Emby-kanalen som førsteklasses kilde: set/favorit virker uden YouTube-bibliotek

Biblioteket er nu valgfrit. Findes hverken kanal eller bibliotek, er svaret
ok=False, og fejler opslaget for en bruger, kastes der, så cyklus-sletningen
aldrig kører uden alle brugeres favoritter.

Claude-Session: https://claude.ai/code/session_012LCTmU2ewo8tnM1dByYAoG

templates/admin_yt.html

@@ -54,7 +54,7 @@
     <div class="setgrid" style="margin-top:.8rem">
       <label class="fld"><span>Karens efter set (dage)</span><input type="number" name="grace_days" min="0" max="365" value="{{ settings.grace_days }}"></label>
       <label class="fld"><span>Undertekst-sprog</span><input name="sub_langs" value="{{ settings.sub_langs }}"></label>
-      <label class="fld"><span>Emby-bibliotek</span><input name="library_name" value="{{ settings.library_name }}"></label>
+      <label class="fld"><span>Emby-kanal/bibliotek</span><input name="library_name" value="{{ settings.library_name }}"></label>
       <label class="fld"><span>Kvalitet (yt-dlp -f)</span><input name="quality" value="{{ settings.quality }}"></label>
     </div>
     <p style="margin-top:1rem;display:flex;gap:.6rem;flex-wrap:wrap">

yt_job.py

@@ -51,6 +51,20 @@ async def main():
         for e in cyc["errors"]:
             print("  cyklus-fejl:", e)
 
+    # Kanal-plugin'ets indeks: playlister (højst ugentligt pr. kanal) og så indekset.
+    try:
+        pl = await ytdl.opdater_playlister()
+        print(f"yt_job: playlister opdateret for {pl['kanaler']} kanaler, {pl['playlister']} i alt")
+        for e in pl["fejl"][:10]:
+            print("  playliste-fejl:", e)
+    except Exception as e:                                        # noqa: BLE001
+        print(f"yt_job: playlister kunne ikke opdateres: {e}")
+    try:
+        ix = ytdl.skriv_indeks()
+        print(f"yt_job: indeks: {ix['channels']} kanaler, {ix['videos']} videoer, {ix['playlists']} playlister")
+    except Exception as e:                                        # noqa: BLE001
+        print(f"yt_job: indekset kunne ikke skrives: {e}")
+
     res = await ytdl.run_cycle(dry_run=False)
     if res.get("error"):
         print("yt_job FEJL:", res["error"])

ytdl.py

@@ -11,12 +11,16 @@ biblioteket — fx det der lå i /slow/shared/yt i forvejen — er urørligt.
 """
 import asyncio
 import fcntl
+import functools
 import json
 import os
 import re
 import shutil
+import subprocess
 import time
 import uuid
+import xml.etree.ElementTree as ET
+from datetime import datetime
 from xml.sax.saxutils import escape
 
 import httpx
@@ -372,6 +376,8 @@ def _ryd_tomme_mapper(saeson_dir: str) -> None:
         for f in KANAL_FILER:
             if os.path.exists(os.path.join(kanal, f)):
                 os.remove(os.path.join(kanal, f))
+        # Playlisternes miniaturer (skrevet af opdater_playlister) hører også til kanalen.
+        shutil.rmtree(os.path.join(kanal, ".playlists"), ignore_errors=True)
         for d in os.listdir(kanal):
             dd = os.path.join(kanal, d)
             if d.startswith("Season ") and os.path.isdir(dd) and not os.listdir(dd):
@@ -395,6 +401,312 @@ def _slet_video(path: str) -> int:
     return size
 
