tests/dataset/test_config_loading.py

from __future__ import annotations

from pathlib import Path
from typing import Any

import pytest

from kneiff.config import StorageConfig
from kneiff.datasets.export.config import load_export_config
from kneiff.datasets.export.types import (
    ImageResizeConfig,
    TrainingExportResult,
)
from kneiff.datasets.export.writer import (
    export_training_dataset as _export_training_dataset,
)
from kneiff.project import ProjectContextError
from tests.captions._helpers import load_rook_vocabulary
from tests.dataset._export_helpers import (
    _manifest_row,
    _write_config,
    _write_image,
    _write_manifest,
)


def export_training_dataset(*args: Any, **kwargs: Any) -> TrainingExportResult:
    """Run config-loading export checks with the Rook fixture vocabulary."""
    kwargs.setdefault("vocabulary", load_rook_vocabulary())
    return _export_training_dataset(*args, **kwargs)


def test_export_config_workers_ignore_process_and_project_dotenv(
    tmp_path: Path,
    monkeypatch: pytest.MonkeyPatch,
) -> None:
    source_root = tmp_path / "SOURCE"
    export_root = tmp_path / "HF" / "ready"
    source_root.mkdir()
    (source_root / ".env.shared").write_text("KNF_WORKERS=4\n", encoding="utf-8")
    monkeypatch.setenv("KNF_WORKERS", "3")
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    _write_config(config_path, source_root, export_root)

    config = load_export_config(config_path)

    assert config.workers == 8


def test_export_workers_allow_explicit_orchestration_override(
    tmp_path: Path,
    monkeypatch: pytest.MonkeyPatch,
) -> None:
    source_root = tmp_path / "SOURCE"
    export_root = tmp_path / "HF" / "ready"
    source_root.mkdir()
    monkeypatch.setenv("KNF_WORKERS", "2")
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    _write_config(config_path, source_root, export_root)

    config = load_export_config(config_path, workers_override=4)

    assert config.workers == 4


def test_export_workers_reject_invalid_override(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    export_root = tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    _write_config(config_path, source_root, export_root)

    with pytest.raises(ValueError, match="KNF_WORKERS"):
        load_export_config(config_path, workers_override=0)


@pytest.mark.parametrize(
    "config_path",
    [
        Path("ready.knf.yaml"),
        Path("configs/ready.yaml"),
        Path("configs/nested/ready.knf.yaml"),
        Path("configs/.knf.yaml"),
        Path("configs/..knf.yaml"),
        Path("configs/ ready.knf.yaml"),
    ],
)
def test_export_config_requires_direct_project_config_path(
    tmp_path: Path,
    config_path: Path,
) -> None:
    invalid_path = tmp_path / config_path
    invalid_path.parent.mkdir(parents=True, exist_ok=True)
    invalid_path.write_text("mappings: {fullbody: [0-FULLBODY]}\n", encoding="utf-8")

    with pytest.raises(ValueError, match="config|Config"):
        load_export_config(invalid_path)


def test_storage_config_owns_project_worker_setting(
    tmp_path: Path,
    monkeypatch: pytest.MonkeyPatch,
) -> None:
    monkeypatch.setenv("KNF_STORAGE", str(tmp_path))
    monkeypatch.setenv("KNF_WORKERS", "4")

    config = StorageConfig()

    assert config.root == tmp_path
    assert config.workers == 4


def test_storage_config_rejects_non_positive_worker_setting(
    tmp_path: Path,
    monkeypatch: pytest.MonkeyPatch,
) -> None:
    monkeypatch.setenv("KNF_STORAGE", str(tmp_path))
    monkeypatch.setenv("KNF_WORKERS", "0")

    with pytest.raises(ValueError, match="KNF_WORKERS"):
        StorageConfig()


def test_export_config_defaults_mirrored_extra_to_false(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
""",
        encoding="utf-8",
    )

    config = load_export_config(config_path)

    assert config.augmentations.mirrored_extra is False
    assert config.augmentations.mirrored_transform == "augmented"
    assert config.augmentations.seed == 12345


def test_export_config_defaults_missing_mirrored_extra_to_false(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
augmentations:
  seed: 12345
""",
        encoding="utf-8",
    )

    config = load_export_config(config_path)

    assert config.augmentations.mirrored_extra is False
    assert config.augmentations.mirrored_transform == "augmented"


def test_export_config_loads_flip_only_mirrored_transform(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
augmentations:
  mirrored_extra: true
  mirrored_transform: flip_only
  seed: 12345
""",
        encoding="utf-8",
    )

    config = load_export_config(config_path)

    assert config.augmentations.mirrored_extra is True
    assert config.augmentations.mirrored_transform == "flip_only"


def test_export_config_rejects_string_mirrored_extra(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
augmentations:
  mirrored_extra: "false"
""",
        encoding="utf-8",
    )

    with pytest.raises(
        ValueError, match="augmentations.mirrored_extra must be true or false"
    ):
        load_export_config(config_path)


def test_export_config_rejects_coerced_augmentation_seed(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
augmentations:
  seed: "12345"
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="augmentations.seed must be an integer"):
        load_export_config(config_path)


