legal-ai/scripts/fix_137_committee_case_number.py

#!/usr/bin/env python3
"""One-off data fix for TaskMaster #137 — committee precedent whose identifier
field was polluted by a full citation.

Background: ``missing_precedent_upload`` (committee branch) fell back to the raw
``citation`` when the form left ``case_number`` blank, so a citation like
``ערר (ת"א 85074-04-25) רפאל לוי ואח' נ' הוועדה … - חולון`` landed as the
identifier / display name (INV-ID2 violation). The code fix (PR for #137)
prevents recurrence; this script corrects the one already-stored row.

Scope: a SINGLE internal_committee row (1bf0bae0). Verified the only
internal_committee row where ``_canonical_case_number(case_number) != case_number``
(the legacy 138 "polluted" matches are external/cited_only PREFIXED numbers —
the X1 §5 external-identifier item, deliberately out of scope here). The row has
``document_id = NULL`` (no file/storage key to rename) and 0 incoming citations,
so only three columns change: ``case_number``, ``case_name``, ``citation_formatted``.

Idempotent: if the row already carries the canonical number it is a no-op.
Dry-run by default; pass ``--apply`` to write.

Run (local, reads ~/.env for POSTGRES_URL):
    HOME=/home/chaim PYTHONPATH=mcp-server/src python scripts/fix_137_committee_case_number.py [--apply]
"""

from __future__ import annotations

import asyncio
import sys

import asyncpg

from legal_mcp import config

# The verified target row + its corrected values (see module docstring).
CASE_LAW_ID = "1bf0bae0-1cb7-4110-ba1b-b956e42b0355"
BAD_CASE_NUMBER = "85074/0425"
NEW_CASE_NUMBER = "85074-04-25"
NEW_CASE_NAME = "רפאל לוי ואח' נ' הוועדה המקומית לתכנון ובניה - חולון"


async def main(apply: bool) -> int:
    conn = await asyncpg.connect(config.POSTGRES_URL)
    try:
        row = await conn.fetchrow(
            "SELECT case_number, case_name, citation_formatted, proceeding_type "
            "FROM case_law WHERE id = $1",
            CASE_LAW_ID,
        )
        if row is None:
            print(f"row {CASE_LAW_ID} not found — nothing to do")
            return 0

        if row["case_number"] == NEW_CASE_NUMBER:
            print(f"already canonical (case_number={NEW_CASE_NUMBER!r}) — no-op")
            return 0

        if row["case_number"] != BAD_CASE_NUMBER:
            print(
                f"UNEXPECTED current case_number={row['case_number']!r} "
                f"(expected {BAD_CASE_NUMBER!r}) — refusing to guess; inspect manually"
            )
            return 1

        # Collision guard: the (case_number, proceeding_type) partial-unique key.
        clash = await conn.fetchval(
            "SELECT id FROM case_law WHERE source_kind='internal_committee' "
            "AND case_number = $1 AND proceeding_type = $2 AND id <> $3",
            NEW_CASE_NUMBER, row["proceeding_type"], CASE_LAW_ID,
        )
        if clash:
            print(f"COLLISION: {NEW_CASE_NUMBER!r}/{row['proceeding_type']!r} "
                  f"already exists as {clash} — aborting")
            return 1

        new_citation = (row["citation_formatted"] or "").replace(
            BAD_CASE_NUMBER, NEW_CASE_NUMBER)

        print("WILL UPDATE:")
        print(f"  case_number: {row['case_number']!r} -> {NEW_CASE_NUMBER!r}")
        print(f"  case_name:   {row['case_name']!r}\n           -> {NEW_CASE_NAME!r}")
        print(f"  citation_formatted: {row['citation_formatted']!r}\n"
              f"                   -> {new_citation!r}")

        if not apply:
            print("\n(dry-run — pass --apply to write)")
            return 0

        await conn.execute(
            "UPDATE case_law SET case_number = $2, case_name = $3, "
            "citation_formatted = $4 WHERE id = $1",
            CASE_LAW_ID, NEW_CASE_NUMBER, NEW_CASE_NAME, new_citation,
        )
        print("\n✓ updated")
        return 0
    finally:
        await conn.close()


if __name__ == "__main__":
    sys.exit(asyncio.run(main("--apply" in sys.argv)))