 
+# ---------------- indeks til Emby-kanal-plugin'et ----------------
+# Emby kan kun vise serie → sæson → afsnit. Kanal-plugin'et (agw/emby-yt-kanaler)
+# viser i stedet Kanal → [Alle videoer | Playlister → Playliste] → videoer, hvor
+# en video kan ligge i flere playlister. Det læser dette indeks, som portalen
+# skriver ud fra disken (NFO'er og billeder) og playliste-cachen. Plugin'et har
+# ingen egen logik for YouTube; alt det ligger her.
+INDEKS_NAVN = ".weircon-index.json"
+PLAYLISTER_PATH = config.DATA / "yt_playlister.json"
+PLAYLISTE_MAKS_ALDER = 7 * 86400
+_VIDEO_ENDELSER = (".mp4", ".mkv", ".webm")
+_SRT_RE = re.compile(r"\.([A-Za-z0-9_-]+)\.(srt|vtt)$")
+
+
+def indeks_sti() -> str:
+    return os.path.join(config.YT_DOWNLOAD_DIR, INDEKS_NAVN)
+
+
+def _laes_nfo(sti: str) -> dict:
+    """Felter fra en NFO (episodedetails/tvshow): tag -> tekst, uniqueid pr. type."""
+    try:
+        rod = ET.parse(sti).getroot()
+    except (OSError, ET.ParseError):
+        return {}
+    ud = {}
+    for el in rod:
+        if el.tag == "uniqueid":
+            ud[f"uniqueid:{el.get('type') or ''}"] = (el.text or "").strip()
+        elif el.tag not in ud:
+            ud[el.tag] = (el.text or "").strip()
+    return ud
+
+
+@functools.lru_cache(maxsize=1024)
+def _varighed_mp4(sti: str, mtime: float):
+    try:
+        from mutagen import File as _MFile
+        m = _MFile(sti)
+        return int(m.info.length) if m and m.info and m.info.length else None
+    except Exception:                                             # noqa: BLE001
+        return None
+
+
+FFPROBE_PATH = config.DATA / "yt_ffprobe.json"
+_ffprobe = {"data": None, "aendret": False}
+
+
+def _brok(v):
+    try:
+        t, n = str(v).split("/")
+        return round(float(t) / float(n), 3) if float(n) else None
+    except (ValueError, ZeroDivisionError):
+        return None
+
+
+def _heltal(v):
+    try:
+        return int(v)
+    except (TypeError, ValueError):
+        return None
+
+
+def _stroemme(sti: str, st_) -> dict:
+    """Video-/lydspor (codec, opløsning, bitrate) via ffprobe, så Emby-plugin'et kan
+    sige til apps at filen kan afspilles direkte. Caches pr. sti+mtime+størrelse."""
+    if _ffprobe["data"] is None:
+        _ffprobe["data"] = _load(FFPROBE_PATH, {}) or {}
+    noegle = f"{sti}|{int(st_.st_mtime)}|{st_.st_size}"
+    if noegle in _ffprobe["data"]:
+        return _ffprobe["data"][noegle]
+    ud = {"bitrate": None, "streams": []}
+    try:
+        r = subprocess.run(
+            [shutil.which("ffprobe") or "ffprobe", "-v", "error", "-of", "json", "-show_entries",
+             "stream=codec_type,codec_name,profile,width,height,pix_fmt,avg_frame_rate,bit_rate,"
+             "channels,sample_rate:stream_disposition=attached_pic:stream_tags=language:format=bit_rate", sti],
+            capture_output=True, text=True, timeout=60)
+        d = json.loads(r.stdout or "{}")
+        ud["bitrate"] = _heltal((d.get("format") or {}).get("bit_rate"))
+        for s_ in d.get("streams", []):
+            if (s_.get("disposition") or {}).get("attached_pic"):
+                continue                      # det indlejrede coverbillede er ikke et videospor
+            if s_.get("codec_type") == "video":
+                ud["streams"].append({"type": "video", "codec": s_.get("codec_name"), "profile": s_.get("profile"),
+                                      "width": s_.get("width"), "height": s_.get("height"),
+                                      "fps": _brok(s_.get("avg_frame_rate")), "bitrate": _heltal(s_.get("bit_rate")),
