HM_project_Viewer_Board/scripts/validate_wehago_anchor_projection.py

from __future__ import annotations

import argparse
import sqlite3
import sys
from pathlib import Path

sys.path.insert(0, str(Path(__file__).resolve().parents[1]))

from runtime_config import DB_PATH


DEFAULT_STATUSES = (
    "voucher_matched",
    "voucher_unmatched",
    "voucher_recheck",
    "voucher_excepted",
)


def latest_signature(conn: sqlite3.Connection, start_year: int, end_year: int) -> str:
    row = conn.execute(
        """
        SELECT signature
        FROM wehago_compare_query_groups
        WHERE start_year = ?
          AND end_year = ?
        GROUP BY signature
        ORDER BY MAX(updated_at) DESC
        LIMIT 1
        """,
        (start_year, end_year),
    ).fetchone()
    return str(row[0] or "") if row else ""


def status_summary(
    conn: sqlite3.Connection,
    start_year: int,
    end_year: int,
    signature: str,
    status_key: str,
) -> dict[str, int | float]:
    row = conn.execute(
        """
        WITH per_group AS (
            SELECT
                g.group_index,
                COUNT(DISTINCT CASE
                    WHEN COALESCE(r.ledger_account_name, '') <> ''
                      OR COALESCE(r.ledger_desc, '') <> ''
                      OR ABS(COALESCE(r.ledger_debit, 0)) >= 0.5
                      OR ABS(COALESCE(r.ledger_credit, 0)) >= 0.5
                    THEN COALESCE(r.ledger_date, '') || '|' || COALESCE(r.voucher_no, '')
                END) AS wehago_keys,
                COUNT(DISTINCT CASE
                    WHEN COALESCE(r.voucher_account_name, '') <> ''
                      OR COALESCE(r.voucher_desc, '') <> ''
                      OR ABS(COALESCE(r.voucher_debit, 0)) >= 0.5
                      OR ABS(COALESCE(r.voucher_credit, 0)) >= 0.5
                    THEN COALESCE(r.draft_no, '')
                END) AS erp_drafts,
                COUNT(*) AS row_count
            FROM wehago_compare_query_groups g
            LEFT JOIN wehago_compare_query_rows r
              ON r.start_year = g.start_year
             AND r.end_year = g.end_year
             AND r.signature = g.signature
             AND r.status_key = g.status_key
             AND r.group_index = g.group_index
            WHERE g.start_year = ?
              AND g.end_year = ?
              AND g.signature = ?
              AND g.status_key = ?
            GROUP BY g.group_index
        )
        SELECT
            COUNT(*) AS groups,
            SUM(CASE WHEN wehago_keys > 1 THEN 1 ELSE 0 END) AS mixed_wehago_groups,
            SUM(CASE WHEN erp_drafts > 1 THEN 1 ELSE 0 END) AS multi_erp_groups,
            COALESCE(MAX(wehago_keys), 0) AS max_wehago_keys,
            COALESCE(MAX(erp_drafts), 0) AS max_erp_drafts,
            COALESCE(MAX(row_count), 0) AS max_rows,
            COALESCE(AVG(row_count), 0) AS avg_rows
        FROM per_group
        """,
        (start_year, end_year, signature, status_key),
    ).fetchone()
    return {
        "groups": int(row["groups"] or 0),
        "mixed_wehago_groups": int(row["mixed_wehago_groups"] or 0),
        "multi_erp_groups": int(row["multi_erp_groups"] or 0),
        "max_wehago_keys": int(row["max_wehago_keys"] or 0),
        "max_erp_drafts": int(row["max_erp_drafts"] or 0),
        "max_rows": int(row["max_rows"] or 0),
        "avg_rows": round(float(row["avg_rows"] or 0), 2),
    }


def sample_mixed_groups(
    conn: sqlite3.Connection,
    start_year: int,
    end_year: int,
    signature: str,
    status_key: str,
    limit: int,
) -> list[dict[str, object]]:
    rows = conn.execute(
        """
        SELECT
            g.group_index,
            g.ledger_date,
            g.voucher_no,
            COUNT(DISTINCT CASE
                WHEN COALESCE(r.ledger_account_name, '') <> ''
                  OR COALESCE(r.ledger_desc, '') <> ''
                  OR ABS(COALESCE(r.ledger_debit, 0)) >= 0.5
                  OR ABS(COALESCE(r.ledger_credit, 0)) >= 0.5
                THEN COALESCE(r.ledger_date, '') || '|' || COALESCE(r.voucher_no, '')
            END) AS wehago_keys,
            COUNT(*) AS row_count
        FROM wehago_compare_query_groups g
        JOIN wehago_compare_query_rows r
          ON r.start_year = g.start_year
         AND r.end_year = g.end_year
         AND r.signature = g.signature
         AND r.status_key = g.status_key
         AND r.group_index = g.group_index
        WHERE g.start_year = ?
          AND g.end_year = ?
          AND g.signature = ?
          AND g.status_key = ?
        GROUP BY g.group_index
        HAVING wehago_keys > 1
        ORDER BY wehago_keys DESC, row_count DESC, g.group_index ASC
        LIMIT ?
        """,
        (start_year, end_year, signature, status_key, limit),
    ).fetchall()
    return [dict(row) for row in rows]


def main() -> None:
    parser = argparse.ArgumentParser(description="Validate WEHAGO voucher projection anchoring.")
    parser.add_argument("--start-year", type=int, default=2025)
    parser.add_argument("--end-year", type=int, default=2025)
    parser.add_argument("--signature", default="")
    parser.add_argument("--sample", type=int, default=5)
    parser.add_argument("--statuses", nargs="*", default=list(DEFAULT_STATUSES))
    args = parser.parse_args()

    conn = sqlite3.connect(DB_PATH)
    conn.row_factory = sqlite3.Row
    signature = args.signature or latest_signature(conn, args.start_year, args.end_year)
    if not signature:
        raise SystemExit("No query projection signature found.")
    print({"db": str(DB_PATH), "signature": signature})
    for status_key in args.statuses:
        summary = status_summary(conn, args.start_year, args.end_year, signature, status_key)
        print({status_key: summary})
        samples = sample_mixed_groups(
            conn,
            args.start_year,
            args.end_year,
            signature,
            status_key,
            args.sample,
        )
        if samples:
            print({"mixed_samples": status_key, "rows": samples})


if __name__ == "__main__":
    main()