mdcaldav/tests/test_parser.py

"""Parser tests against synthetic note fixtures (SPEC 12, tests 1/2/9)."""

from __future__ import annotations

from mdcaldav.config import Config
from mdcaldav.model import Status
from mdcaldav.parser import parse


def titles(tasks):
    return [t.title for t in tasks]


def by_title(tasks, title):
    return next(t for t in tasks if t.title == title)


def test_flat_daily_note(vault, tasks_of):
    tasks = tasks_of(vault, "daily/2026-01-07.md")
    assert titles(tasks) == [
        "Book the soil test",
        "Call the arborist",
        "Tidy the potting bench",
        "Water the seedlings",
    ]
    assert [t.priority for t in tasks] == [1, 5, 9, None]
    assert all(t.status is Status.NEEDS_ACTION for t in tasks)


def test_nesting_and_parentage(vault, tasks_of):
    tasks = tasks_of(vault, "daily/2026-01-06.md")
    parent = by_title(tasks, "Replace the cracked pane")
    assert parent.parent_uid is None
    assert parent.depth == 0

    tasks = tasks_of(vault, "projects/irrigation.md")
    trench = by_title(tasks, "Trench the main line")
    rent = by_title(tasks, "Rent the trencher")
    assert rent.parent_uid == trench.uid
    assert rent.uid in trench.children
    assert (trench.depth, rent.depth) == (0, 1)


def test_deep_nesting_five_levels(vault, tasks_of):
    tasks = tasks_of(vault, "daily/2026-01-05.md")
    deep = by_title(tasks, "Email the council office")
    assert deep.depth == 4

    lookup = {t.uid: t for t in tasks}
    chain, node = [], deep
    while node.parent_uid:
        node = lookup[node.parent_uid]
        chain.append(node.title)
    assert chain == [
        "Check the fence bylaw",
        "Decide on trellis height",
        "Draft the bed layout",
        "Plan the seed order",
    ]


def test_sub_bullet_becomes_description(vault, tasks_of):
    """SPEC 2.3: non-checkbox bullet with no checkbox descendants."""
    tasks = tasks_of(vault, "daily/2026-01-05.md")
    assert by_title(tasks, "Order the cover crop").description == (
        "Arrived on the 12th, invoice is in the binder"
    )
    assert by_title(tasks, "Sharpen the pruners").description is None


def test_multiple_description_bullets_join(vault, tasks_of):
    tasks = tasks_of(vault, "daily/2026-01-06.md")
    assert by_title(tasks, "Replace the cracked pane").description == (
        "Used the offcut from the shed\nSilicone needs a day to cure"
    )


def test_group_node_is_not_a_task(vault, tasks_of):
    """SPEC 2.3: a plain bullet parenting checkboxes contributes context only."""
    tasks = tasks_of(vault, "daily/2026-01-06.md")
    label = "set up the cold frame"
    assert label not in titles(tasks)

    cut = by_title(tasks, "Cut the polycarbonate")
    assert cut.group_path == (label,)
    assert label in cut.categories
    assert cut.parent_uid is None  # re-parented to nearest task ancestor: none
    assert cut.depth == 0


def test_note_bullets_produce_nothing(vault, tasks_of):
    """SPEC 2.3: plain bullets with no task ancestor are dropped entirely."""
    tasks = tasks_of(vault, "daily/2026-01-05.md")
    assert "Buy fewer squash varieties" not in titles(tasks)
    assert not any(
        t.description and "squash" in t.description for t in tasks
    )


def test_tasks_collected_under_any_heading(vault, tasks_of):
    tasks = tasks_of(vault, "daily/2026-01-06.md")
    # heading_path is the full stack, so the H1 note title leads.
    assert by_title(tasks, "Reseal the base").heading_path == (
        "Tuesday, 6 January 2026",
        "Greenhouse repairs",
    )


def test_file_with_no_tasks(vault, tasks_of):
    assert tasks_of(vault, "reference/glossary.md") == []


def test_code_blocks_are_not_tasks(vault, tasks_of):
    """SPEC 2.6: fenced and indented code must not yield tasks."""
    tasks = tasks_of(vault, "projects/irrigation.md")
    assert titles(tasks) == [
        "Trench the main line",
        "Rent the trencher",
        "Price the manifold",
    ]


def test_tilde_fence(cfg: Config):
    src = b"- [ ] real\n\n~~~\n- [ ] fake\n~~~\n"
    assert titles(parse(src, "f.md", cfg)) == ["real"]


def test_states(cfg: Config):
    tasks = parse(b"- [ ] a\n- [x] b\n- [X] c\n- [/] d\n- [-] e\n- [?] f\n", "s.md", cfg)
    assert [t.status for t in tasks] == [
        Status.NEEDS_ACTION,
        Status.COMPLETED,
        Status.COMPLETED,
        Status.IN_PROCESS,
        Status.CANCELLED,
        Status.NEEDS_ACTION,
    ]
    assert tasks[-1].raw_marker == "?"  # unknown marker preserved verbatim


def test_frontmatter_is_not_content(cfg: Config):
    src = b"---\ndate: 2026-01-05\ntags: [daily]\n---\n\n- [ ] only\n"
    assert titles(parse(src, "f.md", cfg)) == ["only"]


def test_spans_locate_exact_bytes(cfg: Config):
    src = b"- [ ] [#A] Ship it\n"
    (task,) = parse(src, "s.md", cfg)
    assert src[slice(*task.source.marker_span)] == b" "
    assert src[slice(*task.source.title_span)] == b"Ship it"
    assert src[slice(*task.source.prio_span)] == b"[#A]"


def test_spans_are_byte_offsets_with_unicode(cfg: Config):
    """Byte spans must stay correct when the line contains multi-byte characters."""
    src = "- [x] café → naïve\n".encode()
    (task,) = parse(src, "u.md", cfg)
    assert src[slice(*task.source.marker_span)] == b"x"
    assert src[slice(*task.source.title_span)].decode() == "café → naïve"


def test_crlf_line_endings(cfg: Config):
    (task,) = parse(b"- [ ] windows\r\n", "w.md", cfg)
    assert task.title == "windows"
    assert task.source.line_span == (0, 13)  # excludes the \r\n


def test_bullet_variants(cfg: Config):
    tasks = parse(b"- [ ] dash\n* [ ] star\n+ [ ] plus\n", "b.md", cfg)
    assert titles(tasks) == ["dash", "star", "plus"]
    assert [t.source.bullet for t in tasks] == ["-", "*", "+"]


def test_tags_become_categories(cfg: Config):
    (task,) = parse(b"- [ ] Fix #1923 for #garden\n", "t.md", cfg)
    assert "garden" in task.categories
    assert "1923" not in task.categories  # bare numbers are not tags
    assert task.title == "Fix #1923 for #garden"  # tags stay in the title


def test_sibling_index_disambiguates_duplicates(cfg: Config):
    tasks = parse(b"- [ ] same\n- [ ] same\n- [ ] same\n", "d.md", cfg)
    assert [t.sibling_index for t in tasks] == [0, 1, 2]


def test_due_syntax_opt_in(cfg: Config):
    src = b"- [ ] Prune the apple tree @due(2026-03-01)\n"
    assert parse(src, "n.md", cfg)[0].due is None  # disabled by default

    cfg.tasks.due_syntax = ["@due(%Y-%m-%d)"]
    task = parse(src, "n.md", cfg)[0]
    assert task.due is not None and task.due.isoformat() == "2026-03-01"
    assert task.title == "Prune the apple tree"