#!/usr/bin/env python3
"""Portable smoke validator: structural and cross-reference checks; not full JSON Schema compliance."""
from pathlib import Path
import json, sys
ROOT = Path(__file__).resolve().parent.parent
errors=[]
for path in sorted((ROOT/'examples/dataspaces').glob('*/snapshot.json')):
    d=json.loads(path.read_text())
    kinds={r['id'] for r in d['kinds']}; ids=[r['id'] for r in d['entities']]
    sources={r['id'] for r in d['sources']}
    if d['schemaVersion']!='0.1': errors.append(f'{path}: schemaVersion')
    if len(ids)!=len(set(ids)): errors.append(f'{path}: duplicate entities')
    for e in d['entities']:
        if e['kindId'] not in kinds: errors.append(f'{path}: undeclared kind {e["kindId"]}')
        for sid in e.get('sourceRefs',[]):
            if sid not in sources: errors.append(f'{path}: missing source {sid}')
    for r in d['relations']:
        for field in ('from','to'):
            if r[field] not in ids: errors.append(f'{path}: dangling {field} {r[field]}')
        for sid in r.get('sourceRefs',[]):
            if sid not in sources: errors.append(f'{path}: missing edge source {sid}')
    manifest=json.loads((path.parent/'manifest.json').read_text())
    if manifest['id']!=d['dataspace']['id']: errors.append(f'{path}: manifest ID mismatch')
    print(f'{d["dataspace"]["id"]}: {len(ids)} entities, {len(d["relations"])} relations, {len(kinds)} kinds')
if errors:
    print('\n'.join(errors),file=sys.stderr);sys.exit(1)
print('Smoke validation OK (full JSON Schema and claims pending).')