+                                      "pix_fmt": s_.get("pix_fmt")})
+            elif s_.get("codec_type") == "audio":
+                ud["streams"].append({"type": "audio", "codec": s_.get("codec_name"), "profile": s_.get("profile"),
+                                      "channels": s_.get("channels"), "sample_rate": _heltal(s_.get("sample_rate")),
+                                      "bitrate": _heltal(s_.get("bit_rate")),
+                                      "language": (s_.get("tags") or {}).get("language")})
+    except Exception:                                             # noqa: BLE001
+        return ud                             # prøv igen næste gang
+    _ffprobe["data"][noegle] = ud
+    _ffprobe["aendret"] = True
+    return ud
+
+
+def _laes_playlister() -> dict:
+    return _load(PLAYLISTER_PATH, {}) or {}
+
+
+def skriv_indeks() -> dict:
+    """Skriv indekset (DOWNLOAD_DIR/.weircon-index.json) ud fra disken. Ingen netværk.
+    Returnerer en lille statistik."""
+    rod = config.YT_DOWNLOAD_DIR
+    kanaler, videoer = [], {}
+    try:
+        navne = sorted(os.listdir(rod), key=str.lower)
+    except OSError:
+        navne = []
+    for navn in navne:
+        kdir = os.path.join(rod, navn)
+        if navn.startswith(".") or not os.path.isdir(kdir):
+            continue
+        tv = _laes_nfo(os.path.join(kdir, "tvshow.nfo"))
+        kid = tv.get("uniqueid:youtube") or f"navn:{navn}"
+        mine = []
+        for d, dirs, fs in os.walk(kdir):
+            dirs[:] = [x for x in dirs if not x.startswith(".")]
+            for f in fs:
+                if not f.lower().endswith(_VIDEO_ENDELSER) or f.startswith("."):
+                    continue
+                sti = os.path.join(d, f)
+                stem = f.rsplit(".", 1)[0]
+                nfo = _laes_nfo(os.path.join(d, stem + ".nfo"))
+                m = re.search(r"\[([A-Za-z0-9_-]{11})\]$", stem)
+                vid = nfo.get("uniqueid:youtube") or (m.group(1) if m else "")
+                if not vid:
+                    continue
+                st_ = os.stat(sti)
+                thumb = os.path.join(d, stem + "-thumb.jpg")
+                subs = []
+                for side in sorted(fs):
+                    if side.startswith(stem + "."):
+                        sm = _SRT_RE.search(side[len(stem):])
+                        if sm:
+                            subs.append({"path": os.path.join(d, side), "lang": sm.group(1)})
+                videoer[vid] = {
+                    "title": nfo.get("title") or stem, "channel": kid, "path": sti,
+                    "thumb": thumb if os.path.exists(thumb) else None,
+                    "description": nfo.get("plot") or "",
+                    "upload_date": nfo.get("aired") or nfo.get("premiered") or "",
+                    "duration": _varighed_mp4(sti, st_.st_mtime), "added": int(st_.st_mtime),
+                    "subtitles": subs, **_stroemme(sti, st_)}
+                mine.append(vid)
+        if not mine:
+            continue
+        mine.sort(key=lambda v: (videoer[v]["upload_date"], videoer[v]["added"]), reverse=True)
+        poster, fanart = os.path.join(kdir, "poster.jpg"), os.path.join(kdir, "fanart.jpg")
+        kanaler.append({"id": kid, "name": tv.get("title") or navn, "description": tv.get("plot") or "",
+                        "poster": poster if os.path.exists(poster) else None,
+                        "fanart": fanart if os.path.exists(fanart) else None,
+                        "videos": mine, "playlists": [], "_dir": kdir})
+
+    # Playlister fra cachen; kun hentede videoer, i YouTubes rækkefølge.
