Initial commit: Aoi (Аой-тян) — Matrix notifier for Navidrome and discovery-playlist

- aoi.py: poller for Navidrome (main + anime libraries), discovery playlists, release watch for liked/rated artists - Last.fm enrichment (bio, tags); Bandcamp search links - config.example.json: safe template; config.json gitignored - deploy/aoi.service: systemd unit (production) - assets/banner.png + aoi-avatar.png: persona banner + bot avatar - Russian README in line with sibling bots (rada/watcher)
2026-04-28 20:15:50 +03:00 · 2026-04-28 20:15:50 +03:00 · 5fa857b9cc
commit 5fa857b9cc
8 changed files with 1155 additions and 0 deletions
--- a/.gitignore
+++ b/.gitignore
@ -0,0 +1,12 @@
+config.json
+*.db
+*.db-*
+*.log
+.venv/
+__pycache__/
+*.pyc
+.pytest_cache/
+*.tmp
+
+
+*.db.bak*
--- a/README.md
+++ b/README.md
@ -0,0 +1,160 @@
+<p align="center">
+  <img src="assets/banner.png" alt="Aoi banner" width="40%">
+</p>
+
+# Aoi
+
+**Aoi** (Аой-тян) — домашний Matrix-бот для музыкальной библиотеки. Она следит за Navidrome и discovery-playlist, аккуратно докладывает о новых альбомах, свежих Discovery-плейлистах и предстоящих релизах любимых артистов, обогащая карточки данными Last.fm и ссылками на внешние источники.
+
+## Что умеет
+
+| Блок | Что происходит |
+| --- | --- |
+| Navidrome albums | Поллинг библиотек `main` и `anime` — новые альбомы превращаются в карточки с обложкой, описанием, тегами и рейтингом. |
+| Discovery playlists | Свежесгенерированные плейлисты discovery-playlist (имя содержит «Discovery») приходят в Matrix отдельным постом. |
+| Release watch | Аналитика по любимым/высоко оцененным артистам: анонсы и фактические релизы из дискографии, отслеживаемой discovery-playlist. |
+| Rich-описания | Last.fm-биография, теги, обложка из `album.getInfo`, ссылки на Last.fm и поиск Bandcamp. |
+| Baseline | На первом запуске можно зафиксировать всё уже существующее как «уже видела», чтобы не получить спам с историей. |
+| State & dedupe | SQLite хранит увиденные альбомы, плейлисты и релизы — повторов не будет. |
+
+## Как устроена логика
+
+Aoi не делает запросов к LLM и не пытается «угадывать» — она опирается на конкретные источники и кеширует увиденное:
+
+1. По расписанию опрашивает Navidrome и discovery-playlist (раздельные интервалы для альбомов, плейлистов и релизов).
+2. Сравнивает с локальным state и забирает только новые сущности.
+3. Обогащает карточки Last.fm-данными при наличии ключа.
+4. Шлёт карточку в Matrix-комнату с обложкой, описанием и ссылками.
+5. Помечает сущность как обработанную в SQLite.
+
+Каждый источник работает независимо, так что падение Navidrome не валит discovery, и наоборот.
+
+## Сообщения в Matrix
+
+Карточка нового альбома выглядит примерно так:
+
+```text
+🎧 Новый альбом в Navidrome
+Artist — Album Title (2026)
+
+━━ ОПИСАНИЕ ━━
+Краткая биография/описание альбома из Last.fm…
+
+━━ ТЕГИ ━━
+shoegaze · indie · dream-pop
+
+🔗 Last.fm: https://www.last.fm/music/Artist/Album
+🔗 Bandcamp: https://bandcamp.com/search?q=Artist+Album
+```
+
+Discovery-плейлисты приходят отдельной карточкой со списком треков, release watch — отдельной с пометкой типа релиза (`announced`, `released`).
+
+## Файлы проекта
+
+| Файл | Назначение |
+| --- | --- |
+| `aoi.py` | Основной сервис. |
+| `config.example.json` | Безопасный пример конфигурации. |
+| `config.json` | Локальная конфигурация с секретами, не коммитится. |
+| `requirements.txt` | Python-зависимости. |
+| `deploy/aoi.service` | systemd unit для production. |
+| `assets/banner.png` | Баннер README. |
+| `aoi.db` | SQLite-состояние, не коммитится. |
+| `aoi.log` | Лог сервиса, не коммитится. |
+
+## Конфигурация
+
+Создать конфиг из примера:
+
+```bash
+cp config.example.json config.json
+```
+
+Заполнить:
+
+- `matrix.homeserver`, `matrix.room_id`, `matrix.access_token`, `matrix.user_id`;
+- `bot.name` (по умолчанию `Аой-тян`) и `bot.avatar_path` для Matrix-профиля;
+- `navidrome.url`, `navidrome.username`, `navidrome.password`, идентификаторы библиотек;
+- `discovery.db_path` — путь к SQLite базе discovery-playlist;
+- `metadata.lastfm_api_key` — ключ Last.fm для био и тегов;
+- `polling.*` — интервалы опроса источников;
+- `polling.baseline_existing_on_first_run` — `true` на первом запуске, чтобы не спамить историей.
+
+Production-путь:
+
+```text
+/storage/scripts/aoi
+```
+
+## Установка
+
+В production Aoi использует виртуальное окружение Watcher (общие зависимости):
+
+```bash
+cd /storage/scripts/watcher
+python3 -m venv .venv
+. .venv/bin/activate
+pip install -r /storage/scripts/aoi/requirements.txt
+```
+
+Можно завести отдельный venv — тогда обновить `deploy/aoi.service`.
