import unittest
from unittest.mock import patch, MagicMock
from services.ssr import extract_headlines
from services.headlines import prepare_headlines
HTML = """
"""
class TestExtractHeadlines(unittest.TestCase):
def test_json_ld_and_next_data(self):
items = extract_headlines(HTML)
headlines = [h for h, _ in items]
# schema.org headline and Next.js title+url are both recovered.
self.assertIn("Court rules on landmark case", headlines)
self.assertIn("Senate passes infrastructure bill", headlines)
# Organization `name` without an article signal is not treated as a headline.
self.assertNotIn("Example News", headlines)
# Date is parsed from JSON-LD datePublished.
by_title = {h: d for h, d in items}
self.assertIsNotNone(by_title["Court rules on landmark case"])
class TestPrepareHeadlinesSSR(unittest.TestCase):
def setUp(self):
self.stopwords = {"the", "a", "an", "in", "on", "and", "of", "to"}
@patch("services.headlines.requests.get")
def test_ssr_source_is_parsed(self, mock_get):
mock_response = MagicMock()
mock_response.status_code = 200
mock_response.content = HTML.encode("utf-8")
mock_response.text = HTML
mock_get.return_value = mock_response
headlines = prepare_headlines(["https://www.example.com/business/"], self.stopwords)
titles = [h.display_text for h in headlines]
# "Markets" is filtered out by the minimum-word headline check downstream.
self.assertIn("Court rules on landmark case", titles)
self.assertIn("Senate passes infrastructure bill", titles)
self.assertNotIn("Markets", titles)
if __name__ == "__main__":
unittest.main()