Skip to content

refactor: replace moviepy with mutagen + pymediainfo for media duration extraction

Summary

app/utils/media_duration.py imports moviepy — a full video editing framework — solely to read the duration of audio and video files. This introduces a heavy dependency for a single metadata field and includes a fragile subprocess re-encoding fallback for corrupted files.

Problem

  • moviepy is a video editing library (transitions, effects, compositing) being used as a metadata reader
  • It pulls in imageio, decorator, proglog, and ffmpeg binary dependencies unnecessarily
  • The file includes a fallback that re-encodes corrupted MP4s via ffmpeg subprocess — adding ffmpeg as a hidden system requirement
  • moviepy's VideoFileClip opens and decodes the file to extract duration, which is orders of magnitude more work than reading metadata headers
  • This makes duration extraction slow for large files and error-prone for unusual encodings

Proposed Solution

Replace with purpose-built metadata libraries:

  • mutagen (LGPL) for audio files (MP3, WAV, M4A, OGG) — reads ID3/Vorbis/M4A tags including duration in milliseconds without decoding audio
  • pymediainfo (MIT) for video files (MP4, MOV, WebM, MKV, AVI) — wraps the MediaInfo library, returns structured track metadata
# Audio (mutagen)
from mutagen import File
audio = File(path)
duration_seconds = audio.info.length

# Video (pymediainfo)
from pymediainfo import MediaInfo
info = MediaInfo.parse(path)
duration_seconds = info.tracks[0].duration / 1000

Library signals:

mutagen pymediainfo
License LGPL MIT
Weekly downloads 2M+ 500K+
Maintenance Active Active
CVEs None known None known

Impact

Dimension Current After
LOC 243 lines (incl. re-encoding fallback) ~30 lines
Dependency weight moviepy + imageio + ffmpeg binary mutagen + pymediainfo (MediaInfo)
Performance Decodes audio/video frames Reads file headers only
Reliability Fragile re-encoding for corrupt files Raises clean exception
Formats covered Same Same (MP3, WAV, M4A, OGG, MP4, MOV, WebM, MKV)

Files Affected

  • app/utils/media_duration.py — replace entirely
  • pyproject.toml — remove moviepy, add mutagen and pymediainfo