cloud-sync/tests/test_state.py

"""state.json read/write + divergence helpers."""

from __future__ import annotations

import json
from datetime import datetime, timedelta, timezone
from pathlib import Path

import pytest

from cloud_sync import scope as scopemod
from cloud_sync import state as statemod
from cloud_sync.sync import (
    _find_modified_in_scope,
    _format_dt,
    _matches_any,
    _parse_backup_summary,
    _parse_restic_time,
    _parse_snapshots,
)


# ---- state.read/write ----


def test_read_missing_returns_none(tmp_path: Path):
    assert statemod.read(tmp_path) is None


_IID = "01H7XJ4WB2KD5MNCYV8RQ6PTAZ"


def test_write_then_read_roundtrip(tmp_path: Path):
    dt = datetime(2026, 6, 5, 12, 34, 56, tzinfo=timezone.utc)
    statemod.write(
        tmp_path,
        statemod.State(
            instance_id=_IID,
            last_pulled_snapshot_id="abc123",
            last_pulled_at=dt,
        ),
    )
    got = statemod.read(tmp_path)
    assert got is not None
    assert got.instance_id == _IID
    assert got.last_pulled_snapshot_id == "abc123"
    assert got.last_pulled_at == dt


def test_write_sets_mode_600(tmp_path: Path):
    statemod.write(
        tmp_path,
        statemod.State(
            instance_id=_IID,
            last_pulled_snapshot_id="x",
            last_pulled_at=datetime.now(timezone.utc),
        ),
    )
    mode = statemod.state_path(tmp_path).stat().st_mode & 0o777
    assert mode == 0o600


def test_clear_idempotent(tmp_path: Path):
    statemod.clear(tmp_path)  # no-op when missing
    statemod.write(
        tmp_path,
        statemod.State(
            instance_id=_IID,
            last_pulled_snapshot_id="x",
            last_pulled_at=datetime.now(timezone.utc),
        ),
    )
    assert statemod.state_path(tmp_path).exists()
    statemod.clear(tmp_path)
    assert not statemod.state_path(tmp_path).exists()


def test_wrong_schema_returns_none(tmp_path: Path):
    p = tmp_path / ".cloud-sync" / "state.json"
    p.parent.mkdir(parents=True)
    p.write_text(json.dumps({
        "schema": 999, "instance_id": _IID,
        "last_pulled_snapshot_id": "x",
        "last_pulled_at": "2026-01-01T00:00:00Z",
    }))
    assert statemod.read(tmp_path) is None


def test_schema_v1_returns_none(tmp_path: Path):
    """Old schema-1 state.json (no instance_id) is treated as missing on
    read. Triggers a fresh sync flow after the schema bump migration."""
    p = tmp_path / ".cloud-sync" / "state.json"
    p.parent.mkdir(parents=True)
    p.write_text(json.dumps({
        "schema": 1,
        "last_pulled_snapshot_id": "x",
        "last_pulled_at": "2026-01-01T00:00:00Z",
    }))
    assert statemod.read(tmp_path) is None


def test_garbage_json_returns_none(tmp_path: Path):
    p = tmp_path / ".cloud-sync" / "state.json"
    p.parent.mkdir(parents=True)
    p.write_text("{not json")
    assert statemod.read(tmp_path) is None


# ---- _find_modified_in_scope ----


def _touch(p: Path, content: str = "x") -> None:
    p.parent.mkdir(parents=True, exist_ok=True)
    p.write_text(content)


def test_no_in_scope_changes_returns_empty(tmp_path: Path):
    _touch(tmp_path / "options.txt")
    # mtime is "now", `since` is the future
    future = datetime.now(timezone.utc) + timedelta(hours=1)
    hits = _find_modified_in_scope(tmp_path, scopemod.Scope(), future)
    assert hits == []


def test_in_scope_file_modified_after_since_detected(tmp_path: Path):
    _touch(tmp_path / "options.txt")
    past = datetime.now(timezone.utc) - timedelta(hours=1)
    hits = _find_modified_in_scope(tmp_path, scopemod.Scope(), past)
    assert any(rel.name == "options.txt" for rel, _ in hits)


def test_excluded_path_skipped(tmp_path: Path):
    # config/packwiz-installer.log matches exclude "**/*.log" AND "config/packwiz*"
    _touch(tmp_path / "config" / "packwiz-installer.log")
    past = datetime.now(timezone.utc) - timedelta(hours=1)
    hits = _find_modified_in_scope(tmp_path, scopemod.Scope(), past)
    assert hits == []


def test_only_excluded_changes_not_a_conflict(tmp_path: Path):
    _touch(tmp_path / ".cloud-sync" / "files-from.txt")  # excluded
    _touch(tmp_path / "config" / "packwiz-installer.log")  # excluded
    past = datetime.now(timezone.utc) - timedelta(hours=1)
    hits = _find_modified_in_scope(tmp_path, scopemod.Scope(), past)
    assert hits == []


def test_walks_subdirectories(tmp_path: Path):
    _touch(tmp_path / "config" / "fabric" / "custom.json")
    past = datetime.now(timezone.utc) - timedelta(hours=1)
    hits = _find_modified_in_scope(tmp_path, scopemod.Scope(), past)
    assert any("custom.json" in str(rel) for rel, _ in hits)


# ---- _matches_any ----


def test_dir_pattern_matches_anything_under(tmp_path: Path):
    assert _matches_any(Path(".cloud-sync/token"), [".cloud-sync/"])
    assert _matches_any(Path("a/b/cache/x"), ["**/cache/"])
    assert not _matches_any(Path("config/options.txt"), [".cloud-sync/"])


def test_glob_pattern(tmp_path: Path):
    assert _matches_any(Path("config/packwiz-installer.log"), ["config/packwiz*"])
    assert _matches_any(Path("foo/bar.log"), ["**/*.log"])


# ---- restic output parsing ----


def test_parse_snapshots_empty():
    assert _parse_snapshots("") == []
    assert _parse_snapshots("null") == []
    assert _parse_snapshots("[]") == []


def test_parse_snapshots_one():
    out = json.dumps([{"id": "abc", "time": "2026-06-05T12:00:00Z"}])
    parsed = _parse_snapshots(out)
    assert len(parsed) == 1
    assert parsed[0]["id"] == "abc"


def test_parse_restic_time_with_nanos():
    dt = _parse_restic_time("2026-06-04T18:33:21.123456789Z")
    assert dt.tzinfo is timezone.utc
    assert dt.year == 2026 and dt.day == 4 and dt.hour == 18


def test_parse_backup_summary_finds_snapshot_id():
    out = (
        '{"message_type":"status","percent_done":0.5}\n'
        '{"message_type":"status","percent_done":1.0}\n'
        '{"message_type":"summary","snapshot_id":"deadbeef","files_new":3}\n'
    )
    assert _parse_backup_summary(out) == "deadbeef"


def test_parse_backup_summary_missing_returns_none():
    assert _parse_backup_summary("") is None
    assert _parse_backup_summary("not json\nstill not json") is None


# ---- _format_dt ----


def test_format_dt_strips_leading_zero_on_hour():
    # 7:12 PM, not 07:12 PM. Day of month also no leading zero.
    dt = datetime(2021, 10, 21, 19, 12, tzinfo=timezone.utc)
    out = _format_dt(dt)
    # Output is local-time-converted, so don't pin the weekday/AM-PM exactly
    # in case CI runs UTC vs PT; just check the formatting shape.
    assert "October" in out or "Oct" not in out  # full month name
    assert ", 2021 at" in out
    assert " AM" in out or " PM" in out
    assert " 0:" not in out  # no leading-zero hour