"""孕妈悄悄话置顶笔记的 100 条模拟祝福评论(11-07 用户:「评论里按正常评论放上 100 个」)。 lines.txt 一行一条,「||」后面是楼中楼回复(其他孕妈/宝妈)。人设从 cm.json 的 1000 人里挑:主评论用孕期人设、回复混宝妈; 尽量挑在别的帖子下出现少的人。确定性(固定种子),重跑结果不变;只追加/覆盖 c[TARGET],不动其他帖子。""" import json, random from collections import Counter TARGET = 'wh-20260928' P = 'guangchang/cm.json' d = json.load(open(P, encoding='utf-8')) used = Counter(c[0] for k, L in d['c'].items() if k != TARGET for c in L) rnd = random.Random(20261107) import re def wk(p): m = re.match(r'孕 ?(\d+)', p[2]); return int(m.group(1)) if m else 0 # 内容多是胎动、产检、待产(孕中晚期的事)→ 主评论只用孕 24 周以上的人设,免得「孕 6 周」说「头朝下了」 preg = sorted([i for i, p in enumerate(d['p']) if wk(p) >= 24], key=lambda i: (used[i], i)) anyp = sorted(range(len(d['p'])), key=lambda i: (used[i], i)) mains = preg[:200]; rnd.shuffle(mains) reps = anyp[:300]; rnd.shuffle(reps) lines = [l.strip() for l in open('tools/whcm/lines.txt', encoding='utf-8') if l.strip()] assert len(lines) == 100, len(lines) out, ri = [], 0 for k, l in enumerate(lines): parts = l.split('||') likes = max(3, int(520 * (0.965 ** k)) + rnd.randint(-18, 18)) rr = [] for t in parts[1:]: rr.append([reps[ri], t]); ri += 1 out.append([mains[k], parts[0], likes, rr]) d['c'][TARGET] = out json.dump(d, open(P, 'w', encoding='utf-8'), ensure_ascii=False, separators=(',', ':')) print('ok', len(out), 'replies', ri)