CheddahBot/tests/test_cora_distribute.py

"""Tests for the Cora distribution watcher (scheduler._distribute_cora_file)."""

from __future__ import annotations

from dataclasses import dataclass, field
from pathlib import Path
from unittest.mock import MagicMock, patch

from cheddahbot.config import AutoCoraConfig, Config, ContentConfig, LinkBuildingConfig


@dataclass
class FakeTask:
    """Minimal ClickUp task stub for distribution tests."""

    id: str = "fake_id"
    name: str = ""
    task_type: str = ""
    custom_fields: dict = field(default_factory=dict)


def _make_scheduler(tmp_path, *, lb_folder="", content_inbox="", human_inbox=""):
    """Build a Scheduler with temp paths and mocked dependencies."""
    from cheddahbot.scheduler import Scheduler

    config = Config()
    config.link_building = LinkBuildingConfig(watch_folder=lb_folder)
    config.content = ContentConfig(cora_inbox=content_inbox)
    config.autocora = AutoCoraConfig(cora_human_inbox=human_inbox, enabled=True)
    config.clickup.enabled = True
    config.clickup.space_id = "sp1"
    config.clickup.api_token = "tok"

    db = MagicMock()
    agent = MagicMock()
    sched = Scheduler(config=config, db=db, agent=agent)
    return sched


KW_FIELDS = {"Keyword": "ac drive repair"}


def _drop_xlsx(folder: Path, name: str = "ac-drive-repair.xlsx") -> Path:
    """Create a dummy xlsx file in the given folder."""
    folder.mkdir(parents=True, exist_ok=True)
    p = folder / name
    p.write_bytes(b"fake-xlsx-data")
    return p


# ── Distribution logic tests ──


def test_distribute_lb_only(tmp_path):
    """LB task matched → copies to cora-inbox only."""
    human = tmp_path / "human"
    lb = tmp_path / "lb"
    content = tmp_path / "content"
    xlsx = _drop_xlsx(human)

    sched = _make_scheduler(
        tmp_path, lb_folder=str(lb), content_inbox=str(content), human_inbox=str(human)
    )

    tasks = [FakeTask(name="LB task", task_type="Link Building", custom_fields=KW_FIELDS)]

    with patch.object(sched, "_get_clickup_client") as mock_client:
        mock_client.return_value.get_tasks_from_overall_lists.return_value = tasks
        sched._distribute_cora_file(xlsx)

    assert (lb / xlsx.name).exists()
    assert not (content / xlsx.name).exists()
    assert (human / "processed" / xlsx.name).exists()
    assert not xlsx.exists()


def test_distribute_content_only(tmp_path):
    """Content task matched → copies to content-cora-inbox only."""
    human = tmp_path / "human"
    lb = tmp_path / "lb"
    content = tmp_path / "content"
    xlsx = _drop_xlsx(human)

    sched = _make_scheduler(
        tmp_path, lb_folder=str(lb), content_inbox=str(content), human_inbox=str(human)
    )

    tasks = [FakeTask(name="CC task", task_type="Content Creation", custom_fields=KW_FIELDS)]

    with patch.object(sched, "_get_clickup_client") as mock_client:
        mock_client.return_value.get_tasks_from_overall_lists.return_value = tasks
        sched._distribute_cora_file(xlsx)

    assert not (lb / xlsx.name).exists()
    assert (content / xlsx.name).exists()
    assert (human / "processed" / xlsx.name).exists()


def test_distribute_mixed(tmp_path):
    """Both LB and Content tasks matched → copies to both inboxes."""
    human = tmp_path / "human"
    lb = tmp_path / "lb"
    content = tmp_path / "content"
    xlsx = _drop_xlsx(human)

    sched = _make_scheduler(
        tmp_path, lb_folder=str(lb), content_inbox=str(content), human_inbox=str(human)
    )

    tasks = [
        FakeTask(name="LB task", task_type="Link Building", custom_fields=KW_FIELDS),
        FakeTask(name="CC task", task_type="Content Creation", custom_fields=KW_FIELDS),
    ]

