Files
legal-ai/mcp-server/tests/test_halacha_quality.py
Chaim fb60dca796 feat(halacha): over-extraction consolidation — fold facets via claude_session (#81.5)
After a precedent finishes extracting, a claude_session pass folds facets of the
SAME legal question (below #82's dedup cosine — the שפר 14-vs-4 / 403-17→89
granularity gap) into one canonical; the rest are marked 'rejected' (reversible:
out of the active corpus AND the review queue, but recoverable). FOLD-ONLY —
never merges distinct legal questions, never invents.

- Engine: claude_session-as-judge (local CLI, zero cost), 'high' effort — folding
  needs careful judgment. One pass per precedent, runs in _extract_impl once all
  chunks are done (the prompt dedups within a chunk; this catches across chunks).
- Pure, unit-tested helpers in halacha_quality: CONSOLIDATE_SYSTEM,
  build_consolidation_prompt, parse_fold_groups (fails SAFE → [] on any malformed
  shape; drops <2-member groups; coerces/dedups indices).
- halacha_extractor._consolidate_precedent picks the canonical per group
  (approved>pending, higher confidence, quote_verified, longer) and rejects the
  rest via the existing update_halachot_batch (#84). Never rejects a canonical.
  Fails OPEN on any error (no CLI / parse fail → 0 folds, data untouched).
- config: HALACHA_CONSOLIDATE_ENABLED/MODEL/EFFORT.

Verified: suite 176 passed (10 new); integration vs dev DB — a 2-facet group
folds to 1 canonical + 1 rejected (tagged), distinct rules untouched, claude
error → 0 folds (fail-open).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-03 16:26:44 +00:00

