sugar/src/molasses/parser_primitives.cpp

#include <algorithm>
#include <cassert>
#include <iostream>
#include "molasses/parser_primitives.h"

namespace molasses {
	parser_context register_integers(parser_context ctx) {
		ctx.types.push_back(std::make_shared<primitive_type>("i8",1));
		ctx.types.push_back(std::make_shared<primitive_type>("i16",2));
		ctx.types.push_back(std::make_shared<primitive_type>("i32",4));
		ctx.types.push_back(std::make_shared<primitive_type>("i64",8));
		ctx.types.push_back(std::make_shared<primitive_type>("u8",1));
		ctx.types.push_back(std::make_shared<primitive_type>("u16",2));
		ctx.types.push_back(std::make_shared<primitive_type>("u32",4));
		ctx.types.push_back(std::make_shared<primitive_type>("u64",8));
		
		return ctx;
	}
	
	std::vector<std::string> operator>>(std::vector<std::string> current_stack, const operation& next_op) {
		{
			auto args = next_op.argument_types();
			while(not (args.empty() or current_stack.empty())) {
				if(current_stack.back() != args.back()) {
					throw TypeInputError();
				} else {
					args.pop_back();
					current_stack.pop_back();
				}
			}
			if(not args.empty()) {
				throw ValueMissingError();
			}
		}
		{
			auto return_types = next_op.return_types();
			std::move(return_types.begin(), return_types.end(), std::back_inserter(current_stack));
		}
		return current_stack;
	}
	
	std::optional<int32_t> try_parse_int32(const std::string& str) {
		int32_t value;
		auto begin = str.data();
		auto end = str.data()+str.size();
		auto result = std::from_chars(begin, end, value, 10);
		// TODO: Add other bases
		if(result.ptr == end) {
			return value;
		}
		return std::nullopt;
	}
	
	auto find_ptr_by_name_in_container(auto container, const auto& name) -> typeof(*std::begin(container)) {
		auto it = std::find_if(std::begin(container), std::end(container), [&](auto elem){
			return elem->name() == name;
		});
		if(it != std::end(container)) {
			return *it;
		}
		return {};
	}
	
	std::shared_ptr<type> parser_context::lookup_type(const std::string & name) const {
		return find_ptr_by_name_in_container(types, name);
	}
	
	std::shared_ptr<operation> parser_context::lookup_operation(const std::string & name) const {
		return find_ptr_by_name_in_container(operations, name);
	}
	
	bool type_check(
		const parser_context& parser_state,
		const lexed_output& lexer_state,
		const std::vector<symbol>& consumed_stream,
		std::vector<std::string> execution_input,
		const std::vector<std::string>& execution_output
	) {
		auto& type_stack = execution_input;
		
		for(const auto& symbol : consumed_stream) {
			const auto& symbol_text = lexer_state.dictionary.at(symbol);
			if(auto is_int = try_parse_int32(symbol_text); is_int) {
				type_stack.emplace_back("i32");
			} else if(auto is_op = parser_state.lookup_operation(symbol_text); is_op) {
				type_stack = type_stack >> *is_op;
			}
		}
		
		return type_stack == execution_output;
	}

    std::vector<std::string> initialize_stack() {
        return {
                ".bss\n",// TODO: make threadlocal
                "stack_instruction:\n",
                "    .quad 0\n",
                ".text\n",
                "initialize_callstack:\n",
                "    movq $9, %rax\n",
                "    movq $0, %rdi\n",
                "    movq $8192, %rsi\n",
                "    movq $3, %rdx\n",
                "    movq $34, %r10\n",
                "    movq $-1, %r8\n",
                "    movq $0, %r9\n",
                "    syscall\n",
                "    movq %rax, (stack_instruction)\n",
                "    retq\n",
        };
    }

	parser_context parse(parser_context ctx, const lexed_output& lexer_data) {
		enum op : int {
			DO_KW = 1,
			SEPARATOR_KW,
			PROC_KW,
			END_KW
		};
		
		lexed_output fake;
		fake.dictionary[PROC_KW] = "__PROC__";
		fake.dictionary[SEPARATOR_KW] = "__--__";
		fake.dictionary[DO_KW] = "__DO__";
		fake.dictionary[END_KW] = "__END__";
		
		auto tokens = concatenate(fake, lexer_data);
		
