"""Cut a video up and print what came out. python -m desk.tools.video python -m desk.tools.video --json > segments.json Videos only. A podcast episode — where the chapter marks and the article links live on the show's own page rather than on YouTube — goes through the layer above: python -m desk.sources.podcast """ from __future__ import annotations import argparse import dataclasses import json import logging import sys import desk.tools.video as video_pkg from desk.tools.video import marks from desk.tools.video.segment import Segment, Video def main(argv: list[str] | None = None) -> int: parser = argparse.ArgumentParser( prog="python -m desk.tools.video", description="Cut a video into one item per chapter mark.", ) parser.add_argument("url", help="a YouTube link or bare video id") parser.add_argument("--json", action="store_true", help="dump everything, transcripts included, as JSON") parser.add_argument("--chars", type=int, default=160, help="how much transcript to preview per segment (default 160)") parser.add_argument("--no-comments", action="store_true", help="don't fall back to chapter marks typed in the comments") parser.add_argument("-v", "--verbose", action="store_true") args = parser.parse_args(argv) logging.basicConfig( level=logging.INFO if args.verbose else logging.WARNING, format="%(levelname)-7s %(message)s", ) # Windows still opens a console in the system codepage, and a video title # with a "ñ" in it — or the arrows this prints — is an unhandled # UnicodeEncodeError half way through the report rather than a bad glyph. if hasattr(sys.stdout, "reconfigure"): sys.stdout.reconfigure(encoding="utf-8", errors="replace") try: video, segments = video_pkg.read(args.url, comments=not args.no_comments) except Exception as exc: print(f"error: {exc}", file=sys.stderr) return 1 if args.json: json.dump(dump(video, segments), sys.stdout, indent=2, default=str) print() else: report(video, segments, args.chars) return 0 def report(video: Video, segments: list[Segment], chars: int) -> None: """The human-readable view. Also used by the podcast module.""" words = sum(len(s.transcript.split()) for s in segments) length = video.duration or (segments[-1].end if segments else 0) print(f"{video.title}") print(f" {video.author or '—'} {video.ref.kind}:{video.ref.key}" f" {marks.hhmmss(length)}") print(f" {len(segments)} segments, {words:,} words of transcript") print() heading = None for segment in segments: if segment.heading != heading: heading = segment.heading print(f" {heading or '(unfiled)'}") preview = segment.transcript[:chars] if len(segment.transcript) > chars: preview += "…" print(f" {segment.timecode:>8} {segment.title[:78]}") print(f" {marks.hhmmss(segment.duration)} long · {segment.url()}") for link in segment.links: print(f" → {link.url}") print(f" {preview or '(no transcript under this mark)'}") print() def dump(video: Video, segments: list[Segment]) -> dict[str, object]: """The shape a stored item is built from, plus the fields a dry run wants to eyeball. The stored half is `Segment.record()` and is defined there; desk/sources/podcast.py turns the rest into a row.""" return { "video": { "ref": dataclasses.asdict(video.ref), "title": video.title, "url": video.url, "author": video.author, "published_at": video.published_at, "duration": video.duration, }, "segments": [ { **s.record(), "external_id": s.external_id, "title": s.title, "heading": s.heading, "timecode": s.timecode, "url": s.url(), "transcript": s.transcript, } for s in segments ], } if __name__ == "__main__": raise SystemExit(main())