Parser / lexer

The parser and the lexer are managed via Bison / Flex. That is, the parser is LALR(1).

Lexer

The file containing rules is src/scanner.ll. To parse tokens, tokens.{cc,hh} files are used (convert int, strings…). To avoid issues with extra line feeds at the start / the end of the file, a new line token is added at the start and at the end of the file. This is done in driver.cc.

Parser

The file containing the grammar rules is src/parser.yy. There are multiple sections :

• Declarations : Functions, types, blocks…
• Statements : Instructions, control blocks…
• Expressions : Every operation that produce a value on the TOS (binary / unary operations…)
• Atoms : Basic expressions such as constants
• Tokens : Compound tokens such as stop which is a sequence of STOP (line feeds)

Expressions

Expressions are complex because of precedence, some expressions must be parsed before others. To avoid conflicts, here are the types of expressions :

• exp : Abstract expression used in statements
• set : Assignment, can be chained
• boolean : Not, And, Or…
• comparison : Binary comparison with boolean comparison operators (==…) or other operators like is / is not…
• binary : Binary arithmetic operation (+ - / *…)
• unary : Unary arithmetic operation (-x…)
• primary : Contains atoms, calls, attribute references, indexings…

Expressions are parsed in order from primary to exp (primary has a greater precedence than exp).