sugar/src/molasses/parser_primitives.cpp

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

namespace molasses {
	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 type_input_error();
				} else {
					args.pop_back();
					current_stack.pop_back();
				}
			}
			if(not args.empty()) {
				throw value_missing_error();
			}
		}
		{
			auto return_types = next_op.return_types();
			std::move(return_types.begin(), return_types.end(), std::back_inserter(current_stack));
		}
		return current_stack;
	}

    auto conditional_begin_int_parse(const auto& str, auto& value, auto& begin, auto& end) {
        if(str.size() > 2 and std::string_view{begin, begin + 2} == "0x") {
            return std::from_chars(begin+2, end, value, 16);
        } else if (str.size() > 1 and std::string_view{begin, begin + 1} == "0") {
            return std::from_chars(begin+1, end, value, 8);
        } else {
            return std::from_chars(begin, end, value, 10);
        }
    }
	
    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 = conditional_begin_int_parse(str, value, begin, end);
        if(result.ptr == end) {
            return value;
        } else {
            if(std::string_view{result.ptr, end} == "_i32") {
                return value;
            }
        }
        return std::nullopt;
    }

    std::optional<int64_t> try_parse_int64(const std::string& str) {
        int64_t value;
        auto begin = str.data();
        auto end = str.data()+str.size();
        auto result = conditional_begin_int_parse(str, value, begin, end);
        if(result.ptr == end) {
            return std::nullopt;
        } else {
            if(std::string_view{result.ptr, end} == "_i64") {
                return value;
            }
        }
        return std::nullopt;
    }
	
	auto find_ptr_by_name_in_container(auto container, const auto& name) -> std::remove_cvref_t<decltype(*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 nullptr;
	}
	
	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(symbol.is_string) {
                type_stack.emplace_back("u8 ptr");
            } else if(auto is_int32 = try_parse_int32(symbol_text); is_int32) {
                type_stack.emplace_back("i32");
            } else if(auto is_int64 = try_parse_int64(symbol_text); is_int64) {
                type_stack.emplace_back("i64");
            } else if(auto is_op = parser_state.lookup_operation(symbol_text); is_op) {
				type_stack = type_stack >> *is_op;
			}
		}
		
		return type_stack == execution_output;
	}

	generate_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 register_pointer_type = [&](std::string full_ptr_type_name) -> void {
			if(auto type = ctx.lookup_type(full_ptr_type_name); !type) {
				ctx.types.push_back(std::make_shared<primitive_type>(std::move(full_ptr_type_name), architecture_ptr_size));
			}
		};
		
		auto compact_type_modifiers = [&](const std::vector<std::string>& type_info) -> std::vector<std::string> {
			std::vector<std::string> ret_val;
			for(const auto& elem : type_info) {
				if(elem == "ptr") {
					if(ret_val.empty()) throw type_expected_with_modifier_error();
					ret_val.back() += " ptr";
					register_pointer_type(ret_val.back());
				} else {
					ret_val.push_back(elem);
				}
			}
			return ret_val;
		};
		
		auto parse_proc = [&](auto it) -> std::pair<decltype(it), std::shared_ptr<procedure_operation>> {
#define check_for_unexpected_stream_end(expected, context)  \
            do{if(it == tokens.symbols.end()) { \
				throw expecting_token_error(expected, context); \
			}}while(false)
			
			if(*it != PROC_KW) {
				throw unexpected_token_error(*it, tokens.dictionary[*it],tokens.dictionary[PROC_KW]);
			}
			++it;
            check_for_unexpected_stream_end(tokens.dictionary[PROC_KW], details::concatenate_builder("In top level, file ",it->file_name, ":", it->line, ":", it->column));
			
			std::string name = tokens.dictionary.at(*it);
            auto& name_symbol = *it;
			++it;
            check_for_unexpected_stream_end("Procedure-Name", details::concatenate_builder("In top level, file ",it->file_name, ":", it->line, ":", it->column));
			
			// 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(tokens.dictionary[SEPARATOR_KW], details::concatenate_builder("Parsing procedure ",tokens.dictionary.at(name_symbol),", file ",it->file_name, ":", it->line, ":", it->column));
			}
			++it;
            check_for_unexpected_stream_end("Procedure-Argument-List", details::concatenate_builder("Parsing procedure ",tokens.dictionary.at(name_symbol),", file ",it->file_name, ":", it->line, ":", it->column));
			argument_types = compact_type_modifiers(argument_types);
			
			// 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(tokens.dictionary[DO_KW], details::concatenate_builder("Parsing procedure ",tokens.dictionary.at(name_symbol),", file ",it->file_name, ":", it->line, ":", it->column));
			}
			++it;
            check_for_unexpected_stream_end("Procedure-Return-List", details::concatenate_builder("Parsing procedure ",tokens.dictionary.at(name_symbol),", file ",it->file_name, ":", it->line, ":", it->column));
			return_types = compact_type_modifiers(return_types);
			
			// Process return types list
			std::vector<symbol> body;
			while(*it != END_KW) {
				body.emplace_back(*it);
				++it;
                check_for_unexpected_stream_end(tokens.dictionary[END_KW], details::concatenate_builder("Parsing procedure ",tokens.dictionary.at(name_symbol),", file ",it->file_name, ":", it->line, ":", it->column));
			}
			++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 procedure_stack_error();
			}
		}

		return {
            tokens,
            ctx,
            parsed_procedures
        };
	}

    std::vector<std::string> procedure_operation::generate(const parser_context& ctx, const lexed_output& lexer_data) const {
        std::vector<std::string> ops = generate_label(name());
	
	    for(auto&& instruction : generate_enter()) {
		    ops.push_back(instruction);
	    }
		
        for(auto elem : _body) {
            auto token = lexer_data.dictionary.at(elem);
            if(elem.is_string) {
                for(auto&& instruction : generate_push_string_ptr(elem)) {
                    ops.push_back(instruction);
                }
            } else if(auto result = try_parse_int32(token); result) {
                for(auto&& instruction : generate_push_int32(result.value())) {
                    ops.push_back(instruction);
                }
            } else if(auto result = try_parse_int64(token); result) {
                for(auto&& instruction : generate_push_int64(result.value())) {
                    ops.push_back(instruction);
                }
            } else if(auto op = ctx.lookup_operation(token); op) {
                for(auto&& instruction : op->emit(ctx)) {
                    ops.push_back(instruction);
                }
            } else {
                throw unknown_token_error(elem);
            }
        }
	
	    for(auto&& instruction : generate_return()) {
		    ops.push_back(instruction);
	    }

        return ops;
    }

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

    std::vector<std::string> generate(const generate_context& ctx) {
        std::vector<std::string> generated;

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

        for(const auto& proc : ctx.procedures) {
            for(const auto& instr : proc->generate(ctx.parser, ctx.lexer)) {
                generated.push_back(instr);
            }
        }

        std::set<int> done;
        for(const auto& value : ctx.lexer.symbols) {
            if(value.is_string && not done.contains(value.id)) {
                for(const auto& instr : generate_string(value, ctx.lexer.dictionary.at(value.id))) {
                    generated.push_back(instr);
                }
                done.insert(value.id);
            }
        }

        return generated;
    }
}