The purpose is to store all the quotes made by the EZA chat during live streams on Twitch. The ability to receive all the chat messages is limited by the archive storage time limit, and the future of the Twitch Rechat API.
As of yet not decided. The architecture must though (at least) consist of three server instances:
- Extraction; e.g. Python or NodeJS
- Database; e.g. MongoDB or MariaDB
- Web
- Front-end; AngularJS, React, Ember
- Back-end; Python or NodeJS The server instances will be cloud hosted; at e.g DigitalOcean or Amazon.
When a new VOD is available (detected by daily API checks), the process of extracting quotes will initiate. The thoughtout process is as now the following:
- Set of quote extracting by a web hook
- Extract quotes using a regular expression pattern
- Store only the quotes in a suitable database, which solution? A NoSQL-based, or SQL-based.
This would follow after the extraction process. As with anything, the contents of extracted quotes can contain harassements etc. or be a gateway for a possible security breach in some way. Controls for this would need to be implemented.
The quotes will be available through a web page. The VOD timestamp of the chat quote is used to provide a Twitch VOD preview, imagine a list of previews and the corresponding quote. Though this will rely on that the quote timestamp and VOD preview timestamp is exactly relative, in terms of its creation and relevance. Though this can be dirty solved by setting the VOD preview timestamp back with e.g. 5 seconds.
The web page will provide the following functions:
- Quote of the day
- Searching
- By keyword
- By VOD
- Within a time span All above can be freely combined
- VOD preview in relation of the chat quote. Though limited by the time availability limit. Perhaps manually link to YouTube archive with a list of all the quotes attached to it in the future? Explore a way to fully automate this as well.
- Twitch API; https://dev.twitch.tv/
- 0.1.0 (2017-04-22)
- Initial description