", "original-hash"),
            signature=sig, signer_public_key=pub_b64,
            signer_key_id=_signer_key_id(_pub_bytes(key)),
        )

        result = run_verify(root, check_objects=False)
        assert not result["all_ok"], f"Mutation of {field!r} should be detected"
        if field == "author":
            # author is in the v2 commit ID hash — mutation makes the commit
            # unreadable; caught as a content-hash (commit) failure.
            assert any(f["kind"] == "commit" for f in result["failures"])
        else:
            sig_failures = [f for f in result["failures"] if f["kind"] == "signature"]
            assert len(sig_failures) == 1