1"""
2context_builder.py — строит саммари по чатам и людям из messages.jsonl
3БЕЗ LLM — чистая статистика и топ-фразы.
4 
5Запускать периодически: python3 context_builder.py
6"""
7 
8import json
9import os
10from pathlib import Path
11from collections import defaultdict, Counter
12from datetime import datetime, timezone
13 
14CONTEXT_DIR = Path(__file__).parent / "context"
15MESSAGES_FILE = CONTEXT_DIR / "messages.jsonl"
16CHATS_DIR = CONTEXT_DIR / "chats"
17PEOPLE_DIR = CONTEXT_DIR / "people"
18 
19CHATS_DIR.mkdir(parents=True, exist_ok=True)
20PEOPLE_DIR.mkdir(parents=True, exist_ok=True)
21 
22OWNER_ID = int(os.getenv("TELEGRAM_OWNER_ID", "0"))
23MAX_SAMPLE = 20  # последних сообщений в саммари
24 
25 
26def load_messages() -> list:
27    if not MESSAGES_FILE.exists():
28        return []
29    msgs = []
30    with open(MESSAGES_FILE, encoding="utf-8") as f:
31        for line in f:
32            try:
33                msgs.append(json.loads(line))
34            except Exception:
35                pass
36    return msgs
37 
38 
39def top_words(texts: list, n=10) -> list:
40    words = []
41    for t in texts:
42        for w in t.lower().split():
43            w = w.strip(".,!?;:\"'()[]{}—-")
44            if len(w) > 3 and w not in {"клав", "это", "что", "как", "для", "все", "там", "так", "тут", "нет", "ну"}:
45                words.append(w)
46    return [w for w, _ in Counter(words).most_common(n)]
47 
48 
49def build_chat_md(chat_id, title, chat_type, msgs) -> str:
50    updated = datetime.now(timezone.utc).strftime("%Y-%m-%d %H:%M UTC")
51    senders = Counter(m.get("sender_name", "?") for m in msgs)
52    top_senders = senders.most_common(5)
53    texts = [m.get("text", "") for m in msgs if m.get("text")]
54    keywords = top_words(texts)
55    last = msgs[-MAX_SAMPLE:]
56 
57    lines = [
58        f"# Чат: {title}",
59        f"**ID:** {chat_id} | **Тип:** {chat_type} | **Сообщений в базе:** {len(msgs)}",
60        f"**Обновлено:** {updated}",
61        "",
62        "## Активные участники",
63    ]
64    for name, count in top_senders:
65        lines.append(f"- {name}: {count} сообщ.")
66 
67    lines += ["", "## Ключевые слова", ", ".join(keywords) or "—", "", "## Последние сообщения"]
68    for m in last:
69        d = "→" if m.get("direction") == "out" else "←"
70        name = m.get("sender_name", "?")
71        text = (m.get("text") or "[медиа]")[:100]
72        ts = m.get("ts", "")[:16].replace("T", " ")
73        lines.append(f"[{ts}] {d} {name}: {text}")
74 
75    return "\n".join(lines)
76 
77 
78def build_person_md(user_id, name, msgs_all) -> str:
79    updated = datetime.now(timezone.utc).strftime("%Y-%m-%d %H:%M UTC")
80    chats_seen = Counter(m.get("chat_title", "?") for m in msgs_all)
81    texts = [m.get("text", "") for m in msgs_all if m.get("text")]
82    keywords = top_words(texts)
83    last = msgs_all[-MAX_SAMPLE:]
84 
85    lines = [
86        f"# {name} (user_id: {user_id})",
87        f"**Сообщений:** {len(msgs_all)} | **Обновлено:** {updated}",
88        "",
89        "## Чаты где встречается",
90    ]
91    for chat, count in chats_seen.most_common():
92        lines.append(f"- {chat}: {count} сообщ.")
93 
94    lines += ["", "## Ключевые слова", ", ".join(keywords) or "—", "", "## Последние сообщения"]
95    for m in last:
96        chat = m.get("chat_title", "?")
97        text = (m.get("text") or "[медиа]")[:100]
98        ts = m.get("ts", "")[:16].replace("T", " ")
99        reply = f" ↩ \"{m['reply_to_text'][:40]}\"" if m.get("reply_to_text") else ""
100        lines.append(f"[{ts}][{chat}]: {text}{reply}")
101 
102    return "\n".join(lines)
103 
104 
105def main():
106    print("[context_builder] Загружаю сообщения...")
107    msgs = load_messages()
108    if not msgs:
109        print("[context_builder] Нет сообщений.")
110        return
111    print(f"[context_builder] Загружено {len(msgs)} записей")
112 
113    # Группируем по чатам
114    chats: dict = defaultdict(lambda: {"title": "", "type": "", "msgs": []})
115    for m in msgs:
116        cid = m.get("chat_id")
117        chats[cid]["title"] = m.get("chat_title", str(cid))
118        chats[cid]["type"] = m.get("chat_type", "")
119        chats[cid]["msgs"].append(m)
120 
121    # Группируем по людям (не owner)
122    people: dict = defaultdict(lambda: {"name": "", "msgs": []})
123    for m in msgs:
124        sid = m.get("sender_id")
125        if not sid or sid == OWNER_ID:
126            continue
127        people[sid]["name"] = m.get("sender_name", str(sid))
128        people[sid]["msgs"].append(m)
129 
130    # Саммари по чатам
131    for cid, info in chats.items():
132        if len(info["msgs"]) < 3:
133            continue
134        print(f"[context_builder] Чат: {info['title']}")
135        md = build_chat_md(cid, info["title"], info["type"], info["msgs"])
136        (CHATS_DIR / f"{cid}.md").write_text(md, encoding="utf-8")
137 
138    # Саммари по людям
139    for uid, info in people.items():
140        if len(info["msgs"]) < 2:
141            continue
142        print(f"[context_builder] Человек: {info['name']}")
143        md = build_person_md(uid, info["name"], info["msgs"])
144        (PEOPLE_DIR / f"{uid}.md").write_text(md, encoding="utf-8")
145 
146    # Общий summary.md
147    with open(CONTEXT_DIR / "summary.md", "w", encoding="utf-8") as f:
148        f.write(f"# Контекст Март 7\n")
149        f.write(f"*Обновлено: {datetime.now(timezone.utc).strftime('%Y-%m-%d %H:%M UTC')}*\n\n")
150        f.write(f"**Всего записей:** {len(msgs)}\n\n")
151        f.write("## Чаты\n")
152        for cid, info in chats.items():
153            if len(info["msgs"]) < 3:
154                continue
155            f.write(f"- [{info['title']}](chats/{cid}.md) — {len(info['msgs'])} сообщ.\n")
156        f.write("\n## Люди\n")
157        for uid, info in people.items():
158            if len(info["msgs"]) < 2:
159                continue
160            f.write(f"- [{info['name']}](people/{uid}.md) — {len(info['msgs'])} сообщ.\n")
161 
162    print("[context_builder] Готово!")
163 
164 
165if __name__ == "__main__":
166    main()
167