Wikiviews API

A simple API to do some basic data aggregation around wikipedia page views

API Docs

The API is read-only and thus, all below resources only support GET. Additionally, all data will be returned as JSON

All instances of the date parameter require an ISO style date stamp (e.g., '2022-10-01'). Any other format will result in a 400.

Endpoints

Most views per week

'/most-viewed/week-ending/{date}'

Returns a sorted list of articles by page view in the last week (as determined by the given date).

Success: 200

{
  "most_viewed": {
    "start_date": {date},
    "end_date": {date},
    "articles": [
      {
        "article": {str},
        "views": {int}
      }
    ]
  }
}

Most views per month

'/most-viewed/month-ending/{date}'

Returns a sorted list of articles by page view in the last month (as determined by the given date).

Success: 200

{
  "most_viewed": {
    "start_date": {date},
    "end_date": {date},
    "articles": [
      {
        "article": {str},
        "views": {int}
      }
    ]
  }
}

Article views per week

'/articles/{article}/views/week-ending/{date}'

Returns the number of views for the given article in the last week (as determined by the given date).

Success: 200

{
  "article_views": {
    "start_date": {date},
    "end_date": {date},
    "article": {str},
    "views": {int}
  }
}

Article views per month

'/articles/{article}/views/month-ending/{date}'

Returns the number of views for the given article in the last month (as determined by the given date).

Success: 200

{
  "article_views": {
    "start_date": {date},
    "end_date": {date},
    "article": {str},
    "views": {int}
  }
}

Development

To run locally for testing and dev, first clone the repo and install poetry

To run the server locally, run . local.sh as provided in the repo.

Similarly, to run tests, run the provided unit.sh and e2e.sh for unit and end to end tests, respectively (the end to end tests will be pretty slow).

Improvements

Some things I didn't have time for

Replace in memory cache with something better (e.g. Redis, or even PostgreSQL)
Fetch data from the API in the background and just use the cache
add a linter
Maybe use threads or asyncio to speed up the bits where we're making a bunch of HTTP calls in a loop
Consider OpenAPI for the docs
Consider Pydantic or something for JSON slinging

sarangj / wikiviews