There is no single lexer in the CAMLE compiler due to the use of
the parser combinator library parsec. We produce a variety of lexers
that will lex different things like identifiers, numbers, reserved
words and operations. The lexers return the parsed token contained
inside of the Parser monad.
The parser is a collection of functions that utilise the lexers, paired with the AST datatype constructors to take a string of source code and produce an AST.
Of particular note is parsing variable assignment and read
calls. There are only two types of variables in the CAMLE language;
strings and integers. They are defined in two different ways:
a := <exp>definesaas an integer variable, since<exp>evaluates to an integer constant by the definition of the language.read(b)definesbas a string variable, or if it exists, replaces the value inside of
The intermediate representation differs from the abstract syntax tree in the following ways:
- Variable names become memory locations
- Assignment becomes move
- Temporary variables are inserted to hold intermediate computations
- Order of evaluation should become insignificant (?)
To implement assignment functionality we need to
- Assign variable definitions memory locations (if it’s the first time we’ve seen it)
- Resolve variable names to memory locations
A hashmap for the symbol table doesn’t really work because you don’t know what has been put into memory yet, i.e. you’d need to pass around an offset as well which would tell you where to put your new variable.