+
+## Запуск
+
+```bash
+/storage/scripts/watcher/.venv/bin/python /storage/scripts/aoi/aoi.py
+```
+
+Healthcheck:
+
+```bash
+curl -fsS http://127.0.0.1:18323/healthz
+```
+
+Ожидаемый ответ:
+
+```json
+{"ok":true}
+```
+
+## systemd
+
+Unit хранится в репозитории:
+
+```text
+deploy/aoi.service
+```
+
+Установка или обновление:
+
+```bash
+sudo cp /storage/scripts/aoi/deploy/aoi.service /etc/systemd/system/aoi.service
+sudo systemctl daemon-reload
+sudo systemctl enable aoi.service
+sudo systemctl restart aoi.service
+```
+
+Операции:
+
+```bash
+sudo systemctl restart aoi.service
+systemctl --no-pager -l status aoi.service
+journalctl -u aoi.service -f
+tail -f /storage/scripts/aoi/aoi.log
+```
+
+## Ручные команды
+
+```bash
+curl -X POST http://127.0.0.1:18323/run/navidrome-albums
+curl -X POST http://127.0.0.1:18323/run/discovery-playlists
+curl -X POST http://127.0.0.1:18323/run/release-watch
+```
+
+## Что не коммитить
+
+- `config.json`
+- `*.db`
+- `*.log`
+- `.venv/`
+- `__pycache__/`
--- a/aoi-avatar.png
+++ b/aoi-avatar.png
--- a/aoi.py
+++ b/aoi.py
@ -0,0 +1,921 @@
+"""Aoi-chan: Matrix notifications for Navidrome and discovery-playlist."""
+
+from __future__ import annotations
+
+import html
+import json
+import mimetypes
+import os
+import re
+import sqlite3
+import threading
+import time
+import traceback
+import uuid
+from collections import Counter
+from datetime import datetime, timezone, timedelta
+from pathlib import Path
+from typing import Any
+from urllib.parse import quote_plus
+
+import requests
+from flask import Flask, jsonify, request
+
+
+BASE_DIR = Path(__file__).resolve().parent
+CONFIG_PATH = Path(os.environ.get("AOI_CONFIG", BASE_DIR / "config.json"))
+STATE_DB = Path(os.environ.get("AOI_STATE_DB", BASE_DIR / "aoi.db"))
+LOG_PATH = Path(os.environ.get("AOI_LOG", BASE_DIR / "aoi.log"))
+STATE_LOCK = threading.Lock()
+
+
+def _load_config() -> dict:
+    example = BASE_DIR / "config.example.json"
+    path = CONFIG_PATH if CONFIG_PATH.exists() else example
+    with path.open("r", encoding="utf-8-sig") as f:
+        return json.load(f)
+
+
+CFG = _load_config()
+
+
+def log(message: str) -> None:
+    line = f"{datetime.now(timezone.utc).isoformat()} {message}"
+    print(line, flush=True)
+    try:
+        with LOG_PATH.open("a", encoding="utf-8") as f:
+            f.write(line + "\n")
+    except Exception:
+        pass
+
+
+def now_utc() -> datetime:
+    return datetime.now(timezone.utc)
+
+
+def parse_dt(value: Any) -> datetime | None:
+    if not value:
+        return None
+    if isinstance(value, datetime):
+        return value if value.tzinfo else value.replace(tzinfo=timezone.utc)
+    text = str(value).strip()
+    if not text:
+        return None
+    text = text.replace("Z", "+00:00")
+    for candidate in (text, text.split(".")[0]):
+        try:
+            dt = datetime.fromisoformat(candidate)
+            return dt if dt.tzinfo else dt.replace(tzinfo=timezone.utc)
+        except Exception:
+            pass
+    for fmt in ("%Y-%m-%d %H:%M:%S", "%Y-%m-%d", "%Y-%m", "%Y"):
+        try:
+            return datetime.strptime(text[: len(fmt)], fmt).replace(tzinfo=timezone.utc)
+        except Exception:
+            pass
+    return None
+
+
+def parse_date(value: Any) -> datetime | None:
+    dt = parse_dt(value)
+    if dt:
+        return dt
+    text = str(value or "").strip()
+    for fmt in ("%Y", "%Y-%m", "%Y-%m-%d"):
+        try:
+            return datetime.strptime(text, fmt).replace(tzinfo=timezone.utc)
+        except Exception:
+            pass
+    return None
+
+
+def strip_html(text: str) -> str:
+    text = re.sub(r"<br\s*/?>", "\n", text or "", flags=re.I)
+    text = re.sub(r"<[^>]+>", "", text)
+    return html.unescape(text).strip()
+
+
+def compact(text: str, limit: int = 700) -> str:
+    text = re.sub(r"\s+", " ", strip_html(text)).strip()
+    if len(text) <= limit:
+        return text
+    return text[: limit - 1].rstrip() + "…"
+
+
+def clean_lastfm_text(text: str) -> str:
+    text = re.sub(r"Read more on Last\.fm.*$", "", text or "", flags=re.I | re.S)
+    text = re.sub(r"User-contributed text.*$", "", text, flags=re.I | re.S)
+    return re.sub(r"\s+", " ", strip_html(text)).strip()
+
+
+def init_state() -> None:
+    with STATE_LOCK, sqlite3.connect(STATE_DB, timeout=30) as conn:
+        conn.execute("PRAGMA journal_mode=WAL")
+        conn.execute("PRAGMA busy_timeout=30000")
+        conn.execute(
+            """
+            CREATE TABLE IF NOT EXISTS sent_events (
+                kind TEXT NOT NULL,
+                event_key TEXT NOT NULL,
+                payload_json TEXT,
+                created_at TEXT NOT NULL,
+                PRIMARY KEY(kind, event_key)
+            )
+            """
+        )
+        conn.execute(
+            """
+            CREATE TABLE IF NOT EXISTS cursors (
+                key TEXT PRIMARY KEY,
+                value TEXT,
+                updated_at TEXT NOT NULL
+            )
+            """
+        )
+        conn.commit()
+
+
+def already_sent(kind: str, key: str) -> bool:
+    with STATE_LOCK, sqlite3.connect(STATE_DB, timeout=30) as conn:
+        conn.execute("PRAGMA busy_timeout=30000")
+        row = conn.execute(
+            "SELECT 1 FROM sent_events WHERE kind=? AND event_key=?",
+            (kind, key),
+        ).fetchone()
+        return bool(row)
+
+
+def mark_sent(kind: str, key: str, payload: dict | None = None) -> None:
+    with STATE_LOCK, sqlite3.connect(STATE_DB, timeout=30) as conn:
+        conn.execute("PRAGMA busy_timeout=30000")
+        conn.execute(
+            """
+            INSERT OR IGNORE INTO sent_events(kind,event_key,payload_json,created_at)
+            VALUES(?,?,?,?)
