feat: Telegram-Bot-Integration — Webhook-Endpoint + Download + Thread-Merge + Auto-Reply

- Neuer Endpoint POST /telegram/webhook (GET=Health-Check)
- Parser für alle Telegram-Nachrichtentypen (Text, Foto, Video, Voice, Audio, Dokument, Sticker, Standort)
- Media-Download via Telegram File API (getFile + Download)
- .mov → .mp4 Konvertierung (ffmpeg)
- 5-Min-Thread-Merge-Logik (gleiches Pattern wie WhatsApp)
- Auto-Reply: Willkommensnachricht für neue Nutzer, Danke für bestehende
- GDPR-Trigger 'Meine Daten löschen' integriert
- #out-Erkennung + ausverkauft-Markierung
- Hintergrund-Transkription + LLM-Auto-Analyse
- TELEGRAM_TOKEN via .env konfigurierbar
This commit is contained in:
Hermes Agent
2026-06-18 18:38:09 +02:00
parent f6581b7de5
commit 5181a09c20
2 changed files with 534 additions and 0 deletions
+264
View File
@@ -0,0 +1,264 @@
# -*- coding: utf-8 -*-
"""Telegram Bot webhook handler — integrated into app.py"""
def _telegram_download_file(file_id):
"""Download a file from Telegram servers. Returns local path or None."""
import requests as _r
if not TELEGRAM_TOKEN:
return None
try:
gr = _r.get("{}/getFile?file_id={}".format(TELEGRAM_API, file_id), timeout=10)
gr.raise_for_status()
gdata = gr.json()
if not gdata.get("ok"):
print("[telegram] getFile failed: {}".format(gdata), file=sys.stderr)
return None
file_path = gdata["result"]["file_path"]
ext = os.path.splitext(file_path)[1].lower() or ".bin"
safe_name = "tg_{}{}".format(file_id, ext)
dest = str(MEDIA_DIR / safe_name)
dl_url = "https://api.telegram.org/file/bot{}/{}".format(TELEGRAM_TOKEN, file_path)
fr = _r.get(dl_url, timeout=60)
fr.raise_for_status()
with open(dest, "wb") as f:
f.write(fr.content)
if ext == ".mov":
mp4_path = dest + ".mp4"
import subprocess
result = subprocess.run(
["/home/hermes/.local/bin/ffmpeg", "-y", "-i", dest,
"-c:v", "libx264", "-preset", "fast", "-crf", "23",
"-c:a", "aac", "-b:a", "128k", "-movflags", "+faststart", mp4_path],
capture_output=True, text=True, timeout=120
)
if result.returncode == 0 and os.path.exists(mp4_path):
os.remove(dest)
return mp4_path
return dest
except Exception as e:
print("[telegram download error] {}".format(e), file=sys.stderr)
return None
def _telegram_autoreply(chat_id, sender_name, is_new_user):
"""Send welcome or thank-you message via Telegram Send API."""
if not TELEGRAM_TOKEN:
return
import requests as _r
try:
if is_new_user:
text = (
"🎲 Hey {}! Cool, dass du da bist — willkommen in der ".format(sender_name) +
"brettspiel-news.de Community!\n\n"
"Schick uns einfach, was dich bewegt: 💬 Text, 🎤 Sprachnachrichten, "
"🖼️ Fotos oder 🎬 Videos — alles rund um Brettspiele. Wir schauen uns alles an!\n\n"
"✍️ Nenn uns deinen Vornamen, wenn du magst — das ist deine Einwilligung, "
"dass wir dich als Quelle nennen, falls wir deinen Beitrag verwenden. "
"Ohne Namen schreiben wir „anonyme Quelle\".\n\n"
"🔒 „Meine Daten löschen\" genügt — und alles ist sofort und vollständig weg."
)
else:
text = "📨 Danke {}! Ist angekommen — wir kümmern uns drum.".format(sender_name)
_r.post(
"{}/sendMessage".format(TELEGRAM_API),
json={"chat_id": chat_id, "text": text},
timeout=10,
)
except Exception as e:
print("[telegram reply error] {}".format(e), file=sys.stderr)
@app.route("/telegram/webhook", methods=["GET", "POST"])
def telegram_webhook():
"""Telegram Bot Webhook. GET=health check, POST=incoming messages."""
if request.method == "GET":
if TELEGRAM_TOKEN:
import requests as _r
try:
r = _r.get("{}/getMe".format(TELEGRAM_API), timeout=5)
data = r.json()
if data.get("ok"):
return jsonify({"status": "ok", "bot": data["result"]["username"]})
except Exception:
pass
return jsonify({"status": "ok", "token_configured": True})
return jsonify({"status": "off", "reason": "TELEGRAM_TOKEN not set"}), 500
# POST: Incoming message
body = request.get_json(force=True, silent=True)
if not body or not TELEGRAM_TOKEN:
return "OK", 200
try:
msg = body.get("message", {})
if not msg:
return "OK", 200
chat = msg.get("chat", {})
sender_id = str(chat.get("id", "unknown"))
sender_name = msg.get("from", {}).get("first_name", "") or ""
last_name = msg.get("from", {}).get("last_name", "")
if last_name:
sender_name = "{} {}".format(sender_name, last_name).strip()
if not sender_name:
sender_name = sender_id
msg_type = "text"
content = ""
media_path = None
file_id = None
if "text" in msg:
msg_type = "text"
content = msg["text"]
elif "photo" in msg:
msg_type = "image"
content = msg.get("caption", "[Bild ohne Text]")
file_id = msg["photo"][-1]["file_id"]
elif "video" in msg:
msg_type = "video"
content = msg.get("caption", "[🎬 Video — Transkription läuft...]")
