settled-reach/tooling/domains/atlas/verify.py

#!/usr/bin/env python3
"""Verify an atlas proposal JSON against integrity checks.

Usage:
    reach atlas verify docs/atlas/proposals/GJ273.json
    reach atlas verify docs/atlas/proposals/*.json
"""
import json
from pathlib import Path

from tooling.core import console


def verify_proposal(path: Path) -> list[str]:
    """Return list of error strings. Empty list = pass."""
    with open(path) as f:
        p = json.load(f)

    errors = []
    bodies = p.get("bodies", [])
    stations = p.get("stations", [])

    # 1. Inhabited bodies must have names
    for b in bodies:
        if b.get("inhabited") and not b.get("proper_name"):
            errors.append(f"Inhabited body {b['body_id']} has no proper_name")

    # 2. Uninhabited bodies MAY have names (lore reasons) — no check needed

    # 3. All stations must have names
    for s in stations:
        if not s.get("proper_name"):
            errors.append(f"Station {s['station_id']} has no proper_name")

    # 4. Body count minimum by star type
    planets = [b for b in bodies if b["body_type"] == "planet"]
    star_type = p.get("spectral_class", "M")
    if star_type.startswith(("G", "F")):
        min_planets = 8
    elif star_type.startswith("K"):
        min_planets = 7
    else:
        min_planets = 6
    if len(planets) < min_planets:
        errors.append(
            f"Only {len(planets)} planets, need {min_planets}+ "
            f"for {star_type} star"
        )

    # 5. Required structures
    if not any(b["body_type"] == "oort_cloud" for b in bodies):
        errors.append("No oort cloud")

    horizons = [s for s in stations if s["station_type"] == "horizon"]
    if not horizons:
        errors.append("No horizon station")
    elif not any(s.get("has_gate_infrastructure") for s in horizons):
        errors.append("Horizon station missing has_gate_infrastructure: true")

    if not any(b["body_type"] == "asteroid_belt" for b in bodies):
        errors.append("No asteroid belt (add one unless wiki contradicts)")

    # 6. Orbit consistency — top-level bodies
    top_level = [b for b in bodies if not b.get("parent_body_id")]
    orbits = [b["orbit_index"] for b in top_level]
    if orbits != sorted(orbits):
        errors.append(f"Top-level orbit_index not monotonic: {orbits}")
    if len(orbits) != len(set(orbits)):
        errors.append(f"Duplicate orbit_index in top-level: {orbits}")

    # 7. Moon orbit consistency
    parents: dict[str, list[int]] = {}
    for b in bodies:
        pid = b.get("parent_body_id")
        if pid:
            parents.setdefault(pid, []).append(b["orbit_index"])
    for pid, idxs in parents.items():
        if idxs != sorted(idxs):
            errors.append(f"Moon orbits under {pid} not monotonic: {idxs}")
        if len(idxs) != len(set(idxs)):
            errors.append(f"Duplicate moon orbit_index under {pid}: {idxs}")

    # 8. Parent references valid
    body_ids = {b["body_id"] for b in bodies}
    for b in bodies:
        pid = b.get("parent_body_id")
        if pid and pid not in body_ids:
            errors.append(
                f"Body {b['body_id']} references missing parent {pid}"
            )
    for s in stations:
        oid = s.get("orbits_body_id")
        if oid and oid not in body_ids:
            errors.append(
                f"Station {s['station_id']} references missing body {oid}"
            )

    # 9. star_type vs spectral_class consistency
    star_type_field = p.get("star_type", "")
    spectral = p.get("spectral_class", "")
    if star_type_field and spectral and star_type_field not in ("binary", "unusual"):
        # Strip dwarf/subdwarf prefixes (d, sd) to get actual class letter
        s = spectral.lstrip("sd").upper()
        spectral_letter = s[0] if s else ""
        if spectral_letter and spectral_letter.isalpha():
            if star_type_field[0].upper() != spectral_letter:
                errors.append(
                    f"star_type '{star_type_field}' conflicts with "
                    f"spectral_class '{spectral}'"
                )

    return errors


# Sol is hand-authored with different rules — named uninhabited bodies and so
# on — so verifying it against the generated-proposal schema would report
# failures for things that are deliberate.
EXCLUDE = {"GJ0.json", "GJ1221.json"}


def verify_all(proposals: list[Path]) -> int:
    """Verify proposal JSONs. Returns the total error count across all files."""
    total = 0
    for path in proposals:
        if path.name in EXCLUDE:
            continue
        if not path.exists():
            console.event(f"SKIP — {path} not found", level="warn")
            continue
        errors = verify_proposal(path)
        if errors:
            console.event(f"FAIL — {path.name}", level="error")
            for error in errors:
                console.event(f"  - {error}", level="error")
            total += len(errors)
        else:
            console.event(f"PASS — {path.name}")
    return total