+            """,
+            (kind, key, json.dumps(payload or {}, ensure_ascii=False), now_utc().isoformat()),
+        )
+        conn.commit()
+
+
+def get_cursor(key: str) -> str:
+    with STATE_LOCK, sqlite3.connect(STATE_DB, timeout=30) as conn:
+        conn.execute("PRAGMA busy_timeout=30000")
+        row = conn.execute("SELECT value FROM cursors WHERE key=?", (key,)).fetchone()
+        return row[0] if row else ""
+
+
+def set_cursor(key: str, value: str) -> None:
+    with STATE_LOCK, sqlite3.connect(STATE_DB, timeout=30) as conn:
+        conn.execute("PRAGMA busy_timeout=30000")
+        conn.execute(
+            """
+            INSERT INTO cursors(key,value,updated_at) VALUES(?,?,?)
+            ON CONFLICT(key) DO UPDATE SET value=excluded.value, updated_at=excluded.updated_at
+            """,
+            (key, value, now_utc().isoformat()),
+        )
+        conn.commit()
+
+
+def request_session() -> requests.Session:
+    sess = requests.Session()
+    proxy = (CFG.get("metadata") or {}).get("proxy") or ""
+    if proxy:
+        sess.proxies.update({"http": proxy, "https": proxy})
+    return sess
+
+
+HTTP = request_session()
+LASTFM_API_KEY_CACHE: str | None = None
+
+
+class MatrixClient:
+    def __init__(self, cfg: dict):
+        self.homeserver = cfg["homeserver"].rstrip("/")
+        self.room_id = cfg["room_id"]
+        self.token = cfg["access_token"]
+        self.user_id = cfg.get("user_id") or ""
+
+    @property
+    def headers(self) -> dict:
+        return {"Authorization": f"Bearer {self.token}"}
+
+    def set_profile(self) -> None:
+        bot = CFG.get("bot") or {}
+        name = bot.get("name")
+        if name and self.user_id:
+            url = f"{self.homeserver}/_matrix/client/v3/profile/{self.user_id}/displayname"
+            r = requests.put(url, headers=self.headers, json={"displayname": name}, timeout=20)
+            if not r.ok:
+                log(f"matrix display name update failed: {r.status_code} {r.text[:200]}")
+
+        avatar_path = bot.get("avatar_path") or ""
+        if avatar_path:
+            path = Path(avatar_path)
+            if path.exists() and self.user_id:
+                mxc = self.upload_file(path.read_bytes(), path.name, mimetypes.guess_type(path.name)[0] or "image/jpeg")
+                if mxc:
+                    url = f"{self.homeserver}/_matrix/client/v3/profile/{self.user_id}/avatar_url"
+                    r = requests.put(url, headers=self.headers, json={"avatar_url": mxc}, timeout=20)
+                    if not r.ok:
+                        log(f"matrix avatar update failed: {r.status_code} {r.text[:200]}")
+
+    def upload_file(self, data: bytes, filename: str, content_type: str) -> str:
+        r = requests.post(
+            f"{self.homeserver}/_matrix/media/v3/upload",
+            headers={**self.headers, "Content-Type": content_type},
+            params={"filename": filename},
+            data=data,
+            timeout=45,
+        )
+        r.raise_for_status()
+        return r.json()["content_uri"]
+
+    def send_text(self, text: str, formatted_html: str | None = None) -> dict:
+        body = {"msgtype": "m.text", "body": text}
+        if formatted_html:
+            body["format"] = "org.matrix.custom.html"
+            body["formatted_body"] = formatted_html
+        txn = uuid.uuid4().hex
+        url = f"{self.homeserver}/_matrix/client/v3/rooms/{self.room_id}/send/m.room.message/{txn}"
+        r = requests.put(url, headers=self.headers, json=body, timeout=30)
+        r.raise_for_status()
+        return r.json()
+
+    def send_image_bytes(self, data: bytes, filename: str, content_type: str = "image/jpeg") -> dict:
+        mxc = self.upload_file(data, filename, content_type)
+        body = {
+            "msgtype": "m.image",
+            "body": filename,
+            "url": mxc,
+            "info": {"mimetype": content_type, "size": len(data)},
+        }
+        txn = uuid.uuid4().hex
+        url = f"{self.homeserver}/_matrix/client/v3/rooms/{self.room_id}/send/m.room.message/{txn}"
+        r = requests.put(url, headers=self.headers, json=body, timeout=30)
+        r.raise_for_status()
+        return r.json()
+
+    def send_card(self, text: str, formatted_html: str, image: tuple[bytes, str, str] | None = None) -> None:
+        if image:
+            try:
+                self.send_image_bytes(*image)
+            except Exception as e:
+                log(f"matrix image send failed: {e}")
+        self.send_text(text, formatted_html)
+
+
+class NavidromeClient:
+    def __init__(self, cfg: dict):
+        self.base_url = cfg["url"].rstrip("/")
+        self.username = cfg["username"]
+        self.password = cfg["password"]
+
+    def _params(self) -> dict:
+        return {
+            "u": self.username,
+            "p": self.password,
+            "v": "1.16.1",
+            "c": "AoiNotifier",
+            "f": "json",
+        }
+
+    def get(self, endpoint: str, **params) -> dict:
+        r = requests.get(
+            f"{self.base_url}/rest/{endpoint}",
+            params={**self._params(), **params},
+            timeout=35,
+        )
+        r.raise_for_status()
+        data = r.json().get("subsonic-response", {})
+        if data.get("status") != "ok":
