marki-docx/tests/test_workflows.py

"""Tests for T04 — Composite workflow orchestration (FR-1300)."""

from __future__ import annotations

import textwrap
from pathlib import Path

import pytest

from markidocx.evidence import EvidenceStore
from markidocx.workflows import (
    SUPPORTED_WORKFLOWS,
    WorkflowError,
    WorkflowResult,
    run_workflow,
    run_workflow_from_content,
)

SIMPLE_MANIFEST = textwrap.dedent("""\
    project:
      name: "Test Document"
      feature_level: level1
      family: article

    sources:
      - path: doc.md

    output:
      dir: ./dist
""")

SIMPLE_MARKDOWN = textwrap.dedent("""\
    # Hello World

    A paragraph with **bold** text.

    ## Section

    - Item one
    - Item two
""")


@pytest.fixture()
def tmp_project(tmp_path: Path) -> Path:
    (tmp_path / "doc.md").write_text(SIMPLE_MARKDOWN, encoding="utf-8")
    (tmp_path / "manifest.yaml").write_text(SIMPLE_MANIFEST, encoding="utf-8")
    (tmp_path / "dist").mkdir()
    return tmp_path


@pytest.fixture()
def store(tmp_path: Path) -> EvidenceStore:
    return EvidenceStore(base_dir=tmp_path / "evidence")


# ---------------------------------------------------------------------------
# Known / unknown workflow names
# ---------------------------------------------------------------------------


def test_supported_workflows_set() -> None:
    assert "single-file-roundtrip" in SUPPORTED_WORKFLOWS
    assert "multi-file-roundtrip" in SUPPORTED_WORKFLOWS
    assert "release-regression" in SUPPORTED_WORKFLOWS
    assert "family-switch-build" in SUPPORTED_WORKFLOWS


def test_unknown_workflow_raises(tmp_project: Path) -> None:
    with pytest.raises(WorkflowError, match="Unknown workflow"):
        run_workflow("no-such-workflow", tmp_project / "manifest.yaml")


# ---------------------------------------------------------------------------
# WorkflowResult shape (FR-1303, FR-1309)
# ---------------------------------------------------------------------------


def test_single_file_roundtrip_result_shape(tmp_project: Path, store: EvidenceStore) -> None:
    result = run_workflow("single-file-roundtrip", tmp_project / "manifest.yaml", store)
    assert isinstance(result, WorkflowResult)
    assert result.run_id
    assert result.workflow_name == "single-file-roundtrip"
    assert result.timestamp
    assert result.classification in ("full", "with-fallback", "partial", "failed")


def test_result_has_steps(tmp_project: Path, store: EvidenceStore) -> None:
    result = run_workflow("single-file-roundtrip", tmp_project / "manifest.yaml", store)
    assert len(result.steps) >= 1
    for step in result.steps:
        assert step.name
        assert step.status in ("executed", "skipped", "failed")


def test_result_has_aggregate_output(tmp_project: Path, store: EvidenceStore) -> None:
    result = run_workflow("single-file-roundtrip", tmp_project / "manifest.yaml", store)
    assert isinstance(result.aggregate_output, dict)


# ---------------------------------------------------------------------------
# Step visibility (FR-1304)
# ---------------------------------------------------------------------------


def test_single_file_roundtrip_step_names(tmp_project: Path, store: EvidenceStore) -> None:
    result = run_workflow("single-file-roundtrip", tmp_project / "manifest.yaml", store)
    step_names = [s.name for s in result.steps]
    assert "validate" in step_names
    assert "build" in step_names
    assert "import" in step_names
    assert "compare" in step_names


def test_failed_validate_stops_early(tmp_path: Path, store: EvidenceStore) -> None:
    """If validation fails, subsequent steps should not execute."""
    bad_manifest = tmp_path / "manifest.yaml"
    bad_manifest.write_text("project:\n  name: x\n", encoding="utf-8")
    (tmp_path / "dist").mkdir()
    result = run_workflow("single-file-roundtrip", bad_manifest, store)
    assert result.classification == "failed"
    step_names = [s.name for s in result.steps]
    assert "validate" in step_names
    # build should not appear after a validation failure
    assert "build" not in step_names


