claudia/tests/test_runner_errors.py

"""Runner error handling test: reporters receive error notifications and summary collects errors."""

from __future__ import annotations

from typing import Any, Dict, List

from claudia.core import Document
from claudia.spiders import BaseSpider
from claudia.db import MemoryDB
from claudia.scheduler import Runner
from claudia.reporter import MemoryReporter


class BoomSpider(BaseSpider):
    """Spider that intentionally raises to exercise Runner error handling."""

    name = "boom"

    def build_payload(self, url: str) -> Dict[str, Any]:
        return {}

    def parse(
        self, url: str, content: str, payload: Dict[str, Any]
    ) -> List[Document]:  # pragma: no cover
        return []

    def fetch(self, url: str, payload: Dict[str, Any]) -> str:  # pragma: no cover
        raise RuntimeError("boom!")


def test_runner_handles_errors_and_reports():
    """Runner should record errors and reporters should receive error notifications."""
    spider = BoomSpider()
    db = MemoryDB()
    reporter = MemoryReporter()
    runner = Runner(db=db, reporters=[reporter])

    summary = runner.run({spider: ["https://example.com/a"]})

    assert summary.total_docs == 0
    assert summary.errors and any("boom" in e for e in summary.errors)
    assert any(e.startswith("error:") for e in reporter.events)