Speed up serienstream season and episode loading

This commit is contained in:
2026-02-01 21:19:28 +01:00
parent fe79cca818
commit 3689aedd23
2 changed files with 62 additions and 29 deletions

5
.gitignore vendored
View File

@@ -6,3 +6,8 @@
# Build outputs
/dist/
# Local tests (not committed)
/tests/
/.pytest_cache/
/pytest.ini

View File

@@ -644,17 +644,14 @@ def resolve_redirect(target_url: str) -> Optional[str]:
def scrape_series_detail(
series_identifier: str,
max_seasons: Optional[int] = None,
*,
load_episodes: bool = True,
) -> List[SeasonInfo]:
_ensure_requests()
series_url = _series_root_url(_normalize_series_url(series_identifier))
_log_url(series_url, kind="SERIES")
_notify_url(series_url)
session = get_requests_session("serienstream", headers=HEADERS)
# Preflight ist optional; manche Umgebungen/Provider leiten die Startseite um.
try:
_get_soup(_get_base_url(), session=session)
except Exception:
pass
soup = _get_soup(series_url, session=session)
base_series_url = _series_root_url(_extract_canonical_url(soup, series_url))
@@ -673,6 +670,8 @@ def scrape_series_detail(
season_links = season_links[:max_seasons]
seasons: List[SeasonInfo] = []
for number, url in season_links:
episodes: List[EpisodeInfo] = []
if load_episodes:
season_soup = _get_soup(url, session=session)
episodes = _extract_episodes(season_soup)
seasons.append(SeasonInfo(number=number, url=url, episodes=episodes))
@@ -689,6 +688,7 @@ class SerienstreamPlugin(BasisPlugin):
def __init__(self) -> None:
self._series_results: Dict[str, SeriesResult] = {}
self._season_cache: Dict[str, List[SeasonInfo]] = {}
self._season_links_cache: Dict[str, List[SeasonInfo]] = {}
self._episode_label_cache: Dict[Tuple[str, str], Dict[str, EpisodeInfo]] = {}
self._catalog_cache: Optional[Dict[str, List[SeriesResult]]] = None
self._popular_cache: Optional[List[SeriesResult]] = None
@@ -830,19 +830,51 @@ class SerienstreamPlugin(BasisPlugin):
self._episode_label(info): info for info in season_info.episodes
}
def _ensure_season_links(self, title: str) -> List[SeasonInfo]:
cached = self._season_links_cache.get(title)
if cached is not None:
return list(cached)
series = self._series_results.get(title)
if not series:
return []
try:
seasons = scrape_series_detail(series.url, load_episodes=False)
except Exception as exc: # pragma: no cover - defensive logging
raise RuntimeError(f"Serienstream-Staffeln konnten nicht geladen werden: {exc}") from exc
self._season_links_cache[title] = list(seasons)
return list(seasons)
def _ensure_season_episodes(self, title: str, season_number: int) -> Optional[SeasonInfo]:
seasons = self._season_cache.get(title) or []
for season in seasons:
if season.number == season_number and season.episodes:
return season
links = self._ensure_season_links(title)
target = next((season for season in links if season.number == season_number), None)
if not target:
return None
try:
season_soup = _get_soup(target.url, session=get_requests_session("serienstream", headers=HEADERS))
season_info = SeasonInfo(number=target.number, url=target.url, episodes=_extract_episodes(season_soup))
except Exception as exc: # pragma: no cover - defensive logging
raise RuntimeError(f"Serienstream-Episoden konnten nicht geladen werden: {exc}") from exc
updated = [season for season in seasons if season.number != season_number]
updated.append(season_info)
updated.sort(key=lambda item: item.number)
self._season_cache[title] = updated
return season_info
def _lookup_episode(self, title: str, season_label: str, episode_label: str) -> Optional[EpisodeInfo]:
cache_key = (title, season_label)
cached = self._episode_label_cache.get(cache_key)
if cached:
return cached.get(episode_label)
seasons = self._ensure_seasons(title)
number = self._parse_season_number(season_label)
if number is None:
return None
for season_info in seasons:
if season_info.number == number:
season_info = self._ensure_season_episodes(title, number)
if season_info:
self._cache_episode_labels(title, season_label, season_info)
return self._episode_label_cache.get(cache_key, {}).get(episode_label)
return None
@@ -852,6 +884,7 @@ class SerienstreamPlugin(BasisPlugin):
if not query:
self._series_results.clear()
self._season_cache.clear()
self._season_links_cache.clear()
self._episode_label_cache.clear()
self._catalog_cache = None
return []
@@ -869,6 +902,7 @@ class SerienstreamPlugin(BasisPlugin):
raise RuntimeError(f"Serienstream-Suche fehlgeschlagen: {exc}") from exc
self._series_results = {result.title: result for result in results}
self._season_cache.clear()
self._season_links_cache.clear()
self._episode_label_cache.clear()
return [result.title for result in results]
@@ -901,27 +935,21 @@ class SerienstreamPlugin(BasisPlugin):
break
if not series:
return []
try:
seasons = scrape_series_detail(series.url)
except Exception as exc: # pragma: no cover - defensive logging
raise RuntimeError(f"Serienstream-Staffeln konnten nicht geladen werden: {exc}") from exc
seasons = self._ensure_season_links(title)
self._clear_episode_cache_for_title(title)
self._season_cache[title] = seasons
return seasons
self._season_cache[title] = list(seasons)
return list(seasons)
def seasons_for(self, title: str) -> List[str]:
seasons = self._ensure_seasons(title)
# Serienstream liefert gelegentlich Staffeln ohne Episoden (z.B. Parsing-/Layoutwechsel).
# Diese sollen im UI nicht als auswählbarer Menüpunkt erscheinen.
return [self._season_label(season.number) for season in seasons if season.episodes]
return [self._season_label(season.number) for season in seasons]
def episodes_for(self, title: str, season: str) -> List[str]:
seasons = self._ensure_seasons(title)
number = self._parse_season_number(season)
if number is None:
return []
for season_info in seasons:
if season_info.number == number:
season_info = self._ensure_season_episodes(title, number)
if season_info:
labels = [self._episode_label(info) for info in season_info.episodes]
self._cache_episode_labels(title, season, season_info)
return labels