import json
from pathlib import Path
from typing import Any

from br_sovereign_llm.cli import main


def test_validate_sources_command(capsys: Any) -> None:
    assert main(["validate-sources"]) == 0
    output = json.loads(capsys.readouterr().out)

    assert output["ok"] is True
    assert output["source_count"] >= 10
    assert "wikimedia_ptwiki" in output["scientific_source_ids"]


def test_validate_project_command_covers_all_registries(capsys: Any) -> None:
    assert main(["validate-project"]) == 0
    output = json.loads(capsys.readouterr().out)

    assert output["ok"] is True
    assert output["benchmarks"]["benchmark_count"] == 10
    assert output["configurations"]["configuration_count"] == 19
    assert output["tokenizer"]["vocabulary_size"] == 32768
    assert len(output["model_ladder"]) == 6


def test_project_root_allows_installed_cli_from_another_directory(
    tmp_path: Path, capsys: Any, monkeypatch: Any
) -> None:
    project_root = Path(__file__).resolve().parents[2]
    monkeypatch.chdir(tmp_path)

    assert main(["--project-root", str(project_root), "validate-project"]) == 0
    output = json.loads(capsys.readouterr().out)
    assert output["configurations"]["configuration_count"] == 19


def test_cli_reports_invalid_registry_without_traceback(
    tmp_path: Path, capsys: Any
) -> None:
    missing = tmp_path / "missing.yaml"

    assert main(["validate-benchmarks", "--registry", str(missing)]) == 2
    captured = capsys.readouterr()
    error = json.loads(captured.err)

    assert captured.out == ""
    assert error["ok"] is False
    assert error["error_type"] == "BenchmarkRegistryError"


def test_template_fingerprint_discloses_unresolved_inputs(capsys: Any) -> None:
    assert (
        main(
            [
                "fingerprint",
                "--study",
                "configs/experiments/scaling-study.yaml",
                "--runtime",
                "configs/runtime/local-cpu.yaml",
            ]
        )
        == 0
    )
    output = json.loads(capsys.readouterr().out)

    assert output["fingerprint_kind"] == "configuration_template"
    assert output["scientific_fingerprint_ready"] is False
    assert "$" + "{TOKENIZER_ID}" in output["study_unresolved_placeholders"]


def test_evidence_report_keeps_every_metric_unmeasured(
    tmp_path: Path, capsys: Any
) -> None:
    output_path = tmp_path / "evidence.json"

    assert main(["evidence-report", "--output", str(output_path)]) == 0
    command_output = json.loads(capsys.readouterr().out)
    report = json.loads(output_path.read_text(encoding="utf-8"))

    assert command_output["eligible_for_public_claims"] is False
    assert report["eligible_for_public_claims"] is False
    assert all(item["state"] == "not_measured" for item in report["evidence"])


def test_tokenizer_manifest_uses_content_addressed_input_identifier(
    tmp_path: Path, capsys: Any
) -> None:
    assert (
        main(
            [
                "train-tokenizer",
                "--output",
                str(tmp_path / "tokenizer"),
            ]
        )
        == 0
    )
    capsys.readouterr()
    manifest_text = (tmp_path / "tokenizer/tokenizer-manifest.json").read_text(
        encoding="utf-8"
    )
    manifest = json.loads(manifest_text)

    assert set(manifest["input_hashes"]) == {
        "urn:sha256:6896439e772f212c2848846f1f908410a606d81cd8845583024c16496a07979f"
    }
    assert "/Users/" not in manifest_text