+    cache = _laes_playlister()
+    for k in kanaler:
+        for pl in (cache.get(k["id"]) or {}).get("playlists", []):
+            vids = [v for v in pl.get("videos", []) if v in videoer]
+            if not vids:
+                continue
+            thumb = os.path.join(k["_dir"], ".playlists", f"{pl['id']}.jpg")
+            k["playlists"].append({"id": pl["id"], "title": pl.get("title") or pl["id"],
+                                   "thumb": thumb if os.path.exists(thumb) else None, "videos": vids})
+        k.pop("_dir", None)
+
+    data = {"version": 1, "generated": datetime.now().astimezone().isoformat(timespec="seconds"),
+            "channels": kanaler, "videos": videoer}
+    sti = indeks_sti()
+    tmp = sti + ".tmp"
+    with open(tmp, "w", encoding="utf-8") as f:
+        json.dump(data, f, ensure_ascii=False, indent=1)
+    os.chmod(tmp, 0o644)
+    os.replace(tmp, sti)
+    if _ffprobe["aendret"] and _ffprobe["data"] is not None:
+        stier = {v["path"] for v in videoer.values()}
+        _ffprobe["data"] = {k: v for k, v in _ffprobe["data"].items() if k.split("|", 1)[0] in stier}
+        _save(FFPROBE_PATH, _ffprobe["data"])
+        _ffprobe["aendret"] = False
+    return {"channels": len(kanaler), "videos": len(videoer),
+            "playlists": sum(len(k["playlists"]) for k in kanaler)}
+
+
+async def _ytdlp_json(url: str) -> dict:
+    """yt-dlp --flat-playlist -J med samme nye forsøg som hentning (403/429)."""
+    err = ""
+    for pause, _ in FORSOEG:
+        if pause:
+            await asyncio.sleep(pause)
+        proc = await asyncio.create_subprocess_exec(
+            config.YT_DLP_BIN, "--flat-playlist", "-J", "--no-warnings", "--ignore-errors", url,
+            stdout=asyncio.subprocess.PIPE, stderr=asyncio.subprocess.PIPE)
+        out, e = await proc.communicate()
+        try:
+            return json.loads(out.decode("utf-8", "replace") or "{}")
+        except ValueError:
+            err = e.decode("utf-8", "replace")
+            if not midlertidig(err):
+                break
+    raise RuntimeError((err.strip().splitlines() or ["ukendt fejl"])[-1][:200])
+
+
+def _stoerste_thumb(thumbs: list) -> str:
+    gode = [t for t in thumbs or [] if t.get("url")]
+    gode.sort(key=lambda t: (t.get("width") or 0) * (t.get("height") or 0), reverse=True)
+    return gode[0]["url"] if gode else ""
+
+
+async def _hent_playliste(pl_url: str) -> dict:
+    d = await _ytdlp_json(pl_url)
+    return {"id": d.get("id") or "", "title": d.get("title") or "",
+            "channel_id": d.get("channel_id") or d.get("uploader_id") or "",
+            "thumb_url": _stoerste_thumb(d.get("thumbnails")),
+            "videos": [e["id"] for e in (d.get("entries") or []) if e and e.get("id")]}
+
+
+async def opdater_playlister(kun_nye: bool = False) -> dict:
+    """Hent kanalernes playlister (og deres video-id'er) til cachen.
+
+    Hver kanal opdateres højst hver 7. dag; kun_nye=True tager kun kanaler uden
+    cache. Portalens egne playliste-job og -abonnementer tages altid med og
+    knyttes til playlistens ejer-kanal. Bagefter skrives indekset.
+    """
+    res = {"kanaler": 0, "playlister": 0, "fejl": []}
+    cache = _laes_playlister()
+    rod = config.YT_DOWNLOAD_DIR
+    kanal_dirs = {}
+    for navn in sorted(os.listdir(rod)):
+        kdir = os.path.join(rod, navn)
+        if navn.startswith(".") or not os.path.isdir(kdir):
+            continue
+        kid = _laes_nfo(os.path.join(kdir, "tvshow.nfo")).get("uniqueid:youtube")
+        if kid:
+            kanal_dirs[kid] = kdir
+    nu = time.time()
+