    with patch.object(sched, "_get_clickup_client") as mock_client:
        mock_client.return_value.get_tasks_from_overall_lists.return_value = tasks
        sched._distribute_cora_file(xlsx)

    assert (lb / xlsx.name).exists()
    assert (content / xlsx.name).exists()
    assert (human / "processed" / xlsx.name).exists()


def test_distribute_no_match(tmp_path):
    """No matching tasks → file stays in inbox, not moved to processed."""
    human = tmp_path / "human"
    lb = tmp_path / "lb"
    content = tmp_path / "content"
    xlsx = _drop_xlsx(human)

    sched = _make_scheduler(
        tmp_path, lb_folder=str(lb), content_inbox=str(content), human_inbox=str(human)
    )

    with patch.object(sched, "_get_clickup_client") as mock_client:
        mock_client.return_value.get_tasks_from_overall_lists.return_value = []
        sched._distribute_cora_file(xlsx)

    assert xlsx.exists()  # Still in inbox
    assert not (human / "processed" / xlsx.name).exists()


def test_distribute_opo_task(tmp_path):
    """On Page Optimization task → copies to content inbox."""
    human = tmp_path / "human"
    lb = tmp_path / "lb"
    content = tmp_path / "content"
    xlsx = _drop_xlsx(human)

    sched = _make_scheduler(
        tmp_path, lb_folder=str(lb), content_inbox=str(content), human_inbox=str(human)
    )

    tasks = [FakeTask(name="OPO task", task_type="On Page Optimization", custom_fields=KW_FIELDS)]

    with patch.object(sched, "_get_clickup_client") as mock_client:
        mock_client.return_value.get_tasks_from_overall_lists.return_value = tasks
        sched._distribute_cora_file(xlsx)

    assert not (lb / xlsx.name).exists()
    assert (content / xlsx.name).exists()


# ── Scan tests ──


def test_scan_skips_processed(tmp_path):
    """Files already in processed/ are skipped."""
    human = tmp_path / "human"
    lb = tmp_path / "lb"
    content = tmp_path / "content"

    # File in both top-level and processed/
    _drop_xlsx(human)
    _drop_xlsx(human / "processed")

    sched = _make_scheduler(
        tmp_path, lb_folder=str(lb), content_inbox=str(content), human_inbox=str(human)
    )

    with patch.object(sched, "_distribute_cora_file") as mock_dist:
        sched._scan_cora_human_inbox()
        mock_dist.assert_not_called()


def test_scan_skips_temp_files(tmp_path):
    """Office temp files (~$...) are skipped."""
    human = tmp_path / "human"
    lb = tmp_path / "lb"
    content = tmp_path / "content"

    _drop_xlsx(human, name="~$ac-drive-repair.xlsx")

    sched = _make_scheduler(
        tmp_path, lb_folder=str(lb), content_inbox=str(content), human_inbox=str(human)
    )

    with patch.object(sched, "_distribute_cora_file") as mock_dist:
        sched._scan_cora_human_inbox()
        mock_dist.assert_not_called()


def test_scan_empty_inbox(tmp_path):
    """Empty inbox → no-op."""
    human = tmp_path / "human"
    human.mkdir()

    sched = _make_scheduler(tmp_path, human_inbox=str(human))

    with patch.object(sched, "_distribute_cora_file") as mock_dist:
        sched._scan_cora_human_inbox()
        mock_dist.assert_not_called()


def test_distribute_copy_failure_no_move(tmp_path):
    """If copy fails, original is NOT moved to processed."""
    human = tmp_path / "human"
    xlsx = _drop_xlsx(human)

    sched = _make_scheduler(tmp_path, lb_folder="/nonexistent/network/path", human_inbox=str(human))

    tasks = [FakeTask(name="LB task", task_type="Link Building", custom_fields=KW_FIELDS)]

    with (
        patch.object(sched, "_get_clickup_client") as mock_client,
        patch("cheddahbot.scheduler.shutil.copy2", side_effect=OSError("network down")),
    ):
        mock_client.return_value.get_tasks_from_overall_lists.return_value = tasks
        sched._distribute_cora_file(xlsx)

    assert xlsx.exists()  # Original untouched
    assert not (human / "processed" / xlsx.name).exists()