Video file names in standard format

This PR updates the video file names in the record package so that they match the format used by EFP and expected by the lookit-api and and the AWS S3-Lambda trigger.

Consent videos: consent-videoStream_{study}_{frame_id}_{response}_{timestamp}_{random_digits}.webm
Non-consent videos: videoStream_{study}_{frame_id}_{response}_{timestamp}_{random_digits}.webm

This naming system requires two arguments from the plugin that calls recorder.start:

whether or not this is a consent recording, which is used for the prefix
the "trial_type" (plugin name), which is used to create the frame_id value

The frame_id value is the same one that lookit-jspsych uses in the response sequence field, and that's used in the lookit-api to identify trials. The format is: {trial_index}-{trial_type}. Trial types are not necessarily unique across the experiment, but the trial indices are.

Implementation note

Our previous file naming system used a set of pre-defined strings as an argument in the recorder.start method: "consent" | "session_video" | "trial_video". It would be nice to still have a set of pre-defined strings for the new trial_type string passed to recorder.start. But I don't think that's makes sense now, since our trial recording extension is meant to be used with any jsPsych plugin, so we'd have to maintain a long list of all (currently ~40) jsPsych trial types ("html-button-response", etc.). So this argument is just an unconstrained string. I'm open to suggestions for ways of constraining or validating it.

TO DO

As discussed here #36, we would like to give researchers the ability to assign their own unique/meaningful trial IDs, and use these index-type IDs as a fallback. But in the interest of meeting our first MVP, I'm just implementing the index-type solution for now, and we can add the support for researcher-provided trial IDs later.

lookit / lookit-jspsych

Video file names in standard format #70

🦋 Changeset detected