@pytest.mark.parametrize(
    ("extra_config", "error"),
    [
        ("mappingz: {}", "Dataset export config contains unknown fields: mappingz"),
        ("augmentations:\n  sed: 1", "augmentations contains unknown fields: sed"),
        (
            "caption:\n  add_domain_to_tags: true",
            "caption contains unknown fields: add_domain_to_tags",
        ),
        (
            "image_resize:\n  min_pixel_area: 512\n  max_pixels: 1024",
            "image_resize contains unknown fields: max_pixels",
        ),
        (
            "publishing:\n  hugginface: {}",
            "publishing contains unknown fields: hugginface",
        ),
        (
            "publishing:\n  huggingface:\n    reop_id: account/data",
            "publishing.huggingface contains unknown fields: reop_id",
        ),
        (
            "caption_outputs_overrides:\n  fullbody:\n    formtas: [tag]",
            "caption_outputs_overrides.fullbody contains unknown fields: formtas",
        ),
    ],
)
def test_export_config_rejects_unknown_fields(
    tmp_path: Path,
    extra_config: str,
    error: str,
) -> None:
    source_root = tmp_path / "SOURCE"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        "mappings:\n  fullbody:\n    - 0-FULLBODY\n" + extra_config + "\n",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match=error):
        load_export_config(config_path)


def test_export_config_rejects_invalid_mirrored_transform(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
augmentations:
  mirrored_transform: sideways
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="augmentations.mirrored_transform"):
        load_export_config(config_path)


def test_export_config_derives_fixed_project_paths(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    export_root = tmp_path / "HF" / "ready"
    _write_image(source_root / "0-FULLBODY" / "scene.png")
    _write_manifest(
        source_root,
        [_manifest_row("0-FULLBODY/scene.png")],
        workbook_name="MANIFEST.knf.xlsx",
    )
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
augmentations:
  mirrored_extra: false
  seed: 12345
""",
        encoding="utf-8",
    )

    config = load_export_config(config_path)
    result = export_training_dataset(config, dry_run=True)

    assert config.source_root == source_root
    assert config.manifest_path == tmp_path / "MANIFEST.knf.xlsx"
    assert config.export_root == export_root
    assert result.source_image_count == 1


def test_export_config_does_not_fallback_to_lowercase_knf_manifest(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    _write_image(source_root / "0-FULLBODY" / "scene.png")
    _write_manifest(
        source_root,
        [_manifest_row("0-FULLBODY/scene.png")],
        workbook_name="MANIFEST.knf.xlsx",
    )
    (tmp_path / "MANIFEST.knf.xlsx").rename(tmp_path / "manifest.knf.xlsx")
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
augmentations:
  mirrored_extra: false
  seed: 12345
""",
        encoding="utf-8",
    )

    config = load_export_config(config_path)

    assert config.manifest_path == tmp_path / "MANIFEST.knf.xlsx"
    with pytest.raises(ProjectContextError) as exc_info:
        export_training_dataset(config, dry_run=True)
    assert "MANIFEST.knf.xlsx" in str(exc_info.value)


def test_export_config_rejects_legacy_densities_key(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
densities: [short]
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="densities is no longer supported"):
        load_export_config(config_path)


def test_export_config_defaults_sfw_subset_export_to_false(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    export_root = tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    _write_config(config_path, source_root, export_root)

    config = load_export_config(config_path)

    assert config.export_sfw_subset is False


def test_export_config_defaults_image_resize_to_none(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    export_root = tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    _write_config(config_path, source_root, export_root)

    config = load_export_config(config_path)

    assert config.image_resize is None


def test_export_config_loads_image_resize_square_edges(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
image_resize:
  min_pixel_area: 768
  max_pixel_area: 1536
""",
        encoding="utf-8",
    )

    config = load_export_config(config_path)

