tests/test_ids.py

"""Tests for `fpgm.data.ids`'s uuid -> DROID raw bucket path derivation.

Pure-stdlib logic, deliberately: no numpy/h5py/requests import happens on
this path, so these tests run in bare environments (e.g. system python3.10,
before the project's conda env exists).
"""

from __future__ import annotations

import pytest

from fpgm.data.ids import (
    DROID_RAW_BUCKET_ROOT,
    episode_base_url,
    episode_date,
    episode_dirname,
    metadata_filename,
)
from fpgm.types import EpisodeId

# Three uuids verified against PointWorld's own `real/droid_paths.txt`.
DOUBLE_DIGIT_DAY = ("AUTOLab+0d4edc83+2023-10-21-19h-07m-04s", "Sat_Oct_21_19:07:04_2023")
SINGLE_DIGIT_DAY = ("SomeLab+abcdef01+2023-10-06-19h-50m-14s", "Fri_Oct__6_19:50:14_2023")
ANOTHER_DOUBLE_DIGIT_DAY = ("IPRL+12345678+2023-01-15-09h-03m-22s", "Sun_Jan_15_09:03:22_2023")


@pytest.mark.parametrize(
    "uuid, expected_dirname",
    [DOUBLE_DIGIT_DAY, SINGLE_DIGIT_DAY, ANOTHER_DOUBLE_DIGIT_DAY],
)
def test_episode_dirname_matches_verified_examples(uuid: str, expected_dirname: str) -> None:
    assert episode_dirname(EpisodeId(uuid)) == expected_dirname


def test_single_digit_day_has_double_underscore() -> None:
    """The `%e` space-pad -> `_` substitution trap: single-digit days must NOT
    collapse to a single underscore like a naive zero-padded format would.
    """
    dirname = episode_dirname(EpisodeId(SINGLE_DIGIT_DAY[0]))
    assert "__6" in dirname
    assert dirname == "Fri_Oct__6_19:50:14_2023"


def test_double_digit_day_has_single_underscore() -> None:
    dirname = episode_dirname(EpisodeId(DOUBLE_DIGIT_DAY[0]))
    assert "_21_" in dirname
    assert "__" not in dirname


@pytest.mark.parametrize(
    "uuid, expected_dirname",
    [DOUBLE_DIGIT_DAY, SINGLE_DIGIT_DAY, ANOTHER_DOUBLE_DIGIT_DAY],
)
def test_episode_id_uuid_roundtrip(uuid: str, expected_dirname: str) -> None:
    """uuid -> (lab, user_id, timestamp) -> dirname round-trips to the verified string."""
    episode_id = EpisodeId(uuid)
    lab, user_id, ts_str = uuid.split("+")
    assert episode_id.lab == lab
    assert episode_id.user_id == user_id
    assert episode_id.timestamp.strftime("%Y-%m-%d-%Hh-%Mm-%Ss") == ts_str
    assert episode_dirname(episode_id) == expected_dirname


def test_episode_date_is_iso_and_independent_of_dirname_format() -> None:
    episode_id = EpisodeId(DOUBLE_DIGIT_DAY[0])
    assert episode_date(episode_id) == "2023-10-21"


def test_metadata_filename_percent_encodes_plus() -> None:
    episode_id = EpisodeId(DOUBLE_DIGIT_DAY[0])
    filename = metadata_filename(episode_id)
    assert filename.startswith("metadata_")
    assert filename.endswith(".json")
    assert "+" not in filename
    assert "%2B" in filename


def test_episode_base_url_percent_encodes_colon_and_has_expected_shape() -> None:
    episode_id = EpisodeId(SINGLE_DIGIT_DAY[0])
    url = episode_base_url(episode_id, "success")
    assert url.startswith(DROID_RAW_BUCKET_ROOT)
    assert "/SomeLab/success/2023-10-06/" in url
    # The scheme's own "://" is not part of the path, so only the path is checked.
    path = url[len(DROID_RAW_BUCKET_ROOT) :]
    assert ":" not in path  # the timestamp's colons must be percent-encoded
    assert "%3A" in path
    assert url.endswith("/")


def test_episode_base_url_rejects_unknown_outcome() -> None:
    episode_id = EpisodeId(DOUBLE_DIGIT_DAY[0])
    with pytest.raises(ValueError):
        episode_base_url(episode_id, "maybe")