# ---------------------------------------------------------------------------
# Workflow classification (FR-1305)
# ---------------------------------------------------------------------------


def test_successful_roundtrip_classification(tmp_project: Path, store: EvidenceStore) -> None:
    result = run_workflow("single-file-roundtrip", tmp_project / "manifest.yaml", store)
    assert result.classification in ("full", "with-fallback")


# ---------------------------------------------------------------------------
# Evidence stored (FR-1309)
# ---------------------------------------------------------------------------


def test_evidence_written_to_store(tmp_project: Path, store: EvidenceStore) -> None:
    result = run_workflow("single-file-roundtrip", tmp_project / "manifest.yaml", store)
    runs = store.list_runs()
    assert result.run_id in runs


def test_validation_report_in_store(tmp_project: Path, store: EvidenceStore) -> None:
    result = run_workflow("single-file-roundtrip", tmp_project / "manifest.yaml", store)
    report = store.get_report(result.run_id, "validation")
    assert report is not None
    assert report.data["status"] == "ok"


# ---------------------------------------------------------------------------
# multi-file-roundtrip and family-switch-build
# ---------------------------------------------------------------------------


def test_multi_file_roundtrip_returns_result(tmp_project: Path, store: EvidenceStore) -> None:
    result = run_workflow("multi-file-roundtrip", tmp_project / "manifest.yaml", store)
    assert result.workflow_name == "multi-file-roundtrip"
    assert result.classification in ("full", "with-fallback", "partial", "failed")


def test_family_switch_build_produces_multiple_steps(tmp_project: Path, store: EvidenceStore) -> None:
    result = run_workflow("family-switch-build", tmp_project / "manifest.yaml", store)
    # Should have one build step per family
    build_steps = [s for s in result.steps if s.name.startswith("build:")]
    assert len(build_steps) == 3  # article, book, website
    assert "builds" in result.aggregate_output


def test_release_regression_returns_result(tmp_project: Path, store: EvidenceStore) -> None:
    result = run_workflow("release-regression", tmp_project / "manifest.yaml", store)
    assert result.workflow_name == "release-regression"


# ---------------------------------------------------------------------------
# run_workflow_from_content (REST/MCP path)
# ---------------------------------------------------------------------------


def test_run_from_content_single_file(store: EvidenceStore) -> None:
    result = run_workflow_from_content(
        "single-file-roundtrip",
        SIMPLE_MANIFEST,
        [{"name": "doc.md", "content": SIMPLE_MARKDOWN}],
        store,
    )
    assert result.workflow_name == "single-file-roundtrip"
    assert result.run_id


def test_run_from_content_unknown_workflow() -> None:
    with pytest.raises(WorkflowError):
        run_workflow_from_content("bad-name", SIMPLE_MANIFEST, [])


# ---------------------------------------------------------------------------
# CLI: markidocx workflow (FR-1308 — CLI interface)
# ---------------------------------------------------------------------------


def test_cli_workflow_command(tmp_project: Path) -> None:
    from typer.testing import CliRunner

    from markidocx.cli import app

    runner = CliRunner()
    result = runner.invoke(app, ["workflow", "single-file-roundtrip", str(tmp_project / "manifest.yaml"), "--json"])
    assert result.exit_code in (0, 1)
    import json

    data = json.loads(result.output.strip())
    assert "run_id" in data
    assert "classification" in data


def test_cli_workflow_unknown_name(tmp_project: Path) -> None:
    from typer.testing import CliRunner

    from markidocx.cli import app

    runner = CliRunner()
    result = runner.invoke(app, ["workflow", "no-such", str(tmp_project / "manifest.yaml")])
    assert result.exit_code == 1