+            raise RuntimeError(f"Navidrome {endpoint} failed: {data}")
+        return data
+
+    def ping(self) -> bool:
+        try:
+            return self.get("ping").get("status") == "ok"
+        except Exception:
+            return False
+
+    def newest_albums(self, music_folder_id: int, size: int) -> list[dict]:
+        data = self.get("getAlbumList2", type="newest", size=size, musicFolderId=music_folder_id)
+        return data.get("albumList2", {}).get("album", []) or []
+
+    def album(self, album_id: str) -> dict:
+        return self.get("getAlbum", id=album_id).get("album", {}) or {}
+
+    def playlists(self) -> list[dict]:
+        data = self.get("getPlaylists")
+        return data.get("playlists", {}).get("playlist", []) or []
+
+    def playlist(self, playlist_id: str) -> dict:
+        return self.get("getPlaylist", id=playlist_id).get("playlist", {}) or {}
+
+    def cover(self, cover_art: str | None) -> tuple[bytes, str, str] | None:
+        if not cover_art:
+            return None
+        try:
+            r = requests.get(
+                f"{self.base_url}/rest/getCoverArt",
+                params={**self._params(), "id": cover_art, "size": 900},
+                timeout=35,
+            )
+            r.raise_for_status()
+            ctype = r.headers.get("Content-Type") or "image/jpeg"
+            ext = mimetypes.guess_extension(ctype.split(";")[0]) or ".jpg"
+            return (r.content, f"cover{ext}", ctype)
+        except Exception as e:
+            log(f"cover fetch failed: {e}")
+            return None
+
+
+MATRIX = MatrixClient(CFG["matrix"])
+NAV = NavidromeClient(CFG["navidrome"])
+
+
+def metadata_headers() -> dict:
+    ua = (CFG.get("metadata") or {}).get("user_agent") or "AoiNotifier/1.0"
+    return {"User-Agent": ua, "Accept": "application/json"}
+
+
+def discovery_setting(key: str) -> str:
+    db_path = (CFG.get("discovery") or {}).get("db_path") or ""
+    if not db_path or not Path(db_path).exists():
+        return ""
+    try:
+        with sqlite3.connect(db_path) as conn:
+            row = conn.execute("SELECT value FROM user_settings WHERE key=?", (key,)).fetchone()
+            return str(row[0] or "") if row else ""
+    except Exception as e:
+        log(f"discovery setting lookup failed for {key}: {e}")
+        return ""
+
+
+def lastfm_api_key() -> str:
+    global LASTFM_API_KEY_CACHE
+    if LASTFM_API_KEY_CACHE is not None:
+        return LASTFM_API_KEY_CACHE
+    LASTFM_API_KEY_CACHE = (CFG.get("metadata") or {}).get("lastfm_api_key") or discovery_setting("lastfm_api_key")
+    return LASTFM_API_KEY_CACHE
+
+
+def lastfm_request(method: str, params: dict, timeout: int = 25) -> dict:
+    api_key = lastfm_api_key()
+    if not api_key:
+        return {}
+    req_params = {
+        "method": method,
+        "api_key": api_key,
+        "format": "json",
+        "autocorrect": 1,
+        **params,
+    }
+    try:
+        r = HTTP.get("https://ws.audioscrobbler.com/2.0/", params=req_params, timeout=timeout)
+        if not r.ok:
+            return {}
+        return r.json()
+    except Exception as e:
+        log(f"lastfm {method} failed: {e}")
+        return {}
+
+
+def lastfm_album_tags(artist: str, album: str, limit: int = 6) -> list[str]:
+    data = lastfm_request("album.getTopTags", {"artist": artist, "album": album}, timeout=20)
+    tags = []
+    for tag in ((data.get("toptags") or {}).get("tag") or [])[:limit]:
+        name = (tag.get("name") or "").strip()
+        if name and len(name) <= 32 and not name.isdigit():
+            tags.append(name)
+    return tags
+
+
+def lastfm_artist_summary(artist: str) -> str:
+    for params in ({"artist": artist, "lang": "ru"}, {"artist": artist}):
+        data = lastfm_request("artist.getinfo", params, timeout=20)
+        bio = ((data.get("artist") or {}).get("bio") or {})
+        summary = clean_lastfm_text(bio.get("summary") or bio.get("content") or "")
+        if summary:
+            return summary
+    return ""
+
+
+def lastfm_album_search_url(artist: str, album: str) -> str:
+    return f"https://www.last.fm/search/albums?q={quote_plus(f'{artist} {album}')}"
+
+
+def bandcamp_album_search_url(artist: str, album: str) -> str:
+    return f"https://bandcamp.com/search?q={quote_plus(f'{artist} {album}')}&item_type=a"
+
+
+def album_source_links(artist: str, album: str, lastfm_url: str = "") -> list[dict[str, str]]:
+    return [
+        {"label": "Last.fm", "url": lastfm_url or lastfm_album_search_url(artist, album)},
+        {"label": "Bandcamp", "url": bandcamp_album_search_url(artist, album)},
+    ]
+
+
+def lastfm_album_info(artist: str, album: str) -> dict[str, str]:
+    for params in ({"artist": artist, "album": album, "lang": "ru"}, {"artist": artist, "album": album}):
+        data = lastfm_request("album.getinfo", params)
+        album_data = data.get("album") or {}
+        wiki = album_data.get("wiki") or {}
+        summary = clean_lastfm_text(wiki.get("summary") or wiki.get("content") or "")
+        if summary:
+            return {"summary": summary, "url": album_data.get("url") or ""}
+
+    tags = lastfm_album_tags(artist, album)
+    artist_summary = lastfm_artist_summary(artist)
+    parts = []
+    if tags:
+        parts.append("Last.fm описывает релиз тегами: " + ", ".join(tags) + ".")