    assert config.image_resize == ImageResizeConfig(
        min_pixel_area=768,
        max_pixel_area=1536,
    )


def test_export_config_rejects_invalid_image_resize_type(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
image_resize: yes please
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="image_resize"):
        load_export_config(config_path)


def test_export_config_rejects_empty_image_resize(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
image_resize: {}
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="min_pixel_area or max_pixel_area"):
        load_export_config(config_path)


def test_export_config_rejects_non_positive_image_resize_value(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
image_resize:
  max_pixel_area: 0
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="image_resize.max_pixel_area"):
        load_export_config(config_path)


def test_export_config_rejects_reversed_image_resize_bounds(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
image_resize:
  min_pixel_area: 1536
  max_pixel_area: 768
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="min_pixel_area"):
        load_export_config(config_path)


def test_export_config_rejects_invalid_sfw_subset_toggle(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
export_sfw_subset: yes please
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="export_sfw_subset"):
        load_export_config(config_path)


def test_export_config_rejects_generated_sfw_subset_name_clash(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  sfw:
    - "0-FULLBODY"
export_sfw_subset: true
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="mappings already defines an sfw subset"):
        load_export_config(config_path)


def test_export_config_rejects_old_source_to_subset_mappings(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  "0-FULLBODY": "fullbody"
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="subset names as keys"):
        load_export_config(config_path)


def test_export_config_rejects_invalid_subject_sex(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption:
  subject_sex: unknown
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="caption.subject_sex"):
        load_export_config(config_path)


def test_export_config_rejects_removed_caption_domain_toggle(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption:
  add_domains_to_tags: true
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="caption contains unknown fields"):
        load_export_config(config_path)


@pytest.mark.parametrize("format_name", ["tag", "prose"])
def test_export_config_rejects_removed_caption_format_aliases(
    tmp_path: Path,
    format_name: str,
) -> None:
    source_root = tmp_path / "SOURCE"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        f"""
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  formats: [{format_name}]
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="must contain only"):
        load_export_config(config_path)


def test_export_config_rejects_unknown_caption_output_subset(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs_overrides:
  missing:
    mode: hybrid_txt
    formats: [tags]
""",
        encoding="utf-8",
    )

    with pytest.raises(
        ValueError,
        match="caption_outputs_overrides references unknown subset: missing",
    ):
        load_export_config(config_path)


def test_export_config_rejects_invalid_caption_output_mode(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  mode: yaml_txt
  formats: [tags]
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="caption_outputs.mode"):
        load_export_config(config_path)


def test_export_config_rejects_invalid_caption_output_format(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  mode: hybrid_txt
  formats: [tags, yaml]
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="caption_outputs.formats"):
        load_export_config(config_path)


def test_export_config_rejects_chroma_mixed_in_hybrid_caption(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  mode: hybrid_txt
  formats: [tags, chroma]
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="cannot combine chroma"):
        load_export_config(config_path)


def test_export_config_loads_global_caption_output_default(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  mode: hybrid_txt
  formats: [chroma]
  tag_scope: all
""",
        encoding="utf-8",
    )

    config = load_export_config(config_path)

    assert config.caption_output_default.mode == "hybrid_txt"
    assert config.caption_output_default.formats == ("chroma",)
    assert config.caption_output_default.tag_scope == "all"
    assert config.caption_outputs == {}


@pytest.mark.parametrize(
    "format_name",
    ["flux2", "z-image", "Z-Image", "z_image", "zimage"],
)
def test_export_config_rejects_removed_nlg_caption_aliases(
    tmp_path: Path,
    format_name: str,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        f"""
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  formats: [{format_name}]
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="must contain only"):
        load_export_config(config_path)


def test_export_config_caption_output_override_inherits_global_formats(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  mode: hybrid_txt
  formats: [chroma]
  tag_scope: all
caption_outputs_overrides:
  fullbody:
    mode: separate_txt
""",
        encoding="utf-8",
    )

    config = load_export_config(config_path)

    assert config.caption_outputs["fullbody"].mode == "separate_txt"
    assert config.caption_outputs["fullbody"].formats == ("chroma",)
    assert config.caption_outputs["fullbody"].tag_scope == "all"


def test_export_config_caption_output_override_inherits_global_mode(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  mode: separate_txt
  formats: [tags, natural]
  tag_scope: all
caption_outputs_overrides:
  fullbody:
    formats: [json]
""",
        encoding="utf-8",
    )

    config = load_export_config(config_path)

