popstas / talks-reducer / 28131157214

push

github

feat: Add video trim (cut begin/end) before converting (#140)

Add an optional trim step that keeps only a chosen [start, end] fragment of each
input before the speed-up pipeline runs.

- CLI: --cut-start / --cut-end (seconds or HH:MM:SS), keep-range in/out semantics;
  0/0 means no trim (behaviour unchanged when off).
- ffmpeg: input-level -ss/-t injected into both audio-extract and final-video
  commands so audio/video stay in sync; pipeline recomputes effective duration so
  progress/target-duration math stays correct on a trimmed clip.
- Desktop GUI: a "Cut video" panel with manual start/end time inputs and a
  frame-scrub thumbnail preview; enabled flag + start/end persist via preferences.
- Web UI (gradio): Cut video checkbox + start/end inputs wired through process_video
  and the remote server path.
- Docs + tests across timecode/cli/ffmpeg/pipeline/models/server/gui (557 tests).

286 of 332 new or added lines in 10 files covered. (86.14%)

8480 of 11318 relevant lines covered (74.92%)

0.75 hits per line