Lib.rs

Background

The goal of this project is to create a complete lexer for the Ruby language using parser combinators.

Differences from MRI

• Control characters can be recursively escaped:

  "\C-\C-\\n" # => "\n"
  

Implementation Notes

Expressions / Statements / Groups

• There is no semantic difference between an expression and a statement in this implementation
• Token::Block contains a list of statements (each item is considered to be a statement)
• Token::Expression contains a list of tokens that make up an individual statement
• Block and expression tokens may be nested via the use of parenthesis:

  (2 + (puts "hi"; 4 - 8;;)) * 5
  #hi
  #=> -10
  

Lexing partial inputs

The Input type can be extended to track a boolean field that denotes whether the lexer's input is complete or partial (such as within IRB's REPL). Combinators that may be partially completed (such as open strings, arrays, etc.) can additionally return a Token::IncompleteInput on the end of their token stream in partial mode to signal that the token has not been completed by the end of the user's input.

I'm not yet sure if start / end tokens should be used for complex objects, but they may be helpful when dealing with partial inputs.

To Do

• https://ruby-doc.org/core-2.7.0/doc/syntax/literals_rdoc.html
  • Rational / Complex numeric literals (r, i, ri)
  • Here Documents (<<, <<-, <<~, UNQUOTED, 'QUOTE', "DBL QUOTE", CMD QUOTE)
  • Arrays
  • Hashes
  • Ranges
  • Regular expressions
  • Procs
  • Percent strings