jbell730
/
anya

import unittest
from unittest.mock import patch, MagicMock
import requests
from services.headlines import prepare_headlines, is_headline, DEFAULT_TIMEOUT, DEFAULT_HEADERS


class TestHeadlinesTimeout(unittest.TestCase):
    def setUp(self):
        self.stopwords = {"the", "a", "an", "in", "on"}

    @patch("services.headlines.requests.get")
    def test_default_timeout_used(self, mock_get):
        mock_response = MagicMock()
        mock_response.status_code = 200
        mock_response.content = b"<html><body><a href='#'>Default Timeout Headline</a></body></html>"
        mock_response.text = "<html><body><a href='#'>Default Timeout Headline</a></body></html>"
        mock_get.return_value = mock_response

        sources = ["https://example.com/news"]
        headlines = prepare_headlines(sources, self.stopwords)

        mock_get.assert_called_once_with("https://example.com/news", allow_redirects=True, timeout=DEFAULT_TIMEOUT, headers=DEFAULT_HEADERS)
        self.assertEqual(len(headlines), 1)
        self.assertEqual(headlines[0].display_text, "Default Timeout Headline")

    @patch("services.headlines.requests.get")
    def test_custom_timeout_used(self, mock_get):
        mock_response = MagicMock()
        mock_response.status_code = 200
        mock_response.content = b"<html><body><a href='#'>Custom Timeout Headline</a></body></html>"
        mock_response.text = "<html><body><a href='#'>Custom Timeout Headline</a></body></html>"
        mock_get.return_value = mock_response

        sources = ["https://example.com/news"]
        headlines = prepare_headlines(sources, self.stopwords, timeout=10)

        mock_get.assert_called_once_with("https://example.com/news", allow_redirects=True, timeout=10, headers=DEFAULT_HEADERS)
        self.assertEqual(len(headlines), 1)

    @patch("services.headlines.requests.get")
    def test_timeout_skips_slow_request_and_processes_others(self, mock_get):
        slow_url = "https://slow-source.example.com"
        fast_url = "https://fast-source.example.com"

        def side_effect(url, **kwargs):
            if url == slow_url:
                raise requests.exceptions.Timeout("Connection timed out after %s seconds" % kwargs.get("timeout"))
            fast_response = MagicMock()
            fast_response.status_code = 200
            fast_response.content = b"<html><body><span>Breaking News Story</span></body></html>"
            fast_response.text = "<html><body><span>Breaking News Story</span></body></html>"
            return fast_response

        mock_get.side_effect = side_effect

        sources = [slow_url, fast_url]
        headlines = prepare_headlines(sources, self.stopwords, timeout=3)

        self.assertEqual(mock_get.call_count, 2)
        self.assertEqual(len(headlines), 1)
        self.assertEqual(headlines[0].display_text, "Breaking News Story")

    @patch("services.headlines.requests.get")
    def test_connect_timeout_and_read_timeout_skipped(self, mock_get):
        connect_timeout_url = "https://connect-timeout.com"
        read_timeout_url = "https://read-timeout.com"

        mock_get.side_effect = [
            requests.exceptions.ConnectTimeout("Connect timeout"),
            requests.exceptions.ReadTimeout("Read timeout")
        ]

        sources = [connect_timeout_url, read_timeout_url]
        headlines = prepare_headlines(sources, self.stopwords)

        self.assertEqual(mock_get.call_count, 2)
        self.assertEqual(len(headlines), 0)


class TestHeadlineSafeguards(unittest.TestCase):
    def setUp(self):
        self.stopwords = {"the", "a", "an", "in", "on", "and", "of", "to"}

    def test_non_headline_link_texts_rejected(self):
        non_headlines = [
            "Account Settings",
            "Follow",
            "Television",
            "Sign In",
            "Home",
            "Politics",
            "About Us",
            "Contact Us",
            "Menu",
            "Search",
            "",
            "   ",
            "123",
            "...",
            "in on a",  # only stopwords
        ]
        for item in non_headlines:
            with self.subTest(item=item):
                self.assertFalse(is_headline(item, self.stopwords), f"{item!r} should not be recognized as a headline")

    def test_valid_headlines_accepted(self):
        valid_headlines = [
            "Breaking News Story",
            "Custom Timeout Headline",
            "Senate passes major infrastructure bill",
            "Scientists discover new ocean species",
            "Federal Reserve holds interest rates steady",
        ]
        for item in valid_headlines:
            with self.subTest(item=item):
                self.assertTrue(is_headline(item, self.stopwords), f"{item!r} should be recognized as a headline")

    @patch("services.headlines.requests.get")
    def test_prepare_headlines_filters_out_navigation_and_non_headlines(self, mock_get):
        html_content = """
        <html>
            <body>
                <a href="/settings">Account Settings</a>
                <a href="/social">Follow</a>
                <a href="/tv">Television</a>
                <a href="/login">Sign In</a>
                <a href="/article1">Senate passes major infrastructure bill</a>
                <span>Home</span>
                <span>Federal Reserve holds interest rates steady</span>
            </body>
        </html>
        """
        mock_response = MagicMock()
        mock_response.status_code = 200
        mock_response.content = html_content.encode("utf-8")
        mock_response.text = html_content
        mock_get.return_value = mock_response

        sources = ["https://example.com/news"]
        headlines = prepare_headlines(sources, self.stopwords)

        self.assertEqual(len(headlines), 2)
        extracted_texts = [h.display_text for h in headlines]
        self.assertIn("Senate passes major infrastructure bill", extracted_texts)
        self.assertIn("Federal Reserve holds interest rates steady", extracted_texts)
        self.assertNotIn("Account Settings", extracted_texts)
        self.assertNotIn("Follow", extracted_texts)
        self.assertNotIn("Television", extracted_texts)
        self.assertNotIn("Sign In", extracted_texts)
        self.assertNotIn("Home", extracted_texts)


if __name__ == "__main__":
    unittest.main()