Files
legal-ai/mcp-server/src/legal_mcp/tools/search.py
Chaim 26d09d648f Practice area separation: multi-tenant axis across DB, RAG, and UI
Adds two orthogonal columns — practice_area (top-level legal domain:
appeals_committee / national_insurance / labor_law) and appeal_subtype
(building_permit / betterment_levy / compensation_197) — denormalized
into cases, documents, document_chunks, decisions, and style_corpus so
vector searches can filter without JOINs.

Why: the system handles two unrelated sub-domains under the same
appeals committee (1xxx building permits and 8xxx/9xxx betterment/197),
with different rules and writing style. Without a separation axis,
search_similar() and the block-writer's precedent lookup were free to
surface betterment-levy paragraphs while drafting a building-permit
decision — a real risk of cross-domain contamination. The same axis
also lets future domains (national insurance, labor law) coexist
without separate schemas.

Schema (V4 migration in db.py):
- ALTER ... ADD COLUMN IF NOT EXISTS on all five tables + composite
  indexes (practice_area first).
- Idempotent backfill: case_number ~ '^1' → building_permit, '^8' →
  betterment_levy, '^9' → compensation_197; propagated to documents,
  chunks, and decisions via case_id; training-corpus rows (case_id NULL)
  default to appeals_committee.

Code:
- New services/practice_area.py with derive_subtype, validate, and
  is_override + enum constants.
- db.create_case / create_document / store_chunks / create_decision
  inherit practice_area from the parent case (or take an explicit
  override for the case_id=None training corpus).
- db.search_similar and search_similar_paragraphs accept practice_area
  + appeal_subtype filters using the denormalized columns.
- tools/search.py auto-resolves the filter from case_number when given.
- block_writer._build_precedents_context now passes the active case's
  practice_area to search_similar_paragraphs — closes the contamination
  hole for the discussion-block precedent fetch.
- tools/cases.case_create auto-derives subtype from case_number; an
  explicit override that disagrees writes a case_subtype_override entry
  to audit_log so we can spot bad classifications later.
- tools/documents.document_upload_training tags new training material
  with practice_area + subtype end-to-end (corpus, document, chunks).

UI (web/static/index.html + web/app.py):
- New-case wizard gets a practice_area dropdown (others disabled until
  national_insurance / labor_law arrive) and an appeal_subtype dropdown
  with JS auto-fill from the case-number prefix; manual edits stick.
- Case header shows a blue badge with practice_area · subtype.
- CaseCreateRequest plumbs both fields through to cases_tools.case_create.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-11 16:36:48 +00:00

170 lines
5.5 KiB
Python

"""MCP tools for RAG search over legal documents and decisions."""
from __future__ import annotations
import json
import logging
from uuid import UUID
from legal_mcp.services import db, embeddings
logger = logging.getLogger(__name__)
async def search_decisions(
query: str,
limit: int = 10,
section_type: str = "",
practice_area: str = "",
appeal_subtype: str = "",
case_number: str = "",
) -> str:
"""חיפוש סמנטי בהחלטות קודמות ובמסמכים — מסונן לפי תחום משפטי.
Args:
query: שאילתת חיפוש בעברית
limit: מספר תוצאות מקסימלי
section_type: סינון לפי סוג סעיף (facts, legal_analysis, ...)
practice_area: תחום משפטי לסינון (appeals_committee/national_insurance/...)
appeal_subtype: סוג ערר לסינון (building_permit/betterment_levy/compensation_197)
case_number: אם סופק, ה-practice_area/subtype יוסקו אוטומטית מהתיק
"""
# Auto-resolve practice_area from case_number if available
if case_number and not practice_area:
case = await db.get_case_by_number(case_number)
if case:
practice_area = case.get("practice_area") or ""
appeal_subtype = appeal_subtype or (case.get("appeal_subtype") or "")
if not practice_area:
logger.warning(
"search_decisions called without practice_area filter — "
"results may mix legal domains"
)
query_emb = await embeddings.embed_query(query)
results = await db.search_similar(
query_embedding=query_emb,
limit=limit,
section_type=section_type or None,
practice_area=practice_area or None,
appeal_subtype=appeal_subtype or None,
)
if not results:
return "לא נמצאו תוצאות."
formatted = []
for r in results:
formatted.append({
"score": round(float(r["score"]), 4),
"case_number": r["case_number"],
"document": r["document_title"],
"section": r["section_type"],
"page": r["page_number"],
"content": r["content"],
})
return json.dumps(formatted, ensure_ascii=False, indent=2)
async def search_case_documents(
case_number: str,
query: str,
limit: int = 10,
) -> str:
"""חיפוש סמנטי בתוך מסמכי תיק ספציפי.
Args:
case_number: מספר תיק הערר
query: שאילתת חיפוש
limit: מספר תוצאות מקסימלי
"""
case = await db.get_case_by_number(case_number)
if not case:
return f"תיק {case_number} לא נמצא."
query_emb = await embeddings.embed_query(query)
# Restricted to case_id — practice_area filter would be redundant.
results = await db.search_similar(
query_embedding=query_emb,
limit=limit,
case_id=UUID(case["id"]),
)
if not results:
return f"לא נמצאו תוצאות בתיק {case_number}."
formatted = []
for r in results:
formatted.append({
"score": round(float(r["score"]), 4),
"document": r["document_title"],
"section": r["section_type"],
"page": r["page_number"],
"content": r["content"],
})
return json.dumps(formatted, ensure_ascii=False, indent=2)
async def find_similar_cases(
description: str,
limit: int = 5,
practice_area: str = "",
appeal_subtype: str = "",
case_number: str = "",
) -> str:
"""מציאת תיקים דומים על בסיס תיאור — מסונן לפי תחום משפטי.
Args:
description: תיאור התיק או הנושא
limit: מספר תוצאות מקסימלי
practice_area: תחום משפטי לסינון
appeal_subtype: סוג ערר לסינון
case_number: אם סופק, ה-practice_area/subtype יוסקו אוטומטית מהתיק
"""
if case_number and not practice_area:
case = await db.get_case_by_number(case_number)
if case:
practice_area = case.get("practice_area") or ""
appeal_subtype = appeal_subtype or (case.get("appeal_subtype") or "")
if not practice_area:
logger.warning(
"find_similar_cases called without practice_area filter — "
"results may mix legal domains"
)
query_emb = await embeddings.embed_query(description)
results = await db.search_similar(
query_embedding=query_emb,
limit=limit * 3, # Get more to deduplicate by case
practice_area=practice_area or None,
appeal_subtype=appeal_subtype or None,
)
if not results:
return "לא נמצאו תיקים דומים."
# Deduplicate by case_number, keep best score per case
seen_cases = {}
for r in results:
cn = r["case_number"]
if cn not in seen_cases or r["score"] > seen_cases[cn]["score"]:
seen_cases[cn] = r
# Sort by score and limit
top_cases = sorted(seen_cases.values(), key=lambda x: x["score"], reverse=True)[:limit]
formatted = []
for r in top_cases:
formatted.append({
"score": round(float(r["score"]), 4),
"case_number": r["case_number"],
"document": r["document_title"],
"relevant_section": r["content"][:500],
})
return json.dumps(formatted, ensure_ascii=False, indent=2)