+    if artist_summary:
+        parts.append(artist_summary)
+    if parts:
+        return {"summary": " ".join(parts), "url": ""}
+    return {"summary": "", "url": ""}
+
+
+def lastfm_album_summary(artist: str, album: str) -> str:
+    return lastfm_album_info(artist, album).get("summary", "")
+
+def musicbrainz_release_hint(artist: str, album: str) -> dict:
+    if not (CFG.get("metadata") or {}).get("musicbrainz_enabled", True):
+        return {}
+    try:
+        query = f'artist:"{artist}" AND release:"{album}"'
+        r = HTTP.get(
+            "https://musicbrainz.org/ws/2/release/",
+            params={"query": query, "fmt": "json", "limit": 3},
+            headers=metadata_headers(),
+            timeout=25,
+        )
+        if not r.ok:
+            return {}
+        releases = r.json().get("releases") or []
+        if not releases:
+            return {}
+        rel = releases[0]
+        return {
+            "date": rel.get("date") or "",
+            "country": rel.get("country") or "",
+            "status": rel.get("status") or "",
+            "score": rel.get("score") or 0,
+        }
+    except Exception as e:
+        log(f"musicbrainz lookup failed: {e}")
+        return {}
+
+
+def lastfm_tags_for_tracks(tracks: list[dict]) -> Counter:
+    api_key = lastfm_api_key()
+    tags = Counter()
+    if not api_key:
+        return tags
+    for tr in tracks[:8]:
+        try:
+            r = HTTP.get(
+                "https://ws.audioscrobbler.com/2.0/",
+                params={
+                    "method": "track.getTopTags",
+                    "artist": tr.get("artist") or "",
+                    "track": tr.get("track") or tr.get("title") or "",
+                    "api_key": api_key,
+                    "format": "json",
+                    "autocorrect": 1,
+                },
+                timeout=20,
+            )
+            if not r.ok:
+                continue
+            for tag in ((r.json().get("toptags") or {}).get("tag") or [])[:5]:
+                name = (tag.get("name") or "").lower().strip()
+                if name and len(name) <= 32:
+                    tags[name] += 1
+        except Exception:
+            continue
+    return tags
+
+
+def album_plain_and_html(album: dict, library_label: str, songs: list[dict]) -> tuple[str, str]:
+    artist = album.get("artist") or "Unknown artist"
+    title = album.get("name") or album.get("album") or "Unknown album"
+    year = album.get("year") or ""
+    duration = int(album.get("duration") or sum(int(s.get("duration") or 0) for s in songs) or 0)
+    minutes = duration // 60 if duration else 0
+    lastfm_info = lastfm_album_info(artist, title)
+    summary = lastfm_info.get("summary") or ""
+    links = album_source_links(artist, title, lastfm_info.get("url") or "")
+    hint = musicbrainz_release_hint(artist, title)
+    release_date = album.get("created") or album.get("starred") or hint.get("date") or ""
+    track_lines = []
+    for idx, song in enumerate(songs[:30], 1):
+        name = song.get("title") or "Untitled"
+        dur = int(song.get("duration") or 0)
+        suffix = f" ({dur // 60}:{dur % 60:02d})" if dur else ""
+        track_lines.append(f"{idx:02d}. {name}{suffix}")
+    if len(songs) > 30:
+        track_lines.append(f"... и еще {len(songs) - 30}")
+
+    header = f"🎧 Новый релиз в Navidrome / {library_label}"
+    facts = [
+        f"{artist} — {title}" + (f" ({year})" if year else ""),
+        f"Треков: {len(songs)}" + (f", длительность: {minutes} мин." if minutes else ""),
+    ]
+    if release_date:
+        facts.append(f"Дата/добавление: {release_date}")
+    if hint.get("country") or hint.get("status"):
+        facts.append("MusicBrainz: " + ", ".join(x for x in [hint.get("country"), hint.get("status")] if x))
+    if summary:
+        description_fact = f"Описание: {summary}"
+    else:
+        description_fact = "Описание: пока не нашла нормальное описание, оставляю чистый треклист."
