Vidyard Scraper

main.py

# /// script
# requires-python = ">=3.10"
# dependencies = [
#   "httpx",
# ]
# ///

import asyncio
import re
import sys
from dataclasses import dataclass
from pathlib import Path

import httpx

client = httpx.AsyncClient()


@dataclass
class Chapter:
    name: str
    vtt_caption_url: str
    mp4_url: str
    hls_playlist_url: str


@dataclass
class Video:
    chapters: list[Chapter]


def input_to_video_id(input: str) -> str:
    if md := re.match(r"^https://play\.vidyard\.com/(.+)$", input):
        return md[1]
    return input


async def get_video(id: str) -> Video:
    resp = await client.get(f"https://play.vidyard.com/player/{id}.json")
    resp.raise_for_status()
    return Video(
        chapters=[
            Chapter(
                name=c["name"],
                vtt_caption_url=[v["vttUrl"] for v in c["captions"] if v["isDefault"]][
                    0
                ],
                mp4_url=[
                    s["url"] for s in c["sources"]["mp4"] if s["profile"] == "1080p"
                ][0],
                hls_playlist_url=[
                    s["url"] for s in c["sources"]["hls"] if s["profile"] == "1080p"
                ][0],
            )
            for c in resp.json()["payload"]["chapters"]
        ]
    )


async def download_file(video_id: str, chunk_url: str, out_path: Path) -> None:
    with out_path.open("wb") as outf:
        async with client.stream(
            "GET",
            chunk_url,
            headers={"Referer": f"https://play.vidyard.com/{video_id}"},
        ) as resp:
            resp.raise_for_status()
            async for chunk in resp.aiter_bytes():
                outf.write(chunk)


async def main() -> None:
    for arg in sys.argv[1:]:
        vid_id = input_to_video_id(arg)
        vid = await get_video(vid_id)
        for chap in vid.chapters:
            await download_file(
                vid_id, chap.vtt_caption_url, Path(f"./{chap.name}.vtt")
            )
            await download_file(vid_id, chap.mp4_url, Path(f"./{chap.name}.mp4"))


if __name__ == "__main__":
    asyncio.run(main())