Opal Compiler

Opal is a source to source compiler. It accepts ruby code as a string and
generates javascript code which can be run in any environment. Generated
code relies on the opal runtime which provides the class system and some
other runtime helpers.

Compiler stages

The compiler can be broken down into 3 separate stages:

• lexing
• parsing
• code generation

Lexer

The opal lexer is implemented in pure ruby using
the StringScanner class from the opal stdlib. The source code is scanned
and tokens are then provided to the parser. This process simply converts
the ruby code given as a string, into a list of tokens representing the
parts of the ruby code.

Parser

The opal parser is implemented using a standard
bison like syntax, but relies on racc, a ruby implementation of yacc/bison
which is again available in the standard library. The parser takes these tokens
generated by the lexer and builds a syntax tree representing the ruby code.
This syntax tree is represented by sexps. As
ruby is such a complex and dynamic language, there is a lot of interaction
between the parser and the lexer, namely through a preserved lex_state.

Code generation

The opal compiler takes these sexps from the parser
and generates ruby code from them. Each type of sexp has its own node type
used to generate javascript. Each node creates an array of one or more
fragments which are the concatendated together to
form the final javascript. Fragments are used as they contain the generated
code as well as a reference back to the original sexp which is useful for
generating source maps afterwards.