anya/tests/test_stories.py

import unittest

from structs.headline import Headline
from structs.story import Story
from services.stories import cluster_stories


def h(text, tokens, url=""):
    return Headline(text, tokens, url)


class TestHeadlineSourceTracking(unittest.TestCase):
    def test_domain_extraction_strips_www_and_lowercases(self):
        headline = h("A story", ["a"], "https://www.CNN.com/us")
        self.assertEqual(headline.domain, "cnn.com")

    def test_domain_extraction_section_paths_collapse(self):
        us = h("A story", ["a"], "https://www.cnn.com/us")
        politics = h("A story", ["a"], "https://cnn.com/politics")
        self.assertEqual(us.domain, politics.domain)

    def test_domain_is_none_when_source_missing(self):
        self.assertIsNone(h("A story", ["a"]).domain)
        self.assertIsNone(h("A story", ["a"], "").domain)
        self.assertIsNone(h("A story", ["a"], "   ").domain)


class TestStory(unittest.TestCase):
    def test_sources_dedup_preserving_order(self):
        story = Story([
            h("Story A", ["a"], "https://www.cnn.com/us"),
            h("Story A", ["a"], "https://foxnews.com/politics"),
            h("Story A", ["a"], "https://www.cnn.com/politics"),
        ])
        self.assertEqual(story.sources, ["cnn.com", "foxnews.com"])
        self.assertEqual(story.source_count, 2)

    def test_representative_is_longest_headline(self):
        story = Story([
            h("Short", ["short"], "https://a.com"),
            h("A much longer descriptive headline", ["long"], "https://b.com"),
        ])
        self.assertEqual(story.representative, "A much longer descriptive headline")


class TestClusterStories(unittest.TestCase):
    def test_transitive_clustering_merges_chain(self):
        # Three headlines sharing tokens pairwise but not as a triple.
        headlines = [
            h("alpha beta", ["alpha", "beta"], "https://a.com"),
            h("beta gamma", ["beta", "gamma"], "https://b.com"),
            h("delta epsilon", ["delta", "epsilon"], "https://c.com"),
        ]
        stories = cluster_stories(headlines, threshold=0.4)
        texts = {s.representative for s in stories}
        # alpha/beta and beta/gamma share "beta" -> one cluster; delta/epsilon separate.
        self.assertEqual(len(stories), 2)
        self.assertIn("alpha beta", texts)
        self.assertIn("delta epsilon", texts)

    def test_non_similar_headlines_stay_separate(self):
        headlines = [
            h("apple pie", ["apple", "pie"], "https://a.com"),
            h("quantum physics", ["quantum", "physics"], "https://b.com"),
        ]
        stories = cluster_stories(headlines, threshold=0.6)
        self.assertEqual(len(stories), 2)

    def test_empty_input(self):
        self.assertEqual(cluster_stories([], 0.75), [])


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