# SPDX-License-Identifier: Apache-2.0
"""Where the tests find data that is NOT shipped with the bundle.

- The **shipped sample** is the synthetic eight-camera frame ``samples/synthetic_8cam.json`` (preset
  ``calib/synthetic_8cam.json``, stored CPU reference ``samples/synthetic_8cam.reference.json``; made by
  ``code/scripts/make_synthetic_sample.py``, Apache-2.0).
- The **PandaSet 019 frame 40 sample** (CC BY 4.0 + PandaSet Dataset Terms), its calibration preset ``pandaset_019``
  and its small goldens (feed sha256, decoded CPU reference, the ``/predict`` reference body) do not ship until the
  user approves PandaSet-derived samples (PLAN.md 6.3, research/DATASETS.md N5). They are found in
  ``$METEOR_PANDASET_DIR`` (a directory laid out like ``staging_samples_pandaset/``, i.e. like ``code/tt_meteor/``),
  else in ``code/tt_meteor/`` once swapped in, else in the git-ignored ``staging_samples_pandaset/`` of the bundle
  checkout. Inside the tt-models workspace a test that needs them FAILS when they are missing (a frozen gate must
  not vanish silently, :func:`need_pandaset`); elsewhere it skips.

The workspace root is ``$TT_MODELS_ROOT`` (default ``/home/ubuntu/experiments/tt-models``).
"""
from __future__ import annotations

import json
import os
from pathlib import Path
from typing import Any, Dict

PKG_DIR = Path(__file__).resolve().parents[1]
CODE_DIR = PKG_DIR.parent
BUNDLE_DIR = CODE_DIR.parent
WORKSPACE = Path(os.environ.get("TT_MODELS_ROOT", "/home/ubuntu/experiments/tt-models"))
GOLDENS_SMALL = PKG_DIR / "tests" / "goldens"
SAMPLES = PKG_DIR / "samples"
SHIPPED_NAME = "synthetic_8cam"
SHIPPED_SAMPLE = SAMPLES / f"{SHIPPED_NAME}.json"                    # the shipped sample
SHIPPED_REFERENCE = SAMPLES / f"{SHIPPED_NAME}.reference.json"
SHIPPED_INPUTS = GOLDENS_SMALL / f"{SHIPPED_NAME}_inputs.json"
STAGING_PANDASET = BUNDLE_DIR / "staging_samples_pandaset"         # git-ignored, never shipped (PLAN.md 6.3)
PANDASET_NAME = "pandaset_019_f40"


def _pandaset_root() -> Path:
    env = os.environ.get("METEOR_PANDASET_DIR")
    if env:
        return Path(env)
    if (SAMPLES / f"{PANDASET_NAME}.json").is_file():                 # swapped in after the user's OK
        return PKG_DIR
    return STAGING_PANDASET


_PS = _pandaset_root()
PANDASET_SAMPLES = _PS / "samples"
PANDASET_CALIB = _PS / "calib"
PANDASET_GOLDENS = _PS / "tests" / "goldens"
PANDASET_SAMPLE = PANDASET_SAMPLES / f"{PANDASET_NAME}.json"          # the request manifest (preset beside it)
PANDASET_SCENE = PANDASET_SAMPLES / PANDASET_NAME                    # METEOR demo-scene layout of the same frame
PANDASET_REFERENCE = PANDASET_SAMPLES / f"{PANDASET_NAME}.reference.json"


def need_pandaset() -> None:
    """Skip (outside the workspace) or fail (inside it) when the staged PandaSet sample is missing."""
    import pytest

    if PANDASET_SAMPLE.is_file() and (PANDASET_GOLDENS / f"{PANDASET_NAME}_inputs.json").is_file():
        return
    msg = (f"the PandaSet 019 frame 40 sample is not at {PANDASET_SAMPLES} (staging_samples_pandaset/ of the bundle "
           "checkout or $METEOR_PANDASET_DIR; it does not ship until the user approves PandaSet-derived samples)")
    if (WORKSPACE / "research" / "meteor").is_dir():
        pytest.fail(msg)
    pytest.skip(msg)


def pandaset_golden(kind: str) -> Dict[str, Any]:
    """``tests/goldens/pandaset_019_f40_<kind>.json`` of the staged sample (``inputs`` or ``reference``)."""
    need_pandaset()
    return json.loads((PANDASET_GOLDENS / f"{PANDASET_NAME}_{kind}.json").read_text())


def shipped_inputs() -> Dict[str, Any]:
    return json.loads(SHIPPED_INPUTS.read_text())