+    facts.append(description_fact)
+    facts.append("Ссылки: " + ", ".join(f"{link['label']}: {link['url']}" for link in links))
+    plain = "\n".join([header, "", *facts, "", "━━ ТРЕКЛИСТ ━━", *track_lines])
+
+    html_parts = [
+        f"<h3>{html.escape(header)}</h3>",
+        f"<b>{html.escape(artist)} — {html.escape(title)}</b>" + (f" ({html.escape(str(year))})" if year else ""),
+        "<ul>",
+        f"<li>Треков: {len(songs)}" + (f", длительность: {minutes} мин." if minutes else "") + "</li>",
+    ]
+    if release_date:
+        html_parts.append(f"<li>Дата/добавление: {html.escape(str(release_date))}</li>")
+    if hint.get("country") or hint.get("status"):
+        html_parts.append(f"<li>MusicBrainz: {html.escape(', '.join(x for x in [hint.get('country'), hint.get('status')] if x))}</li>")
+    html_parts.append("</ul>")
+    html_parts.append(f"<p>{html.escape(description_fact)}</p>")
+    html_parts.append(
+        "<p>Ссылки: "
+        + ", ".join(
+            f'<a href="{html.escape(link["url"], quote=True)}">{html.escape(link["label"])}</a>'
+            for link in links
+        )
+        + "</p>"
+    )
+    html_parts.append("<p><b>Треклист</b></p><ol>")
+    for line in track_lines:
+        html_parts.append(f"<li>{html.escape(line[4:] if re.match(r'^\\d\\d\\. ', line) else line)}</li>")
+    html_parts.append("</ol>")
+    return plain, "\n".join(html_parts)
+
+
+def poll_navidrome_albums(force: bool = False) -> dict:
+    cfg = CFG["navidrome"]
+    libraries = [
+        ("main", int(cfg.get("main_library_id") or 0), "main"),
+        ("anime", int(cfg.get("anime_library_id") or 0), "anime"),
+    ]
+    size = int(cfg.get("album_poll_size") or 40)
+    baseline = (CFG.get("polling") or {}).get("baseline_existing_on_first_run", True)
+    first_run = not get_cursor("navidrome_albums_baselined")
+    sent = 0
+    seen = 0
+    for kind, folder_id, label in libraries:
+        if not folder_id:
+            continue
+        for album in NAV.newest_albums(folder_id, size):
+            album_id = str(album.get("id") or "")
+            if not album_id:
+                continue
+            key = f"{kind}:{album_id}"
+            if already_sent("navidrome_album", key):
+                continue
+            seen += 1
+            if first_run and baseline and not force:
+                mark_sent("navidrome_album", key, album)
+                continue
+            full_album = NAV.album(album_id)
+            songs = full_album.get("song") or []
+            plain, formatted = album_plain_and_html(full_album or album, label, songs)
+            image = NAV.cover((full_album or album).get("coverArt"))
+            MATRIX.send_card(plain, formatted, image)
+            mark_sent("navidrome_album", key, {"album": album, "sent_at": now_utc().isoformat()})
+            sent += 1
+            time.sleep(1)
+    set_cursor("navidrome_albums_baselined", now_utc().isoformat())
+    return {"ok": True, "seen_new": seen, "sent": sent, "first_run": first_run}
+
+
+def discovery_conn() -> sqlite3.Connection | None:
+    db_path = (CFG.get("discovery") or {}).get("db_path") or ""
+    if not db_path or not Path(db_path).exists():
+        log(f"discovery db not found: {db_path}")
+        return None
+    conn = sqlite3.connect(db_path)
+    conn.row_factory = sqlite3.Row
+    return conn
+
+
+def playlist_rows(limit: int = 20) -> list[sqlite3.Row]:
+    conn = discovery_conn()
+    if not conn:
+        return []
+    try:
+        return conn.execute(
+            """
+            SELECT * FROM playlists
+            ORDER BY COALESCE(last_generated, created_at) DESC
+            LIMIT ?
+            """,
+            (limit,),
+        ).fetchall()
+    finally:
+        conn.close()
+
+
+def playlist_tracks(playlist_id: int) -> list[dict]:
+    conn = discovery_conn()
+    if not conn:
+        return []
+    try:
+        rows = conn.execute(
+            """
+            SELECT * FROM playlist_tracks
+            WHERE playlist_id=?
+            ORDER BY score DESC, added_at DESC
+            """,
+            (playlist_id,),
+        ).fetchall()
+        return [dict(r) for r in rows]
+    finally:
+        conn.close()
+
+
+def navidrome_playlist_cover(name: str) -> tuple[bytes, str, str] | None:
+    try:
+        needle = name.strip().lower()
+        candidates = NAV.playlists()
+        chosen = None
+        for pl in candidates:
+            pl_name = (pl.get("name") or "").strip().lower()
+            if pl_name == needle or needle in pl_name or pl_name in needle:
+                chosen = pl
+                break
+        if not chosen:
+            return None
+        full = NAV.playlist(str(chosen.get("id")))
+        if full.get("coverArt"):
+            return NAV.cover(full.get("coverArt"))
+        entries = full.get("entry") or []
+        if entries:
+            return NAV.cover(entries[0].get("coverArt"))
+    except Exception as e:
+        log(f"navidrome playlist cover failed: {e}")
+    return None
+
+
+PROFILE_LABELS = {
+    "deep_cuts": "глубокие находки",
+    "fresh_picks": "свежие рекомендации",
+    "comfort_zone": "комфортная зона",
+}
+
+
+def playlist_plain_and_html(pl: dict, tracks: list[dict]) -> tuple[str, str]:
+    name = pl.get("name") or "Discovery"
+    created = pl.get("last_generated") or pl.get("created_at") or ""
+    top_tracks = tracks[:8]
+    top_lines = [f"• {t.get('artist')} — {t.get('track')}" for t in top_tracks]
+    profile_counts = Counter((t.get("profile") or "").strip() for t in tracks if t.get("profile"))
+    profile_text = ", ".join(
+        f"{PROFILE_LABELS.get(k, k)}: {v}" for k, v in profile_counts.most_common()
+    ) or "смешанный режим"
+    tags = lastfm_tags_for_tracks(tracks)
+    tag_text = ", ".join(tag for tag, _ in tags.most_common(6))
+    if not tag_text:
+        tag_text = profile_text
+    artists = Counter((t.get("artist") or "").strip() for t in tracks if t.get("artist"))
+    artist_text = ", ".join(a for a, _ in artists.most_common(5))
+    summary = (
+        f"В этот раз уклон: {tag_text}. "
+        f"Из заметного: {', '.join(line[2:] for line in top_lines[:4])}."
