Response shape

One object, the same every time. A field that does not apply comes back as null rather than disappearing, so your parser never has to check whether a key exists.

Top level

Field	Type	Notes
`platform`	string	Which network the link belonged to: `instagram`, `youtube`, `tiktok`, `facebook`, `vk`, `threads`, `twitter`, `reddit` or `pinterest`
`type`	string	`video`, `image`, `album`, or `audio` when you asked YouTube for mp3
`title`	string \| null	The caption, or the video's title, when the post carries one
`author`	string \| null	Account or channel name, where the platform reports it
`thumbnail`	string \| null	Preview image
`duration`	number \| null	Length in seconds, where the platform reports it
`media`	array	Always present, always at least one item

Inside media

Field	Type	Notes
`type`	string	`video`, `image` or `audio`
`url`	string	Direct link to one file. A video arrives with its audio already in it, so there is nothing to join afterwards.
`quality`	string \| null	The height VK encoded this rendition at, as a label like `1080p`, or the format you asked YouTube for. `null` everywhere else.
`ext`	string \| null	`mp4` or `mp3` where the platform states it, `null` where it does not

What is always filled in

platform, type and media always carry something. The other four do not: the same platform can answer with an author on one link and null on the next, so treat all four as optional and nothing will surprise you.

The one endpoint with a different shape

/v2/youtube/info downloads nothing, so it has nothing to put in media. It answers with the video's own facts instead: an id, an author_url, every thumbnail size, and a formats array listing each quality with its size in bytes. That body is on the YouTube page.

Writing against it

data = response.json()

if data["type"] == "album":
    print(len(data["media"]), "files")

for item in data["media"]:
    download(item["url"])