file_id = msg["video"]["file_id"]
elif "voice" in msg:
msg_type = "audio"
content = "[🎤 Sprachnachricht — Transkription läuft...]"
file_id = msg["voice"]["file_id"]
elif "audio" in msg:
msg_type = "audio"
content = msg.get("caption", "[🎤 Audio]")
file_id = msg["audio"]["file_id"]
elif "document" in msg:
msg_type = "document"
content = msg.get("caption", "[📎 Dokument]")
file_id = msg["document"]["file_id"]
elif "sticker" in msg:
content = "[Sticker: {}]".format(msg['sticker'].get('emoji', ''))
msg_type = "document"
elif "location" in msg:
lat = msg["location"]["latitude"]
lon = msg["location"]["longitude"]
content = "[📍 Standort: {}, {}]".format(lat, lon)
msg_type = "text"
else:
content = "[Unbekannter Nachrichtentyp]"
msg_type = "text"
if file_id:
media_path = _telegram_download_file(file_id)
# GDPR delete trigger
if msg_type == "text" and content.strip().lower() == "meine daten löschen":
_gdpr_delete_sender(get_db(), sender_id)
if TELEGRAM_TOKEN:
import requests as _r
try:
_r.post(
"{}/sendMessage".format(TELEGRAM_API),
json={"chat_id": int(sender_id),
"text": "✅ Alle deine Daten wurden vollständig und endgültig vom Server gelöscht. "
"Es befinden sich keinerlei personenbezogene Daten mehr von dir bei uns."},
timeout=10,
)
except Exception:
pass
return "OK", 200
db = get_db()
spiel_titel = None
category_override = None
if "#out" in content.lower():
category_override = "ausverkauft"
idx = content.lower().find("#out")
after = content[idx + 4:].strip()
if after:
spiel_titel = re.split(r'[.!?\n]| - | | — | ist | war | in ', after, maxsplit=1)[0].strip()[:200]
halle = extract_halle(content)
THREAD_WINDOW_MINUTES = 5
recent = db.execute(
"""SELECT id, content, media_path, halle FROM submissions
WHERE channel='telegram' AND sender_id=? AND status != 'gdpr_deleted'
ORDER BY created_at DESC LIMIT 1""",
(sender_id,),
).fetchone()
is_new_user = db.execute(
"SELECT COUNT(*) FROM submissions WHERE channel='telegram' AND sender_id=? AND status != 'gdpr_deleted'",
(sender_id,),
).fetchone()[0] == 0
thread_found = False
if recent:
age_check = db.execute(
"SELECT (strftime('%s','now') - strftime('%s',created_at)) < ? FROM submissions WHERE id=?",
(THREAD_WINDOW_MINUTES * 60, recent["id"]),
).fetchone()
if age_check and age_check[0]:
thread_found = True
if thread_found:
prefix = {
"text": "", "image": "\n[🖼️ Bild] ",
"audio": "\n[🎤 Sprachnachricht]", "video": "\n[🎬 Video] ",
"document": "\n[📎 Dokument] ",
}.get(msg_type, "\n[{}] ".format(msg_type))
new_content = (recent["content"] or "") + prefix + content
new_halle = extract_halle(new_content) or recent["halle"]
db.execute(
"UPDATE submissions SET content=?, halle=?, created_at=CURRENT_TIMESTAMP WHERE id=?",
(new_content, new_halle, recent["id"]),
)
if "#out" in content.lower():
out_title = None
idx2 = content.lower().find("#out")
after2 = content[idx2 + 4:].strip()
if after2:
out_title = re.split(r'[.!?\n]| - | | — | ist | war | in ', after2, maxsplit=1)[0].strip()[:200]
db.execute("UPDATE submissions SET category=?, spiel_titel=? WHERE id=?",
("ausverkauft", out_title, recent["id"]))
if out_title:
mark_ausverkauft_in_neuheiten(db, out_title)
if media_path:
merged = media_path
if recent["media_path"]:
merged = recent["media_path"] + "||" + media_path
db.execute("UPDATE submissions SET media_path=? WHERE id=?", (merged, recent["id"]))
else:
cat = category_override or None
db.execute(
"""INSERT INTO submissions
(channel, sender_id, sender_name, type, content, media_path, status, halle, category, spiel_titel)
VALUES (?, ?, ?, ?, ?, ?, 'new', ?, ?, ?)""",
("telegram", sender_id, sender_name, msg_type, content, media_path, halle, cat, spiel_titel),
)
if cat == "ausverkauft" and spiel_titel:
mark_ausverkauft_in_neuheiten(db, spiel_titel)
if not thread_found:
_telegram_autoreply(int(sender_id), sender_name, is_new_user)
if msg_type == "audio" and media_path:
import threading
sub_id = recent["id"] if thread_found else db.execute("SELECT last_insert_rowid()").fetchone()[0]
threading.Thread(target=_transcribe_bg, args=(DB_PATH, sub_id, media_path), daemon=True).start()
if msg_type == "video" and media_path:
import threading
sub_id = recent["id"] if thread_found else db.execute("SELECT last_insert_rowid()").fetchone()[0]
threading.Thread(target=_transcribe_video_bg, args=(DB_PATH, sub_id, media_path), daemon=True).start()
if msg_type in ("text", "image") and not thread_found and content and len(content.strip()) > 20:
import threading
sub_id = db.execute("SELECT last_insert_rowid()").fetchone()[0]
if not halle:
threading.Thread(target=auto_process_bg, args=(DB_PATH, sub_id), daemon=True).start()
db.commit()
except Exception as e:
print("[telegram webhook error] {}".format(e), file=sys.stderr)
return "OK", 200
return "OK", 200