+    )
+    plain = "\n".join([
+        "🧭 Новый Discovery-плейлист",
+        "",
+        name,
+        f"Треков: {len(tracks)}",
+        f"Собран: {created}",
+        f"Профиль: {profile_text}",
+        f"Главные артисты: {artist_text or 'пока не выделяются'}",
+        "",
+        summary,
+        "",
+        "━━ ИНТЕРЕСНОЕ ━━",
+        *top_lines,
+    ])
+    formatted = "\n".join([
+        "<h3>🧭 Новый Discovery-плейлист</h3>",
+        f"<b>{html.escape(name)}</b>",
+        "<ul>",
+        f"<li>Треков: {len(tracks)}</li>",
+        f"<li>Собран: {html.escape(str(created))}</li>",
+        f"<li>Профиль: {html.escape(profile_text)}</li>",
+        f"<li>Главные артисты: {html.escape(artist_text or 'пока не выделяются')}</li>",
+        "</ul>",
+        f"<p>{html.escape(summary)}</p>",
+        "<p><b>Интересное</b></p><ul>",
+        *[f"<li>{html.escape(line[2:])}</li>" for line in top_lines],
+        "</ul>",
+    ])
+    return plain, formatted
+
+
+def poll_discovery_playlists(force: bool = False) -> dict:
+    baseline = (CFG.get("polling") or {}).get("baseline_existing_on_first_run", True)
+    first_run = not get_cursor("discovery_playlists_baselined")
+    sent = 0
+    seen = 0
+    for row in playlist_rows():
+        pl = dict(row)
+        stamp = pl.get("last_generated") or pl.get("created_at") or ""
+        if not stamp:
+            continue
+        key = f"{pl.get('id')}:{stamp}:{pl.get('iteration_number', 0)}"
+        if already_sent("discovery_playlist", key):
+            continue
+        seen += 1
+        if first_run and baseline and not force:
+            mark_sent("discovery_playlist", key, pl)
+            continue
+        tracks = playlist_tracks(int(pl["id"]))
+        plain, formatted = playlist_plain_and_html(pl, tracks)
+        image = navidrome_playlist_cover(pl.get("name") or "")
+        MATRIX.send_card(plain, formatted, image)
+        mark_sent("discovery_playlist", key, {"playlist": pl, "sent_at": now_utc().isoformat()})
+        sent += 1
+        time.sleep(1)
+    set_cursor("discovery_playlists_baselined", now_utc().isoformat())
+    return {"ok": True, "seen_new": seen, "sent": sent, "first_run": first_run}
+
+
+def release_rows(limit: int = 120) -> list[dict]:
+    conn = discovery_conn()
+    if not conn:
+        return []
+    try:
+        rows = conn.execute(
+            """
+            SELECT * FROM artist_release_watch
+            ORDER BY COALESCE(downloaded_at, first_seen_at) DESC
+            LIMIT ?
+            """,
+            (limit,),
+        ).fetchall()
+        return [dict(r) for r in rows]
+    finally:
+        conn.close()
+
+
+def release_plain_and_html(row: dict, mode: str) -> tuple[str, str]:
+    artist = row.get("artist") or "Unknown artist"
+    album = row.get("album") or "Unknown release"
+    date = row.get("release_date") or "дата уточняется"
+    source = row.get("source") or "discovery-playlist"
+    tracks = int(row.get("desired_track_count") or 0)
+    status = row.get("status") or ""
+    if mode == "out":
+        title = "📀 Релиз вышел"
+        lead = f"{artist} — {album} уже можно забирать."
+    else:
+        title = "📡 Анонс релиза отслеживаемого артиста"
+        lead = f"{artist} — {album} появился в release watch."
