Architecture

The process of analysing Ruby source code is broken up into several separate steps. The basic flow of this process looks like the following:

Most third-party code will only deal with the last 3 steps.

This chapter takes a brief look at the various parts that make up ruby-lint. Some of these parts are discussed more in depth in which case a link to these chapters can be found below.

The Parser

The parser converts raw Ruby source code into an AST. The AST is built up by nodes formatted as S expressions. A node for a simple String would look like the following:

(string "hello")

Each node is an instance of RubyLint::AST::Node. Each node can contain a number of child nodes. If one were to format the above S expression as an Array you’d get the following:

[:string, ["hello"]]

The parser used by ruby-lint is the parser gem and as such ruby-lint can support various Ruby implementations and versions.

Virtual Machine

The ruby-lint virtual machine is a stack based virtual machine used for partially evaluating Ruby code in order to build definitions of method definitions, class definitions, variable assignments and so on. For more information on the internals of this class see the documentation of RubyLint::VirtualMachine.

Analysis

This is where the real work start. The analysis step consists out of various Ruby classes of which each performs its own type of analysis. For example, RubyLint::Analysis::UndefinedMethods checks for the use of undefined methods.

Reports

Reports are used for storing a set of RubyLint::Report::Entry instances that contain information about a block of Ruby code that is being analyzed. Each entry contains some location information (line and column number for example) as well as a message (e.g. an error message).

Presenters

Presenters are classes tasked with the job of presenting a report. Currently ruby-lint ships with a presenter that outputs a report in plain text as well as one that outputs a report in JSON.