Skip to content

GLOSSARY.md

Latest commit

 

History

History
55 lines (36 loc) · 3.58 KB

GLOSSARY.md

File metadata and controls

55 lines (36 loc) · 3.58 KB
Raw

LosslessCut Domain Model & Glossary (GLOSSARY.md)

This document establishes the ubiquitous domain language and architectural aggregates for LosslessCut.

1. Core Editing Domain Aggregates

EditingSession (Aggregate Root)

The central domain aggregate managing an active NLE editing workspace for one or more media clips.

  • Interface: Exposes atomic state mutations (splitSegment, updateSegmentBounds, toggleDiscard, undo, redo, save, restore) and reactive state flows (EditingSessionState).
  • Invariants: Enforces minimum segment duration (100ms), single-clip discard restrictions, undo stack limits (30 states), and dirty tracking.
  • Scope & Seam: Placed in :core:domain as a pure JVM module. In-memory workspace state only; downstream export (ExportUseCase) and snapshot extraction (ExtractSnapshotUseCase) consume EditingSessionState snapshots.

MediaClip & TrimSegment

  • MediaClip: Represents an open video or audio file with metadata (URI, duration, stream tracks, sync keyframes).
  • TrimSegment: A bounded timestamp interval (startMs..endMs) within a clip associated with a SegmentAction (KEEP vs DISCARD).

2. Smart Cut & Analysis Entities

SilenceCutConfig & VisualDetectionConfig

  • SilenceCutConfig: Parameterized value object defining silence threshold (dB), minimum silence duration (ms), and padding bounds for automated section removal.
  • VisualDetectionConfig: Configuration object driving frame analysis: frame sampling step (sampleIntervalFrames), pHash difference threshold, SAD luminance delta, and contrast-normalized Laplacian variance threshold.

FrameAnalysis & WaveformResult

  • FrameAnalysis: Single-pass computed frame quality metrics (perceptual hash, luminance delta/sec, and contrast-normalized Laplacian blur score with edge-density gating).
  • WaveformResult: Decoded PCM audio amplitude vector with true RMS energy and perceptual scaling for timeline waveform rendering. Supports multi-track targeting by audio stream index.

3. Media Processing Engine

MuxingPipeline (Engine Module)

The internal :engine processing pipeline hiding keyframe seeking, sample copying, PTS timestamp shifting, track validation, and MediaMuxer lifecycle management behind a clean internal engine seam while satisfying the ILosslessEngine domain interface.

4. Storage & Cache Subsystems

IAnalysisCache & AnalysisCacheImpl

  • IAnalysisCache: Pure JVM domain interface defining cache read/write contracts for extractor-track-ID waveforms (getWaveform(clip, trackId)) and strategy-agnostic visual frame metrics (getVisualAnalysis(clip, sampleInterval)). Visual frame detection is currently Beta.
  • AnalysisCacheImpl: Android file-backed binary cache implementation in :core:data using versioned binary headers (_v2.bin), track-keyed waveform files (_t${trackIndex}_v2.bin), LRU byte cap eviction, and age-based retention expiry in app-private storage.

IMediaFinalizer & StorageUtils

  • IMediaFinalizer: Domain interface handling storage destination registration (IS_PENDING = 0 lifecycle on Android 10+ MediaStore vs SAF DocumentFile SAF tree targets).
  • StorageUtils: Android helper module executing Storage Access Framework tree operations and MediaStore collection queries.

5. UI Geometry & Rendering

Timeline geometry (CustomVideoSeeker internals)

Timestamp-to-pixel mapping helpers inside CustomVideoSeeker (:app/customviews, View-coupled: read the View's width): timestamp↔pixel translations, zoom factor constraints (1x..20x), scroll panning offsets, and hit-testing.