Language Savant
Pull request Compare This branch is 4688 commits behind github:master.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


We use this library at GitHub to detect blob languages, highlight code, ignore binary files, suppress generated files in diffs and generate language breakdown graphs.


Language detection

Linguist defines the list of all languages known to GitHub in a yaml file. In order for a file to be highlighted, a language and lexer must be defined there.

Most languages are detected by their file extension. This is the fastest and most common situation.

For disambiguating between files with common extensions, we use a bayesian classifier. For an example, this helps us tell the difference between .h files which could be either C, C++, or Obj-C.

In the actual GitHub app we deal with Grit::Blob objects. For testing, there is a simple FileBlob API."lib/linguist.rb") #=> "Ruby""bin/linguist") #=> "Ruby"

See lib/linguist/language.rb and lib/linguist/languages.yml.

Syntax Highlighting

The actual syntax highlighting is handled by our Pygments wrapper, pygments.rb. It also provides a Lexer abstraction that determines which highlighter should be used on a file.

We typically run on a prerelease version of Pygments, pygments.rb, to get early access to new lexers. The languages.yml file is a dump of the lexers we have available on our server.


The Language Graph you see on every repository is built by aggregating the languages of all repo's blobs. The top language in the graph determines the project's primary language. Collectively, these stats make up the Top Languages page.

The repository stats API can be used on a directory:

project = Linguist::Repository.from_directory(".")  #=> "Ruby"
project.languages      #=> { "Ruby" => 0.98, "Shell" => 0.02 }

These stats are also printed out by the binary. Try running linguist on itself:

$ bundle exec linguist lib/
100%  Ruby

Ignore vendored files

Checking other code into your git repo is a common practice. But this often inflates your project's language stats and may even cause your project to be labeled as another language. We are able to identify some of these files and directories and exclude them."vendor/plugins/foo.rb").vendored? # => true

See Linguist::BlobHelper#vendored? and lib/linguist/vendor.yml.

Generated file detection

Not all plain text files are true source files. Generated files like minified js and compiled CoffeeScript can be detected and excluded from language stats. As an extra bonus, these files are suppressed in Diffs."underscore.min.js").generated? # => true

See Linguist::BlobHelper#generated?.

Installation is usually running the latest version of the github-linguist gem that is released on

But for development you are going to want to checkout out the source. To get it, clone the repo and run Bundler to install its dependencies.

git clone
cd linguist/
bundle install

To run the tests:

bundle exec rake test


The majority of patches won't need to touch any Ruby code at all. The master language list is just a configuration file.

Almost all bug fixes or new language additions should come with some additional code samples. Just drop them under samples/ in the correct subdirectory and our test suite will automatically test them. In most cases you shouldn't need to add any new assertions.


Sometimes getting the tests running can be too much work, especially if you don't have much Ruby experience. Its okay, be lazy and let our build bot Travis run the tests for you. Just open a pull request and the bot will start cranking away.

Heres our current build status, which is hopefully green: Build Status