184 lines
7.4 KiB
Python

from __future__ import annotations
import pytest
from legal_mcp.services import halacha_quality as hq
# ── non-decision / obiter ──
@pytest.mark.parametrize("text", [
"איני רואה לקבוע מסמרות בשאלה זו",
"אין צורך להכריע בטענה זו",
"למעלה מן הצורך נעיר כי",
"הערה זו ניתנת אגב אורחא",
])
def test_detect_non_decision_hits(text):
assert hq.detect_non_decision(text) is not None
@pytest.mark.parametrize("text", [
"בית המשפט קבע כי ההיתר בטל",
"ועדת הערר מוסמכת לדון בטענת סטייה מתכנית",
"",
])
def test_detect_non_decision_misses(text):
assert hq.detect_non_decision(text) is None
def test_non_decision_scans_all_fields():
# marker sits in the quote, not the abstracted rule
assert hq.detect_non_decision("כלל כללי", "", "וכאן אין צורך להכריע") is not None
# ── truncated quote ──
def test_truncated_dangling_letter():
assert hq.is_quote_truncated("ראוי כי תהיה השפעה על ה") is True
def test_truncated_empty():
assert hq.is_quote_truncated(" ") is True
@pytest.mark.parametrize("quote", [
"ועדת הערר היא הגוף המקצועי האמון על בחינת ההיבטים התכנוניים.",
"אין לועדה סמכות לסטות מתקנות התכנון והבניה", # no period, but full word
"ההיתר תואם את התכנית החלה על האיזור",
])
def test_not_truncated_complete_clauses(quote):
assert hq.is_quote_truncated(quote) is False
# ── thin restatement ──
def test_thin_restatement_near_copy():
quote = "ביטול היתר מחייב טעמים כבדי משקל של אינטרס ציבורי"
rule = "ביטול היתר מחייב טעמים כבדי משקל של אינטרס ציבורי"
assert hq.is_thin_restatement(rule, quote) is True
def test_not_thin_when_abstracted():
quote = "אין חולק כי אין לועדה סמכות לסטות מתקנות"
rule = ("ועדה מקומית לתכנון ובניה אינה מוסמכת לסטות מהוראות תקנות התכנון "
"והבניה, ובכלל זה מהוראות התוספת השנייה, ואין בידה ליתן היתר הסוטה מהן.")
assert hq.is_thin_restatement(rule, quote) is False
def test_thin_handles_empty():
assert hq.is_thin_restatement("", "something") is False
# ── aggregate flags + auto-approve gate semantics ──
def test_clean_halacha_no_flags():
rule = ("ועדת הערר מוסמכת לדון בערר על החלטה ליתן היתר בנייה גם כאשר נטען "
"כי ההיתר סוטה מתכנית, בהתאם למגמת תיקון 43 לחוק.")
quote = ("פרשנות מרחיבה המאפשרת הגשת ערר גם במקרה של מתן היתר כאשר נטען כי "
"ההיתר סוטה מתכנית הולמת את מגמת המחוקק בתיקון 43.")
assert hq.compute_quality_flags(rule, quote, "", quote_verified=True) == []
def test_flags_accumulate():
flags = hq.compute_quality_flags(
"כלל אגב אורחא על ה", "כלל אגב אורחא על ה",
quote_verified=False,
)
assert hq.FLAG_NON_DECISION in flags
assert hq.FLAG_TRUNCATED_QUOTE in flags
assert hq.FLAG_QUOTE_UNVERIFIED in flags
def test_normalize_text_quote_variants():
assert hq.normalize_text('עע"מ 317/10') == hq.normalize_text("עע״מ 317/10")
# ── #81.3 NLI entailment — pure prompt + parser ──
def test_build_nli_prompt_contains_pairs():
items = [
{"rule_statement": "כלל אלף", "supporting_quote": "ציטוט אלף"},
{"rule_statement": "כלל בית", "supporting_quote": "ציטוט בית"},
]
p = hq.build_nli_prompt(items)
assert "כלל אלף" in p and "ציטוט בית" in p
assert "זוג 1" in p and "זוג 2" in p
@pytest.mark.parametrize("raw,n,expected", [
(["entailed", "neutral"], 2, ["entailed", "neutral"]),
(["ENTAILED", "Contradiction"], 2, ["entailed", "contradiction"]), # case-insensitive
([{"verdict": "neutral"}, {"verdict": "entailed"}], 2, ["neutral", "entailed"]), # dict shape
(["entailed"], 2, ["entailed", "entailed"]), # length mismatch -> fail-open
(None, 2, ["entailed", "entailed"]), # non-list -> fail-open
(["bananas", "neutral"], 2, ["entailed", "neutral"]), # unknown label -> entailed
])
def test_parse_nli_verdicts(raw, n, expected):
assert hq.parse_nli_verdicts(raw, n) == expected
# ── _nli_check (async, via claude_session) — fail-open + verdict mapping ──
def test_nli_check_fail_open(monkeypatch):
import asyncio
from legal_mcp.services import halacha_extractor as he
async def boom(*a, **k):
raise RuntimeError("no claude CLI here")
monkeypatch.setattr(he.claude_session, "query_json", boom)
items = [{"rule_statement": "a", "supporting_quote": "b"}]
assert asyncio.run(he._nli_check(items)) == ["entailed"] # never blocks
def test_nli_check_maps_verdicts(monkeypatch):
import asyncio
from legal_mcp.services import halacha_extractor as he
async def fake(*a, **k):
return ["entailed", "neutral"]
monkeypatch.setattr(he.claude_session, "query_json", fake)
items = [{"rule_statement": "a", "supporting_quote": "b"},
{"rule_statement": "c", "supporting_quote": "d"}]
assert asyncio.run(he._nli_check(items)) == ["entailed", "neutral"]
def test_nli_check_empty():
import asyncio
from legal_mcp.services import halacha_extractor as he
assert asyncio.run(he._nli_check([])) == []
# ── #81.5 consolidation — pure prompt + fold-group parser ──
def test_build_consolidation_prompt():
items = [
{"halacha_index": 3, "rule_statement": "כלל גימל", "reasoning_summary": "כי"},
{"halacha_index": 7, "rule_statement": "כלל זין", "reasoning_summary": ""},
]
p = hq.build_consolidation_prompt(items)
assert "[3] כלל גימל" in p and "[7] כלל זין" in p and "היגיון: כי" in p
@pytest.mark.parametrize("raw,expected", [
([[2, 5, 9], [14, 18]], [[2, 5, 9], [14, 18]]),
([[2, 5], [7]], [[2, 5]]), # singleton group dropped
([["2", "5"]], [[2, 5]]), # string ints coerced
([[2, 2, 5]], [[2, 5]]), # dedup within group
([], []), # nothing to fold
("garbage", []), # non-list -> safe
(None, []), # None -> safe
([[1, "x"], [3, 4]], [[3, 4]]), # drop group that falls below 2 valid
])
def test_parse_fold_groups(raw, expected):
assert hq.parse_fold_groups(raw) == expected
def test_consolidation_priority_prefers_approved_then_confidence():
from legal_mcp.services import halacha_extractor as he
approved = {"id": "a", "review_status": "approved", "confidence": 0.7,
"quote_verified": True, "rule_statement": "x"}
pending_hi = {"id": "b", "review_status": "pending_review", "confidence": 0.95,
"quote_verified": True, "rule_statement": "x"}
# approved sorts before higher-confidence pending → kept as canonical
assert min([approved, pending_hi], key=he._consolidation_priority)["id"] == "a"