"""Notes adapter tests: markdown and plain-text parsing, sample ingestion."""

from __future__ import annotations

from pathlib import Path

from helpers import blob

from pmp.adapters.notes import NotesAdapter


def _make_notes_dir(tmp_path: Path) -> Path:
    notes = tmp_path / "notes"
    notes.mkdir()
    (notes / "knee.md").write_text(
        "# Knee check-in\n\nKnee pain came back after the Sunday long run. "
        "Physio suggested reducing mileage for two weeks.\n",
        encoding="utf-8",
    )
    (notes / "shopping.txt").write_text(
        "rye flour\nolive oil\nespresso beans\n",
        encoding="utf-8",
    )
    return notes


def test_ingest_single_markdown_file(tmp_path: Path):
    notes = _make_notes_dir(tmp_path)
    items = list(NotesAdapter().ingest(notes / "knee.md"))
    assert len(items) == 1
    rendered = blob(items[0].content)
    assert "Knee" in rendered
    assert "Physio" in rendered


def test_ingest_directory_picks_up_md_and_txt(tmp_path: Path):
    notes = _make_notes_dir(tmp_path)
    items = list(NotesAdapter().ingest(notes))
    assert len(items) == 2
    combined = blob([item.content for item in items])
    assert "Knee pain" in combined
    assert "espresso beans" in combined


def test_non_text_files_are_skipped(tmp_path: Path):
    notes = _make_notes_dir(tmp_path)
    # A binary blob that is not valid UTF-8 and is not a note format.
    (notes / "IMG_0001.jpg").write_bytes(b"\xff\xd8\xff\xe0" + bytes(range(256)))
    items = list(NotesAdapter().ingest(notes))
    assert len(items) == 2, "binary non-note files must not become note evidence"


def test_items_carry_provenance_fields(tmp_path: Path):
    notes = _make_notes_dir(tmp_path)
    for item in NotesAdapter().ingest(notes):
        assert item.content
        assert item.source_ref
        assert "note" in str(item.source_type).lower()


def test_ingest_is_deterministic(tmp_path: Path):
    notes = _make_notes_dir(tmp_path)
    first = sorted(blob(item.content) for item in NotesAdapter().ingest(notes))
    second = sorted(blob(item.content) for item in NotesAdapter().ingest(notes))
    assert first == second


def test_sample_notes_ingest(samples_dir: Path):
    items = list(NotesAdapter().ingest(samples_dir / "notes"))
    # Five sample note files ship with the repo (four markdown, one txt).
    assert len(items) >= 5
    combined = blob([item.content for item in items]).lower()
    assert "sourdough" in combined
    assert "berlin" in combined
    # Each evidence item is distinct.
    rendered = [blob(item.content) for item in items]
    assert len(set(rendered)) == len(rendered)