- Fix hybrid extract not using maxParallelExtract setting (was hardcoded to 1)
- Fix "Warten auf Parts" label shown for items whose downloads are already complete
- Update hybrid extract progress handler to support parallel archive tracking
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- WRONG_PASSWORD JVM error now falls back to legacy UnRAR extractor
- Added masked password logging for JVM and legacy extractors
- Per-attempt password logging shows which passwords are tried and in what order
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
maxParallelExtract now controls how many archives extract simultaneously
within a single package (e.g. 4 episodes at once). Packages still
extract sequentially (one package at a time) to focus I/O. Progress
handler updated to track multiple active archives independently.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Previously maxParallelExtract allowed multiple packages to extract
simultaneously, splitting I/O across packages. Now packages extract
one at a time in packageOrder so each package finishes faster.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When SevenZipJBinding reports "Archive file can't be opened with any
of the registered codecs", the extractor now falls back to legacy
UnRAR instead of failing immediately. Previously, backend mode "jvm"
(the production default) only allowed fallback for UNSUPPORTEDMETHOD.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Scrapes the Mega-Debrid profile page to display username, premium status,
remaining days, and loyalty points. New "Account prüfen" button in Settings > Accounts.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Encrypt sensitive credentials (tokens, passwords) in backup exports
using AES-256-GCM with PBKDF2 key derivation from OS username
- Backup format v2 with backwards-compatible v1 import
- Show dialog to create non-existent directories when changing
outputDir, extractDir, or mkvLibraryDir settings
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Root cause: SevenZip.initSevenZipFromPlatformJAR() was never called, so
native compression codecs (RAR5, LZMA2, etc.) were not loaded. Archives
could be opened (header parsing is pure Java) but all extractSlow() calls
returned UNSUPPORTEDMETHOD because no native decoder was available.
- Add ensureSevenZipInitialized() with lazy init before extraction
- Pass password to extractSlow(outStream, password) for RAR5 compatibility
- Add UNSUPPORTEDMETHOD -> legacy fallback in extractor.ts as safety net
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Java's getCanonicalFile() resolves subst drives inconsistently,
causing secureResolve() to falsely block valid filenames. JVM handles
long paths natively so subst is only needed for legacy UnRAR/7z.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The JVM sidecar class files were packed inside app.asar where Java
cannot access them. asarUnpack extracts them to app.asar.unpacked/.
Default backend changed from auto to jvm (no legacy fallback).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- New JVM sidecar (resources/extractor-jvm/) using SevenZipJBinding for
RAR/7z/TAR and Zip4j for ZIP multipart, matching JDownloader 2 stack
- Auto/JVM/Legacy backend modes via RD_EXTRACT_BACKEND env variable
- Fallback to legacy UnRAR/7z when JVM runtime unavailable
- Fix isJvmRuntimeMissingError false positives on valid extraction errors
- Cache JVM layout resolution to avoid repeated filesystem checks
- Route nested ZIP extraction through JVM backend consistently
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- setWindowsBackgroundIO (Very Low I/O) now only applied in hybrid mode,
not for all extractions (was causing massive slowdown)
- Hybrid threads changed from -mt1 to half CPU count (e.g. -mt4 on 8-core)
- Move retry count (R9, R22 etc.) from status text to tooltip only
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- normalizeSessionStatuses: reset all queued items to "Wartet" instead of
only checking a few specific patterns (missed Retry, Unrestrict-Fehler etc.)
- Reset completed items with stale extraction status to "Fertig (size)"
- stop(): reset all non-finished items to queued/"Wartet" and packages to queued
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add disk space check before extraction (aborts if insufficient space)
- Add single-level nested archive extraction (archives inside archives)
- Blacklist .iso/.img/.bin/.dmg from nested extraction
- Set real Windows I/O priority (Very Low) on UnRAR via NtSetInformationProcess
- Reduce UnRAR threads to -mt1 during hybrid extraction
- Fix double episode renaming (s01e01e02 pattern)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Items with transient status texts like Provider-Cooldown, Warte auf
Daten, Verbindungsfehler are reset to "Wartet" when the app restarts,
so they don't show misleading status from a previous session.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Debounce: simultaneous failures within 2s count as 1 failure
(prevents 8 parallel unrestrict failures from instant-triggering)
- Raise threshold from 8 to 20 consecutive failures before cooldown
- Escalation tiers: 20→30s, 35→60s, 50→120s, 80+→300s
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Pause now aborts active extractions (previously extraction continued
during pause, showing wrong status like Provider-Cooldown)
- Unpause clears provider circuit breaker for fresh start
- Post-processing status checks account for global pause state
- Reduce stall detection timeout from 30s to 15s for faster retry
- Reduce stall retry base delay from 500ms to 300ms
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add pause check at top of processItem retry loop so items show
"Pausiert" instead of "Provider-Cooldown" when paused
- Lower extraction process priority from BELOW_NORMAL to IDLE
(IDLE_PRIORITY_CLASS on Windows also lowers I/O priority, reducing
disk contention between extraction and active downloads)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Make saveSettings async to stop blocking the event loop during downloads
- Add 120ms minimum gap for forced state emissions to prevent rapid-fire IPC
- Fix circuit breaker feedback loop: reset failure count after cooldown expires
- Add 120s time-decay for failure counter (transient bursts don't snowball)
- Raise circuit breaker threshold from 5 to 8 consecutive failures
- Stop counting network stalls as provider failures
- Items without a provider only check primary provider cooldown, not all
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- New "Jetzt entpacken" right-click option: triggers extraction for
completed packages regardless of paused/stopped state
- Fix 5-10s freeze when pressing Start after Pause: recoverRetryableItems
was calling fs.stat on every item (474+); now only checks failed/completed
- Full IPC pipeline: extractNow in manager, controller, preload, renderer
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
WinRAR doesn't support \?\ prefix (interprets it as UNC network path).