    assert config.caption_outputs["fullbody"].mode == "separate_txt"
    assert config.caption_outputs["fullbody"].formats == ("json",)
    assert config.caption_outputs["fullbody"].tag_scope == "all"


def test_export_config_caption_output_override_can_change_tag_scope(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  mode: hybrid_txt
  formats: [chroma]
caption_outputs_overrides:
  fullbody:
    tag_scope: all
""",
        encoding="utf-8",
    )

    config = load_export_config(config_path)

    assert config.caption_outputs["fullbody"].mode == "hybrid_txt"
    assert config.caption_outputs["fullbody"].formats == ("chroma",)
    assert config.caption_outputs["fullbody"].tag_scope == "all"


def test_export_config_rejects_invalid_global_caption_tag_scope(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  mode: hybrid_txt
  formats: [chroma]
  tag_scope: noisy
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="caption_outputs.tag_scope"):
        load_export_config(config_path)


def test_export_config_rejects_invalid_override_caption_tag_scope(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs_overrides:
  fullbody:
    tag_scope: noisy
""",
        encoding="utf-8",
    )

    with pytest.raises(
        ValueError,
        match="caption_outputs_overrides.fullbody.tag_scope",
    ):
        load_export_config(config_path)


def test_export_config_rejects_unknown_caption_output_override_subset(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  mode: hybrid_txt
  formats: [tags]
caption_outputs_overrides:
  missing:
    formats: [natural]
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="caption_outputs_overrides.*missing"):
        load_export_config(config_path)


def test_export_config_rejects_non_string_caption_output_override_subset(
    tmp_path: Path,
) -> None:
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True)
    config_path.write_text(
        """
mappings:
  fullbody: [0-FULLBODY]
caption_outputs_overrides:
  1:
    formats: [tags]
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="non-empty exact string"):
        load_export_config(config_path)


def test_export_config_accepts_sfw_caption_override_when_generated_subset_enabled(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
export_sfw_subset: true
caption_outputs_overrides:
  sfw:
    formats: [json]
""",
        encoding="utf-8",
    )

    config = load_export_config(config_path)

    assert config.caption_outputs["sfw"].formats == ("json",)


def test_export_config_rejects_sfw_caption_override_without_sfw_subset(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs_overrides:
  sfw:
    formats: [json]
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="caption_outputs_overrides.*sfw"):
        load_export_config(config_path)


def test_export_config_rejects_invalid_global_caption_output_mode(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  mode: yaml_txt
  formats: [tags]
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="caption_outputs.mode"):
        load_export_config(config_path)


def test_export_config_rejects_invalid_override_caption_output_format(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  mode: hybrid_txt
  formats: [tags]
caption_outputs_overrides:
  fullbody:
    formats: [yaml]
""",
        encoding="utf-8",
    )

    with pytest.raises(
        ValueError,
        match="caption_outputs_overrides.fullbody.formats",
    ):
        load_export_config(config_path)


def test_export_config_rejects_subset_entries_in_global_caption_outputs(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  mode: hybrid_txt
  formats: [chroma]
  fullbody:
    formats: [tags]
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="caption_outputs_overrides"):
        load_export_config(config_path)


def test_export_config_rejects_chroma_mixed_in_global_hybrid_caption(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  mode: hybrid_txt
  formats: [tags, chroma]
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="caption_outputs.formats"):
        load_export_config(config_path)


def test_export_config_rejects_flux2_mixed_in_global_hybrid_caption(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  mode: hybrid_txt
  formats: [tags, nlg]
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="caption_outputs.formats"):
        load_export_config(config_path)


def test_export_config_rejects_chroma_mixed_in_override_hybrid_caption(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  mode: separate_txt
  formats: [tags, chroma]
caption_outputs_overrides:
  fullbody:
    mode: hybrid_txt
""",
        encoding="utf-8",
    )

    with pytest.raises(
        ValueError,
        match="caption_outputs_overrides.fullbody.formats",
    ):
        load_export_config(config_path)


def test_export_config_accepts_flux2_with_other_separate_outputs(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
caption_outputs:
  mode: separate_txt
  formats: [tags, natural, json, chroma, nlg]
""",
        encoding="utf-8",
    )

    config = load_export_config(config_path)

    assert config.caption_output_default.mode == "separate_txt"
    assert config.caption_output_default.formats == (
        "tags",
        "natural",
        "json",
        "chroma",
        "nlg",
    )


def test_export_config_rejects_nested_subset(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    _write_image(source_root / "0-FULLBODY" / "scene.png")
    _write_manifest(source_root, [_manifest_row("0-FULLBODY/scene.png")])
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  "nested/fullbody":
    - "0-FULLBODY"
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="mapping subset"):
        load_export_config(config_path)


