| 1 | """The topic subject gets a higher per-author cap than incidental authors. |
| 2 | |
| 3 | The flat cap of 3 is anti-flooding protection and is right for third parties. |
| 4 | But on a person or company topic the subject is the point of the query: the |
| 5 | measured 'Peter Steinberger steipete' run recovered 8 subject-authored posts, |
| 6 | and a flat cap would discard 5 of them after the rest of Phase A worked to |
| 7 | retrieve and keep them. |
| 8 | |
| 9 | The raised cap stays bounded so a prolific subject cannot crowd out all |
| 10 | commentary about them. |
| 11 | """ |
| 12 | |
| 13 | from lib import fusion, schema |
| 14 | |
| 15 | |
| 16 | def _cand(cid: str, author: str) -> schema.Candidate: |
| 17 | url = f"https://x.com/{author}/status/{cid}" |
| 18 | cand = schema.Candidate( |
| 19 | candidate_id=cid, |
| 20 | item_id=f"i{cid}", |
| 21 | source="x", |
| 22 | title=f"post {cid}", |
| 23 | url=url, |
| 24 | snippet="s", |
| 25 | subquery_labels=["primary"], |
| 26 | native_ranks={"primary:x": 1}, |
| 27 | local_relevance=0.5, |
| 28 | freshness=80, |
| 29 | engagement=50, |
| 30 | source_quality=0.68, |
| 31 | rrf_score=0.02, |
| 32 | ) |
| 33 | cand.source_items = [ |
| 34 | schema.SourceItem( |
| 35 | item_id=f"i{cid}", source="x", title=f"post {cid}", |
| 36 | body="b", url=url, author=author, |
| 37 | ) |
| 38 | ] |
| 39 | return cand |
| 40 | |
| 41 | |
| 42 | def _subject_posts(n: int, author: str = "steipete"): |
| 43 | return [_cand(str(i), author) for i in range(1, n + 1)] |
| 44 | |
| 45 | |
| 46 | def test_third_party_author_is_still_capped_at_three(): |
| 47 | kept = fusion._apply_per_author_cap(_subject_posts(8, author="rando")) |
| 48 | assert len(kept) == 3, "the anti-flooding cap must be unchanged for third parties" |
| 49 | |
| 50 | |
| 51 | def test_subject_keeps_more_than_three_posts(): |
| 52 | kept = fusion._apply_per_author_cap( |
| 53 | _subject_posts(8), first_party_handles={"steipete"} |
| 54 | ) |
| 55 | assert len(kept) > 3, ( |
| 56 | "the subject of the topic was capped at 3 of 8 recovered posts, " |
| 57 | "discarding evidence the rest of Phase A worked to keep" |
| 58 | ) |
| 59 | |
| 60 | |
| 61 | def test_subject_cap_is_bounded(): |
| 62 | """A prolific subject must not fill the pool.""" |
| 63 | kept = fusion._apply_per_author_cap( |
| 64 | _subject_posts(40), first_party_handles={"steipete"} |
| 65 | ) |
| 66 | assert len(kept) <= fusion._MAX_ITEMS_PER_FIRST_PARTY_AUTHOR |
| 67 | assert len(kept) < 40 |
| 68 | |
| 69 | |
| 70 | def test_mixed_pool_caps_each_author_by_its_own_rule(): |
| 71 | pool = _subject_posts(8) + [_cand(f"r{i}", "rando") for i in range(1, 8)] |
| 72 | kept = fusion._apply_per_author_cap(pool, first_party_handles={"steipete"}) |
| 73 | subject = [c for c in kept if c.source_items[0].author == "steipete"] |
| 74 | third = [c for c in kept if c.source_items[0].author == "rando"] |
| 75 | assert len(subject) > 3 |
| 76 | assert len(third) == 3 |
| 77 | |
| 78 | |
| 79 | def test_no_handles_behaves_exactly_as_before(): |
| 80 | pool = _subject_posts(8) |
| 81 | assert len(fusion._apply_per_author_cap(pool)) == \ |
| 82 | len(fusion._apply_per_author_cap(pool, first_party_handles=set())) |
| 83 | |
| 84 | |
| 85 | def test_handles_are_matched_case_insensitively(): |
| 86 | kept = fusion._apply_per_author_cap( |
| 87 | _subject_posts(8), first_party_handles={"SteiPete"} |
| 88 | ) |
| 89 | assert len(kept) > 3 |
| 90 | |
| 91 | |
| 92 | def test_ordering_within_an_author_is_preserved(): |
| 93 | """Candidates arrive sorted by quality; the cap keeps the best N.""" |
| 94 | kept = fusion._apply_per_author_cap( |
| 95 | _subject_posts(8), first_party_handles={"steipete"} |
| 96 | ) |
| 97 | assert [c.candidate_id for c in kept] == \ |
| 98 | [str(i) for i in range(1, len(kept) + 1)] |
| 99 | |
| 100 | |
| 101 | def test_weighted_rrf_accepts_first_party_handles(): |
| 102 | import inspect |
| 103 | sig = inspect.signature(fusion.weighted_rrf) |
| 104 | assert "first_party_handles" in sig.parameters, ( |
| 105 | "the fusion entry point must accept the run's resolved handles or the " |
| 106 | "raised cap can never fire in production" |
| 107 | ) |
| 108 | assert sig.parameters["first_party_handles"].default in (None, frozenset()), ( |
| 109 | "must stay optional so the discovery caller is unaffected" |
| 110 | ) |
| 111 |