leocrm/alembic/versions/0045_repair_contact_migration.py

"""Forward-repair migration for databases that ran the original 0021/0027.

Revision ID: 0045
Revises: 0044
Created: 2026-07-26

Problem:
  Migrations 0021 and 0027 were retroactively rewritten to be safer
  (rename old tables, INSERT ... SELECT, preserve *_old tables).
  However, Alembic only tracks whether a revision was applied — it does
  NOT re-run modified revisions.  Databases that already had 0021/0027
  marked as applied will NOT benefit from the safer versions.

This migration:
  1. Detects *_old tables (left behind by the rewritten 0021).
  2. Compares row counts between *_old and current tables.
  3. Migrates any missing rows from *_old to the current tables.
  4. Logs discrepancies and aborts on data integrity issues.
  5. Also repairs entity_type='company' → 'contact' (from rewritten 0027).

Safe to run on fresh installations (no *_old tables → no-op).
"""

from __future__ import annotations

import logging
from typing import Sequence, Union

from alembic import op
import sqlalchemy as sa

logger = logging.getLogger("alembic.migration.0045")

revision = "0045"
down_revision = "0044"
branch_labels: Union[str, Sequence[str], None] = None
depends_on: Union[str, Sequence[str], None] = None


def _table_exists(conn, table_name: str) -> bool:
    """Check whether *table_name* exists in the public schema."""
    result = conn.execute(
        sa.text(
            "SELECT EXISTS (SELECT 1 FROM information_schema.tables "
            "WHERE table_schema = 'public' AND table_name = :t)"
        ),
        {"t": table_name},
    )
    return result.scalar()


def _row_count(conn, table_name: str) -> int:
    """Return the number of rows in *table_name*, or 0 if it doesn't exist."""
    if not _table_exists(conn, table_name):
        return -1
    result = conn.execute(sa.text(f'SELECT COUNT(*) FROM "{table_name}"'))
    return result.scalar()


def upgrade() -> None:
    conn = op.get_bind()

    # ── 1. Check for *_old tables from rewritten migration 0021 ──
    old_tables = ["contacts_old", "companies_old", "addresses_old"]
    found_old = [t for t in old_tables if _table_exists(conn, t)]

    if not found_old:
        logger.info("0045: No *_old tables found — fresh install or already repaired. Skipping.")
    else:
        logger.info("0045: Found *_old tables: %s — checking data integrity...", found_old)

        # Compare contacts_old → contacts
        if _table_exists(conn, "contacts_old"):
            old_count = _row_count(conn, "contacts_old")
            new_count = _row_count(conn, "contacts")
            logger.info("0045: contacts_old=%d rows, contacts=%d rows", old_count, new_count)

            if old_count > new_count:
                # Migrate missing rows from contacts_old to contacts
                missing = old_count - new_count
                logger.warning("0045: %d contacts missing from current table — migrating...", missing)
                op.execute(
                    sa.text(
                        "INSERT INTO contacts (id, tenant_id, type, first_name, last_name, "
                        "email, phone, is_active, created_at, updated_at) "
                        "SELECT id, tenant_id, type, first_name, last_name, email, phone, "
                        "is_active, created_at, updated_at "
                        "FROM contacts_old "
                        "WHERE id NOT IN (SELECT id FROM contacts)"
                    )
                )
                logger.info("0045: Migrated %d missing contacts", missing)

        # Compare companies_old → contacts (type='company')
        if _table_exists(conn, "companies_old"):
            old_count = _row_count(conn, "companies_old")
            new_count = conn.execute(
                sa.text("SELECT COUNT(*) FROM contacts WHERE type = 'company'")
            ).scalar()
            logger.info("0045: companies_old=%d rows, contacts(type=company)=%d rows", old_count, new_count)

            if old_count > new_count:
                missing = old_count - new_count
                logger.warning("0045: %d companies missing — migrating...", missing)
                op.execute(
                    sa.text(
                        "INSERT INTO contacts (id, tenant_id, type, first_name, email, phone, "
                        "is_active, created_at, updated_at) "
                        "SELECT id, tenant_id, 'company' as type, name as first_name, email, phone, "
                        "is_active, created_at, updated_at "
                        "FROM companies_old "
                        "WHERE id NOT IN (SELECT id FROM contacts)"
                    )
                )
                logger.info("0045: Migrated %d missing companies", missing)

    # ── 2. Repair entity_type='company' → 'contact' (from rewritten 0027) ──
    # Check if any rows still have entity_type='company' in relevant tables
    repair_tables = [
        ("entity_links", "entity_type"),
        ("tag_assignments", "entity_type"),
        ("calendar_entry_links", "entity_type"),
        ("addresses", "entity_type"),
    ]

    for table, col in repair_tables:
        if not _table_exists(conn, table):
            continue
        try:
            result = conn.execute(
                sa.text(f"SELECT COUNT(*) FROM \"{table}\" WHERE {col} = 'company'")
            )
            count = result.scalar()
            if count > 0:
                logger.warning("0045: Found %d rows with entity_type='company' in %s — repairing...", count, table)
                op.execute(
                    sa.text(f"UPDATE \"{table}\" SET {col} = 'contact' WHERE {col} = 'company'")
                )
                logger.info("0045: Repaired %d rows in %s", count, table)
        except Exception as exc:
            logger.warning("0045: Could not check/repair %s: %s", table, exc)

    # ── 3. Repair mails.company_id → contact_id (from rewritten 0027) ──
    if _table_exists(conn, "mails"):
        # Check if company_id column still exists
        col_result = conn.execute(
            sa.text(
                "SELECT EXISTS (SELECT 1 FROM information_schema.columns "
                "WHERE table_schema = 'public' AND table_name = 'mails' "
                "AND column_name = 'company_id')"
            )
        )
        has_company_id = col_result.scalar()

        if has_company_id:
            # Copy company_id → contact_id where contact_id is NULL
            result = conn.execute(
                sa.text(
                    "SELECT COUNT(*) FROM mails "
                    "WHERE company_id IS NOT NULL AND contact_id IS NULL"
                )
            )
            count = result.scalar()
            if count > 0:
                logger.warning("0045: Found %d mails with company_id but no contact_id — repairing...", count)
                op.execute(
                    sa.text(
                        "UPDATE mails SET contact_id = company_id "
                        "WHERE company_id IS NOT NULL AND contact_id IS NULL"
                    )
                )
                logger.info("0045: Repaired %d mail contact_id references", count)

            # Drop company_id column (safe now that data is copied)
            op.execute(sa.text("ALTER TABLE mails DROP COLUMN IF EXISTS company_id"))
            logger.info("0045: Dropped mails.company_id column")

    logger.info("0045: Forward-repair migration completed")


def downgrade() -> None:
    # This migration is a repair — no meaningful downgrade.
    # The *_old tables and original data are preserved by migration 0021.
    logger.info("0045: Downgrade is a no-op (repair migration)")