tests/test_delivery_bundle.py

"""
Tests for Stage 68 — bundle export with manifest.

bundle_pilot_output zips a pilot output dir + writes a manifest.json
(sha256 per file). verify_bundle re-hashes and compares. The headline
test is a round-trip: bundle → unzip → verify checksums match.
"""
import hashlib
import json
import zipfile

import pytest

from delivery.bundle import bundle_pilot_output, verify_bundle


def _populate(tmp_path):
    """Build a tiny pilot-shaped dir with predictable content."""
    src = tmp_path / "pilot_out"
    src.mkdir()
    (src / "customer_report.html").write_text("<h1>Report</h1>",
                                                encoding="utf-8")
    (src / "customer_report.pdf").write_bytes(b"%PDF-stub\nfake")
    (src / "decision_queue.html").write_text("<h1>Queue</h1>",
                                                encoding="utf-8")
    (src / "evidence").mkdir()
    (src / "evidence" / "run_xyz.html").write_text("<h1>Evidence</h1>",
                                                      encoding="utf-8")
    return src


def test_bundle_creates_zip_next_to_source(tmp_path):
    src = _populate(tmp_path)
    out = bundle_pilot_output(src)
    assert out.exists()
    assert out.suffix == ".zip"
    assert out.parent == src.parent
    assert out.stem == src.name


def test_bundle_explicit_path(tmp_path):
    src = _populate(tmp_path)
    target = tmp_path / "custom_name.zip"
    out = bundle_pilot_output(src, bundle_path=target)
    assert out == target.resolve()
    assert out.exists()


def test_bundle_includes_every_source_file(tmp_path):
    src = _populate(tmp_path)
    out = bundle_pilot_output(src)
    with zipfile.ZipFile(out) as zf:
        names = set(zf.namelist())
    # manifest + the four files we wrote, with relative paths
    assert "manifest.json" in names
    assert "customer_report.html" in names
    assert "customer_report.pdf" in names
    assert "decision_queue.html" in names
    assert "evidence/run_xyz.html" in names


def test_bundle_manifest_carries_sha256_per_file(tmp_path):
    src = _populate(tmp_path)
    out = bundle_pilot_output(src)
    with zipfile.ZipFile(out) as zf:
        manifest = json.loads(zf.read("manifest.json"))
    assert manifest["source_dir_name"] == src.name
    assert manifest["n_files"] == 4
    assert manifest["total_bytes"] > 0
    # spot-check one file's hash
    by_path = {f["path"]: f for f in manifest["files"]}
    expected = hashlib.sha256(b"<h1>Report</h1>").hexdigest()
    assert by_path["customer_report.html"]["sha256"] == expected
    assert by_path["customer_report.html"]["bytes"] == len(b"<h1>Report</h1>")


def test_bundle_manifest_files_sorted(tmp_path):
    """Deterministic manifest — same input gives same file order, so
    diffing two bundles is meaningful."""
    src = _populate(tmp_path)
    out = bundle_pilot_output(src)
    with zipfile.ZipFile(out) as zf:
        manifest = json.loads(zf.read("manifest.json"))
    paths = [f["path"] for f in manifest["files"]]
    assert paths == sorted(paths)


def test_bundle_missing_source_dir_raises(tmp_path):
    with pytest.raises(FileNotFoundError):
        bundle_pilot_output(tmp_path / "nope")


# --- verify_bundle ----------------------------------------------------

def test_verify_bundle_returns_ok_on_intact_zip(tmp_path):
    src = _populate(tmp_path)
    out = bundle_pilot_output(src)
    report = verify_bundle(out)
    assert report["ok"] is True
    assert report["n_files"] == 4
    assert report["n_mismatches"] == 0
    assert report["mismatches"] == []


def test_verify_bundle_detects_tampered_file(tmp_path):
    """The whole point of the manifest — if a file in the bundle was
    modified, verify must spot it."""
    src = _populate(tmp_path)
    out = bundle_pilot_output(src)
    # rewrite the zip, replacing one of the files with different bytes
    tampered = tmp_path / "tampered.zip"
    with zipfile.ZipFile(out, "r") as zin, \
            zipfile.ZipFile(tampered, "w", zipfile.ZIP_DEFLATED) as zout:
        for name in zin.namelist():
            data = zin.read(name)
            if name == "customer_report.html":
                data = b"<h1>MALICIOUS</h1>"
            zout.writestr(name, data)
    report = verify_bundle(tampered)
    assert report["ok"] is False
    assert report["n_mismatches"] == 1
    assert (report["mismatches"][0]["path"]
             == "customer_report.html")


# --- end-to-end through the pilot CLI -------------------------------

def test_pilot_cli_bundle_flag_creates_zip(tmp_path):
    """The pilot's --bundle flag end-to-end: real pilot run + bundle
    write. Doesn't exhaustively check the bundle (the helper tests
    do that) — just confirms the CLI wiring works."""
    from delivery.pilot import _cli
    out_dir = tmp_path / "pilot_out"
    rc = _cli([
        "--seed", "7",
        "--out", str(out_dir),
        "--bundle",
    ])
    assert rc == 0
    zip_path = out_dir.with_suffix(".zip")
    assert zip_path.exists()
    # the zip is verifiable
    assert verify_bundle(zip_path)["ok"]


def test_pilot_cli_no_bundle_flag_skips_zip(tmp_path):
    """Backward compat — pre-Stage-68 cron lines that don't know
    about --bundle still work as before."""
    from delivery.pilot import _cli
    out_dir = tmp_path / "pilot_out"
    rc = _cli([
        "--seed", "7",
        "--out", str(out_dir),
    ])
    assert rc == 0
    zip_path = out_dir.with_suffix(".zip")
    assert not zip_path.exists()