feat(trip-cluster): coverage detection — completeness flags + outliers

2026-06-27 17:19:06 +02:00
parent e01b30a5b5
commit 8d58ec03a9
2 changed files with 83 additions and 0 deletions
@@ -0,0 +1,48 @@
+from photoflow.core.models import Asset
+from app.clustering import CandidateCluster
+from app.coverage import coverage_members
+
+
+def _a(i, taken):
+    return Asset(immich_id=i, taken_at=taken)
+
+
+def test_coverage_candidate_inside_seeded_window():
+    members = [_a("a", "2019-06-01T10:00:00"), _a("b", "2019-06-01T12:00:00"),
+               _a("c", "2019-06-02T10:00:00")]
+    intruder = _a("x", "2019-06-01T13:00:00")          # in window, untagged
+    outside = _a("y", "2019-07-01T10:00:00")           # out of window
+    by_id = {m.immich_id: m for m in members + [intruder, outside]}
+    cand = CandidateCluster(member_ids=["a", "b", "c"], start_at="2019-06-01T10:00:00",
+                            end_at="2019-06-02T10:00:00", suggested_name="Italy 2019",
+                            confidence=0.95, kind_guess="trip", seed_tag="Italy 2019")
+    out = coverage_members(cand, by_id)
+    flagged = {m.immich_id for m in out if m.flagged_coverage}
+    assert flagged == {"x"}                            # only the in-window intruder
+    assert all(not m.included for m in out if m.flagged_coverage)
+
+
+def test_outlier_member_far_from_bulk():
+    members = [_a("a", "2019-06-01T10:00:00"), _a("a2", "2019-06-01T11:00:00"),
+               _a("a3", "2019-06-01T12:00:00"),
+               _a("z", "2019-09-01T10:00:00")]          # tagged but months away
+    by_id = {m.immich_id: m for m in members}
+    cand = CandidateCluster(member_ids=["a", "a2", "a3", "z"],
+                            start_at="2019-06-01T10:00:00", end_at="2019-09-01T10:00:00",
+                            suggested_name="Italy 2019", confidence=0.95,
+                            kind_guess="trip", seed_tag="Italy 2019")
+    out = coverage_members(cand, by_id)
+    outliers = {m.immich_id for m in out if m.is_outlier}
+    assert outliers == {"z"}
+
+
+def test_non_seeded_cluster_has_no_coverage_or_outliers():
+    members = [_a("a", "2019-06-01T10:00:00"), _a("b", "2019-06-01T12:00:00")]
+    intruder = _a("x", "2019-06-01T11:00:00")
+    by_id = {m.immich_id: m for m in members + [intruder]}
+    cand = CandidateCluster(member_ids=["a", "b"], start_at="2019-06-01T10:00:00",
+                            end_at="2019-06-01T12:00:00", suggested_name="Trip",
+                            confidence=0.4, kind_guess="everyday", seed_tag=None)
+    out = coverage_members(cand, by_id)
+    assert {m.immich_id for m in out} == {"a", "b"}
+    assert not any(m.flagged_coverage or m.is_outlier for m in out)