#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""cek_caption_car10_05okt.py — cek caption album 5 Okt 08:30 IG SS: CJK, hashtag, CTA, link."""
import json
import re
import sys

WS = "/home/mantapman/.hermes/profiles/atun/workspace"
sys.path.insert(0, WS)

docs = json.load(open("/tmp/ss_all_20261005.json", encoding="utf-8"))
IG_SS = "6a38ac33c5ff4ce3a3231232"
CJK = re.compile(r"[\u3000-\u30ff\u3400-\u4dbf\u4e00-\u9fff\uf900-\ufaff\uac00-\ud7af]")
CTA = re.compile(r"komentar|diskusi|pendapat|tulis di|share|pengalaman|menurut (kamu|anda)|setuju", re.I)

for d in docs:
    if d.get("accountId") != IG_SS:
        continue
    if (d.get("scheduleAt") or "")[:16] != "2026-10-05T01:30":
        continue
    desc = d.get("description") or ""
    tags = re.findall(r"#(\w+)", desc)
    print("id        :", d.get("_id"))
    print("status    :", d.get("status"))
    print("panjang   :", len(desc))
    print("CJK       :", CJK.findall(desc) or "0 (bersih)")
    print("hashtag   :", tags, "-> huruf besar:", [t for t in tags if t != t.lower()] or "0")
    print("jumlah hashtag:", len(tags))
    print("mention @ :", re.findall(r"@\w+", desc) or "0")
    print("CTA ekor  :", "ADA" if CTA.search(desc[-160:]) else "TIDAK ADA")
    print("link      :", re.findall(r"https?://\S+", desc))
    print("replies   :", len(d.get("replies") or []))
    print("media     :", len(d.get("medias") or []))