openrun/tests/unit/test_fit_download.py

"""Path B per-second: pulling original FITs from the API instead of a website export.

`download_fit` / `backfill_fits` are mocked at the `garth.download` boundary
(network-dependent, upstream-deprecated — see ROADMAP test conventions). The FIT
extractor is pure and tested directly.
"""

from __future__ import annotations

import io
import zipfile

import pytest

from openrun.ingest import garmin_api as g

# A FIT header carries the literal b".FIT" at bytes 8..12; payload after is opaque here.
FIT_BYTES = b"\x0e\x10\x00\x00\x20\x00\x00\x00.FITdata-goes-here"


def _zip(*members: tuple[str, bytes]) -> bytes:
    buf = io.BytesIO()
    with zipfile.ZipFile(buf, "w") as zf:
        for name, data in members:
            zf.writestr(name, data)
    return buf.getvalue()


# --- _extract_fit_bytes ----------------------------------------------------

def test_extract_from_zip():
    assert g._extract_fit_bytes(_zip(("9.fit", FIT_BYTES))) == FIT_BYTES


def test_extract_from_zip_extensionless_member_picks_largest():
    blob = _zip(("small", b"x"), ("big", FIT_BYTES))
    assert g._extract_fit_bytes(blob) == FIT_BYTES


def test_extract_from_bare_fit():
    assert g._extract_fit_bytes(FIT_BYTES) == FIT_BYTES


@pytest.mark.parametrize("blob", [None, b"", b"not a fit at all"])
def test_extract_rejects_non_fit(blob):
    assert g._extract_fit_bytes(blob) is None


# --- download_fit ----------------------------------------------------------

def _seed_activity(conn, aid: int, atype: str = "running") -> None:
    conn.execute(
        "INSERT INTO activities (activity_id, activity_type, start_time_gmt, raw, fetched_at) "
        "VALUES (?, ?, ?, ?, datetime('now'))",
        (aid, atype, "2026-05-01T10:00:00", "{}"),
    )


def test_download_fit_writes_and_links(tmp_conn, tmp_path, monkeypatch):
    _seed_activity(tmp_conn, 42)
    monkeypatch.setattr(g.garth, "download", lambda path: _zip(("42.fit", FIT_BYTES)))

    ok = g.download_fit(tmp_conn, 42, fit_dir=tmp_path)

    assert ok is True
    dest = tmp_path / "42.fit"
    assert dest.read_bytes() == FIT_BYTES
    row = tmp_conn.execute(
        "SELECT fit_path FROM activity_fit_files WHERE activity_id = 42"
    ).fetchone()
    assert row is not None and row[0].endswith("42.fit")


def test_download_fit_skips_when_already_linked(tmp_conn, tmp_path, monkeypatch):
    _seed_activity(tmp_conn, 7)
    (tmp_path / "7.fit").write_bytes(FIT_BYTES)
    g.record_link(tmp_conn, 7, tmp_path / "7.fit")

    calls = []
    monkeypatch.setattr(g.garth, "download", lambda path: calls.append(path) or b"")

    assert g.download_fit(tmp_conn, 7, fit_dir=tmp_path) is True
    assert calls == []  # no network call when on disk + linked


def test_download_fit_force_redownloads(tmp_conn, tmp_path, monkeypatch):
    _seed_activity(tmp_conn, 7)
    (tmp_path / "7.fit").write_bytes(b"stale")
    g.record_link(tmp_conn, 7, tmp_path / "7.fit")
    monkeypatch.setattr(g.garth, "download", lambda path: _zip(("7.fit", FIT_BYTES)))

    assert g.download_fit(tmp_conn, 7, fit_dir=tmp_path, force=True) is True
    assert (tmp_path / "7.fit").read_bytes() == FIT_BYTES


def test_download_fit_returns_false_on_empty_response(tmp_conn, tmp_path, monkeypatch):
    _seed_activity(tmp_conn, 99)
    monkeypatch.setattr(g.garth, "download", lambda path: None)

    assert g.download_fit(tmp_conn, 99, fit_dir=tmp_path) is False
    assert tmp_conn.execute(
        "SELECT 1 FROM activity_fit_files WHERE activity_id = 99"
    ).fetchone() is None


# --- backfill_fits ---------------------------------------------------------

def test_backfill_only_targets_unlinked(tmp_conn, tmp_path, monkeypatch):
    _seed_activity(tmp_conn, 1)
    _seed_activity(tmp_conn, 2)
    (tmp_path / "1.fit").write_bytes(FIT_BYTES)
    g.record_link(tmp_conn, 1, tmp_path / "1.fit")  # 1 already has a FIT

    pulled = []
    monkeypatch.setattr(
        g.garth, "download",
        lambda path: pulled.append(path) or _zip(("x.fit", FIT_BYTES)),
    )

    got = g.backfill_fits(tmp_conn, fit_dir=tmp_path)

    assert got == 1
    assert pulled == ["/download-service/files/activity/2"]


def test_backfill_respects_type_filter(tmp_conn, tmp_path, monkeypatch):
    _seed_activity(tmp_conn, 10, atype="running")
    _seed_activity(tmp_conn, 11, atype="cycling")
    monkeypatch.setattr(g.garth, "download", lambda path: _zip(("x.fit", FIT_BYTES)))

    got = g.backfill_fits(tmp_conn, fit_dir=tmp_path, activity_type="running")

    assert got == 1
    assert (tmp_path / "10.fit").exists()
    assert not (tmp_path / "11.fit").exists()