theantipopau/slimarr v1.3.0.0

Slimarr - v1.3.0.0

on GitHub

[1.3.0.0] - 2026-05-13

Search diagnostics and degraded-pipeline safety

• Added end-to-end search diagnostics for Prowlarr and direct Newznab searches:
  request URLs with secrets redacted, response status, latency, raw/parsed result counts,
  parser failures, timeouts, malformed payloads, indexer reliability, and rejection summaries.
• Added a dedicated Search Diagnostics page with live requests, indexer responses,
  filtered-vs-accepted counts, failure heatmap, last successful search, reliability metrics,
  and a manual Search Test Harness with raw payload inspection.
• Added degraded-search detection so automation no longer appears healthy while searches are
  effectively failing:
  100 consecutive zero-result searches, repeated all-provider failures, missing search
  providers, malformed payloads, and category mismatch warnings are surfaced to the UI.
• Hardened Newznab parsing so auth errors and malformed XML are no longer silently treated as
  empty result sets.
• Added regression tests for Prowlarr request construction, Newznab parsing, empty results,
  malformed payloads, auth failures, timeout diagnostics, category extraction, and degraded
  zero-result detection.
• Tightened diagnostics security and retention:
  raw payload previews are truncated, API keys/tokens/passwords/auth headers are redacted,
  diagnostics history is bounded, and search diagnostics are included in support bundles.
• Split degraded-search state into warnings and blocking failures so 100+ zero-result searches
  no longer pause automation by itself, while repeated all-provider failures still block cycles.
• Improved Newznab compatibility by parsing namespaced or unprefixed Newznab attributes and
  namespaced error payloads.
• Added persisted diagnostics history stored on disk with pagination and basic text search via
  GET /system/search-diagnostics/history.
• Expanded Search Diagnostics UI with persisted-history controls (filter, search, pagination)
  so incident review is no longer limited to in-memory session events.
• Included persisted diagnostics history in support bundles (system.search.diagnostics.history.json).
• Hardened diagnostics redaction to also mask bearer/basic credential payloads in quoted
  authorization fields.

Media Intelligence and automation safety

• Added Quality Intelligence V2 parsing for source type, resolution, codec, HDR/Dolby Vision,
  audio codec/channels, language markers, subtitle risk markers, PROPER/REPACK, release group,
  and low-quality source detection.
• Added Media Health scoring for candidate releases and local media with ratings:
  Excellent, Good, Acceptable, Risky, and Reject.
• Added Dolby Vision safety mode defaults:
  comparison.avoid_dolby_vision: true and
  comparison.allow_dolby_vision_with_hdr_fallback: false.
• Added language/audio safety defaults:
  comparison.require_english_audio: true, optional dual/multi-audio rejection, and
  hardcoded subtitle rejection.
• Comparison decisions now expose media health score/rating/reasons in search results,
  Movie Detail candidate drawers, and decision audit responses.
• Quality upgrades can now accept a larger file only when the existing copy is clearly poor
  quality and the candidate is a bounded good-source 1080p upgrade.
• Added regression tests for Dolby Vision filtering, language/subtitle safeguards,
  low-quality source rejection, bounded quality upgrades, diagnostics redaction, and
  retention limits.
• Added MediaInfo-backed local stream probing during library scans to enrich missing
  resolution/video codec/audio codec/bitrate metadata from actual files.
• Improved library poster rendering performance with viewport-aware staged image loading,
  skeleton placeholders, and async image decoding for large libraries.