Merge pull request 'feat(halacha): #82.4 provenance-union על dedup-skip + #82.6 over-merge guard' (#192) from worktree-halacha-dedup-provenance-guard into main

mcp-server/tests/test_halacha_dedup_action.py

@@ -0,0 +1,53 @@
+"""Tests for #82.4 / #82.6 — dedup-on-insert decision + over-merge guard.
+
+``halacha_quality.dedup_action`` is the PAIRWISE decision a fresh halacha makes
+against its single nearest same-precedent neighbor: skip (semantic dup), flag
+(lexical tail), or keep. It compares to exactly ONE neighbor and only ever drops
+the *incoming* row, so a chain A~B~C can never collapse to one row — the
+over-merge guard (#82.6). Pure/offline.
+"""
+
+from __future__ import annotations
+
+import pytest
+
+from legal_mcp.services import halacha_quality as hq
+
+# operating point: DEDUP_COSINE=0.93 → dedup_distance=0.07 ; BAND=0.83 → 0.17
+DEDUP_D = 1.0 - 0.93
+BAND_D = 1.0 - 0.83
+
+SIMILAR_A = "מיצוי הליכים הוא תנאי סף להגשת ערר לוועדה"
+SIMILAR_B = "מיצוי הליכים הוא תנאי סף להגשת הערר לוועדה"
+DIFFERENT = "מתחם שיקול הדעת התכנוני של הוועדה המקומית רחב"
+
+
+def test_skip_below_dedup_distance():
+    # cosine ≥ 0.93 (dist ≤ 0.07) → skip, regardless of wording
+    assert hq.dedup_action(0.03, DIFFERENT, SIMILAR_A, DEDUP_D, BAND_D) == "skip"
+    assert hq.dedup_action(0.05, SIMILAR_A, SIMILAR_B, DEDUP_D, BAND_D) == "skip"
+
+
+def test_flag_in_lexical_tail():
+    # in the 0.07–0.17 band AND lexically near → flag (not skip, not keep)
+    assert hq.dedup_action(0.12, SIMILAR_A, SIMILAR_B, DEDUP_D, BAND_D) == "flag"
+
+
+def test_keep_in_tail_when_not_lexically_similar():
+    # in the band but lexically distinct → keep (don't flag a different rule)
+    assert hq.dedup_action(0.12, DIFFERENT, SIMILAR_A, DEDUP_D, BAND_D) == "keep"
+
+
+def test_over_merge_guard_distinct_rule_kept():
+    """Beyond the band, even a lexically-similar rule is KEPT — and because the
+    decision is pairwise (one neighbor, incoming-only drop), a chain A~B~C with
+    A,C distinct never collapses to a single row (#82.6)."""
+    assert hq.dedup_action(0.30, SIMILAR_A, SIMILAR_B, DEDUP_D, BAND_D) == "keep"
+    assert hq.dedup_action(0.50, DIFFERENT, SIMILAR_A, DEDUP_D, BAND_D) == "keep"
+
+
+def test_boundary_exactly_at_band_edge():
+    # dist == band_distance is still within the tail (≤), lexical → flag
+    assert hq.dedup_action(BAND_D, SIMILAR_A, SIMILAR_B, DEDUP_D, BAND_D) == "flag"
+    # just past the band → keep
+    assert hq.dedup_action(BAND_D + 0.001, SIMILAR_A, SIMILAR_B, DEDUP_D, BAND_D) == "keep"