+    for kid in kanal_dirs:
+        gl = cache.get(kid) or {}
+        if kun_nye and gl:
+            continue
+        if gl and nu - gl.get("updated", 0) < PLAYLISTE_MAKS_ALDER:
+            continue
+        try:
+            d = await _ytdlp_json(f"https://www.youtube.com/channel/{kid}/playlists")
+            pls = []
+            for e in d.get("entries") or []:
+                if not e or not e.get("id"):
+                    continue
+                await asyncio.sleep(2)            # skån YouTube
+                try:
+                    pls.append(await _hent_playliste(
+                        e.get("url") or f"https://www.youtube.com/playlist?list={e['id']}"))
+                except Exception as ex:                           # noqa: BLE001
+                    res["fejl"].append(f"{e.get('title') or e['id']}: {ex}")
+            # Bevar egne playlister (fra job/abonnementer) der ikke står på kanalens fane.
+            egne = [p for p in gl.get("playlists", []) if p.get("egen")]
+            ids = {p["id"] for p in pls}
+            cache[kid] = {"updated": nu, "playlists": pls + [p for p in egne if p["id"] not in ids]}
+            res["kanaler"] += 1
+        except Exception as ex:                                   # noqa: BLE001
+            res["fejl"].append(f"{kid}: {ex}")
+        await asyncio.sleep(2)
+
+    # Portalens egne playlister (job med list= og playliste-abonnementer).
+    st = get_state()
+    urls = {j["url"] for j in st["jobs"].values() if "list=" in (j.get("url") or "")}
+    urls |= {s["url"] for s in st["subs"].values() if s.get("kind") == "playlist"}
+    kendte = {p["id"] for c in cache.values() for p in c.get("playlists", [])}
+    for url in sorted(urls):
+        m = re.search(r"list=([A-Za-z0-9_-]+)", url)
+        if m and m.group(1) in kendte:
+            # Allerede i en kanals cache; markér den som egen, så den bevares.
+            for c in cache.values():
+                for p in c.get("playlists", []):
+                    if p["id"] == m.group(1):
+                        p["egen"] = True
+            continue
+        try:
+            await asyncio.sleep(2)
+            pl = await _hent_playliste(
+                f"https://www.youtube.com/playlist?list={m.group(1)}" if m else url)
+            pl["egen"] = True
+            ejer = pl.get("channel_id") or ""
+            c = cache.setdefault(ejer, {"updated": 0, "playlists": []})
+            c["playlists"] = [p for p in c["playlists"] if p["id"] != pl["id"]] + [pl]
+        except Exception as ex:                                   # noqa: BLE001
+            res["fejl"].append(f"{url}: {ex}")
+
+    # Miniaturer til <Kanal>/.playlists/<id>.jpg (kun når de mangler).
+    for kid, c in cache.items():
+        kdir = kanal_dirs.get(kid)
+        if not kdir:
+            continue
+        for p in c.get("playlists", []):
+            mappe = os.path.join(kdir, ".playlists")
+            sti = os.path.join(mappe, f"{p['id']}.jpg")
+            if p.get("thumb_url") and not os.path.exists(sti):
+                os.makedirs(mappe, exist_ok=True)
+                await _hent_billede(p["thumb_url"], sti)
+            res["playlister"] += 1
+    _save(PLAYLISTER_PATH, cache)
+    skriv_indeks()
+    return res
+
+
 # Fejl der ofte går over af sig selv: YouTube afviser en enkelt forespørgsel
 # (403/429), eller forbindelsen falder. yt-dlp gentager ikke selv en 403.
 _MIDLERTIDIG = ("403", "forbidden", "429", "too many requests", "timed out",
@@ -841,35 +1153,79 @@ def _marked_played(ud: dict) -> bool:
     return bool((ud or {}).get("Played"))
 