		std::vector<std::shared_ptr<procedure_operation>> parsed_procedures;
		
		auto parse_proc = [&](auto it) -> std::pair<typeof(it), std::shared_ptr<procedure_operation>> {
#define CHECK_FOR_UNEXPECTED_STREAM_END \
			if(it == tokens.symbols.end()) { \
				throw ExpectingTokenError(); \
			}
			
			if(*it != PROC_KW) {
				throw UnexpectedTokenError();
			}
			++it;
			CHECK_FOR_UNEXPECTED_STREAM_END;
			
			std::string name = tokens.dictionary.at(*it);
			++it;
			CHECK_FOR_UNEXPECTED_STREAM_END;
			
			if(it == tokens.symbols.end()) {
				throw ExpectingTokenError();
			}
			
			// Process arguments list
			std::vector<std::string> argument_types;
			while(*it != SEPARATOR_KW) {
				argument_types.emplace_back(tokens.dictionary.at(*it));
				++it;
				CHECK_FOR_UNEXPECTED_STREAM_END;
			}
			++it;
			CHECK_FOR_UNEXPECTED_STREAM_END;
			
			// Process return types list
			std::vector<std::string> return_types;
			while(*it != DO_KW) {
				return_types.emplace_back(tokens.dictionary.at(*it));
				++it;
				CHECK_FOR_UNEXPECTED_STREAM_END;
			}
			++it;
			CHECK_FOR_UNEXPECTED_STREAM_END;
			
			// Process return types list
			std::vector<symbol> body;
			while(*it != END_KW) {
				body.emplace_back(*it);
				++it;
				CHECK_FOR_UNEXPECTED_STREAM_END;
			}
			++it;
			
			return std::make_pair(it, std::make_shared<procedure_operation>(name, argument_types, return_types, body));
#undef CHECK_FOR_UNEXPECTED_STREAM_END
		};

        auto progress = tokens.symbols.begin();

        do {
            auto [iterator, procedure] = parse_proc(progress);
            ctx.operations.push_back(procedure);
            parsed_procedures.emplace_back(std::move(procedure));
            progress = iterator;
        } while (progress != tokens.symbols.end());

		for(auto& proc : parsed_procedures) {
			if(not type_check(ctx, tokens, proc->_body, proc->_args, proc->_rets)) {
				throw ProcedureStackError();
			}
		}

        std::vector<std::string> generated;

        for(auto instr : initialize_stack()) {
            generated.push_back(instr);
        }

        for(auto proc : parsed_procedures) {
            for(auto instr : proc->generate(ctx, tokens)) {
                generated.push_back(instr);
            }
        }

        for(auto line : generated) {
            std::cout << line;
        }
		
		return ctx;
	}

    std::vector<std::string> generate_call(std::string target) {
        static uint64_t label_count= 0;
        return {
                "    movq return_label_n"+std::to_string(label_count)+", (stack_instruction)\n",
                "    addq $8, stack_instruction\n",
                "    jmp "+target+"\n",
                "    return_label_n"+std::to_string(label_count++)+":"
        };
    }

    std::vector<std::string> generate_push_int32(int32_t target) {
        return {
                "    pushq $" +std::to_string(target)+ "\n"
        };
    }

    std::vector<std::string> procedure_operation::generate(const parser_context& ctx, const lexed_output& lexer_data) const {
        std::vector<std::string> ops;
        ops.emplace_back(name()+":\n");

        for(auto elem : _body) {
            auto token = lexer_data.dictionary.at(elem);
            if(auto result = try_parse_int32(token); result) {
                for(auto&& elem : generate_push_int32(result.value())) {
                    ops.push_back(elem);
                }
            } else if(auto op = ctx.lookup_operation(token); op) {
                for(auto&& elem : op->emit(ctx)) {
                    ops.push_back(elem);
                }
            } else {
                throw UnknownTokenError();
            }
        }

        ops.emplace_back("    // Return to caller\n");
        ops.emplace_back("    addq $-8, stack_instruction\n");
        ops.emplace_back("    movq (stack_instruction), %rax\n");
        ops.emplace_back("    pushq %rax\n");
        ops.emplace_back("    retq\n");

        return ops;
    }

    std::vector<std::string> procedure_operation::emit(const parser_context& ctx) const {
        return generate_call(name());
    }
}