Project is still in progress
The lexical, syntax and semantic analysis are completed. Up next is the machine code generation.
A little bit of info concerning the designs of the main phases:
The Finite state machine which recognizes the atomic elements of the language (lexemes/tokens) is implemented via a table driven design. A stream of Token objects is returned.
The parser is also implemented via a table driven design. It takes as input the previously computed stream of tokens. All the errors contained in the program text file are signaled as the parsing progresses. The parser also builds an abstract syntactic tree. This provides an intermediate representation of the program which is passed on to the semantic analysis phase.
The semantic analysis is done using a tree traversal of the AST tree. The tree is traversed twice. Semantic actions are triggered when critical nodes are encountered. On the first pass it generates symbol tables for the scope of the different functions and classes. It also checks for some possible semantic errors. Here is a list :
"multiple definition (member function)"
"multiple definition (free function)"
"multiple definition (class)"
"multiple identifier"
"undeclared member function"
"undeclared identifier"
"undefined function declaration"
"invalid class"
"WARNING overwritten member"
"could not match with function declaration"
"type mismatch in statement"
"invalid function call or data member"
"index must be of type integer"
The second pass procedes in a similar manner but it uses the symbol table to check types and return types of statements and function calls.
To run the code you must use the **g++**compiler. You also need to change the path of the compiled program in the main (line 24) method to whatever one you want.
I apologize for the lack of clarity in my code. I am probably not taking advantage of key c++ functionalities And I may not even be using some of them in the right way. Anyhow this is by far the most amazing project I have ever worked on.