import json, sys
proj='/Users/bot1/Volumes/root_for_ai/AI工作区/良渚_IP授权PDF直接修改_20260611_1117'
data=json.load(open(f'{proj}/work/ocr_before.json',encoding='utf-8'))
for p in data['pages']:
    if p['page'] in [4,5,6,7,8,9,13,14,15,17,20,21,22,23,25,28,29,30,35,36]:
        print('\nPAGE',p['page'], 'size', p['width'], p['height'])
        items=sorted(p['items'], key=lambda it:(-it['y'], it['x']))
        for it in items:
            print(f"x={it['x']:.3f} y={it['y']:.3f} w={it['w']:.3f} h={it['h']:.3f} conf={it['confidence']:.2f} | {it['text']}")
