Etusivu Tutki Apua
Kirjaudu sisään
lucky
/
news-mcp
1
0
Fork 0
Tiedostot Ongelmat 0 Pull-pyynnöt 0 Wiki
Puu: db1c8dea62
Haarat Tagit
news-mcp
/
test_embedding_support.py

test_embedding_support.py 1.9 KB
Historia Raaka

123456789101112131415161718192021222324252627282930313233343536373839 
  1. from __future__ import annotations
    2. 

  2. 

  3. from news_mcp.dedup.embedding_support import CandidateRules, cluster_is_candidate, cosine_similarity
    4. 

  4. 

  5. 

  6. def test_cosine_similarity_identical_vectors_is_one():
    7. 

  7.     assert cosine_similarity([1.0, 2.0, 3.0], [1.0, 2.0, 3.0]) == 1.0
    8. 

  8. 

  9. 

  10. def test_cosine_similarity_orthogonal_vectors_is_zero():
    11. 

  11.     assert cosine_similarity([1.0, 0.0], [0.0, 1.0]) == 0.0
    12. 

  12. 

  13. 

  14. def test_cosine_similarity_rejects_shape_mismatch():
    15. 

  15.     assert cosine_similarity([1.0, 2.0], [1.0]) == 0.0
    16. 

  16. 

  17. 

  18. def test_cluster_candidate_requires_topic_match_when_enabled():
    19. 

  19.     article = {"timestamp": "2026-04-01T00:00:00Z", "entities": ["Iran"]}
    20. 

  20.     cluster = {"topic": "macro", "timestamp": "2026-04-01T00:00:00Z", "entities": ["Iran"]}
    21. 

  21.     assert not cluster_is_candidate(article, cluster, article_topic="crypto", rules=CandidateRules(require_topic_match=True))
    22. 

  22. 

  23. 

  24. def test_cluster_candidate_allows_topic_mismatch_when_disabled():
    25. 

  25.     article = {"timestamp": "2026-04-01T00:00:00Z", "entities": ["Iran"]}
    26. 

  26.     cluster = {"topic": "macro", "timestamp": "2026-04-01T00:00:00Z", "entities": ["Iran"]}
    27. 

  27.     assert cluster_is_candidate(article, cluster, article_topic="crypto", rules=CandidateRules(require_topic_match=False))
    28. 

  28. 

  29. 

  30. def test_cluster_candidate_requires_entity_overlap():
    31. 

  31.     article = {"timestamp": "2026-04-01T00:00:00Z", "entities": ["Iran"]}
    32. 

  32.     cluster = {"topic": "macro", "timestamp": "2026-04-01T00:00:00Z", "entities": ["Israel"]}
    33. 

  33.     assert not cluster_is_candidate(article, cluster, article_topic="macro", rules=CandidateRules(require_topic_match=False, require_entity_overlap=1))
    34. 

  34. 

  35. 

  36. def test_cluster_candidate_accepts_overlap_and_recency():
    37. 

  37.     article = {"timestamp": "2026-04-01T00:00:00Z", "entities": ["Iran", "US"]}
    38. 

  38.     cluster = {"topic": "macro", "timestamp": "2026-03-31T23:00:00Z", "entities": ["Iran"]}
    39. 

  39.     assert cluster_is_candidate(article, cluster, article_topic="macro", rules=CandidateRules(require_topic_match=False, require_entity_overlap=1, max_age_hours=24))
    40.