@pytest.mark.parametrize("subset_key", ["1", "' fullbody '"])
def test_export_config_rejects_coerced_mapping_subset_names(
    tmp_path: Path,
    subset_key: str,
) -> None:
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True)
    config_path.write_text(
        f"mappings:\n  {subset_key}: [0-FULLBODY]\n",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="non-empty exact string"):
        load_export_config(config_path)


def test_export_config_rejects_duplicate_normalized_mapping_sources(
    tmp_path: Path,
) -> None:
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - identity\\closeups
    - identity/closeups
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="repeats source directory"):
        load_export_config(config_path)


@pytest.mark.parametrize(
    "source_dir",
    [
        "/absolute/source",
        r"C:\absolute\source",
        r"\\server\share\source",
        "../outside",
        "nested/../outside",
        "./nested",
        "nested/./child",
    ],
)
def test_export_config_rejects_unsafe_mapping_source_directories(
    tmp_path: Path,
    source_dir: str,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        f"""
mappings:
  fullbody:
    - '{source_dir}'
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="relative path|path components"):
        load_export_config(config_path)


def test_export_config_normalizes_nested_windows_mapping_source_directory(
    tmp_path: Path,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - 'identity\\closeups'
""",
        encoding="utf-8",
    )

    config = load_export_config(config_path)

    assert config.mappings == {"fullbody": ("identity/closeups",)}


def test_export_config_rejects_backslash_nested_subset(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    _write_image(source_root / "0-FULLBODY" / "scene.png")
    _write_manifest(source_root, [_manifest_row("0-FULLBODY/scene.png")])
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  'nested\\fullbody':
    - "0-FULLBODY"
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="mapping subset"):
        load_export_config(config_path)


@pytest.mark.parametrize("subset_name", ["_TRAINING", "training"])
def test_export_config_accepts_training_named_subset(
    tmp_path: Path,
    subset_name: str,
) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    _write_image(source_root / "0-FULLBODY" / "scene.png")
    _write_manifest(source_root, [_manifest_row("0-FULLBODY/scene.png")])
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        f"""
mappings:
  {subset_name}:
    - "0-FULLBODY"
""",
        encoding="utf-8",
    )

    config = load_export_config(config_path)

    assert config.mappings == {subset_name: ("0-FULLBODY",)}


def test_export_config_loads_huggingface_publish_settings(tmp_path: Path) -> None:
    source_root = tmp_path / "SOURCE"
    tmp_path / "HF" / "ready"
    source_root.mkdir()
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        """
mappings:
  fullbody:
    - "0-FULLBODY"
publishing:
  huggingface:
    enabled: false
    repo_id: example/demo-dataset
    pretty_name: Demo Dataset
    version: v5.1
    optimized_for_model: Chroma1-HD
    license: creativeml-openrail-m
    tags: [chroma, lora]
    provenance: Local source set.
    adult_content: true
    notes: Needs review.
""",
        encoding="utf-8",
    )

    config = load_export_config(config_path)

    assert config.huggingface.enabled is False
    assert config.huggingface.repo_id == "example/demo-dataset"
    assert config.huggingface.pretty_name == "Demo Dataset"
    assert config.huggingface.version == "v5.1"
    assert config.huggingface.optimized_for_model == "Chroma1-HD"
    assert config.huggingface.license == "creativeml-openrail-m"
    assert config.huggingface.tags == ("chroma", "lora")
    assert config.huggingface.provenance == "Local source set."
    assert config.huggingface.adult_content is True
    assert config.huggingface.notes == "Needs review."


@pytest.mark.parametrize(
    ("key", "value"),
    [
        ("source_root", "somewhere"),
        ("manifest_path", "somewhere.xlsx"),
        ("export_root", "somewhere"),
        ("allow_export_inside_source", "true"),
    ],
)
def test_export_config_rejects_project_owned_path_settings(
    tmp_path: Path,
    key: str,
    value: str,
) -> None:
    config_path = tmp_path / "configs" / "ready.knf.yaml"
    config_path.parent.mkdir(parents=True, exist_ok=True)
    config_path.write_text(
        f"""
{key}: {value}
mappings:
  fullbody:
    - "0-FULLBODY"
""",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match=f"remove these key\\(s\\): {key}"):
        load_export_config(config_path)