How to generate a symbol table for a Pascal-subset compiler in C?

Question

I'm writing a compiler in C, which will interpret pseudo-Pascal instructions (their syntax is for now irrelevant) to an asm output. What I know for now, is that I need:

• a syntax scanner which will scan the user input and identify tokens for the parser to process
• a parser which will check if those tokens fit into the defined grammar productions
• a symbol table

I'm a little bit stuck on the most important phase - symbol table. I'm unsure as to what should be included in this table. Certainly any variables (identifiers), with their address. Should I include keywords such as if, for etc? Any guidelines on this would be appreciated.

For now, I think the most logical way would be to define a structure:

struct entry{
    char* name;
    Vartype vartype;
    int address;
}

where Vartype is an enum with available variable types (integer and real). Then ofc I would make an array of structures and expand it when necessary. How and when should I modify it?

Answer 1

Here is an awsome free book: Compiler design in C