Replace with subst drive mapping: maps targetDir to a short drive letter
(Z:, Y:, etc.) before extraction, then removes mapping after. This keeps
total paths under 260 chars even when archives contain deep internal
directory structures.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add longPathForWindows() helper that prefixes extract target directories
with \?\ on Windows, bypassing the 260-char MAX_PATH limit. Applied to
both WinRAR/UnRAR and 7z arguments. Fixes "Die Syntax für den
Dateinamen, Verzeichnisnamen" errors when archive internal directories
create deeply nested output paths.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Right-click packages with extraction errors shows "Extraktion
wiederholen" option to manually retry
- Increase WinRAR error text from 240 to 500 chars for better
diagnostics in logs
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- triggerPendingExtractions() now runs when unpausing, so packages
with extraction errors are automatically retried
- executeDeleteSelection no longer depends on snapshot objects
(prevents unnecessary re-renders with large queues)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Session counter now uses sessionDownloadedBytes (in-memory counter)
instead of summing completed items. Removing packages after extraction
no longer resets the session total.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Save settings every 30s during active downloads (not just session)
- Force settings save on shutdown and run finish
- Preserve live totalDownloadedAllTime when user saves settings
(app-controller's stale copy no longer overwrites the counter)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Pause button is now one-way (orange glow when paused, disabled when
already paused). Start button resumes from pause.
- Fix hybrid extraction attempting incomplete multi-part archives when
paused: disk-fallback now blocks any non-terminal item status, not
just downloading/validating/integrity_check.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Dynamic left padding based on measured label width (no more cut-off numbers)
- Speed history ref lifted to App so chart data survives tab switches
- Pause button uses optimistic UI update for instant visual feedback
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Track totalDownloadedAllTime in settings (persists across restarts)
- Track sessionDownloadedBytes for current app session
- Status bar shows both: Session + Gesamt
- Statistics section shows Heruntergeladen (Session) + (Gesamt)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Track packageId in speed events so package speed uses same 3-second
window as global speed (fixes mismatch between package and status bar)
- Add packageSpeedBps to UiSnapshot, computed from speed events
- Context menu delete actions now respect confirmDeleteSelection setting
- Progress bar visible even when package is collapsed
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Fix Ctrl+Click: mousedown no longer immediately adds item, preventing
onClick from toggling it back off. Drag-select still works via mouseenter.
- Delete key removes selected items/packages with JDownloader-style
confirmation dialog showing count and remaining items.
- "Nicht mehr anzeigen" checkbox disables future confirmations.
- New setting "Vor dem Löschen bestätigen" under Allgemein.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Dead link detection:
- Mega-Web: parse hoster error messages (hosterNotAvailable, etc.) from HTML
and throw specific error instead of returning null
- MegaDebridClient: stop retrying on permanent hoster errors
- download-manager: isPermanentLinkError() immediately fails items with dead
links instead of retrying forever
Extraction race condition:
- package_done cleanup policy checked if all items were "completed" (downloaded)
but not if they were "extracted" — removing the package before the last
episode could be extracted
- Both applyCompletedCleanupPolicy and applyPackageDoneCleanup now guard
against premature removal when autoExtract is enabled
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- HTTP 200 on resume: detect server ignoring Range header, write in truncate mode
instead of appending (prevents doubled/corrupted files)
- HTTP 416 without Content-Range: assume complete if >1MB exists instead of
deleting potentially multi-GB finished files
- Stream handle leak: explicit destroy() after finally to prevent fd exhaustion
- Drain timeout: don't abort controller on disk backpressure, let inner retry
loop handle it instead of escalating to full stall pipeline
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Provider circuit breaker: track consecutive failures per provider with
escalating cooldowns (30s/60s/120s/300s), auto-invalidate Mega-Debrid
session on cooldown
- Escalating backoff: retry delays now scale up to 120s (was 30s max),
unrestrict backoff exponential instead of linear 15s cap
- Shelve logic: after 15 consecutive failures, item pauses 5 min with
counter halving for gradual recovery
- Periodic soft-reset: every 10 min, reset stale retry counters (>10 min
queued) and old provider failures (>15 min), acts like mini-restart
- Mega-Debrid queue timeout: 90s wait limit in runExclusive to prevent
cascade blocking behind stuck calls
- Provider-cooldown-aware retry delays: items wait for provider cooldown
instead of retrying against broken service
- Fix: reconnect/package_toggle now persist retry counters (previously
lost on interruption, defeating shelve logic)
- Mega-Debrid generate: tighter timeouts, progressive reload backoff,
hoster retry limit (5x max)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Left half (blue) shows download progress, right half (green) shows
extraction progress. When no extraction is active, full bar is blue.
Full bar = 50% blue (all downloaded) + 50% green (all extracted).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
