jepp

README

jepp

API fun with Jeopardy! Access >300k Jeopardy clues scraped from j-archive via a simple api.

---

API

The api is backed by a go web server built with gin that exposes a few endpoints to access historical Jeopardy data.

Types

The shape of the data returned from the api aligns with the db schema, this is accomplished via various struct tags on the type definitions.

Struct tags quick reference

• db tag is used by the sqlx library to map the db columns to the struct fields
• json tag is used by the gin library to map the struct fields to the json response
• example tag is used by the swaggo library to generate example responses for the swagger docs
• form and binding tags are used by gin to map query arguments to a struct with some basic validation

for example, the pkg.models.Clue type is defined as follows:

type Clue struct {
	ClueID     int64  `db:"clue_id" json:"clueId" example:"804002032"`
	GameID     int64  `db:"game_id" json:"gameId" example:"8040"`
	CategoryID int64  `db:"category_id" json:"categoryId" example:"804092001"`
	Question   string `db:"question" json:"question" example:"This is the question."`
	Answer     string `db:"answer" json:"answer" example:"This is the answer."`
}

• Struct tags also appear on some helper structs like the pkg.server.Filter type:

// Filter describes the query parameters that can be used to filter the results of an API query.
type Filter struct {
	Random *bool  `form:"random"`
	ID     *int64 `form:"id"`
	Page   *int64 `form:"page,default=0" binding:"min=0"`
	Limit  *int64 `form:"limit,default=1" binding:"min=1,max=100"`
}

Frontend / UI

• The ui is served from the / endpoint and is a simple html page that displays the swagger docs and some other info
• The embedded swagger ui provides runnable request / response examples and type references.

Swagger Docs

• Swagger documentation is generated with swaggo and embedded in the homepage as part of the html template.
• Figuring out the right build/deploy configuration was challenging here, I ran into some problems in my Taskfile task dependencies. The main problem seemed to be multiple tasks with the same set of files listed as sources causing a watched task to continuously rebuild because of some circular dependencies.
• These problems seem to be solved after breaking up and organizing the taskfiles better.

DB

Getting the data into the database started as a manual process, and hasn't been automated yet because the data is all there and I haven't needed to import / export it recently.

Here's how I went about doing it initially:

• For local development I set the DB_HOST, DB_USER, DB_PASS, DB_NAME environment variables to target a mariadb/mysql server running in my home lab.
• Most of the time I play with that local copy of the data, but the public api uses a mysql db hosted on digital ocean
• Initially to populate the prod db I just manually created a backup of my local database and restored it to the prod database, both via an adminer instance running in my home lab.
• Currently the task sql:dump command will create a dump of the database defined by the environment variables and write it to data/dump.sql.gz.
• Recent dumps of the prod database are available in the data directory or as downloads on repository's Releases page.

Data Scraping

The scraper package contains the code to scrape j-archive for jeopardy clues and write the data to a mysql database. Colly is the package use to scrape the data and sqlx is used to write the data to the db. The scraping happened in a few passes, more or less following these steps:

Get all the seasons and populate the seasons table.

• This scrape targeted the season summary page on j-archive and pulled the season number, start date, end date for each season

Get all the games for each season and populate the game table.

• This scrape targets the individual season show pages on j-archive and pulls the game number, air date, taped date for each season

Get all the clues for each game in each season and populate the category and clue tables

• This scrape targeted the individual game pages on j-archive and pulls the clue data from the <table> elements on the page

references / prior art

• jservice
• jservice repo
• jeppy

---

---