settled-reach/tooling/economy-db/economy_import/validators.py

"""Structural-integrity and D-175 coverage validation.

``validate`` is a pre-commit hard blocker; the coverage validators run after
commit (Phase 2 gate, exit 2) — see the entrypoint's main() for the contract.
"""

import sqlite3


def validate(conn: sqlite3.Connection) -> list[str]:
    """Validate structural integrity of imported data.

    Checks FK integrity, chain commodity references, and chain completeness.
    These are hard blockers — broken data must not be committed.

    Coverage validation (commodity/system thresholds) is separate and runs
    after commit via validate_commodity_coverage() and validate_system_coverage().
    """
    errors: list[str] = []

    # FK integrity
    fk_issues = conn.execute("PRAGMA foreign_key_check").fetchall()
    if fk_issues:
        for issue in fk_issues[:10]:
            errors.append(f"FK violation: table={issue[0]} rowid={issue[1]} "
                          f"parent={issue[2]} fkid={issue[3]}")

    # Chain inputs reference valid commodities
    orphan_inputs = conn.execute("""
        SELECT ci.chain_id, ci.input_commodity_id
        FROM chain_inputs ci
        LEFT JOIN commodities c ON ci.input_commodity_id = c.commodity_id
        WHERE c.commodity_id IS NULL
    """).fetchall()
    for chain_id, cid in orphan_inputs:
        errors.append(f"chain_inputs: chain '{chain_id}' references unknown commodity '{cid}'")

    # Chain outputs reference valid commodities
    orphan_outputs = conn.execute("""
        SELECT pc.chain_id, pc.output_commodity_id
        FROM production_chains pc
        LEFT JOIN commodities c ON pc.output_commodity_id = c.commodity_id
        WHERE c.commodity_id IS NULL
    """).fetchall()
    for chain_id, cid in orphan_outputs:
        errors.append(f"production_chains: chain '{chain_id}' outputs unknown commodity '{cid}'")

    # economic_role must be one of the D-194 canonical 10 values
    valid_roles = {
        'manufacturing', 'financial', 'agricultural', 'extraction',
        'service_mixed', 'institutional', 'transit_hub', 'research',
        'military', 'residential',
    }
    bad_roles = conn.execute("""
        SELECT DISTINCT economic_role, COUNT(*) as cnt
        FROM bodies
        WHERE economic_role IS NOT NULL
          AND economic_role NOT IN (
              'manufacturing', 'financial', 'agricultural', 'extraction',
              'service_mixed', 'institutional', 'transit_hub', 'research',
              'military', 'residential'
          )
        GROUP BY economic_role
    """).fetchall()
    for role, cnt in bad_roles:
        errors.append(
            f"bodies.economic_role: non-canonical value '{role}' on {cnt} row(s) — "
            f"valid values: {sorted(valid_roles)}"
        )

    # Chain completeness: every intermediate commodity must have at least one producer
    missing_chains = conn.execute("""
        SELECT c.commodity_id, c.name
        FROM commodities c
        WHERE c.tier = 'intermediate'
          AND c.commodity_id NOT IN (SELECT output_commodity_id FROM production_chains)
        ORDER BY c.commodity_id
    """).fetchall()
    for cid, name in missing_chains:
        errors.append(f"chain completeness: no production chain produces intermediate '{cid}' ({name})")

    return errors


def validate_commodity_coverage(
    conn: sqlite3.Connection, wiki_corps: list[dict], commodity_ids: set[str]
) -> list[str]:
    """3+ corporations per major commodity type (raw + intermediate). D-175."""
    errors: list[str] = []
    major = [
        r[0]
        for r in conn.execute(
            "SELECT commodity_id FROM commodities "
            "WHERE tier IN ('raw', 'intermediate') ORDER BY commodity_id"
        ).fetchall()
    ]

    # Build commodity → corp set from wiki tags filtered to known commodity IDs
    coverage: dict[str, set[str]] = {cid: set() for cid in major}
    for corp in wiki_corps:
        for tag in corp.get("tags", []):
            if tag in coverage:
                coverage[tag].add(corp["corp_id"])

    for cid in major:
        n = len(coverage[cid])
        if n < 3:
            corp_list = sorted(coverage[cid]) if coverage[cid] else ["none"]
            errors.append(
                f"commodity coverage: '{cid}' has {n}/3 corp(s) — {corp_list}"
            )

    return errors


def validate_system_coverage(
    conn: sqlite3.Connection, wiki_corps: list[dict]
) -> list[str]:
    """1+ corporation per inhabited system with population > 100K. D-175.

    Uses wiki_corps headquarters data (not DB corp_presence) so this check
    is accurate in both dry-run and real-run modes.
    """
    covered = {c["system_id"] for c in wiki_corps if c.get("system_id")}
    populated = conn.execute("""
        SELECT se.system_id, ss.proper_name, se.population
        FROM system_economy se
        JOIN star_systems ss ON se.system_id = ss.system_id
        WHERE se.population > 100000
        ORDER BY se.system_id
    """).fetchall()

    return [
        f"system coverage: no corp presence in '{sid}' ({name}, pop={pop:,})"
        for sid, name, pop in populated
        if sid not in covered
    ]