| 1 | """DripStack source: requested-only gating, windowing, and normalization.""" |
| 2 | |
| 3 | from unittest import mock |
| 4 | |
| 5 | from lib import dripstack, pipeline |
| 6 | |
| 7 | |
| 8 | def _api_item(**overrides): |
| 9 | item = { |
| 10 | "publicationSlug": "newsletter.semianalysis.com", |
| 11 | "slug": "nvidia-gpu-debt-backstop", |
| 12 | "title": "Nvidia GPU Debt Backstop", |
| 13 | "subtitle": "Capital, offtake and datacenters.", |
| 14 | "snippet": "Nvidia's backstop economics explained.", |
| 15 | "publishedAt": "2026-07-06T21:53:44.000Z", |
| 16 | "relevanceScore": 84, |
| 17 | "matchConfidence": "strong", |
| 18 | "topicCoverageRatio": 1, |
| 19 | "whyMatched": ["Matched article body for: nvidia", "Hybrid RRF — blended"], |
| 20 | } |
| 21 | item.update(overrides) |
| 22 | return item |
| 23 | |
| 24 | |
| 25 | def test_dripstack_absent_from_available_sources_by_default(): |
| 26 | assert "dripstack" not in pipeline.available_sources({}, None, x_pending=False) |
| 27 | |
| 28 | |
| 29 | def test_dripstack_available_only_when_explicitly_requested(): |
| 30 | available = pipeline.available_sources({}, ["dripstack"], x_pending=False) |
| 31 | assert "dripstack" in available |
| 32 | |
| 33 | |
| 34 | def test_search_goes_through_the_shared_http_choke_point(monkeypatch): |
| 35 | seen = {} |
| 36 | |
| 37 | def fake_get(url, headers=None, **kwargs): |
| 38 | seen["url"] = url |
| 39 | seen["retries"] = kwargs.get("retries") |
| 40 | return {"items": [_api_item()], "matchConfidence": "strong"} |
| 41 | |
| 42 | monkeypatch.setattr(dripstack.http, "get", fake_get) |
| 43 | items = dripstack.search_dripstack("Nvidia earnings", "2026-06-12", "2026-07-12") |
| 44 | |
| 45 | assert len(items) == 1 |
| 46 | assert seen["url"].startswith("https://dripstack.xyz/api/v1/search?") |
| 47 | assert seen["retries"] == 2 |
| 48 | |
| 49 | |
| 50 | def test_search_drops_results_outside_the_window(monkeypatch): |
| 51 | stale = _api_item(publishedAt="2026-05-20T19:06:01.000Z", slug="old-post") |
| 52 | fresh = _api_item() |
| 53 | undated = _api_item(publishedAt="", slug="undated-post") |
| 54 | monkeypatch.setattr( |
| 55 | dripstack.http, "get", |
| 56 | lambda *a, **k: {"items": [stale, fresh, undated], "matchConfidence": "strong"}, |
| 57 | ) |
| 58 | |
| 59 | items = dripstack.search_dripstack("Nvidia", "2026-06-12", "2026-07-12") |
| 60 | |
| 61 | slugs = [item["slug"] for item in items] |
| 62 | assert "old-post" not in slugs |
| 63 | assert "nvidia-gpu-debt-backstop" in slugs |
| 64 | assert "undated-post" in slugs # undated results are kept, not guessed |
| 65 | |
| 66 | |
| 67 | def test_search_failure_returns_empty_list(monkeypatch): |
| 68 | def boom(*a, **k): |
| 69 | raise OSError("connection reset") |
| 70 | |
| 71 | monkeypatch.setattr(dripstack.http, "get", boom) |
| 72 | assert dripstack.search_dripstack("Nvidia", "2026-06-12", "2026-07-12") == [] |
| 73 | |
| 74 | |
| 75 | def test_parse_normalizes_fields_and_relevance(): |
| 76 | parsed = dripstack.parse_dripstack_response([_api_item()], query="nvidia") |
| 77 | |
| 78 | item = parsed[0] |
| 79 | assert item["url"] == "https://newsletter.semianalysis.com/nvidia-gpu-debt-backstop" |
| 80 | assert item["date"] == "2026-07-06" |
| 81 | assert item["relevance"] == 0.84 |
| 82 | assert item["engagement"] == {} |
| 83 | assert item["author"] == "newsletter.semianalysis" |
| 84 | assert "Hybrid" not in item["why_relevant"] |
| 85 | assert item["metadata"]["publication_slug"] == "newsletter.semianalysis.com" |
| 86 | |
| 87 | |
| 88 | def test_parse_handles_missing_fields_conservatively(): |
| 89 | parsed = dripstack.parse_dripstack_response( |
| 90 | [{"title": "", "relevanceScore": "not-a-number"}], query="x" |
| 91 | ) |
| 92 | |
| 93 | item = parsed[0] |
| 94 | assert item["title"] == "DripStack result 1" |
| 95 | assert item["url"] == "" |
| 96 | assert item["relevance"] == 0.5 |
| 97 | |
| 98 | |
| 99 | def test_parse_emits_body_and_normalizer_prefers_it(): |
| 100 | parsed = dripstack.parse_dripstack_response([_api_item()], query="nvidia") |
| 101 | |
| 102 | assert parsed[0]["body"] == "Capital, offtake and datacenters." |
| 103 | |
| 104 | from lib import normalize |
| 105 | normalized = normalize._normalize_dripstack( |
| 106 | "dripstack", parsed[0], 0, "2026-06-12", "2026-07-12" |
| 107 | ) |
| 108 | assert normalized.body == "Capital, offtake and datacenters." |
| 109 | |
| 110 | |
| 111 | def test_dripstack_activates_via_persisted_include_sources(): |
| 112 | """INCLUDE_SOURCES is the .env checkbox for persistent opt-ins (the |
| 113 | LinkedIn/Perplexity pattern); dripstack must honor it, not only --search.""" |
| 114 | available = pipeline.available_sources( |
| 115 | {"INCLUDE_SOURCES": "dripstack"}, None, x_pending=False |
| 116 | ) |
| 117 | assert "dripstack" in available |
| 118 | |
| 119 | |
| 120 | def test_unrelated_include_sources_keeps_dripstack_off(): |
| 121 | available = pipeline.available_sources( |
| 122 | {"INCLUDE_SOURCES": "linkedin,tiktok"}, None, x_pending=False |
| 123 | ) |
| 124 | assert "dripstack" not in available |
| 125 | |
| 126 | |
| 127 | def test_include_sources_tolerates_whitespace_around_commas(): |
| 128 | available = pipeline.available_sources( |
| 129 | {"INCLUDE_SOURCES": "linkedin, dripstack"}, None, x_pending=False |
| 130 | ) |
| 131 | assert "dripstack" in available |
| 132 |