Skip to content

Latest commit

 

History

History
79 lines (53 loc) · 3.75 KB

README.md

File metadata and controls

79 lines (53 loc) · 3.75 KB
Raw

Dungeon Crawler

Download WoW Mythic+ leaderboard data and analyse it. Work in progress!

Tank popularity over time

Using rust to invoke Blizzard API to download leaderboard data and R to draw pretty pictures from it.

For accessing leaderboard data you need to set up API access in Blizzard developer zone. Afer doing so client ID and secret can be specified via CLIENT_ID and CLIENT_SECRET environment variables. These variables can be set in .env file something like this:

CLIENT_ID="ffffffffffffffffffffffffffffffff"
CLIENT_SECRET="xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"

Because we are using credentials flow there's no need for authorization (with password).

Building the web crawler

Leaderboard crawler has been implemented in rust programming language.

If you already have rust installed building the code is as easy as cargo build. Do not be scared of all the dependencies that are downloaded. I have heavily relied on external libraries for web access, deserialization from JSON and serialization to CSV, parallelization, command line parsing etc. To run the executable cargo run -- --help (to see help message) or cargo run -- download --region eu (to download this weeks leaderboard data for entire eu region).

If you want to track log message while downloading a specific week of leaderboards:

$ cargo run -- download --region eu --period 687

Easiest way to install rust is to use rustup and follow the guide on the website. This should work for both linux and osx. I have not tested windows.

Usage of the web crawler

To download most recent week leaderboards of eu realms:

$ dungeon-crawler download --region eu

This creates a CSV file that on every row contains information about individual run and the group composition. Hopefully, the structure of the file is self-evident

Supported regions are 'eu', 'us', 'tw' and 'kr'. Display more options with --help.

High-level overview of project structure:

  • data/static/ contains some static information about WoW M+ (time limits, specs, etc)
  • img/ contains just example output image
  • src/ contains rust source code for crawling Blizzard API. The code uses curl. Mangles JSON data from Blizzard to CSV.
  • Cargo.toml rust project file
  • *.R scripts for drawing charts (more about those below)

Usage of R scripts

I have split R code into two parts:

  • loadCSV.R reads a set of CSV files, preprocess them and stored database as R object db.rds
  • analysis.R read the database file (default db.rds) and draws bunch of charts and graphs from the data.

During preprocessing we remove duplicate entries, convert timestamps to a more usable format, and check if a run was successful or not. We have split data loading into seperate script in order to make testing various charts faster process.

The analysis.R file is very much intended to be modified! In general drawing of each plot has been split into a different function. Scroll all the way down to bottom to see what functionality I have played around with in the past.

The project makes extensive use of data.table library. It's very performant but unfortunately inserting new data is slow.

TODO

  • Command line parameters (week, region).
  • Support chinese realms. Turns out the API for CN is not available.
  • Timebox dungeon time limits.
  • Use proper OAuth2 library. Currently we are making curl requests.
  • Find a better way to represent groups with multiple tank/healer roles.
  • Correlate data with affixes.

Item level analysis is currently beyond the scope of this project.

License

The code is open under MIT license.