How does it work?

When a repository is analyzed two repositories are used: repo_info_extractor and this repository. The repo_info_extractor is responsible to recognize the languages and export the imported libraries. This repository contains a list of supported libraries, imports and technologies they belong to.

In short

Language recognition: repo_info_extractor.
Library recognition: libraries

Structure

The entire list stored as a JSON object, top level key is the language, inside that there are library objects where names are the keys, and the value of such property is a number of attributes such as import pattern, description, etc.

One library can appear in multiple languages and can have multiple import patterns or technologies.

An example entry looks like:

"JavaScript": {
  ..
  "Express": {
    "imports": ["express"],
    "technologies": ["MVC", "Web Development"],
    "description": "Express is a minimal Node.js framework for web and mobile applications.",
    "image": "https://raw.githubusercontent.com/github/explore/80688e429a7d4ef2fca1e82350fe8e3517d3494d/topics/express/express.png"
  },
  ..
}

Deep dive into structure

As mentioned before, top level objects are the languages. Inside a language, there are libraries which are belong to the that language and the key of that object is the library name.

Language defines what language this library belongs to. One library can be mapped to one language only, so in case of ambiguous libraries like Bootstrap, the choice has to be made which language it is mainly. Bootstrap is mainly CSS, even though it involves some JavaScript.

One exception to this rule is, when the same library has different wrappers for different languages. For example mongoose can be used in JavaScript, TypeScript and C. So there are 3 mongoose objects (for each language), with their own import styles.

Imports is an array of strings, that all map to this library. During the analysis of the repository, CodersRank extracts imports that were used in the commits authored by the user. This may have a form of

Symfony\Component\HttpFoundation\Request
github.com/micro/go-micro/client
express

The import name is not the exact string that should be matched, but rather the string the import starts with. Given this rule, the above imports would match the following:

Symfony\
github.com/micro/go-micro
express

Therefore if we import any package from go-micro, e.g. github.com/micro/go-micro/client, it would still match Go-micro library. Same applies for Symfony in PHP. Express is the exact match, but does not look any different from examples above, since the string express starts with the substring express.

Technologies are pretty self-explanatory. This mapping is used to calculate the score for a particular technology, and one library can be mapped to one of more technologies, as in the original example.

Description is a field thast allows the Library description to be provided.

Image allows to attach a logo to the library. This takes a form of the URL pointing to any image renderable by major browsers.

Contributing

All contributions are welcome, and CodersRank relies on such support. The rules are few, but essential:

All fields apart of description and image are required.
Maintain alphabetical order for libraries and languages. Think SQL ORDER BY library.
If providing description, make sure to include the dot . at the end of the line. Also make sure the spelling is correct and the description is accurate.
JavaScript and TypeScript must be identical.

TODO

Currently the image is not used anywhere, but once it is displayed, particular restrictions on format and dimenstions need to be introduced.
More libraries 😀

tsdtsdtsd / libraries