+    facts = [
+        lead,
+        f"Дата релиза: {date}",
+        f"Статус: {status}",
+        f"Источник: {source}",
+    ]
+    if tracks:
+        facts.append(f"Ожидаемый треклист: {tracks} треков")
+    if row.get("error_message"):
+        facts.append(f"Заметка: {compact(row.get('error_message'), 240)}")
+    plain = "\n".join([title, "", *facts])
+    formatted = "\n".join([
+        f"<h3>{html.escape(title)}</h3>",
+        f"<p><b>{html.escape(artist)} — {html.escape(album)}</b></p>",
+        "<ul>",
+        *[f"<li>{html.escape(f)}</li>" for f in facts[1:]],
+        "</ul>",
+    ])
+    return plain, formatted
+
+
+def release_is_out(row: dict) -> bool:
+    dt = parse_date(row.get("release_date"))
+    return bool(dt and dt.date() == now_utc().date())
+
+
+def release_is_announcement(row: dict) -> bool:
+    dt = parse_date(row.get("release_date"))
+    return bool(dt and dt.date() > now_utc().date())
+
+
+def poll_release_watch(force: bool = False) -> dict:
+    baseline = (CFG.get("polling") or {}).get("baseline_existing_on_first_run", True)
+    first_run = not get_cursor("release_watch_baselined")
+    announced = 0
+    out = 0
+    skipped = 0
+    for row in release_rows():
+        row_id = row.get("id")
+        first_seen = row.get("first_seen_at") or ""
+        announce_key = f"{row_id}:{first_seen}"
+        if release_is_announcement(row) and not already_sent("release_announcement", announce_key):
+            if first_run and baseline and not force:
+                mark_sent("release_announcement", announce_key, row)
+            else:
+                plain, formatted = release_plain_and_html(row, "announce")
+                MATRIX.send_card(plain, formatted, None)
+                mark_sent("release_announcement", announce_key, row)
+                announced += 1
+                time.sleep(1)
+        elif not release_is_announcement(row):
+            skipped += 1
+
+        if release_is_out(row):
+            out_stamp = row.get("downloaded_at") or row.get("release_date") or row.get("last_checked_at") or ""
+            out_key = f"{row_id}:{out_stamp}:{row.get('status')}"
+            if already_sent("release_out", out_key):
+                continue
+            if first_run and baseline and not force:
+                mark_sent("release_out", out_key, row)
+            else:
+                plain, formatted = release_plain_and_html(row, "out")
+                MATRIX.send_card(plain, formatted, None)
+                mark_sent("release_out", out_key, row)
+                out += 1
+                time.sleep(1)
+    set_cursor("release_watch_baselined", now_utc().isoformat())
+    return {"ok": True, "announced": announced, "out": out, "skipped_not_current_or_future": skipped, "first_run": first_run}
+
+
+def loop_forever(name: str, interval: int, fn) -> None:
+    while True:
+        try:
+            result = fn()
+            log(f"{name}: {result}")
+        except Exception as e:
+            log(f"{name} failed: {e}\n{traceback.format_exc()}")
+        time.sleep(interval)
+
+
+app = Flask(__name__)
+
+
+@app.get("/healthz")
+def healthz():
+    return jsonify({"ok": True})
+
+
+@app.post("/run/navidrome-albums")
+def run_navidrome_albums():
+    payload = request.get_json(silent=True) or {}
+    return jsonify(poll_navidrome_albums(force=bool(payload.get("force"))))
+
+
+@app.post("/run/discovery-playlists")
+def run_discovery_playlists():
+    payload = request.get_json(silent=True) or {}
+    return jsonify(poll_discovery_playlists(force=bool(payload.get("force"))))
+
+
+@app.post("/run/release-watch")
+def run_release_watch():
+    payload = request.get_json(silent=True) or {}
+    return jsonify(poll_release_watch(force=bool(payload.get("force"))))
+
+
+def start_background() -> None:
+    polling = CFG.get("polling") or {}
+    tasks = [
+        ("navidrome_albums", int(polling.get("navidrome_album_interval_seconds") or 300), poll_navidrome_albums),
+        ("discovery_playlists", int(polling.get("discovery_playlist_interval_seconds") or 180), poll_discovery_playlists),
+        ("release_watch", int(polling.get("release_watch_interval_seconds") or 600), poll_release_watch),
+    ]
+    for name, interval, fn in tasks:
+        threading.Thread(target=loop_forever, args=(name, interval, fn), daemon=True).start()
+
+
+if __name__ == "__main__":
+    init_state()
+    try:
+        MATRIX.set_profile()
+    except Exception as e:
+        log(f"matrix profile setup failed: {e}")
+    if not NAV.ping():
+        log("warning: Navidrome ping failed")
+    start_background()
+    server = CFG.get("server") or {}
+    host = server.get("host") or "0.0.0.0"
+    port = int(server.get("port") or 18323)
+    try:
+        from waitress import serve
+
+        serve(app, host=host, port=port)
+    except Exception:
+        app.run(host=host, port=port)
--- a/assets/banner.png
+++ b/assets/banner.png
--- a/config.example.json
+++ b/config.example.json
@ -0,0 +1,42 @@
+{
+  "matrix": {
+    "homeserver": "https://tuwu.vespahomelab.ru",
+    "access_token": "REPLACE_ME",
+    "room_id": "!roomid:tuwu.vespahomelab.ru",
+    "user_id": "@aoi:tuwu.vespahomelab.ru"
+  },
+  "bot": {
+    "name": "Аой-тян",
+    "avatar_path": ""
+  },
+  "navidrome": {
+    "url": "http://192.168.31.216:4533",
+    "username": "shu",
+    "password": "REPLACE_ME",
+    "main_library_id": 1,
+    "anime_library_id": 3,
+    "discovery_library_id": 4,
+    "album_poll_size": 40
+  },
+  "discovery": {
+    "db_path": "/data/discovery.db",
+    "playlist_name_contains": "Discovery"
+  },
+  "metadata": {
+    "lastfm_api_key": "",
+    "musicbrainz_enabled": true,
+    "proxy": "",
+    "user_agent": "AoiNotifier/1.0 (https://tuwu.vespahomelab.ru)"
+  },
+  "polling": {
+    "navidrome_album_interval_seconds": 300,
+    "discovery_playlist_interval_seconds": 180,
+    "release_watch_interval_seconds": 600,
+    "baseline_existing_on_first_run": true
+  },
+  "server": {
+    "host": "0.0.0.0",
+    "port": 18323
+  }
+}
+
--- a/deploy/aoi.service
+++ b/deploy/aoi.service
@ -0,0 +1,17 @@
+[Unit]
+Description=Aoi Navidrome notifier
+After=network-online.target
+Wants=network-online.target
+
+[Service]
+Type=simple
+User=root
+Group=root
+WorkingDirectory=/storage/scripts/aoi
+ExecStart=/storage/scripts/watcher/.venv/bin/python /storage/scripts/aoi/aoi.py
+Restart=always
+RestartSec=5
+TimeoutStopSec=30
+
+[Install]
+WantedBy=multi-user.target
--- a/requirements.txt
+++ b/requirements.txt
@ -0,0 +1,3 @@
+flask==3.0.3
+requests==2.32.3
+waitress==3.0.2