Challenge 2: Test the complete Behavior — Possible Solution
====================================================================

from tasks.models import Task
from tasks.storage import TaskStorage

def test_completing_a_task_persists(tmp_path):
    storage = TaskStorage(tmp_path / "tasks.json")
    tasks = [
        Task(id=1, title="Buy milk"),
        Task(id=2, title="Walk the dog"),
    ]
    storage.save(tasks)

    tasks[0].done = True
    storage.save(tasks)

    reloaded = storage.load()
    completed = [t for t in reloaded if t.done]

    assert len(completed) == 1
    assert completed[0].id == 1

WHY THIS WORKS AS AN ANSWER
------------------------------
TaskStorage(tmp_path / "tasks.json") reuses this chapter's own
tmp_path fixture usage exactly — pytest supplies a fresh, isolated
temporary directory automatically, so this test never touches a real
~/.tasks.json file, matching the chapter's own test_save_and_load_
round_trip example.

Saving two tasks first, then mutating tasks[0].done = True directly
and calling storage.save(tasks) a second time, mirrors exactly what
the real "complete" command does in the chapter's own cli.py: find the
matching task in the in-memory list, flip its done attribute, then
persist the whole list again — this test exercises that same
sequence, just without going through argparse.

storage.load() reuses the chapter's own load() method to read the
data back fresh from disk, confirming the change was actually
PERSISTED (written to and correctly re-read from the file) rather
than only true in memory before the second save() call.

[t for t in reloaded if t.done] reuses the list comprehension pattern
from Challenge 1 to filter down to only completed tasks. Asserting
len(completed) == 1 confirms exactly one task (not zero, not both) is
marked done, and completed[0].id == 1 confirms it's specifically the
RIGHT task that got completed, not just that some arbitrary task did.