For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).

Python function

validate_video

validate_video()

max.pipelines.context.validate_video(source, max_decoded_bytes=None)

source

Checks a video’s header, decoding no frame.

Intended for trust boundaries (the serving layer) that want an unreadable or oversized video rejected as a client error before it reaches a model worker. The video is decoded exactly once, by whoever consumes it, so damage inside the stream surfaces there rather than here.

Parameters:

• source (str | bytes) – The video to validate.
• max_decoded_bytes (int | None) – Decoded-byte budget available to the video. A frame is allocated at full size before any check can run on it, so the header dimensions are checked against this before decoding.

Raises:

• MediaDecodeError – If the source is not a readable video.
• MediaBudgetExceeded – If one frame would not fit in max_decoded_bytes.

Return type:

None