tests/test_02_skip_existing.py

"""Unit test for `02_collect_news.py --skip-existing` (issue #59).

Verifies that the filter helper correctly drops events whose article files
already exist on disk and keeps the rest. We test the helper rather than
invoking the script end-to-end (which loads an LLM client and hits GDACS).
"""

from __future__ import annotations

import importlib.util
from pathlib import Path

from src.models.schemas import FloodEvent


def _load_collect_module():
    path = Path(__file__).resolve().parent.parent / "scripts" / "02_collect_news.py"
    spec = importlib.util.spec_from_file_location("collect_news_02", path)
    assert spec and spec.loader
    mod = importlib.util.module_from_spec(spec)
    spec.loader.exec_module(mod)
    return mod


def _make_event(event_id: str) -> FloodEvent:
    return FloodEvent(
        event_id=event_id,
        country="X",
        iso="XXX",
        region="Europe",
        location=None,
        latitude=None,
        longitude=None,
        start_date="2020-01-01",
        end_date="2020-01-01",
        disaster_subtype=None,
        origin=None,
        magnitude=None,
        magnitude_scale=None,
        total_deaths=None,
        total_affected=None,
        total_damage_k_usd=None,
    )


def test_filter_skips_events_with_cached_articles(tmp_path: Path) -> None:
    mod = _load_collect_module()
    articles_dir = tmp_path
    (articles_dir / "EVT-A_abc123.json").write_text("{}")
    (articles_dir / "EVT-C_def456.json").write_text("{}")

    events = [_make_event("EVT-A"), _make_event("EVT-B"), _make_event("EVT-C")]
    kept = mod._filter_skip_existing(events, articles_dir)
    assert [e.event_id for e in kept] == ["EVT-B"]


def test_filter_keeps_all_when_articles_dir_empty(tmp_path: Path) -> None:
    mod = _load_collect_module()
    articles_dir = tmp_path  # empty
    events = [_make_event("EVT-A"), _make_event("EVT-B")]
    kept = mod._filter_skip_existing(events, articles_dir)
    assert [e.event_id for e in kept] == ["EVT-A", "EVT-B"]


def test_filter_keeps_all_when_articles_dir_missing(tmp_path: Path) -> None:
    mod = _load_collect_module()
    articles_dir = tmp_path / "does_not_exist"
    events = [_make_event("EVT-A")]
    kept = mod._filter_skip_existing(events, articles_dir)
    assert [e.event_id for e in kept] == ["EVT-A"]