Design notes - mgp/sharebears GitHub Wiki

Entry data type

• matcher: URL -> bool
• fetcher: public API -> model
• converter:
  • serializer: model -> ORM entity
  • deserializer: ORM entity -> model
• renderer: model -> HTML

Transformers

Turns [model]* into [model]*

A transformer can take a sequence of Photo instances and group them into a Slideshow instance composed of those photos.

Where in the pipeline should they be introduced? Upon writing or reading Post instances?

URL types

• YouTube (embed)
• Twitter (embed)
• Images
• Google Docs
• GitHub
• Phabricator

Message format

Loosely it's:

([text]? [url])+ [text]? [hashtag]*

Rules:

• There must be at least one url
• There must be at least one text
• Consecutive url tokens that are images will collapse into a slideshow/carousel

Entity types

User

• id (pk)

GoogleUser

• user_id (fk)
• google_id (pk)

Post

• id (pk)
• creator (fk)
• created_time
• num_stars
• data

Must create an index on (created_time, id) to efficiently return all posts sorted by time.

HashTag

• post_id (fk, pk)
• hashtag (pk)
• created_time

This allows returning all hash tags for a post. To show all posts, sorted by time for a given hash tag, index on (hashtag, created_time, post_id).

StarredPost

• post_id (fk, pk)
• user_id (fk, pk)
• created_time

This allows returning all users that starred a post. To show all starred posts sorted by time for a given user, index on (user_id, created_time, post_id).