 
+async def _kanal_brugerdata(c, users: list, merged: dict, navn: str) -> bool:
+    """Læg set/favorit fra Emby-kanalen (kanal-plugin'et) ind i `merged`, nøglet på
+    filens sti. Kanalens elementer bærer video-id'et i ProviderIds["youtube"]; stien
+    findes via indekset. Samme video ligger i "Alle videoer" og i playlister, og
+    kopierne deler brugerdata, så det er ligegyldigt hvilken vi læser.
+
+    False hvis kanalen eller indekset mangler. Fejler opslaget for en bruger, kastes
+    der: uden alle brugeres favoritter må slettereglen ikke køre."""
+    try:
+        with open(indeks_sti(), encoding="utf-8") as f:
+            stier = {vid: os.path.realpath(v["path"]) for vid, v in json.load(f)["videos"].items()}
+    except (OSError, ValueError, KeyError):
+        return False
+    if not users:
+        return False
+    kanaler = await _emby(c, "/Channels", UserId=users[0])
+    kanal = next((k for k in (kanaler.get("Items") or [])
+                  if (k.get("Name") or "").lower() == navn.lower()), None)
+    if not kanal:
+        return False
+    for uid in users:
+        data = await _emby(c, f"/Users/{uid}/Items", ParentId=kanal["Id"],
+                           Recursive="true", IncludeItemTypes="Video", Fields="ProviderIds,UserData")
+        for it in data.get("Items", []):
+            pid = it.get("ProviderIds") or {}
+            vid = pid.get("youtube") or pid.get("YouTube") or pid.get("Youtube")
+            sti = stier.get(vid or "")
+            if not sti:
+                continue
+            m = merged.setdefault(sti, {"id": it["Id"], "name": it.get("Name") or os.path.basename(sti),
+                                        "path": sti, "played": False, "favorite": False})
+            ud = it.get("UserData") or {}
+            if _marked_played(ud):
+                m["played"] = True
+            if ud.get("IsFavorite"):
+                m["favorite"] = True
+    return True
+
+
 async def emby_yt_items() -> dict:
-    """Alle videoer i YouTube-biblioteket med set/favorit aggregeret over ALLE brugere."""
+    """Alle videoer i Emby med set/favorit aggregeret over ALLE brugere.
+
+    Kilden er kanalen fra kanal-plugin'et og/eller et bibliotek med samme navn
+    (standard "YouTube"). Findes ingen af dem, er svaret ok=False, så intet slettes."""
     settings = get_settings()
+    navn = settings.get("library_name") or "YouTube"
     out = {"ok": False, "items": [], "error": ""}
     try:
         async with httpx.AsyncClient(timeout=45) as c:
-            lib = await _library_id(c, settings.get("library_name") or "YouTube")
-            if not lib:
-                out["error"] = (f"Fandt ikke Emby-biblioteket "
-                                f"'{settings.get('library_name')}'")
-                return out
             users = [u["Id"] for u in await _emby(c, "/Users")]
             merged = {}
-            for uid in users:
-                data = await _emby(c, f"/Users/{uid}/Items", ParentId=lib,
-                                   Recursive="true", Fields="Path,UserData")
-                for it in data.get("Items", []):
-                    path = it.get("Path") or ""
-                    if not path:
-                        continue
-                    key = os.path.realpath(path)
-                    m = merged.setdefault(key, {
-                        "id": it["Id"], "name": it.get("Name") or os.path.basename(path),
-                        "path": key, "played": False, "favorite": False})
-                    ud = it.get("UserData") or {}
-                    if _marked_played(ud):
-                        m["played"] = True
-                    if ud.get("IsFavorite"):
-                        m["favorite"] = True
+            lib = await _library_id(c, navn)
+            if lib:
+                for uid in users:
+                    data = await _emby(c, f"/Users/{uid}/Items", ParentId=lib,
+                                       Recursive="true", Fields="Path,UserData")
+                    for it in data.get("Items", []):
+                        path = it.get("Path") or ""
+                        if not path:
+                            continue
+                        key = os.path.realpath(path)
+                        m = merged.setdefault(key, {
+                            "id": it["Id"], "name": it.get("Name") or os.path.basename(path),
+                            "path": key, "played": False, "favorite": False})
+                        ud = it.get("UserData") or {}
+                        if _marked_played(ud):
+                            m["played"] = True
+                        if ud.get("IsFavorite"):
+                            m["favorite"] = True
+            kanal = await _kanal_brugerdata(c, users, merged, navn)
+            if not lib and not kanal:
+                out["error"] = f"Fandt hverken Emby-kanalen eller -biblioteket '{navn}'"
+                return out
             out["items"] = list(merged.values())
             out["ok"] = True
     except Exception as e:                                        # noqa: BLE001
@@ -878,6 +1234,11 @@ async def emby_yt_items() -> dict:
 
 
 async def refresh_library():
+    # Kanal-plugin'ets indeks følger altid disken: efter hentning og sletning.
+    try:
+        skriv_indeks()
+    except Exception:                                             # noqa: BLE001
+        pass
     try:
         async with httpx.AsyncClient(timeout=30) as c:
             await c.post(f"{config.EMBY_URL}/emby/Library/Refresh",